(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
See other formats





Management and Application o the ' 1 

Nondestructive Testing Labor ing ...... 2 

Field and Maintenance Tests . JJJJ ...... 3 

Fundamental Testing Principle a n ...... 4 

Basic Test Methods . . . ena ....... 5 

Liquid-PenetrantTestPrincipl-'tjes, . ..... 6 

Liquid-Penetrant Test Equip * ..... , 7 

Liquid-Penetrant Test Indicf ' ...... 8 

Filtered-Particle Tests. . Iarge . ...... 9 

Vision and Optics . . . all the , ..... 10 

Visual Inspection Equlpmer 1 fo a ...*. 11 

Optical Projectors and Con. .~ ..... ... 12 

Radiation and Particle Phytallurgy, ..;.* 13 

Electronic Radiation Sourcifll, civil, .,,-. 14 

Isotope Radiation Sources^ 8 tt f .,..--. 15 

Radiation Detection and R<S ....... W 

X-Ray Diffraction and Flue ' ...... 17 

X-Ray and Isotope Gaging ,ff erc d ,,.,.. 18 

Fluoroscopy and X-Ray Ima ( ncnta! s .,,.. 19 

Film Radiography . . . ^ ...... 20 

X-Ray Film Processing , rjj ...... 21 

Xeroradiography . . . ' test .* ^ 

High Voltage Radiography .cof ...... 23 

X-Ray Interpretation JJJJ ...... 24 

X-Ray Control of Weldments , j s j, ,,,.., 25 

Radiation Protection ... 'or ...... 2* 

Attenuation Coefficient Tables k ...... 27 


MAI OCT181938 




Electrified-Partiele Tests ...... : . ... 28 

Electrified-Particle Test Indications ....., 29 

Magnetic-Particle Test Principles 30 

Magnetic-Particle Test Equipment . 31 

Magnetic-Particle Test Indications * 32 

Magnetic-Field Test Principles ,..:.... 33 

Magnetic-Field Test Equipment 34 

Electric Current Test Principles > 35 

Eddy Current Test Principles 36 

Eddy Current Cylinder Tests. 37 

Eddy Current Tube Tests . 38 

Eddy Current Sphere and Sheet Tests , 39 

Eddy Current Test Equipment ,40 

Eddy Current Test Automation 41 

Eddy Current Test Indications ...... 42 

Ultrasonic Test Principles * , * . . 43 

Ultrasonic Transducers , 44 

Ultrasonic Fields * 45 

Ultrasonic Immersion Tests 46 

Ultrasonic Immersion Test Indications ...... 47 

Ultrasonic Contact Tests ............ 48 

Double-Transducer Ultrasonic Tests ...... , . 49 

Ultrasonic Resonance Tests .*.<...,, 50 

Natural Frequency Vibration Tests . . 51 

Brittle-Coating Tests 52 

PhotoclasUo-Coating Tests 53 

Resistance Strain-Gage Tests 54 





Nondestructive touting 

V.2 60-lOl>. : 





Edited for the Society for Nondestructive Testing 




In Two Volumes 



Copyright, , 1959, by 

All Rights Reserved 

No part of this book may be reproduced 

in any form without permission in writing 

from the publisher. 

Library of Congress Catalog Card Number: 59-14660 



Volume II 



































Principles of Test 

Development I 

Advantages and limitations 1 

An elect rifled -particle indication of a crack 
pattern on u porcelain -enameled -sink 

(/, 1) 1 

Major applications 2 

Mechanisms of operation 2 

An electrified -particle gun in operation 

t/.2) 2 

Operation on metal-hacked coatings 2 

A typical elect rified -particle indication ex 
hibiting two-directional stress cracks in 

M porcelain -enamel plate (/, 3) 3 

Operation with liquid penetrant 4 

The triboolectric effect 4 

Electrical field conditions 4 

Nonconducting materials, metal-backed 4 

Electrostatic situation on metal-bucked 
ceramic material prior to application of 

charged powder (/, 4) 4 

Metal-backed ceramic material after ap 
plying charged powder (/. 5) 5 

Porcelain -enamel sample containing a 
crack, after charged powder is applied 

I/.8) 5 

Electrostatic situation after indication is 

built np(/, 7) 6 

Limitations 6 

Nonconducting materials, not metal-backed , . 6 
The electrostatic situation in glass after 
application of penetrant. but prior to 
application of charged powder (/. 8) ,., 6 

Charge movements in penetrant 6 

Effect of application of charged powder to 
sample shown in Fig, 8 (/. 0) 7 

Subsurface indications 

Effect of application of charged powder to 
side not containing defect (sample shown 

in Fig, 8) (/. 10) 

Thin nonconductors 

Nonconducting materials in close proximity to 

Equipment and Materials 

Test materials 


Advantages of calcium carbonate , 


Test equipment 

Powder gun 

Gun -operating instructions 

Technique for use of gun 

Air supply 

. 7 

Control Techniques 

Electrical charging effects 9 

Influence of atmospheric humidity 10 

Environmental powder distribution 10 

Atmospheric water vapor 10 

Water deposition on test objects 10 

Use of heated penetrant 10 

Drying parti surfaces 11 

Excessive test-object temperatures 11 

Loss of indications 11 

Fugitive indications 11 

Standard crack sample 11 

Preparation of cruck samples 11 

Bibliography 12 



Principles of Test 

DEVELOPMENT. The electrified-particle inspection method was invented 
by deForest and Staats in 1945 for the specific solution of a problem concerning 
cracks in glass bottles, too small to be seen by the naked eye. It is known as 
Statiflux (Magnaflux Corporation: method and equipment are covered by one 
or more of U.S. Patents 2,515,396; 2,499,466; 2,499,462; and comparable 
Canadian patents). During development, other uses were found which have 
played a more profound role in industrial nondestructive testing. 

ADVANTAGES AND LIMITATIONS. The electrified-particle method 
works on all nonconducting materials. Representative examples are glass, non- 
porous ceramic materials, nonconductive plastics, and certain paint films. On 
some classes of ceramic materials, cracks are located with a sensitivity com 
parable to penetrant methods. However, on those materials whose surfaces are 
glossy or glazed, the method goes far beyond the ordinary sensitivity of all 
penetrant techniques. The method has the remarkable ability of enabling the 

Magnaflux Corporation 

Fig. 1. An electrified-particle indication of a crack pattern on a porcelain- 
enameled sink. A portion of the indication has been removed to illustrate that the 
actual cracks are completely invisible. 



location of fine cracks which otherwise are completely invisible (as shown in 
Fig. 1). On such cracks, indications are built up which give magnification fac 
tors of approximately 30,000 to 1. An invisible defect, lying perpendicular to 
the surface in a transparent material does not reflect light to the human eye. 
According to the laws of optics, an interface of this width is less than four 
millionths of an inch in width. 

MAJOR APPLICATIONS. The electrified-particle inspection method lias 
three major applications: (1) on porcelain enamel, (2) on glass-to-metal seals, 
and (3) on various materials such as artificial teeth, glass-lined vessels, and other 
nonconducting materials. 

MECHANISMS OF OPERATION. The method depends upon a tribo- 
electric effect, or production of static electricity by rubbing or tearing action. 
Particles of calcium carbonate are blown through a spray gun with a special 
nozzle constructed of hard rubber (Fig. 2). Calcium carbonate lies higher than 

Magnaflu?c Corporation 

Fig. 2. An electrified-particle gun in operation. Note that the positively charged 
particles tend to fly back to the negatively charged gun. 

hard rubber in the triboelectric series and therefore becomes charged positively. 
While many materials exhibit this property, none become charged so readily as the 
particular form of calcium carbonate supplied commercially. During the initial 
development more than 200 varieties of calcium carbonate and other materials 
were tested. Only a few exhibited- 1 the property of forming clean-cut indications 1 
with little background. 

Operation on Metal-backed Coatings. The method has two general modes 
of operation. The first applies where the material to be tested is backed with 
metal, such as porcelain enamel on steel. In this instance the powder emerges 
from the gun charged positively and approaches the porcelain enamel. Electrons 
within the nonconductive glassy layer are probably attracted to the surface, 
but they do not enter the mechanics of the defect location. However, the elec- 


trons in the metal layer are attracted to the interface between glass and metal 
and set up a state of electrical strain between the interface and the surface of 
the porcelain enamel When a defect is present in the glass layer, the dielectric 
strength is much lower at this point. The electrons tend to leak through the 
point of low dielectric strength. For all intents and purposes the crack becomes 
charged negatively. Since positively charged particles are attracted to negative 
areas, a physical indication of appreciable size will build up (Fig. 3). 

Magnaflux Corporation 

Fig. 3. A typical electrified-particle indication exhibiting two-directional stress 
cracks in a porcelain-enamel plate. 



Operation with Liquid Penetrant. The second mode of operation is used 
when a nonconductive surface is not backed with metal and it is impossible to 
locate the defect unless an electron supply is provided in back of or within the 
defect. This situation is handled very effectively by applying a liquid penetrant 
to the material. The penetrant consists of water treated with a highly active 
wetting agent and other substances which provide a good negative-ion source. 
Penetration occurs instantly on glass. After drying, charged powder is applied 
to the surface and an indication develops instantly. In this case the positively 
charged particles attract electrons within the defect. 

THE TRIBOELECTRIC EFFECT. Electrification by friction or contact 
separation is termed triboelectrification. This effect can be demonstrated _ by 
the sudden separation of two contacting pieces of matter. During electrification, 
electrons and protons are neither created nor destroyed. They are merely sepa 
rated or redistributed. It is a fundamental fact that, for every positive charge, 
there must be a corresponding and numerically equal negative charge developed 
somewhere by the same action. A triboelectric series is a list of substances 
arranged in such order that a material higher in the list is postively charged when 
rubbed against a material which is lower in the list. The further apart the mate 
rials are in the series, the more readily they seem to lose or gain electrons. 

The electrified-particle inspection method depends upon the ease with which 
a specific type of calcium carbonate loses its electrons when placed in intimate 
rubbing contact with hard rubber. The powder is unique, since it appears to lose 
its electrons much more readily than any of several hundred varieties of materials 
tested. The location of these two materials at the extreme ends of the tribo 
electric series undoubtedly has something to do with the efficiency of the method. 

ELECTRICAL FIELD CONDITIONS. When an uncharged body is 
brought into an electrical field, a rearrangement of the charges in the body always 
results. If the body is a conductor, the electrons are mobile and free to move. 
If the body is a nonconductor, the electrons associated with the positive nuclei 
in each molecule are displaced or reoriented. The electrons within the body are 
attracted or repelled, depending upon the electrical sign of the external field. The 
movement of electrons in various materials makes it possible to locate cracks 
under certain conditions. In general the method is applicable only to noncon 
ducting materials or coatings. 

static situation on metal-backed ceramic material, such as porcelain enamel, is 
substantially as shown in Fig. 4. When positively charged powder is applied to 



+- - 


Fig. 4. Electrostatic situation on metal-backed ceramic material prior to applica 
tion of charged powder. 



the glass surface, the electrons in both enamel and metal tend to be attracted 
by the charged powder. The nonconductive enamel will allow its electrons to 
reorient but not to flow freely. The conductive metal permits its electrons to 
flow in the direction of the positive potential on the surface of the enamel, as 
illustrated in Fig. 5. If a crack exists in the enamel, the dielectric strength 

* 4 

Fig. 5. Metal-backed ceramic material after applying charged powder. 

of the glass is much less at that point. The negative potential built up in the 
interface will tend to leak through and influence the field in the defect area, as 
shown in Fig. 6, Once this influence is established, additional positively charged 


Fig. 6. Porcelain-enamel sample containing a crack, after charged powder is 
applied. Note leakage influence at defect. 

particles accumulate at the defect, forming a highly visible powder indication, 
as shown in Fig. 7. The fact that electrons flow to the interface may be readily 
verified by testing the metal backing with an electroscope. When positively 
charged powder is applied to the glassy surface of a glass-metal combination, the 
metal always becomes positive. 



Fig. 7. Electrostatic situation after indication is built up. 

Limitations. Decisive indications of cracks are sometimes difficult to form on 
extremely thin, nonconducting, metal-backed materials. Electrons in the 
base metal tend to leak through and influence the surface of the nonconductor to 
the same degree as the defect. In other words, the differential electrostatic situa 
tion is much reduced. Nonconductive coatings which have low dielectric strength 
tend to break down when charged powder is applied to the surface. This break 
down exhibits itself by the formation of pinholes, with a subsequent formation 
of dotlike indications. 

glass does not have available a mobile electron source, it is necessary to provide 
one by introducing electrons into the crack or defect. This is accomplished by 
using liquid penetrant. The penetrant consists mainly of water, wetting agent, 
and other materials to provide a slight conductivity. When a glassy object is 
dipped into penetrant and removed and dried, the crack traps an infinitesimal 
quantity of penetrant (Fig. S). 


Fig. 8. The electrostatic situation in glass after application of penetrant but prior 
to application of charged powder. 

Charge Movements in Penetrant. When positively charged particles aro 
applied to a surface containing a crack, negative ions in the penetrant tend to 
migrate to the top of the crack. The base of the crack becomes charged posi 
tively. Since a negative field is present at the top of the crack, it will attract 
positively charged particles and provide a visible powder indication (Fig. 9). 


28 7 

Fig. 9. Effect of application of charged powder to sample shown in Fig. 8. 

Subsurface Indications. On thin materials, where a crack exists on one face 
and is filled with penetrant, an indication can be developed on the opposite or 
crack-free side. This powder indication has a fuzzy and indeterminate appear 
ance, due to the relatively poor leakage field present. When subsurface indica 
tions are formed, the electrons will be found at the bottom of the crack, as shown 
in Fig. 10. 

Fig. 10. Effect of application of charged powder to side not containing defect 

(sample shown in Fig. 8). 

THIN NONCONDUCTORS. On relatively thin, nonconducting materials 
such as window glass, totally invisible cracks can be located without the use of 
electron-bearing penetrants. Metallic foil or powder, conductive liquids, or the 
human hand placed in back of or below the crack will provide the necessary 
electrons and influence the site of a defect. The effect of the human hand on 
thin, nonconducting materials is most interesting. When the hand is placed 
below and in intimate contact with glass, an outline of the hand will appear 
when charged powder is applied to the glass surface. When the hand or other 
electron-supplying materials are placed below a similar specimen containing a 
crack, indications will form without the aid of penetrants. To verify this effect, 
it is necessary to dry the glass at elevated temperatures to remove water or 
penetrant from the crack. Sometimes atmospheric moisture vapor will enter a 
crack and simulate the action of deliberately added penetrant. 


METALS. Some nonconductive materials have metal in close proximity or 
associated with a crack in glass. The electron-bearing substance may not neces 
sarily be in back of a glassy portion, as is the case with glass-to-metal seals. 
Assuming that known cracks in the glassy portion are completely free of atmos 
pheric vapor and water-based penetrating materials, it is frequently possible to 
develop indications by relying upon nearby metallic parts to supply electrons. 
If useful indications are not produced, penetrating materials must be used. 

Equipment and Materials 

TEST MATERIALS. Suitable powder and, in some cases, liquid-penetrant 
materials are expended during electrified-particle testing. 

Powder. The powder used is a special variety of calcium carbonate commer 
cially available in three colors: white, gray, and black. The gray and black 
powders contain a small percentage of carbon black. The amount of colorant 
affects the charging ability, since the higher the contamination, the lower the 
sensitivity. Of the three colors, white is the most sensitive; black, the least. The 
choice of colors available for blending is limited to those materials which tend to 
become charged on their own account. In addition the particles must be of the 
right size and shape. Most industrial users prefer white powder, since it is the 
least noticeable when left on an inspected piece. The next preference is for light- 
gray powder since it provides a moderate amount of contrast and is not so 
untidy as black. At first glance a white or gray powder indication on a white 
background would appear to have poor contrast and be difficult to see. However, 
tangentially lighted indications cast striking shadows. 

Advantages of Calcium Carbonate. Many varieties of calcium carbonate 
tend to exhibit the electrostatic properties necessary to produce indications. 
Other materials, such as talc, ordinary household flour, and even powdered 
sugar, will develop usable indications when charged appropriately. Despite this, 
when commercially available powder is compared with other materials, the 
results are ^strikingly in its favor. It is completely safe to use, since it cannot 
produce silicosis and will not explode in an air mixture. It has no appreci 
able odor or taste and is completely nonhygroscopic. Very few other materials 
have all these desirable characteristics. 

Penetrant. The penetrant which is commercially available is an especially safe 
and effective concentrate of wetting agents, water, and other materials to guar 
antee a slight amount of conductivity. In use, it is diluted with water. For high 
effectiveness, it is recommended that the penetrant be heated. Penetrant is 
applied by dipping, spraying, or brushing. After application it is allowed to air- 
dry or is wiped with paper toweling. 

TEST EQUIPMENT. The electrified-particle test method requires only 
simple equipment for charging and applying the inspection medium. 

Powder Gun. The powder gun (Fig. 2) is light and simple to use. It consists 
of the basic gun, cup, and charging-nozzle assembly. The charging-nozzle 
assembly is the most important element and consists of a hard-rubber insert, 
metal sleeve, and assembly ring. The hard-rubber insert has an extremely small 
bore, to assist in the development of exceptionally high electrostatic potentials 
When the gun is used repeatedly in service, the bore enlarges. It should be dis- 


carded when inadequate indications are developed on standard crack samples. 
However, a nozzle is still usable up to the point where the bore increases to 
approximately %e in. in diameter. 

Gun-operating Instructions. In use, the cup is filled to half-capacity. The 
gasket and the top edges of the cup are carefully cleaned to assure proper seat 
ing. An accumulation of powder on the gasket prevents the adequate tight 
ening of the cup to the gun. This permits air and powder leakage during use 
and is annoying. A small amount of lubricant applied to the threads of the cup 
assists in the tightening of the gun assembly. 

The gun is connected to a pressure-regulated air supply. Air pressure is not 
critical and is a matter of personal preference, usually ranging from 15 to 25 p.si. 

Two adjusting screws control the air supply. One is located under the trigger 
and controls the amount of air passing through the gun. The setting of this screw 
is a matter of choice. The second adjustment screw is on top of the gun (Fig. 2) 
and splits the air flow between cup and nozzle. When this screw is turned clock 
wise, it reduces the amount of powder introduced into the air stream. It gener 
ally is easier to select the desired air pressure first and then turn the adjustment 
knob to the point where no powder is introduced into the air stream. Using a 
standard test sample, the knob is slowly turned counterclockwise until the 
desired amount of powder is emitted from the nozzle. A slight shaking motion 
of the hand assists the gun in developing the proper powder cloud. 

Technique for Use of Gun. During use, powder tends to lodge within the 
nozzle and on its external surfaces. After a gun has been turned off, this powder 
tends to break loose and blow or drop onto the test piece. This can be irritating; 
however, it rarely produces patterns which are mistaken for genuine defect 
indications. The gun should be held approximately 1 to 3 in. away from the 
surface and shaken gently during use. Some inspectors prefer to blow powder 
in short bursts rather than in one long sequence. 

Air Supply. Portable air compressors can be used to develop adequate air- 
pressure. Pressure-regulated commercial air supplies in most industrial plants 
are adequate for use in spraying powder, although if not blown out occasionally, 
they tend to accumulate drops of water. Occasionally these droplets will ride 
through the line and subsequently through the gun. This has little or no effect 
upon the powder or the development of good indications, providing that exces 
sive moisture is not blown directly on the test piece. Despite this, it is good 
practice to blow out air lines prior to use. Most commercially available devices 
do not provide for moisture condensation. On extremely humid days a com 
pressor takes in moisture-laden air, compresses it, and blows moisture through 
the gun. When this situation is encountered, adding a water trap or silica gel 
drier to the line is a necessity. 

Compressed-air lines may also contain oil droplets which originate from the 
lubrication of the compressor. Oil droplets entering or emerging from the powder 
gun have no effect upon the performance of the gun or subsequent development 
of indications. Excessive oil, however, may prove troublesome and untidy. 

Control Techniques 

ELECTRICAL CHARGING EFFECTS. Under some operating condi 
tions, the operator will receive slight electric shocks from the gun or test piece 
during inspection. This is understandable, since the voltages developed by 


blowing powder through the gun may be as high as 10,000 volts. Little current 
is involved, so that there is little danger to the operator. However, the clement 
of surprise may make him cautious in using the equipment. A ground clip and 
wire are provided on the gun for connection to a nearby ground. ^This tends to 
drain of! the negative potential on gun and operator, thus reducing the possi 
bility of shock. Generally speaking, little trouble is noted on this account when 
the operator is working on a slightly conductive floor such as concrete. Wood 
blocks, asphalt, or rubber covering on floors insulate the operator from ground, 
thus enabling him to build up a potential which may produce an electric shock. 
The aforementioned ground clip assists in the reduction of this potential. 

Influence of Atmospheric Humidity. A shock hazard is still present despite 
the grounding of the gun or operator because the test part itself becomes 
charged. In dry weather it is possible for large objects to receive and hold a 
charge of surprisingly high potential. Ordinarily these will bleed off within a 
few seconds, but in very dry weather the potential lasts several minutes. Ground 
ing of the object itself will help somewhat; however, if the operator knows that 
a shock potential exists, he can avoid touching the part for a short while. 

Environmental Powder Distribution. Excessive powder may be blown and 
coat the operator and surrounding area. This is not hazardous, but it is a 
nuisance. A simple way to avoid this condition is to apply powder in a paint 
spray booth or near an exhaust system. Often it is convenient to use an ordinary 
vacuum cleaner and work with the nozzle near the spot where powder is being 
applied. This is particularly advantageous on large objects which are not con 
venient to place within a booth. 

ATMOSPHERIC WATER VAPOR. Practically all materials will collect 
a microscopic film of moisture on their surfaces under certain atmospheric condi 
tions. This is a natural phenomenon, and ordinarily the film is invisible and 
very thin. On ceramic materials the film appears to dissolve out soluble por 
tions, releasing ions and rendering the surface slightly conductive. When charged 
powder is applied, the sample acts as though it were a metal rather than a non 
conducting material. 

Water Deposition on Test Objects. The amount of water deposition on 
materials is governed by the amount of water vapor in the air. During high 
temperature and humidity conditions, it is possible for much more moisture to 
be in the air than during cold weather. A conductive film of moisture need not 
interfere with effective inspection since it can be removed quite readily. When 
the object is warmed to slightly above room temperature, the water layer will 
evaporate. On large structures such as refrigerator liners, range tops, or glass- 
lined process vessels, this may be very awkward. Under these circumstances, 
commercially available infrared lamps can be strategically placed to warm the 
specific area under test. 

Use of Heated Penetrant. Another useful technique for adding heat to a 
glass surface is to apply heated penetrant. Despite the fact that penetrant is not 
required in the inspection of metal-backed nonconducting materials, its use tends 
to help build sharper indications. If the penetrant is warmed to 140 F. to 
150 F., the heating effect of the penetrant will serve to warm the specimen at 
the same time as the penetration takes place. In addition, a warm specimen is 
relatively easy to dry with paper towels. Once the visible moisture film is 
removed, heat retained by the specimen will drive off the microscopic moisture 


layer. The fact that a glass surface may be warm to the touch is not always 
sufficient evidence that the surface is dry. It is possible for glass to be exposed 
to sunlight and still have extensive water deposition. 

Drying Part Surfaces. Troublesome water-vapor condensation can sometimes 
be removed by wiping the test object with fresh paper toweling and applying 
powder before a new layer deposits. Cloth toweling is not recommended because 
it absorbs moisture and retains it from one sample to the next. Paper toweling 
does not pick up atmospheric moisture vapor readily and will leave few or no 
fibers on the surface. Lint, fibers, or threads sometimes can be confused with 
actual indications. When used to dry surfaces, paper toweling usually is dis 
carded rather than re-used, as is the case with cloth. Many inspectors keep 
paper toweling in one hand and a gun in the other. If an indication is unsatis 
factory, it can be removed quickly and the test repeated. Condensation of 
moisture on glassy surfaces is troublesome during moist, humid weather and the 
summer months. In most parts of the United States, little or no trouble is 
experienced from September to June. 

are warmed in excess of ISO F., electrified particles will precipitate indiscrimi 
nately on the surface. Indications of defects will not appear. A rough gage of 
the correct temperature for applying powder is whether or not the hand can be 
placed on the sample and left in place. A temperature of 140 F. is about the 
maximum which the average human hand can tolerate; defects can be located 
at this temperature. On overheated test pieces, powder blankets the specimen 
and the condition is readily recognized. Normally, indications develop instanta 
neously and excess powder bounces off the defect-free portion of the test piece. 

LOSS OF INDICATIONS. When inspecting large objects like glass-lined 
vessels, it is not uncommon to look back over an area which has just been 
processed and to locate cracks which apparently were missed. Charged powder 
may drift to a crack somewhat out of the field of vision, and the working area 
of the drift powder may be several feet in any direction from the end of the 
gun, Since the normal field of precise vision is less than a foot in diameter, it 
is understandable that indications can be missed. 

Fugitive Indications. Sometimes an indication is fugitive. This phenomenon 
sometimes occurs in highly humid weather. Also, when the impingement force 
of charged particles is higher than the electrostatic forces, indications may refuse 
to stay in place. The air pressure may be reduced until the electrostatic forces 
are sufficiently high to overcome the physical sandblasting effect of the parti 
cles. Sometimes indications refuse to stay in place because the gun is held too 
close to the part. Conversely, if the gun is held too far away from the work, the 
indications become reluctant to form. Experimentation usually will indicate 
optimum test conditions. 

STANDARD CRACK SAMPLE. It is recommended that a standard crack 
sample be created before any inspection work is performed. Because of the possi 
bility that humidity may affect the inspection performance, it is necessary that 
a- known crack sample be available to the inspector for occasional comparison. 
This will assist and convince the operator that the method is always operative. 

Preparation of Crack Samples. Standard crack samples are easy to make. 
Porcelain enamel can be heated and quenched in water. This produces a system 
of extremely tight cracks. This same procedure can be used on most ceramic 


materials. A little experimentation is worthwhile, considering the importance 
of the inspection procedure. In many plants where inspection is performed daily, 
a standard crack sample is hung nearby for occasional spot comparison. In any 
situation where there is a possibility that no cracks will be found, it is always 
helpful to use the standard sample to convince the inspector that his procedure 
is correct. 


For Bibliography, see section on Electrified-Particle Test Indications. 





General Interpretation 


Sensitivity of indications 1 

A transparent piece of glass containing a 

crack (/.I) 1 

Powder indication of the crack shown in 

Fig. 1 (/. 2) 2 

The same glass sample with powder re 
moved (/. 3) 2 

Comparison with liquid penetrants 3 

Types of indications 3 

Indications of subsurface defects 3 

A typical subsurface defect in clear glass 

.(M) 3 

Faint indications 4 

Pinholes 4 

Blinking indications - 4 

Pinhole breakdown 5 

Cracks in thin, metal -backed coatings 5 

Thin paints and resins 5 

Negative indications 5 

Indications on secondary surfaces 5 

Test objects immersed in oil 8 

Oil films as moisture barriers 6 

Blinking indications on secondary surfaces . . 6 

Precautions in explosive atmospheres 6 

Indications on nonhomogeneous materials 6 

False indications 6 

Powder blotting 6 

Residual powder artifacts 6 

Chemically induced artifacts 7 

Loss of indications on slightly conductive 

materials 7 

Mechanisms of Failure of Glass and 
Porcelain Enamels 

Crack growth characteristics 

Causes of cracking in glass 

Factors contributing to the growth of 

cracks in glass (/. 5) 

Mechanisms of porcelain -enamel failure 

Cracks yield engineering information 

Direction of stress 

Crack formation 

Compression failures 


A porcelain -enamel failure due to com 
pression (/. 8) 9 

Tension failures 9 

A tension failure on porcelain enamel (/. 7) 10 

Cyclic loading failures 9 

A porcelain- enamel failure, illustrating 
how cyclic loading contributes to failure 

(/.8) 10 

Industrial Applications 

Inspection of glass containers 11 

Typical indications 11 

Rim defects in glass bottles (/. 9) 12 

Earlier methods for inspection of coatings .... 11 

Brine test 11 

Limitations of brine tests 13 

Modified tesla coil test 13 

Electrified-particle inspection of coatings 13 

Craze patterns 13 

Indications of a craze pattern on a typical 

glass-lined fitting (/. 10) 14 

Stress cracks 14 

Wavy cracks 14 

Pinholes 15 

Cracks through pinholes 15 

Indications from chemical residues 15 

Inspection of glass -to-metal seals 15 

Effects of cracks in seals 15 

Small headers which have been removed 
from equipment, both having identical 

cracks on top and bottom (/. 11) 18 

Electrical leakage 16 

Intermittent faults 16 

Types of cracks 17 

Sketches of typical crack patterns on 

glass-to-metal seals (/. 12) 18 

Evaluation of cracks 17 

Miscellaneous applications 17 

Porcelain teeth 17 

Terracotta 19 

Indication of defect in glaze on terra cotta 

(/. 13) 19 

Glass lenses 20 

Indications of superficial scratches in a 

lens (/. 14) 20 

Brittle- coating tests 20 

Bibliography 21 




General Interpretation 

SENSITIVITY OF INDICATIONS. The electrified-particle inspection 
method is a unique and powerful tool for evaluating the width of a defect in 
nonconducting materials, since it will locate cracks in clear glass which are 
completely invisible to the human eye. When the dimensions of a crack or 
interface in a transparent object become less than four millionths of an inch in 
width, no light is reflected because of an interference phenomenon. A simple 
experiment illustrates this point. Fig. 1 shows a transparent piece of glass con 
taining a crack. The crack is visible because it reflects light to the eyes of the 
observer. When processed with electrified particles, a visible powder indication 
develops as shown in Fig. 2. The powder indication is considerably more exten 
sive than the appearance of the crack to the eyes. If the sample is carefully bent 
to open the crack, the true crack length is now exposed, as seen in Fig. 3, and 

Magnaflux Corporation 
Fig. 1. A transparent piece of glass containing a crack. Crack is visible because 

of reflected light. 

29 1 

29 2 


Magnaflux Corporation 

Fig. 2. Powder indication of the crack shown in Fig. 1. The sample is reversed 
because the crack is in the opposite side. 

Magnaflux Corporation 

Fig. 3. The same glass sample with powder removed. Sample is being bent to 
open the crack so that reflected light can be seen from all portions of the defect. 


agrees exactly with the previous indication. This clearly illustrates the fact that 
cracks, which are four millionths of an inch in width or narrower, can be detected 
by the electrified-particle method. 

Comparison with Liquid Penetrants. Other methods for detecting cracks 
in ceramic materials rely upon introducing colored or fluorescent materials into 
a crack. After excess material is removed, blotting agents are applied to draw 
the material from within the defect and make it visible (see section on Liquid- 
Penetrant Tests) . If the material is transparent, it is possible to detect color- or 
fluorescence-bearing penetrant in the crack down to approximately four millionths 
of an inch in width. At this point the interference phenomenon takes place, 
and despite the presence of material within the crack, no reflected light will be 
emitted from the defect. 

TYPES OF INDICATIONS. The size of powder build-ups on a crack is 
a rough indication of the size of the underlying defect. Generally, large indica 
tions indicate large defects. A wide crack may have a powder build-up which is 
the same size as one formed at the site of a deep defect. Therefore it is difficult 
to tell whether a crack is wide, exceptionally deep, or both, by the superficial 
appearance of an indication. Indications can be developed which are as large as 
% in. in width. A quick calculation will show that this is 30,000 times larger 
than an invisible crack in glass. This illustrates the striking magnifying ability 
of charged powder. 

Indications of Subsurface Defects. The shape of an indication reflects some 
of the characteristics of a crack. Normal indications are almost perfectly half- 
cylindrical in cross-section. Cracks occurring in porcelain enamel sometimes 
branch out at the base of the notch, split, and form two cracks completely below 

Magnaflux Corporation 

Fig. 4. A typical subsurface defect in clear glass. The same piece of glass is illus 
trated in Fig. 1. 


the surface. The indication will reflect this condition on the surface Indications 
of cracks not perpendicular to the surface appear as a shadow or as graduated 
powder build-ups. These usually are fuzzy and less clean-cut. Indications are 
broad and poorly denned on thin specimens where a crack is on one surface and 
not the other. A typical subsurface indication of a defect in glass is shown in 
Fig. 4. Subsurface defects can be located if they are not more than % in. below 
the surface. However, due to the wide difference in dielectric properties of non 
conducting materials, it is not always possible to predict the limit of this measure 
ment. Subsurface defects, including bubbles, can be located only when a defect 
contains negative ions or has an electron supply nearby. 

Faint Indications. It is possible to distinguish between cracks existing in rela 
tive states of tension and compression in thermally shocked enamel. Indications 
of cracks in a high state of compression lack fidelity and form hesitantly. Tight 
cracks provide a weak leakage field. The weaker the field, the poorer the indica 

Indications are sometimes difficult to develop on old or weathered porcelain 
enamel. It has been theorized that the difficulty is due to the infinitesimally 
slow dissolving of the glassy coat in atmospheric water vapor. The inner sur 
faces of cracks in glass are also affected and may encourage crystal growth. This 
may account for their apparent self-sealing ability and resultant faint indica 
tions. The role of atmospheric moisture vapor is extremely significant in relation 
to certain older porcelain enamels. When exposed to moisture vapor for pro 
longed periods of time, soluble portions of the coating leach out, leaving traces 
of moisture within the vacated voids. As leaching progresses, the bubble struc 
ture becomes exposed. Each of the vacated voids then may hold an infinitesimal 
amount of moisture. When a crack occurs in a coating of this type, it is often 
impossible to develop an indication, since the powder is likely to indicate the 
moisture in the exposed bubble structure. In this case, powder deposits as a 
blanket rather than as individual indications. If the sample is warmed until all 
moisture is driven off, indications develop in normal fashion, and the larger pores 
are indicated separately. 

PINHOLES. Inspection for through-to-metal pinholcs on metal-backed 
material is of real importance to the manufacturer of glass-lined process vessels 
and porcelain-enameled goods which are exposed to moisture or corrosive 

Blinking Indications. An indication of a through-to-metal defect may act 
mysteriously and appear to "blink." An indication will build up and blow away, 
whereupon a new indication will form and it, too, will blow away. This sequence 
usually occurs so rapidly that the indications appear to pulsate or blink. The 
blinking effect is caused by the build-up of positive particles directly over the 
breakthrough, with a resultant high negative field occurring just below in the 
metallic layer. When the strain between positive and negative potentials roaches 
a critical point, electrical breakdown takes place. Tiny flashes of light occur 
at the moment of breakdown and confirm that a spark occurred. Once the 
breakdown has occurred, the oncoming material forcibly removes the original 
indication. The initial indication is easily removed, since the electric forces 
which held it in place have been disposed of by the electrical breakdown. Once 
the indication is removed, a new one forms immediately and begins to enlarge 
until it reaches a critical point, whereupon it also breaks down electrically and 
is blown away. The indication does not always blow away completely but 


appears to enlarge and retract in size, giving the impression that it is pulsating. 
The pulsation rate may be as high as 25 c.p.s. 

The blinking effect does not occur if the through-to-metal pinhole is filled 
with conductive material such as water or penetrant. An exposed metallic 
inclusion in the surface of a nonconducting material will act similarly. In these 
cases a powder indication will build up but not pulsate, since an electrical break 
down does not occur. 

Pinhole Breakdown. It is possible to develop d.-c. voltages high enough to 
cause breakdowns of thin coatings. Many coatings contain bubbles which may 
be as large as half or more of the total coating thickness. Such things will not be 
indicated unless the gun is held in one place and a serious attempt made to 
cause a breakthrough and form an indication. A casual movement of the gun 
while blowing powder will rarely break down most coatings, no matter how 
thin they are. Breakdown usually occurs when the gun is held in one place with 
a steady stream of highly charged particles impinging upon the same location. 
The possibility of breakdown of a coating can be reduced somewhat by apply 
ing a sheet of vinylite over the surface in order to increase the total thickness 
of the dielectric film. 

enameling industry has exhibited a trend toward thinner coatings. As a result 
the incidence of cracks in commercial products appears to be much reduced, 
although not eliminated. As the thickness of porcelain enamel diminishes, a 
problem of crack interpretation arises. Generally speaking, the thinner the 
coating, the closer together a group of cracks will be. On exceptionally thin 
coatings (3 mils or less), cracks can occur so closely together that the respective 
indications tend to overlap. 

Thin Paints and Resins. It is difficult to develop a sizable indication of a 
crack on exceptionally thin coatings without increasing the background appre 
ciably. This effect becomes more pronounced as the dielectric strength of the 
coating material is reduced, as with some paints and resins. An indication forms 
because of an electrostatic differential between an area which contains a crack 
and one that does not. As the dielectric strength of the insulating material is 
reduced by a thinning or alteration of its electrical characteristics, the electro 
static differential diminishes to a point where an indication cannot form. 

NEGATIVE INDICATIONS. Under certain atmospheric conditions, pow 
der indications sometimes appear reversed. The crack is indicated by a lack of 
powder at the defect. The powder build-up occurs on both sides of the defect, 
leaving the area directly over the defect free of powder with the possible ex 
ception of a few particles. This effect rarely occurs and is extremely difficult to 

effects produced by electrified particles is the formation of an indication on a 
secondary surface. This effect takes place when a film of oil or a sheet of non 
conducting vinyl is placed directly over a cracked test piece. Indications of 
defects in the original surface will be reproduced on the secondary surface, al 
though the fidelity will be somewhat reduced. The indications form hesitantly 
and resemble subsurface defects. It is essential that the secondary surface be in 
intimate contact with the surface to be inspected. This is easy to accomplish, 
since powder will charge the vinyl and cause it to adhere quite strongly to most 


Test Objects Immersed in Oil. One spectacular demonstration of the second 
ary-surface effect is to place a cracked porcelain-enamel sample in a vessel of 
light petroleum distillate. When charged powder is blown at the surface of the 
oil, particles will eventually migrate to the bottom of the vessel, deposit on the 
sample, and form indications. When oil or a sheet of nonconducting material is 
used as a secondary surface, it is essential that the test surface be completely 
free of moisture vapor, as is the case with normal inspection. 

Oil Films as Moisture Barriers. Another instance where a protective coating 
is useful is during the testing of porcelain or ceramic coatings in the presence 
of excessive moisture vapor or water splashing. The object must be dried 
thoroughly and coated with oil prior to exposure. This will keep the surface 
dry and in condition to detect any cracks which occur. 

Blinking Indications on Secondary Surfaces. The blinking effect described 
earlier occurs on a secondary surface as well as on an original surface. This 
phenomenon can be used to muffle electric discharges. In areas where electric 
discharge may be hazardous, it is possible to cover the specimen with vinyl 
sheeting and proceed with a pinhole inspection, utilizing the blinking effect. The 
electric discharge occurs only between the metallic layer and bottom surface of 
the vinyl layer. Sometimes a secondary layer is useful in preventing the unin 
tentional puncturing of thin coatings. 

Precautions in Explosive Atmospheres. The powder itself will tend to 
muffle electric discharges and especially so if combined with a nitrogen pressure 
source instead of compressed air. Despite these safety features, it is wise to 
avoid using the method in hazardous atmospheres. 

powder is attracted to materials containing free electrons and does not adhere 
readily to nonconducting material. Materials whose dielectric properties lie be 
tween these two extremes will attract or hold charged powder, depending upon 
the degree of mobility of their respective electrons. This suggests that when two 
materials of different dielectric properties are blended together, charged powder 
will adhere more strongly to materials of lower dielectric strength. This is useful 
in the examination of plastic materials or asphalt coatings, since it will indicate 
the flow characteristics or distribution of blended materials. 

FALSE INDICATIONS. Indications which are not genuine occur occa 
sionally, but usually these are due to some physical cause which can be rem 
edied. Moisture streaks left from wiping cloths or toweling commonly produce 
false markings. However, these rarely resemble true crack indications. Lint 
from cloth, rags, or felt often resembles crack indications. When the inspector 
is in doubt, the indication can be pushed with a sharp point to see if it moves. 
Felt is the most common offender, since small fibers left on pieces resemble 
thermal-shock cracks. 

Powder Blotting. Water or oil particles blown on specimens attract pow 
der and resemble pinhole indications. If questionable, these can be removed 
and powder reapplied. False indications produced by the blotting of powder 
will indicate peculiarly after several seconds, since the fluid will spread. 

Residual Powder Artifacts. When a surface has been sprayed with powder, 
it becomes positively charged. If excess powder is brushed from the surface, the 
remaining particles may be preferentially oriented and tend to repel freshly 
charged particles. The surface is charged locally so that, as the next load of 


charged powder is applied, it will tend to distribute itself to areas where there 
is the least amount of charge. The resultant images rarely resemble true indi 

Chemically Induced Artifacts. Strong acid or basic residues, and sometimes 
oily materials, leave streaks or peculiar markings which may be mistaken for 
crack indications. This is particularly true when the bubble structure of glassy 
coatings has been opened. Little can be done about this except to note that a 
true indication is always half-round, sharp, and well defined. Indications pro 
duced by such materials usually are smudged and ill defined, and do not re 
semble true crack indications. 

RIALS. Glass coatings, with the exception of some high-temperature enamels, 
are nonconductive. The same situation exists with most resin coatings. All 
silvery or black coatings should be viewed with suspicion, however, since they 
may be conductive. They may contain stainless steel or aluminum bronze flakes. 
Black coatings invariably contain carbon. Ordinarily these materials are not 
considered conductive when blended with a nonconducting resinous material. 
However, potentials as high as 10,000-volt direct current are developed by 
electrified particles, and at that voltage many materials act as though they 
were conductive. When in doubt the inspector should always test a sample con 
taining a known defect. 

Mechanisms of Failure of Glass and Porcelain Enamels 

CRACK GROWTH CHARACTERISTICS. A characteristic property of 
cracks in glassy materials is that they will grow under certain conditions. A 
crack may grow, encounter a compression area, and stop. If the high stress area 
is relieved by some external cause, the crack may start growing again and enter 
this zone. A stress may be incompletely relieved by the formation of an ex 
tremely small crack. Such a crack has an unpredictable growth potential, de 
pending upon its service treatment. The direction of the crack is nearly always 
such that it tends to relieve locked-up stress; it depends entirely upon the 
stress gradient within the piece. 

CAUSES OF CRACKING IN GLASS. The most common causes of crack 
growth in glass are heat, vibration, or both. Both heat and vibration generally 
apply cyclic loading and, when coupled with secondary causes, will cause growth. 
Factors contributing to the growth of cracks in glass are shown in Fig. 5. 
These are: 

1. Ordinary tension. Tension stresses on the surface of a part tend to tear the 
glass apart. 

2. Heat via arc. If a crack system joins two terminals or a terminal and ground 
where a large voltage differential exists, the ensuing breakdown develops heat 
within the fluid. High vapor pressures develop, thrust outward, and encourage 
crack growth. 

3. .Random-dirt wedge. A crack may receive dirt, and when followed by a com 
pression loading, wedging action contributes to the growth of the crack. 

4. Crack debris wedge. Debris from the inside of the crack may break off and 
drop to the bottom. When compression loading is applied, the debris acts as a 
wedge, and stress levels at the base of the crack increase, causing crack growth. 

5. Mismatch wedge. If the walls of the crack do not mesh under compression 
loading, wedge action will be exerted, resulting in the growth of the crack. 


6. Hydrated-crystal wedge. It is possible for water to combine with certain glass 
constituents and form hydrated crystals. Compression loading in combination 
with wedging action of the crystals will promote crack growth. 

7. Hydrated-crystal growth. It is possible for crystal growth to occur over a 
long period of time and exert considerable outward thrust upon the sidewalls 
of the crack, causing it to split still further. 






Fig. 5. Factors contributing to the growth of cracks in glass. 

enamel, though very strong in many ways and desirable in other respects, is 
brittle. Despite this, it has been used widely for many years without serious 
difficulty. This is not so much a tribute to the ruggediiess of porcelain enamel 
as it is to the manufacturer who has learned to design and ship his product in 
such a way that brittlene^s is overcome. Despite the success of porcelain enamel 
in industry and consumer fields, its inherent brittleness must always be con 
sidered in design, production, shipment, and use. For many years industry and 
the consumer have expected porcelain enamel to crack or fail, and neither was 
aware of the underlying reasons for such failure, 

Cracks Yield Engineering Information. In the past few years the electrified- 
particle inspection method has been accepted as a reliable indicator of cracks in 
porcelain enamel. Further, it has been established that failures invariably occur 
in locations _ which were cracked prior to actual failure. Before this was known, 
shock and impact tests were relied upon to locate potential field failures. While 
some correlation can be obtained in this manner, these tests often introduce 
defects which lead to field failures. 

Direction of Stress. Brittle materials always crack at right angles to the 
direction of the principal applied stress. Subsequent cracks rarely completely 
cross an original crack pattern. The presence of a crack pattern tells the de 
signer that he has a stress concentration which may lead to more serious con 
sequences if neglected. With the electrified-particle method, the stress-indicating 
cracks can be located, studied, and perhaps duplicated under controlled labora 
tory conditions. 



Crack Formation. The formation of a crack requires the stressing of a sample 
to a point where compression is removed and the surface of the enamel goes 
into tension. When a certain point is reached, which may be prior to the yielding 
of the base metal, a crack will form. The crack formed in tension is visual 
evidence of a tension stress that reached a sufficiently high level. When load is 
released, the material returns to its original state, or nearly so, if the base metal 
has not yielded. If the inner surfaces of the crack mesh perfectly, the area at 
the crack is under approximately the same compression stress as the original 
coating. The coating, of course, does not contribute the same amount of strength 
in tension as it did prior to the formation of a crack. 

Compression Failures. If a failed area of porcelain enamel is examined 
closely, it is possible to piece together the chain of events and tell how the defect 
occurred. When enamel is bent in compression in an unreasonable manner, it 
will fail in the manner shown in Fig. 6. The failure always is explosive in nature, 

Magnaflux Corporation 

Fig. 6. A porcelain- enamel failure due to compression. Note ripples in dark under 
coat and the V-beveled edges of failure. 

and the amount of load must be quite high to induce failure. Relatively large 
areas of enamel are removed, and the edges of the failed area always have a 
beveled appearance. If the material contains a ground coat, smooth ripples 

Tension Failures. Porcelain enamel, grossly bent in tension, will fail as shown 
in Fig. 7. Although the enamel is loose, it is usually reluctant to remove itself 
from the base metal. Obviously, yielding of the base metal has contributed to 

Exceedingly high loads are required to cause either tension or compression 
failures. This suggests that it is unlikely that either mild tension or compression 
stresses alone are responsible for field failures of porcelain enamels. 

Cyclic Loading Failures. In Fig. 8 a starlike pattern was formed, with re 
moved chips between the radial cracks. Despite the similarity to impact failure, 
the sample was not subjected to impact loading. This failure was artificially in 
duced by a slow, steady loading on the reverse side until audible cracking was 
heard. With load still in place, the sample was processed, which verified radial 



Magnaflux Corporation 

Fig. 7. A tension failure on porcelain enamel. Note sharp edges of failed area and 
ridges in dark ground coat. 

Magnaflux Corporation 

Fig. 8, A porcelain-enamel failure, illustrating how cyclic loading contributes to 
failure. Note beveled edges indicating compression, and ridges in ground coat 

denoting tension. 


cracking. The defective area was probed with a sharp point, and none of the 
enamel was found to have loosened or failed. When the load was released slowly, 
audible failure and removal of enamel occurred. 

The simple experiment showed that removal of the enamel required a cyclic 
loading, with tension preceding compression. It is obvious that a crack in 
enamel weakens the coating to subsequent tension stress. It is not so obvious 
that a crack weakens the enamel to a subsequent mild compression load. In 
dustrial tests and elaborate tests on hundreds of enamel samples have proved 
that many field failures can be traced to cyclic loading. In other words, failure 
invariably occurs after compression is applied at the site of a crack previously 
formed in tension. Sometimes failure occurs with the mere release of original 
tension. In every case, however, the failed area always shows: 

1. A beveled appearance at the edges of the failed area. 

2. Typical sharp ridges in the ground coat or residual ground coat. 

The ridges are the residue of the original tension stress, and the bevels are an 
indication of compression stresses. 

Several years of experience in working with porcelain enamels indicate that 
these are the conditions under which failure occurs. Despite this, the actual 
mechanism of failure is not apparent. 

Industrial Applications 

INSPECTION OF GLASS CONTAINERS. The electrified-particle 
method was developed initially for the inspection of glass bottles and containers. 
During its early development it was often difficult to obtain uniform test results 
from day to day. To overcome this, a machine was constructed which would 
process bottles of varying size at rates similar to a typical production line. 
Scores of tests were run on many types of beverage bottles manufactured by 
various companies. 

Typical Indications. The types of indications obtained are shown in Fig. 9. 
Fig. 9 (a) illustrates a type of defect that extends from the inside to the outside 
of a sealing rim on a bottle, and which constitutes a path for vacuum or pressure 
leakage. Another type illustrated in Fig. 9(b) does not completely cross the 
sealing rim. Such a crack may grow in service or repeated use. Fig. 9(c) illus 
trates a common defect occurring in glass containers. This type of defect some 
times results in a piece of glass dropping off inside the container. Fig. 9(d) 
illustrates a series of small and tiny indications of cracks which sometimes results 
in an explosion of a bottle during pasteurization. 

of corrosive solvents demands great ingenuity of designers and chemists who 
must develop storage and processing vessels which will resist corrosion. These 
materials range widely in cost, availability, and effectiveness. Some of them 
are chemical porcelain, plastic, stainless-steel clad, glass-lined, and organic-resin 
coatings on steel. Inspection methods are available for most of these materials, 
but until the comparatively recent use of the electrified-particle inspection 
method, the significance of defects has not been fully recognized on glass-lined 
or on organic-resin-lined equipment. To understand its usefulness in this indus 
try, it is necessary to examine other test methods. 

Brine Test. The brine test involves the use of a simple ohmmeter, or an 
electrical device resembling this function. One lead of the instrument is grounded 

2? -12 



Magiuifiux Corporation 

Fig. 9. Rim defects in glass bottles, (a) Indication of a crack extending from inside 
to outside of a sealing rim in a beer bottle, (b) Indication of a defect which does 
not completely cross the sealing rim of a beer bottle, (c) Indications of cracks which 
may lead to chippage into the container, (d) Extremely tiny indications of cracks 
which are sometimes related to explosion in the pasteurization process. 


to a metallic portion of the vessel. A brine-soaked sponge is attached to the tip 
of the other lead and swabbed over the surface of the coating. When the fluid 
on the end of the probe finds a pinhole, it rapidly penetrates into the defect and 
contacts base metal. A reading of lowered resistance immediately appears on 
the ohmmeter. The fluid completes a circuit from ohmmeter to swabbing probe 
to pinhole, through the fluid to metal, and from the grounded probe back to 
the ohmmeter. A variation on. this technique is to fill the tank with a brine 
solution and dip one of the probes into the solution. The other probe, of course, 
is grounded to the base metal. 

Limitations of Brine Tests. Both techniques have the disadvantage of not 
being able to locate a through-to-metal pinhole accurately. The second method 
is particularly disadvantageous in this respect. The probing technique is slow 
and tedious when inspecting extremely large vessels. It is frequently criticized 
because the salt water will attack the base metal at the bottom of a pinhole and 
produce corrosion products which sometimes spread below the coating. Some 
times resin coatings form blisters at this point and peel away from the base 

One serious disadvantage of the probing brine test is that it cannot always 
indicate the actual breakthrough location. When an inspector probes a long 
defect and encounters the end of a crack, the penetrating fluid races to the 
actual breakthrough, which may be several feet away. This gives a reading, 
and the operator will mistakenly assume that the defect is directly underneath 
the probe. 

Modified Tesla Coil Test. Another test utilizes a device which operates on 
110-volt alternating current and which resembles an old automotive induction 
coil. Voltages ranging between 8,000 and 45,000 volts are generated and im 
pressed across the work by a needle protruding from one end of the device. The 
voltage can be varied by a control knob. As the needle is moved over the glass 
or resin lining, a high voltage spark emanates from the end of the probe and 
darts through a through-to-metal defect. Sometimes a spring or chain is placed 
on the end of the probe and the assembly is draped over the coating. This 
permits the inspection of a much larger area. In the hands of a skillful operator 
the Tesla coil is fairly rapid at locating pinholes and is not particularly danger 
ous to coatings. If the voltage is too high and the operator allows the chain or 
probe to dwell in one spot too long, it is possible to break down the coating 
and produce pinholes where none existed before. 

Before the advent of the electrified-particle method, it was believed that the 
electric needle had the ability to locate all breakthroughs to_ metal. This has 
been disproved where a breakthrough coincides with a crack in glass linings. 

ever a defect is completely through to metal, a blinking electrified-particle in 
dication is produced. This is a reliable and positive way to ascertain whether 
or not a defect is through to metal. If a considerable length of crack is broken 
through, a row of round indications appears, and these blink at random. The 
blinking effect only occurs as the powder is being blown at the defect. 

Craze Patterns. One type of crack pattern which occurs on glass-lined ves 
sels resembles the craze sometimes found on chinaware (Fig. 10). A craze pattern 
usually occurs where the coefficient of expansion of glass and base metal are 
improperly matched. Years ago this type of pattern was more common than 
it is now. Increased knowledge and experience have enabled ceramic engineers 


to reduce the incidence of this type of defect. If crazing is encountered, in 
terpretation is difficult because of the great number of defects present. So 
many cracks are present that the job of detecting breakthrough on each and 
every crack indication may be stupendous. 

Despite the fact that manufacturers of glass-lined vessels rarely produce 
crazed coatings, there is the possibility that the user may accidentally do so. 
If spalling or chipping of the glass has already occurred, the damage can be 
visually observed. When the damage is not obvious, charged powder can be 
used to help decide whether or not to continue using the vessel. 

Magnnflux Corporation 
Fig. 10. Indications of a craze pattern on a typical glass-lined fitting. 

Stress Cracks. The most common type of crack found on glass coatings might 
be termed a stress crack, produced by physical bending. This may occur during 
manufacture, shipment, set-up, and in use. Such cracks are found at places of 
bolt-up, changes of section, or where attachments have been welded or bolted 
in place. They always occur perpendicular to the direction of principal stress. 

Stress cracks invariably go through to base metal at some point in their 
length and for this reason are considered a serious defect. They are usually 
invisible to the naked eye and cannot always be found with a Tesla coil. If a 
defect cannot be seen, a plant inspector may assume that none is present. Thus, 
in service, a corrosive chemical may travel through a defect and attack the base 
metal, eventually causing service failure and producing disastrous results. 

Wavy Cracks. Another type of defect resembles the rolling and twisting of 
a roller coaster. This defect does not seem to be related to any unusual applied 
stress and is a manufacturing rather than a user defect. The cracks may range 
from three to several hundred feet in length. They are smooth, curvy, and tend 
to wander in distinctive roller-coaster fashion. The cracks rarely cross one 
another. This type of crack usually occurs only on thick coats of glass such as 
are found on cast-iron sanitary ware and glass-lined vessels. 


The roller-coaster type of crack rarely goes through to metal. Despite this, 
it seems wise to treat it in the same manner as a stress crack. If any portion of 
the defect blinks, the defect is through to metal. Little is known of the growth 
of either the stress or roller-coaster type of defect. 

Pinholes. The occurrence of pinholes is very common in coated vessels and 
is considered a serious defect. After a pinhole is located, the glass is chipped 
away from the localized area. It is then drilled, tapped, and a tantalum or other 
rare-metal plug or bolt is threaded into the side of the vessel. A fluorinated- 
plastic gasket material is used to help complete the seal 

Cracks Through Pinholes. Sometimes a crack runs through and past a pin- 
hole. The inspector may successfully locate the pinhole with a Tesla coil and 
not be aware of the. >f act that a crack also exists in this location. When the 
pinhole is repaired, the manufacturer will assume that the defect has been 
satisfactorily repaired. The chemical used in the vessel may enter the crack 
emanating from either side of the patch and work its way underneath the plugged 
area, producing corrosion. Eventually the plug may drop out. 

If a through-to-metal pinhole is detected with charged powder, its presence 
can be verified by the use of a Tesla coil or a brine test. If a breakthrough occurs 
in 'a crack and is detected with the blinking effect, it may be very difficult to 
verify the exact location by the use of the Tesla coil or a brine test. The Tesla 
coil may not detect the breakthrough because it occurs within a crack. The 
blinking effect will not take place if brine or other conductive material is in the 
through- to-metal defect. 

Indications from Chemical Residues. Glass-lined and other coated vessels 
sometimes will have other types of indications which are related to the material 
that was last in the tank. Very strong acids tend to leave residues, and it is 
sometimes difficult to develop good, clean-cut powder indications. Even though 
a tank subjected to strong acidic material has been thoroughly rinsed, there is 
still the possibility of developing smears and peculiar markings. This is par 
ticularly true where abrasion of the coating has occurred and the bubble structure 
of the coating has opened up. 

Basic materials and water sometimes tend to promote healing of cracks in 
glass-lined vessels. Nothing seems to promote the healing of defects occurring 
in resin coatings. Old cracks rarely heal up completely. The indications obtained, 
however, may be very poor and difficult to develop. 

plex electrical instrumentation in modern aircraft has emphasized the problem 
of maintaining the integrity of every component. As aircraft go higher and faster, 
the demands upon every part become more severe. During World War II a 
major attempt was made to eliminate high-altitude problems in electrical de 
vices. This involved the use of hermetically sealed equipment. As the use of 
this type of equipment increased, problems arose involving glass-to-metal seals. 
Mysterious and unpredictable field failures occurred which were later traced to 
cracks in glass portions of the seal. The electrified-particle method has played 
an important role in the subsequent investigation of these failures and has 
aided in the development of intelligent specifications for industrial purposes. 

Effects of Cracks in Seals. Many cracks in glass-to-metal seals do not go 
through the entire glass thickness. They do not permit gaseous leakage in 
either direction. Fig. 11 illustrates small headers which have been removed from 
equipment, both having identical crack patterns on top and bottom. Cracks 



not coincidental will not cause gaseous leakage. Any crack system winch does 
have coincidental cracks on both sides of a seal should be viewed with suspicion. 
Despite the narrowness of cracks, they may cause gaseous leakage which is un- 
detectible by crude bubble tests. The use of the mass spectrograph on such 
crack systems confirms the fact that through-cracks can leak. 

Mu^nadux Corporation 

Fig. 11. Small headers which have been removed from equipment, both having 
identical cracks on top and bottom. Note that the cracks are not- coincidental. 

Electrical Leakage. From some points of view, electrical leakage of a sealed 
component is a much more serious problem than gaseous leakage, since the 
failure may be abrupt and unpredictable. The designer of electrical equipment 
expects a certain amount of electrical leakage, depending on voltage differentials 
or expected atmospheric conditions. However, it is difficult to design tests to 
account for unpredictable crack systems. 

Atmospheric conditions influence the resistance of the surface of glass. 
Under some conditions, moisture condenses on glass surfaces in an exceptionally 
thin layer. If a crack is present, moisture penetrates it readily and contributes 
to the over-all lowering of resistance of the seal. This fact may be verified by 
checking resistance with a megohmmeter before and after a crack is induced 
in a seal. 

Intermittent Faults. One annoying thing about electrical leakage is that it 
may be intermittent and difficult to detect in service. If a crack contains 
moisture and the equipment is activated, a breakdown may occur until the 
moisture within the crack is heated by the passage of current and is driven 
out. When the equipment cools clown, more moisture may condense on the glass 
surface and the sequence will be repeated. If the equipment is energized con 
tinuously, the chances of electrical breakdown are less, since the moisture is 
driven out and stays out as long as the equipment is warm. However, this same 
heat may contribute to the growth of cracks, and the situation becomes worse. 
Under some conditions carbon will deposit within cracks and provide a perma 
nent breakdown path. This is an easier condition to locate if found in time to 
prevent total loss of the completed assembly. 


Types of Cracks. Generally speaking, there are five types of crack patterns 
which appear on glass-to-metal seals. Some are caused by the assembler of the 
complete hermetic package. Others are caused by improper use in service. The 
five classifications shown in Fig. 12 are pictorial representations indicating only 
one crack of each kind, although more than one classification may occur on the 
same piece. They are listed in order of increasing seriousness with regard to 
vacuum and electrical leakage. 

Fig. 12 (a) shows Type 1 cracking, starting from the surrounding ring or 
terminal. The cracks are short and may not be serious unless they occur 
coincidentally on both sides of the seal. These cracks may grow under the condi 
tions discussed earlier. These generally are typical manufacturing defects. The 
dotted lines indicate cracks on the opposite side. 

Fig. 12 (b) shows Type 2 cracking, which is comparatively rare and consists 
mainly of concentric circular cracks paralleling the rings or the terminals. They 
may be dangerous only if appearing coincidentally on top and bottom of seals. 

Type 3 cracking is shown in Fig. 12 (c). These cracks seem to be a typical 
manufacturing defect and may only be of concern if appearing coincidentally 
on top and bottom surfaces of headers. 

Fig. 12 (d) illustrates Type 4 cracking, typical of thermal shock. This condi 
tion sometimes is noted in new seals, but it is more commonly a result of 
improper soldering techniques. If the cracks are coincidental on top and 
bottom, gaseous leakage may occur. Electrical leakage will occur if the crack 
patterns are appropriately located. 

Fig. 12 (e) shows Type 5 cracking, which is the most common type of defect. 
It seems to be typical of processing techniques. Field repair work also may 
contribute to the creation of this type of defect. This can be dangerous under 
the same conditions mentioned for Type 4. 

Evaluation of Cracks. Some manufacturing concerns inspect hermetic seals 

visually, sometimes aided by a microscope. This is time consuming, expensive, 
and inaccurate in locating all defects. Some companies eliminate those defective 
pieces containing cracks which their customers may see. Their attitude is, ''What 
can't be seen won't hurt." Another attitude is, "We've been making these com 
ponents a long time and have never received any reports of failure." Some 
even say, "It's impossible to eliminate defects." 

The fact that a glass-to-metal seal passes through many hands before ulti 
mately reaching its home within a complex apparatus makes it difficult to 
ascribe the blame for a malfunctioning piece of equipment. The failure of a 
single seal can contribute to the failure of a relay. This may lead to still another 
failure which ends in complete disaster. This chain of failures is complex in 
operation and difficult to trace backwards. Yet it can happen. Since the 
possibility of failure exists due to seemingly insignificant cracks, it is worthwhile 
to locate such defects, particularly since a simple method exists for their detec 

MISCELLANEOUS APPLICATIONS. Several specialized applications 
have been found for electrified-particle tests. 

Porcelain Teeth. The relationship of cracks to the strength of porcelain teeth 
has been thoroughly investigated. It has been proven that invisible defects 
in porcelain teeth can contribute to failure in use. Cracks are generally formed 
by poor laboratory techniques rather than being formed during actual manu 















i'" "^ I 




































































Fig. 12. Sketches of typical crack patterns on glass-to-metal seals, (a) A typical 
Type 1 crack pattern, (b) A typical Type 2 crack pattern, (o) A typical Type 3 
crack pattern, (d) A typical Type 4 crack pattern, (e) A typical 'Typo 5 'crack 




^ Terra Cotta. Fig. 13 is interesting because it represents a peculiar combina 
tion of materials. The base material is porous, whereas the top surface is com 
pletely glazed and nonabsorbent. Here the defect is completely invisible to the 
naked eye, though successfully indicated by powder. This material always re 
quires penetrant in order to detect cracks. The application of charged powder 

Magnaflux Corporation 
Fig. 13. Indication of defect in glaze on terra cotta. 


sometimes electrostatically attracts the fluid within the defect and provides 
blotting action in addition to the electrostatic effect, Tims a much larger crack 
indication is provided. 

Glass Lenses. Fig. 14 shows how charged powder helps delineate superficial 
scratches which may" not be easily visible to the human eye. Material must he 
given penetrant treatment and then carefully subjected to charged powder. The 
deepest ends of the scratches always seem to have a shadowed area in conjunction 

Ma^naflux Corporation 

Fig. 14. Indications of superficial scratches in a lens. Indications like those raroly 
resemble true crack indications. 

with the regular indication. The size of an indication always is dependent upon 
the amount of ionic attraction, which in turn is related to the depth. Thus weak 
indications indicate very slight defects, whereas larger indications always indicate 
substantially more damage. 

Scratch indications are almost always delineated by a shadowed area at the 
end of a crack, and the indications are extremely difficult to keep in place during 
powder application. The best technique is to apply the powder in one quick 
burst and not to attempt to build up the indications, since they invariably blow 
away before an appreciable size is gained. 

Brittle- coating Tests. A recent development has permitted the use of elec 
trified particles in conjunction with the brittle-coating stress analysis method. 
Heretofore it has been difficult to locate cracks in the brittle coating except by 
essentially destructive means. Electrified-particle techniques used in conjunction 


with brittle-coating techniques have opened up whole new areas of experimental 
techniques by permitting the nondestructive location of the stress-indicating 
cracks in the stress-sensitive brittle coatings. (See section on Brittle-Coating 


BETZ, C. E. "Two New Testing Methods for Ceramic Products," Nondestructive 
Testing, 7, No. 2 (1948) : 22. 

COWAN, R. E. "Sonic Method for Measuring Young's Modulus of Elasticity of Porce 
lain Enamel-Metal Composites," Finish, 12, No. 4 (1955) : 40, 62. 

Cox, J. E, and A. I. ANDREWS. "A Study of Hairlining,"' J. ' Am. Ceram. Soc., 37 
(1954): 186. 

DUGGER, E. "The Application of Statiflux for Nondestructive Inspection of Noncon 
ducting Materials," A JF. Tech. Re-pi., No. 5898 (1949). 

"Electrified Particle Nondestructive Test," Report on the State of Research and 
Application of Nondestructive Testing in the U.S. and Canada to the International 
Conference on Nondestructive Testing of Maternal, Brussels, Belgium, May 1955. 
MP-10. Evanston, 111. : Society for Nondestructive Testing, 1955. 

HOWE, E. E. "Special Techniques in Enameling," Proc. Ann. Porcelain Enamel Inst 
15 (1953): 109. 

HUPPERT, P. A. "Lithium Compounds in Porcelain Enamel Compositions," Finish, 
4, No. 6 (1947): 18. 

HUTCHINSON, C. "NST Testing Correlated with Shipping Performance Can Reduce 
Damage Losses," Finish, 9, No. 8 (1952) : ST-5. 

JACKSON, H. J. "Ceramic Stress Coating," Machine Design, 26 (1954) : 184. 

MclNTOSH, R. 0. "Hermetic Seals," Elec. Mfg., 49, No. 4 (1952) : 120, 328, 330 332 

MCMASTER, R. C. "Nondestructive Testing," 26th Edgar Marburg Lecture, Am. Soc. 
Testing Materials, Proc., 52 (1952) : 617. 

MCMASTER, R. C., and S. A. WENK. "Nondestructive Testing of Engineering Mate 
rials and Parts," Materials & Methods, 33, No. 2 (1951): 81. 

MOULD, R. E. "The Behavior of Glass Bottles Under Impact," J. Am. Ceram. Soc. 
35 (1952) : 230. 

ORR, S. C. "Methods for Testing for Enamel Coating Discontinuities," Better 
Enameling, 22, No. 12 (1951) : 16, 31. 

PETERSEN, F. A., T. DEFOREST, and H. N. STAATS. "A New Method for Locating 
Cracks in Ceramic Materials," Finish, 5, No. 3 (1948) : 17. 

PETERSEN, F. A., R. A. JONES, and A. W. ALLEN. "A New Method for Studying 
Fractures of Porcelain Enameled Specimens," J. Am. Ceram. Soc., 31 (1948): 186. 

PIERCE, J. J. "Ultramodern Supersonics and X-Rays," Metal Progr., 56 (1949) : 62. 

PINGEL, M. E. "Torsion Testing as an Aid in Process Control," Proc. Ann. Porcelain 
Enamel Inst., 14 (1952) : 139. 

"Powder Finds Flaws in Liner Coatings," Appliance Mfr., 1, No. 4 (1954) : 1. 

SINGDALE, F. N. "Brittle Coatings for Use in Stress Analysis Under Varying Tempera 
ture Conditions," Nondestructive Testing, 11, No. 6 (1953): 37. 

STAATS, H. N. U A New Tool for Locating Defects in Non-Conductive Coatings," 
Corrosion (in press). 

. "Detecting Cracks in Glass-to-Metal Seals," Electronics, 28, No. 3 (1955) : 284. 

. "Gunning for Bottle Defects," Brewers Dig., 30, No. 10 (1955) : 58. 

. "Observations on the Nature of Cracks in Porcelain Enamel," Am. Ceram. 

Soc. Bull., 31, No. 2 (1952): 33. 

. Principles of Stresscoat, A Manual for Use ivith the Brittle Coating Stress 

Analysis Method. Chicago: Magnaflux Corp., 1955. P. 19. 

. "The Testing of Ceramics," Nondestructive Testing, 10, No. 3 (1952) : 23. 

"Which Nondestructive Test for Finding Defects in Ceramic Parts." Mate 

rials & Methods, 36 ? No. 3 (1952) : 116. 


STAATS, H. N., and S. J. BARANOWSKI. "Calibrated Porcelain Enamel Coatings," Am. 
Ceram. Soc. Bull, 35 (1956) : 143. 

"Symposium on Accelerated Durability Testing of Bituminous Mulc'rials," Am. Stw. 
Testing Materials, Spec. Tech. Publ, 94 (1950). 

Symposium on Electromagnetic Relays, Oklahoma Institute of Technology. Prince 
ton, Ind.: Potter & Brumfeld, 1954. P, 13. 






Magnetic Fields 

Principle of test 1 

Capabilities and limitations 1 

Operational requirements 1 

Magnetic- field principles 2 

Description of magnetic fields 2 

Magnetic -field path in a horseshoe magnet 

(/. 1) 2 

Magnetized ring 2 

Magnetic -field path in a ring magnet with 

air gap (/. 2) 2 

Magnetic-field path in a closed, mag 
netized ring (/. 3) 3 

Effect of cracks in magnetized ring 3 

Magnetic particles attracted to a radial 
crack in a circularly magnetized part 

C/.4) 3 

Bar magnet 3 

Straightening a horseshoe magnet results 

in a bar magnet (/. 5) 4 

Effect of cracks in magnetized bar 3 

A crack in a bar magnet creates magnetic 
poles which attract magnetic particles 

(A 6) 4 

Magnetization with electric current 4 

Circular magnetization 4 

Circumferential magnetic field surround 
ing a straight conductor carrying an 

electric current (/. 7) 4 

Circular magnetization of a test piece 
through which a magnetizing electric 

current passes (/. 8) 5 

Circular magnetization of solid parts 5 

Circular magnetization of typical forms of 

test object (/. 9) 5 

Circular magnetization of hollow parts 6 

Circular magnetization of cylindrical parts 
by use of a central current -carry ing con 
ductor (/. 10) 6 

Longitudinal magnetization 6 

Longitudinal or coil magnetization (/. 11) 7 

Coil magnetization 6 

Cable magnetization 6 

Methods of Magnetization 

Controlling factors 1 

Alloy, shape, and condition of part 7 

Type of magnetizing current 8 

Direct- current magnetization 8 

Half-wave rectified magnetizing current .... 8 


Alternating- current magnetization 8 

Direction of magnetic field 8 

Circular magnetization 8 

Passing current through the entire part 9 

Prod magnetization of large parts 9 

Central conductor magnetization 9 

Limitations of parallel magnetization 9 

Longitudinal magnetization 9 

Important considerations in coil magnetiza 
tion 10 

Coil magnetization of a wheel-like part 

(/. 12) 10 

Yoke magnetization 10 

Longitudinal magnetization of a test part 
by means of an external magnetizing 

yoke (/. 13) 11 

Swinging-field magnetization 11 

Test Procedures 

Sequence of operations 11 

Continuous- field method of inspection 12 

Residual-field method of inspection 12 

Value of flux density 12 

Current requirements for circular magnetiza 
tion 13 

Optimum current magnitudes 13 

Current requirements for longitudinal mag 
netization 13 

Influence of length /diameter ratio 13 

Formula for correct ampere- turns 14 

Typical coil shot currents (amperes) for a 

five-turn coil (/. 14) 14 

Current requirements for flexible cable mag 
netization 14 

Inspection Materials and Their 

Types of magnetic particles 15 

Wet bath materials 15 

Instructions for mixing the oil bath for the 

wet method 15 

Bath capacity of magnetic -particle equip 
ment (/. 15) 16 

Required oil characteristics (/. 16) 16 

Bath strength 16 

Instructions for checking bath strength . . . , 16 

After 30 minutes of settling time, the bath 

strength is indicated by the volume of 

settled particles read on the scale at the 

bottom of the centrifuge tube (/. 17) . . 17 



CONTENTS (Continued) 


Selection of wet method paste 17 

Advantages of fluorescent particles 18 

Water suspension pastes 18 

Particles for the dry method 18 

Application of dry particles 18 

Mechanical powder blower 19 

Other means of applying powder 19 

Special Requirements for Fluorescent 

Black light requirements 19 

Influence of power supply voltage 19 

Checking light intensity 19 

The inspection area 19 

Dark adaptation of inspector 20 

Principles, of Demagnetization 

Residual magnetism 

Reasons for requiring demagnetization 

Causes of magnetization 

Longitudinal and circular residual fields 

Basic principle of demagnetization 

Magnetic hysteresis 

Typical hysteresis curve (/. 18) 

Use of diminishing a.-c. fields 

Diminishing hysteresis curve resulting from 
the alternate reversal and reduction of 

the applied field (/. 19) 

Retentivity and coercive force 

Methods of demagnetization 22 

A.-c. coil demagnetization 23 

Manually operated u.-o. coil demagnetize!* 
complete with roller ciirrmtfe and auto 
matic timing switch (/. 20) 24 

Alternating-current coil demagnetize! 1 in 
corporated into a conveyorizwl unit for 

magnetic-particle inspection (/. 21) 24 

Reversing d.-c. demagnetization 25 

A.-c. circular field demagnetization 25 

A.-c. and d.-c. yoke demagnetization 25 

Methods for Measuring Leakage 
Field Intensities 

Types of measurements 25 

Field indicator 20 

Typical field indicator (/. 22) 26 

Indications 27 

Procedure 27 

Compass indicator 27 

Steel- wire indicator 27 

Limitations with circularly magnetized paiks . 27 

Checking demagnetization 28 

Demagnetization problems 28 

Shielding by magnetic shunting 28 

Small pnits having a small length/ diameter 

ratio 28 

Demagnetizing-coil field strength apparently 

too weak 28 

Fugitive magnetic poles 28 

Bibliography 29 




Magnetic Fields 

PRINCIPLE OF TEST. Magnetic-particle inspection is a nondestructive 
means for detecting discontinuities in ferromagnetic materials. It consists of three 
basic operations : 

1. Establishing a suitable magnetic field in the test object. 

2. Applying magnetic particles to the surface of the test object. 

3. Examining the test-object surface for accumulations of the particles (indica 
tions), and evaluating the serviceability of the test object. 

Capabilities and Limitations. The method can detect all discontinuities at 
the surface, and under certain conditions, those which lie completely under the 
surface. It depends upon the magnetic properties of the test objects and is suit 
able only for metallic materials which can be intensely magnetized. Nonferro- 
magnetic materials, which cannot be strongly magnetized, cannot be inspected 
by this method. Such materials include aluminum, magnesium, brass, copper, 
bronze, lead, titanium, and austenitic stainless steels. With suitable ferromag 
netic materials, magnetic-particle inspection is highly sensitive and produces 
readily discernible indications on the surface of the test parts. A trained in 
spector can, by examining the nature, location, and extent of the indications, 
interpret their causes and evaluate the discontinuities causing the indications. 

Operational Requirements. Parts should be clean before they are subjected 
to magnetic-particle inspection. The ability of the method to provide indications 
depends upon the ability of the magnetic particles to move in response to leak 
age magnetic fields which appear at the surface of the test objects where dis 
continuities are present. During application the particles may be suspended in 
air (dry method) or in liquids such as oil or water (wet method). If parts are 
inspected by the wet method, all dirt, grease, oil, rust, and loose scale must be 
removed; otherwise it may be difficult or impossible to produce indications of 
discontinuities or defects. Random dirt, grease, and oil may wash from the 
part, contaminating the inspection liquid. Rust and loose scale may also con 
taminate the inspection medium or act as electrical insulators to prevent proper 
electrical contact for magnetization of the part. The same requirements for 
cleanliness apply to parts to be inspected with the dry powder technique, but in 
addition, the parts must be dry. Oil, grease, or water will cause the dry pow 
der to stick to the surface of the test object, preventing formation of indications. 
Plating on parts will not interfere with magnetic-particle inspection, provided 
it is less than 0.004 in. thick. Any part made of ferromagnetic material can be 
inspected by the magnetic-particle method; there are no restrictions as to the 
shape and size of the part. 

30 1 


Magnetic-Field Principles. It is essential for reliable inspection that the 
fundamental principles of the magnetic-particle method be clearly understood, 
and that the method be properly applied. If improper techniques are used, some 
discontinuities may be missed and defective parts may be accepted by the in 
spector. With a useful and dependable inspection method, it is not enough to 
find defects; the inspector must be sure that all the defects the specific test 
method can reveal have been located and evaluated. For this reason the nature 
of magnetic fields and the basic principles of magnetic-particle inspection are 
described in an elementary way in the subsequent text. 

DESCRIPTION OF MAGNETIC FIELDS. The magnetic-particle in 
spection method utilizes magnetic fields to reveal material discontinuities. Ferro- 
magnetism is the property of some metals, chiefly iron and steel, to attract 
other pieces of iron or steel. The common horseshoe magnet will attract magnetic 
materials to its ends or poles. Magnetic lines of force, or flux, flow from the 
south pole through a magnet to the north pole, as shown in Fig. 1. Magnets will 

Fig. 1. Magnetic-field path in a horseshoe magnet. 

attract other magnetic material only where the magnetic lines of force leave 
or enter the magnet. When magnetic material is placed across the poles of a 
horseshoe magnet, the lines of force flow from the north pole of tho magnet 
through the material to the south pole (Fig. 1). Magnetic lines of force will flow 
preferentially through magnetic material rather than nonmagnetic material or 

Magnetized Ring. If the horseshoe magnet is bent so that its poles arc close 
together, as shown in Fig. 2, the poles still attract magnetic materials. Iron 
filings 'or other magnetic particles will cling to the poles and bridge tho gap 
between them. If the poles are fused together, the magnetic lines of force will 
be enclosed within the ring (Fig. 3). No external poles will exist, and magnetic 

Fig. 2. Magnetic-field path in a ring magnet with air gap. 


particles dusted over the ring will not be attracted to the ring even though there 
are magnetic lines of force flowing through it. Magnetized materials attract 
externally only when poles exist. A ring magnetized in this manner is said to 
contain a circular magnetic field which is wholly within the part. 

Fig. 3. Magnetic-field path in a closed, magnetized ring. 

Effect of Cracks in Magnetized Ring. Any radial crack in a circularly 
magnetized piece will create a north and a south magnetic pole at the edges of 
the crack. This will force some of the magnetic lines of force out of the metal 
path. These are called leakage flux. Magnetic particles will be attracted to the 
poles created by such a crack, forming an indication of the discontinuity in the 
metal part (Fig. 4). This is the principle of forming magnetic-particle indica 
tions by means of circular magnetization. 


Fig. 4. Magnetic particles attracted to a radial crack in a circularly magnetized 


Bar Magnet. When a horseshoe magnet is straightened, it becomes a bar 
magnet. A bar magnet has poles at each end. Magnetic lines of force flow 
through the bar from the south pole to the north pole (Fig. 5). Magnetic 
particles are attracted to any location where flux lines emerge and particularly 
to the ends of the magnet where the concentration of flux lines is greatest. Since 
the magnetic lines of force within a bar magnet run the length of the bar, it is 
eaid to be longitudinally magnetized or to contain a longitudinal field. 

Effect of Cracks in Magnetized Bar. A crack in a bar magnet (Fig. 6) dis 
torts the magnetic lines of force and creates poles on either side of the crack. 
These poles will attract magnetic particles to form an indication of the crack. 
The strengths of poles formed at a crack depends on the number of magnetic 
lines of force interrupted. A crack at right angles to the magnetic lines of force 



Fig. 5. Straightening a horseshoe magnet results in a bar magnet. Lines of mag 
netic flux pass through the magnet from its south to its north polo. 

interrupts more lines of force and creates .stronger poles than a crack moro 
nearly parallel to the lines of force. Indications of maximum size are formed 
when the magnetic lines of force are at right angles to discontinuities. 



Fig. 6. A crack in a bar magnet creates magnetic poles which attract magnetic 


are used to create or induce magnetic fields in magnetic materials. Since it is 
possible to alter the directions of magnetic fields by controlling the direction of 
the magnetizing current, the arrangement of current paths is used to induce 
the magnetic lines of force so that they are at right angles to a discontinuity in 
the test object. 

Circular Magnetization. Electric current passing through any straight con 
ductor such as a wire or bar creates a circular magnetic field around that 
conductor, as shown in Fig. 7. The magnetic lines of force are always at 
right angles to the direction of the current which induces the magnetic' field. 

\> \j7\7 


Fig. 7. Circumferential magnetic field surrounding a straight conductor carrying 

an electric current. 


30 5 

To remember the direction taken by magnetic lines of force around a conductor, 
imagine that the conductor is grasped with the right hand so that the thumb 
points in the direction of the current flow. The fingers then point in the direc 
tion taken by the magnetic lines of force in the magnetic field surrounding the 
conductor. This is known as the right-hand rule. When a conductor of electric 
current is a magnetic material, the passage of current induces a magnetic field 
in the conductor as well as in surrounding space. A piece magnetized in this 
manner is said to have a circular field or to be circularly magnetized (Fig. 8). 


Fig. 8. Circular magnetization of a test piece through which a magnetizing 

electric current passes. 

Circular Magnetization of Solid Parts. To induce a circular field in a part, 
current is passed through the part, as shown in Fig. 9 (a) . This sets up a circular 









FiK 9 Circular magnetization of typical forms of test object, (a) Circular mag 
netization of test object by passing electric current through part from machine 
contact plates, (b) Production of localized circular field m part by passing electric 
current between contact prods. 



magnetic field in the part, which creates poles on both sides of any crack or dis 
continuity parallel to the length of the part. These poles will attract fine mag 
netic particles and form an indication of the discontinuity. It is also possible 
to induce a circular field in localized areas of the part by using cables and prods 
or contacts to pass current through the area being inspected [Fig. 0(b)]. 

Circular Magnetization of Hollow Parts. With hollow or tubolikc parts the 
inside surfaces may be as important to inspect as the outside. When such parts 
are circularly magnetized by passing the magnetizing current through the part, 
a satisfactory magnetic field is not, produced on the inside surface. Since a 
magnetic field surrounds a current-carrying conductor, it is possible to induce 
a satisfactory magnetic field by sliding the part on to a central copper bar or 
conductor (Fig. 10). Passing current through the bar induces a circular mag 
netic field on both inside and outside cylindrical surfaces of the tubolikc part. 



Fig. 10. Circular magnetization of cylindrical parts by use of a central current- 
carrying conductor. 

Longitudinal Magnetization. Electric current can also be used to create a 
longitudinal magnetic field in magnetic materials. When electric current is 
passed through a coil of several turns, a magnetic field is established lengthwise 
or longitudinally within the coil (Fig. 11). The nature and direction of this 
field are the result of the field around the conductor which forms the turns of 
the coil. Application of the right-hand rule to the conductor at any point in the 
coil illustrated in Fig. 11 (a) shows that the field within the coil' is lengthwise, 
as indicated. 

Coil Magnetization. When magnetic material is placed within a coil, the 
magnetic lines of force created by the electric current concentrate themselves in 
the part and induce a longitudinal magnetic field as indicated in Fig. 1Kb). 
Inspection of a cylindrical part with longitudinal magnetization is shown in Fig. 
11 (c). With a transverse discontinuity in the part, magnetic polos are formed 
on both sides of the crack. These poles will attract magnetic particles to form 
an indication of the discontinuity. A comparison of Fig. ll(e) with Fig. 9 (a) 
shows that in both cases a magnetic field has been induced in the part which is 
at right angles to the defect. This is the most desirable condition for reliable 

Cable Magnetization. Parts too large to fit in a fixed coil can be magnetized 
longitudinally by making a coil of several turns of flexible cable. The use of 
portable magnetizing equipment with cables and prods has immeasurably 
broadened the use of magnetic-particle inspection, since there is no limit to the 



size of the part which can be inspected in this manner. It may be necessary to 
magnetize and inspect small areas sequentially, to cover a large surface area with 
prod magnetization. 









Fig. 11. Longitudinal or coil magnetization, (a) Longitudinal magnetic field within 
a current-carrying magnetizing coil, (b) Longitudinal magnetization of test object 
with coil, (c) Typical arrangement of coil and test object for longitudinal magnetiza 

Methods of Magnetization 

CONTROLLING FACTORS. Five factors must be considered in selecting 
the method of magnetization: (1) alloy, shape, and condition of part; (2) type 
of magnetizing current; (3) direction of magnetic field; (4) sequence of opera 
tions; and (5) value of flux density. 

ALLOY, SHAPE, AND CONDITION OF PART. The particular alloy, 
heat treatment, cold working, and other factors determine the permeability of a 


part, i.e., the ease with which it can be magnetized. It is necessary to consider 
the nature of the alloy in connection with selection of the sequence of operations 
and the value of flux density or magnetic-field intensity. These, in turn, affect 
selection of the magnetization method. The size and shape of the part deter 
mine the most practical method of magnetization with the available equipment. 
The condition of the part, especially the surface condition, influences the 
selection of the magnetization method. Surface conditioners such as Dulite, 
Parkerizing, and lacquer coatings are poor electrical conductors, and therefore 
it is impossible to pass magnetizing current through such coatings. If parts so 
treated can be adequately magnetized in a coil or with a central conductor, they 
can be inspected without removing such coatings. 

TYPE OF MAGNETIZING CURRENT. Many types of magnetizing cur 
rent can be used, of which only one is best for each typo of inspection. 

Direct-Current Magnetization. Direct current (d.c.) obtained from storage 
batteries was first believed to be the most desirable current to use, since direct 
current penetrates more deeply into test specimens than alternating current. 
The big disadvantage of storage batteries as a source of current is that there is 
a definite limit to the magnitude and duration of current which can be drawn 
from the battery before recharging. Battery maintenance is costly and can 
become a source of trouble. Direct current obtained through dry-plate rectifiers 
from a.-c. power lines is similar to battery current and has the advantage of 
permitting an almost unlimited number of magnetizing current f 'shots." Cur 
rent obtained by passing three-phase alternating current through special rec 
tifiers is called three-phase rectified alternating current. It is most commonly 
used in wet horizontal units for any general magnetic-particle inspection and is 
equal to direct current. 

Half-Wave Rectified Magnetizing Current. Half-wave rectified alternating 
current is the most effective current to use for the detection of subsurface and 
surface defects when dry magnetic particles are used. This type of current 
is produced by rectifying single-phase alternating current. Half-wave rectified 
current imparts a very noticeable pulse to the particles. This gives them mobility 
and aids in the formation of indications. 

Alternating- Current Magnetization. Alternating current (a.c.) at line 
frequency is the most effective current to use for the detection of surface dis 
continuities, particularly fatigue cracks. It is important that a.-c. inspection 
equipment be built to include proper current controls. An advantage of a.-c. 
inspection is the ease with which parts so inspected can be demagnetized. 

DIRECTION OF MAGNETIC FIELD. Of the five factors involved in 
determining inspection techniques, the most critical is the direction of the mag 
netic field. The proper orientation of the magnetic field in the part, in relation 
to the direction of the defect, is a more important factor than the magnitude of 
the magnetizing current. For reliable inspection the magnetic lines of force 
should be at right angles to the defect to be detected. If the magnetic lines of 
force are parallel to the defect, there will be little magnetic leakage at the 
defect. If any indication is formed, it is likely to be extremely small or indefinite. 

Circular Magnetization. The magnetizing method which is easiest to control 
in most parts is circular magnetization. This is the method in which the mag 
netizing current is passed directly through the part, setting up circular magnetic 
lines of force at right angles to the direction of current flow. 


Passing Current Through the Entire Part. When the parts can be placed 
between the contact plates of stationary magnetic-particle inspection equipment, 
this is the best way to magnetize the part circularly (see section on Magnetic- 
Particle Test Equipment). Care should be taken to clamp the part firmly be 
tween the soft lead contact plates. Enough surface area of the part must make 
contact with the plates to permit passage of the magnetizing current without 
burning the part. As the area of surface contact decreases, the probability of 
burning increases. On irregular parts it is helpful to use copper-braid contact 
pads between the part and plates, to prevent overheating. When inspecting parts 
with irregular cross-section, it may be necessary to magnetize circularly with a 
low current to inspect the thin areas, and then devise some method of passing 
a higher current through the heavier section for inspection of that area. 

Prod Magnetization of Large Parts. When the part is too big to fit into 
available stationary equipment, or if only portable equipment is available, then 
the part or areas of the part can be circularly magnetized by either of two 
methods. One method is to use prod contacts with cables to transmit the 
magnetizing current from the source to the part [see Fig. 9(b)]. Prods are 
attached to the ends of the cables so that magnetizing current can be passed 
through the part or through an area of the part. Portable equipment should 
have a remote control switch to enable the operator to turn the current on and 
off while moving the prods or viewing the work. Contact clamps can be used 
with cables instead of prods, particularly when the parts are relatively small 
in diameter. Tubular structures can be inspected by positioning the clamps so 
that the current passes through the suspected area and along the line of suspected 

Central Conductor Magnetization. When a circular magnetic field is set up 
in a tubular part by passing current through the tube itself, a satisfactory field 
is not induced on the inside surface of that tube. If the part is hollow or has 
holes through which a central conductor can be passed, it is best to induce a 
circular magnetic field in that part by passing the magnetizing current through 
the conductor. Circular magnetization with a central conductor has the follow 
ing advantages over passing current through the part itself: 

1. It induces a field on the inside diameter of the part, permitting inspection of 
inner as well as outer surfaces. 

2. Direct electrical contact is not made with the part, thereby eliminating the 
likelihood of burning. 

3. Several parts, like washers or nuts, can be suspended on the same conductor 
and inspected in groups. 

Limitations of Parallel Magnetization. The magnetic principle underlying 
circular magnetization with a central conductor is based on the fact that there 
is a circular magnetic field surrounding any electrical conductor. Knowing this 
fact, some operators have assumed that they can induce a circular field in a 
part by placing it next to instead of around a conductor. This is not true. Some 
field is induced in the part by such a procedure, but since a portion of the path 
of the magnetic flux is in air and some through the part, the field in the part is 
greatly reduced, distorted, and unevenly distributed. This procedure is some 
times called "parallel" magnetization. It is not dependable and should not be 

Longitudinal Magnetization. A longitudinal field can be induced in a part 
by placing the part in a fixed current-carrying coil, mounted either on the rails 



of a stationary unit or attached by cables to a portable unit. The effective 
magnetic field induced in a part by a coil extends from 6 to 9 in. beyond either 
end of the coil. If the part is long, it will be necessary to magnetize and inspect 
it in sections along the length of the piece. Longitudinal magnetization with 
portable equipment is accomplished by wrapping current-carrying cable in coil 
fashion around the part. 

Important Considerations in Coil Magnetization. To induce an adequate 
longitudinal magnetic field with a coil, the long dimension of the part should be 
at least twice as great as its short dimension, and the long axis of the part 
should be parallel to the coil axis. This is especially true in the case of irregularly 
shaped parts, since the shape of the part affects the direction of the induced 
field. Take, for example, the case of a wheel-like part. If such a part is placed 
in a coil, as illustrated in Fig. 12, a field will be induced in the white areas of the 



Fig. 12. Coil magnetization of a wheel-like part. Radial defects will be indicated 

only in white areas shown. To reveal radial defects in the dark areas, the part must, 

be rotated through 90 cleg, and remagnetized. 

part in such a direction that radial defects would create indications. However, 
radial cracks in the shaded areas of the part will be parallel (or nearly so) to the 
induced magnetic field, so that few or no indications will be formed. To indicate 
radial defects in the shaded areas, it would be necessary to rotate the part 
through 90 cleg, and remagnetize it. The detection of radial cracks in a part 
of this type is more positively and rapidly done by the use of a central conductor. 

Yoke Magnetization. A longitudinal magnetic field can be induced in a 
part, or in an area of a part, by means of a yoke. A yoke is a C-shaped piece 
of soft magnetic material, either solid or laminated, around which is wound a 
coil carrying the magnetizing current (Fig. 13). When a part is placed across the 
opening of the C-shape and the coil is energized, the part completes the path of 



the magnetic lines of force. This sets up a longitudinal field in the part between 
the ends of the C-shape. Permanent magnet yokes can also be used to create 
a magnetic field. 




Fig. 13. Longitudinal magnetization of a test part by means of an external 

magnetizing yoke. 

SWINGING-FIELD MAGNETIZATION. For complete inspection for 
defects oriented in different directions in a part, two or more magnetizations 
and inspections are necessary. First the part should be circularly magnetized 
and inspected; then, longitudinally magnetized and inspected. Usually the part 
should finally be demagnetized. If the shape of the part is complicated, it may be 
necessary to use more than the two magnetizing operations to be sure that all 
areas have been properly magnetized. 

Defects are shown best when they are at right angles to the magnetic lines 
of force. Also, the effective field induced by a coil on a long part extends only 
6 to 9 in. beyond either end of the coil; therefore a long part, say, 3 ft. long, 
should be given at least two magnetizing shots along its length. 

On certain parts it is possible, by using specially designed equipment, to apply 
two or more magnetizing currents simultaneously to the part in order to find 
defects in all directions. This creates a so-called swinging field. Such equipment 
shortens inspection time considerably, but application of the technique is critical 
and it should be used only under close control and observation by qualified 
personnel. Inspection employing swinging fields can be performed best on 
equipment designed for this purpose and for specified test objects. 

Test Procedures 

SEQUENCE OF OPERATIONS. The magnetic field is strongest while 
magnetizing current is flowing. When the current is cut off, the field drops 
to a lower value, its residual strength. If the part is of hard steel with a high 
carbon content, the residual magnetic field existing after current ceases to 
flow will be relatively strong but still weaker than when current was flowing. 
If the material is soft steel, the residual field will be too weak to attract particles 
and indicate cracks or defects. For greatest sensitivity the particles should 
be present on the part and under the influence of the magnetic field while 
magnetizing current is flowing. 


Continuous-Field Method of Inspection. When the particles are applied 
to the part while the magnetizing current is flowing, the technique is known 
as the continuous method. With the wet continuous method, the inspec 
tion bath is liberally applied to all surfaces of the part, care being taken 
to be sure that all surfaces are wet. The instant that the bath stream is removed 
from the part, the magnetizing shot is applied. This ensures that the par 
ticles will be on the part while the current is flowing, so they can be attracted 
to any leakage fields created by discontinuities. If wet bath is applied during or 
after the magnetizing shot, there is a possibility that the force of the bath 
stream may wash away lightly held indications. For this reason the shot should 
always be given immediately after the bath flow is stopped. 

When the dry continuous method is used, the part is magnetized by using 
prods, clamps, or a coil. While the current is flowing, powder is applied in a 
light cloud to magnetized areas. Excess powder is blown off with a light air 
stream. After excess powder is removed, magnetizing current is shut off. It is 
important that powder be applied as a light cloud and not simply dumped on the 
surface. The operator should watch the inspected area carefully as the powder 
is being applied, since it is usually easy to see indications form even before excess 
powder is removed. If the current flow is stopped before the excess powder is 
blown off, it is possible that lightly held indications will be removed. 

Residual-Field Method of Inspection. When the residual field left in 
parts is high, it may be sufficiently strong to form adequate indications. How 
ever, residual fields are always weaker than magnetic fields present when mag 
netizing current is flowing. Consequently inspection with the residual method will 
not be so sensitive as the continuous method. When using the wet residual 
method, the length of time that the part is covered by the inspection bath is 
of importance. Long exp9sure of a magnetized part to the wet bath increases the 
build-up of magnetic particles forming the indication. It is usually difficult to 
indicate subsurface defects when using the residual method. This fact is 
sometimes used to determine whether an indication is showing a defect on the 
surface or beneath the surface. To determine this, the indication formed by the 
continuous method is wiped off, and the bath or powder is reapplied. If the indi 
cation returns, it usually means a surface discontinuity. 

VALUE OF FLUX DENSITY. The strength of the magnetic field in 
duced in a part is often referred to as the flux density. The proper value of 
flux density, or strength of field, must be determined for each part. Factors 
which affect the strength of field include the test-object size, shape, and material. 
Field strength is one of the factors which determine the success of magnetic- 
particle inspection. A field that is either too weak or too strong will create too 
little or too much leakage at discontinuities. However, the direction of the mag 
netic field is more important than the flux density. 

The behavior of the magnetic particles on the surface of a part dur 
ing magnetization provides an excellent indication of the strength of the induced 
magnetic field or of the amount of magnetizing current being used. This is 
particularly true when dry powder is used, but it is also noticeable with 
the wet method. With dry powder a field that is too strong will frequently 
cause particles to adhere tightly to the surface and hinder their mobility. In 
fact it may sometimes prevent their moving to a defect and may cause them to 
actually stand on end, parallel to the lines of force. On the other hand, when 
no pattern is observable while the current is applied, it is frequently an indica 
tion that insufficient magnetizing current is being used. The use of over- 


strong magnetic fields may not only mask sought-for defects but may also pro 
duce strong leakage fields at projections, off-sets, corners, or angles. This may 
cause an excessive number of particles to adhere at such locations. Removal of 
these accumulations may be so difficult as to require special cleaning operations, 
even after demagnetization. 

Usually, observing the particle pattern, particularly where it can be easily 
observed (as with dry powders), is as satisfactory a guide to proper flux density 
as is the ammeter reading. 

Current Requirements for Circular Magnetization. Only enough cur 
rent should be used to show the indications of sought-for defects. If the 
magnetizing current is too strong, it may bring out things which are not impor 
tant to the inspection. The best gage of magnetizing current strength is to retest 
sample parts with typical indications which can be checked from time to time. 

Optimum Current Magnitudes. It is difficult to assign arbitrary cur 
rent values for a certain size of part. The value for optimum results is affected 
by the magnetic characteristics of the metal, shape and configuration 
of the part, and the type and severity of discontinuities which are pres 
ent. The current values used for overhaul inspection should be lower than 
those used when inspecting for subsurface discontinuities, since in the over 
haul of most equipment, only fatigue or other surface cracks are usually sought. 
The recommended current value for the inspection of new aircraft parts 
during or just after manufacture is from 600 to 800 amp. per linear inch of 
section thickness when circularly magnetized with direct current. Inspection of 
other parts may require as little as 400 amp. per inch; optimum value for a part 
should be determined by experimentation. For example, using these standards, if 
a cylindrical part is 1 in. in diameter, one should pass from 600 to 800 amp. 
through it. The length of the part does not affect this determination except 
that the electrical resistance of a long part may reduce the current magnitude. 
If the part is 2 in. in diameter, 1200 to 1600 amp. should be used for magnetiza 
tion. If exceptionally heavy accumulations of particles occur, particularly 
at abrupt changes in section, the current should be reduced. The proper direction 
of the magnetizing current and concentration of particles are as critical for reli 
able inspection as the magnitude of the magnetizing current. This does not mean 
that the magnitude of magnetizing current is not important, but rather that there 
are other factors which are equally important. 

Current Requirements for Longitudinal Magnetization. Extensive lab 
oratory tests confirmed by mathematical analyses have recently provided a 
sound basis for determining the proper current to be used when a part is to be 
longitudinally magnetized in a coil. This replaces the older and unreliable rule 
of thumb which has been in wide use previously. For example, a current of 
1000 amp. through a five-turn coil creates a magnetizing force of 5000 ampere- 

Influence of Length/Diameter Ratio. For reliable d.-c. coil magnetization 
(longitudinal), the part to be magnetized should be shaped so that its length 
is at least two or three times as great as its diameter. This relationship is 
referred to as the length/diameter, or l/d, ratio. Knowing the l/d ratio 
and the number of turns in the coil, it is possible to determine accurately the 
required current for coil shots, providing the following conditions are met: 

1. Cross-sectional area of part is not greater than one-tenth the area of the 
coil opening. 


2. Part or section of part to be magnetized is no more than 18 in. long. 

3. Part is held against the inside wall of coil and not positioned in center of coil. 

4. Part has an l/d ratio between 2 and 15. 

5. Part is positioned in coil with its long axis parallel to the applied field (coil 

Formula for Correct Ampere-Turns. If all the above conditions are met, 
then the formula for determining correct number of ampere-turns can be stated 

A , 45,000 

Ampere-turns = . , ( 1 ) 

To use the formula, the l/d ratio is reduced to a number and divided into 
45,000. The result will be the number of ampere-turns necessary to magnetize 
the part properly. The number of ampere-turns obtained should then be divided 
by the number of turns in the coil to give the correct current in amperes. These 
calculations apply only to parts not exceeding IS in. in length. For parts longer 
than 18 in., the same calculation is used, but a separate magnetizing shot is 
given for each 18 in. of its length. An example of the calculation of the coil 
shot for a part 8 in. in length and 2 in. in diameter is 

I = 8, d = 2 

l/d =8/2 = 4 

Ampere-turns = 45,000/4 = 11,250 
If the coil has five turns, then 

Magnetizing current = 11,250/5 = 2,250 amp. 
Fig. 14 shows other examples. 

Part Length 

Part Diameter 

l/d Ratio 






































Fig. 14. Typical coil shot currents (amperes) for a five-turn coil. 

Current Requirements for Flexible Cable Magnetization. The correct flux 
density is easier to determine when using portable equipment because it is 
possible, with the use of prods or clamps for circular magnetization, to vary the 
current setting on the equipment and also the space between the prods/ Too 
heavy a particle accumulation between the points of contact is called banding. 
Banding indicates that the field strength is too great and should be reduced by 
lowering the current setting on the equipment or by increasing the space be 
tween the prods. On large parts the use of from 600 to 800 amp. with a prod 
spacing of from 6 to 8 in. is found to be most effective. 


A longitudinal field is usually created with portable equipment by wrapping 
a cable around the part. The effective field strength is measured in ampere-turns 
and is the product of the current and the number of turns. When using d.c., 
the more turns of cable, the stronger the field, but with a.c. there is a limit to 
the number of turns which will increase the flux density. Three to five turns is 
generally the best number. Since the effective field of a coil extends only 6 to 
9 in. beyond either end of the coil, long parts require several successive shots 
along the part. If the powder accumulation on the ends of a longitudinally 
magnetized part is too heavy, it may prevent good inspection. Such heavy 
accumulations indicate that the current should be reduced. 

Inspection Materials and Their Preparation 

TYPES OF MAGNETIC PARTICLES. Two classes of magnetic particles 
are available, depending upon the vehicle or carrying agent used. The wet 
method particles use a liquid vehicle; the dry method particles are borne by 
air. These particles are not ordinary iron filings but are made of carefully 
selected magnetic materials of proper size, shape, magnetic permeability, and 
retentivity. They are colored to give good color contrast with the surface being 
inspected. The wet particles are best suited for the detection of fine surface 
discontinuities such as fatigue cracks. They are commonly used in stationary 
equipment where the bath can be used until contaminated. It is possible to use 
wet particles in field operations with portable equipment, but care must be taken 
to agitate the bath constantly. This means that the spray applicator used to 
apply the bath should be shaken regularly to keep the particles in suspension. 
Dry particles are most sensitive for use on very rough surfaces and for detecting 
defects beneath the surface. They are usually used with portable equipment. 
Reclaiming and re-using dry particles is not recommended. 

WET BATH MATERIALS. The particles used in the wet method are 
ground in oil and are obtained from the manufacturer as a thick paste. The 
paste must be completely dispersed in a liquid bath of proper consistency, color, 
and flash point. Magnetic paste for the wet method is available in black-, red-, 
and fluorescent-particle coatings. These pastes are equally sensitive in most 
applications for forming indications at areas of magnetic leakage. The selection 
of a particular paste color depends upon the visibility or contrast of the paste 
on the part and the required inspection speed. Magnetic particles are available 
for suspension in vehicles other than oil. It is extremely important that these 
particles be used in accordance with manufacturers' recommendations and in 
equipment designed for their use. 

Instructions for Mixing the Oil Bath for the Wet Method. Cleanliness of 
the equipment and bath is vital for reliable inspection. Special attention should 
be given to cleanliness as well as accuracy in proceeding with the following steps: 

1. Before mixing a new bath, the equipment should be cleaned thoroughly. Re 
move and clean agitator pipe. Clean other pipes in the unit and the tank, 
pump, and strainer. 

2. Weigh or measure paste into a clean container. Measure out 1% oz. per gallon 

3 After' tank and hose have been thoroughly cleaned, close all drain cocks and 
add oil in quantities recommended by the equipment manufacturer. This 
amount varies with different suppliers. Fig. 15 may be used as a guide The oil 
used in preparing the bath should be a light, well-refined petroleum distillate 


Equipment Length, Maximum Possible Gallons of Oil 

Distance Between Contact Plates (in.) To Be Used 

48 8 

54 10 

72 12 

96 15 

144 25 

Fig. 15. Bath capacity of magnetic-particle equipment. 

of low sulfur content, should be treated for odor, and should have a relatively 
high flash point. It should have the characteristics indicated in Fig;. 16. Ex 
cessive viscosity makes it more difficult to form indications. If the flash point 
is too low, a distinct fire hazard exists. If the color is incorrect, the oil may 
mask fine indications. If the sulfur content is too high, the odor at the inspec 
tion station will be disagreeable, cand contact with the oil may cause excessive 
irritation to the skin of the operator. 

Viscosity [kinematic at 100 F. (38 C.) I 3 centistokcs (max.) 

Flash point (tag closed cup) 135 F. (57 C.) (min.) 

Initial boiling point 390 F. (199 C.) (min.) 

End point 500 F. (260 C.) (max.) 

Color (Saybolt) Plus25 

Low sulfur available 

(Copper test: ASTM D129-52) 

Fig. 16. Required oil characteristics. 

4. Add a little oil from the tank to the paste in the container and mix thoroughly. 
Continue to add oil to the container and stir until a thin mixture is obtained. 

5. Turn on pump motor in equipment and pour paste and oil mixture from con 
tainer into the equipment tank. Run pump for several minutes, rinsing out any 
sediment from the container with the hand hose. 

6. Test for proper bath strength and adjust as necessary. 

Caution : Never add magnetic paste directly from the can into the equip 
ment tank. Always mix paste with oil in a separate clean container. Paste 
added directly from its shipping can to inspection equipment will not 
disperse in the bath and can clog the system and make inspection un 
reliable. Wet-method bath for use in spray applicators should also be 
mixed with care. The mixing and storing container should be kept clean. 
Always stir or shake thoroughly before filling the spray applicator. 

Bath Strength. The number of magnetic particles per gallon of fluid in the 
inspection bath is called its strength or concentration. If the bath strength is 
not at the proper level, inspection cannot be reliable. Even when the proper 
processing techniques are used with the correct amount of magnetizing current, 
no indications will form if a sufficient concentration of particles is not present in 
the bath. If there are too many particles in the bath, indications may be 

The usable limits of bath concentration are very broad, but for consistent 
results the bath strength should be the same at all times. Deep cracks will 
usually form good indications with a light bath strength, but a heavier particle 


30 17 

concentration will show up fine defects better. The bath concentration which will 
best detect all defects should be determined and maintained constant. The bath 
strength should be checked daily. 

Instructions for Checking Bath Strength. Check the bath strength every 
day as follows: 

1. Let pump motor run for several minutes to agitate a normal mixture of 
particles and oil. 

2. Flow the bath mixture through hose and nozzle for a few moments to clear 

3. Fill a 100-c.c. centrifuge tube to the 100-c.c. line with fluid from the inspection 

4. After 30 min., read the volume of the settled particles on the scale at bottom 
of centrifuge tube. For black or red paste the reading should be from 1.5 to 
2.0 c.c v while for fluoroscent paste the reading should be from 0.2 to 0.4 c.c., 
as indicated in Fig. 17. Do not include dirt particles in centrifuge tube readings. 

5. If reading is higher than allowable, add oil to bath. Repeat centrifuge tube 
test if in doubt about the bath strength after making addition. 

The bath should be changed regularly, once each week in production or up to 
once a month if it is not contaminated or if inspection volume is low. 

Fig. 17. After 30 min. of settling time, the bath strength is indicated by the vol 
ume of settled particles read on the scale at the bottom of the centrifuge tube. 

For black or red pastes the reading should be from 1.5 to 2 c.c. For fluorescent paste 
the reading should be from 0.2 to 0.4 c.c. for proper particle concentration. 

Selection of Wet Method Paste. Since the particles available for the wet 
method are equally sensitive in most applications, the selection depends upon 
the type of particles which provide greatest contrast to the surface inspected. 


Black- or red-paste indications are viewed under ordinary light. Adequate 
daylight is optimum. The fluorescent-paste particles must be viewed under 
near-ultraviolet, or "black light," of adequate intensity, and the equipment or 
inspection area must be darkened to cut off normal light. 

Advantages of Fluorescent Particles. The main advantage of fluorescent 
particles is increased visibility. In some experiments fluorescent-magnetic indi 
cations have been found to be 100 times more easily visible than black- or red- 
particle indications. This increased visibility permits a lighter concentration of 
paste in the fluorescent magnetic-particle bath. Since fluorescent particles are 
easily visible, it takes fewer to form an indication, and it is easier to detect weak 
leakage fields. Fluorescent particles are ideal on threaded parts and coil 
springs. With threaded parts there is frequently a leakage field at the thread 
root which may attract particles that mask or cover true defects. Since there 
are fewer fluorescent particles in the bath, this masking effect is reduced con 
siderably. Coil springs are difficult to inspect because of their shape. Care must 
be taken to view all surfaces. When fluorescent particles arc used on coil 
springs, indications attract the attention of the inspector. 

For high speed and high volume inspection, fluorescent indications permit 
inspection speeds as much as six to ten times those possible when particles are 
viewed under white light. 

WATER SUSPENSION PASTES. Red, black, and fluorescent pastes are 
available for suspension in water. It is extremely important that these particles 
be used in accordance with manufacturers' recommendations and in equipment 
designed for their use. The use of water reduces the fire hazard present with 
oil suspensions and also reduces probabilities of dermatitis. The fluorescent 
paste for use with water is extremely sensitive for fine discontinuities and pro 
duces a minimum of background fluorescence. 

The disadvantages of water suspensions should be considered before using 
them. Water may corrode the inspection equipment as well as the parts in 
spected. Sometimes water suspensions do not readily wet the part and some 
areas may not get any magnetic particles. Wetting agents are added to offset 
this condition, but excessive additives may cause foaming. It is more difficult 
to keep a water bath at optimum balance and to maintain the correct viscosity. 
Evaporation and freezing are factors that must be considered. The possibility 
of water damage to electrical components is the reason that water suspensions 
should be used only in equipment built for that purpose. 

PARTICLES FOR THE DRY METHOD. The particles used in the dry 
method are in the form of a powder. They are available with red, black, gray, 
and fluorescent coatings. The magnetic properties and particle size are similar 
in all colors, making them equally efficient. The choice of powder is then deter 
mined only by which powder will give the best contrast on the test objects. 

Application of Dry Particles. Dry particles depend upon air to carry them 
to the surface of the part, and care must be taken to apply them correctly. The 
particles should float to the inspected surface as gently as possible and 'not be 
thrown against it forcibly. As they float to the magnetized surface, the particles 
are free to be influenced by magnetic leakage fields and form indications. Pow 
der forcibly applied is not equally free to be attracted by leakage fields. Rolling 
a magnetized part in powder or pouring the powder on the part is not recom 
mended. When dry particles are used, it is important to watch the magnetized 
area as the particles are applied to it, when the indications can be seen to form. 


Mechanical Powder Blower. One of the best ways to apply dry powder 
is by specially designed, mechanical powder-blowers. The air stream of such 
blowers is of the low velocity necessary to apply a cloud of powder to the test 
area. Mechanical blowers can also deliver a light stream of air to blow off excess 
powder gently. 

Other Means of Applying Powder. Powder can also be applied with small 
rubber spray bulbs or shakers. In all cases of powder application it is important 
that a gentle cloud of powder be applied to the area being inspected. 

Special Requirements for Fluorescent Inspection 

BLACK LIGHT REQUIREMENTS. The fluorescent coating on the 
magnetic particles is most brilliant when viewed under near-ultraviolet light of 
sufficient intensity, of a specific wavelength (3650 A). This wavelength is be 
tween the visible and ultraviolet in the spectrum and is usually considered as 
noninjurious to the skin or eyes. Illumination must be of proper intensity and 
wavelength, or the effectiveness of the inspection will be greatly reduced. The 
approved light source for producing black light is a 100-watt, reflector-spot-type 
of mercury- vapor bulb. This bulb operates from 110-volt, 60-cycle, single- 
phase, alternating current and requires a special transformer. A special filter is 
fastened to the front of the mercury-vapor bulb to absorb visible light and to 
pass only light of the proper wavelength required for inspection. Cracked or 
broken filters should be replaced immediately. The filter and face of the bulb 
should be cleaned regularly so that dust and dirt do not reduce transmission of 
black light. 

Influence of Power Supply Voltage. Lamps used in black lights are very- 
sensitive to fluctuations in power line voltage, and frequently a small percentage 
of line voltage drop may cause the bulb to be extinguished. It is recommended 
that the circuits which supply the black lights have a constant voltage. If 
necessary a constant-voltage transformer should be used in the circuit. 

Checking Light Intensity. As a general rule, black lights of the recommended 
type emit light of sufficient intensity for average inspection as long as they still 
function. Black light intensity of 90 to 100 foot-candles is suitable for the 
detection of all indications. This intensity is obtained with a normal 100-watt, 
reflector-spot, mercury-vapor bulb at a distance of 15 in. from the face of the 
bulb. If indications are heavy and broad, a light intensity of 20 to 25 foot-candles 
may be sufficient. 

The test for light intensity should be made with a Weston Sight Light Meter 
No. 703, Type 3 (unfiltered), equipped with a 10X multiplier disc or equivalent. 
To measure intensity, the bulb should be equipped with a regular black-light 
filter. Care should be taken to see that bulb and filter are clean. To read the 
light intensity correctly, the meter should face the light source at the distance 
normally used for inspection. 

THE INSPECTION AREA. When using fluorescent particles, the inspec 
tion area should be free from random fluorescent materials, since these are 
likely to confuse the inspector. The operator may experience clouding of vision 
if black light is permitted to shine directly or reflect into the eyeball. This 
cloudy sensation in the eye disappears when such illumination ceases. Because 
it is a somewhat disagreeable sensation, it is desirable to arrange the lights m 


the inspection area so that neither direct nor reflected light shines into the in 
spector's eyes. 

Dark Adaptation of Inspector. For best results with fluorescent particles, 
the inspection should be done in a darkened area. The darker the area of in 
spection, the more brilliant the indication. Dark adaptation from bright, white 
light to levels necessary for reliable fluorescent magnetic inspection may require 
several minutes. The inspector should become adapted to the darkened area 
before starting inspection. He should also avoid going from the dark booth to 
lighted areas and back again without allowing sufficient adaptation time. 

Principles of Demagnetization 

RESIDUAL MAGNETISM. Parts fabricated from ferromagnetic material 
retain a certain amount of residual magnetism (or remanent field) after exposure 
to a magnetizing force. The magnitude of field retained by a part is dependent 
upon the 

1. Magnetic characteristics of the particular material. 

2. Strength of applied magnetizing force. 

3. Direction of magnetization (longitudinal or circular). 

4. Geometry of the part. 

Demagnetization of ferromagnetic material in industry is limited to reducing the 
degree of magnetization to an acceptable level, since complete demagnetization 
is usually impractical. 

sideration should be given to whether or not parts require demagnetization. 
There are four basic reasons for requiring demagnetization. 

1. When subsequent machining or an arc-welding operation is to be performed on 
the part, a strong residual field in the part may interfere. The field may attract 
and hold chips or particles to the surface of the part and adversely affect 
surface finish or cutting action. In arc welding, the presence of strong residual 
fields may deflect the arc. 

2. Subsequent operation of the part may be impaired if its leakage field is exces 
sive. The attraction of chips or particles to rotating parts may cause malfunc 
tions, especially with bearings and bearing surfaces. Strong residual fields are 
a source of excessive friction between moving parts; for instance, between a 
piston and cylinder wall. 

3. The leakage fields emanating from a part may interfere with instrumentation. 
Even slightly magnetized aircraft parts may cause the magnetic compass of an 
airplane to read erroneously. 

4. The presence of a residual field may hamper removal of chips or particles dur 
ing subsequent cleaning operations. 

Unless at least one of the above rules is applicable, demagnetization may be 

CAUSES OF MAGNETIZATION. A ferromagnetic part can become mag 
netized in many ways, such as the following: 

1. The earth's magnetic field can impart a fairly strong residual field to many 
parts. This usually occurs when a part is shocked or vibrated while its long axis 
is parallel to the earth's field. Such residual magnetization may become quite 
significant for long parts subjected to severe vibrations in service. The inten 
sity (horizontal component) of the earth's field in the United States is 


approximately 0.2 oersted, equivalent to the field intensity at the center of a 
five ampere-turn coil, 12 in. in diameter. 

2. Frequently parts become magnetized when subjected to an electric arc-welding 

3. Parts are purposely magnetized in magnetic-particle inspection. 

4. Parts operating or stored in close proximity to high-current electric circuits 
often acquire residual magnetization. 

5. Accidental contact with the pole of a permanent magnet or other highly 
magnetized objects may produce a residual field in a part. 

6. Magnetic chucks are another source of residual fields. Strong residual fields 
may be left if a part is removed from the chuck before the demagnetizing cycle 
has been completed. Sometimes the method of demagnetizing the chuck may 
be inadequate and leave the part partially magnetized. 

7. Low-frequency induction heating can induce very strong residual fields in a 

ence of a residual field is much more evident on parts magnetized in the 
longitudinal direction. This is clue to the relatively high concentration of 
external fields (lines of magnetic flux entering and leaving the part) associated 
with longitudinally magnetized parts. These conditions of magnetization are 
easily detected by indicating devices and by the attraction of other magnetic 
parts or particles. Therefore this type of residual field is usually most objec 
tionable; however, it is also the most responsive to demagnetization. 

Unlike longitudinal residual fields, the circular residual field exhibits little or 
no external evidence of its existence. The field is almost entirely confined to the 
part itself, depending somewhat upon part geometry and method of magnetiza 
tion. For example, if current has passed through a homogeneous piece of ferro 
magnetic bar stock having a circular cross-section, the resulting residual field is, 
for all practical purposes, undetectible without altering the bar in some manner. 
Virtually no leakage fields emanate from its surface, since the magnetic flux 
path is closed upon itself within the part. The internal residual field is much 
stronger than if the part had been magnetized longitudinally by a coil (or 
solenoid) having comparable field intensity. Since there is very little leakage 
flux associated with circular fields, they are not so objectionable as longitudinal 
residual fields. Demagnetization of a circularly magnetized part can be very 
difficult. Reorientation of the circular field into a longitudinal field before 
demagnetization may be advantageous in some instances. 

rials retain a certain amount of residual magnetism after being subjected to a 
magnetizing force. When the magnetic domains of a ferromagnetic material have 
been oriented by a magnetizing force, some domains remain so oriented until an 
additional force in the opposite direction causes them to return to their original 
random orientation. This force is commonly referred to as coercive force. 

Magnetic Hysteresis. This is graphically illustrated by the hysteresis curve 
of Fig. 18. It may be assumed from Fig. 18 that the residual field (B r ) could be 
reduced to zero simply by subjecting the residually magnetized material to a field 
intensity equal to H c in the opposite direction of magnetization. Theoretically 
this is a valid assumption, but it is not a practical method of demagnetization. 
It is not practical because H c is a variable which is dependent upon a number 
of factors, such as the type of material, geometry of the part, homogeneity, 
hardness, and the intensity of the field which originally magnetized the part. 
However, a slight modification of this method does make it quite practical. 


Magnetization, B 




AHc +Hin +H 

Force, H 



Fig. 18. Typical hysteresis curve, 

Use of Diminishing A.-C. Fields. The coercive force //,, is always loss 
than the magnetizing force H m . Therefore, if a part is subjected to a mag 
netizing force initially greater than #, which is alternately reversed in direction 
while being gradually reduced in magnitude, H c and B r will also be reduced in 
magnitude. Both H and B r will eventually approach zero, as shown in Fig, 19. 

Intentionally overshooting H ( . with each reversal can eliminate the guess 
work associated with the theoretically possible, single-shot demagnetize! 1 . This 
principle is the basis for practically all demagnetizing methods used today. 

RETENTIVITY AND COERCIVE FORCE. As a general rule, high 
coercive forces are associated with harder materials and low coercive forces with 
softer materials. Therefore hard materials usually offer more resistance to 
demagnetization and require more intense demagnetizing fields than softer mate 
rials. The fact that a part retains a strong residual field (B r ) is not necessarily 
indicative of a high coercive force (H e ), since some materials retain appreciable 
residual fields and yet are very easily demagnetized. On the other hand, some 
materials which retain only a relatively weak residual field can be extremely 
difficult to demagnetize because of their high coercive force. 

METHODS OF DEMAGNETIZATION. Practically all demagnetizing 
methods in use today are based upon a common procedure. This it? the applica 
tion of a magnetizing force of sufficient intensity to overcome the initial coercive 
force (H c ), and which is alternately reversed in direction and gradually reduced 
in magnitude. There is one exception which should be mentioned. A part can be 
demagnetized by raising its temperature above the Curie point (approximately 

Magnetization, B 


Force, H 

Fig. 19. Diminishing hysteresis curve resulting from the alternate reversal and 
reduction of the applied field. 

1200 F. to 1600 F., depending upon the particular material). Although this 
method will probably result in the most thorough demagnetization, it leaves 
much to be desired in the way of convenience and practicability. 

A.-C. Coil Demagnetization. An a.-c. coil is the most common method of 
demagnetization. The coil is usually designed to operate at line voltage and 
frequency (usually 60 c.p.s.). When a part is placed in the coil, it is subjected 
to a reversing field due to the cyclic action of the current. The magnitude of the 
field can be gradually reduced by slowly withdrawing the part from the coil, 
or the part can be held stationary while the coil is being withdrawn. The coil 
should not be de-energized until the part has been withdrawn to a position be 
yond the influence of the coil field. This method is advantageous for high 
production rates, since a properly designed coil can be energized continuously 
while a steady stream of parts is conveyed through the coil opening. Typical 
a.-c. coil demagnetizes are shown in Figs. 20 and 21. 

Modified versions of this method are sometimes employed. The reduction in 
field intensity is often obtained by reducing the current to the coil in a pre 
determined manner while the part remains within the coil until the current 
has been reduced to zero. Current control is achieved by various means, such as 
a saturable reactor or an autotransformer in conjunction with a tap switch. 

The a.-c. coil method is sometimes ineffective on large parts, due to lack of 
penetration, since the a.-c. magnetic field is confined fairly well to the surface 



Mugnailux Corporation 

Fig. 20. Manually operated a.-c. coil demagnetize! complete with roller carriage 
and automatic timing switch. 

Magiudlux Corporation 

Fig. 21. Alternating-current coil demagnetizer incorporated into a convey orized 
unit for magnetic-particle inspection. 


of the part by the currents induced within the part itself. The higher the 
frequency, the more pronounced the skin effect. 

To minimize the effects of the earth's field, particularly with long parts, the 
demagnetizing coil should be oriented with the central axis pointing east and 
west. Normally, best results are obtained if the long axis of the part is parallel 
to the coil axis. 

Reversing D.-C. Demagnetization. In demagnetization with reversed direct 
current, the desired magnetic field is obtained by means of a coil or passage of 
current through the part itself. The direct current is alternately reversed in 
direction and reduced in amplitude. If a coil is used, the part is left in the coil 
until the demagnetizing cycle has been completed. This method provides deep 
penetration and is usually very effective on parts that are difficult to demag 
netize. Usually, ten reversals and reductions in current will provide satisfactory 
results. However, more reliability can be attained by utilizing a greater number 
of reversals and reductions in current (approximately 30) . 

A.-C. Circular Field Demagnetization. Demagnetizing with an a.-c. circular 
field is similar to the a.-c. coil method, in that the field reversal is provided by 
the cyclic nature of the current. However, in this case, the desired field is ob 
tained by passing current through the part. The magnitude of the current is 
systematically reduced to zero by some suitable device. This method is usually 
employed on rather large parts, just after they have been inspected by the 
magnetic-particle method. Some inspection units have built-in devices for the 
systematic reduction of current, and it is found to be convenient to demag 
netize bulky parts before they are removed from the unit. 

A.-C. and D.-C. Yoke Demagnetization. Yokes are used primarily for 
demagnetizing small parts having very high coercive forces. They are C-shaped 
and are usually designed for demagnetization of a specific type of part. Some 
a.-c. yokes are similar in operation to the a.-c. coil method whereby the part is 
passed between the pole faces (maximum field intensity) and then withdrawn. 
A modified version of this is a solenoidal electromagnet which is light enough 
to be passed over the surface of a part; however, its effectiveness is limited to 
a few special cases. 

Direct-current yokes are usually based upon the reversing d.-c. method; 
however, some designs utilize a damped oscillation to obtain the required revers 
ing and diminishing field. The oscillation is derived from a circuit containing 
capacitance, inductance, and resistance. Design of the circuit is usually based 
upon a specific part. In general, d.-c. yokes probably provide the deepest pene 
tration of any method. 

Methods for Measuring Leakage Field Intensities 

TYPES OF MEASUREMENTS. Leakage field intensities can be measured 
by quantitative or comparative methods. Quantitative measurements usually 
involve the use of instruments, such as D'Arsonval-type meter movements, in 
conjunction with search coils or probes (see section on Magnetic-Field Test 
Principles). Such instruments are often classified as laboratory equipment and 
require competent laboratory personnel for their operation. As a general prac 
tice, leakage field intensities are measured by the comparative method. A few 
of the methods employed are discussed in the subsequent text. 



FIELD INDICATOR. The field indicator, a pocket instrument, is used to 
determine the relative intensity of leakage fields emanating from a part. The 
construction of a typical field indicator is shown in Fig. 22, and the theory of 
operation is quite simple. As indicated in Fig. 22, an elliptically shaped, soft- 
iron vane is attached to a pointer which has pivot points in top and bottom 




! M 



IS \ 






Fig. 22. Typical field indicator. 

bearings. A rectangularly shaped, permanent magnet is mounted in a fixed 
position directly above the soft-iron vane. Since the soft-iron vane is under the 
influence of the magnet, it will align its long axis to correspond with the direc 
tion of the leakage field of the magnet. In so doing, the vane becomes mag 
netized and has a magnetic pole induced at each end of its long axis. On the end 
of the vane, which is below the south pole(s) of the magnet, a north pole will 
be induced. Correspondingly, a south pole will be induced at the other end of 
the vane, below the north pole of the magnet. 


Indications. The pointer normally points to zero on the graduated scale, in 
the absence of external fields. When the magnetic north pole of a magnetized 
part is moved close to the pivot end of the pointer, the south pole of the vane 
will be attracted toward the part. The pointer will consequently move in the 
plus direction. The restraining torque is provided by the tendency of the vane 
to remain aligned with the leakage field of the permanent magnet. 

Procedure. The relative intensity of a leakage field is measured by bringing 
the field indicator to the part and noting the deflection of the pointer. The edge 
of the field indicator case at the pivot end of the pointer should be closest to 
the part being investigated. For small parts it is sometimes advisable to place 
the instrument face up on a nonferrous surface and then slide the part up to 
the instrument. The required degree of demagnetization is usually specified as 
a maximum field indicator reading (perhaps 2.0 divisions). The readings 
obtained with a field indicator are relative units. However, having once deter 
mined the maximum allowable field, the field indicator can be used as a con 
venient comparator. 

COMPASS INDICATOR. A compass is sometimes used for indicating the 
presence of external leakage fields. A compass is placed upon a nonmagnetic 
surface and a magnetized part (aligned due east and west) is moved slowly 
toward the east or west side of the compass case. The presence of an external 
leakage field from the part will cause the compass needle to deviate from its 
normal north-south alignment. However, demagnetized parts will cause the 
needle to deviate from its normal position if the compass case is not approached 
from an easterly or westerly direction. The theory of operation is very similar 
to that for the field indicator, the compass needle being a permanent magnet 
with a magnetic pole at either end. Restraining torque is provided by the 
tendency of the needle to align itself with the earth's north-south magnetic field. 
The sensitivity will depend upon the type of compass. 

STEEL-WIRE INDICATOR. A piece of iron or steel tag wire can be 
fashioned into a fairly good detector when nothing else is available. By forming 
a loop at one end of a piece of tag wire approximately 6 in. long, it can be 
suspended from a second wire supported in the horizontal plane. The part in 
question is then brought into contact, near the free end of the vertically 
suspended wire. The presence of leakage fields will cause the wire to deviate 
from its normal vertical position as the part is slowly withdrawn in a horizontal 
direction Care must be taken to demagnetize the vertically suspended wire 
between each test. Small pieces of tag wire about 1 in. long are sometime^ used 
to indicate the presence of leakage fields. The piece of demagnetized wire is 
placed upon a horizontal nonmagnetic surface and the part in question is placed 
on top of it. If the piece of tag wire can^be lifted off the surface as the part is 
slowly raised, the leakage fields are excessive. 

might be noted that all the methods mentioned pertain to the measurement or 


ofiedl eo.uip.nent. Nevertheless, demagn.liz.t 


to prevent the attraction and holding of fine magnetic particle? to the surface 
of the part. 

Checking Demagnetization. Since circularly magnetized partis can be very 
difficult to demagnetize, merely going through the usual demagnetization opera 
tion does not necessarily mean that ti part has been demagnetized. Some means 
of determining the effectiveness of the demagnetizing method should be em 
ployed. Similar parts that have been scrapped can prove to be very helpful in 
such situations. By magnetizing and demagnetizing them in the proposed man 
ner, effectiveness of demagnetizing methods can in many cases be determined by 
sectioning the part at right angles to the known direction of magnetization. In 
so doing, the path of the circular residual field is interrupted, and any residual 
field will now be easily detcctible in the form of a leakage field. For example, the 
demagnetization of a bearing race magnetized with a central-conductor shot 
can be checked by breaking or cutting the ring-shaped part into two pieces. 
The presence of a residual field will be evident in the form of leakage fields at all 
four exposed ends. 

DEMAGNETIZATION PROBLEMS. Several situations listed here may 
prove helpful when demagnetization of parts becomes problematical. 

Shielding by Magnetic Shunting. Whenever possible, a part requiring de 
magnetization should be demagnetized before being assembled with other parts. 
Occasionally demagnetization of a part is attempted after it has become part 
of an assembly and is adjacent to or surrounded by other magnetic materials. 
In such cases, the demagnetizing field may very well be shunted through the 
adjacent material rather than through the desired part, and demagnetization 
will be ineffective. Small parts should not be passed through a demagnetizing 
coil in bundles or heaped in baskets for the same reason. The parts in the center 
will be shielded from the demagnetizing field by the outer layer of parts. Mag 
netic baskets or containers may be objectionable for the same reason. 

Small Parts Having a Small Length/Diameter Ratio. A demagnetizing 
field may be inadequate for small parts having a length-to-diameter (l/d) ratio 
less than three to one (3:1). This situation can usually be corrected by increas 
ing the l/d ratio by the addition of ferromagnetic pole pieces at either end 
of the part. Pole pieces should be at least 6 in. long and approximately the same 
diameter as the part. With pole pieces at both ends, the part is passed through 
the demagnetizing coil in the usual manner. For large lots of such parts it is 
sometimes convenient to pass the parts through the demagnetizing coil, end 
to end in chain fashion. 

Demagnetizing-Coil Field Strength Apparently Too Weak. The field 
intensity of a coil is a function of the coil diameter as well as of the applied 
ampere-turns. In general, the smaller the coil diameter, the more intense the 
field for the same applied ampere-turns. Therefore, the use of a smaller diameter 
coil of the same ampere-turn rating or greater may be advantageous. However, 
passing the part through the coil as close as possible to the inside wall may be 
all that is required, since the field intensity is greater there than at the center. 

Fugitive Magnetic Poles. Occasionally attempts to demagnetize a part will 
result in chasing magnetic poles from one location on the part to another. This 
may be due to the particular geometry of the part or its orientation with 
respect to the demagnetizing field. This situation can usually be alleviated by 
rotating the part while it is within the influence of the demagnetizing field. In 


particular, ring-shaped parts such as bearing races can be rolled through an 
a.-c. coil to obtain the desired results. 

The mere fact that a part has been passed through an a.-c, coil, or subjected 
to some other method of demagnetization, does not necessarily mean that the 
part has been suitably demagnetized. The results obtained by any method 
should be carefully evaluated before concluding that a given method is applicable 
to the particular part in question. 


CAINE, J. B. "Magnetic Particle Inspection Standards," Foundry, 83, No. 12 (1955): 

CATLIN, F. S. "Testing Methods Valuable in Improving Casting Design," Foundry, 

85, No. 3 (1957) : 168, 
CLARKE, J. E. } R. A. PETERSON, and T. J. DUKSHEATH. "Automatic Inspection of 

Pipe," Welding Erigr,, 36, No. 2 (1951) : 38. 
Cox, R., and W. A. GARDNER, "Maintenance Is Key to Service Fleet Success," Drill- 

ing, 12, No. 5 (1951); 54. 
DOANB, F. G,, and C. E, BETZ. Principles of Magnaflux. 3d ed. Chicago: Photopress, 


GILBERT, L, and W, B. BUNN. "Inspection Techniques for Quality Welding," Weld 
ing J, 32 (1953): 614. 
KELLY, S, G., JR. "New Magnetic Particle Testing Technique Speeds Inspection of 

Ferrous Parts," Materials "& Methods, 33, No. 6 (1951) : 66. 
"Magnetic Particle Inspection," Welding Handbook. 3d ed. New York: American 

Welding Society, 1950. P. 957. 
McCABE, J, L., and B. HIRST, "Use of Tap Water in Magnetic Particle Inspection," 

Metal Progr, 70, No, 1 (1956): 77. 
McCuTCHEON, D. M. "Testing, Inspection and Quality Control," Metal Progr., 68, 

No. 3 (1955): 141. 

MIGEL, H. "Magnetic Particle, Penetrant and Related Inspection Methods As Pro 
duction Tools for Process Control," Steel Processing, 41 (1955) : 86. 
"Standardization Aids Weld Inspection of Tractors," Welding Engr., 42, No. 4 (1957) : 

THOMAS, W. E. "Economic Factors of Nondestructive Testing;," Nondestructive 

Testing, \l. No. 4 (1953): 9. 
. "Magnaflux Inspection," Proc. 29th Ann. Conv. Automotive Engine Rebuild- 

en Assoc., May 7-9, 1951. 
WILSON, T. C. "Nondestructive Testing of Refinery Equipment," Petrol Engr., 27, 

WOLDMAN, N. E. "Some Notes on Fatigue Failures in Aircraft Parts," Iron Age, 

162, No. 24 (1948): 97. 






Selection of Equipment 

Range of available types 

Equipment selection considerations 

Manual Inspection Equipment 

Small, low current units 2 

Hand-held yoke 2 

Hand -held electromagnetic yoke kit for 

magnetic particle inspection (/.I) 3 

Portable a.-c. prod and cable magnetizing 

unit 3 

Magnetic particle crack detector with 
accessories in built-in tool box (/. 2) .... 4 

Small kit for cable magnetizing 3 

Medium unit for cable magnetizing 3 

Medium cable magnetizing unit (/. 3) .... 

Medium and heavy duty equipment 

Medium service a.-c. and half -wave unit .. 
Combination a.-c. and half-wave mag 
netizing unit (2000-amp. output) (/. 4) 
Ratings of a.-c. and half -wave units (/. 5) 
Heavy-duty a.-c. half-wave unit with auto 
matic demagnetizer 6 

Combination a.-c. and half- wave magne 
tizing unit, 3000-amp. type with 30-point 

demagnetizing current control (/. 6) 7 

Three-phase rectifier d.-c. unit 7 

Stationary equipment 7 

Horizontal wet -method a,-c. units 8 

Horizontal wet-method a.-c. units with 

automatic demagnetizer 9 

Heavy-duty a.-c. wet-method magnetic 
particle inspection unit, for parts up to 

96 in. long (/. 7) 8 

Horizontal wet-type d.-c. units 9 


Horizontal wet-type d.-c. units for very 
heavy parts ............................. 9 

Horizontal d.-c., wet-method magnetic- 
particle inspection unit providing both 
longitudinal and circular magnetization, 
with 30-point demagnetizing switch (/. 
8) ....................................... 9 

Wheel-mounted wet-method units ............ 10 

Mechanized Inspection Equipment 

Semi-automatic inspection equipment ........ 10 

Single-purpose semi-automatic equipment .... 10 

Gas-turbine compressor blade test .......... 10 

Three -shot semi-automatic magnetic par 

ticle inspection unit (/. 9) .............. 11 

Conveyorized compressor blade test ........ 11 

Conveyorized magnetic-particle inspection 
unit providing three magnetizing shots 

at one station (/. 10) .................... 12 

Automatic inspection of steering spindles . . 12 

Aircraft engine cylinder inspection ......... 12 

Multi-purpose semi-automatic equipment .... 12 

Automotive f orgings test .................... 12 

Conveyorized magnetic- particle inspection 
unit providing head shot, coil shot, or 

both, as required (/. 11) ................ 13 

Conversion of hand-operated units ......... 13 

Inspection of automotive steering parts .... 13 

Fully-automatic, magnetic- particle test equip- 

ment .................................... 14 

Photoelectric white- light scanner auto 
matically locates and marks defective 
areas revealed by magnetic particles (/. 

12) ...................................... 14 

Equipment selection .......................... 15 

Demagnetizes ................................ 15 

Bibliography .................................. 15 




Selection of Equipment 

RANGE OF AVAILABLE TYPES. More than fifty different varieties of 
standard magnetic-particle test equipment are available for industrial use. Over 
one hundred special-purpose types are used, and new designs are continually 
being developed to meet new needs. This variety in equipment for different uses, 
materials, methods, and techniques of magnetic-particle testing has developed 
because of wide industrial use and the flexibility inherent in the method. Over 
3000 plants in hundreds of industrial groups that make or utilize magnetic mate 
rials, parts, or machinery in the United States currently employ magnetic- 
particle inspection. 

volved in selection of magnetic-particle inspection equipment include: 

1. Type of magnetizing current. 

a. Alternating current (a.c.). 

b. Direct current (d.c.). 

c. Half -wave rectified current (h.-w.d.c.). 

d. Multiple magnetizations (combination currents). 

e. Permanent magnets. 

2. Location and nature of testing. 

a. Portable equipment. Portable magnetic-particle inspection equipment is 
taken to the inspection site. It is used to test large castings and weldments, 
assembled or welded structures, or parts of assemblies tested without dis 

b. Fixed equipment. Large or small test parts are brought to a fixed inspection 
station by conveyor, by plant truck, or by hand. 

c. Production line. Inspection is part of the production line operation. Parts 
can be tested at any stage of manufacture, and handling can be automatic, 
semi-automatic, or manual. Either percentage sampling or 100 percent 
inspection may be done at one or more locations along the production line. 

d. High volume, single part. Where parts of only one design are manufac 
tured in high volume, specialized testing equipment may be best for mini 
mum testing cost per part. An example is testing of automotive connecting 
rods at rates up to 2800 per hour on a two-shift basis. 

e. High volume, various parts. In many plants, inspection of a variety of sizes 
and designs of parts results in very high volume on a single piece of test 
equipment. Eight, ten, or twenty different parts may be inspected on one 
piece of testing equipment in lots of several thousand each, at rates of 
several hundred or several thousand per hour. 

f. Low volume, varying design and size of test objects. This test situation is 
similar to the preceding case (e), but volumes may be much lower. Also, in 
production job shops, the variety of part designs and sizes may be much 
greater, up to several hundred different parts. 



g. Combinations providing flexibility for changing conditions. Some test 
needs vary from time to time between two or more situations. For example, 
percentage production inspection may be carried out until excessively 
numerous defects are located. Then 100 percent testing is used until the 
cause of defects is again under control. Testing equipment in such situations 
must meet various needs. 

3. Test materials used. 

a. Wet bath, visible. 

b. Wet bath, fluorescent. 

c. Dry powder applied manually or automatically. 

4. Purpose of test. 

a. Maximum sensitivity. Maximum sensitivity inspection is used for parts 
which are extremely critical, whose failure would endanger human life or 
cause failure or breakdown of an entire equipment assembly. For such parts, 
100 percent reliability must be assured, and maximum sensitivity may be 
required on all areas of the part. 

b. Controlled intermediate sensitivity. Over-inspection is to bo avoided on 
many normal industrial products or consumer goods. In many cases very 
fine defects need be located only in limited, highly stressed areas on the 
parts. In some cases defects in one direction are serious, while those in 
another direction are of no consequence. 

c. Minimum sensitivity. In some parts of low critical nature or of heavy 
structure, only very severe cracks need be located. Here sensitivity and 
application of the test must be controlled to avoid over-inspection and over- 

5. Area inspected. 

a. Entire part. For very critical parts that carry high stress, or where stresses 
and loads have not been thoroughly evaluated, the entire part often requires 
inspection. The inspection equipment must permit positive inspection at 
lowest possible operating cost. 

b. Selected areas. On many industrial parts only small areas carry a high stress 
and often in only one direction. The inspection, then, is applied only where 

For each test situation, one type of equipment is best suited to achieve the 
required test at lowest cost per piece. Several types of magnetic-particle in 
spection equipment are described in this section, including most of those in widest 

Manual Inspection Equipment 

Approximately twenty standard sizes of portable magnetic-particle inspection 
equipment are in use. They vary from small hand-held yokes which consist of 
permanent magnets or electromagnets energized from 115-volt, a.-c. lines to 
large, 10,000-amp., heavy-duty power units used for the inspection of large 
castings, weldments, or forgings. For purposes of economy to the user, equip 
ment is not built for continuous operation at the maximum amperage rating. 
For example, a 1000-amp. rated unit may have around 400-amp. rating for 
continuous duty, although 1000 amp. can be delivered for a short duty cycle. 

SMALL, LOW CURRENT UNITS. Small, low-cost portable equipment 
types include power supplies for prod and cable magnetization. 

Hand-held Yoke. The hand-held, electromagnetic yoke shown in Fig. 1 sup 
plies a high intensity, unidirectional magnetizing field between the poles when 
placed on a magnetic part and energized. It operates from 115-volt, 60-c.p.s. 
a.-c. lines on less than 6 amp. It can be used with d.c. from a 12-volt auto- 



motive battery, drawing 12 amp. This yoke is primarily used (1) to detect 
surface cracks of moderate to large size, as in gray iron; (2) for locating cracks 
in welds or castings; and (3) for locating fatigue cracks during overhaul of 
large assemblies. 

Magnaflux Corporation 
Fig. 1. Hand-held electromagnetic yoke kit for magnetic-particle inspection. 

Portable A.-C. Prod-and-Cable Magnetizing Unit. European developments 
have led to a small, portable, a.-c. magnetizing unit for prod and cable mag 
netization, as shown in Fig. 2. The compact kit includes a tapped transformer 
for control of magnetizing current magnitude, cables with prod terminals, meter 
to indicate current magnitude, and accessories. The prods may be coupled with 
a special fitting for cable magnetization. A field strength indicator included 
with the kit is intended for checking field strengths on the surface of test objects, 
to aid in determining optimum magnetizing currents. 

Small Kit for Cable Magnetizing. Another small cable magnetizing unit sup 
plies a.-c. magnetizing currents up to 500 amp. for intermittent use, when operated 
from a 115-volt, 60-c.p.s., a.-c. line. Parts are magnetized with prods or cable 
coils. Magnetizing fields are adequate to indicate fatigue and other surface 
cracks in shafts, beams, or machine parts up to several inches in diameter. Dry 
powder or magnetic-particle baths are normally used with this equipment. 

Medium Unit for Cable Magnetizing. A medium cable unit supplies up 
to 500 amp. of magnetizing current when operated from 115-volt a.c. (Fig. 3). 
Both half-wave rectified and a.-c. magnetizing currents are available, and use of 
either is at the option of the operator. Magnetization can be done by cable 
coils or through prods. The unit illustrated can supply a continuous output of 



Rontgen Teehirische Dienst 
Fig. 2. Magnetic-particle crack detector with accessories in built-in tool box. 

500 amp. through 30 ft. of magnetizing cable; it is ruggedly built for general 
shop or field use in maintenance inspection. 

supply 1000 to 10,000 amp. of a.c., d.c., or half-wave rectified magnetizing cur 
rent. In inspection of medium-sized weldments and castings, the equipment most 
widely used supplies a.c. or half-wave current at the option of the operator. 
Sensitivity may be controlled to reveal only surface cracks with a.c., or surface 
cracks and subsurface defects with half-wave current, The very high amperage 
units provide up to 10,000 amp. for over-all magnetization of medium to large 
castings and forgings. 

All equipment rated above 500 amp. operates from either 220- or 440-volt, 
single-phase or three-phase, a.-c. supply lines. The input current requirements 
will depend in part upon the duty cycle required, the magnetizing amperage 
drawn, and electrical characteristics of the magnetizing circuit. For mobile opera 
tion, gasoline generator units are used to supply 220- or 440-volt a.c. to magnetic- 
particle test units. 

Medium Service A.-C. and Half- Wave Unit. Widely used, portable units 
supply a.-c. or half-wave magnetizing currents of 1000 to 2500 amp., selected 


31 5 


Magnaflux Corporation 

Fig. 3. Medium cable magnetizing unit. 



Magnaflux Corporation, 
Fig. 4. Combination a.-c. and half -wave magnetizing unit (2000-amp. output). 

by the operator (Fig. 4). These medium-service units are representative of sev 
eral in the low end of the high current group of heavy industrial portable units. 
Powered by either 220- or 440-volt, 50- to 60-c.p.s. lines, the two units supply 
magnetizing currents as shown in Fig, 5. Where tests are to be made with long 
leads at distances of 80 to 100 ft. from the unit, the unit supplies only 450 amp. 
This is usually less than is required for inspection of heavy weldments or castings. 
Current selection is made with an eight-point switch on both units. 

Unit Rating 

Current Output 

Output Amperes 


Continuous Long Lead Test 




500 450 



1000 450 




600 800 



1250 800 

Fig. 5. Ratings of a.-c. and half-wave units. 

Heavy-Duty, A.-C., Half -Wave Unit with Automatic Demagnetizer. The 

heavy-duty unit shown in Fig, 6 is similar to the 3000-amp., medium service unit, 
but it has a heavier duty electrical system and provides a 30-point, automatic, 
motor-driven, step-down switch for demagnetization of large, irregularly shaped 
parts or frames. The switch also provides much closer control of magnetizing 



Magnaflux Corporation 

Fig. 6. Combination a.-c. and half-wave magnetizing unit, 3000-amp. type with 
30-point demagnetizing current control. 

current magnitudes. This unit supplies high current through long leads and is 
commonly used where heavy weldments, castings, or large structures are in 
spected at distances of j 100 ft. or more from the unit. 

Three-Phase Rectifier, D.-C. Unit. The high-current d.-c. magnetizing unit 
supplies three-phase rectified current. It requires a 220- or 440-volt, three-phase 
supply. Maximum output is 5000 amp.; the continuous rating, 2000 amp. A 
particular advantage of this unit is a 30-point, motor-driven, tap switch 
which automatically reverses the polarity of the d.c. and steps it down in regular 
decrements. This provides effective demagnetization of parts of complex shape, 
such as large crankshafts. 

STATIONARY EQUIPMENT. Hand-operated, stationary, horizontal, 
wet-method equipment is widely used for small manufactured parts. This type 


probably accounts for about 75 percent of the magnetic-particle inspection units 
currently in use in the United States. 

Stationary units normally contain a built-in tank with a pump which agitates 
the wet particle bath and pumps inspection fluid through a hand-held hose for 
application to test objects. A part is clamped within the magnetizing coil 
between the copper contact faces of the head and tail stocks. At the operator's 
option, the parts can be magnetized circularly with current between the heads, 
or longitudinally with current through the coil; or both, if desired. The dual 
magnetization can be sequential on any unit or may be simultaneous on multiple 
magnetization equipment. While the part is magnetized, the operator applies 
the liquid inspection medium and then views the surface for indications. Most 
units are provided with inspection hoocl and black lights so that the fluorescent 
magnetic-particle inspection medium can be used. This increases the speed of 
inspection and reduces the possibility of missing an indication. The most com 
mon multiple-purpose, horizontal, wet-method stationary units are described 
in detail here. 

Horizontal, Wet-Method, A.-C. Units. The smallest horizontal wet-method 
units are. electrically identical. They supply 600-amp., a.-c., continuous mag 
netizing current or a maximum of 2700 to 3000 amp. of intermittent magnetizing 
current. The 54-in. unit accepts parts up to 54 in. long; the longest unit, parts 
up to 96 in. long. Current is controlled by a nine-point tap switch. The units 
can be operated from 220- or 440-volt, a.-c., 50- to 60-c.p.s. power. The bath- 
agitation system and tanks are built into the unit, with a hose to apply in- 

Magnaflux Corporation 

Fig. 7. Heavy-duty, a.-c., wet-method, magnetic-particle inspection unit for parts 
up to "96 in. long. Unit equipped with hood, black lights, and 30-point demagnetizing 




spection medium to the test part. The units are equipped with hoods and black 
lights for normal use with fluorescent particles in industrial manufacture or over 
haul inspection. Since this is an a.-c. unit, parts may be demagnetized with the 
magnetizing coil on the unit itself if required. 

Horizontal, Wet-Method, A.-C. Units with Automatic Demagnetizes 

Fig. 7 shows a typical a.-c. unit with 3000-amp. maximum magnetization current 
which accepts parts up to 96 in. long. A smaller unit accepts parts up to 54 in. 
long. A special advantage of this type of horizontal, wet-type, a.-c. equipment 
is the 30-point tap switch which provides both close current control for mag 
netization and automatic step-down demagnetization cycles for complex parts. 
This equipment can be operated from 220- or 440-volt, 50- to 60-c.p.s., a.-c. 
supplies and is sometimes built for 25-c.p.s. operation or for special voltage 
supplies. Magnetizing coils of various sizes are available. This equipment is 
commonly used with wet fluorescent magnetic-particle tests 

Horizontal, Wet-Type, D.-C. Units. Units supplying d.-c. magnetizing 
currents are available in horizontal, wet-type equipment and can be equipped 
with hoods. The shorter unit accepts parts up to 54 in. long and supplies 
d.c. (full-wave, rectified a.c.) up to 4500 amp. The longer equipment accepts 
parts up to 96 in. long and supplies maximum d.-c. magnetizing currents of 5500 
amp. These units are representative of many similar types in wide use for 
subsurface and surface inspection of aircraft parts. 

Horizontal, Wet-Type, D.-C. Units for Very Heavy Parts. Fig. 8 shows a 
large, horizontal, wet-type unit for heavy castings, forgings, and welded struc- 

Magnaflux Corporation 

Fig. 8. Horizontal, d.-c., wet-method, magnetic-particle inspection unit providing 
both longitudinal and circular magnetization, with 30-point demagnetizing switch. 


tures. These are set on the stainless grille during testing. Both longitudinal 
and circular magnetization are available. Maximum test-object capacities are: 
weight, 1200 Ib; length, 108 in.; and maximum diameter, 24 in. Direct-current 
magnetization is supplied intermittently up to 5000 or 5500 amp., or at 2000 
amp. continuously. A 30-point, automatic-reversing, d.-c. tap switch is provided 
for demagnetization of parts in place on the unit. 

WHEEL-MOUNTED, WET-METHOD UNITS. Analogous to the hori 
zontal units are movable stationary units (mounted on wheels) which can be 
used in one location and quickly rolled to another. Small parts can be run at 
production rates of several thousand per hour. An air compressor is built in. 
Most units operate on 220- or 440-volt a.c. When moved in the plant, they 
require only electrical connection to be operational. Electrically they are built 
to meet Joint Industry Conference design specifications. For high speed opera 
tion, a built-in, automatic, magnetizing cycle is provided with head clamping 
and magnetization actuated by a foot switch. 

Mechanized Inspection Equipment 

types of almost completely automatic magnetic-particle inspection units are used 
in hundreds of plant locations. Inspection processing is carried out automatically 
on parts carried by a continuous conveyor. Loading and unloading may be 
manual or automatic. The inspector is required to view the parts as they pass 
on the conveyor and must only see and react to readily visible indications. Parts 
bearing indications are diverted for later evaluation and salvage or rejection. 
Accepted parts remain on the conveyor and pass through an automatic demag 
netizing coil before being discharged from the unit. This equipment permits 
rapid, low-cost inspection, where slower inspection may not be worth its cost. 

Applications of such inspection include automobile and truck steering parts, 
connecting rods, many types of bolts, and similar mechanical parts. Similar 
automatic handling and operation is used on such critical parts as aircraft gas- 
turbine blades, where volume is very high and it is desirable to inspect at several 
stages of manufacture. Semi-automatic equipment may be special purpose or 

Semi-automatic processing paces the inspection rate much better than 
human operators. It also standardizes and controls the processing steps them 
selves. Thus, fully reproducible test results are achieved on every part tested, 
independent of human operator variations. 

matic equipment can be custom designed for inspection of one type of part. 

Gas-Turbine Compressor Blade Test. The three-shot, multiple-magnetiza 
tion, gas-turbine compressor blade inspection unit shown in Fig. 9 applies circular 
magnetization through the blade and through the dove tail, and also coil 
longitudinal magnetization of the blade, all during a single operating cycle. 
It is jigged for automatic blade handling and automatically applies a gentle dip- 
in bath. Magnetizing currents are applied during part removal and drain-off, 
for maximum sensitivity to detect very fine defects. This unit illustrates how 
special equipment is developed to provide a very special technique required for 
highest sensitivity, in moderate to high volume rates (600 parts per hour) with 
no damage to the delicate blades. 



Magnaflux Corporation 
Fig. 9. Three-shot, semi-automatic, magnetic-particle inspection unit. 

Conveyorized Compressor Blade Test. The three-shot, multiple-magnetiza 
tion, gas-turbine compressor blade conveyorized unit shown in Fig. 10 applies 
three magnetizing shots at one station, with automatic bath application as the 
blades pass through. Up to six inspectors can be used at maximum inspection 
rates of 2400 parts per hour. Thus an all-direction inspection and high sensitivity 
control are achieved through automatic operation, eliminating human process 



Magnafl ux Corporation 

Fig. 10. Conveyorized magnetic-particle inspection unit providing three magnet 
izing shots at one station. 

Automatic Inspection of Steering Spindles. Steering spindles are auto 
matically inspected on an indexing turntable unit. This unit was developed 
for testing a single type of spindle assembly at a rate of 650 units per hour. 
It provides two head shots from the end of the spindle through the two arms 
and a separate coil shot. Alternating-current magnetization is used with water- 
suspended, fluorescent magnetic-particle wet bath. It is fully automatic, with 
automatic demagnetization and automatic conveyor unloading. 

Aircraft Engine Cylinder Inspection. Aircraft cylinders are semi-automati- 
cally inspected with a d.-c. multiple magnetizing unit which processes the entire 
inside diameter and the exposed outside diameter of surfaces of the steel cylinder 
liner. This equipment uses central conductor and coil magnetization, with auto 
matic bath application. The aircraft cylinders enter by roller conveyor and are 
up-ended on a rotating table support. Upon actuation, all clamping, magnet 
izing, bath operations, and 30-point step-down demagnetization following the 
magnetizing cycle are completed automatically. 

There are many other special-purpose units designed and widely used in such 
applications as railroad axle inspection and jet compressor overhaul tests. 

of differing sizes and shapes may be handled on multi-purpose equipment. 

Automotive Forgings Test. Various automotive forgings are inspected on 
a versatile conveyorized unit which provides head shot, coil shot, or both, as 
required for individual parts, at rates up to 900 to 1200 parts per hour (Fig. 
11). The two magnetizations are separate. The head shot may be applied by 
hand, which then permits rapid circular magnetization of parts of various shapes 



Magnaflux Corporation 

Fig. 11. Conveyorized magnetic-particle inspection unit providing head shot, coil 

shot, or both, as required. 

or sizes up to 28 in. long. Bath application can be by hand or automatic for 
the circular shot. For the longitudinal magnetization the part is dropped on the 
mesh conveyor belt and automatically carried through coil magnetization and 
bath application. On rough-surfaced parts, such as forgings, one indication is 
applied on top of the other, with a single inspection further clown the conveyor 
under the black light in the hood. Then the conveyor carries the part through 
automatic demagnetization and ejects it into tote bins or onto a take-away 

Conversion of Hand-operated Units. Auxiliary conveyorized units permit 
conversion of existing standard hand-operated units to semi-automatic opera 
tion at less cost than a complete automatic or semi-automatic unit. Circular 
magnetization is applied on the standard unit where the part is magnetized and 
bath applied. The part is then placed on the auxiliary unit conveyor Ine 
auxiliary unit carries the part through automatic coil magnetization and bath 
application, if longitudinal magnetization is required, and then presents tne 
part to the inspectors. After this, the conveyor carries the parts through auto 
matic demagnetization and ejects them. 

Inspection of Automotive Steering Parts. A general purpose, longitudinal- 
shot unit was developed to handle large forged truck and bus-steering parts of 
many designs and shapes, weighing up to 40 Ib. each It handles the full produc 
tion of assorted designs at one plant. Longitudinal magnetization (with auto 
matic bath application) is the only direction of inspection applied, since only 



transverse defects are serious on these parts. Inspection is by fluorescent 
magnetic-particle bath, and automatic demagnetization is provided. 

MENT. In recent years equipment has been developed which uses photoelectric 
or television viewing of indications to replace the human inspector. This is 
applicable on an automatic basis to parts with simple shapes or to continuous 
simple areas, such as the longitudinal weld on seam-welded pipe. On large pipe, 
white-light scanning is used. For smaller tubes, equipment is designed to use 
fluorescent indications, with black-light scanning. 

Magnaflux Corporation 

Fig. 12. Photoelectric white-light scanner automatically locates and marks defec 
tive areas revealed by magnetic particles. 


Fig. 12 shows an automatic white-light scanner in operation at a steel mill 
which makes large-diameter welded line pipe, The unit automatically magnetizes 
the weld area as the pipe passes, applies a special black, dry powder, and blows 
off and reclaims excess powder. A white strip is applied to the weld area before 
the pipe enters the unit. Thus the indication is a black line on a white back 
ground, and as it passes under the illuminated photoelectric scanner, the crack 
indication is automatically viewed and detected. When a defect is detected, an 
automatic paint spray marks the section of the pipe for later repair. 

EQUIPMENT SELECTION. To meet an inspection need, magnetic- 
particle test equipment should be selected from the available models or the many 
variations by special design. Considerations governing this selection include: 

1. Minimum inspection cost, per piece tested. 

2. Inspection rate requirements. 

3. Flexibility in use. 

4. Sensitivity requirements. 

It is desirable to investigate carefully and select the best equipment to achieve 
the test requirements when a magnetic-particle test is introduced or test condi 
tions change. Thus maximum results may be obtained with minimum total cost. 

DEMAGNETIZERS. Often demagnetizes are required after a magnetic- 
particle test to remove fields resulting from the test itself or from other magnet 
izing operations, such as holding with magnetic chucks, For this need, various 
heavy-duty industrial demagnetizes and procedures have been developed. 
Demagnetizes have small- or large-diameter coils, as required by part size. They 
can be hand operated or automatic and can operate intermittently or con 
tinuously. On semi-automatic units the demagnetizer is commonly integral with 
the test unit and automatically demagnetizes parts as they pass by on the 

The best demagnetizer is selected on the basis of part size, complexity ot shape, 
hardness or magnetic retentivity of test objects, and degree of demagnetization 
required. In selecting magnetic-particle test equipment, this supplemental need 
may or may not exist but should be considered (see section on Magnetic-Particle 
Test Principles for further details) . 


BARTH, V. C. "C. & N.W. Intensifies Inspection of Equipment Parts/' Ry. Mech. 

Engr., 123 (19419): 556. 

"Black Lighting Trouble Spots," Ties (January, 1954). 
CLARKE, J E, R. A. PETERSON, and T. J. DUNSHEATH. "Automatic Inspection of 

Pine' " Weldina Engr., 36, No. 2 (1951) : 38. 
DOANE?F. G , and C. E. BETZ. Principles of Magnaflux. 3d ed. Chicago: Photopress, 

"Fu^Mechanized Automatic Handling Features Pipe Inspection Unit," Steel Equip 
ment and Maintenance News (December, 1956). 

as o j- >-* 

THOMAS,' W. E. "Economic Factors of Nondestructive Testing," Nondestructive 
V J M M H 4 ( Xab?e Magnetic Particle Equipment," Southern Power 

2nd. (October, 1955). 





Interpretation Guide 

Phases of inspection .......................... 1 

Need for training .......................... 1 

Required information ..................... 1 

Classification of indications ................... 1 

Surface discontinuities ...................... 1 

Dry magnetic- particle indications of a 
transverse quench crack and longitudinal 
seam in a railroad locomotive spring 
(/.I) .................................... 2 

Wet magnetic -particle indication of a 

seam in a diesel-engine wrist pin (/. 2) 2 
Subsurface discontinuities .................. 2 

Comparison of wet- and dry-method indi 
cations of subsurface defects at varying 
depths (/. 3) ............................ 3 

Checking indications .......................... 3 

Checking by residual method ............... 4 

Checking with magnifying glass ............ 4 

Indications of nonmetallic inclusions ......... 4 

Characteristics of inclusions in hot-worked 
materials ................................ 4 

Typical examples of inclusions ............. 6 

Dry magnetic- particle indications of non- 

metallic inclusions in a jack screw (/. 4) 4 
Nonplastic inclusions ....................... 6 

Magnetic-particle indication of nonplastic 
subsurface inclusion (/. 5) .............. 5 

Indications of surface seams .................. 8 

Magnetic-particle indication of a longitu 

dinal seam in a rolled steel bar (/. 6) 6 

Indications of cooling cracks .................. 7 

Magnetic-particle indication of a cooling 

crack of varying depth (see Fig. 8) (/. 7) 7 
Cross -sections showing crack depths at 

locations marked on bar of Fig. 7 (/. 8) 7 
Indications of laminations .................... 7 

Indications of forging laps .................... 7 

Forging lap in a propeller-shaft lock nut 


weld in a marine structural weldment 

(/. 12) 11 

Indications of heat-treat cracks 11 

Magnetic-particle indication of a quench 
crack under a bolt head (see Fig. 14) 

(/. 13) 12 

Cross-section, after deep etching, of 
quench crack in bolt head, similar to 

that shown in Fig. 13 (/. 14) 13 

Indications of grinding, plating, or etching 

cracks 11 

Wet-method magnetic -particle indications 
of grinding cracks in a crankshaft (/. 15) 14 

Indications of fatigue cracks 12 

Characteristics of fatigue cracks 13 

Magnetic-particle indication of a fatigue 
crack propagating from an oil hole on 
inside of a hollow crankshaft (/. 16) .... 15 

Development of fatigue cracks 14 

Progressive fracture showing typical 

"fatigue crescents" (/. 17) 16 

Nonrelevant indications 14 

Effect of overmagnetization 14 

Geometric effects 15 

Nonrelevant magnetic-particle indications 
of interior splines of hollow shaft (/. 18) 17 

Recognition of nonrelevant indications 15 

Magnetic writing 17 

Permanent records of magnetic-particle in 
dications 18 

Lacquer fixing of indications 18 

Tape transfer of indications 18 

Photography of indications 18 

Fluorescent indication photography 18 

Indications of forging bursts and flakes ..... 

Magnetic-particle indications of forging 
bursts in upset forging (/. 10) .......... 

Indications of gas porosity ................... 

Dry magnetic-particle indications of sub 
surface gas porosity (blow holes) in a 
cast steel flange (/. 11) .................. 

Indications of incomplete weld penetration ... 

Dry magnetic-particle indication (arrow) 

of incomplete penetration in a fillet 

Nondestructive Supplemental Tests 

Reasons for supplemental tests 19 

Direct surface inspection 19 

Checking indications by residual magnetiza- 

8 tion 19 

9 Surface inspection with magnifiers 19 

Cleaning surface 20 

9 Inspection glasses 20 

9 Binocular microscope 20 

Probing methods 21 

Filing 21 

10 File cut applied to a crack in bar stock 

10 (/. 19) 21 

Grinding 22 

Chipping 22 



CONTENTS (Continued) 


Removal of defects . 
Repair by welding 
Flame gouging 

Destructive Supplemental Tests 

Reasons for destructive tests 22 

Fracturing 23 

Fracturing by hammering 23 

Two rail-joint bars with fracture com 
pleted through fatigue crack (/. 20) 23 

Fracturing in mechanical testing machines . . 23 

Fracturing after blueing by heating 23 

Fracturing after chilling 24 

Sawing 24 

Trepanning or core drilling 24 

Sectioning with cut-off wheels 24 

Effects of overheating while cutting 24 

Examination of cut surface 24 

Spring sectioned on cut-off wheel (/. 21) . 25 


Belt sanding 25 

Etching 25 

Types of etching 25 

Deep etching or macro-etching 25 

Persulphate-etched section of a double- 

Vee butt weld (/. 22) 28 

Persulphate etching 26 

Etching of ingots or coarse costings 26 

Etch cracking 26 

Etching cracks in a twist drill (/. 23) .... 27 

Tempering before etching 26 

Etching technique and equipment 26 

Acid etching 27 

Deep-etched section of a forging, showing 

flow lines (/. 24) 28 

Metallographic examination 28 

Preparation of specimen 28 

Examination of nonetched specimens 29 

Examination of etched specimens 29 

Utility of tests 29 

Bibliography 29 



Interpretation Guide 

PHASES OF INSPECTION. Magnetic-particle inspection involves three 
phases. The first phase includes the mechanical processes necessary to create 
suitable accumulations of magnetic particles at discontinuities, i.e., the formation 
of indications. The second phase is the interpretation of the condition in the 
part which has caused the formation of the indication. The third is the evalua 
tion of the seriousness of the cause of the indication as it can affect the use of 
the part and the disposition of the part: whether it need be scrapped, can be 
reworked, or used as it is. 

Need for Training. The evaluation of indications and of the discontinuities 
which cause them, as well as determining the disposition of the part, should be 
done by individuals whose knowledge of metals and engineering principles enables 
them to make a sound decision. Decisions, or specifications enabling others to 
make such decisions, should be made by skilled and trained men. 

Required Information. In most cases it is possible for an experienced in 
spector to tell, from the location and appearance of a magnetic-particle indica 
tion, what condition in the part has caused the accumulation of magnetic 
particles. Such an appraisal can be made accurately only if the history of the 
part, its metallurgical composition, method of formation, and processing are 

CLASSIFICATION OF INDICATIONS. Magnetic-particle indications 
fall into three broad classes: 

1. Surface discontinuities. 

2. Subsurface discontinuities. 

3. Nonrelevant magnetic disturbances. 

Each class possesses characteristics which generally make its recognition 

Surface Discontinuities. Surface discontinuities produce sharp, distinct pat 
terns which are usually held tightly to the surface of the part. This is particu 
larly true of the sharp, close-lipped surface defects which are difficult to see 
and are most objectionable. 

Fig. 1 illustrates typical surface crack indications. This helical spring of a 
railroad locomotive contains a transverse quenching crack and a longitudinal 
seam. Both are clearly shown by sharp, clean-cut, and tightly adherent dry- 
powder indications. Fig. 2 shows a seam in a wrist pin inspected with the wet 
method. The closely held, compact particle pattern has some height and in some 
cases enough "build-up" to cast a shadow. 

32 1 



Magnaflux Corporation 

Fig. 1. Dry magnetic-particle indication of a transverse quench crack and longi 
tudinal seam in a railroad locomotive spring. 

Magnaflux Corporation 
Fig. 2. Wet magnetic-particle indication of a seam in a diesel-engine wrist pin. 

Subsurface Discontinuities. Fig. 3 clearly illustrates two important effects 
observed m magnetic-particle inspection. These indications are those of holes 
drilled at varying depths beneath the surface of a steel part, The indications 
near the left end of each strip show the artificial defect closest to the surface 
and each successive indication to the right is over a deeper defect This clearly 






Magnafhix Corporation 

Fig. 3. Comparison of wet- and dry-method indications of subsurface defects at 
varying depths. From left to right, subsurface holes were drilled at increasing depths, 
(a) Indications obtained with wet method, continuous magnetization: (1) 500 amp., 
(2) 1000 amp., (3) 2000 amp. (b) Indications obtained with dry magnetic powder, 
continuous magnetization: (1) 500 amp., (2) 1000 amp., (3) 2000 amp. 

shows that the indications become broader and fuzzier as the discontinuity caus 
ing them lies deeper beneath the surface. The results of this test also show that 
the dry powder is much more sensitive than wet particles for subsurface defects. 

CHECKING INDICATIONS. It is usually easy to determine whether the 
discontinuity is surface or subsurface. First, the appearance of the indication 
itself may be all that an experienced operator needs to make a. decision. Second, 



wiping away a part of the indication allows a close examination of the area. 
Often some surface irregularity appears, although its exact nature may not be 
clear. If there is no surface irregularity under the indication, it is very likely that 
its cause is completely beneath the surface. 

Checking by Residual Method. A reapplication of magnetic particles permits 
one to observe whether the indication readily reappears. This is actually an 
attempt to reproduce the indication by the residual method. The residual 
method is always less sensitive than the continuous method. It is obvious that 
if the original indication was produced by the continuous method, the manner 
in which the indications reappear will give some notion as to the severity and 
extent of the discontinuity. Sometimes the reproduction of an indication by the 
residual method is used as a criterion for rejection on certain types of parts. 

Checking with Magnifying Glass. A small hand glass with a magnification 
power of 2 to 7 diam. is a great aid for inspectors. With such a glass it is possible 
first to examine the indication in detail, and later, after wiping away the indica 
tion, to see material surface conditions. The majority of surface cracks and 
seams can be seen with a low-power glass, provided the exact area has been 
previously indicated by the line of magnetic particles. If no surface discontinuity 
is visible under the indication after thorough cleaning, the conclusion that the 
discontinuity is subsurface may be justified. 

discontinuity most commonly found with magnetic-particle inspection is the non- 
metallic inclusion. Even though it is found most frequently, it is often of little 
significance in determining the usefulness of the part. When such inclusions 
occur in areas of high stress or in certain special locations or directions, they 
may be cause for rejection. But it is incorrect to call them "defects" without 
reservation, for they occur in all steels and are not defects unless they reduce the 
strength of the part. For this reason it is important to be able to recognize and 
appraise the true importance of the indications of such nonmetallic inclusions 
(sometimes called stringers). Usually they are very fine and are revealed with 
magnetic-particle inspection only when they occur near the surface. They are 
more likely to be shown on highly finished surfaces inspected by the wet method 
and at high magnetization levels. 

Characteristics of Inclusions in Hot-worked Materials. The indications 
are usually straight and parallel to the rolling direction on rolled products. On 

Magnaflux Corporation 
Fig. 4. Dry magnetic-particle indications of nonmetallic inclusions in a jack screw. 


32 5 

'->';' ' 'i A - v '' >; <h ty "$ 


Magnaflux Corporation 

Fig. 5. Magnetic-particle indication of nonplastic subsurface inclusion, (a) Indi 
cation is not parallel to grain flow in forging, (b) Section through forging beneath 
indication, showing inclusion at 100 X magnification. 


forging, they parallel flow lines which may bo curved. Born use they show only 
when near the surface but are distributed throughout the cross-section, inspec 
tion before machining is not helpful, and if inclusions are considered cause for 
rejection, parts should he inspected after surface finishing. 

Typical Examples of Inclusions. Fig. 4 shows a jack screw machined from 
a rolled bar. The nonmctallic inclusions were not visible without magnetic- 
particle inspection. The straightness of the indications is important, as is the 
fact that different inclusions were brought near the surface in the complex 
machining of the part. 

Nonplastic Inclusions. A subsurface condition which is much more likely to 
be dangerous is the presence of a. nonmetallic inclusion which was not plastic at 
the temperature of the rolling or forging. This may be a fragment of hard 
refractory, and since it is not plastic, it remains large and may be unfavorably 
located near the surface. Fig. 5 (a) shows an indication of such a nonplastic 
inclusion. It is not a stringer because it is not parallel to the grain flow in the 
forging. It is subsurface because its indication is slightly fuzzy and because the 
surface of the material shows no discontinuity when the indication is removed. 
When this part was sectioned and microscopically inspected, as shown in 
Fig. 5(b), the inclusion was shown to be quite large and and at right angles to 
the surface. 

INDICATIONS OF SURFACE SEAMS. Surface seams in rolled bars 
result from cracks or other surface defects in the billets from which they are 
rolled or from some defect introduced into the rolling operation itself. The 
great elongation of the metal draws out such surface defects into long, straight 
seams, usually closely parallel to the direction of rolling. Fig. 6 shows an indica 
tion of such a seam in a rolled bar. Seams are recognizable by their straightness 
and because they are at the surface and are roughly parallel to the direction 
of rolling. 

Magnaflux Corporation 
Fig. 6. Magnetic-particle indication of a longitudinal seam in a rolled steel bar. 


32 7 

INDICATIONS OF COOLING CRACKS. Cooling cracks which occur 
in rolled bars are similar to seams but usually differ in appearance in "some 
respects. Fig. 7 shows an indication of a cooling crack. It is sharp, well defined 
and adherent, but it deviates somewhat from the rolling direction The depth 

Magnaflux Corporation 

Fig. 7. Magnetic-particle indication of a cooling crack of varying depth (see 

Fig. 8). 

of the crack varies, as indicated by the heaviness of the indication. Fig. 8 shows 
sections through the 2%-in. bar shown in Fig. 7. This illustrates how the appear 
ance of an indication may vary with the depth of the crack causing that indica 
tion. When all the variables of the method are controlled, the amount of the 
powder build-up can be used to judge the depth of the crack. 

Magnaflux Corporation 
Fig. 8. Cross-sections showing crack depths at locations marked on bar of Fig. 7. 

INDICATIONS OF LAMINATIONS. Laminations in plate or sheet are 
formed in a manner similar to seam formation in bars. Since laminations are 
parallel to the surface, they cannot be detected with magnetic-particle inspection 
except at an edge or at a hole cutting through the lamination. 

INDICATIONS OF FORGING LAPS. Forging laps are irregular in 
contour, are generally at right angles to the direction of metal flow, and often 

32 8 




Magnanux Corporation 

Fig. 9. Forging lap in a propeller-shaft lock nut. (a) Magnetic-particle indication 
shown by white arrow. ( b) Cross-section through forging lap of (a), magnified 100 X, 

no etch. 



occur at an acute angle to the surface. This last condition usually results in a 
fairly weak leakage field. For this reason indications of forging laps are often 
not well defined. Fig. 9 (a) shows a forging lap and Fig. 9(b) a section through 
the lap, magnified 100 X, which clearly shows the angle of the lap to the surface. 
Destructive tests must often be employed to confirm the inspectors' judgment 
or to determine the exact nature of a defect. Sectioning through the indication 
is extremely useful. Often deep-etching the surface of the cross-section brings 
out grain flow and gives important information as to the cause of the defect. 

metal or alloy at too low a temperature may produce bursts, i.e., ruptures which 
may be wholly internal or which may occur at the surface, as shown in Fig. 10. 

Magnaflux Corporation 
Fig. 10. Magnetic-particle indications of forging bursts in upset forging. 

When masses of metal are cooled too rapidly through a certain temperature 
range, they are subject to internal ruptures called "flakes." Dissolved gases 
coming out of solution in this temperature range contribute to this occurrence. 
Steels of certain composition are more susceptible to flaking than others. 

INDICATIONS OF GAS POROSITY. Porosity in castings, caused by 
gases trapped in the solidification of the molten metal, can sometimes be located 
with magnetic-particle inspection. Subsurface blow holes in a cast steel flange 
are shown in Fig. 11. These indications were brought out with d.-c. magnetizing 
and dry powder. Thermal cracks in castings caused by unequal shrinkage can 
be shown with magnetic-particle inspection. 



Magnafhix Corporation 

Fig. 11. Dry magnetic-particle indications of subsurface gas porosity (blow holes) 

in a cast steel flange. 

netic-particle inspection is used extensively on welds. It is possible to find 
porosity, slag inclusions, shrink cracks, inadequate penetration, and incom 
plete fusion. The indication pointed out by the arrow in Fig, 12 is broad and 
fuzzy, which is typical of a subsurface discontinuity. It is caused by lack of 
penetration at the root of the fillet weld. It may not represent a defect if 100 
percent penetration is not needed in this part. This indication was brought out 
by half- wave magnetizing current and dry powder. Such a fuzzy indication may 
mask a serious defect at the surface, such as a fine shrink crack. However, if 
a.-c. is substituted for d.-c. magnetization, only these surface conditions will be 
revealed. Equipment used for weld inspection should be capable of producing 



Magnaflux Corporation 

Fig". 12. Dry magnetic-particle indication (arrow) of incomplete penetration in a 
fillet weld in a marine structural weldment. 

both half-wave and a.-c. current, to enable the operator to detect either surface 
or subsurface discontinuities. 

INDICATIONS OF HEAT-TREAT CRACKS. Cracks caused by faulty 
heat-treat processing are readily found with magnetic-particle inspection. Such 
cracks may occur during either the heating or quenching cycle and may be an 
enlargement of a condition existing in the part from some previous operation. 
Heat-treat cracks which are created by the quench cycle (quench cracks) 
usually are to be found at sharp changes of section, which cause unequal cooling 
rates, or at fillets or notches which act as stress-concentration points. The indi 
cation under the head of the bolt in Fig. 13 is caused by a quench crack. A simi 
lar indication can be produced by a fold in the upsetting operation. Fig. 14 
shows a deep etch of a section through an indication similar to that shown in 
Fig. 13. Since the flow lines show no distortion in fact the crack cuts directly 
across the flow lines it is evident that the crack was introduced, after the 
upsetting, in the heat-treating operation. The destructive testing performed on 
this bolt demonstrates how laboratory techniques can increase the value of 
magnetic-particle inspection. 

CRACKS. One of the most prolific sources of trouble in highly finished preci 
sion parts is cracking caused by faulty grinding operations. Grinding cracks 
are often critical because they usually occur on surfaces which will be highly 
stressed. They are very fine and sharp and usually occur at right angles to the 
direction of grinding. Fig. 15(a) shows a typical example of grinding cracks 
which . usually are found in groups or .families, although occasionally a- single 



Magnaflux Corporation 

Fig. 13. Magnetic-particle indication of a quench crack under a bolt head (see 

Fig. 14). 

grinding crack may be found on a part. Fig. 15 (b) shows a section through a 
grinding crack at 100X. This crack had an actual depth of 0.021 in. 

Plating eracks or etching cracks are found only in areas where high residual 
stresses remain from some previous process, such as hardening. When such 
parts are plated, that operation may cause those stresses to crack the surface. 

INDICATIONS OF FATIGUE CRACKS. Fatigue cracks are produced 
in service under repeated stress reversals or stress variations. A crack almost 
invariably starts at a highly stressed surface and propagates through the section 
until failure results. A fatigue crack will start more readily where design or 
surface conditions provide a point of stress concentration. Sharp fillets, poor 
surface finish, seams, grinding cracks all act as stress raisers and assist in the 
start of fatigue cracking. 

All magnetic-particle inspection to eliminate seams, inclusions, cooling cracks, 
laps, bursts, porosity, heat cracks, and grinding cracks is for the purpose of 



Magnafhix Corporation 

Fig. 14. Cross-section, after deep etching, of quench crack in bolt head, similar to 
that shown in Fig. 13. Note the crack propagation across forging flow lines. 

preventing fatigue or service failure after the part goes into service. Consist 
ent and studied use of magnetic-particle inspection, as well as other nondestruc 
tive tests in a well-planned preventive-maintenance program, can in many 
cases reduce service failure from fatigue to practically zero. 

Characteristics of Fatigue Cracks. Fatigue cracks occur only in parts that 
have been in service. They are, with few exceptions, surface cracks and can 
therefore be found by using a.-c. current for magnetization. They are sharp, 
close-lipped, and occur in a direction transverse to the direction of local stress. 
They are seldom perfectly straight but usually "wander" a bit from a straight 
line. Fig. 16 shows a fatigue crack in a crankshaft. A close examination of this 



< '.M**\ 


Fig. 15. Wet-method magnetic-particle indications of grinding cracks in a crank 
shaft. (a) Crankshaft assembly magnified approximately I 1 /*. times, (b) Cross- 
section through typical grinding crack with a depth of approximately 0.021 in. (Mag 

nification 100 X, no etch.) 

crack showed that it started at the oil hole on the inside of the hollow .shaft. The 
burr from drilling the hole had not been smoothed away, leaving small, sharp 
irregularities which served as stress raisers. 

Development of Fatigue Cracks. Since a fatigue crack, once started, pro 
gresses through a cross-section, sometimes very slowly, it is sometimes referred 
to as a progressive fracture. The faces of such a crack are smooth and show 
typical "oyster shell" markings, or fatigue crescents. When complete fracture 
occurs, the last part of the section to break shows a rougher face. Fig. 17 shows 
the progression of a fatigue crack. 

NONRELEVANT INDICATIONS. One last group of indications is 
extremely important in any consideration of interpretation. These are nonrele- 
vant indications. The term "false indications" is not an accurate one for this 
group because they are caused in almost all cases by an accumulation of magnetic 
particles held to a particular area by a leakage field. With nonrelevant indica 
tions the leakage fields are caused by conditions which have no bearing on the 
suitability of the part for service and are therefore nonrelevant to the proper 
evaluation of the part. 

Effect of Overmagnetization. Perhaps the most commonly found nonrelevant 
indication is the particle adherence due to local poles, i.e., leakage fields at sharp 


Magnaflux Corporation 

Fig. 16. Magnetic-particle indication of a fatigue crack propagating from an oil 
hole on inside of a hollow crankshaft. The burr left from drilling served as a fatigue 


edges or at the ends of parts longitudinally magnetized. The presence of such 
patterns is often caused by overmagnetization. 

Geometric Effects. A constriction in the metal path through which the 
magnetic field must pass may cause the field to be forced out of the part, creating 
poles which will attract particles. In Fig. 18 the splines on the inside of the 
hollow shaft create a particle pattern or nonrelevant indication on the outside 
surface. The field is crowded out by the reduced metal path at the base of the 
splines. Similar crowding at the inside corners of sharp fillets or at the root of 
threads may cause the same type of nonrelevant indication. 

Recognition of Nonrelevant Indications. Nonrelevant indications can usually 
be recognized when : 

1. Given the same magnetizing technique, indications appear on all parts and in 
the same locations. 

2. Indications always can be related to some feature of construction or cross- 
section which accounts for such a leakage field by a constriction in the metal 

3. Indications seldom appear, to an experienced inspector, like indications of 
actual discontinuities. 



Magnaflux Corporation 
Fig. 17. Progressive fracture showing typical "fatigue crescents." 



Magnaflux Corporation 

Fig. 18. Nonrelevant magnetic-particle indications of interior splines of hollow 


Magnetic Writing. Another type of nonrelevant magnetization indication 
is caused by the creation of local poles set up by contact between a hardened 
part and another piece of magnetized material. A magnetized part may produce 
poles on an unmagnetized part. Poles may also be produced on a circularly 
magnetized part by contact with unmagnetized steel This type of nonrelevant 
indication is known as magnetic writing and can be recognized because the indi 
cation is poorly delineated and loosely held and usually runs in haphazard 

Nonrelevant indications are sometimes found at borders of local areas of cold 
working of a part or along the weld line in the parent metal, where the heat of 


welding has changed the permeability of the metal. Particle patterns will fre 
quently be found at the boundary of the change of permeability. 

TIONS. Magnetic-particle inspection is a rapid method in which indications 
of discontinuities are quickly formed on the part and remain there for interpre 
tation and evaluation. There is no need to study an oscilloscope screen or to 
develop and study a negative and then relate that evidence to the part or area 
in question. It is^ however, sometimes desirable to make a permanent record of 
magnetic-particle indications. There are four ways that permanent records can 
be made: 

1. Lacquer-fixing of indication on part. 

2. Transparent-tape fixing of indication on part. 

3. Transfer of indication onto record sheet. 

4. Photography. 

The simplest visible record is a sketch of the part showing the location and 
extent of the indications. On large parts it is sometimes sufficient to sketch only 
the critical area. 

Lacquer-fixing of Indications. If the part itself is to be preserved, the indi 
cation can be fixed permanently by using a transparent lacquer which should 
be thinned somewhat and applied by spraying or gently flowing over the indica 
tion. Before applying lacquer, the surface must be dry. If the indication was 
formed with the wet method, the vehicle used in making the bath must be 
allowed to dry before lacquering. Drying of the indication can be accelerated 
by heat or by gently flowing on a volatile solvent. 

Indications can 'also be fixed on a part by covering with pressure-sensitive 
transparent tape. While this is faster than using lacquer, it may be difficult to 
get a smooth fit if the surface is irregular. 

Tape Transfer of Indications. Transparent tape can also be used to lift 
indications from a part and transfer them to a permanent record sheet or card. 
Before using this method to transfer powder indications, it is essential that 
excess dry powder be blown off. Wet-method indications should not be trans 
ferred until the surface is dry. Tape transfers of indications are usually slightly 
broader than the original indications because of the flattening effect of tape 

Photography of Indications. Photographs of indications can also be taken 
for record purposes. Enough of the part should be shown to make it possible to 
recognize the part and the position of the indication. It is also helpful to include 
a ruler or some familiar object in the photograph, to show relative size. Care 
must be taken to avoid highlights and reflections on highly polished parts, 
since they may make the indication less outstanding. 

Fluorescent Indication Photography. Photography of fluorescent indications 
calls for special photographic techniques but, with practice, excellent results can 
be obtained. Fast panchromatic film is preferred. Pictures are taken in a 
darkened room with at least two 100- watt black lights positioned at a distance 
which gives evenly distributed illumination of the area being photographed. 
Sometimes better results are obtained by using white light during part of the 
exposure to record a faint outline of the part. Exposure time may vary from 
20 min. to 1 hr., and a better result is usually obtained by using either a G 
(orange) or K2 (yellow) filter. When photographing highly reflective surfaces, 


the use of polarizing filters may improve results but longer exposure times will 
be needed with such filters. Random fluorescence on the part will reduce the 
contrast of ^ the indication arid the background and should be removed before 
photographing. Excessively high magnetizing currents may create a generally 
fluorescent background which can be reduced by reprocessing the part, using 
lower current values. Fluorescent oil or finger" marks can be removed with 

Nondestructive Supplemental Tests 

REASONS FOR SUPPLEMENTAL TESTS. In specific, difficult prob 
lems ^ of interpretation, various confirming tests help in determining the nature 
of discontinuities which produce magnetic-particle indications. Knowledge of 
what to look for is indispensable to intelligent interpretation of indications. On 
the other hand, the ability to recognize magnetic-particle indications of disconti 
nuities, so as to identify them definitely, depends equally on actual experience 
with indications of typical discontinuities. To gain such experience, one must 
actually identify sufficient cases to be acquainted with the kinds of indications 
produced by various discontinuities. Therefore, various supplemental tests are 
needed. These tests often involve cutting up a typical specimen to see what the 
discontinuity really is and where it is located. 

Many confirming tests are simple and can be performed with limited labora 
tory or testing equipment, Others require specialized equipment and experience 
involving use of testing laboratories and methods of metallographical investiga 

DIRECT SURFACE INSPECTION. Often an indication cannot be satis 
factorily accounted for within the inspector's experience, and no decision can be 
made regarding such indications without supplemental tests. However, each test 
improves the inspector's knowledge and makes it easier to identify similar indi 
cations later. 

Some of the simplest tests are almost instinctively applied when an indication 
appears. For instance, when one observes magnetic-particle build-up, one al 
most automatically wipes it off the surface to see what may lie underneath. 
Sometimes some sort of surface irregularity appears immediately under the indi 
cation, though its exact nature may not be clear. 

Checking Indications by Residual Magnetization. Usually the powder or 
liquid bath is reapplied to determine if the indication readily reappears. This is 
first tried without demagnetizing or remagnetizing ; that is, one attempts to 
reproduce the indication by the residual method. Since the residual method is 
always less sensitive than the continuous method, the manner in which the resid 
ual indication reappears helps to indicate its severity and extent. As a matter of 
fact, this criterion of reproduction by residual magnetization is made the basis 
of acceptance or rejection on many aircraft engine parts under the recommended 
practice of the U.S. Navy, Bureau of Aeronautics. Often it is worthwhile to 
demagnetize the part and repeat the test from the start to confirm the indication. 

a small hand glass with magnifications from 2 to perhaps 7 diam. Examining the 
indication under low-power magnification often helps the inspector to form an 
opinion as to its cause. Such a magnified view may show that the powder 
build-up is dense and tightly packed, as from a sharp discontinuity lying at or 
just under the surface. Powder somewhat spread out and loosely packed would 


indicate a discontinuity farther below the surface; or it could mean that the 
indication is due to some nonrelevant condition. Again, wiping away part of the 
indication and looking just under it with a low-powered glass may reveal a 
surface condition responsible for the powder pattern. The majority of surface 
cracks and seams can be seen with a low-power glass after they have been located 
by magnetic-particle inspection. 

Cleaning Surface. If the surface being examined is scaled, rusted, or covered 
with a thin coat of paint, examination with a low-power hand glass is often 
facilitated by using fine emery paper to clean the surface where the indication 
appears. Such cleaning clears away surface material which may obstruct a view 
of the crack itself. Sometimes a new application of powder over the clean surface 
will be helpful. Sometimes an indication appears on a rough-machined surface 
where the roughness of the surface itself may obscure the crack even though the 
magnetic particles clearly indicate its presence. Here again, smoothing of the 
surface with emery paper (this time perhaps somewhat coarser so that some 
actual surface metal is removed) makes it possible to see the crack with a low- 
power glass. 

Inspection Glasses. Most inspection-department equipment includes large- 
diameter, low-power, magnifying glasses set on stands with pome sort of lamp to 
illuminate the specimen (see section on Visual Inspection Equipment). Such 
glasses are extremely useful, not only for examination of particular specimens 
but often in the course of routine inspections as well. The nature of the parts 
or the character of the defects sought may make such low-power magnification of 
the surface helpful in seeing the indication in the first place. These glasses 
usually have a magnification of about 4 diam. but provide a large viewing area 
so that a considerable portion of the surface can be seen at one time. 

Binocular Microscope. Often a surface crack may be extremely fine or closed 
at the surface, so that it cannot be seen with low-power magnifying glasses. 
An extremely useful piece of supplemental equipment for the inspector, there 
fore, is a binocular microscope which permits examination of surfaces at higher 
magnification (see section on Visual Inspection Equipment). Binocular micro 
scopes are easy to use and do not require a skilled technician. These microscopes 
come mounted on stands for bench use, but special mounting supports are also 
provided which permit the microscope to be swung out on an arm at any desired 
height so that large and irregularly shaped piece? can be examined. The principal 
requirement for such examinations, in addition to the microscope itself, is good 
light. Since the area which can be magnified and observed at one time is much 
smaller than that examined with lower-power lenses, better light on the surface 
is proportionately more important. 

Binocular microscopes of this type are fitted with various lenses to permit 
magnification ranges from 10 to about 32 diam. Such magnifications make 
nearly all surface discontinuities visible, although, again, some surface prepara 
tion may be necessary. Such surface preparation may involve cleaning with 
emery cloth or smoothing with a file or coarse emery. Sometimes light etching 
of the surface is useful. Of course, if none of these methods of surface examina 
tion reveals a surface discontinuity which will account for the magnetic-particle 
indication, and the indication persists after the surface is smooth and clean, th? 
presumption is that the discontinuity is below the surface. Interpretation of 
subsurface discontinuities may be confirmed by the appearance of the powder 
particles when examined under one or another of these magnifying devices. These 


step in the 

32 . 21 

PROBING METHODS. Often information beyond that revealed bv micro 
scopic surface examination of a discontinuity is required. Various ^ 

; ne ? f the sim P lest P robin S methods is the use of either a 
or flat file used cornerwise. Usually one selects a point where the ap 
of the powder pattern suggests that the discontinuity is most severe and 

Fig. 19. File cut applied to a crack in bar stock. Crack is shown with magnetic 


across the indication with the file. Such cutting, of course, will quickly show 
whether the indication is caused by a very shallow surface condition or whether 
it penetrates appreciably into the metal. The file cut is usually continued until 
the bottom of the crack or seam is reached or until the cut has gone to such a 
depth that it becomes apparent the defect is severe enough to warrant rejection 
of the piece. Checking the seam or crack at the file cut by reapplying magnetic 
powder or liquid indicates whether or not the cut has gone to the bottom of the 
defect, as shown by Doane and Betz, Principles of Magnaflux, in Fig, 19. 

The file method of inspection for seams in hot-rolled bars, particularly alloy 
and tool steel bars, is common practice in all mills. It is sometimes the only 
means of inspection and serves as an aid to the eye of the inspector where 
magnetic-particle inspection is not being used to locate seams. The file method 
of checking may, of course, be used on any surface. However, filing on finished 
parts may destroy usefulness for service. On some parts, such depth probing 



may determine whether the defect can be removed by finish machining or 

Grinding. A similar test for determining depth and extent of surface discon 
tinuities is obtained by grinding the surface at the indication. For this purpose 
a flexible-shaft grinding wheel is often used. Another very useful tool is the 
small hand grinder, provided with a number of specially shaped grinding heads 
which permit grinding down at the location of the crack with a minimum removal 
of metal beyond that strictly local to the crack. Such grinding methods permit 
exploring a crack over a greater length more readily than by filing. 

In the inspection of small forgings, for instance, it is common practice to 
investigate indications of forging laps with the grinding wheel. Shallow laps are 
usually ground out entirely, so that the objectionable condition is completely 
removed from the part. Caution must be exercised to determine that the area 
ground has sufficient tolerance to permit grinding. 

Chipping. An additional method for probing the depth of surface seams or 
cracks is by chipping. Chipping either by hand or by power hammer, using 
properly shaped chipping tools, is much faster than grinding and cuts more 
quickly' to the bottom of deep cracks. Chipping methods are most commonly 
used on welds, castings, and forgings, not only that the inspector may deter 
mine the actual depth of the defect but also in preparation for salvaging. 

In such chipping operations, the split chip test is also useful. This merely 
involves observation of the manner in which the chip comes away from the chisel 
as the cut proceeds either along or across the crack. If the crack extends com 
pletely through the chip being removed, the chip would, of course, separate into 
two parts. When the cut extends below the bottom of the crack, the chip will 
not separate. A magnetic-particle check, however, is a sure safeguard. 

REMOVAL OF DEFECTS. Grinding and chipping methods often serve 
the double purpose of determining the extent of the defect and at the same time 
removing the defect from the part. Forgings are commonly ground to remove 
entirely a discontinuity which does not penetrate below the minimum dimensional 
tolerance of the part. Frequent rechecking with magnetic particles is used to 
ensure complete removal of such defects. 

Repair by Welding, In case of large castings, forgings, and heavy welds, a 
crack may be chipped out and the cavity filled with weld metal if the operation 
is carried out with suitable precautions. For the purpose of such salvage, it is 
essential that the crack be entirely removed. A magnetic-particle check of the 
bottom of the chip groove is necessary to ensure that no trace of the crack 
remains before welding is begun. 

Flame Gouging. In the particular case of castings, removal of the defect 
before welding may be accomplished by flame gouging. By suitable manipulation 
of the cutting torch, metal may be very rapidly removed in a defective area. By 
watching the surface of the metal under the flame before actual cutting com 
mences, it is often possible to observe the presence of a surface crack. However, 
magnetic-particle checking before welding is again a safe precaution. 

Destructive Supplemental Tests 

REASONS FOR DESTRUCTIVE TESTS. To derive the maximum pos 
sible information concerning a questioned indication, numerous useful tests 
involve complete destruction of a part. Such tests, of course, cannot be regularly 



applied in the course of routine inspection, but they are invaluable to the in 
spector who is attempting to improve his own knowledge of the nature of defects 
occurring in the products for which he is responsible. Such tests also are fre 
quently used after a part has been rejected because of an indication, in order to 
confirm the judgment of the inspector making the rejection. This, of course, 
again leads to increased knowledge to aid further decisions. 

FRACTURING. One of the easiest destructive confirming tests is to break 
the part or attempt to break it through the discontinuity. Usually preliminary 
tests have shown that the indication is caused by a surface crack or discontinuity 
of some sort. 

Fracturing by Hammering. The part may be placed in a vise in such a 
manner that the metal on one side of the crack is held firmly. The rest of the 
piece projects from the vise so that when struck heavy blows with a hammer, it- 
will tend to break at the crack. Such procedures usually work best on hard 
parts, although they are. sometimes applied with good results on relatively soft 
materials (Fig. 20).' 

, CM 
Magnafhix Corporation 

Fig. 20. Two rail- joint bars with fracture completed through fatigue crack. 

Fracturing in Mechanical Testing Machines. Another way of completing 
the fracture, if the section is too large to handle by the hammer and vise 
method, is to put the part in a large testing machine and apply a load in such 
fashion that the part tends to break at the crack. Such methods are particularly 
effective in the case of fatigue cracks or in the case of heat-treat and grinding 

Fracturing After Blueing by Heating. Sometimes the value of a test can be 
increased by heating the part before fracturing to a sufficiently high temperature 
so that the"surface becomes blued. When fracture at the crack is completed, the 
original face of the crack itself will appear dark, while the fracture produced 
at the time of breaking will be bright. 


Fracturing After Chilling. To facilitate fracturing, particularly on small 
parts, the part may be chilled in dry ice and then fractured imme'diately. At 
temperatures far below F., most steels, even those which may be relatively soft 
at ordinary temperatures, become brittle enough to break easily. 

SAWING. Often it is desirable to examine a cross-section of the crack or 
discontinuity. Various methods are used for sectioning a part. If the material 
is soft, the easiest procedure is to saw across the crack to obtain a view of the 
depth and direction of the crack in a cross-section. This may be done with a 
hand or power hacksaw or with a slotting saw on a milling machine. 

TREPANNING OR CORE DRILLING. When large forgings, castings, or 
other bulky parts are being investigated, a complete section is difficult or costly 
to make. It is simpler to remove a piece of metal by various methods so as to 
include a cross-section of the crack for its entire depth. Slotting saws, chisels, 
and various ingenious methods are sometimes necessary to get such a piece out 
of the larger part. One rather easy technique is to use a core drill. This is a 
circular hollow cutter which drills a core about an inch in diameter. Cores may 
extend to any required depth. This method of cutting a section (often called 
"trepanning") is extensively used in the investigation of welds in heavy plate. 

are either too slow and difficult or cannot be used at all to make a cross-sectional 
cut on very hard materials because the saws are not so hard as the part itself. 
In this case, sectioning by means of the high-speed abrasive wheel or disc 
is easy. The disc revolves at very high speeds and is quite thin, often only Me in. 
thick. It is made up of a fine abrasive cutting material molded into shape with 
hard rubber or plastic which preferably should not be brittle. A brittle wheel 
might break if a deep cut were being made and the part not carefully fed. Such a 
cutting device should never be used unless it is fully guarded to minimize danger 
to the operator if breakage does occur, nor should the part ever be hand fed 
against the wheel. Many such cutting wheels are provided with a mechanical 

Effects of Overheating While Cutting. Great care must be taken to avoid 
overheating of the part during this cutting operation. Heating at the cut is very 
rapid, and even slight local temperature increases may cause the crack being 
investigated to extend beyond its original length or depth. Such heat may also 
change the structure of the metal, which may be a matter of importance. 
Frequently it is necessary to remove the specimen and cool it under water or 
adopt other means to keep the cut cool. Some wheels operate under water. This 
probably is the best technique, although some, which are also very good, apply 
a constant stream of water to the cut. 

EXAMINATION OF CUT SURFACE. Some cut cross-sections can be 
examined without further treatment, either directly with a glass or microscope 
or by the application of magnetic particles after suitable magnetization (Fig. 21). 
More often, however, it is better to smooth the surface of the cut in some manner. 
This can sometimes be done by using a flat file held crosswise of the work so 
that, by grasping the ends of the file with both hands, the back-and-forth motion 
will produce a perfectly flat and relatively smooth surface. This may be fol 
lowed by polishing with emery cloth which may be wrapped around the file and 
applied with similar motion. Alternatively, the emery can be laid on a flat surface 
and the cross-section laid against it and smoothed by a back-and-forth motion. 



Magnaflux Corporation 

Fig. 21. Spring sectioned on cut-off wheel. No surface preparation. Depth of crack 
on section is shown with magnetic particles. 

Belt Sanding. Draw filing and emery polishing are usually best done in a 
direction transverse to the crack. The belt sander can be used as a substitute for 
draw filing. This is especially useful for a hard part. In this case the emery or 
carborundum cloth moves in the form of an endless belt across a solid, flat 
backing surface. The face of the section of the part should be held firmly against 
the surface of the moving belt. This produces the same effect as the hand motion 
on the flat emery cloth but can be done more rapidly and with less effort Bv 
using successively finer grades of cloth, a surface which is flat and free of coarse 
scratches can be obtained. After such preparation, magnetic particles may be 
applied, or the depth and contour of the crack may be examined under the binoc 
ular microscope. 

Etching. Acids or other etching reagents can be used to obtain information 
with respect to discontinuities, either on the original surface of a part or on a cut 
cross-section. "Etching" is the term applied to the process of attacking the 
surface under examination by means of acids or other chemical reagents. Such 
etching materials eat out the surface but not uniformly. For instance the edges 
of a crack will be more rapidly attacked than the flat surface adjoining these 
edges. The result is that the surface opening of the crack is exaggerated, often 
making a very fine crack easily visible to the eye. Minute variations in compo 
sition of^the metal also affect the rate of attack of the etching material. Grain 
boundaries or flow lines indicating the grain flow in the metal in rolling or 
forging can be brought out in relief. 

Types of Etching. Etching can be conducted in various ways for various 
purposes, and the technique varies widely, depending on the information sought. 
The purpose of etching may be to bring out the coarse structure or surface 
characteristics of the metal. In this case it is referred to as macro-etching. 
Etching designed to bring out microscopic grain structure and characteristics 
of the metal is referred to as micro-etching. 

Deep Etching or Macro-etching. Macro-etching is most frequently applied 
as a severe acid attack (often called "deep etching") for the purpose of bringing 
out flow lines and exaggerating cracks and segregated areas. In this case, 


acid etching reagents are used. When the purpose of the etch is to show grain 
structure, a solution of ammonium persulfate is used instead of acids. Such an 
etch applied to a cross-section through a weld will reveal the boundary of the 
weld area and distinguish between base metal and weld metal, as shown by 
Doane and Betz, Principles of Magnaflux, in Fig. 22. 

mm^ ] v v " 

Fig. 22. Persulfate-etched section of a double- Vee butt weld. 

Persulfate Etching. Persulfate etching is carried out with a 10 to 20 percent 
solution of ammonium persulfate in water. This solution is used cold and 
swabbed on the surface to be etched with a cotton swab. Etching proceeds quite 
rapidly, and when complete to the desired point, the surface is washed, dried, 
and lacquered. 

Etching of Ingots or Coarse Castings. Persulfate etching on a cross-section 
of an ingot or a nonheat-treated casting will show the coarse casting structure 
resulting from relatively slow freezing of the molten metal. The use of etching 
methods to reveal the presence of cracks in surfaces is of long standing. Before 
the advent of magnetic-particle inspection, it was much more commonly used 
than at present as a method of locating surface cracks in suspected areas. 

Etch Cracking. If a hardened part is to be etched, one precaution must be 
consistently kept in mind. Heat-treated or ground articles which have not 
been thoroughly stress-relieved by tempering or drawing contain residual 
stresses to such an extent that if the surface fibers of the metal are attacked by 
etching reagents, these stresses may cause cracks to appear in the surface as the 
stresses relieve themselves. Such cracking (often spoken of as "etch cracking") 
appears on the etched surface (as shown, by Doane and Betz, Principles of 
Magnaflux, in Fig. 23) and is often indistinguishable from cracks which may 
have been present before etching. 

Tempering Before Etching. Failure to recognize this occurrence has often 
led to false conclusions regarding the presence of cracks in such articles. If, there 
fore, the surface of a highly hardened part is to be etched as a means of confirm 
ing the presence of cracks or of investigating cracks revealed by magnetic- 
particle test, the specimen should first be freed of residual stresses by heating 
to temperatures which probably need not exceed 700 F., cooling slowly, and 
then submitting to the etching reagent. 

Etching Technique and Equipment. Surfaces to be etched may or may not 
require preliminary preparation. They should, at least, be free of oil and grease. 



Fig. 23. Etching cracks in a twist drill. 

Scale or other corrosion products usually need not be removed prior to deep 
etching, since the acid itself dissolves or removes scale and rust and exposes the 
etched surface without additional help. Cut surfaces should be relatively smooth, 
and usually some sort of dressing, as described previously, is desirable. 

Acid Etching. Deep etching for grain flow (illustrated by Doane and Betz, 
Principles of Magnaflux, in Fig. 24) requires a less smooth surface than per- 
sulfate etching for structure. The usual deep etching reagent consists of equal 
parts of commercial hydrochloric acid and water and is used hot at a temperature 
of from 160 F. to 175 F. Sufficient attack on most steels can be obtained in 
15 to 45 min. with this technique. The same reagent can be used cold, but in 
that case several hours may be required to complete the etch. 


Another acid mixture that is considered satisfactory for such etching is a 
mixture of 38 percent by volume of commercial hydrochloric acid, 12 percent of 
concentrated sulfuric acid, and 50 percent of water. Caution: Concentrated 
sulfuric acid can cause severe burns and must be handled with great care. It is 
important to add the sulfuric acid to the cold mixture, rather than vice versa, 
to avoid the danger of spattering. Temperature and etching time are approxi 
mately the same "as for the hydrochloric-acid mixture. The etching should be 

Fig. 24. Deep-etched section of a forging, showing flow lines. 

carried out in glass or porcelain vessels. Since the fumes from the process are 
corrosive, care must be taken that such fumes do not damage metallic equipment 
in the laboratory. When etching is complete, the specimen is washed under hot 
running water and scrubbed with a stiff brush. It is usually a good plan then to 
wash the etched surface with dilute ammonia to neutralize any remaining acid, 
after which scrubbing in hot water is repeated. The part should then be dried 
rapidly and protected from rusting by the application of a thin, clear lacquer. 

METALLOGRAPHIC EXAMINATION. Often a great deal of informa 
tion as to the character and origin of a crack or other discontinuity which ^has 
produced a magnetic-particle indication can be obtained by examining sections 
at much higher magnification than provided by the usual binocular microscope. 
Satisfactory microscopic examination requires very special equipment and special 
skill and training on the part of the person who undertakes such investigations. 
Usually this is a job for a metallurgical laboratory, but the methods employed 
are of general interest. 

Preparation of Specimen. Specimens for examination under the microscope 
should be small and must, of necessity, be free of all surface scratches which 
could interfere with proper observation of the surface at high magnification. A 
section is usually cut across the crack or over a questionable area. Special care 
must be exercised, particularly if the cutting-off disc is used, so that local heat 


does not change the metal structure or the extent of the crack at the section. The 
cross-section is (1) brought to a flat surface on the belt sander or coarse polish 
ing wheel, (2) polished on succeeding grades of emery of increasing grain fineness, 
and finally (3) polished to a scratch-free surface on special polishing wheels 
employing fine polishing materials such as rouge, magnesium oxide, and alumina. 

Examination of Nonetched Specimens. Properly polished specimens are 
usually examined under a microscope at magnifications of 50 to 200 diam. 
Microscopes must be provided with special methods of illuminating this surface, 
since at high magnification the lens must be extremely close to the metal surface. 
The character and extent of cracks and other discontinuities, as well as large and 
small nonmetallic inclusions, can be studied by such examination. Most metal 
lurgical microscopes are provided with camera attachments so that typical 
portions of the surface can be photographed at high magnifications (see section 
on Visual Inspection Equipment). 

Examination of Etched Specimens. Much additional information can be 
obtained from specimens by etching their surfaces and examining under magnifi 
cations of 100 diam. or more. Such etching differs greatly from macro-etching. 

The most common etching fluid, especially for carbon steels, is a mixture of 
5 parts of nitric acid to 100 parts of alcohol. The reagent is called "nital." It 
reveals grain structure and carbon distribution in plain carbon steels and most 
steels of low alloy content. Numerous other etching reagents are used for various 
steels and for various purposes. Micro-etching requires considerable skill and 
experience and is often done in only a second or two. In examining a cracked 
specimen after micro-etching, information may be obtained as to (1) whether 
the crack has followed grain boundaries or cut across them, (2) whether the 
edges of the crack have been oxidized with a resultant loss of carbon in the layers 
of metal immediately adjoining the crack, (3) whether there is a decarburized 
surface layer of metal that may have affected the start of a fatigue crack, and 
much other information along similar lines. The interpretation of structure is the. 
field of the trained metallurgist 

UTILITY OF TESTS. The supplemental tests just described are for the 
most part simple and require equipment usually already available in most 
plants. All the equipment, with the exception of metallurgical microscopes and 
polishing facilities, is relatively inexpensive and can be readily obtained if not 
available. In the utilization of the results of nondestructive testing for defects, 
such tests are obviously of great value, since they confirm the inspector's judg 
ment in interpreting such results and build up his experience in diagnosing 

Often much more elaborate tests and experimental programs are well 
worthwhile for the purpose of determining definitely the reason for the occur 
rence of certain defects. Such tests can lead more positively to design, material, 
or process improvement. The avoidance of defects, rather than their mere 
location, should be the goal of all nondestructive testing of this type. 


ALLEN A. H. "Improved Tools Expedite Magnetic Particle and Penetrant Inspec 
tion," Metal Progr., 65, No. 1 (1954) : 161. 

BAXTER, L. E. "Magnetic Particle Inspection of Welds," Can. Welder, 46, No. 6 (1955). 
"Black Light Finds Flaws in Axles," Mod. Railroads, (January, 1949). 
CAJNE J. B. "What's Wrong With Castings," Foundry, 79, No. 6 (1951) : 106, 267. 


CATLIN, F. S. "Nondestructive Sample Testing for Cracks Aids Heat Treating," 

Metal Treating, (March-April, 1956). 
"Checking Equipment Parts Saves Time and Money at Dowcy Cement Plant," Pit 

and Quarry, (February, 1954). 
COVER, H. T. "How the PRR Uses Non-destructive Testing," A*//. Age, 135, No. 26 

(1953) : 48. 
DEHN, G. L. C. "Magnaflux for Quality Control of Pipe-Lino Welds," Welding J., 32 

(1953) : 721. 
."Magnetic Particle Inspection in the Oil Industry," Drilling Contractor, 

(April, 1950). 
DEVRIES, A. J. ''Productive Inspection Assures Quality and Reduces Scrap/ 1 Chicago 

Midwest Metalworker, (January, 1956). 
DOANE, F. G., and C. E. BETZ. Principles oj Magnaflux. 3d cd. Chicago: Photopress, 


"Double Inspection Assures 'Safe Service' Castings," Iron Age, 178, No. 1 (1956) ; 67. 
GEIST, C. M. "Magnetic Particle Inspection," Nondestructive Tenting, 10, No. 3 

(1951-1952) : 27. 
HARRER, J. R. "Greater Acceptance of Welding Through the Use of Inspection 

Methods," Welding ] 36 (1957) : 252. 

"Latest Improvements in Diesel Overhaul," Southern Power <( 1ml., (October, 1949). 
LEE, F. E. "Fleet Shop Detective," Fleet Owner (October, 1949). 
LINDGREN, A. R. "Complete Weld Check Setup Prevents Leaks on Indiana Line," 

Gas, 29 (1953) : 132, 134, 136, 138. 
LOVE, J. L. "The Value of Scientific Inspection," Proe. i?7tli Ann. Conv. Automotive 

Engine Rebuild era Assoc., (May 19-21, 1949). 
MAcPHBRSON, J. R., and A. H. PRUCKNICKI. "Detecting Flaws in Traction Gears," 

Ry. Locomotives and Cars, 125, No. 12 (1955) : 66. 

MARADUDIN, A. P. "Quality Control of the Fabrication of High Yield-Strength Pipe 
lines," Gas, 25, No. 8 (1949) : 77. 
MCPARLAN, J. L. "How Nondestructive Testing Aids Power Station Maintenance," 

Power Eng., 61, No. 3 (1957) : 57. 

OYE, L. J. "Nondestructive Testing of Structures," Welding J., 33 (1954) : 223. 
PATTERSON, N. C. "Armco Avoids Accidents with Lifting Inspection Program," 

Power Eng., 59. No. 11 (1955) : 80. 

PETERSON, R. S. "Engineering Life Savers," Speed Age, (April, 1950) : 18. 
PLATT, D. M. "How Magnetic Particle Inspection Aids Maintenance at Camaa," 

Paper Trade J., 138, No. 20 (1954) : 98. 
ROBINSON, A. "Magnetic Particle Inspection of Jet Engine Parts," Nondestructive 

Testing, 10, No. 3 (1951-1952) : 19. 
SIMKINS, M. K. "Parts Inspection Pays in Three Ways," Com. Car J., (October, 

SKEIE, K. "Increased Foundry Profits Through Non-destructive Testing," Trans. Am. 

Foundrymen's Soc., 63 (1955) : 730. 
SMITH, 0. G. "Magnetic Particle Technique Makes Billot Inspection Positive and 

Efficient," Iron Age, 175, No. 18 (1955) : 99. 
SWEET, J. W. "Weldment Inspection in Aircraft. Construction,' 1 Nondestructive 

Testing, 10, No. 2 (1951): 34. 
THOMAS, W. E. "Economic Factors of Nondestructive Testing," Nondestructive 

Testing, 11, No. 4 (1953): 9. 
TOTARO, S. C. "Nondestructive Tests Chock Forging Quality, Spot Equipment 

Troubles," Iron Age, 174, No. 7 (1954) : 117. 

"Tracking Down Cracks and Hidden Flaws," Bus Transportation, (August, 1950). 
UPSON, F. A. "Magnaflux Inspection of Woldod Stonigo Tanks," Welding J., 29 

(1950): 27. 
VIGLIONI, J. "Fatigue Strength of Bolts Reduced by Longitudinal Flaws," Prod. Eng., 

28, No. 3 (1957): 203. 
WENK, S. A., and H. M. BANTA. "A Now Nondestructive Tost for Corrosion Fatigue 

Cracks in Drill Pipe," Oil Oan J., 50, No. 2 (1951) : 90. 






Magnetic-Field Probes 

Types of probes 1 

Foerster probe arrangements 1 

Probe arrangements used for various prob 
lems (/. 1) 2 

Longitudinal field probe 1 

Field difference (gradient) probe 1 

Tangential gradient probe (tangential com 
ponent) 2 

Tangential gradient probe (normal com 
ponent) 2 

Astatic magnetometer probe 2 

Tape or gap probe 2 

Probe-coil operation 2 

A.-c. hysteresis loop 3 

Magnetization curve of a ferromagnetic 

material (/. 2) 3 

Hysteresis loop with d.-c. bias 3 

Circuit for measuring d.-c. magnetic fields .. 3 
Circuit arrangement for measuring d.-c. 

magnetic field (/. 3) 4 

Derivation of secondary- voltage wave 
forms for the field probe arrangement 

(/.4) 5 

Secondary voltage 4 

Second harmonic voltage 4 

Circuit for measuring field difference 8 

Derivation of secondary- voltage wave 

forms for the gradient (/. 5) 7 

Secondary voltage 7 

Accuracy of field gradient measurements .... 8 

Design of instruments 8 

Probe-Type Magnetic Field Meters 

Precision magnetic-field meter 8 

Circuit arrangement 9 

Block diagram of the electric circuit of the 

field meter (/. 6) 9 

Meter calibration 9 

Types of pick-up probes 9 

Measurements of uniform fields 10 

Measurements of nonuniform fields 10 

Cancellation of earth's field 10 

Field gradient measurements 10 


Tape-probe measurements 10 

Tape test technique 11 

Tape transport 11 

Recording instruments 11 

Hall-Generator Magnetic Field Meters 

Hall generators 11 

Hall generator element used for static 

magnetic-field measurements (/. 7) .... 12 

Materials for Hall -effect probes 11 

Measurement of tangential field strength .. 12 

Hall detector probes 12 

Arrangements of Hall probe measurement 

heads (/. 8) 13 

Use of Hall detector instruments 14 

Measurement of strong magnetic fields 14 

Arrangement of Hall generator with com 
pensation of the Hall voltage by a 
potentiometer arrangement, for sensitive 
comparison of strong magnetic fields 

(/. 9) 14 

Application In Magnetic-Particle 

Surface field strength measurements 14 

Surface flux density 15 

Demagnetization by free poles 15 

Distribution of magnetic field in a com 
plicated test part, showing the demag 
netizing action of free magnetic poles 

(/. 10) 15 

Intensity of demagnetization 15 

True field strength 16 

Calculation of demagnetization factor 16 

Determining true field strength 16 

The field recorder method 16 

Field distribution on magnetized camshaft . 17 

Distribution of the true field strength 

(Ht) on a camshaft at a coil field (H) 

of 190 oersteds (/. 11) 17 

Measuring impulse fields 17 

References 17 

Bibliography 18 




Magnetic-Field Probes 

TYPES OF PROBES. During recent years there has been a great increase 
in use of magnetic-field instrumentation. The following static magnetic-field 
measurement methods are now used in industry: 

1. The rotating coil. 1 

2. The oscillating coil. 2 

3. The second-harmonic magnetometer (Foerster probe). 3 

4. The impedance variation of bismuth wires in magnetic fields. 4 

5. The Hall generator. 5 

6. The deflection of an electron beam in vacuum. 

7. Nuclear resonance. 6 

A method using the Hall generator has recently led to several new measure 
ment methods and promising applications in nondestructive testing. 7 These 
developments are a result of the recent discovery of new substances with an 
unusually large Hall effect and a low temperature sensitivity. 8 In the field of 
nondestructive testing in European industry, the method utilizing the Foerster 
probe 9 is especially successful because of the small probe dimensions and its high 

FOERSTER PROBE ARRANGEMENTS. The heart of the Foerster 
device is a pair of specially constructed probes having cores of high-permeability 
material that saturates at a relatively low flux density. For some types of in 
strumentation utilizing the Foerster probe, the two cores with their associated 
coils are mounted in separate probes. In other cases the two cores are mounted 
in fixed positions with respect to each other in the same probe. Fig. 1 sum 
marizes probe-coil arrangements used for various problems. 

Longitudinal Field Probe. The arrangement shown in Fig. l(a) is used to 
measure the static magnetic-field component along the longitudinal axis of the 
probe coil. This arrangement finds applications in geophysics and ship de 
gaussing. The same arrangement, in miniaturized form, is used to determine 
true field strength on the surface of magnetically soft material and to evaluate 
effects of static screening. 

Field Difference (Gradient) Probe. The arrangement shown in Fig. l(b) 
measures field differences independently of average field strength. Designed in 
miniaturized forms, it measures the field gradient. Practical applications in 

1. Iron search instrument. 

2. Detection of ferromagnetic impurities. 

3. Astatic coercive-force measurements. 


33 2 


4. Wall-thickness meters for nonferrous materials. 

5. Local hardness measurements, 

Tangential Gradient Probe (Tangential Component). Fig. l(c) illustrates 
the probe-coil arrangement for measurement of the tangential component of the 
tangential field gradient. This arrangement serves for: 

1. Measurement of the tangential component in Ihe point-pole method. 

2. Local coercive-force measurement. 

3. Nondestructive determination of the magnetic anisotropy of sheets. 






Institut Dr. Foerster 

Fig. 1, Probe arrangements used for various problems, (a) General field strength 
measurement (geophysics, ship measurements, etc.). (b) Field difference measure 
ment (iron search instrument, astatic coercive-force measurement arrangement, pole 
searcher, wall thickness meter for nonferroius materials), (c) Tangential component 
of the tangential gradient (local coercive-force measurement, nondestructive deter 
mination of the magnetic anisotropy). (d) Normal component of the tangential 
gradient (crack testing, coercive-force measurement with double-point method), 
(e) Tangential component of the normal gradient (astatic magnetometer), (f) Gap 
probe for the measurement of the state of magnetization in small zones (sound-tape 


Tangential Gradient Probe (Normal Component). Fig. l(cl) presents the 
arrangement for measurement of the normal component of the tangential field 
gradient. This arrangement is used for : 

1. Crack testing. 

2. Local coercive-force measurement with the double-pole method. 

3. Wall-thickness measurements on ferromagnetic materials. 

Astatic Magnetometer Probe. The probe arrangement of Fig. He), for 
measurement of the tangential component of the normal gradient, is used in the 
astatic magnetometer for determination of sample magnetization. 

Tape or Gap Probe. Fig. l(f) illustrates the gap probe for measurement of 
the magnetization in small zones having an extent of a few microns. This method 
is used in industry for the measurement of magnetic tapes and magnetic 
recorder wires. 

PROBE-COIL OPERATION. Each probe core carries a primary and 
secondary winding. In some cases the primary coils arc connected in series 
opposition. At any instant the alternating-current (a.-c.) field due to the first 
primary coil acts in the opposite direction to that of the second coil. The two 


33 3 

secondary coils are connected in series so that the two secondary voltages add. 
Conversely, the primary coils may be connected in series adding and the second 
ary coils in series opposing. 

A.-C. Hysteresis Loop. Fig. 2 (a) shows the magnetization curve of a ferro 
magnetic material (solid line). The magnetization curve is assumed to be free 
from hysteresis, which means that the magnetic induction B is zero when the 
field strength H is zero. When an alternating current is applied to a coil con- 


Fig. 2. Magnetization curve of a ferromagnetic material. 

taining the ferromagnetic material, the magnetization field strength H varies 
symmetrically about the zero axis [Fig. 2(b)]. When a direct-current (d.-c.) 
magnetic field (for example, the earth's field) is applied to the core material 
in addition to the a.-c. field, the curve is displaced horizontally along the H axis 
to the position shown by the dotted line. 

Hysteresis Loop with D.-C. Bias. Let H d . c . be the simultaneous d.-c. field. 
When the a.-c. field passes through zero, the induction B has not yet reached 
zero because the field # d-c> is still applied to the coil. The field # d . c . corresponds 
to an induction B^ [see Fig. 2 (a)]. The entire field H, and consequently the 
induction B, equals zero only when the instantaneous value of the a.-c. field is 
equal and opposite to the d.-c. field (point b). That is, under the action of the 
d.-c. field, the (dotted) magnetization curve in Fig. 2 is displaced by an amount 
# d>c- in the horizontal direction from zero, opposite to the direction of the field. 

shows an arrangement of two cores, K and K 2 , each having a magnetization 
coil through which magnetizing current flows. The magnetization coils are wired 



so that at any instant, the magnetizing field strength at K-^ is equal in mag 
nitude to but opposite in direction from that at K 2 . Fig. 4 (a) shows the two 
magnetization loops plotted on the same graph. If a d.-c. field, # d-( ,, is applied 
so that it has the same direction in both strips, the magnetization 'curve of ^ 
is displaced from zero to &, and the magnetization curve of K> 2 is displaced in 


Institut Dr. Foerster 
Fig. 3. Circuit arrangement for measuring d.-c. magnetic fields. 

the opposite direction from zero to a. Fig. 4(b) shows the sum of the two 
displaced curves, B^ + B 2 . In the absence of a d.-c. field, the sum B^ + B,, is 
zero because of the opposite magnetization of the two loops. However, under the 
influence of a _ d.-c. field, a horizontal displacement of both curves results. The 
sum B 1 + Bo is then different from zero. 

Secondary Voltage. The secondary voltage e flec at the terminals of the two 
secondary coils in Fig. 3 is 




Fig. 4(d) shows the secondary voltage which appears under the influence of the 
d.-c. field. 

Second Harmonic Voltage. Fig. 4(d) indicates that the frequency of the 
secondary voltage contains a strong second harmonic component. When the 
sum B 1 + B 2 goes through half a period, the secondary voltage goes through its 
full period. The dependence of the secondary voltage on the d.-c. field to be 
measured can be analytically investigated. The time variation of the secondary 
voltage, Fig. 4(d), can be expanded as a Fourier series, 10 and closed expressions 
can be obtained for the fundamental wave and harmonics. 

The development of the afore-mentioned mathematical process is very involved, 
so only the final results are given here. The second harmonic component of the 




Fig. 4. Derivation of secondary-voltage wave forms for the field probe arrange 

secondary voltage can be filtered out of the spectrum. This component can be 
used to measure the d.-c. field. The maximum amplitude e 2 is 

4 . /. H, 

sin I 2 -7T- 

_ H. HK 
cos I 2 -A- -?=- 


where w* = the number of secondary turns on the probe coil (Fig. 3). 
/ = the magnetization current frequency. 
F = the cross-sectional area of the probe-coil core. 


H A = the initial relative permeability of the coil core (slope of the straight 
portion of the curve where it passes through zero induction). This value 
is dependent upon the ratio (l/d) or form permeability. 
H, = the saturation field strength (the intersection of the continuation of the 

straight portion of the loop with the saturation magnetization line). 
HK = the bending field strength which results from the approximation of the 

hysteresis curve by an extension of its straight-line portion to H. 
H ^ the maximum amplitude of the magnetization field strength, 
//d.c. the d.-c. field to be measured. 

For a given curve shape for the core (H s , H K , \a A ), as well as a specific mag 
netization field strength #_, at the location of the probe core, the term 

/ ff, H K \ 
cos ( 2 -77- ) 
4 . (^ H,\ \ H^ #/ _ L 

sin I 2 A I -r-. fr 7F~\~2 h* 

jc \ // / i_ /4 H, H K \ 2 

~ l U//_ ' Hj 


can be combined as a constant /u. Investigation has shown that the value of the 
expression /io is relatively insensitive to changes in the ratios of H K /H K and 


Therefore the voltage generated by the field probes is 


where the constant B contains the operational and constructional data of the 
probe coil as well as the magnetic characteristics of the core. The magnetization 
field strength //_ should be much greater than the highest d.-c. field strength, 
// d . c> , to be measured, so that 

sin ( ^- ) = sin 6 ^ 8 

can be replaced by the angle 0. The result is that the secondary voltage increases 
linearly with the d.-c. field strength to be measured; i.e., 

& = AH A . C . (6) 

The proportionality between the d.-c. field strength and the secondary voltage 
generated by the probe coil is an essential characteristic of the field-strength 
meter. Fig. 3 shows the arrangement of the two probe coils for field-strength 
measurement. Note the opposing directions of the primary windings (magnetiza 
tion windings) as well as the similar winding sense of the secondary windings. 

able characteristic of this field instrumentation method that the field-strength 
meter is converted into a difference-field meter simply by turning one probe 
coil into the opposite direction. The latter indicates only the difference in field 
strength at the locations of the two probe coils, independently of the d.-c. field 

For difference-field measurements, the primary winding of one of the probe 
coils shown in Fig. 3 is reversed, without changing the balance of the circuit. 
The magnetization direction Hi is now the same in both halves of the probe coil, 
while the secondary windings are connected in phase opposition. If a uniform 
d.-c. field is superimposed on the a.-c. field and is parallel to it at the location of 
the two probe coils, the horizontal curve-displacement depicted in Fig. 2 occurs 



in the same direction in both loops [Fig. 5(a)]. In this case B l B 2 equals 
zero under the influence of a uniform d.-c. field [Fig. 5(b)]. As soon as a field 
difference appears between the two probe coils, however, the displacement of 
the two curves will be of different magnitude [Fig. 5(c)l. A field difference, 

B r B f *6forH p *0 

Fig. S. Derivation of secondary-voltage wave forms for the gradient. 

2H n causes the curve K, to be displaced by an amount H D to the left The 
cross-over point of the two curves, measured from the zero reference of H^,, 
equals # d . c .. 


difference magnetization [Fig. 5(e)] results in a voltage in the secondary coil, as 
shown in Fig. 5(f). In the field-difference meter, the second harmonic com 
ponent is again filtered out of this voltage. The voltage > of the field-strength 
difference meter is 

ft*' cos (2 ? ) sin (2 ^) (7) 

H~ H ~ 

Accuracy of Field Gradient Measurements. Eq. (7) agrees with Eq. (3) 
except for one term. Instead of the d.-c. field strength H Ac , the difference field 
strength H D appears. Only the term cos (2# d . /$_) shows that the difference- 
field indication depends on the d.-c. field H Ac in which the field-difference meas 
urement takes place. However, Eq. (7) shows that the field-difference indication 
may be obtained independently of the d.-c. field if the a.-c. field excitation is 
chosen considerably larger than the d.-c. field strength # c i. c .- The expression 
cos(2F dc /#.J will then be very nearly unity. For example, if the a.-c. field 
amplitude reacting on the probe-coil cores is chosen ten times larger than the 
earth's field, the error which appears in a difference-field measurement because 
of the simultaneous action of the earth's field is below 0.5 percent. In case of a 
twenty-fold a.-c. field amplitude, the error caused by a d.-c. field decreases to 
0.1 percent. This means that difference-field (field gradient) measurements, 
which are carried out in the direction of the earth's field and perpendicular to it, 
differ by not more than 0.1 percent in the two cases. 

In case of a large a.-c. field amplitude, 

so that, for the voltage generated from the difference-field measurement arrange 
ment, a nearly linear relationship with the difference field strength exists; i.e., 

& = AH D (8) 

DESIGN OF INSTRUMENTS. The complete theory of the field and 
difference-field measurement methods 10 provides data for optimum probe-coil 
dimensioning at a given oscillator power and at a chosen oscillator frequency for 
various requirements. These requirements include (1) measurement of a maxi 
mum d.-c. field (or difference field) at a desired linearity, and (2) independence 
of the difference-field measurement from the environmental d.-c. field. 

If the probe pair is arranged as shown in Fig. l(a), the absolute value of the 
d.-c. field strength is indicated by means of an instrument deflection which is 
proportional to the field strength. The alternate arrangement of the probe pair 
shown in Fig. l(b) gives an indication of the field difference between the two 
probe halves, independent of the ambient d.-c. field. In each case only the field 
component or the field difference along the longitudinal axis of the probe is 

Probe-Type Magnetic-Field Meters 

PRECISION MAGNETIC-FIELD METER. One example of equipment 
making use of magnetic-field probes is the precision magnetic-field meter. Direct 
measurement can be made of weak magnetic fields (10~ 5 to 10 -1 oersted for 
one scale division, or 0.001 to 1.0 oersteds for full-scale deflection). Direct-current 
field-measuring instruments for field strengths up to 1000 oersteds have been de 
veloped which use this principle. Field strength is indicated on a large precision 
mirror-scale microammeter with an error of less than 1 percent. 



Circuit Arrangement. Fig. 6 shows a block diagram of the electric circuits 
of the field meter and the difference-field meter. A stabilized oscillator serves 
as an excitation current source for the probes. The second harmonic component 
of the secondary voltage in the probe is filtered out and amplified in a selective 
amplifier. Rectification of the a.-c. voltage output from the amplifier takes 
place in a phase detector. Contrary to standard detectors, north-south and south- 


o 1 

o 1 
o 1 

2ND ' 







Institut Dr. Foerster 
Fig. 6. Block diagram of the electric circuit of the field meter. 

north field polarities are indicated by opposite deflections of the meter from its 
zero center position. The control voltage necessary for the phase detector is 
obtained from the oscillator by means of a transformer followed by a frequency 
doubler. Through the use of phase detection, the operation is independent of 
small hysteresis variations in the core strips. In addition, exact linearity is 
obtained down to zero signal, while in a standard rectifier nonlinearity exists for 
small values of signal. 

Meter Calibration. Calibration of this equipment is checked by depressing a 
push button on the front panel. This introduces a controlled amount of d.-c. 
current into the secondary windings of the probes. This current produces a 
magnetic field which is measured by the instrument. If the meter needle does 
not deflect from zero to the full-scale mark, the calibration control should be 
adjusted until the proper value is obtained. This calibration procedure is ap 
plicable only when the field probes are being used. 

Types of Pick-up Probes. Four types of pick-up probes may be used with 
the instrument: 

1. The field probe is the principal pick-up device used. Its use is explained in 
detail later. The indicating meter of the instrument is direct-reading in milli- 
oersteds for this pick-up device only. 

2. The microprobe is the smallest of the pick-up devices used with the equip 
ment. It is intended to be used in locations where the larger size of the field 
probes prevents their use. The microprobe has one-tenth the sensitivity of the 
field probe. 

3. The gradient probe contains two tiny sensing elements arranged to indicate the 
differences in field strength between them. The instrument is not direct-reading, 
but readings are repetitive. A calibration curve can be obtained by placing the 
probe in fields of known gradient. 


4. The tape probe is intended for static measurements of the field on magnetic 
tape and wire. Tape or wire is placed in the channel of the probe assembly, in 
contact with the pick-up device. The gap in the pick-up device is 0.002 in. wide. 
The readings are relative, since the instrument is not calibrated in absolute 
units for use with this probe. 

Measurements of Uniform Fields. The operation of the meter with the 
field probes is quite simple. Suppose it is desired to measure the field at the 
center of a solenoid coil which is considerably larger than the probes. With 
the solenoid coil current turned off, both field probes are placed inside the 
solenoid at the measurement point. The field probes should be side by side, 
pointing in the same direction and parallel with the field, as shown in Fig. 1 (a) . 
If the field of the solenoid alone is desired, the effect of the earth's field must be 
eliminated. To do this, the solenoid is turned until the probes point approxi 
mately east or west, so that the field meter reads zero in the appropriate range. 
When current is applied to the solenoid, the field meter will indicate the absolute 
value of the field acting on its probes. 

Measurements of Nonuniform Fields. If one probe is turned ISO deg., as 
shown in Fig. l(b), the reaction of one probe is canceled by an equal and 
opposite reaction in the other probe. This probe arrangement results in a zero 
meter reading in uniform fields. In nonuniform fields the meter reads one-half 
the difference in field between the two probes. This suggests that this probe 
arrangement could be used to measure field gradients. 

Cancellation of Earth's Field. Two probes can be arranged to cancel out the 
earth's field. One probe is placed inside the solenoid producing the unknown 
field. The other is placed far enough from the solenoid so that the effect of the 
solenoid field is negligible. This second probe is oriented experimentally to zero 
the field meter, with the solenoid current turned off. The instrument calibration 
should be checked, as described previously, with the probes in position. "Upon 
turning on the solenoid current, the field meter will read one-half the magnitude 
of the unknown field. However, this method should not be used to compensate 
for unwanted fields stronger than five times the earth's field. Partial saturation 
of the probes may result, causing errors in the readings obtained. 

Field Gradient Measurements. The gradient probe contains two sensing 
elements, positioned so as to be equivalent to the arrangement shown in Fig. 
l(d). The readings obtained with this probe are proportional to one-half the 
difference in field (field gradient) between the two probes. The probe should 
be oriented so that the field is parallel with the probe axis for gradient indi 
cations. The readings obtained are relative unless the instrument is calibrated 
by pla'cing this probe in a field with a known gradient. In this case the mag 
nitude of the calibrating field should be about the same as the field to be ex 
plored, to avoid saturation effects. 

In some applications it is desirable to explore the surface of steel samples for 
leakage fields. Hard wear plates are provided on the end of the probe to 
facilitate sliding the probe along the surface of samples. 

Tape-Probe Measurements. The tape probe [sketched in Fig. l(f)] should be 
fastened to a wooden work table, by means of nonmagnetic screws in the 
screw holes provided, before using it for precise measurements. All tools or other 
articles made of steel, iron, or other magnetic material must be removed from 
the vicinity of the probe. The meter is adjusted to read zero by means of the 
knob on the probe assembly. This knob controls the position of a small per- 


manent magnet which balances out the effect of the earth's field on the probe. 
This adjustment should be made first at a low sensitivity setting of the range 
switch and repeated as the range switch is advanced to the desired sensitivity. 

Tape Test Technique. The tape or wire sample is placed in contact with 
the pole pieces of the pick-up in the channel of the probe [see Fig. Iff)]. The 
gap is 0.002 in. wide. The highest frequency that can be measured on the tape 
is a function of the speed at which the signal was recorded on the tape. Read 
ings obtained are directly proportional to the field strength of the magnetic 
recording on the tape, but the field meter is not calibrated directly in oersteds 
for this probe. With the aid of the probe, the linearity of modulation of the 
tape can be investigated for any signal strength ranging from the very low to the 

Tape Transport. It may be convenient to use an automatic tape-transport 
device and a recording oscillograph in this application. A tape speed of to 2.5 
in. per minute is recommended. 

Recording Instruments. Two jacks are provided on the back of the field 
meter for connecting a recorder. Both these terminals are at a potential of 195 
volts with respect to chassis. Accidental shorting of one of these jacks to chassis 
may damage the front panel meter. Standard banana plugs fit the instrument 
jacks. A switch connected with one jack disconnects the front panel meter while 
the recorder is connected. 

Two types of recorders are suitable. Recording galvanometers (without 
electronic amplifiers) having a zero-center scale, a d.-c. resistance of about 1500 
ohms, and a current sensitivity of 50-^ia full scale are suitable. No connection 
shou'd be made to the chassis when using a recording instrument of this type. 

Recorders using electronic amplifiers with a push-pull, high-impedance input 
are also suitable. The shunt resistance from each jack terminal to ground should 
be about 1 megohm or higher. The recorder should have a sensitivity of about 
75-mv. full scale. 

The speed of response of the field meter is much faster (about 1 msec.) when 
used with a recorder than when used with the front panel meter. 

Hall-Generator Magnetic-Field Meters 

HALL GENERATORS. In 1879 Hall discovered that a flat conductor, 
carrying a longitudinal component of electric current 7 and penetrated perpen 
dicularly to its largest surface by a magnetic field H, exhibits a potential differ 
ence U H at its narrow sides (A, A'). (See Fig. 7.) The following equation holds 
true for the Hall voltage, U H : 

where RH = a material constant, known as the Hall coefficient. 
/ = the current through the Hall conductor. 
H = the magnetic d.-c. field strength perpendicular to the largest surface. 

t = the thickness of the Hall plate, 

Materials for Hall-Effect Probes. A material is especially suited for the 
measurement of d.-c. magnetic fields if it has 

1. A large Hall coefficient. 

2. Good electrical conductivity. 

3. A low dependence of the Hall coefficient and the electrical conductivity upon 




k ' 

t > 




Fig. 7. Hall generator element used for static magnetic-field measurements. 

Semi-conductor combinations such as indium-antimony (InSb) and indium- 
arsenic (InAs) discovered recently 5 fulfill these requirements especially well. 
In miniaturized form these Hall generators permit the point measurement of 
magnetic d.-c. fields in fractions of 1 oersted up to the highest producible field 

Measurement of Tangential Field Strength. The Hall generator method is 
especially well suited for the determination of the tangential field strength, that 
is, the true field strength during magnetization of a test part, because of the 
small dimension of the measurement probe. For complicated test parts such 
as crankshafts, measurement of the tangential field strength with the Hall micro- 
generator determines the density B of the lines of force in a test part. 

In magnetic-particle testing, this field strength is proportional to the mag 
netic crack-leakage flux. An induction of 11,000 gausses reportedly corresponds 
to a true field strength of 157 oersteds (i.e., 125 ampere-turns per centimeter) for 
most steels. To determine the tangential surface field strength (true field 
strength), a Hall microprobe is placed on the surface of the test object during 
magnetization. Thus one can directly determine whether the field strength is 
adequate for magnetic-particle inspection. 

As far as convenience of measurement is concerned, the Hall-generator method 
surpasses the field recorder method (see The Field Recorder Method in this sec 
tion). Furthermore it is insensitive to the opposing field ( N /) which appears 
after magnetization and which can alter the indications of the field recorder 
method in case of hard magnetic materials (see True Field Strength in this sec 
tion) . 

Hall Detector Probes. A hand-held measurement head has been designed 
whose Hall generator can be attached to the handle in each of three directions 
for various d.-c. field measurement problems. Fig. 8 (a) illustrates the position 
ing of the head for measuring the surface field strength of a circularly mag 
netized part. Fig. 8(b) shows the procedure for measuring true field strength 
in a longitudinally magnetized test object. Positioning to measure the true 
circular field strength on the inside surface of a tube is shown in Fig. 8(c). 













Institut Dr. Poerster 

Fig. 8. Arrangements of Hall probe measurement heads, (a) Arrangement for 
measurement of surface field strength in a circular field produced by a longitudinal 
current, (b) Arrangement for determination of true longitudinal field strength within 
a coil, (c) Arrangement for determination of true circular field strength inside of tube. 



USE OF HALL DETECTOR INSTRUMENTS. Various types of instru 
mentation are available for d.-c. magnetic-field measurement with Hall genera 
tors. One small instrument is easy to carry and operates from self-contained 
batteries. Its most sensitive measurement range is 200 oersteds. Its least sensi 
tive measurement range is 10,000 oersteds for full-scale deflection. A larger 
instrument has a maximum sensitivity of 0.5 oersted for 100 scale divisions, which 
can be reduced, in ten sensitivity steps, to 20,000 oersteds for 100 scale divisions. 
It has a cathode-ray tube which serves for the simultaneous measurement of a.-c. 
fields superimposed on the d,-c. fields and which indicates the form and magnitude 
of magnetization impulses. 

Measurement of Strong Magnetic Fields. The Hall generator method is 
especially well suited for the measurement of high magnetic-field strengths of 
several thousand oersteds. In a potentiometer arrangement (Fig. 9) the Hall 



Institut Dr. Foerster 

Fig. 9. Arrangement of Hall generator with compensation of the Hall voltage by 
a potentiometer arrangement, for sensitive comparison of strong magnetic fields. 

voltage U H which appears at a specific d.-c. field is suppressed to zero by 
displacement of the potentiometer slide. If the sensitivity of the instrument is 
then increased by a factor of 10 or 100. deviations from the original d.-c. field for 
which the potentiometer slide was calibrated can be measured with a ten- to 
hundred-fold increase in sensitivity. This method can be used to compare strong 
magnetic fields at different locations. 

Application in Magnetic-Particle Testing 

particle tests, the test objects are often magnetized with a direct current or a 
static magnetic field. To develop crack indications, the density of the magnetic 
lines of force (induction B) in the surface of the test object must exceed a 
minimum value so that ferromagnetic powder particles will gather at the crack. 
The magnetic-particle indications are a result of the crack-leakage flux. This 
flux is proportional to the flux density, B, in the surface of the specimen. 


tic flux densit y B " the surface of a test 

t true an ena e ^engt 

the surface of the test part. The permeability ^ is known approximately for 

he field s rnZ ^? " ^* *"**** is necessary omy to deterrmne 
tne Held strength # # to obtain the flux density 5. 

Demonetization by Free Poles. It is extremely difficult to determine the 

f r , com P licated P^t S such as crankshafts. The 
( ' f the teSt bject Can M V by a factor of 10 

f n + i, - trength ff -- The coil field stren S th can be calcu- 

ted from the magnetizing current and the coil factor of the magnetizing coil 
This condition is of special significance in magnetic-particle tests 

Fig. 10. Distribution of magnetic field in a complicated test part, showing the 
demagnetizing action of free magnetic poles. 

For example, Fig. 10 shows a specimen with two thick sections (C-D). 
It is assumed to be magnetized within a coil of field strength H a . At ends A and 
B and in areas C and D of the two thick sections, free magnetic poles appear 
as a result of magnetization. It is well known that magnetic lines of force exist 
between two opposite magnetic poles; for example, A and B or C and D. These 
lines of force pass from one pole to the other, both through the air and through 
the test part. Lines of force emerging from the induced magnetic poles oppose 
and weaken the externally applied magnetic field # fl , as indicated in Fig. 10. 

Intensity of Demagnetization. Lines of force resulting from free magnetic 
poles and reacting on the original magnetic field H u , become denser 

1. As the magnetization I of the test part increases. 

2. As the distance between the free poles decreases. 


As is indicated in Fig. 10, the lines of force emerging from the magnetic poles 
A and B weaken the original field H a less than those emerging from the more 
closely spaced poles C and D. The longer air path, AB, has a greater reluctance 
to the returning lines of force than the shorter path, CD. 

True Field Strength. The true field strength H t is the difference between 
the coil field strength H a and the reverse field strength H p resulting from the 
free poles; i.e., 

The field strength resulting from the free poles is calculated to be 

ff P = NI (12) 

FromEqs. (11) and (12), 

H t = H n -NI (13) 

Here, / is the magnetization of the test part. The demagnetizing pole strength 
is proportional to the magnetization /. N is the so-callod demagnetizing factor. 
N becomes larger as the test part becomes shorter, i.e., as the distance AB 
becomes smaller. For a complicated test part, demagnetization increases as poles 
C and D get closer together. 

Calculation of Demagnetization Factor. The demagnetizing factor N can 
be calculated exactly only for the rotation ellipsoid. 11 Empirical values are 
known for cylindrical bodies. 11 When the demagnetizing factor N is known, even 
approximately, the true field on the surface of the test part can be calculated 
from Eq. (13). If the permeability |i is known, the density B of the lines of 
force can be calculated from Eq. (10). This flux density B must reach a specific 
minimum value if cracks are to be indicated with the magnetic-particle method. 
McClurg has discussed these relationships thoroughly. 12 

Determining True Field Strength. Calculation of the true field strength H t 
is impossible for complicated bodies like that of Fig. 10 because the demag 
netizing factor N varies from point to point. However, the true tangential field 
strength H t on the surface and directly below the surface is constant on either 
side of the boundary between the material and air. Thus, if it is possible to 
measure the true tangential field strength in the air close to the surface of the 
test part, then the true field strength H t on and immediately below the surface 
can be assumed to be identical. With this field strength and the approximate 
permeability ^, the flux density can be obtained from Eq. (10) . 

The field strength in air due to the leakage flux of a crack is proportional to 
the magnetic field strength in the crack. The field strength in the crack is also 
identical with the flux density B in the test part, With complicated shapes of 
specimens or closely spaced demagnetizing poles, the field strength must be 
measured closer to the surface to obtain the true field strength H t on the surface. 
The highest surface field strength H t exists at the location of the smallest 
specimen diameter (for example, at EF in Fig. 10). At these points the direction 
of the lines of force from the poles C and D have the same direction as the field 
strength H a (see arrow direction). 

THE FIELD RECORDER METHOD. By means of the "field recorder 
method" it is possible to measure the field strength closer than 0.0004 in. to the 
surface of the test part. This is done by placing a 0.0004 in. thick, % X %-in. 
piece of special magnetic alloy with characteristics similar to "magnetic record- 



ing tape" on the surface of the test object. After magnetization this recorder 
strip retains a residual magnetization proportional to the highest field attained 
during magnetization. 

Field Distribution on Magnetized Camshaft. 'The residual magnetization of 
the strip is measured with a field-measurement instrument and microprobe to 
determine H t . Fig. 11 illustrates the field strength distribution obtained with this 
method on a complicated camshaft. The magnetlzing-coil field strength H a was 
190 oersteds. The field strength distribution H t of the camshaft is noted at 
several points directly on the surface and is indicative of the field in the surface. 
The fact that the true field strength for complicated test parts varies to such 
an extent from point to point can be determined readily with the "field recorder 
method" or by means of the Hall-generator probe instrument. 

43 Oe 101 Oe <100e 145 Oe <100e 170 Oe 

-190 Oe 

Institut Dr. Foerster 

Fig. 11. Distribution of the true field strength (Ht} on a camshaft at a coil field 

(H a ) of 190 oersteds. 

Measuring Impulse Fields. In the "field recorder method," the period of 
the field reaction on the recorder tape can be less than 10~ 3 sec. Thus the 
maximum field strengths obtained with impulse magnetization can be deter 
mined readily. For example, the field recorder method has been used to measure 
lightning currents, since the magnetic field of the peak lightning current leaves 
a residual magnetization in recorder tapes at various locations. In the United 
States these magnetic recording elements are known as surge crest ammeters. 


1. WILLS, M. S. "A Rotating Coil Fluxmeter," J. Sci. Instr., 29 (1952) : 374. 

2. KIESSKALT, S. "Aus der Praxis der zerstoerungsfreien Schweissnahtpruefung 

(Nondestructive Testing of Welded Seams)," Autogene Metdlbearbeitung, 27 
(1934): 65. 

3. FOERSTER, F. "A Method for the Measurement of B.C. Fields and D.C. Field 

Differences and Its Application to Nondestructive Testing," Nondestructive 
Testing, 13, No. 5 (1955) : 31. 


4. DEWAR, J., and J. A. FLEMING. "On the Electrical Resistivity of Bismuth at the' 

Temperature of Liquid Air," Pmc. Roy. Sac. London, 60 (1896-1897) : 72. 

5. BARLOW, H. E. M. "Hall Effect in Semi-Conductors," Elec. Rev., 155 (1954) : 848. 
,6. BLOCK, F. "Nuclear Induction," Phys. Rev., 70 (1946): 460; BLOCK, F., W. W. 

HANSEN, and M. PACKARD. "The Nuclear Induction Experiment/' Phys. Pev 
70 (1946): 474. 

7. FOERSTER, F. ''The Application of Hall Generators for Nondestructive Testing," 

Z. Metallk. (In press.) 

8. WELKER, H. "Neue Werkstoffe mit grossem Hall-Effekt und grosser Widorstands- 

aenderung im Magnetfeld (New Materials with Great Hall Effect and Great 
Change of Resistance in the Magnetic Field)," Elektrotcch. Z., 76 (1955) : 513. 

9. FOERSTER, F. "Bin Verfahren zur Messung von magnet ischen Gleichfeldern und 

Gleichfelddifferenzen und seine Anwendung in der Metallforschung und Technik 
(A Method for the Measurement of Steady Magnetic Fields and Differences of 
Steady Fields and Its Application to Metallurgical Research and Industry)," 
Z. Metallk., 46 (1955): 358. 

10. Zerstoerungfreie Werkstofiprucfung mit elcktmc.hen und magnctischen Verfahven 

(N widest motive Material Testing with Electrical and Magnetic Methods), 
Berlin, Goettingen, Heidelberg: Springer-Verhig. (In press.) 

11. BOZORTH, R. M. Ferromagnetism. 2d ed. New York: D. Van Nostrum! Co. Inc 

1953. Pp. 846, 847, 849. 

12. MCCLURG, G. 0. "Theory and Application of Coil Magnetization," Nondestruc 

tive Testing, 13, No. 1 (1955) : 23. 


BOROWIK, A. "Ferromagnetic Metals. Identification and Measurement of Internal 

Stresses, Part 1. Magnetic Tests," Iron and Steel, 21 (1948) : 3, 39. 
BOZORTH, R. M., and J. M. WILLIAMS. "Effect of Smiill Stresses on Magnetic Prop 

erties," Rev. Mod. Phys., 17 (1945) : 72. 
BROWN, R. J., and J. H. BRIDLE, "A New Method of Sorting Steels," Engineer 176 

(1943): 442. 
DIAMOND, M. J. "Hardness Tester Sorts Auto Engine Parts," Electronics, 27, No 12 

(1954): 160. 
FOERSTER, F. "Nouveaux precedes d'essai Slectronique nondestructif des materiaux 

(New Procedures for the Electronic Nondestructive Testing of Materials)," 

Metaux (Corrosion-Inds.), 26 (1951) : 497. 
- . "Theoretische und experimented Gnmdlagen der eloktromagnel.ischon 

Qualitaetssortierung von Stahlteilen, IV. Das Restfeldverfahren (Theoretical and 

Experimental Foundation of the Electromagnetic Quality Sorting of Steel Parts. 

IV. The Residual Field Method)," Z. Metallk., 45 (1954) : 233. 
FOERSTER, F v and G. ZIZELMANN. "Die schnelle zerstoerungsfreie Bestimnumg der 

Blechanisotropie mit dem Restpunktpolverfahrcn (The Rapid Nondestructive De 

termination of Sheet Anisotropy with the Residual Point Pole Method)," Z 

Metallk., 45 (1954) : 245. 

FOERSTER; F., and K STAMBKE. "Magnetische Untersuchungen innerer Spanmmgen, I. 
Eigenspannungen beim Recken von Nickeldraht (Magnetic Investigations of 
Internal Stresses, I. The Internal Stress of Nickel Wire During Stretching) " 

HAINZ, R. "Beispiel einer Einfuehrung der elektronmgnetischcn Sortierverfahren in 

die Fertigung von Dieseleinspritzpumpen und duescn (Example of the Use of 

Magnetic Sorting in the Production of Components for Diesel Injection Pumps 

and Nozzles)," Z. Metallk., 45 (1954) : 238. 
KEHOE, W. K. "Magnetic Testing to Control Quality of Ferrous Alloy Parts" Gen 

Elcc. Rev., 47, No. 4 (1944) : 59. 
MIEKIEJOHN, W. H. "A Non-Destructive Magnetic Hardness Tester," Ekctromc 

Inds. and Instrumentation, 1, No. 10 (1947) : 14, 45. 


ORTHBIL, J. "Kritische Betrachtungen der magnetischen Sortierung von Stahlteilen 
in der Massenfertigung (Critical Considerations of Magnetic Sorting of Steel Parts 
in Mass Production)," Z. Metallk., 45 (1954) : 243. 

SANFORD, R. L. "An Alternating-Current Magnetic Comparator, and Testing of Tool- 
Resisting Prison Bars," J. Research Natl. Bur. Standards, 16 (1936) : 563. 






Coercive-Force Tests 

Advantages of magnetic tests 1 

Applications in metallurgical tests 1 

Advantages of measuring coercive force and 

permeability 1 

Technique of measuring coercive force 1 

Coercive- force meter 2 

Block diagram of equipment for meas 
urement of coercive force of materials 

(/. 1) 2 

Coercive- force measurement equipment 

(/. 2) 3 

Compensation of earth's field 3 

Range of applications 3 

Measurement of magnetic-material properties 1 
Coercive -force measurements at high temper 
ature 4 

High -temperature magnetometer arrange 
ment for indication of true magnetiza 
tion and true field strength (/. 3) 4 

Automatic sorting of production parts by 

coercive force 5 

Residual magnetization as a measure of 

coercive force 5 

Magnetization loop of a short production 

part (/. 4) 6 

Measuring residual magnetization with ex 
ternal probes 5 

Automatic sorting arrangement ..... 6 

Arrangement for automatic sorting of 
production parts by residual field values 
and (for many alloys) hardness (/. 5) . . 7 
Advantages of automatic sorting by coercive 

force 6 

Applications of sorting by coercive-force tests 7 

Ball bearings 7 

Control of hardness of tempered parts .... 7 

Hardened steel parts 7 

Tempered castings 8 

Statimat graph of the frequency dis 
tribution of annealed cast parts (/. 6).. 8 
Electrical relay parts 8 

Point-Pole Tests 

The point-pole method -_ 

Measuring anisotropy in sheet materials 

Probe and magnet arrangement used to 
determine grain orientation of sheet 
steel (/. 7) 


Polar diagram of magnetic field about a 
point pole for various types of steel 

sheet (/. 8) 10 

Hysteresis loss of electrical steel sheets .... 10 

Internal stress in structural steel 10 

Analysis of carbon content 11 

Point-pole indication as a function of the 
carbon content for carbon steel (/. 9) . . 11 

Sorting mixed stocks 11 

Frequency distribution in a lot of two 
mixed alloys, obtained in a steel plant 

(/. 10) 12 

Sorting of two similar alloys with an ex 
panded measurement scale (/. 11) 12 

Hardness testing 13 

Relation between point -pole indication 
and mechanical hardness of centrifugally 

cast tubes (/. 12) 13 

Depth of case hardening 13 

Other applications 13 

Scale expansion with compensating magnet 14 

Double-pole tests 14 

Difference- probe arrangement for mag 
netization with a pot magnet (double- 
pole pole) to suppress influence of the 
earth's field (/. 13) 14 

Thickness Tests 

Probe measurement of nonmagnetic wall 


Probe array with permanent magnet for 
wall thickness measurement of nonfer- 
rous metals (independent of wall mate 
rial) (/. 14) 

Scale expansion 

Cast nonferrous parts 

Automotive piston heads 

Large wall thickness 

Multiple-point measurements of wall thick 

Sheet and plate thickness measurements .. 
Multiple thickness- measurement device 
with simultaneous indications on the 
screen of a cathode-ray tube (/. 15) ... 
Continuous- sheet thickness measurements . . 
Hall -generator measurement of wall thick 

Arrangement of Hall probe for measure 
ment of wall thickness (/. 16) 



CONTENTS (Continued) 


Sheet thickness meter 17 

Nonfeiroinagnetic- coating thickness meter . 17 
Arrangement of permanent magnet with 
Hall generator for measurement of 
thickness of nonferromagnetic layers of 

iron and steel (/. 17) 18 

Eccentricity of a conductor in insulating 

covering IS 

Probe arrangement for direct representa 
tion of position of conductor in an in 
sulating covering (/. 18) 19 

Feedback control 19 

Measurement of the eccentricity of tubes 19 

Field within an eccentric tube carrying cur 
rent ! 19 

Arrangement for measuring the eccentricity 
of a nonferrous metal tube by measur 
ing the magnetic field inside the tube 
produced by a current in the tube wall 
(A 19) 20 


Other Applications 

Crack depth measurement 20 

Effect of crack on path of current (f. 20) 21 
Determination of iron impurities in nonmag 
netic material 21 

Arrangement for rapid, continuous meas 
urement of iron content in non ferro 
magnetic materials (/. 21) 22 

Detecting ferrous material in packaged 

foods 22 

Iron in nonferrous scrap 22 

Iron core-wire in nonferrous castings 22 

Unexploded blasting cartridges 22 

Survey markers 23 

Buried, unexploded bombs, pipelines, cable 

boxes 23 

Liquid-level control 23 

Flow meters 23 

Detecting movement of nonferroua materials 23 

References 23 




Coercive-Force Tests 

ADVANTAGES OF MAGNETIC TESTS. Static magnetic field tests have 
found many applications in nondestructive testing in recent years for the follow 
ing reasons: 

1. Test indications are representative of the entire cross-section of the test object 
and are related to important material characteristics. 

2. Static field methods used in defect testing have a greater depth of penetration 
than varying fields, since skin effect is absent. 

3. Wall thickness indications for nonferromagnetic materials are independent of 
the electrical properties of the test object. 

electronic techniques has increased considerably in industrial measurement, test, 
and control techniques for metallurgical tests. However, the well-known classical 
measurement methods are still widely applied. The metallurgist measures the 
physical properties of his material for various reasons. First the measurement 
serves to identify or determine the application of the material. For example, 
the specific electrical conductivity of materials used for electric circuits or the 
coercive force of materials used for permanent magnets can be determined. 
Second, and this represents the predominating case, the measurement can provide 
a means of observing metallurgical processes such as transformations, age- 
hardening effects, ordering processes, aging, and other phenomena. 

Advantages of Measuring Coercive Force and Permeability. The physical 
factors with the greatest variation range in ferromagnetic materials are coercive 
force and permeability. Most physical properties such as conductivity, density, 
or elasticity will change by a factor of only 10 to 100 over the entire metal- 
lurgically interesting range, whereas the variation in coercive force and per 
meability is about 10 5 . In permeability measurements, the shape of the test 
sample enters decisively into the result, but the coercive force represents a 
factor which can be measured independently of shape. Therefore a rod or smaU 
grindings of a given material will give, without recalculation, the same measured 
values. In addition, by suitable instrumentation, the coercive force can be 
measured as a function of temperature more readily than any other factor. 

Technique of Measuring Coercive Force. Coercive force is defined as the 
demagnetization field strength H c required to bring about the disappearance of 
the magnetization intensity / or the residual induction B in a material, tal 
lowing saturation. Therefore the measurement process consists of increasing the 
field strength in a material to saturation, followed by a reversal in field strength 
back to zero. The sample then has a residual magnetization. If a field ot re- 




versed polarity with increasing strength is applied to the sample, magnetic induc 
tion will decrease to zero. The strength (in oersteds) of the opposing field at 
which the magnetic moment passes through zero is the coercive force H c . 

COERCIVE-FORCE METER. Fig. 1 shows the block diagram of a com 
mercially available instrument designed to measure the coercive force with high 
accuracy in a few seconds, A variable transformer, in conjunction with a rectifier 
and filter circuit, supplies direct current to the magnetic coil. The current is 
continuously variable by means of a knob on the front panel of the equipment 
(see Fig. 2). By turning this control once from zero to full scale and back again, 









^220 V A 

Institut Dr. Foerster 

Fig. 1. Block diagram of equipment for measurement of coercive force of 


the sample is magnetized to saturation and returns to the residual field point. 
The field meter now shows a deflection corresponding to the residual magnetiza 
tion of the sample. By throwing the current-reversing switch from the "mag 
netizing" to "measuring" position and turning the variable control transformer 
once more, a field appears whose direction is opposite to the original saturation 
field. This opposing field is now increased by means of the variable transformer 
until the deflection caused by the sample magnetization passes through zero. 
At this moment the absolute value of coercive force appears on the calibrated, 
precision, coercive-force meter. 

Operation of the equipment depends on the use of high-sensitivity field 
probes in conjunction with a magnetic-field meter. The probes are aligned so 
that the lines of force from the magnetizing field coil cannot influence the field 
probes. However, as shown in Fig. 1, the lines of force emanating from the 
sample are measured, and thus the magnetic-field strength in the sample may 



ind r ndentl y f the field in the main magnetization 

complete set-up is illustrated in Fig. 2. 

me^sSlhfn f - ^f^ ^ Additio1 ^ P recauti * must be taken when 
measuring the coercive force of very soft magnetic materials. The effect of the 

Institut Dr. Foerster 
Fig. 2. Coercive-force measurement equipment. 

earth's magnetic field will cause a considerable error at lower levels of measure 
ment. Compensation of this field at the sample location in the coil may be done 
with the same equipment and will completely eliminate this extraneous effect. 

Range of Applications. The sensitivity of the probe-coil method for measur 
ing coercive force is so great that accurate results may be obtained even in the 
case of austenitic steels and nonferrous metals with small ferromagnetic additions. 
Relay parts or hard metal parts may be sorted according to coercive-force values. 

34 4 



Magnetic properties, such as coercive force, saturation flux density, remanence, 
permeability, or hysteresis loss, permit nondestructive determination of impor 
tant mechanical and metallurgical conditions. For example, coercive force aids 
in evaluating the quality of tungsten carbide, since coercive force increases, in 
accordance with Neel's theory, 1 as the cobalt is more finely distributed in the 
tungsten carbide. A fine distribution of cobalt is required for high wear resist 
ance. Many hard metals produced in Europe are tested with the coercive-force 
meter with Foerster probes. 2 Measurement of magnetic-saturation flux density 
indicates material properties such as residual austenite content and ferrite 
content. A universal arrangement for determining the entire d.-c. hysteresis 
loop as a function of temperature is especially suited to practical application. 3 
It measures the true field strength inside the test object. 

TURE. Fig. 3 illustrates a magnetizing coil surrounding an oven containing the 
test object. A specimen is selected with a dimension ratio whose specific demag 
netization factor N is known. Desirable shapes include the cylinder with 
cone-shaped ends or the rotation ellipsoid. In this arrangement, probes for meas 
urement of magnetization intensity of the test object are placed outside the mag- 

Fig. 3. 


Institut Dr. Foevster 

High-temperature magnetometer arrangement for indication of true 
magnetization and true field strength. 

netizing coil. Thus, by using an oven in the magnetizing coil, this arrangement 
can be used for measurement of the variation of coercive force with temperature. 
Coercive force is a sensitive indicator of metallurgical changes such as conver 
sion, precipitation, and age hardening 4 in the test specimen. The lines of force 
emerging from the specimen are detected by two d.-c, field-measurement probes, 
PI and P 2 . The field strength resulting from the test body at probe P l is propor 
tional to the magnetization 4jt/ of the specimen. The distance of probe P l from 


the specimen is adjusted initially so that the ratio of the field strength at the 
location of the probe P 1 to the magnetization 4jt/ in the test body is exactly 
1:10,000. Thereafter, all field strengths measured at the probe P x need only be 
multiplied by 10,000 to obtain the magnetization inside the test specimen. 

Probe P 2 is located within an auxiliary coil energized by the same current as 
the magnetizing coil. Thus the field generated in the auxiliary coil is proportional 
to the field H a of the magnetizing coil at the test specimen. The field resulting 
from magnetization of the specimen reacts simultaneously on probe P%. The 
distance of probe P 2 from the test body is fixed so that the field strength result 
ing from the magnetization, 4jt/, of the test body is N-L Thus two fields, 
namely, H a and N-I, react on the probe P 2 . These two fields oppose each other. 
Together they give the true field strength H t = H a N-I. The two meters in 
Fig. 3 show the true magnetization and the true field strength of the specimen, 
respectively. The test specimen can be heated above the Curie point in the 
furnace within the coils. 

CIVE FORCE. 4 ' 5 - 6 The coercive force is the most important magnetic charac 
teristic of test materials. It is related to significant mechanical and metallurgical 
properties such as hardness, tensile strength, depth of case, alloy content 
(especially carbon content) and aging conditions. These properties can be 
nondestructively determined in many cases by measuring the coercive force. 
However, before magnetic sorting methods are used, extensive experiments are 
needed to check the relationship between the coercive force and the significant 
properties, such as hardness or tensile strength. It is essential that no reversal 
appear in the function relating hardness to coercive force. Such reversals can 
appear with several alloys, especially after repeated heat treatment. However, 
coercive-force measurement requires three steps: 

1. Saturation. 

2. Reduction of the magnetization force H to zero. 

3. Increase of the opposing field until the magnetization is reduced to zero. 

This requires so much time that it is unsuitable for production sorting. 

Residual Magnetization as a Measure of Coercive Force. The residual 
magnetic field which remains after saturation magnetization is approximately 
proportional to the coercive force for normal production test parts. 7 Short pro 
duction parts do not exhibit the typical material permeability \L. Instead, they 
exhibit a greatly reduced form-permeability [ip. This is a function only of the 
ratio of specimen length to diameter (I/d). It is a result of the strong demag 
netization caused by free magnetic poles. For cylindrical parts the apparent 
form-permeability \JL F is 8 9 

However, the permeability is generally given as the increase of the magnetization, 
, with the field strength H, so that from the triangle of Fig. 4 we have 

/4jt/ r \ C9\ 

tan = \JLF = ( -JJ-- J W 

Measuring Residual Magnetization with External Probes. The residual 
magnetization 4jt/ r generates a magnetic field H r outside the test specimen. This 
external field E T is proportional to the residual magnetization I r . It is a function 



Fig. 4. Magnetization loop of a short production part (length, I; diameter, d). Note 
the proportionality between coercive force and the residual field of the test object. 

of the distance a of the probes from the test part and the length / of the test part 
only. From Eqs. (1) and (2), the residual field H r measured by the probe is 

H r = 

I =#//,( 64- 


The constant K is calculated from the distance a of the probo from the test 
object and the length I of the test object. The exact calculation is presented by 
Foerster. 7 

Eq. (3) indicates the proportionality between the field measured with the 
probe and the coercive force of the test specimen. It is the basis of several fully 
automatic sorting instruments. 

Automatic Sorting Arrangement. Fig. 5 illustrates such an automatic sorter 
schematically. The test parts are fed to the conveyor belt at specific intervals. 
They then pass through the magnetizing coil. At a specific distance beyond the 
coil, they pass the probe. The latter measures the residual field (proportional 
to the coercive force) within 1/1000 sec. Gates are electronically controlled to 
sort test parts according to their coercive-force values. These values are related 
to specimen hardness, tensile strength, carbon content, or other factors. 

An astatic arrangement of four probes 7 can be used to eliminate the influ 
ences of the earth's magnetic field and inexact centering of test parts on the 
conveyor belt. Test indications are independent of the speed of parts passing 
the probe. This has lead to a very simple test arrangement. Test parts are 
placed in an inclined groove, slide through the magnetizing coil, pass the probe r 
and are automatically sorted according to their hardness range. Just before 
entering their particular containers, parts pass through a demagnetizing coil to 
remove residual fields which may affect subsequent machining operations. 

Advantages of Automatic Sorting by Coercive Force. Automatic coercive- 
force sorting has three significant advantages: 

1. Low sensitivity to test-object dimensions and shape. 

2. Uniform response to the entire cross-section of test objects. 

3. High sorting speed. 





A J&. 



7 7 





Institut Dr. Foerster 

- F r f ample ' a weU -^ ball-bearing plant daily tests 
* b r m ?. pms * nd a corresponding number of rollers and baU-LS 

K + v, Parts ' Other ^^ tert vital auto 

subject to high wear, to attain uniform tempering Before the 
change-over rom mechanical hardness testing of each individual ' part to h 
much faster electronic testing, one plant made comparison tests between mechan! 
ical and magnetic hardness testing on 40,000 parts from various heats This 
pensive test showed that the deviation between mechanical and dectriS 
determinations of the hardness lay within the normal variation of Rockwell hard 
ness measurements. Individual mechanical tests could not have given more exact 
results than the much faster nondestructive sorting by hardness values. 

Hardened Steel Parts. In the magnetic sorting of high quality diesel motor 
parts" subject to high wear (such as injection pumps and nozzle needles) it 
was proved statistically that a considerable increase was obtained in the average 
service life. Material structure could be kept to much closer tolerances as a 
result of the residual-field measurement. Mechanical hardness testing obviously 
could not be applied to 100 percent of production 



Tempered Castings. In tempered cast parts, 32 it is important to control 
carbon content. Excessive carbon results in a poor machineability ; insufficient 
carbon results in difficulties in the threading dies. The material is too soft, or it 

Residual fields remaining after magnetization of tempered cast parts corre 
sponds exactly to the carbon content; so, tempered cast parts are sorted auto 
matically according to the mechanical machineability. The results of sorting 




80 WO 




1 III 











I I 


Institut Dr. Foerster 

Fig. 6. Statimat graph of the frequency distribution of annealed cast parts. Sorting 
limit between 30 and 75 scale divisions. 

are easily observed with the Statimat instrument, a newly developed method for 
the automatic indication of statistical distributions. The twelve indicating chan 
nels of this instrument show the statistical frequency distribution of measured 
properties such as hardness, tensile strength, machineability, or carbon content. 
Fig. 6 shows the Statimat graph resulting from sorting of tempered cast parts 
for machineability. 

Electrical Relay Parts. In the examples previously discussed, measurement 
of coercive force was only an auxiliary means for indicating hardness. European 



plants in the electrical industry are also using magnetic-sorting instruments for 
production sorting of relay parts into various coercive-force groups. Coercive 
force represents an important characteristic of relay parts. Excessively high 
coercive force leads to erratic relay operation. 

Point-Pole Tests 

THE POINT-POLE METHOD. In cases of strong demagnetization (for 
example, of short production parts), the residual magnetic field remaining after 
previous saturation is proportional to the coercive force. This important relation 
ship between residual field and coercive force 13 is valid both in case of magneti 
zation of entire test parts and for local point-pole magnetization. 14 







Institut Dr. Foerster 

Fig. 7. Probe and magnet arrangement used to determine grain orientation of 

sheet steel. 

If a test object is brought into brief contact with the suitably contoured point 
of a permanent magnet, a residual magnetic pole remains in the test object. 
The larger the demagnetization factor N, the more exact the proportionality 
between the pole's residual field and coercive force. The surface of a test part 
(mathematically described as an infinitely large half-plane) has an especially 
strong demagnetizing action on the point pole. This corresponds to a large 
demagnetization factor. Consequently the requirement for proportionality 
between point-pole strength and coercive force is fulfilled. The residual field 
strength indicates the coercive force at the point pole. 

A series of measurements on sheets, machine parts, and forgings has confirmed 
this proportionality between the point-pole indication, obtained with the probe, 
and the local coercive force. 13 

tion of a rod magnet with a probe arrangement, as shown in Fig. 7, is used for 
determination of the anisotropy of sheets. Pushing the spring-loaded magnet 
down onto the sheet surface will cause a point magnetic pole to appear on the 
sheet surface. The tangential component of the point-pole field is measured by 



the difference probe pair. The deflection of the meter connected to the pair is 
proportional to the coercive force at the location of the point pole. This permits 
quick and nondestructive sorting of sheets by coercive-force values. These 
measurements indicate certain engineering; properties such as elasticity or deep- 
drawing ability. If the probe pair is rotated around the point pole, a constant 
deflection of the meter will be obtained only in case of isotropic material. 

Fig. S is a polar diagram of the instrument deflection for various sheets. 
Here all maximum deflections are normalized to the same value. This method 
is of importance in the electrical industry where grain-oriented sheets are used, 
as well as in the sheet-metal industry where too heavy an orientation leads to 
undesirable lobe-formation during deep drawing. 






Fig. 8. Polar diagram of magnetic field about a point pole for various types of 

steel sheet. 

Hysteresis Loss of Electrical Steel Sheets. The point-pole method is used 
extensively in sorting of electrical steel laminations. Generally the hysteresis loss 
of electrical sheets increases with coercive force, i.e., with widening of the 
hysteresis loop. Thus the point-pole strength of electrical sheets represents the 
static component of the wattage loss. This hysteresis loss is measured very 

Internal Stress in Structural Steel. An especially important application of 
the point-pole method is the nondestructive determination of internal stress in 
steel structures. Normally, structural steels do not exhibit magnetic anisotropy. 
Under tensile or compressive stress, anisotropy results in the material. This 
causes a deformation of the circle. The internal stress is determined from the 
ratio of maximum to minimum deflection and the calibration for the correspond- 



ing steels. Thus the point-pole method has possible uses as a nondestructive 
method for instantaneous determination of the magnitude and direction of inter 
nal stresses. The stress direction is indicated by the direction of maximum 
deformation of the circle. 

ANALYSIS OF CARBON CONTENT. On large machine parts the point- 
pole method tests the uniformity of heat treatment or of carburized or induction- 
hardened layers. In steel plants, large pieces of steel are tested for carbon con- 





0,1 0,2 0.3 0,4 0,5 0,6 %C 

Fig. 9. Point-pole indication as a function of the carbon content for carbon steel. 

tent. Fig. 9 illustrates the relationship of the point-pole instrument indication 
(H c ) to carbon content. The band width in Fig. 9 indicates the accuracy of 
the magnetic carbon analysis with the point-pole method. 

Sorting Mixed Stocks. The point-pole method can also be used to sort semi 
finished steel parts into various alloy groups, as long as their coercive-force 
values are sufficiently distinguishable. Frequently a point-pole measurement at 
the ends of bars in stock piles is sufficient to separate mixed alloys. 

Fig. 10 illustrates the frequency distribution of indications from point-pole 
sorting of mixed lots of C1040 and C1060 steel rods. 15 The spread from 30 to 40 
scale divisions for the C1040 steel is a result of slight variations in the rnrbon 



20 30 

40 50 60 70 

60 90 


Fig. 10. Frequency distribution in a lot of two mixed alloys, obtained in a 

steel plant. 

content from various heats. Fig. 10 shows reliable separation of both steels, as 
indicated by the large distance between their indications. 10 The test data of Fig. 
10, obtained by counting, are found automatically by testing; with the Statimot 
instrument. Generally, two alloys can be separated if at least one channel re 
mains empty between the distributions of the two alloy groups. Fig. H illustrates 








85 90 95 

ll lllllllllllll 





Fig. 11. Sorting of two similar alloys with an expanded measurement scale. 

Feasibility of reliable separation is indicated by empty sorting chnnnnls between the 

two frequency groups. 



the distribution of indications for two similar alloys (C1034, C1045), obtained 
with an expanded measurement range. The empty "indication channels indicate 
that reliable porting; i^ possible. 

Hardness Testing. The point-pole method can be used for rapid sorting of 
centrifugally cast tubes for surface hardness. Foerster shows the relation 
between the meter indication and the mechanical hardness of centrifugally cast 
tubes in Fig. 12. 6 This relationship must first be established experimentally with 
control samples whose alloy content, heat treatment, and microst nurture are 
held constant. Hardness is indicated nondestructively and quickly after pressing 
down the rod-shaped permanent magnet. 












300 400 500 



Fig. 12. Relation between point-pole indication and mechanical hardness of cen 
trifugally cast tubes. 6 

Depth of Case Hardening. Another application of this method is the local 
determination of the case depth on crankshafts and other machine parts. Since 
the static field method has greater penetration depth than the a.-c. field method, 
much heavier case depth and induction-hardened layers can be measured. The 
point-pole method can generally be used even on parts with large dimensional 

Other Applications. Small parts and fine cast iron (for example, in the sew 
ing machine industry) are normally cast in a cluster. By measuring only a 
section of this cluster, the quality of a number of test parts is determined 
simultaneously. The fine cast method uses carbon-arc furnaces, and occasionally 



carbon particles falling into the melt increase the carbon content undesirably. 
Such cast parts are easily detected by their excessive coercive force and high 
point-pole indications. 

Another application is hardness sorting of textile spindle tips. A point pole is 
induced in the spindle tip with a permanent magnet of suitable shape. The pole 
strength is proportional to the mechanical hardness of the outer tip. The method 
permits indirect hardness measurements at locations not accessible to mechanical 
hardness testing; for example, at the base of a %-in. cliam. injection-nozzle hole 
of a diesel engine. 11 

Scale Expansion with Compensating Magnet. The probe method permits 
convenient expansion of the measurement scale. This is clone with an adjustable 
compensation magnet near the upper gradient probe. The unused portion of the 
scale from to 75 scale divisions is suppressed to obtain the results shown in 
Fig. 11, Only indications above 75 scale divisions result in deflections. If a higher 
sensitivity is selected, the range between 75 and 100 scale divisions can be 
expanded to cover the entire scale. 

DOUBLE-POLE TESTS. A double magnetic pole 6 produced with a pot 
magnet is used with the gradient probe (tangential gradient of the normal com 
ponent) for tests on magnetized parts or tests in the presence of other interfer 
ing fields. Fig. 13 illustrates the pole form and probe arrangement for the 





Institut Dr. Foerster 

13. Difference-probe arrangement for magnetization with a pot magnet 
(double-pole pole) to suppress influence of the earth's field. 

double residual-pole test. The lines of force of the residual pole caused by the 
pot magnet at the first probe are opposite in direction to those at the second 
probe. This field distribution is measured by the gradient probe. Magnetization 
of the entire rod (caused, for example, by the earth's field) creates magnetic lines 
of force having the same direction at each probe. The gradient probe is insensitive 
to such fields. 



.Thickness Tests 


NESS." The gradient probe shown in Fig. 14 is not affected by the earth's field 
(normal gradient of the normal component). If it is placed on one side of a 
wall while a small permanent magnet is held on the opposite side, the field 
strength at the probe is a function only of the distance from magnet to probe. 
-The measurement is not a function of the characteristics such as electrical con 
ductivity of the wall material, as are alternating field methods which have been 
suggested for this purpose." The indication scale is identical for all nonferrous 
metals as well as for glass, plastic, or wood. 





Institut Dr. Foerster 

Fig. 14. Probe array with permanent magnet for wall thickness measurement of 
nonferrous metals (independent of wall material). 

Scale Expansion. Expansion of the thickness indication scale into a desired 
range is often convenient. The field of the main permanent magnet is compen 
sated at the probe with a small, adjustable magnet in the probe head. The 
"required" value is centered to H = on the wall thickness scale. Thus, by 
selecting a suitable sensitivity, a 1/10-, 1/100-, or l/1000-in. deviation 
from the required value results in a full-scale deflection. 

. The probe with difference coils (Fig. 14) contains a screw cap which can be 
adjusted in or out for the desired measurement range. The instrument uses the 
same scale for various thickness ranges. By adjusting the cap, the distance from 
magnet to probe for the selected required value is always the same. 



Cast Nonferrous Parts. For wall thickness measurement of cast nonferrous 
parts, the probe is placed outside the test part (Fig. 14). The very small perma 
nent magnet, mounted on a flexible wire, can be inserted even into complicated 
hollow bodies, 18 

Automotive Piston Heads. An automatic instrument has been devised for 
sorting automotive piston heads into various thickness groups. The difference 
probe is held lightly against the bottom side of the piston by a, spring. As soon 
as the center of the piston is below the permanent magnet, the magnet is pressed 
down against the piston head. At the same instant the light beam of the instru 
ment indicates the piston thickness. A color marker, triggered by photocells 
on the instrument scale, marks the piston according to its thickness group. 

Large Wall Thickness. Wall thicknesses far above 100 in. are measured with 
large permanent magnets (for example, 1X4 in.). The probe method is used in 
mining to measure the thickness of walls remaining between two galleries drilled 
from opposite sides. 


Probe-and-magnet wall thickness measurement devices are relatively small and 
are independent of the rigidity requirements of a mechanical caliper. This 
method is especially suitable for multiple measurements where numerous points 
are measured simultaneously. 

Sheet and Plate Thickness Measurements. Fig. 15 illustrates a ten-point 
wall thickness measurement device for nonferrous metal, wood, glass, plastic, and 
similar materials. The probes are pressed lightly by springs against the bottom 
of the surface to be measured. The permanent magnets are located opposite the 




Institut Dr. Foerster 

Fig. 15. Multiple thickness-measurement device with simultaneous indications on 
the screen of a cathode-ray tube. 

probes, on the top surface. These magnets can be conveniently lowered to the 
surface to be measured by an electromagnetic device. With electronic scanning 
of all measurement points, indications appear simultaneously on a cathode-ray 
tube. This display shows deviations from standard values and indicates the 
shape of the test object. 



Continuous-Sheet Thickness Measurements. The method is also suitable 
for a continuous operation in which the individual thickness indications and the 
cross-section of the test object appear continuously as a luminous picture on the 
screen. Thicknesses which exceed or fall below the tolerance trigger a signal. 
Methods using a roller guide for the probes and magnets are used in the lino 
leum and plastic industry. The thickness of the material to be tested is recorded 
during production. 

Test indications are entirely independent of the rigidity of the supporting 
fixture which carries the probes and the small permanent magnets (Fig. 15). 
Continuous thickness measurements on flat nonferrous materials can be carried 
out far from the edge without constructing an especially rigid fork to carry the 
probe and the magnet. 


Fig. 16 illustrates the application of the Hall generator for the wall thickness 
measurement of ferromagnetic material from one side. 19 A pot-shaped yoke 
magnet magnetized by direct current is placed on the wall whose thickness is 
to be measured. The Hall generator is placed in the induced flux of the center 








\ /v : "~* 




\ \ 

\ \ 

\ v 

\ \ 

:rir r 

\ \ / / 



t 1 


\ I 
1 1 








Institut Dr. Foerster 

Fig. 16. Arrangement of Hall probe for measurement of wall thickness. 

leg. When the yoke magnet is placed on the wall, the increase in magnetic flux is 
proportional to the product of wall thickness and saturation magnetization of the 
wall material. For many materials, such as deep-drawing sheets, boiler walls, 
and boiler tubes, the saturation magnetization is constant, so that the instru 
ment deflection can be calibrated directly in wall thickness. 

Sheet Thickness Meter. By means of the electrical compensation, a small 
portion of the thickness range can be represented on the measurement scale. A 
sheet thickness meter with Hall generator is used for automatic sheet-metal 
thickness sorting. Adjustable microphoto cells, providing a maximum of 30 
sorting groups, are located on the scale of the light-beam galvanometer for 
adjustment of the desired thickness in each sorting group. 

Nonferromagnetic-coating Thickness Meter. Another application of the 
Hall generator is the measurement of thickness of nonferromagnetic coatings on 



a ferromagnetic base. A series of instruments is available for coating thickness 
measurements. These measure the force required to lift a permanent magnet 
from the surface of the layer to be measured. A greater lifting force indicates 
a stronger magnetic induction in the tip of the permanent magnet. 

With the miniaturized Hall probe it is possible to measure the magnetic induc 
tion in the tip of the permanent magnet. The indicating instrument of Fig. 17 
can be calibrated directly in wall thickness values. 



Institut Dr. Foorster 

Fig. 17. Arrangement of permanent magnet with Hall generator for measurement 

of thickness of nonferromagnetic layers of iron and steel. The density of the lines 

of force in the tip is measured by means of tho Hall generator. 

ING. If current flows in a conductor, a ring-shaped magnetic field appears, 
surrounding the conductor. The field strength H is 

H = 



where H = the field strength in oersteds. 

7 = the current through the conductor, in amperes. 

R = the distance in centimeters from the center of the conductor to the point 
at which the field strength is to be measured. 

The displacement of the conductor from its concentric position is measured 
with a suitable probe arrangement moved by means of small rollers on the 
insulated conductor. Fig. 18 illustrates the arrangement with which the position 
of the conductor in its covering is indicated on a cathode-ray tube screen. 

The direct current flowing through the conductor generates a magnetic field H 
at the probe array. The probe array is constructed so that no indication is 
obtained when the conductor is centered with reference to the probes. With an 
eccentric conductor, the field reaction on the difference-probe pairs, Pi-Pa and 
Ps-P*) will no longer cancel. By connecting probes P l and P 2 (which measure the 
horizontal deviation of the conductor from its concentric position) to the hori- 



zontal plates and probes P 3 and P 4 to the vertical plates, the position of the 
beam of the cathode-ray screen indicates the position of the conductor in the 
insulating cover. If the eccentricity of the conductor in the insulating cover 
exceeds a specific value, the beam moves out beyond the tolerance circle on the 

Cathode Ray Screen 

Institut Dr. Foerster 


Fig. 18. Probe arrangement for direct representation of position of conductor in 

an insulating covering. 

Feedback Control. The voltages produced by the probes can be used to guide 
the conductor back to its concentric position by means of a suitable control 
mechanism. The direct current which flows through the conductor is simulta 
neously used for the sensitivity control of the probe amplifier. In this way the 
eccentricity measurement is independent of the magnitude of the direct current 
through the conductor. A similar arrangement can be used to determine the 
eccentric position of the metallic core in covered welding electrodes. 

current passes through the wall of a concentric tube, the static magnetic field 
outside the tube will be exactly the same as though the current were concen 
trated in a central conductor within the tube. The inside of a concentric tube 
is entirely free of magnetic field. 

Field Within an Eccentric Tube Carrying Current. In an eccentric tube 
with nonuniform wall thickness, a completely homogeneous magnetic field 
strength, # ecc , appears. As shown in Fig. 19. this field is perpendicular both^to 
the tube axis and to the line connecting the point of minimum wall thickness with 
that of maximum wall thickness. 

The eccentricity e in the tube results in a field # ece . calculated as 

Here e, as indicated in Fig. 19, is the displacement of the center point of the 
inside waU of the tube with respect to the center point of the outer wall of the 
tube. In a concentric tube, both center points coincide (e = 0), and 

Ri the outside radius. 
R a = the inside radius of the tube. 
/ = the current in amperes which flows parallel to the tube axis. 



By using two probe pairs placed perpendicular to each other and connected 
to the deflection plates of the cathode-ray tube, the magnitude and direction of 
eccentricity of a tube can be read directly from the screen. 

Eccentricity in wall thickness of tubes can also bo determined by measuring 
the radial component of the field strength outside the tube. In a concentric 
tube the radial component of field strength is always zero. In this case the mag 
netic lines of force are concentric circles around the tube, perpendicular to the 
direction of the radius. However, the radial surface field strength for an eccen 
tric tube is never zero. 

Institut Dr. Foerster 

Fig. 19. Arrangement for measuring the eccentricity of a nonferrous metal tube 
by measuring the magnetic field inside the tube produced by a current in the 

tube wall. 

Foerster has described the mathematical derivation of the field distribution of 
an eccentric tube - and has presented examples of nondestructive eccentricity 
measurement on other production parts. 

Other Applications 

CRACK DEPTH MEASUREMENT. The small dimensions of the Hall 
microprobe make possible the measurement of the magnetic leakage flux of 
cracks on the surface of test objects. Thus, the Hall probe can be used for the 
experimental verification of theoretical calculations 21 for the magnetic leakage 
flux over cracks. 

Fig. 20(a) illustrates the arrangement of the Hall probe for crack depth 
measurement. The crack lies between the two contacts which apply a direct cur 
rent to the test part. The Hall microprobe is located directly above the crack 
so that the magnetic d.-c. field is perpendicular to the largest surface of the Hall 
crystal. The instrument connected with the Hall probe indicates the magnetic- 
field strength generated by the direct current. In the absence of a crack, Fig. 



20 (b), the main portion of the current flows directly below the surface. Zero 
crack depth corresponds to a specific d.-c. field strength at the location of the 
Hall probe for this current distribution. The Hall voltage at zero crack depth 
can be compensated by means of a compensation voltage E$ obtained from the 
d.-c. circuit across a resistor R. If a crack is present, the d.-c. current path is 
forced deeper into the test part. However, the greater length of the current path 
results in a weakening of the magnetic d.-c. field. This weakening increases with 
crack depth. Thus it seems practical for crack depth applications to determine 
the magnetic field instead of the potential difference above the crack. The non- 
contacting Hall probe requires much lower currents and eliminates interference 
caused by thermoelectric forces. 

Institut Dr. Foerster 

Fig. 20. Effect of crack on path of current, (a) Current path around crack, 
(b) Current path without a crack. 

MATERIAL. For many applications it is important that a material be as free 
as possible of magnetic impurities. For example, an iron-free material is required 
for the construction of electrical measurement instruments, such ^as compass 
housings. Fig. 21 illustrates a simple arrangement for the quantitative determi 
nation of very small amounts of free, undissolved iron in small, rod-shaped test 
samples The arrangement utilizes a Foerster probe at a specific distance above 
a strong permanent magnet. The field of the large permanent magnet is com 
pensated at the probe by a small permanent magnet whose distance above the 
probe is adjustable. In the absence of a test object, the probe indicates zero field 
strength. If a specimen is brought into the field, a large measurement effect 
appears even in case of low iron impurities. This permits quantitative deter 
mination of the iron content. The instantaneous indication of iron content 
permits continuous operation as rods and tubes pass between the probe and tne 
permanent magnet. 



Detecting Ferrous Material in Packaged Foods. Induction methods for 
detection of iron contamination cannot be used with metal-foil wrapped packages. 
However, the static magnetic-field test can be applied. The packages pass 
through the field of a permanent magnet for sorting. The probe is placed in an 
astatic gradient arrangement independent of the earth's field and perpendicular 
to the field of the permanent magnet. For example, iron- wire particles weigh 
ing less than 1()- 3 gram produce additional lines of force detectible by the probe. 
These can be utilized for the automatic elimination of contaminated packages. 


Compensation magnet 

Probes \FERROUS 

Permanent magnet 

Institut Dr. Foerster 

Fig. 21. Arrangement for rapid, continuous measurement of iron content in non- 
ferromagnetic materials. 

Iron in Nonferrous Scrap. Specially designed search instruments are used to 
locate compact iron parts in scrap packages. In tightly pressed nonferrous metal 
scrap, the presence of iron parts is disadvantageous. Those iron parts can be 
detected from the outside with a special search instrument. 

Iron Core-Wire in Nonferrous Castings. In the casting of complicated non- 
ferrous castings, it frequently happens that iron core-wires or chills sometimes 
remain inside the cast bodies. These later cause difficulties. These wires are 
easily located with a small model of the search instrument. 

Unexploded Blasting Cartridges. In the following example, the word 
"nondestructive" is of special significance because it does not refer to test parts 
but to people. In mining and in construction of tunnels, large numbers of blast 
ing cartridges are placed and simultaneously detonated. Occasionally a cartridge 
does not explode. A worker may later accidentally explode an unfired cartridge, 
resulting in injuries or death. One European country which has to construct an 
especially large number of tunnels and power plants is considering equipping all 
blasting cartridge with small magnets. These would be pulverized during the 
explosion. As soon as a detonation of many cartridges occurred, the blast-off 
wall would be scanned with a gradient probe, which operated independently of 
the earth's field but reacted to the field of the small magnets. Each unexploded 


cartridge would be immediately indicated on the gradient probe instrument by 
a flashing signal lamp. 

Survey Markers. Another application of the gradient probe is in geodesy. 
In the measurement of land boundaries, markers are posted to record the 
survey. These, however, can be displaced as time passes. If a sealed glass tube 
containing a permanent magnet the size of a pencil is buried at a depth of 3 ft., 
the exact location can be identified within an accuracy of 1 in. with a simple 
gradient-probe instrument on the ground surface. 

Buried, Unexploded Bombs, Pipelines, Cable Boxes. Most European 
countries use search instruments with Foerster probes for the location of iron 
parts such as dud bombs in the ground. These instruments are also used to 
locate hidden pipelines. The Telephone and Telegraph Division of the German 
Federal Postal Department uses these search instruments to relocate buried 
cable boxes. 

LIQUID-LEVEL CONTROL. The combination of the (L-c. probe with a 
small permanent magnet has numerous other measurement possibilities. The 
liquid level in a closed container can easily be observed and controlled by using 
a small permanent magnet sealed in a glass tube as a float. The approach of this 
magnet to the built-in gradient-probe array can be quantitatively indicated on 
the outside of the container. Liquid level is indicated not only at a predetermined 
limit but over the entire measurement range. Probe signals can be used for 
control of the liquid level. 

Flow Meters. Another application is the measurement of the speed of flow 
of a liquid in a closed tube. A small turbine wheel placed in this tube carries a 
small permanent magnet. The wheel rotates in proportion to the speed of flow. 
The reaction of the permanent magnet on the probe outside the tube continu 
ously indicates the number of rotations of the wheel. The small permanent 
magnet reacts on the probe even through iron tubes. The speed of rotation meas 
ures the speed of flow. Electronic integrating counters can measure the total 
amount of liquid passing the measuring point. 


probe is located in the field of a rod-shaped permanent magnet so that lines of 
force of the magnet are perpendicular to the probe, no field strength is indicated 
by the probe. However, as soon as a sheet of nonferromagnetic metal is moved 
between the probe and the magnet, the lines of force of the permanent magnet 
are apparently carried along because of the well-known eddy current effect. In 
consequence the probe is no longer perpendicular to the lines of force. The field 
strength measured while a metallic body moves between the probe and the mag 
net is proportional to the thickness of the metallic body, its electrical conduc 
tivity, and its speed. With this arrangement it is possible to obtain large 
deflections from movements of 1/1000 in. per second. 


1 NEEL, L. "Nouvelle theorie du champ coercitif (New Theory of the Coercive 
Field), "Physica, 15 (1949): 225. 

2. FOERSTER, F. "New Measuring Methods," Ind. Ameiger, 77 (1955): 220. 

3. . "An Astatic Magnetometer for the Determination of Properties and its 

Temperature Dependence," -4rc/i. tech. Messen. (in press). 


4. FOERSTER, F. "Die schnelle und genaue Messung der Koerzitivkraft (A Quick 

and Accurate Measurement of the Coercive Force)," Arch. tech. Messen., Lfg. 
254 ( 1957) : 65 ; Lfg. 255 (1957): 87. 

5. ."A Method for the Measurement of D. C. Fields and D. C. Field Differ 
ences and its Application to Nondestructive Testing," Nondestructive Testing, 
13, No. 5 (1955): 31. 

6. . ki Ein Verfahren zur Messung von magnetischen Gleichfeldern und Gleich- 

felddifferenzen und seine Anwendung in der Metallforschung und Technik 
(A Method for the Measurement of Steady Magnetic Fields and Differences of 
Steady Field? and its Application to Metallurgical Research and Industry)," 
Z. Metallk., 46 (1955) : 358. 

7. . "Theoretische und experimentclle Grundlagcn der olektromagnetischen 

Quiilitiietssortierung von Stahlteilen, IV. Das Restfeldverfahren (Theoretical 
and Experimental Foundation of the Electromagnetic Quality Sorting of Steel 
Parts. IV. The Residual Field Method)," Z. Metallk., 45 (1954): 233. 

8. BORSORTH, R. M. Ferromagnotixm. 2d ed. New York: D. Van Nostrand Co., 

Inc., 1953. Pp. 846, 847, 849. 

9. McCuiRG, G. 0. " Theory and Application of Coil Magnetization," Nondestruc 

tive Testing, 13, No. 1 (1955): 23. 

10. ORTHEIL, J. "Kritische Betrachtungen der magnetischen Sortierung von Stahl 

teilen in der Massenfertigung (Critical Considerations of Magnetic Sorting of 
Steel Parts in Mass Production)," Z, Metallk., 45 (1954): 243. 

11. HAINZ, R. "Beispiel einer Einfuehrung der olektromagnetischen Sortiervorfahren 

in die Fertigung von Dieseleinspritzpumpen und -duesen (Example of the Use 
of Magnetic Sorting in the Production of Components for Diesel Injection 
Pumps and Nozzles)," Z. Metallk, 45 (1954): 238. 

12. FOERSTER, F. "Nouveaux precedes d'essai electronique nondest motif des 

materially (New Procedures for Electronic Nondestructive Testing of Mate 
rials)," Metaux (Conosion-Inds.), 26 (1951): 497. 

13. FOERSTER, F., and G. ZIZELMANN. "Die schnelle zerstoenmgsfreie Bestimmung der 

Blechanisotropie mit dem Restpunktpolverfahren (The Rapid Nondestructive 
Determination of Sheet Anisotropy with the Residual Point Pole Method),' 1 
Z. Metallk., 45 (1954) : 245. 

14. . 'The Quality Sorting of Sheets with Electrical and Magnetic Methods," 

Blech, 1 (1952): 1. 

15. MICHALSKJ, A. "Nondestructive Testing with the Magnetic Leakage Field 

Method," Z. Metallk. (in press). 

16. FOERSTER, F. "New Methods of the Quality Control from the Viewpoint of 

Automation," Tech. Mitt., (April, 1957). 

17. COLTEN, R. B. "Noncontacting Gages for Nonferrous Metals," Electronics, 29, 

No. 3 (1956) : 171. 

18. SCHNEIDER, PH., and P. DEKKER. ''Die Wandstaerkemessung von Leichtmetall- 

gusstellen mit dem Sondenkawimeter nach Dr. Foerster (The Measurement of 
Wall Thickness with the Foerster Kawimeter)," Metatt, 3 (1949): 321. 

19. FOERSTER, F. "The Application of Hall Generators for Nondestructive Testing," 

Z. Metallk. (in press). 

20. . Zers.toeru,ngsfreie Werkstoffpniejung mit ekktmchen und magnetischen 

Verfahren (Nondestructive Material Testing with Electrical and Magnetic 
Methods). Berlin, Goettingen, Heidelberg: Springer-Verlag (in press). 

21. . "Theoretical and Experimental Foundation of the Leakage Field," Z. 

Metallk. (in press). 






Measurement of Wall Thickness 
and Crack Depth 

Principle of test 1 

Test development 1 

Test requirements 1 

Applications and limitations 1 

Test equipment 2 

Schematic illustration of equipment used 
in direct- current conduction testing of 

wall thickness and crack depth (/. 1) . . 2 

Electrodes 2 

Electrode spacing 2 

Current source 2 

Potential measurement 2 

Selection of components 3 

Preparatory measurement procedure 3 

Electrode spacing 3 

Surface preparation 3 

Electrode contact 3 

Wall thickness measurement procedure 3 

Crack depth measurement procedure 3 

Interpretation of Test Indications 

Flat plate thickness 4 

Electrode arrangement 4 

Calibration curves for case of four- elec 
trode array on an infinite plate (/. 2 ) . . . . 5 

Thickness -potential relation 4 

Thin-plate modification 4 

Calibration 4 

Edge corrections 5 

Values of K f for equally spaced array 
(BIA = 1) near an edge of a semi-infinite 

plates (/. 3) 8 

Positioning electrodes near edges 5 

Temperature control 7 

Thickness of curved walls 7 

Crack depth indications 7 

Long perpendicular surface cracks 7 

Calibration curves for equally spaced 
array on "ideal" crack in finite thickness 

Plate (/. 4) 7 

Short perpendicular surface cracks 9 

Calibration curves for finite-length cracks 
having rectangular shapes in semi- 

infinite medium (/. 5) 8 

Calibration curves for finite-length cracks 
having circular shapes in a semi -infinite 

medium (/. 6) 8 

Other crack conditions 9 


Guide to Proper Application of 
Test Method 

Wall thickness measurement 9 

Selecting electrode spacing 9 

Selecting current magnitude 9 

Corrosion wall-thinning 9 

Crack depth measurement 10 

Selecting electrode spacing 10 

Selecting current magnitude 10 

Forging cracks 10 

Wall thickness measurement with square elec 
trode array 10 

Sensitivity of square array of electrodes ... 11 

Crack propagation study 11 

Tests of Railroad Rails 

Principle of test 11 

The magnetic field 11 

Magnetic field surrounding current-carry 
ing conductor (/. 7) 12 

Effect of a change in conductor cross-sec 
tion 11 

Induced electromotive force 12 

Magnetic field around sound rail 12 

Magnetic field about a current-carrying 

railroad rail (/. 8) 13 

Magnetic field around defective rail 12 

Searching coil reactions 13 

Rail test equipment 13 

Rail detector car (/. 91 14 

Main rail-current generator 13 

Main brush carriages 13 

Main brush carriage of rail detector car 

(/. 10) 14 

Third brush carriages 13 

Third -brush carnage which supplies pre- 

energizing current (f. 11) 15 

System for providing magnetizing current 
to main brush carriage and to third 
brush carriage (pre- energizing current) 

of rail detector car (/. 12) 15 

Detector unit carriage 16 

Searching units 16 

Power supplies 16 

Amplifiers 16 

Pen unit 17 

1 Rail detector car multiple-channel tape- 
recording unit (/. 13) 16 

Detector car tape 17 


CONTENTS (Continued) 


Paint guns 10 

Hand teat equipment 19 

Hand test equipment for determining 
exact location, size, and type of rail de- 

feets (/. 14) 17 

Personnel requirements 20 

Interpretation of indications 20 


Typical defects detected 20 

Typical rail defects (/. 13) ig 

Interpretation procedures 20 

Classification of defects 20 

Marking defects in rail 21 

References 21 

Bibliography 22 




Measurement of Wall Thickness and Crack Depth 

PRINCIPLE OF TEST. If four electrodes are placed in contact with an 
electrically conducting object of resistivity p (ohm-centimeter), and a current / 
(amperes) is passed between two of the electrodes, a potential V (volts) will be 
produced between the other two electrodes such that, 


j = Kp (1) 

Here, K is a constant having the units of length (centimeter) and depending 
only upon the geometry of the electrode arrangement and the geometry of the 
object. For a fixed electrode arrangement and one material, an observed change 
in the ratio V/I will therefore represent a change in geometry of the part. This 
is the basic principle used for the nondestructive measurement of wall thickness 
and crack depth by the direct-current conduction method. 

Test Development. Thornton and Thornton 7 ' 8 used four-point electrodes 
for measuring the wall thickness (from one side only) of a large variety of plates, 
pressure vessels, boiler tubes, ship hulls, and castings. During the course of this 
work they discovered that a surface crack interposed between the electrodes 
caused anomalous readings, and thereafter they employed the method for crack 
detection. 6 Other workers 3 4 5 used this principle to detect cracks, and some 
efforts were made to devise a calibration technique for crack depth measure 
ment. The method was finally developed l to a point where crack depths can be 
measured with fair accuracy in many cases. 

Test Requirements. Both wall thickness and crack depth determinations 
depend upon the existence of homogeneous and isotropic electrical resistivity 

within the test object. Fortunately, to the accuracy required in most cases, this 
is true for most metallic materials. The need to make electrical contact at four 
points and to measure very small potentials (about 30 jiv.) constitute the main 
difficulties of the measurements. The principal advantage is that only one 
material property, the resistivity, enters into the measurement. This simplifies 
interpretation of test indications. 

Applications and Limitations. The direct-current test method can be used 
for wall thickness measurements from one side only, where other techniques 
may not be applicable. Edge effects and other calibration difficulties tend to 
limit the usefulness of the method, but these can usually be overcome. The tech 
nique is very useful for crack depth measurement and is often the only method 
available. Its greatest field of application is to single cracks previously delineated 
by visual observation, magnetic-particle, or penetrant techniques. The method 
cannot be easily applied in continuous inspection problems, due to the difficulty 




in balancing: out changing thermal electromotive forces at the contacting 

TEST EQUIPMENT. Thu basic equipment For measuring wall thickness or 
crack depth is the same and is shown schematically in Fig. 1. Contact is made 
to the metal through the four electrodes, 0^ C> 2 , P 3 , and P.,, mounted either in a 
single insulating head (Fig. 1), or in two separate heads, as used by Thornton. 8 

200 Ohm 3Rv. 

Critical Dompiny HtI 'P* Wlth 
Resistor Multi-turn Dial 


-45 pa/mm 
Deflection Galvanometer 

Dept. of Mitien and Technical Surveys (Canada") 

Fig. 1. Schematic illustration of equipment used in direct-current conduction 
testing of wall thickness and crack depth. 

Electrodes. The electrodes are spring-loaded to facilitate contact on uneven 
or curved surfaces. A convenient electrode material is hardened drill steel which 
will retain a sharp point. 

Electrode Spacing. The electrode spacings depend upon the application. 
Large spacings (% in. or greater) are generally used in measuring thickness. 
Small spacings (% in. and less) are commonly used for crack depth measure 
ment. Unequal spacings are often used in wall thickness measurement, but equal 
spacings give best results in crack depth measurement. 

Current Source. The current 7 is passed through the metal between current 
electrodes d and C 2 . It is supplied from a battery or other d.-c. source capable 
of delivering up to 10 amp., depending upon the application. The current is 
controlled by a rheostat and read on an ammeter. 

Potential Measurement. The potential V is measured either by a potentio- 
metric circuit consisting of a 1%-volt battery, dropping resistor, voltage divider 
and null galvanometer; or more simply (but less accurately) by measuring the 
deflection produced on a moving-coil galvanometer. With either system, poten 
tials of about 30 to 100 pv. must be measured. Readings which are proportional 
to / and JP 7 are quite adequate, since changes in the V/I ratio are always used in 
determining thicknesses or crack depth. 


Selection of Components. It is difficult to assign fixed values to the electrical 
components in the current circuit of Fig. 1 because they depend somewhat upon 
the^ spacing and the material. However, once the electrode geometry has been 
decided upon from considerations of sensitivity and range, the current required 
to produce 30 pv. can be calculated, and the rheostat and battery values can 
be found. In the potential circuit in Fig. 1, values are assigned to the components 
which are satisfactory for measuring 30 \LV. or greater. Instruments can be 
designed to cover a wide range of commonly occurring values and can also be 
made to read directly in inches of wall thickness or crack depth by calibrating 
either the potential-measuring device (constant-current method) or the ammeter 
(constant-potential method). 

steps given here are used in advance of each calibration and test measurement 
described later. 

Electrode Spacing. Select a suitable electrode spacing for the particular 
problem at hand (see Guide to Proper Application of Test Method, in this 
section) . 

Surface Preparation. Prepare the surface of calibration and test specimens 
so that electrical contact will be made at all electrodes. Freshly machined 
surfaces need no preparation, and mildly oxidized surfaces can be prepared 
easily by rubbing with emery cloth. For corroded or scaled surfaces a con 
venient method is to scribe a heavy mark on the surface in the line where the 
electrodes are to be placed. Punch marks at each electrode have also been used. 

Electrode Contact. Apply the electrodes to the surface and make a steady 
contact. If the electrode head is hand-held, it should be designed so that the 
head itself can rest against the metal as the electrodes are pushed upward. A 
magnet attached to the head is useful for magnetic materials. Clamping or 
weighting of the electrode head may also be used. 

wall thickness, after the preparatory steps proceed as follows: 

1. Energize the potential circuit, keeping the current circuit open. Balance out 
the thermal e.m.f. produced between potential electrodes Pi and P z or possibly 
from other points in the potential circuit. This can be done either with a 
bucking-out circuit or more simply by adjusting the zero on the galvanometer. 

2. Apply a current 7 and measure V, the potential produced by this current flow 
(or some reading proportional to V) on the calibration specimen of known 
thickness and the same resistivity as the specimen. 

3. Repeat measurement steps 1 and 2 on the specimen of unknown thickness, 
using either the same current as in the calibration procedure (constant current 
method), or the same potential (constant potential method). 

4. Obtain the thickness from the ratio Vtest/Vcai. (or 7 te8 t//oai.) and a cali 
bration chart (see Interpretation of Test Indications, in this section). 

A variation in this method 8 utilizes six electrodes. It permits measurement of 
wall thickness on material of unknown resistivity. A calibration plate of known 
thickness is not required. 

eated either visually or by some other method such as magnetic-particle or liquid- 
penetrant inspection. Two surface areas, one over the crack and the other 


remote from the crack, are prepared as described previously. Measurement 
steps are: 

1. Apply the electrodes and make a steady contact on the surface of sound metal 
remote from the crack on the specimen to be tested. For deep cracks the 
nearest electrode should be at least five times Ihe electrode spacing A away 
from the crack. Balance out the thermal e.rni. as described in step 1 unrtrr 
Wall Thickness Measurement Procedure, in this section. 

2. Apply a current 7 and measure V, the potential in this remote calibration area. 

3. Place the electrode array centrally over the crack and at right angles to it, 
balance out the thermal e.m.f., and measure I or V, using cither tho same 
current or the same potential as before. 

4. Obtain the crack depth from the Fcrack/Fnormm (V<,/V n ) ratio or the 
-Zorack//normai ratio, or from other observed geometry such as the thickness of 
the part or the calibration data (see Crack Depth Indications, in this section). 

Interpretation of Test Indications 

FLAT PLATE THICKNESS. The simplest case of interpretation, where 
this method is most usefully employed, is that of measurements on flat plates 
having lateral dimensions about ten times the thickness or greater. The analytical 
solution to the problem has been established, and calibration curves can be 
readily obtained for this case. 

Electrode Arrangement. In the schematic diagram in Fig. 2, four electrodes 
are placed in line, contacting an infinite slab of homogeneous, isotropio material 
having a resistivity p (ohm-centimeter) and a thickness T (centimeter). The 
potential electrodes P l and ? 2 are separated by a distance B (centimeter), while 
spacings CJP^ and P 2 C 2 are both equal to A (centimeter) . 

Thickness-Potential Relation. If a current / (amperes) is passed through 
the material between C x and C 2 , and the potential V (volts) is measured between 
PI and P 2 , it can be shown that 


Here M is a function calculated and tabulated by Uhlir. 2 The above relation is 
plotted in Fig. 2 with K f as ordinate, T/A as abscissa, and B/A as a family 
parameter. For particular electrode spacings other than those plotted in Fig. 2, 
the calibration data can be calculated, using the tables for M given by Uhlir. 2 

Thin-Plate Modification. For thin plates Eq. (2) becomes 

This formula holds within 1 percent for T/A <i 0.5. In this case it can be seen 
that 7/7 is inversely proportional to the thickness T. Utilizing an array such 
that T/A is less than 0.5 simplifies interpretation considerably, since in that case 
Eq. (3) holds. The potential V is inversely proportional to T if the constant 
current method is used, or / is directly proportional to T if the constant poten 
tial method is used. 

Calibration. Knowing the electrode spacings, the pertinent calibration curve 
can be chosen. A reading taken upon a plate of known thickness establishes tho 
scale factor by which readings can be converted to values of K'. Then subsc- 




Dept. of Mines and Technical Surveys (Canada) 
Fig. 2. Calibration curves for case of four-electrode array on an infinite plate. 

quent readings on plates of the same material but of unknown thicknesses can 
be converted to K' values and hence to thickness. 

Edge Corrections. It is not always possible to utilize spacings large enough 
with respect to the thickness to produce the condition T/A ^ 0.5 due to 
Idle effects Fi* 3 (Uhlir*) shows the effect of edge distance on K' of an 
eqSly S ed ic'rode array (B/A = l) for the two cases of (1) the array 
placed f parallel to an edge, and (2) the array placed pe 'Pedicular to an edge^ 
Here L is the distance from the nearest electrode to the edge of the plate The 
fqual spacing between adjacent electrodes in the straight-line array is identified 
as A The ratio of edge distance to electrode spacing is L/A. 


rH rH CO CO 

r-J i-i ^J CC 

" ' 





i>" id 

CO CO "^ OS 
^H ^ O O 

rH i-H QQ CO 



N O3 CO 

rH rH CO CO 





* -* oo co 
o co b- 10 

O O O rH 

rH rH ]> -^i 


W3 TJH 10 O 


ca co cd j> 





smaller corrections are necessary for the array placed perpendicular to an edge. 
For this case L/A should be greater than 5 if no correction is to be made without 
incurring errors greater than 1 percent. 

Temperature Control. Care should be taken to ensure that the test plate 
and the calibration plate are at the same temperature. The resistivity increases 
with temperature and will change the readings accordingly. 

Thickness of Curved Walls. The determination of wall thicknesses can also 
be carried out for other than flat plates. In general the theoretical solution for 
the current flow and potential distribution in these cases is not known. Interpre 
tation must be based on experimentally established calibration curves. It is not 
necessary to use the test material to obtain calibration curves, provided the 
calibration material and the test material obey Ohm's law. 

CRACK DEPTH INDICATIONS. Crack depths are most readily evalu 
ated for surface cracks perpendicular to the exposed surface. 

Long Perpendicular Surface Cracks. The first case to be considered is an 
ideal surface crack of depth D, whose plane is perpendicular to the surface of 
a slab of thickness T. The crack length is assumed long in comparison with 
its depth. Fig. 4 illustrates this case and gives the calibration curves VJV* 
plotted against D/A, with T/A as the family parameter. These curves were 
established experimentally, 1 using mercury as the metal and a bakelite strip as 
the "crack." They can be used to determine the crack depth for this case. 
Equally spaced electrodes are used throughout this analysis. 

'S fo" 20 30 40 50 8-0 7* M 9* 10-0 


Dept. of Alines arid Technical Surveys (Canada I 

Fig. 4. Calibration curves for equally spaced array on "ideal" crack in finite thick 
ness plate. 






3.0 40 


9.0 10.0 1 1.0 

5.0 6,0 7.0 

Dept. of Mines ami Technical Surveys (Canada) 

Fig. 5. Calibration curves for finite-length cracks having rectangular shapes in 

semi-infinite medium. 


8.0 9.0 10.0 11.0 

Dept. of Mines and Technical Surveys (Canada) 

Fig. 6. Calibration curves for finite-length cracks having circular shapes in a 

semi-infinite medium. 


Short Perpendicular Surface Cracks. In practice, many cracks are not long 
in comparison with their depth but have a finite length, and sometimes their 
shapes are irregular. Two cases of finite-length cracks have been investigated: 

1. Rectangular cracks having a constant depth over their entire length. 

2. Cracks of circular shape having their maximum depth D at the center of their 

These two cases, along with the calibration data (also established by the 
bakelite-strip-in-mercury technique), are illustrated in Figs. 5 and 6. In these 
graphs LI A. is the family parameter, and all the results apply to cases where 
material thickness T is large with respect to electrode spacing A. For a given 
value of V c /V n obtained from readings on a finite-length crack, these two graphs 
permit the placing of approximate limits on the actual depth. By assuming the 
existence of a rectangular crack, a smaller depth will be indicated than that for 
a circular crack. 

In practice it is often possible to choose such a spacing A that one will be 
operating on the envelope curve of each of the sets of curves given in Figs. 4, 
5, and 6. This is discussed more fully under Guide to Proper Application of 
Test Method, in this section. 

Other Crack Conditions. The preceding discussion does not strictly apply 
to cracks which are not at right angles to the surface nor to cracks on curved 
surfaces. However, where the angle of inclination is not excessive or where the 
surface curvature is not too great, quantitative results can still be obtained. 
Nonsurfacing cracks can be detected with this technique, but analysis of crack 
depth is not generally possible. 

Guide to Proper Application of Test Method 

WALL THICKNESS MEASUREMENT, The first step in applying the 
direct-current test technique to a specific measuring problem is to choose the 
electrode spacing which covers the range of thicknesses to be encountered and 
which provides sufficient sensitivity to obtain the required accuracies. 

Selecting Electrode Spacing. For best sensitivity, T/A should be less than 
0.5, but unfortunately this sometimes makes A too large to be practical without 
making sizeable corrections due to edge effects. The curves in Fig. 2 and Eq. 
(3) permit calculation of the best spacing for the problem. Another consideration 
is that this method samples an area roughly circular in shape and encompassing 
all the electrodes. 

Selecting Current Magnitude. Having determined the spacing, the current 
required to produce a value for V of at least 30 |iv. should be calculated from 
Eq. (2) or (3). Provision should be made to supply this current with smooth 
control over the required range. Larger currents are required for good conduc 
tors such as copper, aluminum, or brass than for poorer conductors such as 
alloy steel. 

Corrosion Wall-Thinning. As an example, consider the problem of measur 
ing corrosion thinning on mild steel hull plates which are nominally % in. 
thick and whose thicknesses are to be measured from one side only to an ac 
curacy of Vo4 in. or 3 percent. As a trial spacing, assume A = I in. 
(T/A-0.5) and B/A = 1. A comparison of K! calculated from Eq. (3) and 
the value from Eq. (2) shows that they do not differ by 1 percent (2.77 and 


2.7S). Therefore, with this spacing, / would be proportional (o 7' with the con 
stant voltage method. A 3 permit change in thickness would give, a 3 percent 
change in meter reading, which would be quite readable. However, assuming 
a value of p equal to 15 X 10~ (l ohm-cm., it can he shown by Rq. (3) that a 
current of 11.5 amp. is required to produce 1 a potential of 30 |iv. This current 
could be provided, but if a current less than 10 amp. wore desirable, a value of 
2 for B/A would increase K f to 4.40 and reduce the current requirement to 
about 7 amp. The latter choice of electrode spacing would sample a larger area 
but would be subject to larger edge effects. For this spacing, a rheostat pro 
viding smooth control over the range from S to about 4 amp. (assuming a 
reduction in thickness of 50 percent) should be chosen. 

CRACK DEPTH MEASUREMENT. For best results in measuring crack 
depths, the electrode spacing 4 should be chosen so that the minimum crack 
depth of interest produces a value of V,./V H of 1.1 or greater. This can be done 
by reference to Figs. 4, 5, and 6. 

Selecting Electrode Spacing. Except for cracks with small lengths L, the 
spacing A should be about ten times the crack depth. Generally speaking, if 
readings of 7 c /7,, larger than about 2 arc obtained on a crack, Ihe accuracy of 
the result will be lowered, and a larger spacing should be used. 

Selecting Current Magnitude. Once the spacing is established, the current 

1 which will produce 30 ^.v. for V. n can be obtained from Fig. 2 (for the case 
B/A = 1) and the value of p for the material under test. If the material is 
thick, T/A > 3, the current can be found with sufficient accuracy from the 

/ = 2nA - (4) 


Forging Cracks. As an example, consider cracks in forcings whose depths 
must be established to determine the possibility of removing them by machining. 
Suppose that the maximum allowable machining is % in. A convenient electrode 
spacing, A, for this case is %2 in. (0.156 in.), which permits using Vfc-in. drill 
steel for electrodes. For this spacing a crack depth D of % in. produces a value 
of V c /V n of about 1.6, which would be a convenient operating point. If the 
material were an alloy steel with a resistivity p equal to 40 X 10 " r> ohm-cm., 
Eq. (4) shows that a current 7 of 1.9 amp. would produce the* required potential 
of 30 jj,v. for 7 n . If the constant current method is employed, facility to supply 

2 amp. with fine control over a small range (to allow for variations in contact 
resistance) and facilities for measuring potentials up to about 100 j.iv. should 
be provided. The potential circuit shown in Fig. 1 would be satisfactory for 
this purpose. 

This technique can also be used as a crack detector where only small areas 
of a part are suspected of being defective. When used in this manner, it is 
useful to know that an array placed so that only one electrode is over the crack 
will result in a reading for 7 less than 7,,, as has been shown analytically for 
the case of very deep cracks. 1 

TRODE ARRAY. A special application of this method to wall thickness - 
measurements is to place four electrodes in the form of a square instead of in 
line; the two current electrodes are on one side of the square, and the two 
potential electrodes are on the other. 9 The measuring apparatus is essentially 
the same as for the in-line array. 


Sensitivity of Square Array of Electrodes. The sensitivity of this array to 
thickness changes can be calculated from the formula 

Here A is the electrode spacing along the sides of the square. It can be shown 
that for the same value of A, this expression is always less than the value of K r 
given by Eq. (2) ; i.e., the square electrode array is always less sensitive than 
the in-line array. However, it does have a smaller edge-correction factor for 
the same spacing than an equally spaced in-line array. 

CRACK PROPAGATION STUDY. The detection and measurement of 
edge cracks is an application requiring special treatment. A particular example 
is the study of the rate of propagation of cracks in the trailing edges of turbine 
blades during thermal cycling. The location of the cracks can be readily 
determined by using knife-edge electrodes with a small spacing (say, 0.075 in.) 
and by plotting voltage/current readings for consecutive probe positions. Then, 
placing the probe symmetrically over each crack in turn, the depth can be 
determined from the V c /V n ratio and a calibration curve. 

A theoretical curve has been derived for the two-dimensional case, and good 
agreement has been obtained with an experimental calibration using a thin brass 
plate with machined edge cracks. Subsequent cleaning of the blades, followed 
by measurement of the cracks with a traveling microscope, gave results within 
20 percent of those determined from the V c /V n ratios and the curve. The cracks 
were markedly asymmetrical, however, and although the mean of the micro 
scope readings was taken in each case, this was obviously only an approximation. 
It is considered that better agreement would probably be obtained by using 
a calibration made under conditions more closely simulating those of the turbine 

Tests of Railroad Rails 

PRINCIPLE OF TEST. The electric current method of rail testing can be 
described briefly as follows: 

1. A heavy current is passed through the rail to be tested, thus establishing a 
strong magnetic field around the rail. 

2. Searching coils are moved through the established magnetic field at a fixed rate 
and distance above the rail. 

3. Electromotive-force pulses from the coils are recorded. These are the result of 
changes in the magnetic field around the rail at points where internal defects 
cause deflection of the electric current path. 

The Magnetic Field. Whenever a current is passed through an electrical 
conductor, a magnetic field is set up surrounding that conductor. This magnetic 
field can be considered as a region composed of concentric lines of force, each 
of which is perpendicular to the axis of the conductor. The intensity or strength 
of the magnetic field, measured by the degree of concentration of lines of force, 
is greatest at the surface of the conductor and decreases as distance from the 
conductor increases. Fig. 7 (a) is a conventional representation of the magnetic 
field around a sound circular conductor. 

Effect of a Change in Conductor Cross-Section. If the cross-section of 
the circular conductor shown in Fig. 7 (a) is altered to a semi-circle, the mag- 



netic field assumes the shape shown in Fig. 7(b). When the conductor returns 
to its original circular shape, the magnetic field is restored to its original position 
and shape. 

Induced Electromotive Force. If a coil is moved through a magnetic field 
so as to cut a changing number of lines of force, an electromotive force (e.m.f.) 
is induced in the coil and is measurable as a voltage output. If the same coil is 
moved through an unchanging magnetic field at a constant speed, no e.m.f. is 
developed. If a coil moves along the length of a circular conductor at a constant 
speed and constant distance from its center, no e.m.f. is developed until it reaches 
the point where the cross-section diminishes to a semi-circle. At that point 
it would cut a decreasing number of lines of force and an e.m.f. would be in 
duced. The e.m.f. is induced slightly before the coil passes tho actual change in 
cross-section, since the current path is deflected downward. When the con 
ductor again becomes circular in cross-section, an e.m.f. of opposite polarity is 
generated, since the coil then cuts an increasing number of linos of force. 




Fig. 7. Magnetic field surrounding current-carrying conductor, (a) Magnetic field 

around circular conductor, (b) Magnetic field around serai-circular conductor, 

simulating a defect in circular cross-section conductor. 

Magnetic Field Around Sound Rail. In the Sperry system of rail testing, 
current is introduced into the rail head and sets up a magnetic field approxi 
mately as shown in Fig. 8. The field thus established remains constant in both 
strength and shape as long as both (1) the effective cross-section of the rail and 
(2) the test current remain constant. The lines of force pass not only through 
the air around the rail but also through the rail as they seek the path of least 
reluctance in completing their circuits. 

Magnetic Field Around Defective Rail. Any defect in a rail reduces the 
cross-sectional area by an amount equal to the transverse area of that defect. 
This is due to the inability of low voltage current to jump the gap presented 
by the defect. The current is forced into the sound' portions of the rail in 
order to flow around the defect and then fans out once more after passing the 

The change in direction and density of the current as it flows around a defect 
is accompanied by a corresponding change in the direction and position of the 
magnetic field around the defective portion of the rail. Since the magnetic field 
is always perpendicular to the axis of the current path, transverse deflection of 
the current as it flows past a defect causes the magnetic lines of force to bo 
displaced proportionately. 



Searching Coil Reactions. The reaction of searching coils to a change in the 
direction of the magnetic field is dependent to some extent upon their positions 
relative to the magnetic field. This factor permits some coils to move closer to 
a, defect before being affected by the changing magnetic field than others whose 
axes are arranged in a different direction. Similarly, some coils, by virtue of 
their position with respect to the rail, are affected only by magnetic-field changes 
caused by defects in the field side of the rail head; others, only by gage side 
changes; and still others, only by longitudinal defects. The efficiency and 
thoroughness of rail testing is based upon the application of this selective prin 

Lines of Force 

Fig. 8. Magnetic field about a current-carrying railroad rail. 

RAIL TEST EQUIPMENT. The rail detector car developed and used by 
the Sperry Rail Test Service is shown in Fig. 9. Its equipment components and 
their functions are described in the subsequent text. Such cars test about 50 
miles of track per day. 

Main Rail-Current Generator. The main rail-current generator and its ac 
cessories provide a controllable current, possessing the proper characteristics, 
which can be introduced into rails to set up the magnetic field necessary for 

Main Brush Carriages. The main brush carriages provide the means of 
introducing the output of the rail-current generator into the rails to be tested 
(Fig. 10) . The forward cluster of each main brush carriage also provides a path 
for the introduction of a pre-energizing current into the rail. 

Third Brush Carriages. The third brush carriages (Fig. 11) are used to 
provide a return path for the pre-energizing current which was introduced into 
the rail through the leading brushes of the main carriages. In operation, the 
third brush carriages are held against the rail by a pneumatic pressure system. 
These carriages operate in unison with the corresponding main brush carriages 
(see Fig. 12) . 



Sporr.v Hail Test Service 

Fig. 9. Rail detector car. 

Sperry Rail Test Service 
Fig. 10. Main brush carriage of rail detector car. 



Sperry Rail Test Service 
Fig. 11. Third-brush carriage which supplies pre-energizing current. 

FTo Motor-Generator Set 

llOv. DC 1 

Bus Bar |^. 




1 ' 






p^ATo Left Side Main 
[ GENERATOR ^^ Brush Carriage 
PER) iL -- From Left Side Main 
i^^ Brush Carriage 

JT6 Left Side Third 
Brush Carrage 


r (UP 







1 1 




1 1 


1 1 



1 T L- 



Sperry Rail Test Service 

Fig. 12. System for providing magnetizing current to main brush carriage and to 
third brush carriage (pre-energizing current) of rail detector car. 



Detector Unit Carriage. The function of the detector carriage is to carry 
the searching units at a fixed distance and in a fixed position above the running; 
surface of the rail. Means are provided for adjusting this distance when neces 
sary. One detector unit carriage is suspended from each main brush carriage. 

Searching Units. The function of each searching unit is to produce an 
electrical impulse or signal every time it passes over a defect in the rail being 
tested. This defect signal initiates operation of the indicating unit* of the 
detection equipment. Each searching unit contains several sets of coils arranged 
in pairs and positioned effectively to cover the entire magnetic field about the 
rail. The axis and winding of the various coils are designed so that a defect will 
be detected by at least one set of coils. This arrangement permits preliminary 
interpretation, or classification, of the indicated defect by consideration of the 
coils involved in initiating the pen response, 

Power Supplies. An engine-driven generator provides a source of filtered 
alternating current at a constant 115 volts for the detection equipment. Power 
packs provide the voltages necessary for the proper operation of the amplifiers, 
relays, and other detection equipment. 

Amplifiers. Amplifiers increase the signal strength from searching unit coils 
to operate relays controlling the indicating units of the detector equipment. The 
amplifiers operate conventionally except that selective filtering action is used 
to improve response. The amplifier outputs are fed to a series of sea-Hive relays 
which in turn actuate pens on the operator's recording tape. These ivlays and 

Sperry Rail Test Service 
Fig. 13. Rail detector car multiple-channel tape-recording unit. 



associated circuitry are interconnected so that improper functioning of any 
portion of the detection circuit will be indicated on the record tape. Sensitivity 
of the amplifier is normally set to respond to signals caused by defects having 
an effective transverse area of 10 percent or more of the cross-sectional area of 
the rail head. 

Pen Unit. The pen unit provides a visible indication on the detector car tape 
of signals from the various coils of the searching units. One pen provides a 
visible warning when the rail current drops below the operating standard. A 
land mark pen is used to inscribe the land mark code and a time interval 
marker. The indicating pens for both sides of the track are grouped together 
in a single pen unit mounted on a recording table (Fig. 13). 

Detector Car Tape. The detector car tape provides a visual indication of 
defects and rail conditions found by the detection equipment at the time of test. 
This facilitates interpretation by the operator of the pen indications and rail 
surface. The table which supports the tape, tape drive, and pen unit is located 
in front of the middle rear window in the rear compartment of the rail test car 
(Fig. 13). The unit consists of a working surface, tape guide, pen unit mounting 
ami tape drive mechanism. The tape is driven by a power take-off from the 
rear axle. A Via-in. movement of the tape corresponds to 1-ft. movement along 
the track. 

Sperry Rail Test Service 

Fig. 14. Hand test equipment for determining exact location, size, and type of 

rail defects. 





(c) Sperry Rail Test Service 

Fig. IS. (Continued.) (c) Engine burn fracture. 

Paint Guns. Paint guns provide a visible indication on the inside of the rail 
web at any point where the magnetic field is distorted sufficiently to cause an 
indication on the tape. The paint marks are used to locate suspected sections 
of rails when interpreting the cause of the indication. Two paint guns are 
mounted on each brush carriage adjacent to the front end of the searching unit 
and point at the center of the web on the inside of the rail. Each gun is con 
nected by piping to a pressurized paint tank. 

Hand Test Equipment. Hand test equipment provides a means of deter 
mining the exact location, size, and type of defects. Hand testing is based on 
the potential rise principle rather than on the magnetic induction principles 
discussed earlier. 10 The hand contactor is placed on the rail close to the paint 
mark indicating the location of a possible defect (Fig. 14). If a transverse 
separation is present, the deflection of the current will cause an increased differ 
ence in potential to exist between the two sides of the separation. When the 
hand contactor is moved along the rail and its two spring contacts enter the 


area directly above the separation, the rise in potential at that point (directly 
above the defect) will cause a corresponding rise in the millivoltmeter reading. 
The rise confirms the presence of the defect. The amount of rise indicates the 
size of the defect. The rate of rise indicates the general type of defect. By 
the use of this equipment, defects are finally and accurately pinpointed to within 
% in. 

PERSONNEL REQUIREMENTS. The standard (Hector car crew con 
sists of a chief operator assisted by two operators. The chief operator is in full 
charge of the car and is responsible, through a supervisor in a specified territory, 
to the operating manager. 

During testing, there are three operating stations manned by (1) a tape 
operator, (2) an examining operator, and (3) a, driving operator. Testing speed 
varies from 6 to 12 miles per hour, depending on rail conditions. Potentially 
dangerous areas are indicated on a continuously moving detector car tape ami 
may require verification by the tape operator. 

INTERPRETATION OF INDICATIONS. In both new and old rail the 
extent and development of progressive fractures from microscopic irregular 
ities is of great concern. The location anil size of these defects determine the 
capacity of the rail to withstand the stresses imposed upon it in service. Each 
year, approximately 150,000 miles of track are tested by Sperry induction cars, 
which locate approximately 50,000 defects. Since 192S this method has been used 
for inspection of more than 3 million miles of rail, equivalent to 000 million 
tons of steel. More than l l /z million defective rails, equivalent to 2 million tons 
of steel, have been marked for removal. 

Typical Defects Detected. Typical rail defects detected are shown in Fig. 
15. More than half the defects detected are classified as transverse defects, and 
one-third as vertical split heads; the balance are horizontal split heads and 

Interpretation Procedures. One problem that confronts the interpreter is to 
distinguish indications caused by harmless surface defects or track structures 
from those caused by. dangerous internal separations or defects in the rail. Three 
procedural steps assist him in making his decision: 

1. Visual examination from the curs of tho condition of the rail in the immediate 
vicinity of the paint mark which corresponds to the pen indication to bo 

2. Ground examination of the rail at the point, whore the indication occurred. 
This method becomes necessary when tho indication cannol lu verified from 
the car. The pen indication provides clue as to what to look for and where 
to look for it. By examining the rail at close range, external signs of an 
internal defect may often be found and tho cause of the indication thus verified. 

3. Hand testing must be used when tho visual inspections fail to roveul the cause 
of the indication. Hand testing is also used when the operator doubts that a, 
visible surface defect is the true cause of the indication. In many cases an 
internal defect has been found directly below the surface defect. 

Classification of Defects. After an indication has been verified as being the 
result of an internal defect, it is necessary to decide what kind ami what size 
defect is present. This process is called "classification." In classification, as in 
interpretation, the operator must make use of a number of sources of information 
to determine the type of defect that has been Found. 


pen indication gives the first clue to the nature of the defect. Although 
all defects of a given type will not cause identical pen indications, the operator 
will soon learn to associate certain combinations of pen responses with certain 

The second and most important factor used in classifying a defect is the 
visual appearance of the rail, seen either from the car or" from the ground, at 
the point where the indication occurred. With the exception of transverse defects 
that have not progressed to the surface, most defects develop certain recog 
nizable characteristics. Experience is the best teacher in developing the ability 
to recognize such visible characteristics. 

Hand testing is the third factor in determining the proper classification of 
a verified defect. With the exception of longitudinal defects, each general type 
of defect will produce different meter reactions when a hand test is made. By 
carefully observing the meter reaction in relation to the position of the hand 
contactor on the rail, the operator can classify the defect as to size. A hand 
test, made to classify a visible defect, provides more exact knowledge of the size 
and location of that defect. 

Marking Defects in Rail. Each rail containing a verified internal defect is 
marked by yellow crayon with the following information for the guidance of the 
contracting railroad's maintenance personnel: 

1. Type or types of defects detected. 

2. Estimated size of transverse separation, if present (except in the case of engine 
burn fractures). 

3. Serial number of defect. If two or more defects are found in one length of rail, 
the serial number is marked below the most serious defect. 

4. Location or longitudinal extent of the defect. 


1. BUCHANAN, J. G., and R. C. A. THURSTON. "The Measurement of Crack Depths 

by the Direct-Current Conduction Method," Nondestructive Testing, 14, No. 5 
(1956): 36,43. 

2. UHLIR, A. ''The Potentials of Infinite Systems of Sources and Numerical Solu 

tions of Problems in Semiconductor Engineering," Bell System Tech. J., 34 
(1955): 105. 

3. JACKSON, L. R., H. M. BANTA, R. C. MC-MASTER, and T. P. NORDIN. "Electric 

Current Conduction Tests, Fifth Progress Report on Nondestructive Testing of 
Drill Pipe," Drilling Contractor (August, 1948) : 68. 

4. ARMOUR, A. M. "Eddy Current and Electrical Methods of Crack Detection," 

J.Sci.Imtr.,25 (1948): 209. 

5. HIRST, G. W. C. "Apparatus for an Electrical Determination of the Depth of 

Transverse Cracks or Fissures in Magnetizable Circular Prisms," J. Inst. 
Engrs. Australia, 19 (1947) : 145. 

6. THORNTON, B. M. "The Detection of Cracks in Castings by an Electrical 

Method," Foundry Trade J., 68 (1942): 277. 

7. THORNTON, B. M., and W. M. THORNTON. ''On Testing the Wall Thickness of 

Castings," Foundry Trade J., 65 (1941) : 253. 
8 . "The Measurement of the Thickness of Metal Walls From One Surface 

Only, by an Electrical Method," Proc. Inst. Meek. Engrs., 140 (1938): 349. 
9. SPERRY, E* A. (To Sperry Development Co.) ''Fissure Detector for Metals." U.S. 

Patent No. 1, 820, 505, (Aug. 25, 1931). 
10. . (To Sperry Products, Inc.) "Fissure Detector for Steel Rails." U.S. 

Patent No. 1, 804, 380, (May 5, 1931). 



BARNES, W. C., and H. W. KEEVIL. "Method and Apparatus for Detecting Flaws in 
Rails." U.S. Patent No. 2, 410, 803, (Nov. 12, 1946). 

CAMPBELL, J. E., H. 0. MclNTiRE, and G. K. Manning. "Summary Report on Iho 
Examination of Rails Which Contain Detail Fractures to Joint Contact Com 
mittee on Rails of American Railroads and American Iron and Steel Institute and 
Committee on Rails of American Railway Engineering Association," Am. /?//. Eng. 
Assoc.BulL, 51 (1950): 608. 

CRAMER, R. E. "Investigation of Failures in Railroad Rails," Am, ft//, Kng. Amw. 
Bull., 51 (1950) : 543. 

FRICKEY, R. E., and C. W. McKEE, (To Welding Service, Inc,) "Method and Appara 
tus for Detecting Flaws in Rails." U. S. Patent No. 2, 388, 683, (Nov. 13, 1945). 

KEEVIL, H. W. "Method and Apparatus for Detecting Flaws in Metallic Bodies. 11 
U.S. Patent No. 2, 089, 967, (Aug. 17, 1937). 

MANNING, G. K. "Summary Report on the Examination of Shelled Rails to Joint 
Contact Committee on Rails of Association of American Railroads and American 
Iron and Steel Institute," Am. Ry. Eng. Amic. Butt., 50 (1949) : 542. 

McBHiAN, R. 3 C. J. CODE, C. B, BRONSON, G. F. HAND, J. G. RONKY, and A. A. 
SCHILLANDER. "Rail Failure Statistics, 13 Am. Ry. Eng, Awc. Bull., 51 (1950) : 550. 

Rail Defect Manual 4- Danbury, Conn.: Sperry Rail Test Service, 1940. 

SPERRY, E. A. "Non-Destructive Detection of Flaws," Iron Age, 122 (1928) : 1214. 






Basic Principles 

Principle of test. 

Typical arrangements of probe coil and 

test object (/. 1) 

Test coil characteristics 

Representation of test coil characteristics 

on impedance plane (/. 2) 

Development of basic principles 

Summary of eddy current test problems 
solved by Institut Dr. Foerster (/. 3) .. 

Effective Permeability 

Assumptions used in analysis ................. 9 

Determination of effective permeability ...... 10 

Schematic drawing of test object in "feed- 
through" test coil with primary (excita 
tion) and secondary (pick-up) windings 
(M) .................................... JO 

Empty coil voltage ......................... 10 

Influence of test object ..................... 11 

Schematic representation of magnetic-field 
conditions with test object in test coil 

Secondary coil voltage with test object ..... 11 
Simplified representation of magnetic con- 


ditions within test object in test coil 

(/-6) .................................... 12 

General case of a cylindrical test object .... 12 

Limit frequency, fg ......................... 13 

Effective permeability as a function of fre 

quency ratio ............................ 13 

Variation in effective permeability (/. 7) . . 14 
Tabular listing of values of effective per 

meability (/. 8) ......................... 15 

Sample calculations for effective permeabil 

ity and secondary coil voltage ............ 13 

Test object smaller than secondary coil diam 

eter ..................................... 16 

Test coil containing a test object of 

smaller diameter, d (/. 9) .............. 17 

Secondary coil voltages ..................... 16 

Complex permeability plane ................ 17 

Parameter as a function of complex per 

meability (/. 10) ................ . ....... 18 

Correction for ferromagnetic test object ... 19 

Impedance characteristics of a single test coil 19 
Arrangement of cylindrical test object in a 

single test coil (/. 11) .................. 19 

Normalized coil characteristics .............. 20 

Example of nonferromagnetic test object .. 20 
Three complex planes for eddy- current test 

data ....................................... 20 

References ..................................... 21 




Basic Principles 

PRINCIPLE OF TEST. In many eddy current (or electromagnetic induc 
tion tests), the test object is placed in the varying magnetic field of a coil or an 
array of conductors carrying an alternating current (a.c.). The a.-c. magnetic 
field induces eddy currents in the test object. These eddy currents, in turn, 
produce an additional a.-c. magnetic field in the vicinity of the test object. Fig. 
l(a) illustrates conditions for a coil placed on the surface of a test object. The 
vector H p represents the primary a.-c. field of the test coil, whereas H 8 indi 
cates the secondary a.-c. field resulting from eddy currents in the test object. 
Fig. l(b) represents conditions for the arrangement in which the test coil sur 
rounds the test object. In each case two a.-c. magnetic fields are superimposed. 
The magnetic field near the test coil is modified if a test object is present. Other 
arrangements of eddy current test coils are summarized in Fig. 3 of this section. 
Coil shapes include: solenoid coils, inside coils, hand-probe coils, fork-shaped coils, 
and special coil shapes for specific test objects. (The illustrations in this section 
and those following on eddy current tests are from the Institut Dr. Foerster, 
except where otherwise indicated.) 

Test Coil Characteristics. In general the test coil is characterized by two 
electrical impedance quantities: 

1. The inductive reactance X L = 2jt/L, where / is the frequency of the a.-c. field 
in cycles per second (c.p.s.) and L is the self-inductance of the coil. 

2. The ohmic resistance R. 

It is common practice to plot the reactance X L as ordinate and resistance R as 
abscissa in the impedance plane. In this way the test coil impedance Z is repre 
sented by a point P, formed by two perpendicular components, X L = 2jt/I/ = col/, 
and R, on the impedance plane. In the absence of a test object, the empty test 
coil has a characteristic impedance with coordinates X Lo = ooL and RQ, shown 
on the impedance plane by the empty point P of Fig. 2. If a test object is 
placed in the field of the test coil, the original field of the empty test coil is 
modified by the superimposed field of the eddy currents. This field modification 
has exactly the same effect as would be obtained if the characteristics of the 
test coil itself had been changed. The influence of the test object can be de 
scribed by a variation in test coil characteristics. The apparent impedance of 
the "empty" coil represented by P is displaced to P l (corresponding to new 
values of reactance col/i, and resistance RI) under the influence of the test 
object (see Fig 2). 

The magnitude and direction of the displacement of the apparent impedance 
from P to P! under the influence of the test object are functions of the properties 




(G.D.ju, cracks) 

IiiHtitnt Dr. Koorstor 

Fig. 1. Typical arrangements of probe coil and test object. (:0 Prohr coil above 
flat surface of a metallic teat object, (b) Tost objrot, within M cylindrical lost coil. 

H p = primary field of coil in absence of test object. 

H 8 = secondary field created by eddy currents in test, object. 

of the test object and of the characteristics of the instrumentation. Significant 
properties of the test object include: 

1. Electrical conductivity (tf). 

2. Dimensions (such as diameter of rods). 

3. Magnetic permeability (n). 

4. Presence of discontinuities such as cracks or cavities. 

Significant instrument characteristics include: 

1. Frequency of the a.-c. field of the teat coil. 

2. Size and shape of the test coil. 



3. Distance of test coil from test object (coupling between the coil and the speci 

The influence of several physical properties of the test specimen upon the 
impedance characteristics of the probe coil can be calculated for various test 
frequencies in certain cases. Often it is possible to determine from the im 
pedance change (distance P to P-J, and frequently to measure quantitatively 



^ mmmmmm ^ if 



v >v 

/.)/ 4, 



) *"/ 


H * 














Fig 2. Representation of test coil characteristics on impedance plane -Point 
P, , = impedance of coil in absence of test object. Point Pi = impedance of test coil 

with test object. 

and independently of each other, not only the conductivity, dimensions, and 
magnetic permeability of the test object but even the magnitude and direction 
of cracks The optimum frequency for a specific test problem is determined by 
theory or by experiment and provides the highest sensitivity obtainable with 
eddy current methods for detection of: 

1 Variations in electrical conductivity, used in sorting alloys. 

2! Variations in test-object diameter, used in dimensional control. 

3. Cracks. 

Indications are obtained, without electrical contact with the test object, in ex- 
t^AoA time intervals, such as 1/1000 sec Physical propert -s measured 
include, alloy, heat treatment, hardness, depth of case, surface decarbunzation, 
magnitude of defects, and dimensions. 



Methods of Testing of Solid Cylindrical Parts 



Test Arrangement 


Test Conditions 
Determined by 


Solenoid Coil 



/7V777 m 

i i 








Solenoid Coil 

Shape. Size . 
Depth. Length. 
Position of 

Model Test 

i////// n 

i i - 


Coil With 
Probe Coils 


Model Test 

Two - Layer 

Solenoid Coil 

of Each Layer 


fun/ & 

t. i 


Fig. 3. Summary of eddy current test problems solved by Institut Dr. Foerster. 



Application, and Equipment For Tests of Solid CylinJri,,,! P,, rt , 


Shape of Coil 



Instrument Designation 


American Type 


Solenoid Coil 

Quality control, 
classification for 
ttnsile strength 
and dimensions 

Iron and 

Magnatest Q 

Magnatest FS 300 




Magnatest FW300 
Magnatest FS 200 




Quality control 




Solenoid Coil 


Iron and steel 

Magnatest D 



Bar and 

Magnatest FW 200 
Magnatest FW400 
Magnatest FS 200 



Coil With 
Probe Coils 

Defect detection, 
cracks, overlaps, 
soft spots 

Metals and 



Two- Layer 

Soltnoid Coil 

Plating thickness, 
case depth, 



Magnatest FS 200 



Magnatest Q 

Magnatest FS300 
Magnatest FS 200 



Applications and Equipment For Tests of Hollow Tubular Parts 


Shape of Coil 



Instrument Designation 


European Type 


Solenoid Coil 

Quality control, 
classification for 
alloy components, 
conductivity test, 
hardness and 
tensile strength 



Magnatest FW300 
Magnatest FS 200 



Magnatest Q 

Magnatest FS300 
Magnatest FS200 



Solenoid Coil 



Bar and 

Magnatest FW 200 
Magnatest FW400 
Magnatest FS 200 



Magnatest D 

Magnatest FS200 


Two- Layer 

Solenoid Coil 

Plating thickness, 







Magnatest FS 200 


of Wall 

Solenoid Coil 




Magnatest FS 200 
Magnatest FW300 



Magnates> Q 

Magnatest FS200 
Magnatest FS300 



Inside Coil 

Inner Defects 



Magnatest FW300 
Magnatest FS 200 


(Reference numbers refer to references listed in section on Eddy Current Test Indications.) 



Methods of Testing of Spherical and Ellipsoidal Parts 


Shape of Coil 

Test Arrangement 


Test Conditions 
Determined by 

Ellipsoid of 

Solenoid Coil 




Ball of 

Solenoid Coil 




Ball with 

Solenoid Coil 



Model Test 

Ball with 



Modtl Test 



. = 


Methods of Testing of Sheets and Plates 


Shape of Coi 

Test Arrangement 


Test Conditions 
Determined by 


Hand Probe 
Coil on One 
Side of 

, ^ 

E[ , 


Model Test 



Coils on Both 
Sides of 





I ^ 

Two -Layer 
With Spacing 

Hand Probe 
Coil on One 
Side of 

i kl_j 

of Each Layer, 
Between Layers 

Model Test 

^^MiimHim & 


Hand Probe 
Coil on One 
Side of 


a. <sr, 

Plating Thickness. 
of Each 

Model Test 

mm <? 2 

Plating on 

Hand Probe 
Coil on One 
Side of 


d, *, 

Plating Thickness, 
of Each 

Model Test 

OTram,^W3JTO ^ 

With Defects 

Hand Probe 
Coi 1 on One 
Side of 


i >^ __ ^_^_ 


Depth of cracks, 
Shrinkage , 


Model Test 




Fig. 3. (Continued.) 



Applications and Equipment For Tests of Spherical and Ellipsoidal Parts 


Shape of Coil 



Instrument Designation 


European Type 

American Type 

Ellipsoid of 

Solenoid Coil 

Quality control 
short pieces 



Ball of 

Solenoid Coil 

Quality control 
(Alloy content, 

Steel and 


Magnatest FS 200 


Ball with 

Solenoid Coil 

Defect detection 

Steel and 




Ball with 


Defect detection, 
soft spots 


Crack Detector 
for Balls 



Applications and Equipment For Tests of Sheets and Plates 


Shape of Coil 



Instrument Designation 


European Type 

American Type 


Hand Probe 
Coil on One 
Side of 

Quality control, 
conductivity test, 
alloy composition, 
hardness, porosity, 
wall thickness 



Magnatest FM 100 



Magnatest Q 

Magnatest FS300 



Fork- Shaped 
Coils on Both 
Sides of 

Thickness meas 
urement of foils 
sheet and strips 
square resistivity] 


Sheet and foil 
Square Resistivity 

Magnates* FT 200 
Magnatest FT 100 


Two- Layer 
With Spacing 

Hand Probe 
Coil on One 
Side of 

Separation of 
Layers , 
Spoiled Coatings 






Hand Probe 
Coil on One 
Side of 

Thickness meas 
urement of con- 
layers on 
nonferrous base 



Magnatest FT600 


Plating on 

Hand Probe 
Coil on One 
Side of 

Thickness meas 
urement of metal 
coating on iron or 
steel (tin layers 
on sheets) 

Tin Plate 


With Defects 


Defect detection, 
cracks, shrink 
ages, porosity 



Magnatest FM 100 



(Reference numbers refer to references listed in section on Eddy Current Test Indications.) 

36 8 


Methods of Testing of Plates With Insulating Layers 



Test Arrangement 


Test Conditions 
Determined by 

Coating on 
Non ferrous 

Hand Proba 
Sid. of 

* kl . 


Model Tost 

t <jmm3S3m*i 

Non ferrous 
on Insulating 

Hand Probe 
Coil on One 
Side of 

; W 1 *, 

of Metallic 



Fig. 3. (Continued.) 

Because of its versatility, the eddy current test method is finding increased 
industrial use, particularly in fully automatic tent units. Some Ion million parts 
such as rods, tubes, and automotive parts are tested daily in Europe by this 
method at the present time. This increase in use, particularly in recent years, 
is based upon extensive theoretical and experimental developments which' have 
established the fundamental principles underlying this test method. These prin 
ciples indicate the capabilities and limitations, as well as the practical applications 
of the method, both for existing and new test problems. 

derlying eddy current test methods were developed in two ways. First, those 
test problems amenable to mathematical analysis were calculated quantitatively 
on the basis of reasonable assumptions concerning test-object shape and material 
properties. Cases included in this group, for example, were: 

1. The cylinder with single and multiple layers. 48 

2. The tube with single and multiple lavcrs! 48 

3. The sphere. 39 

4. The rotation ellipsoid. 

5. The probe coil inside the tube. 

6. The pick-up coil for testing thin metallic layers. 33 

7. The fork coil used for noncontacting thickness measurement, or Iho measure 
ment of resistance per unit square, for metal foils and lamina lionn. :iB 

8. The cylinder within a fork coil (with the direction of the fiold perpendicular to 
the axis of the cylinder). 

The first four items in the preceding list refer to the case of a test object within 
a cylindrical probe coil (a "feed-through" coil). 

In the second group of test problems, exact mathematical solutions of eddy 
current effects were impossible because of the boundary conditions involved 
However, a similarity law applicable to eddy current problems make* it feasible 
to obtain quantitative information by means of measurements on models These 
model test data can then be applied to practical test objects whoso dimensions 
electrical conductivity, and relative magnetic permeability are given. Included' 
in tms group are cases such as: 

ith craekSi cavitiesi * othcr 

2. The eccentric tube. 



Applications and Equipment For Tests of Plates with Insulating Layers 


Shape of Coil 



Instrument Designation 


European Type 

American Type 

Coating on 

Hand Probe 
Coil on One 
Side of 

Thickness meas 
urement of insu 
lated coating on 
nonferrous metals 






on Insulating 


Hand Probe 
Coil on One 
Side of 

Thickness meas 
urement of con 
ductive nonferro- 
magnetic coatings 
(square resistivity] 


Equivalent silver 
coating -thickness 


(j) Institut Dr. Foerster 

(Reference numbers refer to references listed in section on Eddy Current Test Indications.) 

3. The sphere containing cracks. 

4. Pick-up coil tests of thick metal plates with variations in material electrical 
conductivity, plate thickness, and distance of the coil from the plate surface. 30 

5. Pick-up coil tests of crack depth and location in plate or cylinder. 30 

The first three items in the list refer to the case of a test object within a concen 
tric coil ("feed-through" coil). 

The typical problems solved are shown schematically with their test arrange 
ments in Fig. 3 (a)-(j). Numerous other nondestructive test problems can be 
related to these standard cases which have already been evaluated. 

Effective Permeability 

ASSUMPTIONS USED IN ANALYSIS. The concept of effective perme 
ability considerably simplifies the analysis of eddy current test arrangements. It 
will be illustrated by the example of a long, cylindrical test object magnetized 
within a cylindrical coil carrying alternating current. The following assumptions 
are made for this example: 

1. The test coil produces an essentially uniform magnetizing field, H 0) throughout 
the cross-section within it. 

2. The test object is infinitely long, i.e., end effects are neglected. 

3. The test object is cylindrical in shape, of constant diameter d. 

4. The electrical conductivity a and the magnetic permeability M- of the test 
object are constant, both throughout its volume and with varying states of 

5. Initially it is assumed that the test object fills the test coil. Later, the case oi 
smaller test objects which incompletely fill the coil are considered. 

6. The varying states of magnetization, resulting from the shielding effects of 
eddy currents which oppose the magnetizing field, can be represented by means 
of an "effective permeability," u C ff., which compensates for varying conditions 
within the test object. 

7. The magnetizing field is assumed sinusoidal and of a single frequency. Ine 
effects of harmonic frequency components generated by nonlinear character 
istics of the test object are ignored. Thus the test coil can be Characterized by 
"impedances" similar to those used in analysis of linear a.-c. circuits. 

8. The effects of magnetic hysteresis, shown to be of small magnitude in compari 
son with the effects of eddy currents in the test object, are neglected. 



trates schematically the most widely used arrangement for eddy current testing. 
A primary coil, carrying an a.-c. current of frequency /. generates the primary 
a.-c. magnetic field # . The secondary coil is located inside the primary coil. 
The secondary coil is characterized by its number of turns, w, and the average 
area of the opening within the winding, jt0 a A where D w the diameter of the 
secondary coil. The material to be tested, such as a wire, tube, rod, or equivalent, 
passes through the center of the concentric primary and secondary windings. 


Institut Dr. Focrster 

Fig. 4. Schematic drawing of test object in "feed-through" test coil with primary 
(excitation) and secondary (pick-up) windings. 

Empty Coil Voltage. The basic problem in eddy current testing is to obtain 
valid information concerning the test material from the secondary coil voltage, 
$ seCi . The method becomes quantitative and useful only by exact analysis of 
how test-object conditions influence the secondary voltage, so that physical prop 
erties can be determined from test indications. In the absence of a test object, 
the voltage appearing across the terminals of the secondary coil is calculated 
to be: 

ec. = (2jtf)(n) (n r oi.)#o X 10- 8 volts 


where / = test excitation frequency, c.p.s. 
jtD 2 /4 = area within secondary coil. 
D = diameter of secondary coil. 
n = number of turns in secondary coil, 

Urn. = relative magnetic permeability (approximately 100 for iron, Fo, oqinil to 
unity for aluminum, copper, and other nonforromagnetic metals and 
alloys) . 
Ho = excitation field strength inside the test coil, in oersteds. 


36 11 

In the absence of a test object, the density of magnetic lines of force measured 
in gausses in the air within the test coil is numerically equal to the exciting field 
strength # , measured in oersteds, of the primary coil. 

Influence of Test Object. If a test object such as a metal rod is inserted 
within the coil, eddy currents arc induced within it by the primary a.-c. field. 
The test object acts like a short-circuited winding of an air-core transformer. 
The eddy currents, in turn, generate a magnetic field which tends to weaken the 
primary magnetic field H a within the test object. The resultant field-weakening 
decreases the secondary coil voltage g ec ., in accordance with Eq. (1). 

The weakening of field strength within the test object by the action of the eddy 
currents can be described in two ways. The first, which corresponds to physical 
reality, is illustrated schematically in Fig. 5. The field strength outside the test 
object corresponds to the field H resulting from the primary coil (not shown). 





Fig. 5. Schematic representation of magnetic-field conditions with test object in 

test coil. 

Within the test object, the field strength H t is weakened by the effect of the eddy 
currents and decreases in intensity toward the center as a consequence of skin 
effect. An additional effect (not shown in Fig. 5) is a phase displacement of the 
a.-c. magnetic-field strength associated with this decreased field intensity. This 
phase displacement (in time) occurs to an increasing extent toward the center 
of the test object. 

Secondary Coil Voltage with Test Object. To calculate the secondary coil 
voltage, # sec , by Eq. (1) would require that the test object be divided into an 
infinite number of cylindrical elements, each of area dF and field strength Hi 
(which would vary from area element to area element both in magnitude and 
phase angle). The secondary voltage # sec ., would be calculated from the summa- 



tion of the products of the area and the actual field strength within each cylin 
drical element, as follows: 

*.I. X 10- 8 1 H*d f (2) 

where H v = actual magnetic-field strength within element r. 
d v = cross-sectional area of element v. 

As an alternative to this complex representation, the tost, condition can be 
indicated more simply as shown in Fig. 6. Here it is assumed that the applied 



Fig. 6. Simplified representation of magnetic conditions within test object in test 
coil, assuming an effective permeability less than unity and a constant field 

strength #. 

field strength H is uniform and undisturbed over the entire cross-section of the 
test object. However, it is assumed that the test object can bo assigned an 
effective permeability, u, ff , which is (1) constant over the entire ero^-section 
of the test object, and (2) a complex number with a magnitude .mailer than 
unity. The computed secondary voltage sec , is the same whether one uses the 
summation of area elements of Eq. (2) or the assumed constant field strength 
H multiplied by the fictitious complex material constant u,, ff The latter 
assumption leads to an expression for the secondary voltage as Mows: 

General Case of a Cylindrical Test Object. Extensive mathematical calcu 
lations have been made for the effective permeability ^ for the general case of 

1. Electrical conductivity, a, of the test object. 

2. Relative magnetic permeability, n rol ., of the test object. 


3. Diameter of the test object, d. 

4. Test frequency, /. 

The homogeneous cylinder or rod of uniform material results as a special case of 
the general computation. 48 In the calculation of the effective permeability, the 
following expression occurs as the argument A of the Bessel functions : 

where / frequency of the primary exciting field, c.p.s. 

a = electrical conductivity of test material, meter/ ohm-mm. 2 . 
d = diameter of test object, cm. 

Hr-i. = relative permeability of test material (approximately 100 for iron, unity for 

Limit Frequency, / /; . Fortunately the test frequency, /, can always be se 
lected so that the factor A in Eq. (4) will be equal to unity. This frequency at 
which A = 1 for given values of a, d, and [a reL , is called the limit frequency f g . 
Substituting this limit frequency / for the frequency / in Eq. (4), and taking 
factor A = 1, the expression can be solved for f g as follows: 

, 5066 / f> \ 

! ' = - c - p - s - (5) 

If test frequencies are selected so that the frequency ratio f/f g = 1/2,3 . . . , 
the quantity A in Eq. (4) also assumes the values A = 1,2,3 .... 

Effective Permeability as a Function of Frequency Ratio. Since the effec 
tive permeability |i eff . is a function only of the magnitude of the quantity A in 
Eq. (4), considerable simplification is attained if u eff . is expressed as a function 
of the frequency ratio f/f ff , i.e., as a multiple of the limit frequency f ff of Eq. 
(5). Fig. 7 shows computed values of pi eff . (both real and imaginary components) 
for the case of the solid cylindrical test object. 48 This curve gives values for 
jipff., as a function of the frequency ratio f/f g , in the complex effective perme 
ability plane. Points corresponding to multiples of the limit frequency j ff are 
marked on the curve throughout the range from f/f ff = (at the top) to f/fg = 
(at the bottom). In this curve the real component of [i eff is plotted vertically, 
while the imaginary component is plotted horizontally. Fig. 8 lists precise 
values for ^ ff . corresponding to points marked in the curve of Fig. 7. 

Sample Calculations for Effective Permeability and Secondary Coil Volt 
age. Since the data of Figs. 7 and 8 form the basis of eddy current tests of 
cylindrical objects, the calculation of n off . and the determination of the second 
ary coil voltage E wc from Eq. (3) will be illustrated by a practical example. 
Let us assume a copper rod of diameter d = 1 cm., with electrical conductivity 
a = 50.6 meter/ohm-mm. 2 , and a relative magnetic permeability u^i. = 1. The 
limit frequency / is calculated to be f g = 100 c.p.s., from Eq. (4). If the rod is 
placed in a coil excited at a frequency of 100 c.p.s., the frequency ratio f/fg is 
unity Under these conditions the copper rod exhibits an effective permeability, 
LU , corresponding to /// = ! in Figs. 7 and S. From Fig. 8 it is found that 
tff' (rean = ^798 and [lefMimn,., = - 1216 - However, if one were now to select 
It measurement frequency of 1000 c.p.s, the ratio ///, would be 10. In this case 
Fig. S gives the components of the effective permeability as p, eff . (real) - 0-4678 
and' m. (lmn g.,= 0.3494. 



0,1 0,2 0,3 0,4 

Institul Dr. Koerster 

Fig. 7. Variation in effective permeability ^ ff . with frequency ratio /// plotted i 
the complex permeability plane. 










































































Institut Dr. Foerster 

Fig. 8. Tabular listing of values of effective permeability \i e tt. for specific values of 
frequency ratio ///, from Fig. 7. 

The voltage induced in a one-turn secondary winding (n=l), for H Q = 1 
oersted, can be computed from Eq. (3) for a frequency / = 100 c.p.s. Since this 
is equal to the limit frequency, f g = 100, the frequency ratio f/f ff = 1. In this 
case the real component of secondary voltage is 

^Hoc.(rcal) = 0.600 X 10- 6 VOlt 

and the imaginary component is 

^e.dmM.) =4.83 X ID' 6 VOlt 

111 accordance with the terminology of electrical engineering, the real component 
of voltage results from multiplying Eq. (3) by the imaginary component of the 
effective permeability; hence 

#roi = 2ji/w(jJV4)^rci.|x..ciin.ijffo X 10' 8 volt (6a) 

The imaginary component of voltage is similarly obtained by multiplying by the 
real component of the effective permeability; hence 

E lmne = 2jc/rc(jcdV4)^i.|^ff.(reni>#o X 10~ 8 volt (6b) 

These basic eddy current test equations permit computation of the components 
of secondary voltage for various test-object materials and dimensions. The steps 
involved are: 

1. Substitute the numerical values of test-object electrical conductivity a, relative 
magnetic permeability ni., and dimensional constant d into Eq. (5) to deter 
mine the limit frequency } 9 . 


2. Divide the operating test frequency / by iho limit frroiueiiry / to obtain the 
frequency ratio ///. t 

3. Use this frequency ratio and Figs. 7 or 8 to obtain the components, of the effec 
tive permeability n-off.. 

4. Calculate the components of secondary coil voltage by substituting these values 
of effective permeability into Eqs. (6, a and b). 

For all nonferromagnetic materials, the value of U IM ,I. = 1 is used in Eq. (6, 
a and b). 

Values of ^ rel for ferromagnetic materials are obtained from experience or by 
tests on the specific material. To illustrate the procedure, the secondary voltage 
components will be calculated for a ferromagnetic test object for which: 

Electrical conductivity a = 10 motcr/ohm-mm, a 

Diameter d = 1 cm. 

Relative permeability MTI. == 100. 

The secondary coil will be assumed to be a one-turn winding {;/ = 1). The pri 
mary coil field strength is taken as 1 oersted (# = 1). For n larger number of 
turns or higher field strength, this calculated voltage would he multiplied by the 
corresponding factors. The limit frequency f g is found from Eq. (5) to be 5.07 
c.p.s. If the test frequency is / = 50 c.p.s., one obtains 

/// = 10 (approx.) 

M'Pff.(rc-ttl) = 0.47 
lAefMlm.R.) = 0.35 

The secondary coil voltage components obtained from Eq, (()) are 

#rea) = 86.6 X 10" fl VOlt, 

-Em,.*. = 116.4 X 10- volt 

ETER. In the preceding examples, it, was assumed that a test object of diameter 
d completely filled a secondary coil with the inside diameter 7), so that d = D. 
In practice, however, the coil diameter D must be considerably larger than the 
test object diameter d, so that the test object can pass through the secondary coil 
freely and without contacting it at high test speeds. Fig. 9 illustrates a cross- 
section of a secondary coil enclosing a smaller test object. The annular air space 
between the test object and the coil has an area of 

and permeabilities of 

M-rei. = 1 M-off. = 1 (a real number only) 

Secondary Coil Voltages. The primary magnetizing field 7/ () produces two 
components of voltage in the secondary coil. The first component, contributed 
by the field in the annular air space, is purely imaginary (since [,i offi is real) ; 

# fl oo. (ai r ,!*> = 2jt/?i(jt/4) (D 2 - rf 2 )#<> X lO' 8 volt (7) 

The second component, contributed by the presence of the test object, is a 
complex number: 

^soe.aost object) = 2jC/ (jl/4)d 2 |A M l.^f fj/o X lO'* Volt (8) 



Fig. 9. Test coil containing a test object of smaller diameter d, partially filling the 


The total secondary coil voltage is obtained by the vectorial addition of the 
components of Eqs. (7) and (8) ; hence 

fi'MM-. (total) ^Hpr.dilr rliiR) 4" -#<><:. (tout object) 

= 2jt/n(jtD 2 /4)[l - (d/D) 2 + (d/ZV,i Heff.ltfo X 10- 8 volt 

The ratio (d/D) 2 indicates the fraction of the secondary coil area filled by the 
test object. This is known as the fill factor, r], for the coil; i.e., 

n = (d/D)* (10) 

In the absence of a test object (d = 0), the term within the brackets in Eq. (9) 
becomes unity. For the empty coil the secondary coil voltage becomes 

Eo = 2nfn(nD 2 /4)H X 10' 8 volt (11) 

In the general case in which a nonferromagnetic test object partially fills the 
test coil, the coil voltage can be expressed as 

E = E (l - r\ + i\n. t t.) (12) 

The parenthesized expression, (1 r| + T]u ef f.), in Eq. (12) is an important and 
useful parameter. By multiplying it by the voltage E Q of the empty coil, one 
obtains the secondary voltage with a test object within the coil. 

Complex Permeability Plane. Fig. 10 shows values of the parenthetical ex 
pression from Eq. (12), plotted on the complex permeability plane for two fill 
factors, as 

TI = 1 (The test object completely fills the test coil.) 

V] = % (The cross-sectional area is only one-half the area within the secondary 
winding of the test coil.) 



For rj = 1, the parenthetical expression has the same value MS ji t . ff> . Figs. 7 and S 
apply for \] = 1. The outermost curve in Fig. 10 is thus identical with the curve 
of Fig. 7. For the case in which TJ = %, the steps involved in using Fig. 10 are 

1. Plot the value of (1 - TI) = 0.5 along the ordinate in Fig, 10, i.r., in the direction 
of M-orr.(rcai). Here the term (1 TI) corresponds lo the contribution of the 
annular air ring and is not influenced by the properties of tho test object. 
With an infinitely high test frequency, p,, f f. =0 (from Fig;. 8). In this case 
only the air ring influences the value of the parenthetical expression in Kq. 
(12). The parenthetical expression then reduces to (1 t]). 

2. The value of the complex permeability M-off. multiplied by the fill factor 11 is 
then plotted vectorially from the end of the vector (1 TI) at point B. For 
example, the runoff, value for /// = 4 is plotted in Fig. 10 as line HA. 

3. The total value of the parenthetical expression in Eq. (12) is now given by 
the vector from the origin to the point A (distance OA). Us real and 
imaginary components can be read from the coordinates of Fig. 10. 

Fig. 10. Parameter (1 - TI + mio".) as a function of complex permeability 
Hoff., for fill factors r\ = 1 and TI = %. 



Summary. In Fig. 10: 

Distance OB represents the term (1 - v\) 

Distance BA represents the term riM-off. 

Distance BC represents the product T]jioff.(rcai) 

Distance CA represents the product TUicfMiman.) 

Distance OA represents the entire parameter (1 TJ + Ti|Li a ff.) 

Correction for Ferromagnetic Test Object. If a ferromagnetic test cylinder 
with the relative magnetic permeability (.ij. el< is placed in the test coil of Fig. 9, 
it is necessary to modify Eq. (12) as follows: 

E = Eo(l - Tl + -nHrol.|A.M.) (13) 


The impedance characteristics of a single test coil (see Fig. 11) can be deter- 

Fig. 11. Arrangement of cylindrical test object in a single test coil characterized 
by an inductive reactance coL and an ohmic resistance J?. 

mined with the aid of the preceding development. In the absence of a test object, 
the empty coil of Fig. 11 is characterized by an impedance whose components are 

coLo inductive reactance (empty coil). 

7? = ohmic resistance (empty coil). 

The self -inductance of a cylindrical coil with no test object is 

Lo = K(n*A/l) 


whore K = constant. 

A = average coil area. 
?i = number of series turns in coil winding. 
I = coil length. 

When a portion of the coil area is filled by a cylindrical test object of diameter d, 
conductivity a, and relative permeability [A rel ., the self-inductance L is calculated 
by analogy to Eq. (13) as 

L = L (l T] 


The. voltage drop due to self-inductance E L is given by the product of the coil 
current 7 and the inductive reactance col/; thus 

EL = I(2nfL) = 



The voltage drop due to ohmic resistance in the test coil is given by the prod 
uct of the coil current by the resistance 

E R - IR (real) 
In Eqs. (6) through (12), the excitation field strength of a current -carrying coil 

where A' is a constant, n is the number of magnetizing turns, and / is the current 
through the coil. 

Normalized Coil Characteristics. The voltages induced in a test coil contain 
ing a test object are said to be "normalized" when they are divided by the induc 
tive reactance component of voltage, EL Q) of the same coil when empty (see 
Eq. 11). The normalized imaginary component of coil voltage becomes 

%it = ^- = [1 - -n + THirH.H*rr.froi>l (16a) 

JLr lQ col/o 

The normalized real component of coil voltage becomes 

Erwil /V llo r T 

-~^- = - =r-l = [TUlrM.H-f f.(ii.uir.>l 
tii^ COLo 

These equations indicate that the curves of Fig. 10 apply to either 

1. The normalized secondary coil voltage components, or 

2, The normalized values of the inductive reactance (coL/coLn) and the resistance 
(R flo)/a>Ln of a single test coil. 

The normalized resistance (real) and the normalized inductive reactance 

(imag.) of the test coil are the components of its "normalized" impedance. 

Example of Nonferromagnetic Test Object. In the case of nonferromag- 
netic test materials such as stainless steel, copper, or aluminum, |A n ,i. = 1. If the 
test cylinder entirely fills the test coil (fill factor >] = !), Eqs. (If) a and b) 
reduce to 

__ (tiL __ Eimng. 

imng. . ^ . 

aT (17a) 

_ R Rn _ Evrni fii \ 

^"-> - ~^ZT ~ ~W (17b) 

Three Complex Planes for Eddy Current Test Data. Thus the effective 
permeability represented in Figs. 7 and S is identical to the secondary voltages 
and coil data normalized to the empty coil values. Fig. 7 equally well represents 
three valuable forms of eddy current test data as follows: 

1. The complex permeability plane, showing components of n<.ff. (real and 
imaginary) . 

2. The complex impedance plane, showing the components of tho normalized 
impedance (real and imaginary). 

3. The complex plane of secondary coil voltages normalized to the "empty coil" 

For example, the impedance plane will be referred to in later discussions of test 
methods using a bridge arrangement of primary coils only. On the other hand, 
test arrangements with both primary and secondary coils will be referred to the 
complex voltage plane. Eqs. (17 a and b) indicate that both test arrangements 
are described by the same equations. 


The effective permeability, obtained from Eq. (5) or Figs. 7 or 8, is a funda 
mental factor in all eddy current tests. It is used to obtain information on 
physical properties or discontinuities in test objects. 


Reference numbers in this section refer to the references at the end of the section 
on Eddy Current Test Indications. 






Conductivity, Permeability, and Diam 
eter of Cylindrical Test Objects 

Assumptions used in analysis 1 

Separation of effects of conductivity and 

diameter (nonmagnetic cylinder) 1 

Variation of conductivity c 2 

Characteristics of the complex impedance or 

voltage planes 2 

Complex impedance and voltage planes for 
various fill factors (nonferromagnetic 

materials) (/. 1) 3 

Selection of frequency for separation of 

diameter and conductivity variations 2 

Effect of high test frequencies 3 

Effect of low test frequencies 3 

Effect of variation in fill factor 4 

Determination of limit frequency }g 4 

Quantitative test indications 4 

Separation of effects of conductivity, diameter, 
and relative magnetic permeability 

(ferromagnetic cylinders) 28 4 

Complex impedance and voltage planes for 
ferromagnetic cylinders with various 

relative permeabilities (/. 2) 5 

Advantage in crack detection 5 

Effect of changes in cylinder diameter 6 

Effect of changes in electrical conductivity . 6 

Effects of magnetic hysteresis 6 

Separation of variable parameters 6 

Field Strength and Eddy Current 
Distribution in Cylinders 48 

Field strength distribution 6 

Amplitude and phase angle of field 
strength distribution within a metallic 

cylinder (/. 3) 7 

Eddy current distribution 8 

Amplitude and phase angle of eddy cur 
rent distribution within a metallic cylin 
der (/. 4) 8 

Examples of eddy current distributions .... 9 

Magnitudes of field strength and eddy 

current density as functions of radial 

position in cylinder (/. 5) 9 

Penetration depth 9 

The penetration depth P at which field 
strength and eddy current density are 
1/e = 37 percent of surface values, as a 
function of frequency ratio (/. 8) 10 


Control of eddy current magnitudes 10 

The similarity law in eddy current testing .... 10 
Application to detection of discontinuities . . 11 

Application to model tests 11 

Experimental verification of mathematical cal 
culations 11 

Mercury model for experimental verifica 
tion of field strength and eddy current 
distribution in cylindrical test object 

(A 7) 12 

Comparison between calculated and ex 
perimentally determined field strength 
distribution in mercury model cylinder 

(/ 8) 12 

Model tests as a basis for quantitative tests of 

discontinuities 13 

Model arrangement for simulating any 
given discontinuity in the mercury cylin 
der (/. 9) 13 

Electric circuit used in model tests with 

mercury cylinder (/. 10) 13 

Summary of types of discontinuities in 
cylindrical specimens studied in mercury 
model tests (/. 11) 14 

Test Coil Arrangements for Feed- 
through Testing 

Classification of feed-through coil arrange 
ments 15 

Arrangements for test coils for feed- 

through tests (/. 12) 16 

Complex voltage or impedance plane for feed- 
through coils 15 

Complex voltage and impedance planes for 

test coil arrangements of Fig. 12 (/. 13) 17 

Single primary test coil 15 

Primary and secondary coil arrangement .... 17 

Bridge arrangement of test coils 17 

Differential or self -comparison arrangements 18 

Tests for Cracks in Solid Cylinders 

Cracks in nonferrous cylindrical test objects . 18 

Impedance variations caused by surface 

and subsurface cracks in nonferromag- 

netic cylinders, at a frequency ratio 

///, = 5 (/. 14) 19 

Impedance variations caused by surface 
and subsurface cracks in nonferromag- 
netic cylinders, at a frequency ratio 
///, = 15 (f. 15) 20 



CONTENTS (Continual} 


Impedance variations caused by surface 
and subsurface cracks in nonferromug- 
netic cylinders, at a frequency ratio 

///, = 50 (/- 16) 21 

Impedance variations caused by surface 
and subsurface cracks in nonferromag- 
netic cylinders, at a frequency ratio 

///, = 150 (/. 17) 22 

Test coil voltages 10 

Differential test coil arrangement contain 
ing sound tind defective cylindrical test- 
specimens (/. 18) 22 

Suppression of variations in cylinder diam 
eter 21 

Frequency selection for surface crack de 
tection 23 

Changes in effective permeability |A/iti<ff.| 
caused by surface cracks of various 
depths in nonferromagnetic cylinders 

(/. 19) 23 

Changes in component of effective perme 
ability |A/4oft.| perpendicular to direc 
tion of diameter changes, caused by 
surface cracks of various depths in non- 

ferromagnetic cylinders (/. 20) 24 

Frequency selection for subsurface crack de 
tection 24 

Changes in effective permeability |A/uff.| 
caused by subsurface cracks of 30 per 
cent crack depth at various distances 
below the surface of nonferromagnetic 

cylinders (/. 21) 25 

Changes in component of effective perme 
ability |A/*eff.| perpendicular to direc 
tion of diameter changes, caused by 
subsurface cracks of 30 percent crack 
depth at various distances beneath outer 
surface of nonferromagnetic cylinder 

(/. 22) 28 

Cracks in ferromagnetic cylinders 25 

Suppression of variations in cylinder diam 
eter 25 

Changes in component of effective perme 
ability |A0eff.| perpendicular to the di 
rection of diameter and relative perme 
ability changes, caused by surface cracks 
of various depths in ferromagnetic cylin 
ders (/. 23) 27 

Frequency selection for crack detection 26 

Quantitative measurement of crack depths , , . 27 
Experimental determination of frequency 

ratio fjfg 27 

Variation of angle a with frequency ratio 
flfg or product ffd* in nonferromagnetic 

cylinder (/. 24) 28 

Crack evaluation with comparison specimen 28 
Explanation of complex plane conditions 
for comparison specimen test arrange 
ment (/. 25) 29 

Conductivity or diameter evaluation with 
comparison specimen 30 


Instrument sensitivity calibration 30 

Quantitative crack-depth calculation}* 31 

Quotient y Ajt<'ff./( absolute value) as a 
function of frequency ratio /'/,/ for 
surface (-nicks of various depths iti non- 
ferromagnetic cylinders (/. 26) 30 

Computing crack depth from instrument 

indication 32 

Quotient Q = A/iorr. /(absolute value) as a 
function of crack depth, for various fre 
quency ratios, ///(/, with noiiferwmng- 
netic cylinders (/, 27) 31 

Applications of Test Instruments 

Operation of test, insl rument s 32 

Simplified schematic diagram of eddy cur 
rent test instrument for quantitative 
measurement of crack effects, conductiv 
ity effects, and diameter effects i-n non- 
ferrous cylinders (/. 28) 32 

Analysis of screen patterns 33 

Char act eristics of elliptical screen pattern, 
resulting from sinusoidal delleetion volt 
ages displaced in phase angle, on eddy 

current test instrument (/. 20) 33 

Adjustment of the phnse shifter 34 

Instrument sensitivity characteristics 35 

Impedance variations (amplitude and 
phase) for a 1 percent variation of con 
ductivity (f, 30) 34 

Impedance variations (amplitude and 
phase) for a 1 percent variation in diam 
eter (/. 31) 35 

Variation in effective permeubilHy A/ttoff,, 
for 1 percent conductivity variation in 

nonferromagnetic cylinder (/. 32) 36 

Variation in effective permeability A/i.ff., 
for 1 percent diameter variation in non- 
ferromagnetic cylinder (/. 33) 36 

Quantitative meaHurements of conductivity or 

diameter effects 35 

Quotient Q = A/ttoff. /(absolute, valup), for 
1 percent conductivity variation in non- 

ferrouwKnetie cylinder (/. 34) 37 

Quotient Q = A/ii'ff. /(absolute value), for 
a 1 percent diameter variation in non- 
ferromagnetic cylinder (/. 35) 37 

Conductivity measurement 36 

Diameter measurement. 38 

Setting tolerance limits 38 

Detecting cracks independently of diameter 

variations 38 

Component of quotient Q A/^fr./ (abso 
lute value), perpendicular to diameter 
direction in effective permeability plane 

(/. 36) 39 

Crack evaluation with diameter variations . 39 
References 39 



Conductivity, Permeability, and Diameter of Cylindrical 
Test Objects 

ASSUMPTIONS USED IN ANALYSIS. The influence of test-object 
electrical conductivity o, magnetic permeability [j rel ., and diameter d will be 
explored for the case of a long, cylindrical test object within- a cylindrical test 
coil arrangement, with the same basic assumptions listed in the preceding section 
for the analysis of effective permeability. These test-object properties influence 
the value of the characteristic function; i.e., 

1 Tl + TlM.rol.Hcff. 

used in calculating coil impedance characteristics (Eqs. 13, 15, 16, and 17 in the 
section on Eddy Current Test Principles) or secondary coil voltages. This func 
tion contains all the basic factors which influence eddy current tests where : 

1. The effective permeability \i tt t. is determined by the frequency ratio ///,. 

2. The limit frequency /, is a function of the physical properties of the test 
object, including conductivity a, permeability M-I. and diameter d. 

3. The test frequency / is a characteristic of the test instrument. 

4. The fill. factor t\ is determined by the diameter of the test object, d, and the 
inside diameter of the test coil, D. 

Fortunately for the development of a number of practical eddy current test 
instruments, the electrical conductivity <r appears only in the term |i effi of the 
characteristic function. (See Eq. 5 in the section on Eddy Current Test Prin 
ciples.) On the other hand, the test-object diameter d influences both the effec 
tive permeability fi eff . and the fill factor TJ. 

ETER (NONMAGNETIC CYLINDER), To illustrate the possibilities for 
separation of the effects of variations in electrical conductivity from those of 
test-object diameter, the case of a nonferromagnetic cylinder will be considered. 
The test object will be assumed to be an aluminum rod (with a conductivity 
a=35 meter/ohm-mm. 2 , a diameter d = l2 cm., a magnetic permeability 
[i rel =1) which entirely fills the test coil so that fill factor r) = l. The limit 
frequency f g is calculated from Eq. (5), Eddy Current Test Principles, to be 
f ff = 100. The test frequency of the instruments is assumed to be / = 10,000 c.p.s., 
so that the frequency ratio }/f p = 100. 

From the outermost curve of Fig. 10 in the section on Eddy Current Test 
Principles (which applies for 7] = 1), the point }/f ff = 100 (point D) is selected in 
order to obtain the secondary voltage of the test coil normalized to the "empty 
coil" value (see Eq. 16, Eddy Current Test Principles). Now suppose that the 
aluminum rod is ground down from its original diameter (1.2 cm.) to 0.85 cm. 



From Eq. (5), Eddy Current Test Principles, the frequency ratio /// is now 50. 
From Eq. (10) in the same section, the fill factor 1] is calculated ns 0.5. 

To compute the new position on the impedance piano corresponding to the 
reduced bar diameter (see Fig. 10 in the section on Eddy Current Test Prin 
ciples) : 

1. The value (1 TJ) =0.5 is measured along the ordinate. Thus distance OB 
corresponds to the influence of the annular air ring upon the secondary coil 

2. The components of the product Tin-off, (where TI = 0.5 and the components of 
M-cff. are found in the section on Eddy Current Tost Principles from Fig. 8 for 
ilia = 50) are plotted vectorially from point B. The real component (BE} = 
0.5 X 0.20 is plotted along the ordinate axis. The imaginary component 
(EF) = 0.5 X 0.18 is plotted parallel to the abscissa. 

3. The characteristic function corresponding to the reduced bur diametor is repre 
sented by point F on the complex voltage plane. The vector OB represents the 
term (1 TI), the vector BF corresponds to the term ru^rc., and their vector 
sum OF corresponds to the entire characteristic function (1 TI + iifioff.). 

Thus, the reduction of bar diameter from 1.2 to 0.85 cm. has displaced the normal 
ized secondary voltage (or the normalized impedance values) from point D to 
point F. 

If the diameter of the aluminum rod is still further decreased to 0.6 cm., the 
new frequency ratio f/f ff will equal 25, and the fill factor j] will equal 0.25. In 
the complex plane of Fig. 10 in the section on Eddy Current Test Principles, the 
distance OG corresponds to the term (1 r|) = 0.75. Addition of the components 
of r)u, eff for ///^ = 25 locates point / for the characteristic function for the 
aluminum rod of 0.6-cm. diam. 

Variation of Conductivity a. If the electrical conductivity of the original 
1.2-cm. diam. aluminum rod had been reduced from its original value of a = 35 
to a new value of a = 17.5 (for example, by heating), point D on the outermost 
curve of Fig. 10 in the section on Eddy Current Test Principles would be displaced 
from the value corresponding to f/f ff = 100 to the point A' whose /// = 50 (see 
Eq. 5, Eddy Current Test Principles). Thus, conductivity variations take place in 
a different direction from diameter variations in the complex impedance or voltage 
planes. This introduces the possibility (later discussed in more detail) of separat 
ing conductivity variations (used, for example, in sorting alloys) from diameter 

VOLTAGE PLANES. Fig. 1 shows curves of the characteristic function on the 
complex impedance or voltage plane for fill factors of t] = 1, 0.75, 0.50, and 0.25. 
Various frequency ratios f/f ff are marked on the various fill-factor curves. The 
directions of conductivity variations are indicated by arrows Libeled with o*. 
The directions of diameter variations are shown by arrows labeled d. The 
characteristics discussed here and illustrated by Fig. 1 are used as a basis for 
several eddy current instruments to be described later. 

Selection of Frequency for Separation of Diameter and Conductivity 
Variations. At low test frequencies, where f/f g takes on low values, the char 
acteristic function is represented by points near the top of the curves of Fig. 1. 
Here the angle between the directions of diameter and conductivity variations 
decreases considerably, and the separation of these points is difficult. The separa 
tion of diameter and conductivity effects is better where the angle between the 


37 3 

two variations is greater. For best results it is desirable to work at frequency 
ratios f/f ff larger than 4. 

Effect of High Test Frequencies. At high test frequencies, where the f/f ff 
ratio is of the order of 10 or greater, the characteristic function is represented by 
points on the lower halves of the curves of Fig. 1. As the frequency ratio ap- 


Inatitut Dr. Foerster 

Fig 1 Complex impedance and voltage planes for various fill factors (nonferro- 

magnetic materials). 

Broaches infinity (very high test frequencies), the direction of changes due to 
Sete variations approaches the vertical, the direction of the self-mductm or 
of the imaginary component of secondary coil voltage. On the other hand varia 
tions in electrical conductivity a cause displacements at an angle of about 45 deg. 
to the abscissa in the complex plane. 

Effect of Low Test Frequencies. For very small ///, ratios, the direction of 
diameter changes approaches the direction of changes in resistance (parallel to 


the abscissa). This is the direction of the real component of voltage, 7 roal . As 
f/fg becomes smaller (near the tops of the curves), the directions of changes in 
apparent test coil resistance R and test-object conductivity approach each 

Effect of Variation in Fill Factor. Lines connecting the point at which 
M-eff. 1 (top left of impedance plane) to specific f/f g points on the curve for fill 
factor T] = 1 intersect the curves for all other fill factors nt exactly the same f/f p 
values, In other words, a straight line drawn from the point for j.i pffp = 1 shows 
the effect of variation in fill factor at a constant test frequency. 

Determination of Limit Frequency f ff . The straight lines connecting points 
of various fill factors for a given test frequency also permit evaluation of an 
eddy current test without detailed coil data. The angle ^> between the ordinate 
axis and the straight line corresponding to a specific f/f ff ratio is independent 
of the fill factor r|; that is, independent of the test coil data. Consequently the 
f/f ff ratio for the test object can be determined simply by measuring only the 
angle <. Since an instrument normally operates at a specific frequency /, the 
limit frequency f g is given by the angle 0. The diameter of the test cylinder can 
be easily measured with a caliper. Thus, with the aid of Eq. (5), Eddy Current 
Test Principles, and the values of f ff and d t the value of the electrical conductivity 
a can be obtained quantitatively without specific coil data. 

Quantitative Test Indications. Practical eddy current test instruments to be 
described later have the following capabilities. They 

1. Read directly the product ad 2 (electrical conductivity times square of cylinder 
diameter) . 

2. Indicate quantitatively, for each sensitivity step, the effect of 1 percent varia 
tion in electrical conductivity or in diameter. 

3. Show crack depths of specific magnitudes, such as 5, 10, or 20 percent of the 
cylinder diameter. 

NETIC CYLINDERS). 28 Eq. (16), Eddy Current Test Principles, used in the 
analysis of the response of the test coil to nonferrous cylinders, can be modified 
to show the entirely different effect of ferromagnetic test materials. The relative 
magnetic permeability of ferromagnetic materials is normally far greater than 
unity (^ reL 1). When the fill factor T| is not too small, the term (1 T|) can 
often be neglected in comparison with the much larger term tifi ro i.|X e ff - In con 
sequence Eqs. (16 a and b) may be modified for this case as follows: 


fi o ^ = nn.roi.H.eff.croai) (approx.) (la) 

#real __ R 

-~JT- "^ ^ nMrel.M-eff.dmag.) (appI'OX.) (lb) 

Fig. 2 shows the complex impedance plane for the case in which a ferromagnetic 
test cylinder entirely fills the test coil (fill factor TJ = 1). This curve differs from 
preceding curves in that the values of [A C f Mrea i) and He a . (lmag ., are increased by 
the factor [i reh . 

The effective permeability components, |i C M.(reni) and M*ff.<imag) depend only 
upon the value of the frequency ratio; thus 

J_ __ /Urol.gQ? 2 

j, ~ 5066 (2) 


37 5 

In Eq. (1), the factor r]|i rel . = (d 2 /D 2 )^ &l appears in the same manner as in 
Eq. (2) ; i.e., as the product of the relative permeability and the square of the 
test-object diameter. Because they are associated together in this manner, varia 
tions in jj, rel> and d appear in the same direction on the impedance plane (Fig. 2) . 

'jj,d direction 

Institut Dr. Foerster 

Fig. 2. Complex impedance and voltage planes for ferromagnetic cylinders with 
various relative permeabilities (fill factor TJ = 1). 

Consequently it is impossible to distinguish whether a change in this direction is 
caused by a permeability or by a diameter effect. 

Advantage in Crack Detection. Fortunately, in surface crack testing of 
steel, for example, the common direction of diameter and permeability changes 


on the impedance plane is at a relatively large angle to the direction of changes 
caused by cracks in test cylinders. Thus the interfering effects of variations in 
diameter and permeability can be separated from the indications of surface cracks. 
Such permeability variations can arise from conditions such as internal stress fol 
lowing straightening. They are often very annoying difficulties in eddy current 
tests of steel parts. 

Effect of Changes in Cylinder Diameter. With nonferromagnetic materials, 
an increase in cylinder diameter resulted in a decrease in the effective permeability 
(see Fig. 1). However, an opposite effect is obtained with ferromagnetic materials 
in the useful range (see Fig. 2) of measurements (0 < f/f ff < 200). Hero the 
increase in the quantity of magnetic material in the test coil (which increases 
the secondary voltage) outweighs the weakening of the field by eddy currents 
(which tends to reduce the secondary voltage) . 

Effect of Changes in Electrical Conductivity. In the impedance or voltage 
plane of Fig. 2, the direction of changes in electrical conductivity a is identical to 
the direction of changes in effective permeability with variations in frequency 
ratio f/f ff . Conductivity changes can be separated from the effects of changes in 
diameter or permeability most effectively in the upper half of the impedance 
plane of Fig. 2 ; where the angle between the conductivity direction and the 
direction of diameter and permeability changes is greatest. For f/f ff ratios less 
than 15, changes in conductivity (which are a function of alloy content and of 
structural conditions related to hardness) can be measured independently of 
the effects of permeability changes due to mechanical stress after cold-working 
such as drawing or straightening. Such measurements can also be made inde 
pendent of the effects of the diameter variations always present in commercial 

Effects of Magnetic Hysteresis. Magnetic hysteresis effects add to the hori 
zontal or real component of the apparent impedance of the test coil, in the direc 
tion of [x e ff.(imag.) ^ Fig. 2. However, this component is always small in 
comparison with the component due to eddy currents and will be neglected in 
this discussion. 

Separation of Variable Parameters. The preceding observations indicate the 
degree of separation feasible, at any given test frequency and for a given coil 
diameter (fill factor y]), of the variables: 

1. Effects of electrical conductivity a. 

2. Effects of magnetic properties M,rei.. 

3. Effects of geometrical properties (diameter of cylinder d). 

The direction of changes of the apparent impedance of the test coil indicates 
which effect was responsible for the impedance variations. Instruments have been 
developed for the continuous testing of nonferrous wires and rods, for example, 
which measure electrical conductivity and diameter directly with an accuracy of 
1 percent while samples pass through the test coil rapidly and without contacting 
it (see section on Eddy Current Test Equipment) . 

Field Strength and Eddy Current Distribution in Cylinders 4lS 

FIELD STRENGTH DISTRIBUTION. The distributions of magnetic- 
field strength and of eddy current density within the cylindrical test object placed 
in the a.-c. field of the test coil determine the depth sensitivity of the test 


37 7 

method. The formulas given in the literature for the field strength penetration 
depth apply only for very high frequency ratios, where the Bessel functions can 
be approximated by means of exponential functions. However, the range of high 
/// ratios is of little interest in eddy current nondestructive tests. At high test 
frequencies the sensitivity to conductivity and crack effects is reduced consider 
ably, whereas the sensitivity to diameter changes increases considerably. To 
obtain the true field strength and eddy current distributions within the test 


Fig. 3. Amplitude and phase angle of field strength distribution within a metallic 
cylinder. Curves show various frequency ratios ///,. Numbers on curves show frac 
tion of cylinder radius from surface (r = r ) . 

cylinder the ///, ratio must be calculated by Eq. (5), Eddy Current Test Prin 
ciples for the test object. This ratio characterizes its physical properties of 
electrical conductivity 0, relative magnetic permeability ^ rel ., and diameter d, 
as well as the test frequency /. ,..,* 

Fig 3 shows the field strength distribution within cylindrical specimens for 
various frequency ratios. The vector from the zero point of the coordinate system 
to the point r = 1 corresponds in magnitude and direction to the field strength 
directly at the surface of the test cylinder (the external field strength). Num 
bers alono- the curves indicate radial location in the cylinder, in fractions of the 



cylinder radius. For example, the value of 0,6 corresponds to a point at a radius 
of 60 percent of the distance from the center line to the outer surface r of the 
cylinder. Such a point would lie beneath the cylinder (surface by a distance corre 
sponding to 20 percent of the diameter of the cylinder (2r ). For low multiples 
of the limit frequency f ff (small frequency ratios ///,), the field strength suffers 
a phase displacement toward the center of the cylinder. At high frequencies, in 
addition to the phase displacement, the field strength also diminishes rapidly 
toward the center of the specimen. 

EDDY CURRENT DISTRIBUTION. Fig. 4 illustrates the distribution of 
eddy current density from the surface to the interior of the cylinder, in fractions 




Fig. 4. Amplitude and phase angle of eddy current distribution within a metallic 
cylinder. Curves show various frequency ratios ///,. Numbers on curves show frac 
tion of cylinder radius from outer surface (r ). 

of the specimen radius, for various frequency ratios. The current density values 
given in this figure correspond to an external field strength of 1 oersted in the test 
coil. Defects or discontinuities within the cylindrical specimen can be detected 
only when the eddy currents are interrupted or deflected -by the discontinuity. 
In contrast to the field strength, the eddy current density always disappears at 
the center of the cylinder. Thus discontinuities which lie exactly along the 
center line (such as central piping) cannot be detected. The depths at which 
discontinuities can be detected beneath the surface will be described quantitativelv 
later. J 



Examples of Eddy Current Distributions. Fig. 5 shows the absolute magni 
tudes of eddy current and field strength distributions (without regard to phase 
angles) at whole multiples of the limit frequency throughout the cross-section of 
the cylinder. Its use will be explained by means of an example. Assume that the 
test object is a cylindrical rod of an aluminum-magnesium-copper alloy whose 
electrical conductivity a = 20 meter/ohm-mm. 2 and whose diameter d = 1.6 cm. 
The limit frequency is calculated from Eq. (5), Eddy Current Test Principles, as 
f ff = 100, At a test frequency of 10,000 c.p.s. = 100 f ff , the field strength is prac 
tically zero at 0.3 of the radius from the center line, while at 0.9 of the radius, the 
field strength is decreased to approximately one-half of the external field strength. 


-r - 0,2040,60,81,0' 

Diameter - 

r -0 0,2 04 060,8 1.0 T 

Diameter * 

Fig. 5. Magnitudes of field strength and eddy current density as functions of 
radial position in cylinder, for various frequency ratios ///,. 

Penetration Depth. The penetration depth P is generally defined as the depth 
below the surface at which the field strength has decreased to l/e = 36.8 percent 
of the surface field strength. This value is shown in Fig. 5 by the line PP. Thus 
the penetration depth can be obtained in fractions of cylinder radius for each f/f ff 
ratio. The frequency ratio f/f ff is calculated by Eq. (5), Eddy Current Test Prin 
ciples, from the physical constants of the test object and the test frequency. Fig. 6 
shows the penetration depth P as a function of the frequency ///,. It indicates 
that for frequency ratios f/f ff < 13, the field strength in the center of the 
cylinder is greater than 36.S percent of the surface field strength. In this case the 
penetration depth P is greater than the cylinder radius r . In Fig. 6 the solid 
curve labeled H shows the penetration depth P of the magnetic field strength 
H. The solid curve labeled G denotes the penetration depth of the eddy current 
density. The dotted curve gives the theoretical penetration depth for both field 
strength and eddy current density as calculated from the common approximation 

formula. ?:>* 

In Fig. 5 it is evident that the eddy current densities at a distance oi 0.6r troin 
the center of the cylinder are the same at 100 / as at 1 f a . However, the eddy 



GiH approx. 


J0 50 





1000 f/ f 

Fig. 6. The penetration depth / at which field strength and eddy current density 
are l/e = 37 percent of surface values, as a function of frequency ratio ///. Solid 
curve H shows field strength. Solid curve G shows eddy current density. Dotted 
curve shows penetration depth calculated by approximation formula. Frequency 

ratio, ///. 


current density at the surface of the cylinder is 20 times larger than at 1 /. Con 
sequently the response to surface cracks is greatly reduced at low /// f/ ratios 
(low test frequencies) because of the low eddy current density at the surface. 

Control of Eddy Current Magnitudes. Fig. 4 shows the amplitude of ecldy 
current densities for various frequency ratios and radial positions in the cylinder. 
In this case, with an applied field strength of 1 oersted (produced by the coil in 
the absence of the test object), the current density at the surface of the cylinder 
would amount to 7.7 amp./cm. 2 . If the magnetizing field strength were increased 
to 100 oersteds, the eddy current density would also increase a hundredfold. Thus 
Fig. 4 makes it possible to determine at each test frequency the permissible field 
strengths which do not exceed practical limits beyond which the test objects 
would be heated by excessive eddy current density. A temperature rise during 
testing could result in changes in electrical conductivity in the test object, which 
might result in erroneous test indications. 

of the test object upon the test coil is determined entirely by the ratio of the test 
frequency / to the limit frequency f ff} as indicated by Eqs. (5) and (16), Eddy Cur 
rent Test Principles. As indicated in Fig. 7 in the section on Eddy Current Test 
Principles, the critical eddy current parameter ^ the effective permeability, 
is determined entirely by the magnitude of the frequency ratio ///.. Similarly it is 
indicated in Figs. 3 and 4 here that the geometrical distributions of' field 
strengths and of eddy current densities within the test cylinder are functions 
only ot f/J g . These conditions lead to the fundamental similarity law for eddv 
current tests, as follows : 

Qti^ aS - We11 aS the S eometri cal distributions of the field 
strength and eddy current densities, is the same for two different test objects if the 
frequency ratio /// is the same for each test object 


From Eq. (5), Eddy Current Test Principles, it is evident that this condition of 
similarity is met if 

/jUrol.iCWi 2 = /.^re,.^;. 2 (3) 

where the subscripts 1 and 2 refer to the properties of test objects 1 and 2, re 
For example, this similarity law indicates that: 

1. An aluminum cylinder with an electrical conductivity a = 35 meter/ ohm-mm. 2 , 
with the diameter d = 10cm., whose limit frequency j g = 1.45 c.p.s., tested at 
a frequency of 145 c.pjs., so that /// = 100. 

exhibits the same effective permeability and geometrically similar distributions of 
field strength and eddy current density as: 

2. A thin iron wire with an electrical conductivity a = 10 meter/ ohm-cm. 2 , a rela 
tive peremability M-rei. = 100, with a diameter d = 0.01 cm., whose limit 
frequency ] = 50,660 c.p.s., tested at a frequency of 5.07 Me. per second, so 
that jjjg also equals 100. 

Application to Detection of Discontinuities. The similarity law provides a 
basis for the experimental determination of the effects upon the apparent im 
pedance of the test coil of cracks having a specific depth, shape, and location. 
Although the influence of the physical properties and dimensions of cylindrical 
test objects could be determined mathematically, it is not possible to calculate the 
effects of defects having specific depth, shape, and location, upon the apparent 
impedance of the test coil. The similarity law, applied to discontinuities in the 
material, states: 

Geometrically similar discontinuities (such as cracks with specific depth and width 
measured in percentage of cylinder diameter) will result in the same eddy current 
effects and in the same variation of the effective permeability if the /// ratio is the 

Thus a crack of 0.0005-cm. depth in a tungsten wire of 0.01-cm. diam. would 
result in the same variation in effective permeability as a crack of 0.5-cm. depth 
in an aluminum cylinder of 10-cm. diam., or also a crack of 0.005-cm. depth in an 
iron wire of 0.1-cm. diam., if, for all three examples, the same f/f ff value were 
chosen by means of a suitable selection of test frequencies. 

Application to Model Tests. If model measurements are once made with 
artificial discontinuities for the entire f/f p range, to determine the variation of 
effective permeability as a function of crack depth, crack shape, and crack loca 
tion, the similarity law indicates that: 

The variations in the effective permeability caused by discontinuities obtained by 
model measurements can be transferred quantitatively to any given test object of 
cylindrical shape. 

This technique will be discussed later. 

CULATIONS. The similarity law can be used in experimental verification of the 
preceding mathematical indications which are basic in eddy current testing. For 
these tests a glass cylinder filled with mercury, with a field test coil wound on 
the outer surface of the cylinder, can be used (Fig. 7) . The field strength within 
the metallic mercury cylinder is measured by a very small microcoil which is 



moved along the radius of the cylinder by means of a micrometer. The a.-c, field 
strength measured by the microcoil appears as a voltage across the coil. This 
signal voltage is split into two components by two phase-controlled amplifiers. 
The two components of the voltage are then applied to the deflection plates of a 


Institut Dr. Foerster 

Fig. 7. Mercury model for experimental verification of field strength and eddy 
current distribution in cylindrical test object. 

cathode-ray tube. The screen display of the cathode-ray tube reproduces the 
field strength plane (see Fig. 3), including both phase 'and amplitude. This 
arrangement permits direct experimental reproduction of the field strength dis 
tributions calculated by means of Bessel functions. 

Fig. 8 shows the results of field strength measurements as a function of the 
radial position of the probe coil in the mercury tube. The experimental points are 
plotted as small circles, whereas the calculated results are shown as smooth 
curves. The high degree of correlation between experimental points and calculated 
values indicates the reliability of the mathematical calculations. 

math, calculation 

o model test measured 

15 mm 

Institut Dr. Foerster 

Fig. 8 Comparison between calculated (solid lines) and experimentally deter 
mined (o) field strength distribution in mercury model cylinder; for two test fre 



DISCONTINUITIES. Mercury is an ideal material for measurements of the 
effects of discontinuities because any given discontinuity can be easily represented 
by the insertion of a suitable shape of insulating material. Fig. 9 illustrates the 


Hg artificial defect 

Institut Dr. Foerster 

Fig. 9. Model arrangement for simulating any given discontinuity in the mercury 


model test arrangement. A coil is wound on a glass cylinder, Insulating bodies 
of any desired shape can be inserted within the mercury-filled glass tube. The 
position of these insulating bodies, representing artificial defects or discontinu 
ities, can be changed accurately with the aid of micrometer drives. During inser 
tion of the artificial discontinuities, the mercury can be emptied from the glass 
cylinder by means of a movable storage reservoir. 

Fig. 10 shows the electric circuit usedior precise measurement of the effects of 
cracks on the apparent impedance of the test coil throughout a wide frequency 
range. The bridge arrangement, fed by a variable-frequency generator, contains 




Hg Cylinder 

Institut Dr. Foerster 
Fig. 10. Electric circuit used in model tests with mercury cylinder. 



a precision resistance decade in one arm and a precision capacitance decade in 
series with the mercury-filled test coil in the other arm. If the variable capacitance 
and resistance arms are balanced so that zero voltage appears at the input to the 
amplifier in the diagonal arm of the bridge, the following relations exist : 

1. The reactance of the test coil, coL = 2jt/L, is equal to the capacitivo roac-tance 
of the decade capacitor, l/co(7, where / is the test frequency. 

2. The resistance of the test coil, R, is equal to the resistance- of the variable 
resistor decade, R D . 

The corresponding values for the test coil without mercury arc coL and R (} (ompty 
coil values) . From Eqs. (16, a and b), Eddy Current Test Principles, the mercury- 
filled cylinder, prior to insertion of insulating discontinuities, is described by 


t ZLO 



If a discontinuity of specific magnitude, location, and shape is simulated in the 
mercury by the insertion of a piece of plastic, a voltage appears across the 
diagonal of the measurement bridge of Fig. 10. This voltage is again adjusted to 


kind of defect 


character of defect 



' surface crack 






inner crack 


O O 


O O 






Institut Dr. Foerster 

Fig. 11. Summary of types of discontinuities in cylindrical specimens studied in 

mercury model tests. 


zero by varying the capacitive and resistive arms of the bridge, to obtain values 
ol (. k and tf crack . The vanation of the effective permeability Am* , result- 

crac . cve permeaty 

" 1 PrCSenCe thC CfaCkl iS alCUlated by means Of 


. ,r> 

A lAeff.dmtf.) = CKcrack - ) (5b) 

In addition to the known fill factor for the coil arrangement, Eqs. (5 a and b) 
contain only factors measured very exactly with the precision condenser and 

tT nm ge ClrCUit> ^ g ' n illustrates the sc P e of a P' ^ <* *ort 

than 50 000 measurements to determine the influence of given discontinuities 
upon the effective permeability of cylindrical specimens over the entire ran-e of 
frequency ratios. Once obtained, these results can be transferred to any given 
material and test cylinder in accordance with the similarity law. The results of 
these model measurements indicate if and how discontinuities with specified shape 
depth, and location will be indicated by eddy current tests. 

Test Coil Arrangements for Feed-through Testing 


MENTS. Various coil arrangements are used in the field in eddy current testing 
where cylindrical test objects are inserted into the coil for measurement Coil 
types include: 

1. Single primary coil, Fig. 12(a). 

2. Primary and secondary coils, Fig. 12(b). 

3. Bridge arrangement of two primary coils, Fig. 12(c). 

4. Bridge arrangement of two primary and two secondary coils, Fig. 12 (d). 

5. Comparison arrangement of two primary coils, Fig. 12(e). 

6. Differential arrangement of two primary and two secondary coils, Fig. 12 (f). 

With these various coil arrangements, quantitative measurements can determine: 

1. The absolute value of electrical conductivity a. 

2. Phase and amplitude variations caused by diameter or conductivity variations 
or by a surface crack of specific depth. 

These quantitative measurements are possible even if the coil data (fill factor ri) 
or the electrical conductivity a of the test cylinder are unknown. 

THROUGH COILS. Fig. 13 shows the apparent impedance plane, or the com 
plex voltage plane, for the various coil arrangements of Fig. 12. It is constructed 
for the assumed case in which the fill factor 7] = 0.75 and the frequency ratio 
///0 = 9- As indicated previously, the same complex plane diagram applies 

1. The ordinate represents the normalized inductive reactance, coL/coLo, and the 
abscissa represents the normalized coil resistance R/toL , or 

2. The ordinate represents the imaginary component of the normalized coil 
voltage, Eimzg./Eo, and the abscissa represents the -real component of the 
normalized coil voltage, E T6 ai/Eo. 

Single Primary Test Coil. Fig. 12 (a) shows the very simple case in which the 
test coil consists of a single primary winding. In this case, the vector OP of Fig. 



co/.*/? Sample 

P S ^Sample 

V.*-S, / 























1 &1 




P L 

J * 



O O 1 




S<> 9 


s [ 



2 \ 


1 ^ 





Institut Dr. Foerstcr 

) Single primary test 



13 represents the apparent impedance of the empty coil or the complex voltage 
across the empty coil. When the test cylinder is inserted, the new apparent im 
pedance vector OP appears across the coil. 

Primary and Secondary Coil Arrangement. For the primary and secondary 
coil arrangement of Fig. 12 (b), the distance OP in Fig. 13 represents the second 

Fig. 13. Complex voltage and impedance planes for test coil arrangements shown 

in Fig. 12. Fill factor TJ = 0.75. Frequency ratio j/j 9 = 9. Relative permeability 

M-rel. = 1. 

ary coil voltage in the absence of the test object. The vector OP represents the 
secondary coil voltage with the test cylinder inserted into the coils. 

Bridge Arrangements of Test Coils. In the bridge arrangements with two 
primary coils [Fig. 12 (c)] or two primary and two secondary coils [Fig. 12 (d)], 
no voltage appears across the diagonal of the bridge circuit or across the differ 
entially connected secondary coils in the absence of the test cylinder. The 
bridge is balanced, or the two secondary coil voltages cancel each other in the 


"empty state." When the test cylinder is inserted into either of these coil arrange 
ments, the voltage vector P P appears in Fig. 13. 

If another test cylinder with the same physical properties is now inserted into 
the second (empty) coil, the voltage across the bridge terminals, or at the output 
of the differentially connected secondary coils, disappears. If the physical prop 
erties (for example, the conductivity) of one test specimen vary from those of the 
comparison specimen, their effective permeabilities will also differ. In this case a 
voltage difference will appear across the terminals of the bridge of Fig. 12(c) or 
the secondary coils of Fig. 12(d). This voltage difference, AE, will be related 
to the difference in effective permeabilities, A|A C ff. = [AI cf f. fx 2 off ., as follows: 

) (6) 

where r\ is the fill factor of the coil and E G the coil voltage in the absence of the 
test object. But to draw conclusions concerning the physical cause of the signal 
voltage in this case, the following information must be determined in advance: 

1. The fill factor r) of the coil and test cylinders. 

2. The effective permeability ^tt. of the test cylinder. 

3. From these, the frequency ratio /// used in the test. 

The variation in effective permeability caused by a specific conductivity, diameter, 
or crack effect is a function of the specific frequency ratio used in the test, which 
in turn depends upon the effective permeability of the test cylinder. With known 
values for a, f,i reL , and d, the frequency ratio f/f ff can be determined from Figs. 
7 or 8 in the section on Eddy Current Test Principles. 

Differential or Self-Comparison Arrangements. In the bridge arrangements 
of Figs. 12(c) and (d), the second test cylinder served only for compensation of 
the fundamental voltage (? P) of Fig. 13. The test coil arrangements of Figs. 
12(e) and (f), on the other hand, employ a different portion of "the same cylin 
drical test object as a standard of comparison. This is often called a "self- 
comparison" arrangement. In this case a voltage appears at the output terminals 
of the coils only if the local effective permeability of the test cylinder differs 
at the locations of the first and second test coils. If the test cylinder is inserted 
into only one of the two coils, the voltage (P P) of Fig. 13 will again appear at 
the coil output terminals. 

_ This "self-comparison" arrangement is often used in crack-test instruments, 
since the crack depth normally varies from location to location along the test 
cylinder. (Note, however, that a long, uniform crack is indicated only at its ends 
and not along the uniform, continuous portions.) Variations in physical properties 
(such as conductivity, relative magnetic permeability, and diameter) from one 
test cylinder to another have no effect with these coil arrangements, since their 
influence on the effective permeability is compensated by the differential connec 
tions of the double test coils. Slowly varying physical properties in the test 
cylinders are also suppressed if the two self-comparison coils are close to each 
other. The fact that cracks of constant depth cannot be indicated is a limitation 
oi the self-comparison test method, which will be discussed later in more detail. 

Tests for Cracks in Solid Cylinders 

14 through 17 show apparent impedance variations resulting from cracks of 
various depths and radial locations in nonferromagnetic cylinders for frequency 



ratios of 5, 15 50, and^lSO. The frequency ratio ///, = 5 is particularly useful, 
since it permits detection of surface and subsurface cracks with the same 

As indicated in Eq. (17), Eddy Current Test Principles, the components of the 
effective permeability are equal to those of the apparent impedance or those of 
the secondary coil voltage if the fill factor Tj = l. Point in Fig 14 corre 
sponds to the case of a nonferromagnetic cylinder without a crack. This point 
appeared on the effective permeability curve of Fig. 7 in the section on Eddy 
Current Test Principles at the location marked ///, = 5. Thus Fig. 14 represents 

of crack in % of diameter 
30 P 

-0,1 -0,08 -0,06 -0.04 -0,02 


Inatitut Dr. Foerster 

Fig. 14. Impedance variations caused by surface and subsurface cracks in non- 
ferromagnetic cylinders, at a frequency ratio f/j t = 5. 

the area near f/f ff = 5 in Fig. 7 (section on Eddy Current Test Principles) but on 
a considerably enlarged scale. Figs 15, 16, and 17 are corresponding enlargements 
of the areas near f/f g ratios of 15, 50, and 150 in Fig. 7 (section on Eddy Current 
Test Principles) . Figs, 14 through 17 permit direct determination of secondary 
coil voltages corresponding to surface or subsurface cracks at specific depths 
and locations. 

Test Coil Voltages. Fig. 18 illustrates a pair of test coil units consisting of two 
primary and two secondary coils. When a nonferromagnetic cylinder free from 
cracks is placed through both coils, no voltage appears at the output terminals of 



lustitut Dr. Foerster 

Fig. 15. Impedance variations caused by surface and subsurface cracks in non- 
ferromagnetic cylinders, at a frequency ratio /// = 15. 

the secondary coils. However, as soon as a crack appears in one of the two test 
coil locations, the differential voltage A# appears. The normalized components 
of the crack signal voltage are 




Here E is the voltage of the empty secondary coil, and TI is the fill factor of the 
secondary coil (see Eq. 10, Eddy Current Test Principles) . The factors A|x e f Mre ai) 





0.04 wL 

Institut Dr. Foerster 

Fig. 16. Impedance variations caused by surface and subsurface cracks in non- 
ferromagnetic cylinders, at a frequency ratio f/J, = 50. 

and Aj^eff.dmag.) are shown as coordinates in Fig. 14. For a surface crack of 30 
percent depth, point P has coordinates of 

An, 6 .<r ca i> = 0.11 
Ap.ff.cim.f.> = -0.05 

These factors, when inserted in Eqs. (7, a and b) for a given crack, indicate the 
corresponding test coil voltages for this crack. 

Suppression of Variations in Cylinder Diameter. Particular attention is 
directed to those components of the variation in effective permeability which are 
perpendicular to the direction of diameter variations in the complex impedance 
plane Those physical effects of the test object which produce displacements in the 
impedance plane in the same direction as changes in cylinder diameter cannot be 





0.0 5 J loJIlLJilV ' ^/^I^\CRACK 



0.05 wL < 

Jnstitut Dr. Foerster 

Fig. 17. Impedance variations caused by surface and subsurface cracks in non- 
ferromagnetic cylinders, at a frequency ratio ///, = 150. 

Institut Dr. Poerater 

Fig. 18. Differential test coil arrangement containing sound and defective 
cylindrical test specimens. 



separated from the ever-present interfering variations in diameter. However, 
as will be shown later, these interfering diameter effects can be entirely sup 
pressed if test indications can be limited to those which are perpendicular to the 
direction of diameter variations in the impedance plane. For example, in Fig. 14, 
the vector PD represents the change in effective permeability or apparent im 
pedance normal to the diameter direction for a crack of 30 percent depth, tested 
at a frequency ratio f/f g = 5. 

Frequency Selection for Surface Crack Detection. Fig. 19 presents an 
evaluation of the impedance plane data of Figs. 14 through 17. The ordinates 

depth of crack 
in % of diameter 

700 750 

Fig. 19. Magnitude of changes in effective permeability |Anofr.|, caused by surface 
cracks of various depths in nonferromagnetic cylinders, as a function of frequency 

ratio, ///. 

represent the values of the effective permeability variations |Api eff J resulting from 
surface cracks. The abscissae show the frequency ratio f/f g . Various crack depths 
are shown by different curves. For example, the data for 30 percent crack depth 
correspond to the distance OP in Fig. 14 for f/f g = 5. 

Fig. 20 illustrates the magnitudes of surface crack signals in a test instrument 
which suppresses the interfering influence of diameter variations. Here the 
ordinates represent only the component of the change in effective permeability 
perpendicular to the direction of diameter changes on the complex impedance 
planes. Thus, for the evaluation of surface cracks in nonferromagnetic cylinders, 
independently of variations in cylinder diameter, the optimum frequency range 
as indicated in Fig. 20 is for f/f ff ratios between 10 and 50. 



depth of crack, 
in % of diameter 

Fig. 20. Magnitude of changes in component of effective permeability |Anf.| 

perpendicular to direction of diameter changes, caused by surface cracks of various 

depths in nonferromagnetic cylinders, as a function of frequency ratio ///,. 

Frequency Selection for Subsurface Crack Detection. Fig. 21 provides a 
similar evaluation for subsurface cracks in nonferromagnetic cylinders. A crack 
depth of 30 percent of the cylinder diameter is taken as an example in this illus 
tration. However, families of curves for any other specific crack depths can be 
derived from these model curves. The ordinates in Fig. 21 represent magnitudes 
of effective permeability variations |A[A e ff.| resulting from subsurface cracks. The 
abscissae indicate the depth of cracks from the outer surface of the cylinder, in 
percentage of cylinder diameter. A crack with zero-percent distance is a surface 
crack. The great decrease in magnitude of crack signals with increasing depth 
from the surface, particularly at high frequency ratios, is clearly indicated in 
Fig. 21. 

Fig. 22 is analogous to Fig. 20 in that it shows only the component of the change 
in effective permeability whose direction is perpendicular to the direction of 
diameter variations in the complex impedance plane. Thus Fig. 22 corresponds 
to the magnitudes of crack signals obtained in instruments which suppress diam 
eter variations. The curve for f/f ff = 5 is important. It illustrates that in the 
range of this f/f ff value, crack indications on instruments which suppress diameter 
variations are practically independent of the distance of the crack below the 
surface of the test cylinder, provided this distance is not excessive, This fortunate 
independence of depth in the crack indication results from the fact that varia 
tions m effective permeability for subsurface cracks result in phape rotations in 
the^ complex impedance plane into the direction perpendicular to the diameter 
variations. Thus the component of |Au. eff .| perpendicular to the diameter direc- 



subsurface crack 30 % depth 

2.5 5 7.5 
% of diameter 


Fig. 21. Magnitude of changes in effective permeability |Ajie.| caused by sub 
surface cracks of 30 percent crack depth at various distances below the surface of 
nonferromagnetic cylinders. Frequency ratios shown on curves. Distance below 
surface given in percentage of cylinder diameter. 

tion remains practically constant with increasing distance of cracks beneath the 
surface, whereas the magnitude of Afx ef f. itself decreases considerably, as shown 
in Fig. 21. Consequently the optimum test frequencies for subsurface crack 
detection correspond to frequency ratios f/f g which lie between 4 and 20. 

ysis of cracks in nonferrous cylinders can be extended to the case of ferromagnetic 
cylinders, at least in so far as the magnitude of the variation in effective perme 
ability (see Fig. 19) is concerned. For a ferromagnetic cylinder the signal voltage 
components of Eqs. (7, a and b) need only be multiplied by the relative perme 
ability jj, rel . to apply to the ferromagnetic cylinder. For example, if pi rel . equals 
100, the secondary voltage signal resulting from the presence of a specific crack 
is 100 times larger than for the nonferromagnetic cylinder, provided both tests 
have been made at the same frequency ratio f/f g . 

Suppression of Variations in Cylinder Diameter. The directions of diameter 
variations are entirely different for ferrous and nonferrous materials in the com- 




subsurface crack 30% depth 

2,5 5 7,5 
% of diameter 

Fig. 22. Magnitude of changes in component of effective permeability |A|i f f.| 
perpendicular to direction of diameter changes, caused by subsurface cracks of 
30 percent crack, depth at various distances beneath outer surface of nonferro- 
magnetic cylinder. Frequency ratios shown on curves. Distance* below surface given 
in percentage of cylinder diameter. 

plex impedance plane (compare Figs. 1 and 2). Fig. 23 shows the component of 
the variation in effective permeability, perpendicular to the direction of diameter 
variation, for ferromagnetic cylinders whose relative magnetic permeability 

Hrel. I- 

Frequency Selection for Crack Detection. In contrast to the situation with 
nonferrous cylinders, the directions of diameter and of crack effects form a large 
angle with each other at low frequency ratios in the complex impedance and 
voltage planes. Therefore, to separate crack and diameter effects for steel cyl 
inders, the optimum frequencies correspond to f/f g ratios of less than 10. How 
ever, for ratios of f/f g < 10, the interfering effects of both diameter and perme 
ability variations (such as result from cold working or straightening) have the 
same direction, perpendicular to the direction of crack effects. Thus these two 
interfering effects are simultaneously separated from the indications of cracks in 
crack-detection instruments. The preceding observations refer only to the funda 
mental wave. Effects caused by the appearance of harmonics (hysteresis loops) 
will be discussed in subsequent sections. 


d component 


10 15 

Fig. 23. Magnitude of changes in component of effective permeability |A|ie.| 
perpendicular to the direction of diameter and relative permeability changes, 
caused by surface cracks of various depths in ferromagnetic cylinders, as a func 
tion of frequency ratio ///,. 

crack model tests have made it possible to measure crack depths quantitatively 
and to calibrate eddy current test instruments in terms of crack depth and loca 
tion. Alternatively, the test instruments can be used to measure variations in 
diameter or electrical conductivity or the value of the electrical conductivity by 

Experimental Determination of Frequency Ratio f/f g . The frequency ratio 
f/fg can be determined experimentally, even with a fixed-frequency test instru 
ment. When a cylindrical test specimen is inserted into the measurement coils 
of Fig. 12(d), for example, the voltage vector P P appears in the complex voltage 
plane of Fig. 13. This voltage vector is characterized both by its magnitude and 
by the angle a, which the new vector P P makes with the voltage vector OP of 
the empty test coil. Fig. 24 shows the relationship between the angle a and the 
frequency ratio f/f ff for the case of nonferromagnetic cylinders. If the test 
instrument has a fixed operating frequency, say, 1000 c.p.s. (1 kc.), the f/f g scale 
of the instrument can be replaced by a scale reading in terms of the product ad 2 
of the electrical conductivity and the square of the cylinder diameter. In the 
case of a cylinder of unknown conductivity, the diameter can be measured with 
a caliper, and the conductivity can be determined directly from the angle a and 
Fig. 24. This measurement of conductivity is independent of the properties of the 
test coil or fill factor r|. 





./ KC P S) 

Fig. 24. Variation of angle a with frequency ratio /// or product ar/ 2 in nonferro- 

magrietic cylinder. 

Crack Evaluation with Comparison Specimen. In normal testing with a 
comparison standard specimen, the standard (crack-free) specimen is placed in 
one of the two test coils and the unknown specimen in the other coil. The com 
parison standard serves to compensate the voltage P P of the unknown cylinder 
(see Fig. 25) . If the two cylinders are equivalent, the differential connection of the 
secondary coils causes the two vector voltages P P and PP to cancel each other, 
so that the output voltage is zero. However, if the unknown specimen contains a 
crack, the output voltage vector of its test coil will change from PP to CP , for 
example (see Fig. 25). Now the voltage vectors of the crack-free standard 
(P P) and the cracked specimen (CP Q ) no longer cancel. Instead, a complex 
voltage signal, hE/E Q = PC, appears at the terminals of the differentially con 
nected secondary coils. 

If the test cylinders do not fill the test coils but have, for example, a fill factor 
T] = 0.75, the following important relationship is obtained from the similar 
triangles of Fig. 25: 





where point P l lies on the curve for r\ = 1. Here the vector P^ corresponds to 
the variation in effective permeability, A[X e ff., caused by a crack, as described pre 
viously (see Fig. 14) . These crack effects are shown in Figs. 14 to 17 for a fill 
factor T| = 1. It follows from Eq. (8) that 

PC = Tl(A|leff.Ccraclo) PoP = 

From Eqs. (8) and (9), one obtains 









Fig. 25. Explanation of complex plane conditions for comparison specimen test 


The vector P P represents the voltage which appears when one coil contains a 
cylinder with the fill factor i] while the other coil is empty. (This quantity is 
known in European literature as the absolute value of the test cylinder.) From 
Eq. (10), the unknown fill factor of the coil can be determined from the quo 
tient of the measured effect (for example, the variation PC caused by a crack) 
and the "absolute value" (vector P P) . The variations in effective permeability 
found in crack model tests (Figs. 14 to 17), divided by the magnitude of the 
vector P P (connecting P to the corresponding f/f ff point PI on the r\ = 1 curve), 
correspond to the ratio of the measured effect (PC) to the "absolute value" 
(P P), both measured by the test instrument. 



Conductivity or Diameter Evaluation with Comparison Specimen. A simi 
lar technique is used in analysis of conductivity or diameter variations. The Au.,, ff . 
values (from Fig. 1) are divided by the distance from point P,, In tho correspond 
ing f/fg values on the u,>. curve of Fig. 13 (i.e., by (he "absolute value"). 

Instrument Sensitivity Calibration. Eddy current tost instruments (such as 
those manufactured by the Institut Dr. Focrstcr in Reutlingon, West Germany) 
can be equipped with a sensitivity switch indicating what percentage of the 
"absolute value" (P P) corresponds to a 1-in. deflection on the cathode-ray screen 



c/epff) of crack in % 
of diameter 

,30 % 






Fig. 26. Quotient Q = A M, ff./( absolute value) as a function of frequency ratio f/j a 

for surface cracks of various depths in nonferromagnetic cylinders. Crack depth in 

percentage of cylinder diameter. 



for any given sensitivity step. The frequency ratio f/f ff can also be read from the 
phase-shifter of the instrument through a method described later. These 
calibrations make the test method quantitative in measuring test-object properties. 

Quantitative Crack-Depth Calculations. Fig. 26 can be used as an example 
for calculating crack depth in nonferromagnetic cylinders from the data of Fig. 19. 
The ordinate in Fig. 26 corresponds to the ratio 

Q = 

(absolute value) 

of the variation in effective permeability to the "absolute value" (vector P P) . 
Fig. 26 shows curves for Q for cracks having depths of 10, 15, 20, 25, and 30 
percent of the cylinder diameter. 

depth of crack 
30 in % of diameter 

Fig. 27. Quotient Q = Ajief*./ (absolute value) as a function of crack depth, for 
various frequency ratios f/f g , with nonferromagnetic cylinders. 

To calculate the deflection corresponding to a crack of 10 percent depth, 

for example, at a given instrument sensitivity setting, the steps are: 

1. The frequency ratio ///, of the test object is determined from the ///, scale 
(phase shifter) of the test instrument. (For this example assume that the value 

2 Determine 7 Q (percentage ratio of AM*. to the "absolute value") from Fig. 26. 
(For the assumed 10 percent crack depth, the ordinate at point B indicates 
that Q = 2.9 percent.) 

3 - Determine the sensitivity setting of the test instrument from the sensitivity 

control knob setting. (For example, the knob may indicate N = 2 percent pel- 
inch, in which case & deflection of 1 in. on the cathode-ray screen corresponds 
to 2 percent of the "absolute value.") 



4, To obtain the deflection corresponding to a 10 percent crack depth at this 
sensitivity setting, divide the value of Q by N. (Thus a crack of 10 percent 
depth would be indicated by a deflection of Q/N = 2.9/2 = 1.45 in.) 

Computing Crack Depth from Instrument Indication. In normal testing, 
a deflection appearing on the screen of the test instrument must bo converted into 
crack depth. Fig. 27 is used for this conversion. The steps involved are: 

1. Read the instrument deflection. (For example, an unknown crack may be 
indicated by a deflection A = 2.5 in. when the sensitivity control setting is 
N = 1 percent per inch and the phase-shift control indicates a frequency ratio 

iff, = 15.) 

2. Multiply the deflection A by the sensitivity setting JV. (In this case, N X A 
= (1%/inch) X (2.5 in.) = 2.5% of the "absolute value. 1 ') 

3. On Fig. 27 find the intersection of the ordinate (Q = 2.5%) with the crack- 
depth curve (for /// ff = 15). Read the corresponding crack dopth on the 
abscissa scale. (In this example point D corresponds to a crack dopth D = 9%.) 

The preceding explanation applies to surface cracks in nonferrous cylinders. The 
position and depth of subsurface cracks may be established also, 42 

Applications of Test Instruments 

^ OPERATION OF TEST INSTRUMENTS, Fig. 2S shows a simplified 
circuit diagram for a typical ellipse-method eddy current instrument (Sigmaflux 
or Wire Crack test instrument developed by Institut Dr. Foerstcr) . An oscillator 
feeds the two pairs of test coils (PA, P 2 *So) with the magnetizing field current. 
The secondary windings of the test coils act on the amplifier. The amplifier sen 
sitivity control N indicates the percentage of the "absolute value" corresponding 
to a 1-in. deflection on the cathode-ray screen. The amplifier output acts on the 
vertical deflection plates of the cathode-ray tube. The primary magnetizing field 
current acting through transformer T induces a secondary voltage whose phase 
angle corresponds to the phase angle of the voltage of a secondary test coil in the 

j[ Screen 

Phase shifter 

Institut Dr. Foerster 

Fig. 28. Simplified schematic diagram of eddy current test instrument for quan 
titative measurement of crack effects, conductivity effects, and diameter effects in 

nonferrous cylinders. 



absence of a test object. The secondary voltage of the transformer T is applied to 
a precision phase-shifter whose phase angle is indicated exactly. The output 
voltage FG of the phase .shifter is applied to the horizontal deflection plates of the 
cathode-ray oscilloscope. 

Analysis of Screen Patterns. Before a cylindrical test specimen S is placed in 
test coil PI&I, no voltage is applied to the vertical plates of the oscilloscope. A 
horizontal straight line appears on the screen. Insertion of the test specimen pro 
duces a voltage on the vertical plates. This voltage is proportional to the vector 
P Q P of Fig. 29. It has the phase angle a, measured between the horizontal and 
vertical deflection voltages, if the phase-shifter is adjusted to deg. (the direction 
of the normalized inductive reactance coI//coZ/ ) . Two sinusoidal voltages of single 
frequency, displaced in phase by the angle a, form an ellipse on the cathode-ray 
screen. The character of the ellipse has special significance in eddy current testing. 

? P _, 

Fig. 29. Characteristics of elliptical screen pattern, resulting from sinusoidal 
deflection voltages displaced in phase angle, on eddy current test instrument. 

Fig. 29 serves to explain the most important characteristics of the elliptical 
screen pattern. The vector voltage P P appears as the maximum vertical deflec 
tion of the ellipse when a test cylinder is inserted into the measurement coil of 
Fig. 28. The maximum horizontal deflection of the ellipse has the amplitude P (} K. 
The vertical section PR from the screen center to the upper side of the ellipse is of 
special significance. This is the component of the vertical measurement voltage 
perpendicular to the direction of the horizontal base voltage. It permits separa 
tion of diameter and crack effects. When the voltage at the horizontal plates 
lies exactly in the diameter direction (on the effective permeability plane), diam 
eter effects acting on the vertical plates can appear on the screen only as in 
clined straight lines. (This corresponds to no phase displacement between the 
vertical and horizontal deflection voltages.) In this case, widening or splitting of 
the ellipse (PR in Fig. 29) cannot occur, Thus the splitting (PR), if it appears 
in the test indications, can be caused only by cracks or conductivity variations. 



The components of crack and conductivity effects which lie perpendicular to the 
direction of diameter changes on the effective permeability plane (see Fig. 14) 
are significant for separation of diameter effects from quantitative measurements 
of crack and conductivity effects. 

Adjustment of the Phase Shifter. The elliptical screen pattern which appears 
when the test cylinder is inserted into the test coil shows two important character 
istic factors, as follows: 

1. The magnitude of the "absolute value" of the specimen (vector PoP). 

2. The angle a, 

With these factors known, Fig. 24 can be used to determine the frequency ratio 
///, and at a given test frequency /, the product ad 2 of the electrical conductivity 
and the square of the cylinder diameter. 

The angle a is adjusted with the precision phase shifter until the ellipse changes 
into a straight line. The phase-shifter angle at which the ellipse disappears can be 
determined exactly. The scale of the phase shifter in Fig. 24 is calibrated in terms 
of the product crd 2 for a test frequency of 1000 c.p.s. When the phase shifter is 
adjusted to convert the. ellipse to a straight line, the frequency ratio f/f ff and the 
product crcf 2 can be read directly from its scale. ,By measuring the cylinder diam 
eter with a caliper, the electrical conductivity can be determined. 

ooO +0 

XIO" 2 + 





X f 9 

s ' 

r L< 



. O 1- 




> / 







XIO" 1 



f\ Tl 



>v - 


XIO" 2 





Fig. 30.^ Impedance variations (amplitude and phase), for a 1 percent variation of 
conductivity, as a function of frequency ratio ///,. Nonmagnetic materials (M. = 1). 



+.05x10* +QI +REALAR 

~~ 1 ri ' L 


Fig. 31. Impedance variations (amplitude and phase), for a 1 percent variation in 
diameter, as a function of frequency ratio ///,. Nonmagnetic material (|i = 1). 

Instrument Sensitivity Characteristics. Figs. 30 and 31 show sensitivity 
diagrams for conductivity and diameter effects. These diagrams show the 
variations in effective permeability or complex impedance when conductivity or 
diameter of the test cylinder vary by 1 percent. Figs. 32 and 33 show the ampli 
tude of the variations in effective permeability for these 1 percent changes. The 
secondary coil voltage change, AE, resulting from an M-percent change in con 
ductivity or diameter effect is 


where r\ is the fill factor of the test coil (see Eq. 10, section on Eddy Current Test 
Principles), #0 is the voltage of the empty coil (BC in Fig. 28), and A^ e ff.<i%) is 
shown as a function of f/f ff in Figs. 32 or 33 for 1 percent variations in conductivity 
or diameter. 

DIAMETER EFFECTS. Unknown conductivity or diameter effects can be 
measured quantitatively, even with unknown fill factors, with calibrated eddy 
current instruments. The variation in effective permeability A^M. can be ob- 



Fig. 32. Variation in effective permeability Apioff., for 1 percent conductivity varia 
tion in nonferromagnetic cylinder, as a function of frequency ratio j/j g . 

tained from Figs. 32 and 33. Figs. 34 and 35 show the quotient Q = A ^./(abso 
lute value) as a function of f/f ff for 1 percent diameter or conductivity effects. 
The example here will illustrate the quantitative measurement of a given con 
ductivity or diameter variation even when fill factor, specimen conductivity, and 
test frequency are all unknown. 

Conductivity Measurement. When an unknown test cylinder is inserted into 
one of the two test coils connected in series opposition, an ellipse appears on the 
screen of the cathode-ray tube, as described earlier in this section. The precision 
phase shifter of Fig, 28 is adjusted until the ellipse forms ,a straight line. The f/f ff 


Fig. 33. Variation in effective permeability AM,,,,., for 1 percent diameter variation 
in nonferromagnetic cylinder, as a function of frequency ratio ///,. 





IQ-KT 2 

Fig. 34. Quotient Q = AM^*./ (absolute value), for 1 percent conductivity variation 
in nonferromagnetic cylinder, as a function of frequency ratio ///*. 

ratio is obtained from the resultant phase angle and Fig. 24. When another test 
specimen is inserted into the second coil, the voltage output of the opposed second 
ary coils again becomes zero. Now assume that a conductivity variation occurs 
during testing, resulting in a beam deflection A of 1.5 in. If the sensitivity knob 
of the instrument is set at N = 4% per inch, a 1-in. deflection corresponds to 4 per 
cent of the "absolute value" (P P in Figs. 13 and 25). Thus the measured effect 
has a magnitude of A X N = (1.5) X (4) = 6 percent of the "absolute value." 
Suppose that the phase shifter now indicates a frequency ratio f/f g = 10. From 
Fig. 34 the corresponding value of Q is found to be 0.47 percent for a 1 percent 






Fig 35 Quotient Q = A^ff./ (absolute value), for a 1 percent diameter variation 
in nonferromagnetic cylinder, as a function of frequency ratio ///,. 


variation in conductivity. Thus the variation in conductivity is found to be 
(6)/(0.47) = 12.6% in this example. More generally, 

A<y AN 

V (%)= -Q- (12) 

where Aa/a(%) = percentage change in electrical conductivity. 

A = deflection of beam (in inches) caused by 'the conductivity varia 

N = instrument sensitivity setting (given in percentage of the "abso 
lute value" for 1-in. beam deflection). 

Q = quotient read from Fig. 34 for the ///, ratio shown on the phase 

Diameter Measurement. The magnitude of a diameter variation can be 
measured in the same manner as a conductivity variation. When the cylinder 
diameter varies from that of the reference standard, the cathode-ray beam is again 
deflected in the test instrument. If the deflection A = 2.5 in. at a' sensitivity set 
ting N = 2% per inch, the diameter effect is N X A = (2) X (2.5) = 5% of the 
"absolute value." Suppose that the phase shifter indicates the frequency ratio to 
be t/fg = 20. Fig. 35 indicates that a diameter variation of 1 percent at'/// = 20 
results in a measurement effect of Q = 2.35 percent of the "absolute value." The 
diameter effect for the example in question is calculated to bo 

In the general case, 

Ad AN 

(%) =r - /-ION 

d Q 'W 

where kd/d(%) = percentage change in cylinder diameter. 

A = measured deflection on screen of eddy current teat instrument, 

caused by a diameter effect. 
N = instrument sensitivity setting (given in percentage of the "absolute 

value" for 1-in. beam deflection). 

Q = quotient read from Fig. 35 for the f/f, ratio shown on the phase 

Setting Tolerance Limits. If tolerances are specified for conductivity or diam 
eter variations, the beam deflections corresponding to these tolerance limits can 
be established for each sensitivity setting of the test instrument. Suppose that the 
specified tolerance for conductivity or diameter is T percent. The beam deflec 
tion corresponding to this tolerance percentage is then 

where Q = ttokisa from Figs. 34 or 35 for the f/f. ratio indicated on the 

c V^. trumen * s e( ^ ui PP ed with Baling or automatic sorting 
S g /T tS Can be * djUSted as indicated above for a specified con- 

tT ( Y S0rt ? ng) ' haidneSS tolerance ' or diameter Wenrnfle, for 
each sensitivity step, in accordance with Eq. (14). 

* ndependentl y of Diameter Variations. In practical test 

ing, crack effects and diameter variations often appear simultaneously (for exam 
ple, in tungsten and molybdenum wires). In this case the crack depth can still 



be determined quantitatively by means of the ellipse-analysis method (see 
Analysis of Screen Patterns in this section and Fig. 29). If the voltage on the 
horizontal deflection plates of the cathode-ray tube is adjusted precisely in phase 
angle to the direction of diameter variations (see Fig. 14), the intersection of 
the ellipse with the vertical center line (PR in Fig. 29) represents the com 
ponent of the crack signal perpendicular to the diameter direction on the complex 
permeability plane of Fig. 14. Fig. 36 represents the component of the quotient 
Q, perpendicular to the diameter direction (Fig. 27), as a function of frequency 
ratio f/f g . It provides a means of quantitative crack evaluation even when 
diameter variations are present in the test cylinder. This is true even though the 
eddy current effects, A[i e .> f diameter variations can be many times larger than 
those resulting from crack effects. 

Q . netf j_ d componen t 

Fig. 36. Component of quotient Q = A^etf. /(absolute value), perpendicular to 
diameter direction in effective permeability plane, as a function of frequency ratio 

f/f ff . Curves show crack depths independently of cylinder diameter variations. Crack 
depths in percentage of diameter of nonferrous cylinder. 

Crack Evaluation with Diameter Variations. The evaluation of cracks in the 
presence of diameter variations will be illustrated by means of an example. When 
a crack is present in the test cylinder, an ellipse will appear upon the screen of 
the indicating instrument. The voltage applied to the horizontal deflection plates 
is selected to conform with the direction of diameter variations. The vertical 
intercept (PR in Fig. 29) is read as A = 1.25 in., for example. Suppose that the 
sensitivity control indicates N = 4% per inch of "absolute value/' The product 
AN = 05 The frequency ratio, indicated by the phase shifter, is f/f ff = 13. The 
coordinates of Q = 0.05 and ///, = 13 (point B in Fig. 36) correspond to a crack 
depth of 22 percent. 


Reference numbers in this section refer to the references at the end of the 
section on Eddy Current Test Indications. 





Tests of Thin- Walled Tubes 

Significant characteristics of tubes 1 

Thin-walled nonferrous tubes 1 

Complex impedance or voltage plane 1 

Curve of effective permeability jUoff. as a 
function of frequency ratio ///// for 
thin-walled, nonferromagnetic tube fill 
ing the teat coil (/.I) 2 

Construction of frequency ratio scale 3 

Sample calculations for effective permeabil 
ity and secondary coil voltage 8 

Direct computation of components of effec 
tive permeability from frequency ratio, 

Mi 3 

Effect of changes in outside diameter of tube 
(with proportional changes in tube wall 

thickness) 3 

Normalized apparent impedance plane for 
thin -walled, nonferromagnetic tubes 
with varying outside diameters and a 
constant ratio of inside to outside d'am- 

eter (/. 2) 4 

Effect of changes in outside diameter of tube 

(with inside diameter held constant) ... 4 
Normalized apparent impedance plane for 
thin-walled, nonferromagnetic tubes 
with varying outside diameters and fixed 
inside diameter, and constant conductiv 
ity c (/. 3) 5 

Thin -walled ferromagnetic tubes 6 

Sensitivity of tests of thin-walled tubes 6 

Sensitivity diagram for eddy current tests 

of thin -walled tubes (/. 4) 6 

Maximum test sensitivity 6 

Experimental determination of frequency 

ratio ill 9 ' 

Variation of angle et with frequency ratio 
llh, or product (ffdiW), in nonferro 
magnetic tube (/. 5) 7 

Cracks in thin- walled tubes 7 

Frequency selection for crack detection 7 

Tests of Thick- Walled Tubes 

Thick-walled nonferrous tubes 8 

Range of variation of effective perme 
ability for thick-walled, nonferrous tubes 

(shown by shaded area) (/. 6) 8 

Effect of variations in inside diameter (with 
outside diameter held constant) 8 


Complex permeability or impedance plane 
for nonferromagnetic tubes of various 

wall thicknesses (/. 7) 9 

Sample calculation for variation of inside 

diameter 10 

Effect of variations in electrical conductivity 10 
Effect of variations of outside diameter 
(with ratio of inside to outside diameter 

held constant) 10 

Complex impedance or permeability plane 
for nonferromagnetic tubes with wall 
thickness of 20 percent of outside tube 
radius as outside diameter do of tube 

varies (/. 8) . 11 

Maximum test sensitivity 12 

Calculation of frequency ratio, ///?, for 

maximum test sensitivity 12 

Frequency ratios Jjf g for optimum test 
sensitivity in thick-walled tubes with 
various ratios of wall thickness W to 

outside radius r (/. 9) 12 

Cracks in nonferromagnetic tubes 12 

Quantitative evaluation of crack depths .;,. 12 
Apparent impedance plane showing effects 
of cracks for four different tube wall 
thicknesses, at a frequency ratio jljg 
= 5, for nonferromagneiic tubes (/. 10). 13 
Apparent impedance plane showing effects 
of cracks for four different tube wall 
thicknesses, at a frequency ratio f/f g 
= 15, for nonferromagnetic tubes (/. 11) 13 
Apparent impedance plane showing effects 
of cracks for four different tube wall 
thicknesses, at a frequency ratio ///$ 
= 50, for nonferromagnetic tubes (/. 12) 14 
Apparent impedance plane showing effects 
of cracks for four different tube wall 
thicknesses, at a frequency ratio jlfg 
150, for nonferromagnetic tubes (/. 13) 14 
Characteristics of crack evaluation curves . . 14 

Eddy Current Tests of Tubes With 
Internal Test Coil 

Applications of internal probes 15 

Internal probe test arrangements 15 

Test coil inside cylinder (/. 14) 15 

Heavy- walled tubes 15 

Arrangement for measuring the screening 

effect of tubes (/. 15) 16 

Decrease of field strength upon insertion 
of tube between roils of Fig. 15, for 



CONTENTS (Continued) 


copper, brass, and stainless steel tubes of 
0,8-in, outside diameter and 0.040-in, 

wall thickness (/, 16) 16 

Evaluation of thin -walled non ferromagnetic 

tubes 17 

Evaluation of thick-walled nonferromagnetic 

tubes 17 

Apparent impedance diagram of test coil 
inside heavy- walled nonferrous tube (/. 

17) 18 

Separation of material properties and inside 

corrosion thinning 17 

Maximum test sensitivity 17 

Evaluation of medium- thickness tube walls 17 

Penetration Depth Limitations 

Significance of penetration depth 19 

Penetration depths for different instrument 
sensitivities (10- 2 , 10-3, 10-4, an( j io-) 
as a function of frequency ratio flfg 

(M8) 19 

Significance of instrument sensitivities 19 

Limitations in detection of concentric cavi 
ties 20 

Determination of secondary coil voltages 

with tubes 20 

Magnetic permeability limitations on depth 

penetration 20 

Relative magnetic permeability of a car 
bon steel as a function of external mag 
netic-field strength (/. 19) 20 

Limitations in use of high magnetizing field 

Penetration depths for carbon steel tubes 
of 0.040, 0,080, and 0,120 in, wall thick 
ness (/.20) 22 

Shielding tests at 50-c,p.s, frequency 21 

Shielding tests at 5-c.p,s. frequency 22 

Tests of Small Diameter Tubing 

Principle of test 23 

Impedance plane characteristics 23 

Impedance plane for coil encircling non- 

ferromagnetic metal tubes (/, 21) 24 

Controlling factors 24 


Types of coil systems 25 

Test coil conh'Ktinitioib fur eddy current 
testing of smalNiliameter tubing (/, 22) 25 

Sensitivity and signal (liMtiminnf urn 28 

Differential coil system 26 

Absolute roil system 26 

Tube-feeding mechanism 27 

Encircling-coil assembly and feed mecha 
nism for tubing inspection (/. 23) 27 

Instrumentation 27 

Resistive-component sensing instruments ,,,, 28 
Resistance-loss test indications as a func 
tion of tubing wall thickness with ///r as 

a parameter (/. 24) 28 

Flaw sensitivity 29 

Eddy current signal trace and photo 
micrograph of defective U X 0.049-in, 

Hastelloy tubing (/, 25) 29 

Detection of intcrgnmiilai 1 corrosion 30 

Eddy current signal trace and photo 
micrograph of intergrunular corrosion on 
inside surface of Nimonic tubing (/. 26) 30 
Eddy current .signal truce and photomicro 
graphs of defective 3/16 X 0,025-in. In- 

conel tubing (/, 27) 32 

Sensitivity to small discontinuities 31 

Wall thickness and tube diameter 31 

Frequency selection 31 

Comparison of resistive and reactive com 
ponent signal traces with actual dimen 
sional variations in n 0,229 X 0.025-in. 

Inconel tube (/. 28) 33 

The surface probe-coil system 3,5 

Coil holder assembly 35 

Eddy current probe coil and holder for 
wanning nnall-diamater tubing (/, 29) ., 34 

Control of coil-to-tube spacing 36 

Inspection of tubing with an eddy current 

probe (/. 30) 35 

Advantages of probe-coil system 36 

Probe-coil eddy current B-sciui photo 
graphs (/, 31) 37 

Limitations in test automation 37 

Selection of signal-display system 38 

Eddy current signals from defective 3/16 

X0.025-in. Inconel tubing (/, 32) 38 

Application areas 39 

References 39 




Tests of Thin- Walled Tubes 

tubes in industry is increasing continuously. Typical applications include the 
construction of boilers and heat exchangers, reactors, refrigerating systems, 
chemical and oil refining plants, and many process industries. Therefore non 
destructive testing of large quantities of tubular products for discontinuities, 
eccentricity, wall thickness, alloy, hardness, and other properties is an especially 
important problem. However, additional difficulties are encountered in eddy 
current testing of tubes, since additional factors, not present in testing solid 
cylinders, influence tube tests. Among the characteristics and possible discon 
tinuities which must be considered in tube testing are: 

1. The electrical conductivity a. 

2. The relative permeability of ferromagnetic materials. 

3. The outside diameter d* 2r a , where r a is radius. 

4. The inside diameter di = 2n, where n is radius. 

5. Wall thickness W = W. - di)/2 = r. - n. 

6. Subsurface cracks of specific depth, d (in percentage of wall thickness). 

7. Outside-wall surface cracks of specific depth, Ca (in terms of wall thickness). 

8. Inside-wall surface cracks of specific depth, Cm (in percentage of wall thick 
ness) . 

Eccentricity, in percentage of the wall thickness, is given as the ratio of the thin 
nest wall thickness Tf min . to the nominal wall thickness W of the concentric tube; 

W mln, ^, IAA 

Eccentricity (%) = 


THIN-WALLED NONFERROUS TUBES. The thin-walled tube is an 
especially clear example for introducing the mathematical treatment of the eddy 
current problem. Its calculations can be carried out with elementary formulas. 
The two components of the effective permeability, or apparent impedance of 
the test coil in which the tube is placed, can be expressed by closed solutions. The 
simple mathematical relationships for the thin-walled tube are derived from the 
rather complicated system of formulas applicable to the tube with a given wall 

Complex Impedance or Voltage Plane. The basic curve for the effective 
permeability (shown for the solid metallic cylinder in Fig. 7 of the section on Eddy 
Current Test Principles) becomes an exact semi-circle with diameter d = 1 for the 
thin-walled tube which entirely fills the test coil, r\ = 1 (Fig. 1) . The argument, 

38 1 

38 2 


a; 0.2 a3 cu as 


Inatitut Dr. Foerster 

Fig. 1. Curve of effective permeability ^ Qtt . as a function of frequency ratio /// 
for thin- walled, nonferromagnetic tube filling the test coil (i\ = 1). 

A, of the basic function for calculation of eddy current behavior in thin-walled 
tubes, appears as the product 

A=i ^ W (1) 

where/ = test frequency, c.p.s. 

a = electrical conductivity, meter/ohm-min. 2 . 
di = inside diameter of tube, cm. 
W = wall thickness of tube, cm. 

[Compare with Eq. (4), Eddy Current Test Principles, for the solid cylinder.] 

If, as in the case of the solid cylinder, the test frequency / is selected so that the 
argument A in Eq. (1) equals unity, this frequency is called the limit frequency 
f g for the tube; i.e., 

/ ' <ttt " ) ~ 

_ 5066 

[Compare with Eq. (5), Eddy Current Test Principles, for the solid cylinder.] 


Construction of Frequency Ratio Scale. The f/f ff scale is determined by a 
simple geometrical construction on the semi-circle of effective permeability for the 
thin-walled tube (Fig. 1). A linear horizontal scale is constructed from the 
ordinate point ji e ff.( re ai) =!/ = !, and values x = 0.1, 0.2, etc., are marked on this 
horizontal scale. A straight line drawn from the origin of the effective perme 
ability coordinates to any given point on this horizontal scale will intersect 
the semi-circle at the corresponding j/j g point. For example, the line drawn from 
to A = 0.5 in Fig. 1 intersects the semi-circle at f/f ff = 0.5 (point A') . Similarly 
a straight line drawn from to any point N of the horizontal scale at y = 1 will 
intersect the semi-circle exactly at the point where j/j g = N. 

Sample Calculations for Effective Permeability and Secondary Coil Volt 
age. As an example, consider a thin-walled tube of stainless steel for which the 
electrical conductivity a = 2 meter/ohm-mm. 2 , relative magnetic permeability 
IVei. = 1, insi de diameter ^ = 2.5 cm., and wall thickness W = 0.1 cm. From Eq. 
(2) the limit frequency is calculated to be f ff = 10,000. If the test frequency is 
also taken as 10,000 c.p.s., the frequency ratio f/f ff = 1. From Fig. 1 the com 
ponents of the effective permeability are found to be [i e ff.(reai) =0.5 and 

M-efMtmnR.) =0.5. 

The two components of the apparent impedance of the test coil, or of the 
secondary coil voltage in coil arrangements such as shown in Fig. 12(c) and (d) 
in the section on Eddy Current Cylinder Tests, are 

COL Eimnf. - / s 

= - = 1 "'"*""-" (3a) 

- j- cff.dmac.) 
COL/o -&o 

where TI = (d a /d e ) 2 = fill factor of coil, 
(L = outside diameter of tube. 
do = average diameter of teat coil. 

Eqfc-i. (3) for the thin-walled tube are comparable to Eq. (16), Eddy Current Test 
Principles, for the solid cylinder. 

Direct Computation of Components of Effective Permeability from Fre 
quency Ratio, f/f g . With thin-walled tubes the components of effective per 
meability jXeff. can be calculated directly from the frequency ratio f/f ff by the 
simple formula 

*".Crl>= 1 + (///f)a (4a) 

H..c...> = ! + (/ / /f) . (4b) 


J/h 5066 

for the thin-walled, nonferrous tube (see Eq. 2). These components of ^ e ff. can 
be inserted into Eq. (3) to obtain the normalized apparent impedance components 
or the normalized components of secondary coil voltage. In this way the response 
of the test coil to the thin-walled tube can be calculated quantitatively by means 
of closed formulas. 

Effect of Changes in Outside Diameter of Tube (with Proportional 
Changes in Tube Wall Thickness). If the ratio (d^d a ) of inside to outside 



tube diameter is held constant while the outside tube diameter vanes, the clumges 
in effective permeability are as shown in Fig. 2. The outormort cnirveot F g 2 
corresponds to the semi-circle of Fig. 1, for which the till factor n = 1- The 
of Fig. 2 correspond to those for the solid cylinder shown in Pig;. 1, m the .section 

= const 

0.1 0,2 0.3 0,4 0,5 

Institut Dr. Foerster 

Fig. 2. Normalized apparent impedance plane for thin-walled, nonferromagnetic 

tubes with varying outside diameters and a constant ratio of inside to outside 

diameter (dt/da. = constant), 

on Eddy Current Cylinder Tests. Variations of electrical conductivity a, inside 
diameter d it and wail thickness W at constant outside diameter rf a are also indi 
cated by these curves. 

Effect of Changes in Outside Diameter of Tube (with Inside Diameter 
Held Constant). If the outside diameter of the tube decreased while the inside 
diameter was held constant, a very small reduction in outside diameter would 



result in a large decrease in tube wall thickness W. Fig 3 show, the apparent 
impedance plane for this case. For example, suppose that the wall tmcKnesb 
w= 10% and the inside tube radius is 90 percent of the outside tube radius. 
If the outside diameter now decreases by 5 percent while the inside diameter 

-/""Ax- '?' D c , 

\\J^\ V I fi-.JLffifli 

a; 0,2 as u wi "0 


diameter > 

impedance variation lies on 

ting the original ///, 
t g 




in diameter corresponds to a decrease in fill factor of approximately 10 percent, 
since TJ = (d/D)-, where d is test-object diameter and D is coil diameter, from 
Eq. 10, Eddy Current Test Principles.) Thus, if either wall thickness or electrical 
conductivity varies while outside diameter remains constant, the apparent im 
pedance or secondary coil voltage varies along the curved lines labeled "d a , W 
variation" in Fig. 3. 

ment referred to nonferromagnetic tubes for which ^ re] . = 1. For a ferromagnetic 
tube for which j.i rel . 1, the procedure is developed in the same manner as in 
Eqs. (1, a and b), in the section on Eddy Current Cylinder Tests, for the solid 
cylinder. However, with thin-walled steel tubes, only the volume of the tube wall 
contributes to the magnetization. These relationships for ferromagnetic tubes are 
treated more thoroughly in the literature. 42 The characteristics of ferromagnetic 
tubes tested at high field strengths, where the relative permeability varies with the 
field strength, will be discussed more thoroughly subsequently in this section under 
Limitations in Use of High Magnetizing Field Strengths. 

cates the variations in effective permeability resulting from 1 percent changes in 
tube characteristics such as electrical conductivity a, relative magnetic perme 
ability ^ rou inside diameter d b or wall thickness W. If a, |A n ,i., or the product 

Fig. 4. Sensitivity diagram for eddy current tests of thin-walled tubes. Vector OP 

indicates direction and magnitude of variation in effective permeability for 1 percent 

variations of conductivity a, inside diameter d\, or wall thickness W. 

(diXW) vary by 1 percent, the magnitude and direction of the change in effec 
tive permeability are given by the vector from the origin to the f/f ff point on the 
circle corresponding to the specific tube (see Eq, 2). 

Maximum Test Sensitivity. In accordance with a general law in the theory 
of eddy current testing, maximum test sensitivity is attained at that point on the 



effective permeability curve where the imaginary component (along the horizontal 
or z-axis) ^ e ff.(imag.) goes through a maximum. In Fig. 1, for example, this 
maximum lies at f/f g = 1. The corresponding point is shown at the bottom of the 
circle in Fig. 4. At maximum sensitivity (///,= !), a 1 percent variation in 
conductivity or in the product (d t X W) results in 0.5 percent variation in the 
effective permeability. This variation occurs in the vertical direction along the 
vector from the point f/f g = to the point for ]/f g = 1 in the fx e ff.(reai) direc 
tion (see Fig. 4). Fig. 4 indicates the magnitude and direction of apparent im 
pedance variation caused by a change in tube characteristics (a, d|, or W) in the 
apparent impedance plane. 

Experimental Determination of Frequency Ratio f/J g . The frequency ratio 
f/f g can be determined experimentally, even with a fixed-frequency test instru 
ment. The procedure is similar to that presented earlier for solid cylinders 
(see Fig. 24 of the section on Eddy Current Cylinder Tests.) Fig. 5 shows 






I 1 





J f 'f9 








- - 



t (c 








/ i 




K f lfg 



. i 



5 10 

/ 5 10 50 100 

5. Variation of angle a with frequency ratio j/] g , or produc 
ferromagnetic tube. 

the relation between the angle a and the frequency ratio /// for the case of 
nonferromagnetic tubes. From this curve and known values of test frequency /, 
wall thickness W, and inside diameter d i} the electrical conductivity of the tube 
material can be determined directly from Eq. (2) . 

Cracks in Thin-walled Tubes. Cracks in thin-walled tubes have the same 
eddy current effect as a decrease in the wall thickness W. Thus, outside cracks 
produce variations in effective permeability which lie in the same direction on the 
complex permeability plane as a decrease in outside diameter d^ with inside 
diameter held constant (see Fig. 3). Inside cracks produce variations in the 
same direction as a decrease of wall thickness with constant outside diameter. 
Thus, inside cracks are indicated by variations toward smaller f/J ff values along 
the semi-circles in the permeability plane of Fig. 2. 

Frequency Selection for Crack Detection. In crack testing of thin-walled 
tubes, the crack sensitivity curve corresponds in general to Fig. 4. Thin-walled 
tubes should be tested for cracks, alloy, or wall thickness at frequency ratios 
between /// = 0.4 and /// = 2.4. These limits correspond to a decrease from 
maximum test sensitivity of l/x/2, or 71 percent. 



Tests of Thick- Walled Tubes 

THICK-WALLED NONFERROUS TUBES. The characteristics of a 
test coil containing a thick-walled tube lie between those for a thin-walled tube 
(semi-circles of Fig. 1) and those of a solid cylinder (Fig. 7 in the section on Eddy 
Current Test Principles). The shaded area in Fig. 6 shows the range of effective 
permeability within which all variations in tube properties occur. 

0,2 0,4 


Institut Dr. Foerster 

Fig. 6. Range of variation of effective permeability for thick-walled, nonferrous 

tubes (shown by shaded area) . Variations in conductivity, inside diameter, and wall 

thickness occur in this range when outside diameter is held constant. 

Effect of Variations in Inside Diameter (with Outside Diameter Held 
Constant) . To illustrate the case of inside diameter variations, assume that a test 
cylinder is initially solid. The effective permeability or apparent impedance values 
for the test coil lie along the curve of Fig, 7 (of the section on Eddy Current Test 
Principles) at the j/j g value calculated from Eq. (5) in the same section. Now 
suppose that successively larger holes are drilled along the longitudinal axis of 
the tube. The inside diameter d t increases, while the wall thickness W decreases. 



For a given thin-wall 'thickness, the permeability curve of Fig. 1 is applicable. 
When the wall thickness is reduced to zero (W = 0), the effective permeability 
falls at fXeff.(reai) o r coL/coL = 1 (top left of Fig. 1). In the absence of the test 
object, this point corresponds to the effective permeability of the coil when empty. 
Fig. 7 shows the paths followed by effective permeability for four solid cylinders 
with frequency ratios of f/f g = 4, 9, 25, and 100, as wall thickness varies from 
that of the solid cylinder (W = d a /2) to zero (empty test coil). The electrical 

, d a = const/ 
dj,W= variation 

dj,d a ,W = const. 
G,f = variation 


0/1 0,2 03 0,4 

Institut Dr. Foerster 
Fig. 7. Complex permeability or impedance plane for nonferromagnetic tubes of 


conductivity and outside diameter d a are assumed constant for these curves, 
while only inside diameter d 4 and wall thickness W vary. To simplify the 
presentation, the f/f g value for the solid cylinder is taken for the entire curve 
extending from solid cylinder to empty coil conditions. Only the inside diameter 
' varies along these curves for specific f/f ff ratios. The ratios of inside diameter 
to outside diameter, di/d a = rjr^ in percentages, are marked on those dashed 

Sample Calculation for Variation of Inside Diameter. To obtain the effec 
tive permeability or apparent impedance of the test coil for the thick- walled tube, 
the steps are : 

1. Calculate the ///, ratio for the solid cylinder from Eq. (5), Eddy Current 
Teat Principles. 

2. Select the tube curve in Fig. 7 which starts at this }/} ff value on the curve for 
the solid cylinder and ends at the point where jn,nff.<r<>,in = 1 for wall thickness 
W = (at upper left of Fig. 7). 

3. Find the point 'which corresponds to the diameter ratio d ( /d a , for the tube in 
question, along the selected curve. 

4. The real and imaginary components of the effective permeability of this point 
on Fig. 7 are inserted into Eqs. (16, a and b), Eddy Current Test Principles, to 
determine the components of the apparent impedance of the tost coil containing 
the thick-walled tube. 

For example, point A in Fig. 7 corresponds to a solid cylinder with a frequency 
ratio f/f p = 9. If this cylinder is drilled out until the inside diameter is 60 percent 
of the outside diameter, the effective permeability is given by point B. Point C 
corresponds to a diameter ratio di/d a = 70% ; and point D corresponds to 80 

Effect of Variations in Electrical Conductivity. Dashed curves in Fig. 7 con 
nect points with the same diameter ratio, c?</d tt = ^/r,,, between the various f/f ff 
curves,. For example, at dt/d a = 95%, the wall thickness is 5 percent of the out 
side radius r a of the tube. The frequency ratio f/f ff increases along these curves, 
corresponding to changes in either the electrical conductivity or the test frequency. 

Effect of Variations of Outside Diameter (with Ratio of Inside to Outside 
Diameter Held Constant). Fig. 8 shows the apparent impedance plane for tubes 
with varying outside diameter d a for the special case in which wall thickness W 
is 20 percent of the outside radius (W/r a = 20%). The diameter ratio, di/d a , is 
constant, as well. The apparent impedance for this case is obtained by combining 
data from Figs. 7 and 8. The solid curves of Fig. 7 show the effects of varying the 
ratio of d i to d a . The family of curves of Fig. 8 shows the effect of the variations 
in fill factor r\, which correspond to various outside diameters in relative units. 
Since T) = (dJD)*, where d a is the outside diameter of the tube and D is the inside 
diameter of the test coil, the fill-factor curves represent outside diameters as 

<f . = 1 ti = 1 
d a = 0.9 t| = 0.81 
d* = 0.8 TJ = 0.64 

and so forth. 

The outermost curve of Fig. 8, labeled "T| = 1," corresponds to the curve for 
di/d a of Fig. 7. In Fig. 8 the directions of diameter and conductivity effects 
are clearly indicated. The d a curves form large angles with the a curves, so that 
diameter and conductivity variations are readily separated from each other. 



0,1 0,2 0,3 04 

Institut Dr. Foerster 

Fig 8 Complex impedance or permeability plane for nonferromagnetic tubes with 

wall thickness of 20 percent of outside tube radius as outside diameter d. of tube 

varies. Curves labeled d. correspond to variations of outside diameter. Variations 

of electrical conductivity or test frequency are indicated along j/J, curves. 

The /// ratio for the solid cylinder, from Eq. (5), Eddy Current Test Prin 
ciples is used for the thick-walled tubes in Fig. 7, where conductivity a and out 
side diameter d a were assumed to be constant while wall thickness W varied from 
dJZ (solid cylinder) to zero (empty coil). Each curve in Fig. 7 which starts at a 
specific ///. value for the solid cylinder and ends at the empty value (upper lelt) 
for W = crosses the curve for d./d. = 80 percent at one point These p_omts 
were used to plot the curve for v\ = I (outermost curve of Fig. 8) for d t /d a - U.S. 


Maximum Test Sensitivity. The maximum tost sensitivity for effects of con 
ductivity, wall thickness, and cracks occurs where the real component of apparent 
impedance, R/($L ()) or of normalized secondary coil voltage, K ml > f E (]t reaches a 
maximum. In the case shown in Fig. S, this occurs near /// = 13 (right extremity 
of j/jg curve for T] = 1). Thus, if a tube with 20 percent wall thickness is to be 
tested, the frequency ratio f/f ff for a solid cylinder of the same material (with the 
same electrical conductivity a) and the same outside diameter d a should be in the 
vicinity of 13. 

Calculation of Frequency Ratio, f/f ff , for Maximum Test Sensitivity. 

The frequency ratio J/f ff for optimum sensitivity in tube tests is calculated, from 
Eq. (5), Eddy Current Test Principles, for the solid cylinder, and Eq. (2) of this 
section, for the thin-walled tube as 

For the thick-walled tube of Fig. 8, where W/r a = O.S, the ///,, ratio calculated 
from Eq. (5) is found to be f/f ff = 12.5. This correlates well with the maximum 
deflection to the right in Fig. 8. The maximum test sensitivity thus lies exactly at 
i/jg = 12.5 (previously estimated from the curve to be near 13) . 

Fig. 9 tabulates frequency ratios for optimum test sensitivities for various 
ratios of wall thickness W to outside tube radius r ft . If the tube wall is not too 
thin, this optimum f/f g ratio can be varied from 0.7 to 1.4 times the optimum 
without decreasing test sensitivity below SO percent of the optimum. 

W/r a ///,,Cyl.Opt. ///t,Rango 






















Fig. 9. Frequency ratios ///, for optimum test sensitivity in thick-walled tubes 
with various ratios of wall thickness W to outside radius r. 

boundary conditions prevents exact mathematical solution of the problem of tubes 
containing cracks, data were obtained by means of extensive model experiments. 
The mercury model technique (see Figs. 7 to 10 of the section on Edcly Current 
Cylinder Tests) was used to determine the influence of outside-surface, inside- 
surface, and subsurface cracks of various depths over a wide frequency range. 
Thousands of individual measurements were needed to account for the effects of 
tube wall thickness and material conductivity upon crack indications. 

Quantitative Evaluation of Crack Depths. Only a small portion of the exten 
sive crack-effect studies can be reproduced here. Figs. 10 to 13 show the influence 
of cracks at various locations and of various depths, for various tube wall 
thicknesses, and for frequency ratios of f/f ff = 5, 15, 50, and 150, respectively. 
The analysis procedure is similar to that used previously for analysis of cracks in 



\ IN % OF W 

-0,02 -0,01 0,01 0,02 

Fig. 10. Apparent impedance plane showing effects of cracks for four different 
tube wall thicknesses, at a frequency ratio f/f a = 5, for nonferromagnetic tubes. 


0.01 002 007 0,02 0.01 0,02 wi-o 
- T'*% - 


Fig. 11. Apparent impedance plane showing effects of cracks for four different 
tube wall thicknesses, at a frequency ratio ///, = 15, for nonferromagnetic tubes. 



f/ f "50 (CYLINDER) 
T 9 


IN % OF W 

Fig. 12. Apparent impedance plane showing effects of cracks for four different 
tube wall thicknesses, at a frequency ratio f/f a = 50, for nonferromagnetic tubes. 

solid cylinders. A family of curves is obtained for each }/f ff value for the solid 
cylinder, showing the apparent impedance values for various crack depths and 
locations. From these, a new family of curves is obtained for each tube wall thick 
ness. Figs. 10 to 13 show such families of crack curves for the following wall- 
thickness ratios: PF/r a = 33%, 26%, 20%, and 13%. Here, W is the wall" thick 
ness and r a is the outside tube radius r a = d a /2. 




-001 +0.01 0,01 ( 
J1L.2S% 4^-- 

f / f -/SO (.CYLINDER 3 
f 9 





Fig. 13. Apparent impedance plane showing effects of cracks for four different 
tube wall thicknesses, at a frequency ratio /// = 150, for nonferromagnetic tubes. 

Characteristics of Crack Evaluation Curves. With higher frequency ratios 
and thicker tube walls, the phase displacement between field strength of outside 
and inside tube surfaces increases. For W/r a ratios of only 13 percent (curves on 
right of Figs. 10 and 11), no phase displacement exists in the impedance curves 
for outside and inside surface cracks. It is also interesting to note from Fig. 11 
that the influence of a subsurface crack (which reaches neither inside nor out 
side tube surfaces) is slightly smaller than for a surface crack of the same depth 
(compare distances OA and OB). (This is analogous to the accumulation point of 
a stream of water. A board placed transversely to the stream direction in the 
'center of the stream causes less current resistance that does the same board at the 
edge 'of the stream.) 


The response to a crack-free tube can be determined from Fig. 7 for specific 
f/fg ratios for the solid cylinder [calculated from Eq, (5) of the section on Eddy 
Current Test Principles, the tube material properties, and W/r a ratio]. Figs. 10 
to 13 are plotted to scale so that the components of relative permeability, 
Apteff., obtained from these curves can be inserted into Eqs. (7, a and b), of the 
section on Eddy Current Cylinder Tests, to obtain the voltage variations &E of 
the te,st coil, in response to the cracks. 

W Eddy Current Tests of Tubes with Internal Test Coil 

current test problems in which the test coil is located within the test object. For 
example, large numbers of tubes are required for boilers, condensers, coolers, and 
heat exchangers in oil refineries, steam power plants, chemical plants, and sugar 
refineries. Such installations usually consist of large numbers of tubes located 
close together and enclosed in a vessel so that their external surfaces are not 
accessible. These tubes can be tested only by inserting the test coil into the tubes 
from the tube header or other access points. In other cases it becomes neces 
sary to test deep, small-diameter drilled holes in large metal parts (for discon 
tinuities, segregation, corrosion, or other conditions) by inserting the test coil 
within the hole. 

Another important problem is the internal testing of thick-walled tubes 
which, although their external surfaces are accessible, do not permit detection of 
dangerous inside-surface cracks because their heavy wall thickness limits the eddy 
current penetration depth. This situation arises in cases of stress corrosion in 
the chemical industry, for example. 

test coil of diameter d placed within a tube of inside diameter D t . To calculate 
how the coil characteristics vary with test object characteristics, such as (1) inside 


Df d 

I T 

Fig. 14. Test coil inside cylinder. 

diameter of tube, D i} (2) tube wall thickness, W, (3) electrical conductivity a of 
tube material, (4) relative magnetic permeability p, rel> of tube material, three 
distinguishing cases must be considered: 

1. The thin-walled tube, 

2. The tube with medium wall thickness. 

3. The heavy-walled tube. 

Heavy-walled Tubes. Fig. 15 can be used to illustrate the definition of the 
limit between medium and heavy-tube wall thicknesses. In Fig. 15 the in- 



.Sec. Prim. 


Institut Dr. Foerster 
Fig. 15. Arrangement for measuring the screening effect of tubes. 

ternal coil is connected to a source of alternating current and serves to establish 
an a.-c. magnetizing field. The external coil serves as a secondary winding. If the 
tube in Fig. 15 is removed from the coil system, the a.-c. field of the internal 
primary coil induces a voltage E in the external secondary winding. If the tube 
is again placed between the primary and secondary coils of Fig. 15, a reduced 
voltage B appears across the secondary coil terminals. Fig. 16 shows the experi 
mentally determined ratio of the voltage E, obtained with a tube in place, to the 
voltage E of the empty coil system, as a function of test frequency. Curves are 
presented for three tubes, of copper, brass, and stainless steel, each with an 
outside diameter d a = 0.8 in. and a wall thickness W = 0.040 in. Whether a tube 
is considered heavy-walled or medium-walled depends upon the test frequency, 
as illustrated by Fig. 16. If less than 10 percent of the magnetizing field of the 

2 4 6 8 10 12 14 /CCPS 

Institut Dr. Foerster 

Fig. 16. Decrease of field strength upon insertion of tube between coils of Fig. 15, 
for copper, brass, and stainless steel tubes of 0.8-in. outside diameter and 0.040-in. 

wall thickness. 


inside primary coil penetrates the tube and is indicated by secondary coil voltage, 
the tube is considered to be heavy-walled. Thus the copper tube with wall thick 
ness of 0.040 in., tested at frequencies exceeding 3000 c.p.s. (3 kc.), as well as the 
brass tube at frequencies above 11 kc., appear to be thick-walled. 

Evaluation of Thin-walled Nonferromagnetic Tubes. The evaluation of 
thin-walled, nonferromagnetic tubes is attained through the use of Figs. 1 to 4, 
which apply also for the testing of thin-walled tubes with an inside coil. The 
apparent impedance of the test coil is the same, whether the test coil is placed 
directly inside the inner surface or directly on the outer surface of a thin-walled 
tube. In both cases the lines of magnetic flux are present within the tube material, 
so that the field of the short-circuited eddy currents has the same reaction on the 
inside coil as on the outside coil. 

Evaluation of Thick-walled Nonferromagnetic Tubes. The thick-walled 
tube, on the other hand, has an entirely different reaction. A limit frequency f ff 
is defined, as with the solid cylinder, by equating to unity the argument of the 
mathematical function describing the problem. For the inside coil and thick- 
walled tube, the limit frequency is defined as 

. 5066 ,, , 

fa = rTTar < 6a ) 

and the frequency ratio f/f g as 


where a = electrical conductivity of tube material, meter /ohm-mm. 2 . 
Mrei. = relative magnetic permeability of tube material. 
Di = inside diameter of tube, cm. 

Before insertion in the tube, the test coil exhibits the apparent impedance 
values o)L and R . After insertion, these values change to coL and R. Fig. 17 
shows the normalized impedance components for various frequency ratios 
1/jg and four different fill factors TJ = (d/D^. The apparent impedance 
curve for the fill factor Y] = 1 is identical with the effective permeability. The 
ordinate coL/o)L corresponds to the real component, and the abscissa 
(R R Q )/toL to the imaginary component of the effective permeability. 

Separation of Material Properties and Inside Corrosion Thinning. When 
the material conductivity or relative permeability characteristics vary, while the 
inside diameter DI of the tube remains constant, the apparent impedance values 
of the test coil move along the f/f ff curves. In the case of variation of the inside 
diameter of the tube caused, for example, by corrosion wall-thinning, the 
apparent impedance of the test coil moves along the D i curves of Fig. 17. Thus a 
corrosion effect (variation in JDJ lies in a different phase direction in the 
apparent impedance plane than does a variation in material characteristics from 
tube to tube. Consequently, corrosion effects can be measured with the inside 
test coil independently of tube material properties. 

Maximum Test Sensitivity. The greatest indication sensitivity for conductiv 
ity a, relative permeability jj, reL , and crack effects lies at test frequencies for which 
the frequency ratio f/f ff is in the range from 1.5 to 12. 

Evaluation of Medium-Thickness Tube Walls. The theory of the inside 
coil with medium-thickness tube walls, in which inside as well as outside cracks 
influence the coil indications, is discussed more thoroughly in the literature. 42 





Fig. 17. Apparent impedance diagram of test coil inside heavy-walled nonferrous 
tube, for various fill factors T]. Institut Dr. Foerster. 




Penetration Depth Limitations 

tion of the effective permeability values of tubes with a given wall thickness 
permits determination of the penetration depth, a fundamental factor in eddy 
current testing. The calculation indicates the maximum tube wall thickness 
detectible. If a tube is placed in the field of the test coil instead of a solid cylinder, 
the calculation indicates at what diameter ratio (Di/D a ) the difference in effec 
tive permeability between the solid cylinder and the tube is sufficiently large to 
produce reliable test indications. 

Depending upon the sensitivity of the test instrument, it is able to "see" 
more or less deeply into the inside of the cylinder. Fig. 18, based upon an exact 
calculation of the difference in permeability between cylinder and tube, indicates 

Fig. 18. Penetration depths for different instrument sensitivities (10~ 2 , 10~ 3 , 10' 4 , 
and 10~ 5 ) as a function of frequency ratio ///,. 

to what extent the center of the cylinder can be drilled out before the test 
instrument can detect the variation from the apparent impedance of the solid 
cylinder. This maximum detectible wall thickness is a function of the fre 
quency ratio and of the sensitivity of the test instrument. Fig. 18 shows curves 
for fill factor r) = 1, corresponding to four different instrument sensitivities. 

Significance of Instrument Sensitivities. An instrument sensitivity of 
10~ 2 or 10~ 5 for example, means that a variation of 10~ 2 or 10 5 in 
effective permeability (as described by Fig. 7 of the section on Eddy Current 
Test Principles) is still clearly visible in the instrument indication. Since 
Fig 18 applies for fill factor T] = 1, the variation in effective permeability must 
be twice as large to be detectible for a fill factor n = %. Fig. 18 indicates the 
penetration depth for a wide range of frequency ratios /// for instrument sensi 
tivities of lO- 2 , 10- 3 , 10- 4 , and 1Q- 5 . Alternatively, Fig. 18 indicates how far 
the central cavity in a cylinder can increase before it is just detectible at specific 



instrument, sensitivity and f/f ff values. It shows the relative insensitivity of 
eddy current methods to concentric cavities. 

Limitations in Detection of Concentric Cavities. Even in the most favor 
able case, with f/f ff = 6.25 and an instrument sensitivity as high as 10~ s , the 
central 6 percent of the diameter of a cylinder cannot be inspected with eddy 
current measurements. With an instrument sensitivity of 10~ 3 , even with the 
most favorable frequency ratio of ///^ = 6.25, the central 23 percent of the 
cylinder diameter cannot be inspected. However, in most cases, discontinuities 
such as cracks and cavities which influence serviceability have an eccentric 
location and are more easily determined. 

Determination of Secondary Coil Voltages with Tubes. To calculate the 
variation in test coil voltage when a tube rather than a solid cylinder is present, 
the A^ieff. values from Fig. IS are inserted into Eq. (7). 

Atf/J0 = n(An..) (7) 

The A^eff. values on the curves of Fig. 18 are determined, of course, by the avail 
able instrument sensitivities. When voltage values have been calculated for a 
specific instrument sensitivity by means of Eq. (7), Fig. IS can be used to obtain 
penetration depths in percentages of the outside radius of the cylinder. 

Magnetic Permeability Limitations on Depth Penetration. Ferromagnetic 
test materials are subject to a phenomenon which generally causes considerable 
decrease in the effective depth of penetration of eddy current measurements. Fig. 
19 shows the relative magnetic permeability |A rel . as a function of the magnetizing 

SO 100 


150 H 0e 

Fig. 19. Relative magnetic permeability, \i ro i. of a carbon steel as a function of 
external magnetic-field strength. Note rapid increase in initial permeability with 

increasing field strength. 


field strength for a typical carbon steel. At low field strengths the ferromagnetic 
material normally shows a very small initial permeability, ji^. As the field 
strength is increased, the relative permeability increases greatly and eventually 
reaches the high maximum permeability, ji max . At still higher magnetizing field 
strengths, the permeability values decrease more slowly. 

The theory of penetration depth, as summarized in Figs. 3, 4, 5, and 6 in 
the section on Eddy Current Cylinder Tests, indicates that penetration depth 
decreases with increases in the product (D 2 a[x) of the square of cylinder diameter, 
electrical conductivity, and relative magnetic permeability. Thus Fig. 19 would 
appear to indicate that it would be practical to work with very low field strengths, 
to take advantage of the very low initial permeability. If this were feasible, it 
might permit a greater penetration depth and better indication of subsurface 
cracks. However, the use of weak field strengths is impossible because of the 
influence of inhomogeneous stresses caused, for example, by the process of 
straightening rods and tubes, which influence the magnetic characteristics of the 
test material. The initial permeability is especially changed by inhomogeneous 
mechanical stress conditions. In consequence, various initial permeability 
values appear from point to point, in accordance with previous deformations, in 
the testing of rods or tubes. 

Limitations in Use of High Magnetizing Field Strengths. The known de 
crease of field strength toward the inside of the cylinder (see Fig. 5 of the section 
on Eddy Current Cylinder Tests) appears also in the use of higher a.-c. 
magnetizing field strengths. As soon as the field strength within the cylinder 
approaches the value at which the maximum permeability occurs, a large and 
rapid decrease occurs in field strength as a result of the increasing product 
( D 2 G\i) . At test frequencies of 60 c.p.s., even with high external field strengths, 
field strength values corresponding to the maximum permeability are obtained 
immediately below the surface of the cylinder or tube. This layer of maximum 
permeability offers considerable resistance to deeper penetration of the a.-c. field. 
Fig. 20 presents experimental proof of the existence of this condition, which 
becomes quite important in nondestructive testing of iron and steel. 

In the experiments used to provide the data of Fig. 20, three steel tubes with 
low carbon content were prepared with diameters of 1.2 in. and wall thicknesses 
of 0.040, 0.080, and 0.120 in. Cylindrical plugs were made to fill the entire interior 
volume of each tube. First, the three plugged tubes, whose physical behavior is 
the same as that of solid cylinders, were inserted into the test coil and resulting 
test coil voltages were measured. Then the cylindrical core of each tube was 
removed, and the test repeated. The voltages obtained with the empty tubes, 
expressed in percentages of the voltages of the filled cylinders, were plotted as the 
ordinates in Fig. 20. 

Shielding Tests at 50-C.p.s. Frequency. Fig. 20 shows the differences in 
voltages between the solid cylinder and the hollow tube as a function of external 
field strength (measured in terms of magnetizing amperes). In tests made at 
low field strengths and 50 c.p.s., removal of the cylindrical core from the 
0.040-in. wall tube resulted in 38 percent voltage variation (see point A, Fig. 20). 
As field strengths were increased toward the condition of maximum relative 
permeability, the influence of the core decreased. When the maximum per 
meability was reached, the influence of the core was only 3.5 percent (see point 
B). The effect of the core was reduced tenfold because of the decrease in pene 
tration depth as magnetization increased from the initial permeability conditions 
at low field strengths to the maximum permeability condition. At still higher field 



strengths, the relative permeability decreases. The penetration depth increases 
above the maximum permeability condition of magnetization, so that the effect of 
the core is again more distinct. 

Shielding Tests at 5-C.p.s. Frequency. At the low test frequency of^5 c.p.s., 
the penetration depth is greater than at 50 c.p.s. Removal of the cylindrical core 


Institut Dr. Foerster 

Fig. 20. Penetration depths for carbon steel tubes of 0.040, 0.080, and 0.120 in. 
wall thickness, 1.2-in. diameter, at 5 and 50 c.p.s. 

from the 0.040-in. wall tube results in an 80 percent voltage variation at low field 
strengths. In the initial permeability range, the influence of the core is about 
twice as great at 5 c.p.s. as at 50 c.p.s. With tubes of greater wall thickness, the 
difference in penetration depth between 5- and 50-c.p.s. test frequencies is con 
siderably larger. 


At medium field strengths (corresponding to a magnetizing current of 1.5 
amp. in Fig. 20), changing from 50 to 5 c.p.s. increases the influence of the core, 
by a ratio of 16:1, with tubes of 0.080-in. wall thickness. With even greater wall 
thicknesses, the superiority of 5-c.p.s. test frequencies in increasing the pene 
tration depth is even more obvious. The relatively low penetration depths of 
eddy current tests of ferromagnetic materials makes them most suitable for 
detection of surface discontinuities only, for the reasons discussed above. 

Tests of Small Diameter Tubing 

PRINCIPLE OF TEST. Electromagnetically induced, alternating, eddy 
current flow is a method of determining the quality of small-diameter, nonferro- 
magnetic, metallic tubing. Fluctuations in the field of an exciting coil system are 
produced by variations in eddy current flow in a tube moved relative to the coil. 
In general, two distinctly different effects on the field of the coil are produced by 
eddy current flow : 

1. The flow of induced current through a metal part of finite resistivity produces 
a power loss. This must be supplied by the exciting field, which in turn must 
be supplied from the alternating current feeding the exciting coil. Thus the 
exciting coil impedance must have a resistive component whose magnitude 
depends upon the power loss in the metal. 

2. The field of the induced current opposes the exciting field, decreasing the field 
flux of the exciting coil. The inductance of the exciting coil (the ease with 
which it produces magnetic flux) is reduced by an amount depending upon the 
induced current magnitude. The reactive component of the exciting coil im 
pedance is reduced accordingly. 

As the tube moves through the coil, fluctuations occur in both the resistive and 
reactive components of the coil impedance. These correspond to variations in 
the magnitude, phase, and distribution of the induced eddy currents. 

curves of Fig. 21 show variations in amplitude and phase of impedance for a 
test coil encircling a cylindrical metal tube. Here the amplitude and phase values 
are resolved into two quadrature components, the reactive and resistive com 
ponents. These values are normalized to the pure reactive impedance of the coil 
in the absence of the metal tube. The frequency ratio /// c permits this plot to be 
used for tubing of any conductivity, diameter, and wall thickness. The quantity 
f c is a characteristic frequency defined by the relationship 

where M- = permeability, henries/meter, (4jt X 10' 7 for nonmagnetic material). 

Y = conductivity, mhos/meter. 
Dp = outside diameter of part, meters. 

The plot of Fig. 21 was made for a ratio of part diameter to coil diameter, 
v = D P /D C = 1, and four different values of /// c . The dotted curve is for varying 
/// ratios, the heavy dashed lines show the effect of varying outside diameter, and 
the solid lines show the effect of varying wall thickness. These curves differ^ con 
siderably in shape from those representing the impedance of a coil encircling a 
solid rod. The wall thickness of the tube is as important in influencing the 
impedance of the coil as are the conductivity and outside diameter of the tube. 


= PERMEABILITY OF AIR (henries/m) 

O/ 7.5% 5% 


0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 

Oak Ridge Nut'l. Lab. 

Fig. 21. Impedance plane for coil encircling nonferromagnetic metal tubes. 

Controlling Factors. The alteration of the field produced by the induced eddy 
currents is a function of 

1. The frequency of the exciting current. 

2. The electrical conductivity and magnetic permeability of the tube. 

3. The physical dimensions of the tube and its location in the field. 

4. The presence and location of discontinuities or "inhomogeneities" within the 
tube wall. 

Since the mechanical and thermal history of the tube influence its electrical 
conductivity and, in the case of the austenitic stainless steels, its magnetic per 
meability, they also influence the field of the coil. Mechanical, metallurgical, or 
chemical variations within the tube, as it moves through the coil, affect the coil 
impedance. The numerous sources of signals from this type of test place a large 
burden of interpretation upon the inspector. 

The exact nature of the signals obtained and their interpretation depend 
entirely upon the type of coil system and instrumentation used. In addition the 



manner in which signals are displayed and the method used to move the tube 
through the coil system are exceedingly important in the interpretation. 

TYPES OF COIL SYSTEMS. Fig. 22 shows three basic coil systems used 
on small-diameter tubing. 

1. Encircling, or "feed-through," coils. 

2. Inside-probe, or "bobbin," coils. 

3. Surface probe coils. 

Also illustrated for each case is the use of both "absolute" measurements (with 
out a direct reference) and differential methods. The encircling-coil system is 
more common because of its ease of utilization and relatively high inspection 
speeds. The mechanical difficulties involved in the application of bobbin coils 
are considerable, particularly in the case of long lengths of very small-diameter 
tubing. Despite this, their use may be warranted for inspection of the bore of 
heavy-wall tubing or duplex tubing. The surface probe coil is not in general 
use for inspection of small-diameter tubing because of inherent mechanical prob 
lems and slow inspection speeds. Its high sensitivity to small discontinuities and 
definitive ability, however, are very attractive for critical inspection and will be 
considered in detail later. 

The use of single and double coil arrangements is also illustrated in Fig. 22. 
In the case of the double encircling-coil system, the a.-c. excitation is applied to 





Oak Ridge Xat'l. Lab. 
Fig. 22. Test coil configurations for eddy current testing of small-diameter tubing. 


the outer coil, and the voltage induced in the inner "pick-up" coil is examined. 
The advantage of this arrangement is that the pick-up coil may have a very small 
dimension in the longitudinal direction and thus be sensitive to very short dis 
continuities. At the same time the exciting field in the vicinity of the pick-up coil 
may be made almost a geometrically perfect cylinder of flux parallel to the axis 
of the coil and of uniform density in absence of the tube. This is done by the use 
of an elongated exciting coil. The double-coil system does not appear to offer a 
great practical advantage over the single-coil system with a large ratio of outside 
diameter of tube to effective coil diameter, since the effect of field distortion in the 
vicinity of the tube wall produced by the use of a thin single coil is small. 

Sensitivity and Signal Discrimination. In employing an encircling coil to 
inspect tubing, the problem is more than one of sensitivity to variations in the 
tube. As suggested previously, the eddy currents will, in general, be influenced by 
all possible variations occurring in the tube. By exercising a little ingenuity, 
several instrument types can be contrived which will have a very sensitive response 
to these variations. The problem is one of separation and classification of these 
variations by their nature and extent. More specifically, it is desirable to dis 
tinguish signals resulting from dimensional variations from those resulting from 
discontinuities in the tubing. In addition, it would be advantageous to separate 
outside-surface from inside-surface discontinuity signals and also to separate the 
effects of wall-thickness and diameter changes. 

Differential Coil System. One approach to the problem of distinguishing 
between signals from discontinuities and those resulting from dimensional varia 
tions is the use of the differential-coil system illustrated in Fig. 22. In this scheme, 
signals from coils encircling two adjacent sections of the tube are subtracted. 
Dimensional variations generally occur more slowly as a function of tube lengths 
than do the discontinuities, and if the two coils are closely spaced, the signals 
from the dimensional variations tend to be eliminated. Several disadvantages and 
limitations are imposed by the use of this technique. If the system is successful, 
the dimensional signals are completely eliminated from the test indications instead 
of being separated for observation or monitoring. An additional absolute coil 
must be used to obtain the dimensional information when it is desired. 

If the differential system is to be insensitive to dimensional variations, the two 
coils must be very close together. All drawn tubing contains a periodic varia 
tion in both wall thickness and diameter, varying in magnitude within a period 
of 2 to 8 in., depending upon the material, tube size, and the drawing and anneal 
ing practices used. Thus, if the two coils are not close together, the dimensional 
signals ^may be added rather than subtracted. Assuming that the coils are in 
proximity, there is a further and more serious disadvantage : A signal produced 
by this system measures only the size of a discontinuity when the length of the 
discontinuity is less than the distance between the coils. For discontinuities 
longer than this distance, the signal measures the size of the leading and trailing 
edges of the discontinuity over a length equal to the coil spacing, and the rate of 
change with respect to tube length at all intermediate points. Thus signals from 
such a system may be very difficult to interpret, even to the extent of beino- 

Absolute Coil System. The absolute encircling-coil system does not inherently 
distinguish between dimensional variations and discontinuities. However, the 
resultant signals usually can be identified separately if the signal is recorded as a 
function of tube length, as will be illustrated later. 



Tube-feeding Mechanism. Interpretation of signals from encircling-coil sys 
tems is often hampered by spurious signals caused by the wobble of the tubing 
in the coil. A tube-feeding mechanism, such as that illustrated in Fig. 23 has been 
very effective in eliminating such spurious signals. In this system the testing coil 
is wound on a laminated-plastic guide tube, having a length approximately equal 
to ten tubing diameters. The tube under inspection slides inside this guide tube. 
The coil assembly is then supported by very light springs so that it can move 
freely with the tube. 


TUBING OD + 0.015 in. 




Fig. 23. 


Oak Ridge Natl. Lab. 
Encircling-coil assembly and feed mechanism for tubing inspection. 

INSTRUMENTATION. Four basic types of measurements can be applied 
to the inspection of small-diameter tubing: 

1. Measurement of the change in magnitude of the total impedance of the testing 

coil, regardless of the phase, _ 

2 Measurement of only the resistive component of the coil impedance (core loss). 
3' Measurement of only the reactive component of the testing-coil impedance. 
4^ Phase-sensitive measurements which separate the resistive and reactive com 
ponents of the coil impedance. 

38 28 


Regardless of the type of instrumentation used, signals from the test coil behave 
primarily as indicated by the impedance plane of Fig. 21. Of the four basic instru 
ment types, those employing phase-sensitive detectors have gained the greatest 
popularity, since two forms of information are obtained rather than one. For 
example, refer to the impedance plot of Fig. 21 and consider a lube having a wall 
thickness of 10 percent of its diameter. If a frequency is chosen such that 
f/fc = 25, the effects of varying diameter and varying wall thickness are separated 
in phase by approximately 90 cleg. Thus, by employing phase-sensitive detectors, 
it is possible simultaneously to obtain separate signals for varying wall thickness 
and varying diameter. However, to attain a 90-deg. separation between the effects 
of varying diameter and va tying wall thickness, two conditions must be obtained. 

1. The operating frequency must be varied according to the square of the tube 
diameter and with the first power of the tube conductivity. 

2. The f/fc ratio must be varied as the percentage of wall thickness varies. 

This illustrates the inadequacy of fixed-frequency, phase-sensitive, inspection 

which measure only one component on the impedance plane of Fig. 21, and those 
which measure the total vector magnitude of a change on the impedance plane, 
can be used to advantage in the inspection of tubing. An example of this is the 
use of an instrument which measures only the resistivity component of changes 
in the testing-coil impedance. Fig. 24 is a plot of the instrument reading (at a 

fc = 


/x = PERMEABILITY (henries/meter) 
r = CONDUCTIVITY (mhos/meter) 

5 10 15 20 25 30 35 40 45 50 

Oak Ridge Natl Lab. 

Fig. 24. Resistance-loss test indications as a function of tubing wall thickness with 

f/fc as a parameter. 

reduced sensitivity) as a function of tube- wall thickness, with the ratio f/f c as a 
parameter. This read-out is a distorted inverse measure of the resistive com 
ponents on the impedance plane of Fig. 21. The nonlinear characteristic of this 
instrument near the zero or "quench" of the oscillator causes the read-out to be 
distorted. It also accounts for the high sensitivity of the instrument. 



FLAW SENSITIVITY. The optimum outside diameter and inside diameter 
flaw sensitivity is obtained for a value of f/f c which causes the curve to peak for 
a value of wall thickness near that of the tubing to be tested. This is the condi 
tion of maximum power loss or core loss in the tube. In general, discontinuities 
within the tube wall cause indications similar to those from reductions in wall 




\ \ 

\ \ 

\ 1 



Y 17292 \ \ 


FREQUENCY: 160 kc 

Oak Ridge Nat'l. Lab. 

Fig. 25. Eddy current signal trace and photomicrograph of defective % X 0.049-in. 

Hastelloy tubing. 

thickness. Therefore operating points are selected just to the left of nega 
tive peaks in the curves of Fig. 24. Discontinuities, wall-thickness reductions and 
diameter reductions then cause positive deBections in the instrument read-out, 
whereas increases in the wall thickness or diameter, or foreign-metal pick-up 
cause negative deflections. Although the signals obtained from this system are not 



definitive in themselves, interpretation can be enhanced by recording the signals 
as a function of tube lengths. Signals due to relatively large discontinuities are 
usually very easily separated from those resulting from dimensional variations in 
the tube because of their large amplitude and very abrupt nature. This is illus 
trated by the sharp positive spikes in the trace of Fig. 25, caused by an outside 
radial crack of 0.036-in. depth in a % X 0.049-in, Hastelloy B tube. 

Detection of Intergranular Corrosion. Intergranular attack on the inner 
surface of tubing produces signals similar to the indications illustrated in Fig. 26, 


T / r r r~ v_404-7* ' r r r 








TUBING: y 2 - x 0.0625-in. NIMONIC 


Oak Ridge Nat'l. Lab. 

Fig. 26. Eddy current signal trace and photomicrograph of intergranular corrosion 
on inside surface of Nimonic tubing. 


which appear less sharp than signals caused by cracks. The indicated signal was 
produced by the 0.002-in. deep, corrosive condition shown in the photomicrograph. 
The smaller signals appearing on the trace were produced by dimensional varia 
tions and less severe conditions of intergranular attack. Foreign metallic 
particles are sometimes found embedded in the inside wall of redrawn small- 
diameter tubing. These bodies are usually picked up from tools used in manufac 
turing of the tubing and generally have a high magnetic permeability; hence, 
they are very easily detected by eddy currents in a nonferromagnetic tube. Fig. 
27 shows a photomicrograph of two such pick-up areas on the inside wall of a 
% 6 X 0.025-in. Inconel tube, and the corresponding eddy current trace. The 
presence of these areas is indicated by the two sharp negative spikes. These 
spikes are in the opposite direction to those caused by the cracks of Fig. 25 and 
the intergranular attack shown in Fig. 26. The smooth cyclic variations in this 
trace are due to dimensional variations in the tube. 

Sensitivity to Small Discontinuities. The signal traces shown in Fig. 27 indi 
cate a limitation imposed by the use of the encircling coil for the detection of very 
small discontinuities. A defect cannot be resolved if the signal it produces is small 
compared with the signals from all other variations, including dimensional varia 
tions, observed as the tube is moved through the coil. The maximum practical 
test sensitivity which may be achieved with this type of test is the detection of 
discontinuities which 

1. Have lengths comparable to, or longer than, that of the coil. 

2. Have depths of 5 percent or more of the wall thickness. 

The sensitivity to average dimensional changes and to relatively large areas of 
intergranular attack is, of course, much greater than this, as has been shown in the 
previous illustrations. 

Wall Thickness and Tube Diameter. In many critical uses of small-diameter 
tubing the physical dimensions of the tubing are important. In such cases, ^ the 
consumer must be able to determine the conformity of the tube to the required 
dimensional tolerance range. The only promising approach to a fast accurate 
continuous gaging of tube dimensions is that of the encircling-coil, eddy current 


The difficulty encountered in utilizing an instrument measuring only resistance 
losses in this application is that signals due to both diameter and wall thickness, 
in addition to signals from the other variables previously .mentioned appear 
together in the single channel read-out. Furthermore, all encircling-coil tests are 
inherently more sensitive to diameter changes than to waU-thickness changes due 
to the attenuation of the eddy current density as a function of depth. Therefore 
the difficulty in interpreting the diameter and wall-thickness signals in the single 
channel read-out is further increased. Although it is theoretically possible to use 
phase-sensitive instruments to separate these effects, available fixed-f equenc} 





thickness The use of high frequencies to produce 90-deg. separation results in 
undesirably high attenuation of the eddy current density at the inside surface. 
Use oi lower frequencies yields operating points near the resistive peak of the 
impedance plane of Fig. 21. Because of the resultant lower attenuation consider 
able improvement is obtained in the sensitivity to defects on the inner surface 
but at the expense of less phase separation between outside and inside diameter 
effects. An operating point is thus selected for a particular tube, such that small 
changes in wall thickness produce little or no change in the resistive components 

- 0.0300 

a 290 








" -^ 

/- . 



9g] 0.2300 
^t 0.2290 



< 0.2280 

10 15 20 

LENGTH (in.) 




Oak Ridge Nat'l. Lab. 

Fig. 28. Comparison of resistive and reactive component signal traces with actual 
dimensional variations in a 0.229 X 0.025-in. Inconel tube. 

but produce maximum change in the reactive component. Thus changes in the 
outside diameter produce impedance variations at an angle oblique to both re 
sistive and reactive components. These produce signals in both the resistive 
and reactive channels. The controls of the resolving bridges provided in the 
instrument can then be adjusted until one channel responds primarily to varia 
tion in diameter and the other primarily to variation in wall thickness. Small 
defects on the inner wall of the tubing cause impedance excursions which are 
nearly in phase with those from wall-thickness variations. Small defects on the 
outer surface cau&;e excursions which are nearly in phase with those from diam 
eter variations. Therefore small inside defects appear on the same channel with 
wall-thickness variation and small outside defects appear on the same channel 





with diameter variation. Gross defects cause very large impedance excursions 
at some phase angle between the effects of varying diameter and wall thickness; 
thus the large resultant defect signal is present in both channels. 

The ability of this system to distinguish between diameter and wall-thickness 
variations is illustrated in Fig. 28. The signal traces are recordings from the 
two signal channels for two 35-in. lengths of 0.229 X 0.025-in. Inconel tubing 
which were previously stretched to accentuate their dimensional variation. Above 
the traces are shown plots of the average wall thickness as measured with the 
ultrasonic-resonance thickness gage. Plots of the average outside diameter, as 
measured with mechanical micrometers, are shown below the traces. It is readily 
seen that close correlation exists between the instrument signal and the other 
independent measurements. These traces were made on the 0.229 X 0.025-in. 
Inconel tubing at a linear tube speed of approximately 1 f.p.s., and at an operat 
ing frequency of 189 kc., for which /// c = 15. 

THE SURFACE PROBE-COIL SYSTEM. From the standpoint of inter- 
pretability, the probe-coil system illustrated in Fig. 22 is very attractive. In such 
a system the probe mechanically scans the tube surface in a helical pattern. Sig- 



Oak Ridge Nat'l. Lab. 
Fig. 30. Inspection of tubing with an eddy current probe. 


nal variations aro displayed on a persistent-screen oscilloscope as a function of 
tube rotation. The advantage of the probe coil is its inherent ability to be de 
finitive, Because it is a surface probe, it is not affected by changes in diameter. 
Changes in wall thickness are indicated by changes in the signal level, which is 
a function of the position along the tube but independent of the lube rotation. 
Eccentricity is indicated by changes in the signal level, which varies slowly as a, 
function of tube rotation. Discontinuities within the wall of the tube, because of 
their abrupt nature, cause sharp indications in the pattern. 

Coil Holder Assembly. The probe coil and holder assembly illustrated in Fig. 
29 nre typical of those which have been constructed and successfully utilized, 
Probe coils as small as M,o-in. diam. have been constructed and found to be very 
sensitive to minute defects. This small size results in very slow scanning rates. 
Therefore elongated probes must be utilized, although the sensitivity to minute 
discontinuities is reduced in proportion to the increased length. The coil mounting 
is spring-loaded to hold the coil face in contact with the tube surface of the tube. 
Replaceable inserts allow the mechanism to be used for a range of tube 

Control of Coil-to-Tube Spacing. The difficulty with the probe-coil system 
is that even with a very carefully designed coil-holding mechanism, wobble of the 
tubing does occur, and the output signal varies with the coil-to-tube spacing. 
These signals sometimes look like those obtained from cracks. To obtain a first- 
order discrimination against wobble or lift-off signals, an instrument has been 
built which utilizes phase-sensitive detectors, as illustrated in the block diagram 
of Fig. 30. In this system the probe is composed of two coils: the exciting coil, 
fed by alternating current of constant amplitude, and the pick-up coil, placed in 
proximity to the exciting coil. The voltage from the pick-up coil is compared with 
that from adjustable balance circuits in the instrument. The difference is amplified 
and detected in phase-sensitive circuits in which the reference phase is continu 
ously variable. Thus, lift-off signals can be eliminated from one of the detector 
channel outputs, leaving only signals from defect or wall-thickness changes. Such 
a system has three major undesirable features : 

1. The phase of the lift-off signals varies with the amount of the lift-off. The 
phase-control adjustment which eliminates the lift-off signal of one magnitude 
will not, in general, remove the lift-off signal of a different magnitude. 

2. Discrimination against signals from lift-off will, in general, result in discriminat 
ing against signals from very small discontinuities on the outside surface of 
the tube. 

3. Successful operation of these circuits depends upon choice of n tost frequency 
according to the wall thickness and tube material, such that phase separation 
between lift-off signals and wall thickness signals approaches 90 dog. 

The difficulties are also encountered with phase-sensitive detectors in instrumenta 
tion employing encircling coils. 

Advantages of Probe-Coil System. Despite difficulties encountered in the 
application of the probe-coil system, its superior definitive abilities and ease of 
signal interpretation cannot be overemphasized. The typical oscilloscope pat 
terns shown in Fig. 31 demonstrate the ease with which signals obtained from this 
system can be interpreted. This method is mechanically difficult and is inherently 
a much slower method than that of the encircling coil. However, its superior 
definitive abilities can more than offset its disadvantages in many inspection 



(a) No indication of defects. 

(b) Indication of tube wall eccentricity. 

(c) Indication of small defect. (d) Indication of large defect. 

Oak Ridge Nat'L Lab. 
Fig. 31. Probe-coil eddy current B-scan photographs. 

LIMITATIONS IN TEST AUTOMATION. The techniques and instru 
mentations of eddy current methods presently employed for small-diameter 
tubing have not yet advanced to the level of automation. Eddy current testing 
cannot be made fully automatic until the method can be standardized to give, at 
least, semi-quantitative indications of defect depth and length. The problem of 
standardization will be very difficult because : 

1 . Every variation in the tubing may produce indications which obscure or confuse 
the defect indication. 

2. If the signals are differentiated, either in terms of time or distance, the signal 
amplitude will be dependent on the rate of change rather than on the defect 

3. The sensitivity of the test to a given defect is dependent on the selected test 
conditions, since these test parameters describe a position on the impedance 
plane, as shown in Fig. 21. 

4. The density of the induced currents decreases with depth into the tube wall. 
Hence, similar defects at the inner and outer surfaces of the tube wall give 
signals of different amplitudes. Eddy current tests will be difficult to stand 
ardize unless the location of the defect can be determined. 

5. The impedance change caused by a defect is a change in both amplitude and 
phase. These changes are dependent not only on the type of defect but also 
on the defect location in the tube wall. 

The complexities of the eddy current method make automation difficult unless 
the inspection is intended to locate only large flaws. For more critical inspection 
applications, the method requires inspectors with the experience and judgment of 
trained engineers to separate and evaluate the signals from the large number of 
possible variables. Furthermore, present-day eddy current instrumentation avail- 



able for use on small-diameter tubing is not sufficiently versatile to provide the 
inspector with adequate information in a form that is readily interpretable. 

strumentation used in a given eddy current test, the type of signal display system 
utilized is of primary importance in determining signal inter pretability. Experi 
ence has indicated that interpretation can be greatly enhanced by introducing 

(a) Probe-coil eddy current B-scan fre- (b) Encircling-coil eddy current trace fre 
quency, 200 kc. quency, 200 kc. 

f r '^-^' r ^%^^^^^^Wa^ 

(c) Cross-section of crack in tube. Oak Ridge Nat'l. Lab. 
Fig. 32. Eddy current signals from defective % 6 X 0.025-in. Inconel tubing. 

positional intelligence into the signal-display system. This is illustrated in Fig. 
32, which compares indications caused by a gross crack found in a length of 
Me X 0.025-in. Inconel tubing inspected by both the probe-coil method and the 
encircling-coil method. A cross-section of the crack is shown in Fig. 32(c). The 


trace from the encircling-coil test is shown in Fig. 32 (b) and indicates the presence 
of the crack by the sharp upward spike. The trace from the probe coil for the 
same defect is shown in Fig. 32 (a) . In both cases the amplitude of the signal from 
the crack was not large enough, compared with the other signals obtained from 
dimensional variations in the tube, for reliable detection. When displayed either 
as a function of tube length or tube rotation, however, the sharpness of the signals 
was sufficient to allow the detection and interpretation to be made easily and 

APPLICATION AREAS. At their present stage of development, eddy 
current techniques can, if properly utilized, perform valuable inspection on small- 
diameter tubing. For the detection of gross defects in commercial-grade tubing, 
the encircling-coil technique is the fastest and most effective nondestructive test 
which exists today. The inspection of premium-quality tubing for critical 
applications by this method allows: 

1. High-speed detection of intergranular corrosion on the inside surface. 

2. The detection of foreign metal pick-up on the inside surface. 

3. High-speed continuous gaging of the dimensions of the tubing. 

None of these inspections can be accomplished by any other method. 


Reference numbers in this section refer to the references at the end of the 
section on Eddy Current Test Indications. 





Eddy Current Tests of Spheres 

Significance of spherical test objects 1 

Determination of the limit frequency fg 1 

Effective permeability, /tteff., of the sphere .. 3 

Effective permeability for a sphere of high 

relative magnetic permeability, #roi. 

(/. 1) 2 

Normalized impedance and secondary voltages 

for a sphere 3 

Sphere in various test coil arrangements 

(/.2) 3 

Characteristics of the complex voltage or im 
pedance plane 4 

Apparent impedance diagram for a short, 
circular test coil containing a sphere 

(/. 3) 5 

Effect of changes in fill factor 77 4 

Effect of variations in conductivity and rela 
tive permeability 4 

Separation of diameter changes 4 

Effect of extending teat coil length 4 

Effect of low test frequency or high relative 

permeability 4 

Shape permeability and demagnetization factor 4 
Demagnetization factor N for a sphere 6 
Effect of variation in relative permeability 

of material 6 

Apparent impedance diagram for a short, 
circular coil containing a sphere, for fill 
factor TI = 1 and relative magnetic per 
meabilities /irei. =50, 5, 2, and 1 (/. 4) 7 

Shape permeability for short cylinders 6 

Short cylinder with axis transverse to axis of 

test coil 

Computing effective permeability 7 

Computing initial points of apparent im 
pedance curves ? 

Prolate ellipsoids 8 

Apparent impedance curves for short cylinder 

or ellipsoid 8 

Complex voltage and impedance planes for 

non ferromagnetic spheres 8 

Complex impedance or voltage plane for 
circular coil with nonferromagnetic 

sphere (/. 5) 9 

Application to small metallic impurities in 

nonmetallic test objects 10 

Crack testing of spheres 10 

Three main positions of cracks in spheres, 
relative to the direction H of the a.-c. 

magnetizing field (/. 6) 10 

Apparent impedance plane for spheres with 
cracks ^ 


Apparent impedance plane of a test coil 
containing spheres free from cracks 
(open-circle points) and with cracks 

(black dots) (/. 7) 12 

Determination of fill factor i) 11 

Field strength and eddy current distribution 
in spheres 11 

Eddy Current Tests of Sheets and Foils 

Applications of sheet and foil tests 12 

Principle of fork coil tests 12 

Arrangement of a fork coil system consist 
ing of a primary magnetizing coil and a 
secondary pick-up coil between which a 
flat metallic sheet, M, is located (/. 8) 13 

Transmission coefficient T 13 

Limit frequency fg 13 

Independence of test -object position in 

forked coil 14 

Plane of complex transmission coefficient T ... 14 
Complex plane of transmission coefficient 
T, for a flat metallic sheet, as a function 

of frequency ratio flfg (/. 9) 15 

Analysis of the complex T plane 14 

Similarity law for tests of flat conductors . . 14 

Reaction of eddy current field 14 

Maximum test sensitivity 15 

Sensitivity diagram for indication of ef 
fects of conductivity, thickness, and 
magnetic permeability variations in flat 

sheet materials (/. 10) 16 

Tests for conductivity and thickness of flat 

conductors 16 

Differential circuit arrangement 17 

Circuit for measuring the product of con 
ductivity and thickness of flat metallic 
sheets by frequency adjustment (/. 11) 17 
Determining the product of conductivity and 

thickness of metallic sheets 17 

Application of calibrated RC oscillator .... 18 

Arrangement for measuring the thickness 

of flat metallic sheets with a calibrated 

RC oscillator (/. 12) IS 

Direct indication of thickness of metallic 

sheets and foils 18 

Direct measurement of resistance per unit 

square . 


Control of film resistance during metal dep 
osition 19 

Differential arrangement for the measure 
ment of resistance per unit square for 

thin metallic foils (/. 13) 19 

References 19 




Eddy Current Tests of Spheres 

for spheres placed in the electromagnetic field of an eddy current test coil is im 
portant for several reasons. Most production test parts, such as bolts and bear 
ing rollers, are relatively short. In these cases the demagnetizing influence of the 
ends of the test part becomes more obvious. The assumptions used in the analysis 
of the infinitely long cylinder in a test coil providing a homogeneous field are no 
longer applicable. The sphere is the more typical case for the short production 
part. Because of its symmetrical form, the case of the sphere is approachable 
with exact mathematical calculations. The results so obtained for the sphere can 
then be related to other short production parts, and several important character 
istics of the behavior of the short part in the a.-c. magnetic field can be derived. 

Another application of the basic methods for analysis of the sphere is in the 
detection of small ferromagnetic or nonferromagnetic impurities in a non- 
metallic substance such as paint or food products. The theory provides exact 
formulas for the measurement voltage caused by a small metallic particle passing 
through the metal detection coil. This voltage is a function of the particle diam 
eter, electrical conductivity, and magnetic permeability, as well as of the test fre 
quency and the coil size. 

the effect of a sphere in an a.-c. electromagnetic field, Maxwell's equations are 
transformed into spherical coordinates. The solution of these differential equations 
involves the usual boundary conditions : 

1. Continuity of the tangential components of the electric and magnetic field 

2. Continuity of the normal component of the magnetic induction, leading to 
solutions in the form of hyperbolic functions. 

As discussed in the section on Eddy Current Cylinder Tests, a limit frequency 
i a is defined as the frequency at which the argument of the functions used for the 
calculation becomes equal to unity. In the case of the sphere this is the argument 
of the hyperbolic functions. The following equation holds for the limit fre 
quency f g of the sphere : 

. _5066|Arel. m 

/ <~~^ r ~~ U> 

where a = electrical conductivity of material in sphere, meter/ohm-mm. 2 . 

B - diameter of the sphere, cm. 
Hrei. = relative magnetic permeability of material in sphere. 

In the case of nonmagnetic materials, [i reli = 1. 

39 1 



Instititt Dr. Foerster 
Fig. 1. Effective permeability for a sphere of high relative magnetic permeability, 



permeability [x eff . introduced for the sphere is similar to that used previously in 
analysis of cylinders. Its real and imaginary components are obtained from Fig. 1. 
The effective permeability of the sphere is obtained by calculating the frequency 
ratio f/f g from 

where / is the test frequency. These f/f g ratios are shown along the curve of 
Fig. 1. The real and imaginary components of the effective permeability are 
read from the coordinates of these points. For example, if f/f g = 1, these com 
ponents appear as 

^ff.(roal) = +0.25 


jU,eff.(ininR.) H~0.31 

FOR A SPHERE. Fig. 2 shows coil arrangements used for testing of spheres 
and short production parts. Fig. 2 (a) shows an arrangement of a single test coil 

Fig. 2. Sphere in various test coil arrangements, (a) Sphere in a single test coil, 
(b) Sphere in a coil arrangement of a primary coil (not shown here) and a secondary 

coil (shown). 

whose impedance components are col/ and R. Fig. 2(b) shows an arrangement 
with primary and secondary coils, for which the secondary coil voltage compo 
nents are # ima g. and # real . The normalized components of impedance and sec 
ondary voltage for this case are 

joL _ ginug. =1 , 
coLo Eo 

I M-eff.(real) 

M-cff ( 





- l,Upff.<imnB.) 


where B/D is the ratio of the sphere diameter B to the coil diameter D. As in 
the case of the cylinder, the coil fill factor r[ is defined as i| = (B/D)' 2 . The 
expression \/l + (l/D)' 2 takes the coil dimensions into account, where / is the 
length of the coil. 

For the short, circular coil where I D, the denominators in Eqs. (3) 
become unity. An extension of the length of the test coil merely decreases the 
influence of the sphere upon the coil impedance, since with a long coil a portion 
of the coil is not affected by the insertion of the sphere. For example, in a coil 
whose length I is equal to its diameter D, so that (l/D) = 1, the influence of the 
sphere is l/\/2 = 71 percent of the effect for a short, circular coil. 

PEDANCE PLANE. Fig. 3 shows the test coil characteristics as a function of 
the frequency ratio /// calculated from Eqs. (3). It illustrates the influence of 
the characteristics of the sphere (diameter B, conductivity a, and relative perme 
ability |A re i.) and of the test frequency / on the normalized impedance or second 
ary coil voltage. It also illustrates the variation in coil characteristics (inductive 
reactance coL and coil resistance R) when the diameter of the sphere decreases 
from the value (B/D) = 1 (at which the sphere entirely fills the test coil) to 
smaller diameters. 

Effect of Changes in Fill Factor i]. Fig. 3 shows curves for a number of 
values of fill factor TJ = (B/D) 2 . The bending of the sphere diameter curves 
(labeled B) with decreasing sphere diameters results from the simultaneous varia 
tion in the frequency ratio f/f g [see Eq. (2) ] when the sphere diameter B becomes 

Effect of Variations in Conductivity and Relative Permeability. Variations 
in the conductivity and permeability of the sphere, as well as variations in the 
test frequency, result in displacements in the ooL and R values along the f/f g direc 
tion of the circular arcs. Especially interesting is the fact that with infinitely high 
values of relative permeability, the impedance ratio is always f/f ff = 0. Thus, at 
any test frequency, but with infinitely high permeability, the same apparent im 
pedance results as for f/f g = 0. 

Separation of Diameter Changes. The relatively large angles between the 
directions of the diameter curves (B) and the curves for changes in conductivity 
or permeability (a, |A re i.) in Fig. 3 indicate the degree to which diameter variations 
for spheres can be separated from material property variations. 

Effect of Extending Test Coil Length. As indicated by Eqs. (3), an exten 
sion in the length I of the test coil has the same effect as a decrease in the fill 
factor T), without influencing the frequency ratio f/f g . Lengthening the test coil 
to twice the coil diameter (I = 2D) results in the same fill-factor curve in Fig. 3 
as a decrease in sphere diameter from (B/D) = 1 to (B/D) = 0.76 [see Eqs. (3)]. 

Effect of Low Test Frequency or High Relative Permeability. Fig. 1, 
showing the effective permeability values of the sphere for any given frequency 
ratio, is accurate only for large values of relative permeability ([A re i. ~ 100). For 
very small frequency ratios (static case) or for very large relative permeability 
values, the normalized inductive reactance, a)Z//o)L = 3 for fill factor TJ = 1. 


Generally, in short production parts, the apparent or shape permeability, ^ app ., 
is considerably smaller than the relative permeability of the test material because 



Institut Dr. Foerster 

Fig. 3. Apparent impedance diagram for a short, circular test coil containing a 


39 5 


of the high demagnetization effects. [Sec Eq. (1) in the section on Magnetic-Field 
Test Equipment for discussion of shape or form permeability effects.] In general, 

where A r is the so-called demagnetization factor, which becomes larger as the 
length-to-cliameter ratio of the test part becomes smaller. 

Demagnetization Factor N for a Sphere. For a sphere, the demagnetization 
factor is 

W = 4ic/3 

so that the apparent permeability [from Eq. (4)] is [_I. M)I)I = 3. Now the self- 
induction of a coil is proportional to the permeability of its core. Thus, if the 
sphere has an apparent permeability of 3, the self-induction of the small, circular 
coil shown in Fig. 2 must be three times larger when the sphere is present, (at 
f/f ff = 0, the static case) and fill factor T) = 1, than without a sphere within the 
coil. Thus, as shown in Fig. 3, the normalized inductive reactance o)L/(oZ; = 3. 

Effect of Variation in Relative Permeability of Material, Fig. 4 illustrates 
curves for the apparent impedance of the test coil for four different values of 
relative magnetic permeability of the test material. The curve for ^ rol> = 1 applies 
for nonferromagnetic spheres; for example, of stainless steel. The difference in 
the apparent impedance curves for small relative permeability values is especially 
obvious for low frequency ratios f/f g . For values of pi rol . > 5 and }/f ff values 
exceeding unity, the apparent impedance curves all coincide (lower half of outer 
most curve in Fig. 4). Thus the calculations of effective impedance arc exact 
down to the very low permeability value of pi rel< = 5, corresponding, for example, 
to permanent magnets or to tungsten carbide. 

Shape Permeability for Short Cylinders. As previously indicated, the appar 
ent impedance curve for the short test coil, for which l D and \\ = 1, starts 
at f/f ff = at the value of the apparent shape permeability for the sphere with 
u app . = 3, namely, coL/o)L = 3. The shape permeability for a short test cylinder 
can be obtained from Eq. (4) and the demagnetization factor N. Values of de 
magnetization factors N given by Bozorth, n can be used for cylinders with 
different length-to-diameter ratios. For example, the value given for the de 
magnetization factor N is 3.4 for a short cylinder whose length is equal to its 
diameter (l/D = l). From this, the apparent permeability |i ni)1 >. = 3.7 is calcu 
lated from Eq. (4) . The apparent impedance curve for such a short cylinder starts 
at GoZ//coL = 3.7 at f/f g = 0, as shown in Fig. 4. 

Short Cylinder with Axis Transverse to Axis of Test Coil. If the short 
cylinder is turned so that its longitudinal axis is perpendicular to the coil axis 
while within the coil, its apparent permeability decreases from 3.7 to 2.0. The 
demagnetizing factor N of the cylinder perpendicular to its longitudinal axis is 
calculated as N = 2jt. Substituting this value in Eq. (4) gives the value of 
[i app- = 2. The normalized reactance of the test coil also changes from 
coL/o)L = 3.7 to coZ/AoZ/o = 2 as the short cylinder turns from the longitudinal 
to the transverse position within the coil. Thus an inexact positioning of the 
cylinder in the longitudinal axis of the coil can cause considerable variation in the 
apparent impedance. This positioning must be carefully controlled in eddy 
current testing of roller-shaped test objects. 




Institut Dr. Foerster 

Fig 4. Apparent impedance diagram for a short, circular coil containing a sphere, 
for fill factor i\= 1 and relative magnetic permeabilities JI M I. = 50, 5, 2, and 1. 

Dashed curve is initial portion of curve for a cylinder with l/D - 1. 

and R of the apparent impedance of the test coil are obtained by Eqs. (3) from: 

1. The test coil dimensions (I, D). 

2. The coil fill factor r). 

3. The components of effective permeability, M..<rn and M-eff.cima B .). 

The effective permeability components can be taken from Fig. 1 for any given 
frequency ratio ///, of the sphere. The components of the apparent impedance 
are derived from the effective permeability values and are shown in Jng. 6. 

Computing Initial Points of Apparent Impedance Curves. The initial point 
of the apparent impedance curve corresponds to the frequency ratio /// - U, tne 
static case. For fill factor T[ = 1, 

'coL\ _.. (5) 


Prom Eqs. (3a) and (5), the initial point of the effective permeability curve at 
f/f ff = is generally found to be 

IWf.Creal)] (f/fy=0) = (|A.pp.)/2 ~ 0.5 (6) 

According to McClurg, 70 the approximate relationship for the apparent perme 
ability of a cylindrical body of length I and diameter D, where 2 < (l/D) < 10, is 

Happ. =6/Z-5 (7) 

Thus the initial point of the apparent impedance curve for short cylinders is 
calculated to be 

[fioff.(ra.l>] lf/ r g=0 ) = 3(Z/Z - 3 (8) 

No approximation formula exists for computing the apparent permeability for 
short cylinders with length/diameter ratios between 1 and 2. In this case Eq. (4) 
can be used for the calculation if the demagnetizing factor N is taken from 
Bozorth's curves e for any given l/D ratio for the cylinder. 

Prolate Ellipsoids. For prolate ellipsoids with ratios of major to minor diam 
eters between 1 and 10, Foerster 2 has derived the following approximation 
formula : 


where I is the large axis and D the small axis of the ellipsoid. From this approxi 
mation the initial point for the effective permeability curve at f/L = Q is 
obtained as ' 

[Mf.(real)] <r / tgSffi = 1.5(J/D) - 0.5 (9b) 

Either Eq. (6) or Eqs. (9) is used for calculating initial points, depending upon 
whether the shape of the short test object more nearly approximates a cylinder 
or an ellipsoid. 

Apparent Impedance Curves for Short Cylinder or Ellipsoid. To calculate 
the normalized reactance a)Z//coL of the test coil containing either a short cylinder 
or ellipsoid at /// = 0, the approximation formulas [Eqs. (7) or (9a)] are inserted 
into Eq. (5). Since the apparent impedance curves for short cylinders and short 
ellipsoids are similar (see Fig. 3), the behavior of short production parts in the 
a.-c field can be estimated. However, an exact calculation for the short cylinder 
in the a.-c. field is not possible because the magnetization is not homogeneous over 
the length of the cylinder and decreases toward the ends as a result of the de 
magnetization. As the demagnetizing influence of the ends of the cylinder 
diminishes, however, the apparent permeability more closely approaches the 
relative permeability of the material. Under these conditions the apparent im 
pedance curves of Fig. 4 tend to approach the curves for the infinitely long, solid 
cylinder shown in Fig. 2 of the section on Eddy Current Cylinder Tests. 

FERROMAGNETIC SPHERES. The performance of the test coil containing 
a nonferromagnetic sphere is of special interest, since it approximates its behavior 
with short nonferromagnetic objects. This situation is intermediate between the 
case of the long, solid cylinder and that of the sphere. The impedance or complex 
voltage plane for a test coil containing a nonferromagnetic sphere is computed 
from fcqs. (3) and the apparent impedance values for ^ el = 1 in Fig 4 Fig 5 
snows the resultant family of curves. These are very similar to the curves for the 
infinitely long cylinder given in Fig. 2 of the section on Eddy Current Cylinder 




Institut Dr. Foerster 

Fig. 5. Complex impedance or voltage plane for circular coil with nonferro- 
magnetic sphere. Direction B corresponds to increasing sphere diameter. 

Tests, although the path of the apparent impedance curve of Fig. 5 in this section 
is not identical with that of Fig. 2. 

Eq. (5) of the section on Eddy Current Test Principles, and Eq. (1) of this 
section for the limit frequency /^ are the same for the nonferromagnetic sphere 
and the nonferromagnetic cylinder (fi reL = 1). Fig. 5 shows that the same varia 
tions in apparent impedance are obtained for the sphere at an j/j g ratio approx 
imately twice as high as for the cylinder. By comparing the test specimen with a 



reference standard in differentially connected test coils (the so-called absolute 
method), the quantitative measurement of the electrical conductivity is pos 
sible by means of Fig. 5. 

Application to Small Metallic Impurities in Nonmetallic Test Objects. 

The preceding theory of the influence of the sphere on the apparent impedance 
or secondary voltage of the test coil can be used to compute the effects resulting 
when an a.-c. coil is used to search for metallic impurities (assumed to be sphere- 
shaped) in nonmetallic substances. If the smallest voltage variation which can 
just be indicated is known for the test coil, the minimum detectible impurity 
size can be calculated in terms of : 

1. The test coil size (diameter D and length 0- 

2. The properties of the spherical impurity (conductivity a and relative per 
meability (Irol.). 

3. Fig. 1 for ferromagnetic sphere-shaped impurities. 

4. Fig, 5 for nonferromagnetic sphere-shaped impurities. 

^CRACK TESTING OF SPHERES. The influence of cracks in spheres was 
detennrnied -ex-perimentally, since complicated boundary conditions made mathe- 

~ (a) 


~ (b) 


~ (c) 

Fig. 6. Three main positions of cracks in spheres, relative to the direction H of 
the a.-c. magnetizing field, (a) Large, position for greatest crack indication sensitiv- 
iiv (b) Medium, position for medium crack indication sensitivity, (c) Small, 
position for minimum crack indication sensitivity'. "" 



matical solutions impossible. Surface cracks in spheres can have three main 
positions, depending upon the location of the crack in the a.-c, field of the test coil 
(see Fig. 6). These positions are: 

1. Long dimension of crack paralleling direction of magnetizing field, Fig. 6 (a), 
producing a large test indication. 

2. Short dimension (depth) of crack paralleling direction of magnetizing field, 
Fig. 6(b), producing an intermediate test indication. 

3. Crack transverse to direction of magnetizing field, Fig. 6(c), producing mini 
mum test indication. 

In the optimum crack position, which produces a large test indication, the crack 
surface is perpendicular to the equatorial plane. The lines of force of the primary 
magnetizing field are parallel to the length of the crack. The equatorial plane 
is the one having the greatest eddy current density, which explains the large 
sensitivity of crack indications. In the intermediate position, Fig. 6(b), the crack 
surface in a region of medium eddy current density is perpendicular to the eddy 
current pole axis. When the crack is in the least-detectible position, Fig. 6(c), 
the plane of the crack is parallel to the direction of eddy current flow. Here it 
has no reaction upon the eddy currents and produces no weakening of the eddy 
current density. 

Apparent Impedance Plane for Spheres with Cracks. Fig. 7 shows a small 
portion of the apparent impedance plane of Fig. 3, showing variations found 
experimentally as a result of cracks in spheres. The apparent impedance values 
of several crack-free spheres are shown by open circles in the dotted rectangle 
near the bottom of the graph. The slight variations of the positions of these 
points (open circles) result from the small degree of anisotropy found in the 
spheres (probably a consequence of the grain orientation in the section of rod 
from which the spheres were made). Varying the position of crack-free spheres in 
the test coil had very little effect upon the external impedance of the test coil. 

The indications obtained from spheres containing cracks are shown by solid 
black dots in Fig. 7. At the bottom of the graph, within the dotted rectangle, are 
black dots corresponding to tests of spheres with cracks in the position of 
minimum detectibility [see Fig. 6(c)]. However, if the crack in a sphere is 
brought into the position of medium sensitivity [see Fig. 6(b)], its indications 
appear near the middle portion of the curves of Fig. 7, as noted by ''Medium." 
With cracks oriented into the optimum test position [see Fig. 6 (a)], the test 
indications are changed considerably, as shown by the black dots near the top of 
the curves of Fig. 7. The "empty value" of the impedance, o)L , of the test coil 
in the absence of a test sphere corresponds to the point eo^/co^o = 1 in Fig. 3. 
Thus cracks in spheres result in an increase of the normalized reactance along 
the apparent impedance curves of Fig. 3 or 7. 

Determination of Fill Factor r|. The fill factor r| = (B/D) 2 can be deter 
mined for spherical test objects, as follows: Suppose that the impedance coordi 
nates in Fig. 3 for a test coil containing a sphere are, as shown by the light coordi 
nate lines, (o/G)L = 1.11 and (R JS )/coL = 0.1. These coordinates indicate 
that the fill factor rj = 0.3 because the (dashed) apparent impedance curve for 
T] = 0.3 intersects the point given by these coordinates. 

Field Strength and Eddy Current Distribution in Spheres. The distribu 
tion of field strength and eddy current densities on the surface and within the 
interior of spheres, as well as the quantitative influence of crack conditions, will 
be discussed more thoroughly in the literature. 42 














Fig. 7. Apparent impedance plane of a test coil containing spheres free from 
cracks (open-circle points) and with cracks (black dots). 

Eddy Current Tests of Sheets and Foils 

of application for the eddy current test method is in noncontacting measure 
ments of the thickness, electrical conductivity, and magnetic permeability of flat 
metallic sheets, foils, and surface films. These techniques have been developed 
for direct measurement of the electrical conductivity and for the quantitative 
measurement of corrosion attack upon sheet materials. Instruments utilizing 
these test methods are described in the section on Eddy Current Test Equipment. 

PRINCIPLE OF FORK COIL TESTS. Fig. 8 shows the arrangement of 
the fork coil for tests of sheet materials. A generator sends an alternating current 
(a.c.) of frequency / through the primary or magnetizing coil. The test material 
is placed between the primary coil and a secondary or pick-up coil. The field of 
the primary coil at the secondary coil location, at a distance A from the primary 
coil, induces an a.-c. voltage in the secondary coil. In the absence of the interven 
ing test sheet, the secondary voltage is E Q . As soon as a metallic sheet M is in 
serted between the coils, the primary magnetizing field produces eddy currents 
in the sheet whose fields are superimposed upon the weakened fields of the 
primary coil. Thus, the insertion of the metallic layer M between the primary 


and secondary coils changes the original primary field at the location of the 
secondary coil. As a result of the screening action of the metal sheet, a new 
a.-c. voltage, E M , appears Across the terminals of the secondary coil. 



f \ 



%=>^ r 



Institut Dr. Foerster 

Fig. 8. Arrangement of a fork coil system consisting of a primary magnetizing 
coil and a secondary pick-up coil between which a flat metallic sheet, M, is located. 

Transmission Coefficient T. The ratio of the secondary coil voltage E M , with 
the test sheet inserted, to the "empty coil" voltage E Q in the absence of the test 
object is termed the transmission coefficient T. This transmission coefficient is a 
function of : 

1. The electrical conductivity a of the metallic sheet. 

2. The relative magnetic permeability M-rei. of the sheet. 

3. The thickness D of the metallic sheet. 

4. The test frequency /. 

5. The distance A of the primary coil from the secondary coil. 

Similarity laws, analogous to those developed for the case of solid cylindrical test 
objects (see Similarity Law in Eddy Current Testing in the section on Eddy 
Current Cylinder Tests), illustrate the relations between these controlling vari 
ables and the transmission coefficient, T = E M /E . 

Limit Frequency f g . The theory of the forked coil was solved mathematically 
as a boundary-value problem, using Maxwell's equations. The solution of the 
differential equations had to satisfy the following boundary conditions : 

1. Continuity of the tangential components of electric and magnetic-field 

2. Continuity of the normal component of the magnetic induction through the 
metallic surfaces. 

The extensive calculations, which cannot be discussed further here, again resulted 
in a limit frequency, f s , chosen by equating to unity the argument of the function 
which appeared in the calculation of the transmission coefficient T. The limit 
frequency for the case of a flat conductor within the forked coil is 

/, = T^T (WW 

and the frequency ratio f/f s is 

///. = ^ (W 

where a = electrical conductivity of sheet material, meter/ohm-mm. 2 . 

D = thickness of flat metallic conductor, cm. 
HP.I. relative magnetic permeability of flat metallic sheet. 

A = distance (cm.) of the primary coil from the secondary coil (a fixed con 
stant of the test instrument in most cases) . 


It should be noted that the factor A, representing the geometry of the coil 
arrangement (see Fig. 8), is contained in the expression for the limit frequency 
[Eq. (10a)] . This is contrary to the situation in the case of the solid cylinder and 
the hollow tube in the feed-through coil arrangement, where the limit frequency 
was independent of the specific coil arrangements. This dependency upon A can 
be used to advantage in computing the "lift-off" effect of the test coil, described in 
the literature. 4 - 

Independence of Test-Object Position in Forked Coil. The distance A be 
tween primary and secondary coils appears in Eq. (10), but the distance of the 
metallic sheet from, the coils does not appear. As indicated by theory, the influ 
ence of the metallic layer upon the secondary coil characteristics is independent 
of the position of the metallic layer between the two coils. 


transmission coefficient consists of two components, a real component, jT rett i, and 
an imaginary component, T im&gn . These are plotted on a complex plane in the same 
manner as the components of the effective permeability for the cylinder and tube. 
Various frequency ratios are noted along the transmission-coefficient curve of 
Fig. 9, which shows the characteristics of T = E M /E^ as obtained by extensive 
calculations. With zero sheet thickness (no material within the forked coil), the 
frequency ratio /// tf = 0, and the transmission coefficient T f =T = l (a real 
number) . 

Analysis of the Complex T Plane. To illustrate the use of the complex plane 
of the transmission coefficient T, assume that the test object is an aluminum foil 
with the following characteristics : 

1. Thickness, D = lOp, = 0.001 cm. 

2. Electrical conductivity, a = 38meter/ohm-mm. 2 . 

3. Relative magnetic permeability, (i PO i. = 1. 

The distance of the primary coil from the secondary is assumed to be 10 cm. The 
limit frequency is calculated from Eq. (10) as f g = 6660 c.p.s. If the test fre 
quency is chosen as 6660 c,p.s., the frequency ratio f/f g = 1. In this case the vec 
tor representing the transmission coefficient T for the foil extends from the origin 
to the point at which f/f ff = 1 on the curve of Fig. 9. Before insertion of the 
foil into the test coil system, the vector representing the transmission coefficient 
T extended from the origin to the point T = +1, along the vertical (real) axis. 
Upon insertion of the test material, the transmission coefficient vector under 
went both a shift in phase and a reduction in amplitude to 87.5 percent of its initial 
value. This change was caused by the field of the eddy current induced in the foil. 
If the test frequency for the preceding foil test had been selected as / = 66,600 
c.p.s., the vector representing the transmission coefficient T would have extended 
from the origin to the point where f/f y = 10 on the curve of Fig. 9. Thus, from 
Eq. (10) and Fig. 9, the influence upon secondary coil voltage can be determined 
for any given metal thickness, electrical conductivity, and magnetic permeability. 

Similarity Law for Tests of Flat Conductors. The similarity law for eddy 
current testing of flat conductors states: 

Any given metallic material and any given flat sheet thickness result in the same 
transmission coefficient T if the test is made at the same frequency ratio j/j g as given 
byEq. (10). 

Reaction of Eddy Current Field. The reaction field of the eddy currents in a 
test sheet is superimposed upon the original field, T = 1, of the test coils in the 



absence of the test object. The effect of the eddy current field is described by the 
vector W in the complex plane of Fig. 9. Thus the undisturbed primary field 
(described by vector T ) and the eddy current field (described by vector W) act 
simultaneously upon the secondary coil. Their vector difference is the vector T. 
Inversely, by adding the vector representing the field acting upon the secondary 
coil when a test object is present (vector T) to the secondary field in the metal 
plate (vector W), one obtains the undisturbed primary field (vector T = 1). 

T reat 

Institut Dr. Foerster 

Fig. 9. Complex plane of transmission coefficient T, for a flat metallic sheet, as a 
function of frequency ratio ///. 

Maximum Test Sensitivity. The highest indication sensitivity for the metallic 
sheet in the forked coil arrangement is determined, as a function of the electrical 
conductivity and thickness of the test sheet, in a manner analogous to the deter 
mination for optimum test sensitivities for cylinders and tubes, presented pre 
viously. Fig. 10 shows the sensitivity diagram for the fork-coil test method. 
The line connecting the origin to a specific ///, value on the curve represents the 



variation AT in the transmission coefficient (shown as ATF in Fig. 10) for 1 per 
cent variation in conductivity or thickness in the test material. As in all eddy 
current problems, the maximum indication sensitivity occurs at the maximum of 
the imaginary component of T, corresponding to the maximum eddy current 
loss. In Fig. 10 this point is located at /// = 2.75, at the bottom of the circle. 

-2-KT 3 -1-1Q- 3 

+2-KT 3 - 

Fig. 10. Sensitivity diagram for indication of effects of conductivity, thickness, 

and magnetic permeability variations in flat sheet materials. Vector diagram shows 

variation of transmission coefficient, AT = AT7, for a variation of 1 percent in test 

object conductivity or thickness. 

For this frequency ratio, a conductivity or thickness variation of 1 percent corre 
sponds to a variation of 3.85 X 10~ 3 in the transmission coefficient T. At this 
point of optimum sensitivity, the direction of the T variation is parallel to the real 
axis (vertical direction) in Fig. 9 (the direction from /// = 2.6 to /// = 3.0 in 
Fig. 9). 

DUCTORS. The sensitivity diagram of Fig. 10 provides a basis for a method of 
determining the magnitude of the product (aD) of conductivity and thickness 

of flat metallic sheets. In this arrangement the scale of the indicating meter has 
the same calibration (in percentage of the deviation factor aD) for both positive 
and negative variations and is independent of the specific values of conductivity 
and thickness for the metal sheet. 



Differential Circuit Arrangement. Fig. 11 shows the circuit used for measur 
ing the magnitude of the product 0D, which is described in European literature 
as the "absolute measurement method." A variable-frequency generator feeds 
the primary magnetizing coil P, which acts identically upon the two similar sec 
ondary coils S 1 and S 2 . The voltages induced in secondary coils Si and S 2 are 
separately amplified, and after rectification, are connected in series opposition as 
d.-c. voltages. To calibrate the test instrument, the switch at the input to the 





Institut Dr. Foerster 

Fig. 11. Circuit for measuring the product of conductivity and thickness of flat 
metallic sheets by frequency adjustment. 

second amplifier is turned to C (calibration). In the absence of a test object be 
tween the primary P and secondary coil S lt the amplifier output voltages must 
be identical, and the indicating meter will show zero. As shown in Fig. 9, the 
distance T from the origin to the f/f g = 2.75 point (where highest measurement 
sensitivity is attained) amounts to 68 percent of the "empty coil" voltage T = 1 
(no test object). To measure the product aD of a metallic sheet or foil between 
P and Si, the input switch of the second amplifier is turned to M (measurement), 
which decreases the voltage from S 2 to 6S percent of its true magnitude. 

Determining the Product of Conductivity, and Thickness of Metallic 
Sheets. Insertion of a metallic sheet between the coils P and Si of the instrument 
shown in Fig. 11 causes the output voltage of secondary coil S x to diminish. If 
the oscillator feeding the primary coil is varied in frequency until the indicating 
meter shows a zero deflection, the amplitude of the field of the primary coil at the 
location of the secondary coil S l is also decreased to exactly 68 percent. The 
transmission coefficient T now corresponds to the line connecting the origin to 
the point at which ///, = 2.75 in Fig. 9. The magnitude of T is now 68 percent of 
T =l In this way one has selected the frequency which provides the maximum 
indication sensitivity (see Fig. 11) for conductivity and thickness variations. 
At this point, ///, = 2.75, Eq. (10) indicates that 

_ /7000\ /1\ (I'M 

aD = ( T-) (-7) u; 

The instrument constant A (the effective distance from the primary to the 
secondary coil) can be determined with a metal foil (for example, of aluminum) 



of known conductivity and thickness from Eq. (11). The determination of the 
product aD by Eq. (11) is based upon a frequency measurement which can be 
carried out with high accuracy. 

Application of Calibrated RC Oscillator. A calibrated resistance-capacitance 
(RC) oscillator can be used as the generator in the system shown in Fig. 12. As 
is well known, the frequency of such an oscillator is inversely proportional to the 
resistance R and the capacitance C of the RC network. In this case, as indicated 
by Eq. (11), the product aD is proportional to the resistance R. of the oscillator 
(if a fixed capacitance C is employed) . 

Institut Dr. Foerster 

Fig. 12. Arrangement for measuring the thickness of flat metallic sheets with a 
calibrated RC oscillator. Thickness is read directly from the scales of three decade 
resistances. Variations of electrical conductivity with alloy or temperature are com 
pensated with the capacitance C. 

Direct Indication of Thickness of Metallic Sheets and Foils. The capaci 
tance C of the oscillator in the system shown in Fig. 12 can be adjusted to 
correspond with the electrical conductivity a of the material being inspected. 
Variations in alloy or in temperature during testing can thus be adjusted with 
the capacitor. If the resistance variation is made with precision decade resistors, 
the thickness can then be indicated directly by the settings of the decade resist 
ance. The calibration of the indicating meter in terms of the percentage devia 
tion in sheet thickness remains constant, since all tests are made at the same 
frequency ratio f/f ff = 2.75 on the transmission-coefficient curve of Fig. 9, without 
regard to the individual values of conductivity or thickness of the test material. 

Direct Measurement of Resistance per Unit Square. In many measurements 
of conducting films, resistance is measured per unit square. The resistance per 
unit square, R n , is defined as the electrical resistance between two opposite edges 
of a square of conducting film, and is given in units of ohms per unit square. (The 
dimensions of the square are of no significance in this measurement.) Eq., (10), 
applied to the resistance per unit square, indicates that the limit frequency is 

Iff = 2.5 X 10 7 (R n /A) (12a) 

and the frequency ratio is 

tit = ^ ^ 10~ 8 (iA/R ) (12b) 

where A = distance between primary and secondaiy coil. 
R n = resistance per unit square of conducting film. 



Control of Film Resistance During Metal Deposition, To illustrate an 
application of the fork-coil eddy current test of resistance per unit square, con 
sider the problem of continuous observation of the vacuum vapor-deposition 
process of making condenser foils. In this case the resistance per unit square, 
B D , of the metallized paper is of interest. Fig. 13 shows the schematic circuit 
diagram of a measuring instrument designed for direct, continuous indication of 
the resistance per unit square during the process of metal deposition in vacuum. 




Institnt Dr, Foerster 

Fig. 13. Differential arrangement for the measurement of resistance per unit 
square for thin metallic foils, 

A fixed-frequency oscillator energizes the primary coil P, which acts on the two 
balanced secondary coils Si and > so that their induced voltages cancel. With no 
metallic layer between P and S 1; the indicating meter shows no deflection. The 
oscillator frequency is selected so that the test will be carried out at very low 
i/jg ratios (f/f g < 0.4) in the range of interest, In this range the vector W of 
Fig. 9 is proportional to l/JR n , as shown by the approximation formula for small 
j/j g ratios. As a result of this simple relationship, the indicating meter of the 
resistance-per-unit-square test instrument in Fig. 13 can be calibrated without a 

Instruments developed for production control of rolling processes and for con 
trol of metal deposition processes for foils or condenser paper in industry are 
described in the section on Eddy Current Test Equipment. 


Reference numbers in this section refer to the references at the end of the 
section on Eddy Current Test Indications. ' 





Impedance-Magnitude Tests 

Classification of eddy current .test equipment 1 

Types of test coils 1 

Test coil circuit arrangements 1 

Impedance- plane response characteristics .. 4 
Type of presentation of test indications ... 4 
Identification of eddy current test instru 
ments 4 

Classification of eddy current instruments 
for measuring material properties and 

discontinuities (/. 1) 2 

Principle of operation of impedance- magni 
tude tests 5 

Simplified diagram of impedance-magni 
tude test instrument, showing primary 
and secondary coil connections, with 
comparison standard test specimen (/. 2) 5 
Simplified diagram of impedance-magni 
tude teat instrument, with bridge cir 
cuit and two primary coils, with com 
parison standard test specimen (/. 3) .. 6 

Test indications 5 

20-Mc. tungsten wire test 6 

Balancing of test bridges 6 

Circuit for balancing an eddy current test 
bridge in both <aL and R directions 

(/. 4) I 

Limitations of impedance-magnitude tests. 7 

Reactance Magnitude Tests 

Principle of operation 7 

Frequency of oscillations 7 

Frequency variations resulting from conduc 
tivity changes in nonferromagnetic test 

cylinders ; 8 

Sample calculation of frequency variation 
resulting from 1 percent conductivity 

change 9 

Relative frequency variation, caused by 
1 percent variation in electrical conduc 
tivity of a nonferromagnetic test cylin 
der, as a function of frequency ratio 
1lf g for four coil fill factors, 13 = 1.0, 

0.75, 0.5, and 0.25 (/. 5) 9 

Frequency variations resulting from diameter 
changes in nonferromagnetic test cylin 
ders 10 

Relative frequency variation, A/// caused 
by a 1 percent variation in test -object 
diameter, as a function of frequency 
ratio ///, (/. 6) 10 


Frequency variations resulting from cracks in 

nonferromagnetic test cylinders H 

Frequency variations, caused by surface 
cracks of depth equal to 10 percent of 
diameter of nonferromagnetic cylinders 
placed hi test coils of oscillator circuits, 
as a function of frequency ratio ///? 

(/. 7) 11 

Frequency variations, caused by surface 
cracks of depth equal to 20 percent of 
diameter of nonferromagnetic cylinders 
placed in test coils of oscillator circuits, 
as a function of frequency ratio fffy 

(/. 8) I 2 

Frequency variations, caused by surface 
cracks of depth equal to 30 percent of 
diameter of nonferromagnetic cylinders 
placed in test coils of oscillator circuits, 
as a function of frequency ratio Ufa 

(/, 9) 13 

Test frequency selection H 

Frequency selection for temperature com 
pensation 12 

Calculating frequency ratios for temperature 

compensation 13 

Radiofrequency crack detector 14 

Limitations of reactance magnitude tests .... 14 

Feedback-controlled Impedance Tests 

Principle of operation 

Circuit analysis 

Basic oscillator circuit used in feedback- 
controlled impedance, eddy current test 

method (/. 10) 

Amplitude of oscillations 

Condition for no test indication 

Direction on impedance plane for no test 


Impedance plane for feedback- controlled 

impedance test method (/. 11) 

Direction on impedance plane for maximum 

test indications ' 

Response characteristics of the cyclograph .. 

Cyclograph-response impedance plane 

Impedance plane for feedback- controlled 
impedance tests showing direction of 
conductivity <r, diameter d, crack effects 
(black areas), and test measurement 

directions C (/. 12) 

Response characteristics of the Sedac instru 



CONTENTS (Continued) 


It espouse characteristics of the Zijlstra in 
strument 19 

-Response characteristics of the Cornelius in 
strument 20 

Suppression of undesired test effects 20 

Cathode-Ray Tube Vector Point Tests 

Principle of operation 20 

Electric circuit 20 

Wiring diagram of Foerster Multitest in 
strument with automatic sorting device 
(American designation: Magnatest FS- 

200) (/. 13) 21 

Calibration 21 

Test indications 22 

Impedance plane displays of the vector- 
point test instrument (Foerster Multi- 
test) (/. 14) 22 

Sorting 22 

Applications 23 

Cathode-Ray Tube Ellipse Tests 

Principle of operation 23 

Wiring diagram for Foerster ellipse 

method test instrument (/. 15) 24 

Phase rotation 23 

Impedance- plane analysis of ellipse patterns 24 
Impedance plane showing diameter direc 
tion EOC and crack direction 0.4 (/. 16) 25 
Analysis of impedance- plane measurement 
effects from elliptical screen patterns 

(/. 17) 25 

Separation of crack and diameter effects ... 26 

Sigmaflux eddy current test instrument 26 

Meter indications 26 

Sigmaflux test equipment 27 

Ellipse test instruments (/. 18) 27 

Test coils for ellipse test instruments (/. 

19) 28 

Applications of the ellipse test method 28 

Linear Time-Base Tests 

Principle of operation 29 

Circuit analysis 29 

Wiring diagram of linear time-base test 

instrument (/. 20) 29 

Calibration 30 

Slit analysis technique 30 

Screen image of linear time-base test 
instrument, with sinusoidal signals (/. 

21) 31 

Section of complex voltage plane showing 
vector voltages OA and OB of Fig. 21 

(/. 22) 32 

Impedance-plane analysis of slit values .... 32 
Complex voltage plane of test coil, con 
taining a ferromagnetic test cylinder, for 
a fixed test frequency of 5 c.p.s. (/. 23) 33 

Sample calculations for iron rod 32 

Crack depth analysis 35 

Analysis of Magnatest Q screen patterns 
and impedance plane characteristics (/. 
24) 34 


Self-comparison crack depth tests 35 

Response of difference- coil method to 
surface cracks in cylindrical test ob 
jects (/. 25) 36 

Combined absolute and comparison testy .. 36 

Secondary coil voltage signals 30 

Selection of test frequency 37 

Applications 38 

Test coils for Foerster Magnatest Q in 
struments (American designation, Mag 
natest FS-300) (/. 26) 37 

Suppression of Undesired Effects 

Significance of undesired effects in test indi 
cations 38 

Phase-controlled rectifier method 38 

Basic circuitry of the phase- controlled 

rectifier (/. 27) 38 

In-phase operation 39 

Analysis of operation of phase-controlled 

rectifier (/. 28) 40 

Out -of -phase operation 39 

Operation at intermediate phase angles .... 39 

Suppression of undesired signals 39 

Suppression of undesired harmonics 41 

Analysis of harmonics in signal voltage 

(/. 29) 41 

Condition for suppression of wth harmonic 42 
Resonant circuit impedance variation test .... 42 
Circuit for complete suppression of unde 
sired effects such as "lift-off" effect by 
partial compensation of voltage across 
a condenser in the resonance circuit 

(/. 30) 42 

Vector relations in the circuit of Fig. 30 

(/. 31) 43 

Circle diagram 43 

Suppression of lift-off effects 44 

Impedance plane showing "lift-off" direc 
tion ABC, with conductivity <r and 

crack directions (/. 32) 44 

Crack detection 45 

Applications of suppression techniques 46 

Crack-depth teats 47 

Defectometer probe-coil instrument for 
detection of cracks and crack- depth 
measurement on both ferromagnetic and 

nonferromagnetic materials (/. 33) 45 

Deflection on the Defectometer instru 
ment of Fig. 33 above the cross -section 
of a sound (left) and a cruckecl piston 

head (right) (/, 34) 45 

Porosity test 47 

Defectometer indication of porosity in the 
cross-section of a light, metal block (/. 

35) 46 

Cracks within drilled holes 47 

"Crack detector" for light -alloy com 
pressor blades (/. 36) 47 

Tests for thickness of nonmetallic layers .. 47 
Eddy current instrument for the meas 
urement of -the thickness of insulating 

layers (Foerster Isometer) (/. 37) 48 

References 48 




Impedance-Magnitude Tests 


theory developed and presented in the preceding sections on eddy current tests 
provides data on the optimum test conditions, test frequency, variations in meas 
ured quantities, suppression of undesired influences, and related parameters for 
various test problems in the field of eddy current testing. However, many eddy 
current test instruments were developed on an empirical basis before this theory 
was available. In consequence their designs sometimes deviate considerably from 
the optimum indicated by theory for specific test problems. The many different 
design variations are indicated in the classifications of eddy current instrument 
characteristics which are discussed in the subsequent text. 

Types of Test Coils. The most common types of test coils used in eddy 
current test equipment are: 

1. Feed- through coils, in which test objects such as rods, tubes, extruded or rolled 
shapes, wire, and other production parts are passed through a hole within the 
test coil. 

2. Inside test coils, which are inserted within cavities such as the interior of 
hollow tubes or drilled holes. 

3. Probe coils, which are placed on the surfaces of test objects. 

4. Forked coils, through whose arms test objects such as sheets, foils, metallized 
paper, and similar conducting materials pass during tests. 

Test Coil Circuit Arrangements. Test coils are connected to the electronic 
test instruments in various circuit arrangements which produce varying measure 
ment effects, as follows: 

1. Resonance-circuit methods, in which the test coil is part of a resonant circuit 
within the test instrument. 

2. Time-differentiated resonance methods, in which a resonance coil is connected 
to a circuit with time differentiation to suppress slowly changing variations in 
the test objects. 

3. Differential arrangements with a comparison coil (called the "absolute coil" 
method in European literature). In these the measurement voltage from the 
coil containing the unknown test object is compensated by that from a second 
coil containing a standard test object. 

4. Time-differentiated comparison arrangements, utilizing comparison coils, but 
with time differentiation in the indicating circuit to suppress slowly changing 
variations in the test objects. 

5. Self-comparison differential coil methods, in which one portion of the test 
object is compared with another portion of the same test object. 

6. Time-differentiated self -comparison methods, in which the output of a differen 
tial coil system is subjected to time differentiation in the indication circuit so 




^ Si 

4J B 

xw'd'g o 


s h g a 

- u 



60 -S -tf 

f I '-E 

I > 5 1 




S S 

ed ed 
O O 

P-l r-l O 

QJ OJ "tj 

a a d 







































bfl bo M M 

T3 Id T3 T3 

'S 'fi 'C 

-a -a -a -a 
s I g g 

^ ^ ja ^ 



















* S| 

E "58 

S ^ 

.s .sis 

3 I 


I I 

ft! U 

test FS- 











09 ft 







rting and 

i'S w 

epth m 

C3 o> F 
"? r; "^3 



** 2 


" Ji 

s *3 












fa o 
^ S 





| ft 
fa * 







S s 






i ft 

s s 

03 ft 

& a 
a s 

03 ft 

fa r- 


^ u 



BQ P -rj 









S fa 


03 , 







that only effects with a rapid time variation of the impedance difference are 

Impedance-Plane Response Characteristics. Discriminating circuits used in 
eddy current test instruments vary with respect to those characteristics of im 
pedance-plane changes to which they respond. Various methods, include: 

1. Impedance magnitude methods, which indicate only the magnitude of varia 
tions in the total impedance Z, regardless of the phase angle or direction in 
which it occurs on the impedance plane. (See section on Eddy Current Cylinder 
Tests for explanation of the characteristics of impedance planes for eddy 
current tests.) 

2. Inductive-reactance magnitude methods, which indicate only the magnitudes of 
changes in inductive reactance, coL, without regard to changes in the resistive 
component of the total impedance. In many of these methods, changes in 
reactance are detected by changes in the frequency of tuned circuits containing 
the test coils. 

3. Feedback-controlled impedance methods, which are also sometimes designated 
as core-loss or loss-sensing methods. In these methods the phase angle or ratio 
L/R of the inductance to the resistance of the test coil circuit influences the 
feedback factors of self-excited oscillator circuits. 

4. Impedance vector analysis methods, which indicate both the magnitude and 
phase angle (or direction on the impedance plane) of impedance variations. 

5. Methods which suppress the indications of undesired or insignificant test object 

Type of Presentation of Test Indications. Various methods are used to 
present eddy current test indications to the observer or to automatic selection or 
control devices. Among these are: 

1. Direct meter indication with pointer-type instruments. 

2. Meter indication with phase-controlled rectifier. 

3. Cathode-ray tube indicators, showing a generalized representation of the test 

4. Cathode-ray tube point method, in which the apparent impedance plane is 
represented by the screen of the cathode-ray tube, and measured values appear 
as a luminous point on the screen. 

5. Cathode-ray tube ellipse method, in which the measurement effect reacts on 
the vertical deflection plates of a cathode-ray tube, and a reference voltage, 
whose direction on the impedance plane can be selected at will, is applied to the 
horizontal deflection plates. 

6. Linear time-base method, in which the measurement effect influences the 
vertical deflection of the cathode-ray beam. Here a linear saw-tooth timing 
voltage, whose phase position can be selected at will, is applied to the horizontal 
deflection plates. 

7. Time-slit selection method, in which a specific instant or phase-angle point is 
selected in the sinusoidal signal or control voltage, and measured values at that 
time instant only are indicated or evaluated. 

Identification of Eddy Current Test Instruments. The preceding classifica 
tions can be used to explain the basic principles of operation of eddy current- 
test instruments, as shown in Fig. 1. Commercial designations are included 
to aid in identification of these particular instrument types. The following discus 
sion of eddy current test equipment will be devoted to those types whose perform 
ance can be described in terms of their impedance plane characteristics, since these 
characteristics are intimately related to the nature and operation of the tests. Only 
those methods which are widely used in industry or show promise of increased 
future acceptance are discussed in detail. 



Principle of Operation of Impedance-Magnitude Tests. Eddy current test 
instruments that indicate only the magnitude of variations in the total impedance 
of the test coil were widely used in the past. Their principle of operation is indi 
cated in the diagram of Fig. 2. An alternating current (a.c.) flows through two 
primary coils. One coil contains the unknown part to be tested, while the other 
contains a comparison or reference standard specimen. If both test specimens are 
identical, no voltage appears at the terminals of the secondary coils, since these 
are connected in series opposition and therefore their voltages cancel. A difference 
voltage appears across the secondary terminals if the unknown test object deviates 
in any of its properties from those of the reference standard. The test response 
may be caused by variations in test-object conductivity, magnetic permeability 
and dimensions, or by the presence of cracks or other discontinuities. The bridge 
circuit arrangement of Fig. 3 can also be used for measuring the magnitude of 
impedance variations, in place of the arrangement shown in Fig. 2. 





Institut Dr. Foerster 

Fig. 2. Simplified diagram of impedance-magnitude test instrument, showing 
primary and secondary coil connections, with comparison standard test specimen. 

Test Indications. The various methods for measuring the magnitude of im 
pedance variations are also distinguished by the type of indication of the difference 
voltages. Schirp, 78 ' 79 Mathes, 89 and others indicate the difference voltage by 
means of a meter. Brown and Bridle 8 display the variations of the two secondary 
coils by the deflection of the beam of a cathode-ray tube. In the "Magnetic 
Sorting Bridge" of the Salford Electrical Instrument Company, 2 a horizontal line 
appears on the screen of a cathode-ray tube when the unknown test object and 
the standard are the same. If the test specimen deviates from the reference 
standard specimen, the cathode-ray beam is also deflected vertically by a 
distance which corresponds to the degree of variation between the test specimen 
and the standard. The deflection is proportional to the magnitude of the variation 
between the sample and the standard. However, it is not possible to identify a 
given factor in the screen display with an individual property of the test material. 
The reference voltage applied to the horizontal deflection plates is fixed in 
apparent impedance direction, which makes an impedance analysis impossible. 



20-Mc. Tungsten Wire Test. A bridge similar to that shown in Fig. 3 was 
used by O'Dell 7S for the indication of splitting in tungsten wires. However, its 
test frequency was 20 megacycles (Me.) per second. Thus, in practice, only 
surface cracks could be found. In addition, at this high test frequency, the sen 
sitivity for crack indications is very low compared to the sensitivity of the 
indication to diameter variations. For tungsten wire of 2-mm. thickness, the limit 
frequency f ff = 7000 c.p.s. [see Eq. (5) in the section on Eddy Current Test Prin 
ciples for the definition of the limit frequency for solid cylinders]. A test fre 
quency of 20 Me. is 2860 times this limit frequency. Here the crack sensitivity is 
only 4 percent of that for a test frequency which is six times the limit frequency. 





Institut Dr. Foerster 

Fig. 3. Simplified diagram of impedance-magnitude test instrument, with bridge 
circuit and two primary coils, with comparison standard test specimen. 

Model tests and the similarity law for eddy current tests indicate that at a fre 
quency of 20 Me., a surface crack of 10 percent depth in 2-mm. tungsten wire is 
indicated with the same sensitivity as a diameter variation of only 0.15 percent. 
However, since tungsten wire is usually swaged, which results in diameter varia 
tions of 2 to 3 percent, there is considerable difficulty in separating dangerous 
cracks from harmless diameter variations. 

Balancing of Test Bridges. Various eddy current bridge test methods are 
distinguished not only by differences in the presentation of indications but also 
in the methods used for bridge balancing. It is impossible to wind test coils and 
compensation coils so exactly alike that no residual difference voltage appears. 
Thus it is necessary to balance residual bridge voltages by means of additional 
circuitry. Balancing can be entirely eliminated in simple bridges such as that 
developed by Brown and Bridle, 8 for example, and in the "Ferrometer" and 
"Magnetrie" bridges of French construction. Other bridges can be balanced for 
variations in the reactance direction by means of a resistor. The bridge used by 
Schirp and Mathes 78 permits balancing of both the reactance and 'the resistive 
components. Fig. 4 shows a schematic diagram of this bridge. The reactance con 
trol balances variations in the coL direction, while the resistance control balances 
variations in the R direction. 



Limitations of Impedance-Magnitude Tests. All impedance-magnitude eddy 
current tests similarly indicate desired effects, such as conductivity variations or 
crack effects, and undesired effects, such as minor diameter variations. These 
allowable tolerances in diameters of semi-finished products are the reason why 
methods which indicate only the magnitude of apparent impedance variations 
are often unsuccessful in practice. The dimensional tolerances far outweigh, in 






Institut Dr. Foerster 

Fig. 4. Circuit for balancing an eddy current test bridge in both o>L and R 


their eddy current response, the effects of variations in electrical conductivity 
which are often used in sorting alloys. For example, at a frequency ratio of 
f/f ff = 100, the effect of a 1 percent diameter variation is 20 times as strong as the 
effect of a 1 percent conductivity variation. However, a 1 percent diameter 
variation is well within commercial tolerance and is insignificant, whereas a 20 
percent variation in conductivity can be highly significant in some instances. 

Reactance Magnitude Tests 

PRINCIPLE OF OPERATION. Another series of eddy current test 
methods utilizes variations of the fundamental frequency of an oscillatory elec 
trical circuit whose inductance is determined in part or entirely by the test coil. 
The fundamental frequency of an oscillatory circuit is a function of its self- 
inductance but (to a first approximation) is not significantly influenced by its 
resistance JR. Consequently all physical effects of the test object which cause a 
change in the reactance of the test coil can be indicated. (See section on Eddy 
Current Test Principles for an explanation of inductance variations resulting irom 
test-object properties.) 

Frequency of Oscillations. Methods in which oscillation frequency is influ 
enced by the test coil are widely used in eddy current test devices. The quantita 
tive theory of this method indicates, for any given test material or test frequency, 


the change in oscillator frequency resulting from 1 percent variations in con 
ductivity or diameter or from a crack of given depth. For an oscillator circuit 
consisting of a self-inductance L (henries) and a capacitance (farads), the 
fundamental frequency / (c.p.s.) is 

The self -inductance L of the test coil has been derived [see Eq. (15) in the sec 
tion on Eddy Current Test Principles] as 

L = Lo [1 T] + T|plrfll.|Apff.(ral)] (2a) 

where L = self-inductance of the test coil in the absence of a test object. 

T) = coil fill factor. 

Hoi. = relative magnetic permeability of test material. 
tbff.creai) = real component of the complex effective permeability. 

The effective permeability pieff. is a function of the frequency ratio 

_ frivol, d 2 , 2M 

f/fff - 5066 (2h) 

where j g = limit frequency. 

/ = test frequency, c.pjs. 

a = electrical conductivity of test material, meter/oh m-mm. a . 

d = diameter of test cylinder, cm. 

(See Effective Permeability in the section on Eddy Current Test Principles for 
an explanation and quantitative values for these parameters.) 


The relative frequency variation, A///, can be determined for given variations in 
test-object conductivity, relative magnetic permeability, or dimensions, as well as 
from the presence of specific surface or subsurface cracks. For eddy current tests 
whose oscillator frequency is influenced by the test coil, only the real component 
of the effective permeability, [x eff . (real) , need be considered, since only this com 
ponent influences the frequency. In this case the frequency variation caused by 
a 1 percent variation in the electrical conductivity of the test material amounts to 

A,,, = (-V200) [1 e 


where A/// = relative frequency variation of the oscillatory circuit whose self- 
inductance is produced by the test coil when the electrical con 
ductivity of the test object in the coil varies by 1 percent. 

TI =fill factor of coil = d 2 /D 2 (for a solid cylinder). 

d = diameter of test cylinder. 

D = diameter of test coil. 

h = Bessel function of first order (Argument A = 


Jo = Bessel function of zero order (Argument A = ^1!^ 

\ 5066 

M-eff. = effective permeability. 

(Values for the effective permeability are shown in Figs. 7 and 8 of the section on 
Eddy Current Test Principles for specific frequency ratios /// r ) Note that the 
effective permeability |i eff> is a complex number, so that 


f f.(i ma|t .> 



Values of the Bessel functions J : and J Q for given frequency ratios are found from 
Bessel function tables. 

Sample Calculation of Frequency Variation Resulting from 1 Percent 
Conductivity Change. Fig. 5 shows an evaluation of the frequency variations of 
Eq. (3) as a function of frequency ratio f/f ff for four different coil fill factors v\. 


Fig. 5. Relative frequency variation, caused by 1 percent variation in electrical 

conductivity of a nonferromagnetic test cylinder, as a function of frequency ratio 

f/fg for four coil fill factors, TJ = 1.0, 0.75, 0.5, and 0.25. Frequency variation given 

in percentage of test frequency. 

These frequency variations correspond to variations of 1 percent in test material 
conductivity for the case of solid nonferromagnetic cylinders. To illustrate the use 
of these curves, suppose, for example, that an aluminum rod has the following 
characteristics : 

Diameter, d = 1 cm. 

Electrical conductivity, a = 35 meter/ ohm-mm. 2 . 

Relative magnetic permeability, p-rei. = 1. 

Assume that this rod is inserted into the test coil and that the frequency of the 
oscillator circuit is adjusted to / = 2860 c.p.s. by varying the tuning capacitor. 
The frequency ratio f/f ff now equals 20 (approximately), from Eq. (2b). If the 
fill factor r| = d 2 /D 2 = 0.75 for the specific test coil in use, point A in Fig. 5 is 
selected. The corresponding ordinate is about 0.12 X 10~-. This value of A///, 
multiplied by the test frequency / = 2860 c.p.s., corresponds to a frequency varia 
tion of 3.44 c.p.s. This frequency variation is the consequence of a change of 
1 percent in the electrical conductivity of the aluminum rod, which could possibly 
result from a temperature variation of 2.5 C. 




quency variations caused by a 1 percent change in the diameter of a nonferro- 
magnetic test cylinder placed in the test coil of an oscillator circuit are given by 

The symbols are explained below Eq. (3). The terms 
depend upon the specific frequency ratio f/f ff used in the test. 

S ) and Aureal) 

1000 fc 

Fig. 6. Relative frequency variation, A///, caused by 1 percent variation in test- 
object diameter, as a function of frequency ratio ///,. Curves are shown for four 
, . different coil fill factors t\ t corresponding to Fig. 5. 

Fig. 6 shows frequency variations computed from Eq. (4) as a function of 
frequency ratio f/f ff and coil fill factor T). At a frequency ratio of f/f g = 10 and 
a coil fill factor TJ = 0.75, for example, a diameter variation of 1 percent causes a 
frequency variation of 1 percent. At a frequency ratio of ///, = 100, with the 
same 1-cm. diam. aluminum rod whose properties were listed earlier, the fre 
quency variation caused by 1 percent diameter change is 20 times as great as that 
caused by a 1 percent change in conductivity. 



which form the core of test coils in oscillator circuits also influence the frequency 
variations in such test instruments. The variation in effective permeability, Au eff , 
caused by surface and subsurface cracks in nonferromagnetic test cylinders 
has been determined quantitatively. (See Figs. 14 to 17, inclusive, in section on 
Eddy Current Cylinder Tests.) Similar determinations have been made for out 
side surface, inside surface, and subsurface cracks in tubes. (See Figs. 10 to 
13 in section on Eddy Current Tube Tests.) Only the real component, Aji eff (rea i), 
of the variation in effective permeability due to cracks causes a frequency varia 
tion in oscillator test instruments. The quantitative relationship between the 
crack effect, A^ e ff.(reai), and the resultant variation in oscillator frequency, 
A///, is 

M/f = ~"(l/2)Heff.Creal) 

(1/TUlrel.) +M-eff.(real> 


Here the term [x e ff.(reai) is the real component of the effective permeability for 
the specific frequency ratio f/f g at which crack tests are made. (The values for 
H-eff.(reai) are given for various frequency ratios for solid test cylinders in Figs. 7 
and 8 of the section on Eddy Current Test Principles.) 

-j- for 10% depth of Crack 



/} -0,75 _ 

Fig. 7. Frequency variations, caused by surface cracks of depth equal to 10 percent 
of diameter of nonferromagnetic cylinders placed in test coils of oscillator circuits, 
as a function of frequency ratio f/f a . Curves are shown for coil fill factors r\ = 1.0, 

0.75, 0.50, and 0.25. 

Figs. 7, 8, and 9, based on Eq. (5), show the relative frequency variations, 
resulting from cracks of various depths in nonferromagnetic test cylinders, 
as a function of frequency ratio f/f g . As shown in Fig. 9. a crack having a depth 
equal to 30 percent of the cylinder diameter, tested as a frequency ratio f/f g 50, 
causes the same frequency variation as either a 2.4 percent change in cylinder 
diameter or a 36 percent change in material conductivity. 

TEST FREQUENCY SELECTION. Eq. (5) can be used not only for the 
case of solid metallic cylinders but also in the same manner to calculate the fre 
quency variations resulting from cracks in tubes and spheres and from surface 




for 20% depth of Crack 

Fig. 8. Frequency variations, caused by surface cracks of depth equal to 20 per 
cent of diameter of nonferromagnetic cylinders placed in test coils of oscillator 
circuits, as a function of frequency ratio ///. Curves are shown for coil fill factors 
T) = 1.0, 0.75, 0.50, and 055. 

cracks in metallic sheets. (Curves showing the variations in effective permeability 
for these cases are presented in the preceding sections on eddy current tests.) 
Figs. 5 through 9 permit calculation of the frequency variations resulting from 
various conductivity, dimensional, and crack effects for any test frequency and 
coil fill factors. Fig. 5 shows that the maximum test sensitivity for specific con 
ductivity effects, at which the largest frequency variations result, is attained at 
frequency ratios f/f ff between 4 and 10. Fig. 6 indicates that maximum sensitivity 
to diameter changes in nonferromagnetic cylinders increases steadily with in 
creasing frequency ratios f/f g . Figs. 5 and 6 are used to obtain an important 
result which can be applied to measurement of intergranular corrosion effects by 
an eddy current method (see the section on Eddy Current Test Indications), 

Frequency Selection for Temperature Compensation. The heating of a 
metal rod results in an increase in its diameter. As indicated in Fig. 6, this results 
in a decrease of oscillator frequency. However, at the same time, the higher 
temperature results in a decrease in electrical conductivity which, in turn, in 
creases the self-inductance. As indicated in Fig. 5, this results in an increase of the 
oscillator frequency. Now, for example, the temperature coefficient of expan 
sion of aluminum is approximately 200 times smaller than the temperature 
coefficient of electrical conductivity. Consequently one can select the frequency 
ratio f/f ff so high that the frequency response to dimensions becomes 200 times as 
great as the frequency response to conductivity changes. Thus, for any given 
temperature coefficients of expansion and conductivity, a frequency ratio is avail 
able at which the frequency influence of the conductivity variation and the 
diameter variation during heating or cooling compensate each other. At the fre- 


tor 30 % depth of Cmck 









tj -0,75 



Fig. 9. Frequency variations, caused by surface cracks of depth equal to 30 per 
cent of diameter of nonferromagnetic cylinders placed in test coils of oscillator 
circuits, as a function of frequency ratio ///,. Curves are shown for coil fill factors 

YI = 1.0, 0.75, 0.50, and 0.25. 

quency ratio where test response is independent of temperature, diameter varia 
tions as small as, for example, 10~ 5 , possibly caused by corrosion, can be 
measured with an extremely high sensitivity, independently of temperature 

Calculating Frequency Ratios for Temperature Compensation. For a given 
test cylinder diameter d and electrical conductivity a, it is possible to calculate 
the frequency ratio f/j g at which the frequency influence of the expansion is 
compensated by the frequency influence of electrical conductivity. For a non- 
ferromagnetic test cylinder with a temperature coefficient of electrical con 
ductivity, a and a temperature coefficient of expansion (3, this frequency ratio is. 

j, = (a/P) 2 /8 



The limit frequency j g is a function of test cylinder diameter and conductivity 
[see Eq. (2b)], so that this relation can be written as 

/i= S (a/P)a ' (6b) 

where /$ is the temperature-invariant test frequency at which the conductivity 
influence and the diameter influence due to temperature variations compensate 
one another. This frequency is selected, for example, to measure corrosion effects 
of external and intergranular corrosion (which reduces cylinder diameter) 
with extremely high sensitivity independent, of temperature variations. 

Crack Detector" (Salford Electrical Instrument Co.) operates at test frequencies 
between 50 kc. and 5 Me. In its operation, two transmitters feed a mixer tube 
which reproduces the difference frequency of the two transmitters: If the fre 
quencies of the transmitters are equal, the difference frequency disappears. As 
noted previously, all apparent impedance effects which have a component in the 
inductive reactance (coL) direction can- be- converted into, a frequency effect. If 
one of these effects, such as a variation in conductivity, permeability, or diameter, 
or the presence of cracks, causes the frequency of the transmitter containing the 
test object in its oscillator coil to vary, a difference frequency appears at the 
output of the mixer tube. After rectification this output appears as a meter 
indication. The mixer tube operates on a resonant circuit. This causes the instru 
ment deflection to be proportional to the difference frequency and thus to the 
variation of the reactance (coL) of the test coil. A similar arrangement is used by 
Plant and Manual 74 to detect defects in thin-walled stainless steel tubes. 

in effective permeability caused by surface cracks have a specific component in 
the reactance direction on the complex plane. (See, for example, Figs- 14 through 
18 in the section on Eddy Current Cylinder Tests.) Unfortunately, how- 
ever, variations in diameter of test cylinders also produce variations in the 
reactance direction. Since semi-finished industrial parts, as already noted, are 
subject to certain allowable diameter variations, it is impossible to distinguish 
between crack effects and diameter variations with reactance magnitude tests. At 
15 times the limit frequency, a crack of 5 percent depth causes the same reactance 
effect as a diameter variation of 1 percent (see Fig. 15 in the section on Eddy 
Current Cylinder Tests). However, for some applications; a crack of 5 percent 
depth is undoubtedly a serious defect. 

The diameter variations corresponding to 5 percent crack depth in their 
reactance variations, for nonferromagnetic cylinders, are an given in the accom 
panying table. 

Frequency Ratio, f/f g , of Test Diameter Variation Corresponding to 

Frequency to Limit Frequency 5 Percent Crack Depth (Percent) 

15 1 

50 0.47 

100 0.33 

300 0.22 

1000 0.112 

3000, 0.062 

Here the frequency ratio f/f g is computed from the physical properties of the 
test cylinder and the test frequency [see Eq. 2(b) in this section]. For example, 



at test _ frequencies of 1000 times the limit frequency, a diameter variation of 
approximately 0.1 percent causes the same variation in inductive reactance as a 
crack of 5 percent depth. Thus, frequency-response tests that operate at high 
multiples of the limit frequency are limited to detection of surface cracks with 
extremely accurately ground test objects and do not approach the optimum test 
sensitivity. The disadvantage of eddy current tests in which oscillator frequency 
is influenced by the test coil lies generally in the fact that it is impossible to 
separate desirable measurement effects (such as material conductivity or presence 
of cracks) from less significant effects such as trivial diameter variations which 
produce larger test indications. Of all effects present on the impedance plane, only 
the components in the inductive reactance direction are indicated. 

Feedback-controlled Impedance Tests 

PRINCIPLE OF OPERATION. Several eddy current test methods have 
a common feature in that the test coil forms the self-inductance of a feedback 
oscillator circuit. In these tests the ratio of the inductance to the resistance of 
the test coil influences the feedback factor of a self-excited oscillator circuit. Such 
tests are sometimes described as loss-sensing tests, since energy losses in the 
test object tend to damp out the oscillations. Commercial test instruments in this 
class include: 

1. The Cyclograph. 

2. The Sedac instrument. 

3. The Cornelius instrument. 

4. The Crack -Test instrument by Zijlstra. 

Since the Cyclograph is well known in the United States, the theory underlying 
its operation and that of other instruments in this group will be discussed in rela 
tion to impedance-plane effect indications. 

CIRCUIT ANALYSIS. Each of the instruments in the preceding list can be 
related to the basic circuit shown in Fig. 10. The test coil represents the self- 

Fig. 10. Basic oscillator circuit used in feedback-controlled impedance, eddy cur 
rent test method. 


inductance of the oscillator circuit and at the same time determines the feedback 
factor. The feedback voltage is fed back to the oscillator through the ohmic 
resistance R K . Oscillations are self-excited in the circuit arrangement of Fig. 10 
when the feedback requirement 

KV = 1 

is fulfilled. Here K is the feedback factor and V is the amplification factor of the 
oscillator tube circuit. From Fig. 10 the feedback factor K is given by 

K = A/B 

For example, if the amplification factor of the oscillator tube circuit is V = 100, 
undamped oscillations are excited as soon as K equals or exceeds 1/100. The self- 
excited oscillation always occurs at the natural frequency of the oscillating 
circuit, composed of the capacitance C, the resistance R. and the self -inductance 
L. Since the oscillating circuit always operates at the resonance point, it repre 
sents the equivalent of an ohmic resistance of magnitude L/CR. Thus the feed 
back factor K for the circuit of Fig. 10 is 

Since R K is normally much larger in magnitude than L/CR, the latter can be 
neglected to a first approximation in the denominator of Eq. (7a) . 

Amplitude of Oscillations. The amplitude A of the self-excited oscillations of 
the circuit of Fig. 10 increases in a certain range in proportion to the feedback 
factor K. Thus this amplitude can be approximated by 

4cc A'cc (1/R X C)(L/R) (7b) 

The factors C and R K are fixed for a specific test object. Therefore the amplitude 
of self -excited oscillations for the circuit of Fig. 10 is a function only of the 
ratio L/R. Consequently it is easily understood that all physical effects of test 
objects which cause a variation of impedance in which the ratio L/R remains 
constant cannot be indicated with feedback-controlled, impedance oscillator test 
methods of the type illustrated in Fig. 10. 

Condition for No Test Indication. The amplitude A of the self-excited 
oscillation remains constant, (A^/A = 0), for a variation of self -inductance AL 
and of resistance A.R if the following condition is fulfilled: 

In other words, if a variation AZ/ in the self-inductance L and a variation AJ? in 
the resistance R occur simultaneously and are such that the ratio (L + AZ/)/ 
(R + A#) after the impedance variation remains equal to the ratio L/R before 
the impedance variation, no change occurs in the amplitude of the self-excited 
oscillation. Consequently no indication of such an effect occurs in the feedback- 
controlled impedance test method. From Eq. (8) it follows that the condition for 
no change in indicated amplitude of oscillations also can be written as 

AL/AJ? = L/R for &A/A = (9) 

The test coil of the feedback-controlled impedance test method has two functions. 
The ratio L/R determines the amplitude of oscillations, whereas the factor L alone 
determines the frequency of the self-excited oscillation. If the self-inductance L 



increases by 1 percent, the frequency /, and thus the factor co = 2jt/, decrease by 
y 2 percent in accordance with Eq. (1). The result is an increase in the product 
coL, or inductive reactance, by y 2 percent. 

Direction on Impedance Plane for No Test Indication. The condition upon 
the components of the impedance variation (AcoL and A.R), for which the ampli 
tude A of the self-excited oscillations remains constant despite changes in test 
object properties, 

AcoL coL ,, AX 

is obtained from Eqs. (1) and (9). Eq. (10) indicates the direction of impedance 
variations on the complex impedance plane which have no influence upon the 
amplitude of the self-excited oscillation, and which therefore cannot be indicated 
with a feedback-controlled, impedance test method. 

tnv. Direction 

Fig. 11. Impedance plane for feedback-controlled impedance test method. The 

insensitive direction is parallel to line Pi. The sensitive direction is parallel to line 
SS. Test indications result only from impedance variations parallel to the sensitive 

direction SS. 

This impedance variation direction (corresponding to no change in oscillation 
amplitude) is given on the impedance plane of Fig. 11 by a straight line drawn 
from the point B (ojLj/2 on the ordinate axis) to the point PI (corresponding to 
the impedance values coL and R). An impedance variation AcoL and A# plotted 
in the direction of line BP^ fulfills the requirements of Eq. (10). Such an im 
pedance variation causes no change in the oscillation amplitude. Consequently 
such a change is not indicated by feedback-controlled impedance tests. 


Direction on Impedance Plane for Maximum Test Indications. On the 
other hand, impedance variations in the &S direction on the impedance plane of 
Fig. 11, normal to the line BP l} cause a maximum variation in oscillation ampli 
tude. The condition for this direction of maximum test sensitivity is obtained 
from Eq. (10), in accordance with the laws of analytic geometry, as 


The maximum sensitivity direction SS calculated from Eq. (11) represents the 
general direction of test measurements. In the case of impedance variations caused 
by physical effects such as variations in conductivity, permeability, or diameter or 
crack effects in the test object, the variation in any given direction is split into two 
components. The component parallel to the direction of line 5P 3 in Fig. 11 has 
no effect upon test indications. Only the component parallel to the sensitive direc 
tion SS causes a change in test indications. 


effect M measured by the test indications of the Cyclograph in response to a test 
coil impedance variation of 

is given by 

Mcydosraph bA/A ~ [ V(AcoL) a + (A#P ] cos (12) 

Here the angle <j> is the direction between the sensitive direction SS in Fig. 11 [as 
given by the condition of Eq. (11)] and the direction of the actual impedance 
variation AZ. The manner in which the Cyclograph indicates a variation in con 
ductivity, dimensional, or crack effects in a cylinder, tube, sphere, or sheet at any 
test frequency is given by Eq. (12) and the theoretical characteristics of the im 
pedance planes given previously. (See section on Eddy Current Cylinder Tests 
for detailed analysis of impedance-plane response to various test-object charac 

Cyclograph-Reponse Impedance Plane. Fig. 12 shows the impedance plane 
and the directions of conductivity, diameter, and crack effects for nonferromag- 
netic materials. The measurement direction C of the Cyclograph is plotted as 
solid lines intersecting the f/f ff curves, in accordance with the conditions of Eq. 
(11). Only test object effects which have an impedance component in the sensi 
tive measurement direction of the Cyclograph cause an indication, in accordance 
with Eq. (12). Fig. 12 shows that conductivity and diameter as well as crack 
effects cause an indication effect because all these effects have a larger or smaller 
component in the direction of Cyclograph sensitivity. Thus suppression of the 
influence of cylinder diameter is not possible for nonferrous test objects in the 
Cyclograph indications. On the other hand, Fig. 2, in the section on Eddy Current 
Cylinder Tests, shows that the direction of diameter changes for ferromagnetic 
materials can be made perpendicular to the sensitive direction of the Cyclograph 
by selecting a specific test frequency (a specific coL value). 


The Sedac instrument is designed to indicate the depth of seams and cracks. 
With this instrument, a point in the impedance plane can be reached by selecting 
a suitable test frequency at which the direction of the undesired "lift-off" effect 
is perpendicular to the sensitive S direction on the impedance plane given by 
Eq. (11). To do this, the test coil data (L, R) and the test frequency are selected 




Crack 30% 

f i -*\/ 

\Surface^\ | j Subsurface 

Institut Dr. Foerster 

Fig. 12. Impedance plane for feedback-controlled impedance tests showing direc 
tion of conductivity cr, diameter d, crack effects (black areas), and test measure 
ment directions C. 

so that Eq. (10) applies for the "lift-off" effect. Experience indicates that this 
condition is met for ferromagnetic materials in the vicinity of 10-kc. test fre 

MENT. Zijlstra 8S uses the feedback-controlled impedance test principle for the 
detection of cracks in metallic wires, particularly fused-in lead wires through 
glass envelopes. However, lead wires of tungsten and molybdenum in particu 
lar exhibit considerable thickness variations, since their thickness is reduced by 


means of swaging. It has previously been shown how cracks appear in the im 
pedance plane and how their effects compare with those of thickness variations 
(see preceding sections on eddy current tests). As theory indicates, thickness 
variations far outweigh crack effects in their influence upon the L/R ratio in feed 
back-controlled impedance tests. The separation of serious cracks from in 
significant variations in dimensions does not seem feasible in theory. Actually, as 
Zijlstra 88 points out at the end of his paper, this method can also be used to 
detect changing wire thicknesses. However, at high f/f ff ratios, a diameter increase 
reacts upon the L/R ratio exactly like a small surface crack, and thus it affects 
instrument indications. 

The Zijlstra instrument operates with a frequency of 5.6 Me. and is used for 
wires of 0.7 to 2.5-mm. diam. A test frequency of 5.6 Me. represents about 400 
times the limit frequency for a 1.5-mm. diain. tungsten wire. At this frequency 
the defect sensitivity is considerably lower than at 10 times the limit frequency. 

MENT. The Cornelius test instrument 10 (Wickman Co., England) also utilizes 
the feedback-controlled impedance method. The inventor indicates application 
fields for the instrument, including measurement of hardness, geometrical dimen 
sions, cracks, thickness of insulating layers, and electrical conductivity. All these 
effects influence the apparent impedance of the test coil, and thus the amplitude 
of the self-excited oscillations. However, separation of the various effects in a 
"one-dimensional" test indication is impossible in principle. Variations in several 
factors usually occur in practical testing. 

controlled impedance tests, it is possible to suppress certain undesired effects, 
such as dimensional variations in the test object or the lift-off effect of the test 
coil. To do this, the test frequency is selected so that the requirements of Eq. (10) 
are met for the small impedance variations caused by an undesired effect. How 
ever, the test frequency indicated by Eq. (10) for suppressing undesirable effects 
need not be identical with the optimum test frequency for the specific test prob 
lem. Impedance vector-analysis methods, described next, can usually be better 
adapted to test problems requiring suppression of such effects. 

Cathode-Ray Tube Vector Point Tests 

PRINCIPLE OF OPERATION. Variations in physical properties such as 
the conductivity, permeability, or discontinuities in test objects are represented 
by specific magnitudes and directions of changes in the apparent test coil im 
pedance. (See preceding sections on eddy current tests for details of these 
changes.) The Multitest instrument permits the impedance plane of the test coil 
to be represented quantitatively upon the screen of a cathode-ray tube. Im 
pedance values calculated from theory for given test-object conductivities and 
diameters appear as luminous points on this screen. 

Electric Circuit. Fig. 13 shows the block diagram of a Multitest instrument. 
An oscillator, 1, feeds two primary windings, PI and P 2 , of two test coils. Oppo 
sitely connected secondary coils, Si and $3, are connected to two compensators, 
3 and 4- Compensator 3 serves for compensation of imaginary voltages corre 
sponding to the inductive reactance, col/, direction on the impedance plane. Com 
pensator 4 compensates the real component of the voltage corresponding to the 
resistance, R, direction on the impedance plane. These compensators permit the 



operator to move the field of view on the screen of the cathode-ray tube (in 
the imaginary ooL and the real R directions) to any point on the complex im 
pedance or voltage planes. The result is analogous to moving the field of view of a 
microscope by operating the mechanical stage which carries the object under 

Compensators 3 and 4 obtain their voltages from the transformers 2\ and T 2 
through which the primary magnetizing coil current flows. The amplifier 2 
amplifies the compensated signal voltages from the test coils. Its function is 
analogous to the magnification adjustment of a microscope; at very high amplifi 
cations only a very small portion of the impedance plane is represented on the 
cathode-ray screen. 

Institut Dr, Foerster 

Fig. 13. Wiring diagram of Foerster Multitest instrument with automatic sorting 
device (American designation: Magnatest FS-200). 

Calibration, The amplifier # has ten sensitivity steps. The amplification 
doubles with each higher step, so that the area of the impedance plane shown on 
the screen is reduced to one-fourth. At the highest step (No. 10), the impedance 
plane area shown on the screen is 10~ of that shown at the lowest sensitivity 
step (No. 1). Only a very small portion of the impedance plane is shown at the 
highest step, but here a change of only 10~ 5 in conductivity or diameter effects is 
readable in the beam deflection. A calibration ring, calibrated in terms of per 
centage of "absolute value," indicates the percentage of the "absolute value 1 ' 
corresponding to 1-in. deflection. This permits quantitative measurements not 
dependent upon the selection of the test coil, and the comparison of test results 
from differing test conditions. 



Test Indications. The output of amplifier 2 is applied to the horizontal and 
vertical deflection amplifiers 7 and 8 of the oscilloscope. Amplifier 8 for horizontal 
deflection is designed to amplify only signal components corresponding to a 
specific, selected direction in the impedance plane. The vertical deflection am 
plifier 7 responds to signal components in the direction perpendicular on the 
impedance plane to the direction selected for the horizontal deflection. (The 
method is described in this section under Suppression of Undesired Effects.) 

Actually, the deflection amplifiers 7 and 5 contain phase-controlled rectifiers. 
The control voltage from a phase-shifter, 5, controls the response of the vertical 
deflection amplifier 7. Similarly the phase-controlled rectifier in the horizontal 
amplifier 8 obtains its control voltage from phase shifter 6. The control voltages 
from these two phase shifters are always at 90 deg. from each other. The phase 
angle of the phase shifters can be controlled arbitrarily without changing their 
90-deg. phase displacements from each other. 


Fig. 14. Impedance plane displays of the vector-point test instrument (Foerster 
Multitest). (a) Impedance plane showing circular field of view on cathode-ray 
screen, (b) Displacement of field of view with compensators to bring point P into 
field of view, (c) Rotation of point P to P' through the angle by moans of phase 
shifters 5 and 6 in Fig. 13. 

Fig. 14 (a) shows the impedance plane with the section represented on the 
cathode-ray screen within the circle. Let be the center of the screen. This point 
can be displaced in the vertical direction by adjusting the col/ compensator 3. 
It is displaced horizontally by means of the R compensator 4- For example, to 
bring the test point P onto the screen, the coL compensator is displaced by AcoZ> 
and the R compensator by Aft, as shown in Fig. 14(b). Adjustment of the phane 
shifters 5 and 6 moves point P on a circular path through an angle A< at constant 
radius from 0. The angle POP' is identical with the phase displacement of the 
phase shifters. In this way any desired direction on the impedance plane can be 
selected for representation along the horizontal axis of the oscilloscope. 

Sorting. The horizontal deflection voltage controls the automatic sorter 10 (see 
Fig. 13). Sorting gates are opened automatically whenever the horizontal deflec 
tion voltage exceeds selected magnitudes (a, 6, c, etc.). Since the horizontal 
deflection voltage direction can be selected at will to correspond with any desired 
direction in the impedance plane, it is possible to suppress undesirecl effects, such 
as diameter variations in sorting for alloy, or conductivity variations when sorting 
for dimension. To do this, the phase shifters S and 6 are displaced so that the 


direction of undesired effects is perpendicular to the direction of the horizontal- 
deflection amplifier signals, and such effects are then represented by vertical deflec 
tions only. Since the automatic sorter responds only to horizontal deflection 
signals, the undesired effects are suppressed in the sorting operation. 

In sorting tests, a normal and acceptable test object is placed in test coils PI 
and S-L (see Fig. 13). The resultant secondary coil output voltage is largely com 
pensated by a compensation specimen, S C o mp ., which is then placed in test coils 
P 2 and S 2 . The residual voltage from the differentially connected secondary coils 
is next compensated (by compensators 3 and 4) so that the beam rests at the 
center point of the cathode-ray screen, with the normal test object in P^. 

If, for example, the test objects are to be sorted for conductivity variations, 
the normal test object in coils P^ is removed and its diameter varied by filing or 
grinding (a decrease of 0.5 to 1 percent in diameter is sufficient). The specimen 
with reduced diameter is again inserted into test coil P^. The beam of 
the cathode-ray tube then moves away from the screen center as a result of the 
variation in impedance caused by the decrease in diameter. The position of the 
cathode-ray beam is now adjusted by means of phase shifters 5 and 6 until it rests 
on the vertical center line through 0. The undesired-dimension variation direction 
now lies perpendicular to the horizontal deflection used for sorting. Consequently 
the dimensional variations no longer influence the sorting operation. However, 
the position of the beam, as shown upon the screen, indicates all impedance varia 
tions, from whose magnitude and direction the causes can often be determined by 
impedance-plane analysis. (See preceding sections on eddy current tests.) 

APPLICATIONS. The Multitest method, which produces a point indication 
on the cathode-ray screen for each test object, is especially well suited for testing 
of production parts. Test objects such as spheres, rollers, pins, and other produc 
tion parts can fall freely through the test coil at very high speeds (four to five 
small parts per second) . The very high test sensitivity permits sorting of watch 
springs weighing only 10 ~ 3 grams (for ladies' wrist watches) according to the so- 
called thermoelastic coefficient. In some cases, parts must be tested at a specific 
point, as in hardness tests of the tips of drills or sewing machine needles. For 
these applications the Multitest instrument test coil is equipped with an elec 
tronically controlled stop which holds the test object in the desired position for 
about 1/100 sec. The sorting gate is then reset by the approach of the next test 
object. (Additional applications of Multitest methods are described in the sec 
tion on Eddy Current Test Indications.) 

Cathode-Ray Tube Ellipse Tests 

PRINCIPLE OF OPERATION. In many respects the equipment for the 
ellipse test method is similar to that used in the vector point method. As shown 
in Fig. 15, the test coils, the coL and R compensators, and the vertical amplifier 
correspond to the components of the Multitest instrument of Fig. 13. In the 
ellipse method, however, the amplified voltage from the secondary test coils acts 
directly upon the vertical deflection plates of the cathode-ray oscilloscope. The 
voltage applied to the horizontal deflection plates corresponds to the primary 
magnetizing current in the test coils and is taken from this current through 
transformer T 2 . 

Phase Rotation. The phase of the horizontal deflection voltage can be turned 
to any desired direction on the impedance plane by means of the phase shifter 5. 
The voltage output of the phase shifter 5 is amplified by the horizontal deflection 

40 24 


amplifier 7. The output of this amplifier is applied both to the horizontal plates 
and to the 90-deg. phase shifter 8. This 90-deg. phase shifter produces the control 
voltage for the phase-controlled rectifier 9. The output of rectifier 9 indicates 
only the component of the test-coil signal voltage which is perpendicular on the 
impedance plane to the voltage direction of the signal applied to the horizontal 
deflection plates. The signal used for sorting or for marking of defective test 
objects is also influenced only by this component of the vertical voltage which is 
perpendicular to the horizontal deflection voltage. 


Institut Dr. Foerster 

Fig. 15. Wiring diagram for Foerster ellipse method test instrument. Wire Crack 
Test Instrument (FW-200) ; Rod Crack Test Instrument, (FW-400) ; Sigmaflux 

Instrument (FW-300). 

Impedance-Plane Analysis of Ellipse Patterns. Various physical effects 
shown by displacements on the impedance plane (such as variations in dimen 
sions, conductivity, permeability, or crack conditions) are indicated by the 
shapes of the elliptical patterns on the screen of the cathode-ray oscilloscope. 
Fig. 16 shows the impedance plane with the direction of diameter variations 
shown by the line EOC and the direction of crack effects assumed to be OA. If 
the voltage OC is applied to the horizontal deflection plates, a horizontal trace of 
length COC' (Fig. 17) appears on the screen as long as no voltage is applied to 
the vertical deflection plates. 

Suppose that the voltage OE (Fig. 16), corresponding to an increase in test- 
object diameter is now applied to the vertical deflection plates. In this case an 
inclined line such as W OE" (Fig. 17) appears in the cathode-ray tube display. 
A straight-line trace results because two voltages having the same phase displace 
ment which are applied simultaneously to the two sets of deflection plates of a 
cathode-ray oscilloscope always produce a straight-line trace. In this inclined line 
W OE", the horizontal half-amplitude OC corresponds to the magnitude of vector 
OC in Fig. 16. 

The vertical half-amplitude OE in Fig. 17 represents the magnitude of vector 
OE in Fig. 16. However, if a crack effect is present, as represented by vector OA 



Fig. 16. Impedance plane showing diameter direction EOC and crack direction OA. 

in Fig. 16, the voltages applied to horizontal and vertical deflection plates differ in 
phase. In this case the cathode-ray tube pattern is an ellipse, as traced in Fig. 17. 
The relationships given here are important in analysis of elliptical screen pat 
terns.- The vertical half-amplitude OA of the ellipse in Fig. 17 corresponds to the 
magnitude of the vector OA resulting from a crack effect in the impedance plane 
of Fig. 16. If the horizontal deflection reference voltage COC' is shorted out by 
means of a short-circuit button, the distance OA appears as a vertical trace on 

Fig. 17. Analysis of impedance-plane measurement effects from elliptical screen 



the cathode-ray oscilloscope (CRO) screen. The labels on the elliptical display of 
Fig. 17 are defined as follows: 

00 = b = reference voltage in the diameter direction, 

OA =c = signal corresponding to a crack effect. 

= phase angle between direction of crack effect and direction of diameter varia 
tions (reference voltage) in Fig. 16. 

OB = component of crack signal perpendicular to the diameter direction in Fig. 16. 
OB = OA sin 0. 

OD ^component of reference voltage OC perpendicular to direction of crack effect 
in Fig. 16. OD = b sin 0. 

OE = increase in diameter of test object. 

Separation of Crack and Diameter Effects. The important feature of the 
elliptical pattern is the intersection of the ellipse with the vertical axis (direction 
OA). The vertical intercept (line OB) or opening of the ellipse is proportional 
to the crack component perpendicular to the diameter direction in the impedance 
plane of Fig. 16. Diameter effects, as previously noted, cannot cause an opening of 
the ellipse. If diameter ar^l crack effects appear simultaneously, as is often the 
case in practical testing, the opening of the ellipse always corresponds to the crack 
effect, independently of the diameter variations. 

The magnitudes of the components of crack effects perpendicular to the diameter 
direction on the impedance plane have been determined as a function of crack 
depth and frequency ratio f/f g . (For example, see Fig. 36, in the section on Eddy 
Current Cylinder Tests, for curves for the case of nonferromagnetic test cylinders.) 
These components of the crack effect perpendicular to the diameter direction are 
represented by the opening OB of the elliptical screen patterns. 

the electrical block diagram of the Sigmaflux instrument of the Institut Dr. 
Foerster (American designation: Magnatest FW-300). It is a typical example of 
the ellipse test method. An oscillator sends an a.-c. magnetizing current through 
the pair of primary coils PI and Po. The residual secondary voltage, with test 
objects in the coils, is compensated to zero by means of the complex compen 
sators 3 and 4, is amplified, and then is applied to the vertical deflection plates. 
These compensators are energized from the primary coil current through trans 
former TV A voltage, which also originates from the magnetizing primary coil 
current energizing transformer T 2 , is applied to the phase shifter. The output 
voltage of this phase shifter 5 is applied, after amplification, to the horizontal 
deflection plates and also to the input of the 90-deg. phase shifter 8. The 90-deg. 
phase shifter provides a control voltage to the phase-controlled rectifier 9. 

Meter Indications. If the phase shifter 5 in Fig. 15 is adjusted so that the 
horizontal deflection voltage corresponds to the diameter direction of the im 
pedance plane of Fig. 16, diameter variations appear as inclined straight lines on 
the cathode-ray tube display (as previously noted). The secondary coil voltage 
applied to the vertical deflection plates is also applied to the phase-controlled 
rectifier 9, which obtains its control voltage from phase shifter S. The indicating 
instrument or meter of the instrument shown in Fig. 15 reads zero in case of 
diameter variations such as are shown by a straight line on the tube screen. Only 
signal voltages which have a component in phase with the 90-deg. control signal 
from phase shifter 8 are amplified by the phase-controlled rectifier 9. Since this 
reference voltage is displaced 90 deg. in phase from diameter-direction voltages, 
diameter effects cannot be amplified by amplifier 9. 


The meter in Fig. 15 shows only components perpendicular to the diameter 
direction in the impedance plane of Fig. 16; that is, parallel to the vector OB The 
meter indication thus corresponds to the opening OB of the ellipse in Fio- 17 The 
signal used for sorting or marking defective parts is likewise influenced onlv bv 
detects and not by diameter effects. " " 

Sigmaflux Test Equipment. Fig. 18 shows the Sigmaflux instrument widelv 
used in European industry for defect testing, conductivity measurements, sortm* 
of alloys, dimensional testing, and in fully automatic test units. Fig. 19 shows the 
wide range of test coil types, of varying dimensions, used with these ellipse test 

Institut Dr. Foerster 

Fig. 18. Ellipse test instruments, (a) Sigmaflux eddy current test instrument 

(American designation : Magnatest FW-300) . (b) Foerster wire crack-test instrument 

(American designation: Magnatest, FW-200). 



instruments in testing of semi-finished parts in industry. The upper row of coils 
in Fig. 19 shows complete coil assemblies. The center row shows coil inserts for 
the self-comparison difference coil method in which one section of the test object 
is compared electrically with another section of the same test object. The lower 
row of coils shows the inserts for tests with a separate comparison specimen, the 
so-called absolute method. 

Institut Dr. Foerster 
Fig. 19. Test coils for ellipse test instruments. 

The coil holders shown in the top row in Fig. 19 contain devices which con 
tinuously adjust the coil opening to match the diameter of the test parts. In addi 
tion these coil holders contain a device consisting of a light source, optics, and 
mlcrophotocell which assures that the electronic components will be inactive unless 
the light beams are interrupted at both ends of the coil by a test piece extending 
entirely through the coil. The coil inserts are of the plug-in variety which auto 
matically make electrical contact when inserted into the coil holders! 

test method is used with feed-through coils for the following applications : 

1. Alloy porting of semi-finished metallic parts (such as rods, tubes, and wires) by 
electrical conductivity measurements, as in sorting mixed lots. 

2. Hardness sorting of age-hardenable light-metal alloys. 

3. Control of the uniformity of hardening of semi-finished parts quenched imme 
diately after extruding. 

4. Testing of rods, tubes, and wires for surface and subsurface cracks. 

5. Measurement of the diameter of rods, tubes, and wires without direct contact 
and independent of test-object material. 



The diameter range of the ellipse test method extends from 0.41 to 100 mm 
Since the conductivity range of most test parts lies between the approximate 
limits of 1 and 60 meters per ohm-mm. 2 , the corresponding ran- e of limit fre 
quencies extends from /, = 1 to /, = 3 X 10. For testing of small or very short 
discontinuities in test parts, a time selection circuit can be used to indicate only 
rapid variations and to suppress slowly varying effects. Commercial forms of 
ellipse test instruments include a wire crack-test unit, a rod crack-test unit, and 
the Sigmaflux instrument (American designations are the Magnatest FW-200 400 
and 300, respectively) . 

Linear Time-Base Tests 

PRINCIPLE OF OPERATION. In the cathode-ray tube vector-point 
and ellipse test methods described previously, only the fundamental frequency of 
the a.-c. magnetization was used to produce test indications. This fundamental 
frequency appears alone in the output signal from the test coils in tests of mate 
rials whose relative magnetic permeability is a constant, independent of the field 
strength of the test coil. However, in ferromagnetic test-object materials, the 
relative magnetic permeability |i reL is more or less a function of the magnetizing 
field strength. (For example, see Fig. 19 in the section on Eddy Current Tube 
Tests.) The nonlinear magnetization curve of ferromagnetic materials causes odd 
harmonic frequencies, such as the third, fifth, and seventh harmonics of the 
fundamental frequency, to appear in the output voltage of the secondary test coils. 
Even harmonics appear only if a direct-current (d.-c.) field reacts upon the test 
object in addition to the alternating-current (a.-c.) field, or if the test object 
exhibits a preferential direction of magnetization as a consequence of a previous 
d.-c. magnetization. By contrast, the linear time-base eddy current test method is 
especially well suited for the analysis of mixtures of fundamental and harmonic 
frequencies in test signals. In this method the harmonic frequencies play an im 
portant role in the interpretation of the test indications. (Examples of applica 
tions of this test method are given in the section on Eddy Current Test Indica 

Circuit Analysis. Fig. 20 shows the block electrical diagram for a linear time- 
base eddy current test instrument (the Foerster Magnatest Q instrument; Arner- 


Institut Dr. Foevster 

Fig. 20. Wiring diagram of linear time-base test instrument. (Foerster Magnatest 
Q; American designation, Magnatest FS-300.) 


lean designation, Magnatest FS-300). The two primary coils, P l and P 2 , are 
energized from an adjustable current regulator having sinusoidal output fre 
quencies of 5, 60, or 400 c.p.s. The residual voltages from the differentially con 
nected secondary coils of the comparison coil system are compensated to zero for 
normal test specimens by means of the complex compensators 3 and 4- The 
operating voltages for these complex compensators are obtained from the mag 
netizing field-current regulator through transformer T lm 

The voltage which results from a variation in the conductivity, permeability, 
dimensions, or discontinuities of an unknown test object in test coils P^ is 
amplified in amplifier 2 and applied to the vertical deflection plates of a cathode- 
ray oscilloscope. Another voltage from the magnetizing field-current regulator is 
applied through transformer T 2 to the phase shifter 5. By adjusting this phase 
shifter, this voltage can be turned to any desired direction, in the impedance 
plane. The sinusoidal output voltage of the phase shifter is next converted into a 
saw-tooth wave form by the saw-tooth generator 6. This voltage, after amplifica 
tion in amplifier 7, is applied to the horizontal deflection plates of the oscilloscope 
to provide a linear, horizontal sweep signal. When a signal is applied from the 
secondary test coils to the vertical deflection plates, a sinusoidal curve (which 
can contain harmonics) appears on the screen. This curve can be displaced 
horizontally to any desired position by means of the phase shifter 5. The inter 
esting portion of the curve can be moved to the center of the screen, where a 
reading slit is located, for quantitative evaluation. 

Calibration. The signal amplifier 2 is provided with a calibration control which 
indicates, for any given sensitivity step, the percentage of the "absolute value 7 ' 
corresponding to 1-in. beam deflection. The absolute value is defined as the 
voltage effect produced by the test object in the absence of a compensating test 
object in the comparison test coil. This sensitivity indication is independent of 
both the selection of the test coil type and of the particular test object employed. 
This indication of the absolute value for each sensitivity step on the Magnatest Q 
instrument is especially significant in quantitative interpretation of test indica 

Test indications corresponding to the depth of case-hardened layers, mechanical 
hardness, carbon content, edge decarburization, or other factors can be given in 
terms of the percentage of absolute value. This percentage is independent of the 
specific instrument, test coil, and test object involved. Consequently test indica 
tions, obtained in different plants at various times on test objects of differing 
dimensions and in various sizes of test coils, can be compared with each other. 
(Examples of quantitative evaluation of test indications are presented in the sec 
tion on Eddy Current Test Indications.) 

Slit Analysis Technique. The switch 8 (Fig. 20) in the test coil signal circuit 
permits application to the vertical deflection plates of one of three signals : 

1. The true secondary coil voltage signal. 

2. The fundamental frequency only of the signal voltage. 

3. The odd harmonics of the signal voltage. 

The choice depends upon the nature of the testing problem. A choice also exists 
in selection of the test frequency between 5, 60, and 400 c.p.s. The 5-c.p.s. fre 
quency increases the applicability of the method considerably because of the 
greater penetration depth of eddy currents at this low frequency. (Applications 
of each frequency and signal selection are discussed in the section on Eddy Current 
Test Indications.) 



A slit is provided in the center of the cathode-ray tube screen for quantitative 
analysis of signal wave forms. The interesting portion of the signal wave form 
can be moved to the slit by means of the phase shifter. Fig. 21 shows the CRO 
screen with a sinusoidal signal of amplitude OA selected in phase with the voltage 
from the phase shifter (solid curve). Here the amplitude OA corresponds to the 
vector OA on the impedance plane of Fig. 22, corresponding to a measurement 
effect. The instantaneous value of this signal voltage at the vertical center slit is 
zero. Now, if a measurement effect perpendicular to the base direction OA appears 


Fig. 21. Screen image of linear time-base test instrument, with sinusoidal signals. 

on the impedance plane (for example, vector OB), the wave form displayed on 
the CRO screen corresponds to the dashed curve in Fig. 21. Its amplitude at the 
slit is OB and is designated as the slit value M. In general, 

M = A sin < 


where A is the amplitude of the measurement effect and < is the angle between 
the measurement effect and the base voltage on the impedance plane (Fig. 22). 
The linear horizontal sweep of the oscilloscope is synchronized with the base 
voltage or output of the phase shifter. The beam starts from the left side of the 
screen at the zero point (< = 0) of the base voltage and travels linearly to the 
right side during one cycle (< = 360), retraces rapidly, and repeats this sweep 
procedure. The slit value M therefore indicates the component of a given meas 
urement effect which is perpendicular to the direction of the base voltage on the 
impedance plane. Thus the slit value M is identical to the opening of the ellipse 
in the elliptical presentation (see Fig. 17). 




Fig. 22. Section of complex voltage plane showing vector voltages OA and OB 

of Fig. 21. 

Impedance-Plane Analysis of Slit Values. The complex voltage plane for a 
ferromagnetic test cylinder is shown in Fig. 23. This diagram differs from pre 
ceding impedance-plane illustrations where points along the curves were identified 
in terms of the frequency ratio ///<,, defined by 

B " 5066 

Here, the curves are labeled for the specific test frequency of 5 c.p.s, in terms of 
the product a|A re i.c? 2 , where 

a = electrical conductivity of test material in mctor/ohm-mm. 2 . 
Uroi. = relative magnetic permeability (usually between 10 and 500 for ferromagnetic 

materials used in industry) . 
d = diameter of cylindrical test object, cm, 

Fig, 23 shows curves for two values of relative permeability, [i pol = 50 and 

Hrel. = 100. 

The directions of variations in relative permeability and in diameter coincide 
(as indicated in Fig. 2 in section on Eddy Current Cylinder Tests) . Variations 
in conductivity occur along the heavy curves for various (ji reli values. The effects 
of surface cracks of 30 percent depth are shown by arrows at four points where 
the product a^d? assumes the values of 5,000, 15,000, 50,000, and 150,000. 

Sample Calculations for Iron Rod. Assume that an iron rod has the follow 
ing characteristics : 

Electrical conductivity, a = 10 meter/ohm-mm. 2 . 
Relative permeability, M, P M. = 100. 
Diameter, d = 2 cm. 

In this case, the product 

d a = 4000 



Assuming that the test cylinder fills the test coil, the coil fill factor TJ = 1. The 
test point is shown at C on the outer curve of Fig. 23. Here the normalized 
imaginary component of the test coil voltage # imag ./# = 77.4, the ordinate of 
point C. E is the secondary test coil voltage in the absence of a test object. Thus, 
inserting the ferromagnetic cylinder into the test coil increases the imaginary com 
ponent of its output voltage by 77.4 times the signal voltage available with the 
coil empty. The real component (abscissa) of point C is E Te&l /E = 34.5. 

Institut Dr. Foerster 

Fig. 23. Complex voltage plane of test coil, containing a ferromagnetic test 

cylinder, for a fixed test frequency of 5 c.p.s. Values of product (an.rei.rf 2 ) shown 

on curves. Coil fill factor 11 = 1 . 

40 34 



Institut Dr. Poerster 

Fig. 24. Analysis of Magnatest Q screen patterns and impedance plane charac 
teristics. (a) Enlargement of area around point A in complex voltage plane of Fig. 
23, showing directions of diameter and permeability variations (vector AS), crack 
effects (vector AD}, and conductivity effects (vector AC}, (b) Screen indications 
corresponding to vector effects of Fig. 24(a) for Foerster Magnatest Q instrument. 
(c) Enlarged portion of the impedance plane of Fig. 23, showing crack effects along 
line AD and direction of diameter and permeability variations. 


Suppose now that the phase direction AB (corresponding to variations of 
permeability or diameter) is selected for the base voltage to which the horizontal 
sweep of the instrument is synchronized. In this case all variations in diameter or 
magnetic permeability which appear in the signal voltage on the CRO screen 
have no influence upon the signal at the slit. Hence the wave form passes through 
zero at the slit, and the slit value M equals zero. 

If, however, a physical effect such as a variation in carbon content of the iron 
rod causes its electrical conductivity to increase from a = 10 to a = 12.5, the test 
point moves from point C (where the product au, re i.tf 2 = 4000) to point A (where 
the product an re i. d 2 = 5000). The direction of vector CA in Fig. 23 lies almost 
perpendicular to the direction of diameter and permeability changes. Thus the 
conductivity variation is indicated sensitively by the slit method, with relatively 
little interference from variations in diameter or permeability. The vector AD, 
corresponding to the effect of a crack of 30 percent depth, also has a large com 
ponent in the direction AE, perpendicular to the base voltage. Thus crack effects 
and conductivity variations are indicated similarly and cannot be separated 
effectively under this test condition. 

CRACK DEPTH ANALYSIS. In the complex voltage plane of Fig. 24(a), 
various physical effects for the test object (ferromagnetic cylinder), including 

1. The diameter and permeability effect (vector AB}, 

2. The conductivity effect (vector AC}, and 

3. The crack effect (vector AD} for a 30 percent depth of crack 

are replotted from Fig. 23. These vectors are transferred to the Magnatest Q 
screen pattern in Fig. 24 (b). An amplification factor of unity has been selected 
in the transfer from Fig. 24(a) to Fig. 24(b) so that the distance AB on the 
screen pattern is equal to the distance AB on the complex voltage plane. Fig. 
24 (c) shows an enlargement of a portion of the complex voltage plane for the case 
in which the product a|j, rel .rf 2 = 5000. Crack effects are shown along the line AD. 
Distance AB corresponds to a 10 percent crack depth, AC to a 20 percent crack 
depth, and so on, according to the crack depth scale plotted along line AD. The 
direction of diameter and permeability effects is also indicated. 

Self-Comparison Crack Depth Tests. In the Magnatest Q method described 
earlier, the voltage of the test coil containing the unknown test object is always 
compensated (approximately to zero) by means of the compensation coil con 
taining the reference-standard test object. However, the Magnatest D method 
utilizes the technique of self-comparison in which one portion of the test rod is 
electrically compared with another portion of the same test rod. [See Fig. 12 (f) 
in the section on Eddy Current Cylinder Tests for an illustration of this ar 
rangement.] Fig. 25(a) shows a test rod with a crack which increases in depth 
from to 30 percent of the rod diameter, passing through a pair of differentially 
connected secondary test coils, Si and >. If coils S x and S 2 are both over defect- 
free portions of the test rod, the net secondary voltage equals zero. However, if 
a crack of 10 percent depth [point B in Fig. 25(a)] moves into coil S l while the 
portion of the rod within coil S x is still free from defects, a signal voltage appears 
across the terminals of the secondary coils. This voltage corresponds to the vector 
AB in Fig. 24(c) . The corresponding test indication is shown as AB in Fig. 25(b) . 
As the rod progresses through the coils, the secondary coil S l will contain the 
point B (of 10 percent crack depth) when coil S 2 contains point C fof 20 percent 
crack depth). At this instant the difference voltage is given by the vector BC in 



Fig. 24(c) and indicated as BC in Fig. 25(b). As the rod proceeds until the 
uniform crack depth of 30 percent lies within both test coils, no difference voltage 
appears at the terminate of the test coils, and the indication returns to zero, as 
shown in Fig. 25 (b). 

Thus the difference test method indicates only the variations in crack depth. 
Therefore a, crack having a constant depth cannot be indicated. However, 
shallow cracks normally vary continuously in their depth; they usually begin 
and end several times over the entire length of a rod. Only in the case of stress 
cracks or of a seam-welded tube which has completely burst open along the seam 
will the defect progress through the entire length at n constant depth. 



Fig. 25. Response of difference-coil test method to surface cracks in cylindrical 
test objects, (a) Metallic rod with surface crack depth increasing from to 30 percent 
of diameter, (b) Test indication or "slit value A/" on calhodo-ray screen as crack 

passes test coils. 

Combined Absolute and Comparison Tests. The crack test method, using 
difference coils for self-comparison on the same test "specimen, can be combined 
with the ''absolute method," in which the unknown test rod is compared with a 
reference standard rod. With this combination, cracks of constant depth can 
also be indicated, since small quality variations in alloy, lot, heat treatment, and 
so on, from rod to rod are outweighed by the effects of deep cracks. However, for 
detection of fine cracks, it is necessary to use only the difference method (Magnn- 
test D), in which one point of the rod is compared electrically with another point 
of the same rod. This follows because variations in physical constants can be con 
siderably greater than the effects caused by fine cracks. 

Secondary Coil Voltage Signals. The magnitude of the coil voltage M< 
appearing at the slit in the cathode-ray screen in response to a crack, is 

M ^?oTi|Liroi.(A^ff.) sin 




where E - voltage of a single secondary coil in the absence of a test object. 
T] = coil fill factor = (d/D) 2 . 
d = diameter of test cylinder. 
D = diameter of test coil. 

Mrei. = relative magnetic permeability of test-object material. 
AM-eff. = variation in effective permeability caused bv the crack (see distance AD 

in Fig. 23). ~ 

= phase angle between direction of diameter and permeability variations, 
and direction of crack effects, on the complex voltage' plane [see 

In the difference test method, A[i eff . corresponds to the difference voltage of the 
two secondary coils resulting from the crack effects (see Fig. 25). 

Selection of Test Frequency. Fig. 23 shows that the angle between the direc 
tion of permeability and diameter effects and the direction of crack effects de 
creases as the product a|i re i.d 2 becomes larger. Thus the possibility of separating 
serious crack effects from insignificant dimensional and permeability effects is 
diminished at higher frequency ratios f/f fft Consequently, lower test frequencies 
would be preferred for this application. Fig. 23, of course, refers to the funda- 

Institut Dr. Foerster 

Coil 1 = 300-mm. diam. 

Coil 2 = 1-mm. diam. 

Coil 3 = square cross-section, 400 X 400mm. 

Coil 4 = rectangular cross-section, 300 X 100 mm. 

Coil 5 = rectangular cross-section for sheet metal, 6 X 1100 mm. 

Fig. 26. Test coils for Foerster Magnatest Q instruments (American designation, 

Magnatest FS-300). 



mental test frequency. The influence of test-object properties upon harmonic fre 
quencies is described in the literature. 42 

APPLICATIONS. Fig. 26., showing test coils for Magnatest Q instruments, 
illustrates the wide range of test-object dimensions tested with this instrument 
in European industry. The diameter range of the test coils extends from 1 to 
400 mm. Coils with rectangular cross-sections are used for steel blocks and similar 
test objects. Entire metal sheets are tested with the large slit coils. The Magna- 
test Q instrument is the magnetic-quality test instrument most widely used in 
European steel industry. The practical application as well as the interpretation of 
the test results obtained with this instrument will be discussed in the section on 
Eddy Current Test Indications. 

Suppression of Undesired Effects 

TIONS. In the practical test instruments developed during the past 20 years, 
various methods, as appropriate to specific test problems, are used for the suppres 
sion of undesired effects. 23 ' 44 45 The basic theory for nondestructive testing with 
eddy current methods has shown that it is essential to suppress undesired effects 
in test indications. For example, dimensional effects and "lift-off" effects are fre 
quently considerably greater in magnitude than desired effects, such as crack 
effects and conductivity effects. Two methods which will now be described are 
the phase-controlled rectifier and a resonance test system. 

ment of Fig. 13 and the Sigmaflux instrument of Fig. 15 employ an arrangement 
which is termed a "phase-controlled rectifier." This system indicates only those 
components of the test coil voltage which have the same phase direction as the so- 
called control voltage. 

Fig. 27 shows a simple design of the phase-controlled rectifier. The resistors 
#! through # 4 , together with the rectifiers r and r 2 , form a bridge circuit. The 
voltage to be measured, E H , is applied to one diagonal of the bridge, while the 
control voltage # C ontr. is applied to the other bridge diagonal. The control voltage, 

Institut Dr. Foerster 

Fig. 27. Basic circuitry of the phase-controlled rectifier. E M = measurement 
voltage, jEUatr. = control voltage. 


assumed to be of rectangular wave shape, causes a current /^tr. to flow through 
the two rectifiers r x and r 2 during those half-cycles when the control voltage is in 
the forward direction of the rectifier. In the "other half -cycle, the two rectifiers 
f! and r 2 do not conduct, since the voltage E coniTr is in the reverse direction of the 
rectifier. However, the current 7 coutr does not result in a deflection of the 
meter M, since in the absence of a measurement voltage, points A and B 
exhibit the same potential because of the symmetrical bridge circuit. Thus the 
control voltage turns the two rectifiers r : and r 2 into two switches. They are 
"closed" when the current / contri flows in the forward direction of the rectifier. 
They are "open," i.e., they block the current, when the control voltage is in the 
reverse direction of the rectifier. 

In-Phase Operation. Therefore, the measurement voltage E M is applied to the 
meter M only during the positive half-cycle, while the control voltage operates 
in the forward direction of the rectifier. When the control voltage is in-phase with 
the measurement voltage, the results shown in Fig. 28(a) are obtained. From 
point F to G, i.e., during a half-cycle, the measurement voltage of amplitude A is 
applied to the instrument. During the next half-cycle, the measurement voltage 
at the instrument is turned off, as discussed previously. Thus, during the entire 
cycle FH, the average current passing through the instrument corresponds to the 
vertically shaded area over FG. This operation is repeated during the next cycle, 

Out-of -Phase Operation. Fig. 28 (b) shows the conditions with a phase dis 
placement of 90 deg. between measurement and control voltage. The measure 
ment voltage at the meter is now between the points L and M, so that the current 
through the meter corresponds to the shaded area between L and M. However, 
the meter M is a direct-current instrument which averages the positive and nega 
tive portions of the wave form. The mean current value in Fig. 28 (b) equals 
zero because of the equally great positive and negative currents. With a phase dis 
placement of 90 deg. between the measurement and control voltage, the control 
voltage is suppressed in the indication of the meter. The deflection of the meter 
M for the in-phase condition corresponds to the mean value HB plotted in Fig. 
28 (a), since the area under the sine curve from F to G is equal to the area 
(FH) X (HB). 

Operation at Intermediate Phase Angles. With a general phase displacement 
of angle <f> between control and measurement voltage, the mean current value 
through the meter is 

Here E M{m& ^) is the maximum amplitude of the measurement voltage and R is 
the resistance of the meter plus that of the rectifier in the forward direction. 

Suppression of Undesired Signals. An undesired effect can be suppressed if 
one selects a control voltage which is perpendicular to the direction of this 
undesired effect in the impedance plane or complex voltage plane. All other 
effects which are not in the direction of the undesired effect are indicated accord 
ing to their components in the direction of the control voltage; i.e., according 
to their components perpendicular to the direction of the undesired effect. 

The influence of various effects such as conductivity, diameter, permeability, 
or crack effects on the impedance or complex voltage plane of the test coils has 
been described in the preceding sections on eddy current tests. From these discus- 



sions, one can see easily which effects can be separated from each other in the 
indication. For a metallic test rod in alloy sorting, for example, one selects the 
control voltage perpendicular to the diameter direction. On the contrary, to 
register the dimensions of wires, independent of the material, the control 














Fig. 28. Analysis of operation of phase-controlled rectifier, (a) Measurement 
voltage Ex and control voltage E c ^t r . are in phase. Current flowing from the phase- 
controlled rectifier through the instrument A/ corresponds to the area under the sine 
wave from F to (7. The height HB is the arithmetic average value, (b) A 90-deg. 
phase displacement between the control voltage and Ex. The average value of the 
area from L to M equals zero; i.e., there is no meter indication. 

voltage is selected perpendicular to the conductivity direction. An eddy current 
instrument with two phase-controlled rectifiers (the so-called a-D instrument) 
has been developed to measure the variation of the specific electrical conductivity 
and the diameter, independently of each other. It can be used, for example, 
during tensile tests on metal rods. 


40 41 

Suppression of Undesired Harmonics. In Fijs. 28 the effective duration of 
the control voltage FG or LM was assumed to be ISO cleg.; i.e., a half-cycle. 
There are several problems in testing of ferromagnetic materials in which it is 
possible to suppress specific undesired effects, which appear in the harmonic 
frequencies, by selecting an effective duration of the control voltage which is less 
than a half-period (ISO cleg.). Fig. 29(a) shows a distorted voltage, observed 
frequently in induction testing of ferromagnetic objects, which consists of a funda 
mental and a third harmonic frequency. Fig. 29(b| shows the fundamental 
frequency and the third harmonic plotted separately by Fourier analysis. 



Fig. 29. Analysis of harmonics in signal voltage, (a) Composite wave consisting of 
the fundamental frequency and third harmonic, appearing in the testing of ferro 
magnetic materials, (b) Fundamental frequency and third harmonic separated by 

Fourier analysis. 

If the control voltage in the circuit of Fig. 27 acts from point P to S, the funda 
mental frequency is reproduced. In the third harmonic, the area extends above 
PQ and below QR, so that only the area above RS (i.e., one-third of the entire 
effect) is reproduced. Generally, the phase-controlled rectifier circuit responds to 
the nth harmonic with 1/n of the amplitude, as long as the effective width of the 
control voltage is exactly a half-cycle (ISO deg.) of the fundamental frequency. 

By selecting a different effective width for the control voltage, one can entirely 
suppress undesired harmonics in the indication of the phase-controlled rectifier. 
For example, if one selects not ISO deg. but 120 deg. as the effective width of 
control voltage corresponding to the distance PR in Fig. 29 (b), a full cycle of the 

40 42 


third harmonic always lies in this effective range. Thus, with an effective control 
voltage interval of 120 deg., the positive and negative areas of the third harmonic 
always cancel each other in the indication. In this case the third harmonic is 
suppressed entirely, independent of the phase position of the control voltage. 

Condition for Suppression of nth Harmonic. In general, the nth harmonic 
is suppressed in the indication of the phase-controlled rectifier if the effective width 
A< of the control voltage is selected from the following values : 

A0 = ^- (m) (16) 


/ i \ 

1 9 Q . 

1, 4, o 

\ ** s 

For example, to suppress the seventh harmonic (ft = 7), Eq. (16) indicates the 
value of the effective width A< of the control voltage as Ae/> = (360/7) (1), 
(360/7) (2), and (360/7) (3). For an effective width A< of the control voltage, 
the current at the fundamental frequency is 

1 2/ (17) 

Here <f> is the angle between the measurement voltage and the control voltage, 
and A< is the effective width of the control voltage. Inserting 180 deg. for A< 
in Eq. (17) results in Eq. (15). In general, Eq. (17) permits 

1. The suppression of specific undesired effects in the impedance plane by the 
selection of the phase angle of the control voltage. 

2. In addition, for materials whose permeabilities depend upon magnetizing 
field strengths, the suppression of specific harmonics by the selection of the 
effective width A0. 

method, developed in 1947, 30 permits complete suppression of an undesired factor 
in the impedance plane; for example, the "lift-off" effect at coil distances from 
zero to infinitv. 


Institut Dr. Foerster 

Fig. 30. Circuit for complete suppression of undesired effects such as "lift-off" 
effect by partial compensation of voltage across a condenser in the resonant 
circuit. Eo = voltage of an oscillator. E K = compensation voltage. E* = measure 
ment voltage. E e = voltage across condenser. 

Fig. 30 shows a resonant circuit, consisting of the test coil with impedance 
values coZ and R, in series with the capacitance C. The current / through th 
oscillating circuit is to 


As soon as coL = 1/coC, the maximum current in the resonance circuit is 

/(max.) = : | L (19) 

The resistance R can be expressed by the vector OA in Fig. 31, which shows a 
section of the impedance plane. At point on the eoL axis, it is assumed that 
coL = 1/coC. If the resonant circuit is detuned by AcoL, due to a variation of the 
reactance, the resonance current is 

(AcoL) 2 


Fig. 31. Vector relations in the circuit of Fig. 30. 

Circle Diagram. By using the inversion method, one can construct the current 
in the resonance circuit for any given coL value. The inversion method is ex 
plained by Fig. 31. The triangle OAB is similar to the triangle ODA because the 
three angles of the triangle are equal. Thus OD/OA = OA/OB] i.e., 
OD = (OA) 2 /OB. The line OB intersects the half-circle at D, where the distance 
OD is inversely proportional to the distance OB. At the same time OE = (OA) 2 / 
OC and OF = (OA) 2 /OD. The distance OB is identical to the expression 
OB = \/R 2 + (AcoL) 2 in Fig. 31. From Eq. (20) the current in the resonant 
circuit of Fig. 30 is proportional to the distance OD, OE, or OF when the reso 
nance circuit is detuned by the distance AcoL = AB } AC, or AD, respectively. In 
other words the graphic representation of the current in the resonant circuit is the 
half-circle shown above the resistance in the impedance plane at the resonance 
point coL = 1/coC, 



Suppression of Lift-off Effects. Fig. 32 shows the impedance plane with 
lift-off, crack, and conductivity directions designated. The point A represents the 
impedance of the probe coil placed on the test body. Suppose now that the 
resonant circuit is not tuned to resonance according to the coZ/ value G in Fig. 32, 
but to the point 0, where OA is selected perpendicular to the impedance direction 
of the undesired effect. In this case the half-circle ADEG represents the locus of 
the current in the resonant circuit of Fig. 31 while the coil is lifted from the test 
object. When the impedance value of the test coil, while being lifted, changes 
along the lift-off effect curve ABC, the current of the resonant circuit passes 
through the values OA f OD, and OE. 



3 mm DEPTH 


Institut Dr. Foerster 

Fig. 32. Impedance plane showing "lift-off" direction ABC, with conductivity 

a and crack directions. 

Point K (i.e., the center point of the half-circle of diameter OA) is always the 
same distance from the points A, D, and E. In the resonant circuit of Fig! 30, a 
compensation voltage E K (corresponding to the distance OK in Fig. 32) is 
added to the voltage drop across the condenser E c . Under these conditions the 
lift-off can be entirely suppressed in the measurement voltage E M because the 
measurement voltage rotates around the point K, with constant amplitude (KA< 
KD, KE). However, with a crack (for example) having a depth of 0.5 mm. 
(point L) or of 1 mm. (point M), the amplitude of the resonant current is cal 
culated from the relations: OH = (OA) 2 /OL and 01 = (OA)*/OM. Thus cracks 
L and M result in a reduction of the measurement voltage E M of- the resonant 
circuit of Fig. 30, corresponding to the distances KH and KL respectively in 
Fig. 32. -- i 


40 45 

Institut Dr. Foerster 

Fig. 33. Defectometer probe-coil instrument for detection of cracks and crack- 
depth measurement on both ferromagnetic and nonferromagnetic materials. 

Crack Detection. Compensation of the lift-off effect (or any other undesired 
effect) has a strange measurement consequence. As one places the test coil on the 
specific metal (point A, Fig. 32) or removes it entirely, the measurement voltage 
does not react at all. As soon as a crack is present in the metal below the test 
coil, however, an indication is obtained. Thus it is possible to move the test coil 
rapidly and without special care over the surface of a metal. An effect can be 
expected only when the test coil moves over a crack. 


60 r 



20 40 




Fig. 34. Deflection on the Defectometer instrument of Fig. 33 above the cross- 
section of a sound (left) and a cracked piston head (right). 



described here for suppressing undesired effects in the impedance plane are but 
two examples from a number of methods used in eddy current test instruments. 
In the Sigmaflux instrument (American designation, Magnatest FW-300), the 
influence of the test-object dimension is suppressed by means of a phase-controlled 
rectifier of special construction. In the instrument for measuring the diameter of 
wires which pass rapidly through the test coil, the influence of material con 
ductivity is eliminated. 

In the Sigmatest probe-coil instrument (American designation, Magnatest 
FM-100) for conductivity measurements independent of test-object shape, the 
lift-off effect of small coil distances is suppressed. The deflection of the indicating 
meter of the Sigmatest instrument, plotted as a function of the distance from the 
coil to the metal surface, is a cosine curve. 30 - As is well known, the cosine value 
changes only very slightly at small arguments, which has the effect of suppressing 
the lift-off effect. 



Fig. 35. Defectometer indication of porosity in the cross-section of a light metal 




Crack-Depth Tests. In the Defectometer, shown in Fig. 33, the suppression 
of the lift-off effect is obtained by means of a new principle. The Defectometer 
was especially developed for crack detection and crack-depth measurement. For 
example, it is used in the nonferrous metal industry for detection and measure 
ment of cracks in rods, pressed parts, and pistons. Fig. 34 shows the deflection 
curve of the Defectometer above the cross-sections of a sound and of a defective 
piston rod. The small deflections above the sound rod are caused by segregation; 
i.e., by conductivity variations. 

Porosity Test. Fig. 35 shows the influence of porosity in light metal blocks. 
The curves 1 to 5 show the deflection of the Defectometer instrument while the 
probe coil passes over five cross-sections of a light metal block. The lift-off effect, 
of the test coil is entirely suppressed. The presence of porosity is indicated by 
a large deflection. 

Cracks Within Drilled Holes. The semi-automatic crack test instrument of 
Fig. 36 was developed for the detection of cracks in small drilled holes; for 
example, within pinholes in turbine blades. The turbine blade rotates on the 
measurement table. The scanning microprobe coil, which is perpendicular to the 
inside surface of the hole, is moved in the longitudinal direction of the hole. This 
results in a spiral scanning of the inside hole surface. This method does not react 
to distance variations between coil and metal. A defect is indicated by a signal 
This instrument for testing for inside cracks has proved successful for numerous 
test parts. A correlation of defect indication and defect, size was obtained for 
even the smallest size of cracks. 

Tests for Thickness of Nonmetallic Layers. A number of examples can be 
given from the field of nondestructive testing with eddy current methods in which 
a method could be adapted for industrial use only after the suppression of one or 
several undesired effects. The Isometer (American designation, FT-400), shown 

Institut Dr. 
Fig. 36. "Crack detector" for light-alloy compressor blades. 



Institut Dr. Foerster 

Fig. 37. Eddy current instrument for the measurement of the thickness of insulat 
ing layers (Foerster Isometer). 

in Fig. 37, is an example of an instrument which applies this principle to tho 


SSI* to t 






Industrial Test Applications 

Significance of test speed 1 

Automatic testing with the Multitest instru 
ment 1 

Applications 2 

Three automatic Multitest sorters for sort 
ing drills by hardness values, with elec 
tromechanical positioning of test parts 

in the test coil (/. 1) 1 

Fully automatic unit for sorting watch 
springs according to the thermoelastic 
coefficient, with automatic calibration- 
checking device and statistical-evaluation 

instrument (/. 2) 2 

Examples of test objects sorted with the 
Multitest instrument for hardness, alloy, 

heat treatment, or defects (/. 3.) 3 

Handling of test parts 2 

Automatic control 3 

Noncontactmg foil-thickness meter with 
automatic feedback control of rolls (/, 4) 4 

Metallizing of condenser paper 4 

Vacuum metallizing unit with control of 
metal-layer thickness by means of the 

square resistance meter (/. 5) 4 

Automation of the Magnatest Q test 5 

Sorting heat-treated forgings 5 

Three Magnatest Q curves of forged parts 

(A 6) 5 

Bands formed by a large number of Mag 
natest Q curves resulting from forged 
parts in the conditions A, B, and C of 

Fig. 6(/. 7) 6 

Selecting test conditions 6 

Establishing sorting limits 6 

Speed of testing 6 

Adjusting sorting limits 7 

Magnatest Q picture of group C from Fig. 

7 at four times higher sensitivity (/. 8) 7 
Automation of sorting and quality control 

data 7 

Block diagram of the automatic Magnatest 

Q instrument (/. 9) 8 

Handling of large parts 8 

Conveyorized systems 8 

Automatic Magnatest Q unit with con 
veyor belt (/. 10) 9 

Other automation applir.it ions S 

A number of Magnates! Q unit.*, in opprn- 
tion for 100 percent quality control in n 

large European steel plant ( /. 11 ) 10 

Automation of crack testing 10 

Magnetic-analysis equipment 11 

Zuschlag magnetic-analysis crack tot 
instrument with conveyor for semi 
finished parts (/. 12) ....' 11 

Automation of Statistical Quality 

Significance of statistical test data 12 

Automatic statistical control indicator 12 

Statimat used for automatic recording of 
statistical distribution during nrmde- 

structive testing (7. 13) 12 

Statistical records of alloy or carbon content 13 
Statistical distributions obtained for alloys 

{/. 14) 13 

Automat ic sorting systems 13 

Automatic unit for sorting into 13 grow 

during nondestructive tests (/. 15) 14 

Ball-bearing tests 14 

Two automatic instruments for rapid juo- 
duction-Rorting of balls for defects (/. 

16) - 15 

Magnatest Q picture of test parts of ball 
bearing steel (/. 17) 15 

Other applications 17 

Magnatest Q spread bands for chain husli- 
ings with various depths of case, to- 
gether with frequency- distribution of 

case depths (/. 18) 16 

Multitest point picture with Statimat pic 
ture (/. 19) 18 

Evaluation of statistical information 18 

Quality figure for continuous test material,"! IS 

Combined automatic tests 19 

Automatic tester for sorting defective 
ball-shaped test bodies and simultaneous 
sorting into 12 dimensional groups which 
deviate from each other by 40 pn. (/. 

20) IS 

References 19 



Industrial Test Applications 

SIGNIFICANCE OF TEST SPEED. Eddy current methods are distin 
guished from other nondestructive test methods by their unusually high test 
speeds. This lends itself readily to automation of the test processes. Most of the 
Foerster test instruments discussed in the section on Eddy Current Test Equip 
ment can operate fully automatically in industry. In the subsequent text the 
automation of several test processes is discussed. 


Possibilities for the automation of the Multitest method have been pointed out 
previously. (See Fig. 13 in the section on Eddy Current Test Equipment.) Pro 
duction parts, such as balls, needles, or rollers, slide or roll freely at high speed 
through the test coil. The maximum measurement value which occurs when the 
part passes the coil center requires only about 1/1000 sec. for the control of the 
corresponding sorting gates. Fully automatic units have been developed for sort 
ing balls for cracks. Some of these units sort with a speed of more than 10,000 
parts per hour. Such parts are passed freely through the test coil when the 
maximum measurement value in the coil center is sufficient. However, there are 
many parts which have to be tested at a specific point; for example, drills at their 
hardened tips or sewing-machine needles at the needle tip. For this purpose, 
electronically controlled devices were developed which fix the test part at a specific 
point in the coil for about Ho sec. 

Institut Dr. Foerster 

Fig 1 Three automatic Multitest sorters for sorting drills by hardness values, 
with electromechanical positioning of test parts in the test coil. 




Applications. Fig. 1 shows three automatic Multitest sorters which sort the 
main portion of the production in a large European drill factory into three hard 
ness groups. Here, by means of an electromechanical-positioning feature, only 
the tips of the drills are measured. Pressed, light-metal parts are sorted auto 
matically in the same manner for hardness, since, for age-hardened aluminum 
alloys, a linear relationship exists between the electrical conductivity and the 
hardness. Furthermore, automatic Multitest devices serve for sorting arc-light 
carbons according to the thickness of the copper plating. Metal parts are sorted 
for cracks, and steel parts for heat treatment and cracks. 

Fig. 2 shows a fully automatic unit for sorting watch springs according to the 
so-called thermoelastic coefficient. The zero point of the instrument is auto 
matically checked during each rotation of the conveyor belt. The thermoelastic 





Institut Dr. Foerster 

Fig. 2. Fully automatic unit for sorting watch springs according to the thermo 
elastic coefficient, with automatic calibration-checking device and statistical- 
evaluation instrument. 

coefficient ('which indicates how many seconds per day a watch will be fast or 
slow at a 1 C. temperature variation) has a linear relationship to the magnetic 
property measured by the Multitest instrument. The automatic sorter sorts the 
watch springs into three quality groups. An electrical device subdivides the test 
results into 12 groups. Each of the 12 groups controls a channel of the statistical 
evaluation instrument shown in Fig. 2. This instrument automatically indicates 
the frequency distribution curve for the thermoelastic coefficient during sorting 
and permits the production quality to be observed at a glance. Fig. 3 shows a 
selection of production parts (against a centimeter rule) which are automatically 
sorted for quality with the Multitest instrument. 

HANDLING OF TEST PARTS. An automatic conveyor belt is used in 
most of the instruments for the automation of alloy sorting and crack testino- of 
nonferrous and ferrous metal rods. Test parts are passed through the coil on^the 
belt at a constant speed. When a defect appears, a color marking is usually 
applied to the specific point with the aid of a spray gun. In several automatic 
units, two color markings (red and white) are used to indicate the beginning and 
the end of a defect by different colors. 



In the difference-coil method, defects having a constant depth are not indi 
cated. Only variations of the crack depth are shown, as is well known. However, 
the beginning and the end of a crack are indicated by a measurement effect in an 
opposite phase position. This is used to control the color marking with correspond 
ing colors. By means of the different colored markings at the beginning and end 

Institut Dr. Foerster 

Fig. 3. Examples of test objects sorted with the Multitest instrument for hard 
ness, alloy, heat treatment, or defects. 

of a defect, it is possible, especially in the case of long rods, to cut out defective 
pieces. The defective rods are pushed out on the "waste" side after they have 
passed the test coil. In some cases this is accomplished by compressed air control; 
in other cases, by means of a magnetic ejector; and in still others, by a motor- 
driven ejector. 

In several units the test parts are put through the coil of the crack test instru 
ment, as well as through the coil of the Sigmaflux alloy-testing instrument, to 
eliminate cracked and mixed materials at the same time. Still other designs per 
mit dimensional measurements. 

AUTOMATIC CONTROL. The theory of thickness measurement of non- 
ferrous metal foils and layers has been discussed in the section on Eddy Current 
Sphere and Sheet Tests. Fig. 4 shows a foil-thickness meter whose ligfct 
beam galvanometer is equipped with microphotocells. These photocells can be 
moved arbitrarily to any tolerance value on the scale of the instrument. When 
the light-beam indicator exceeds or falls short of the previously adjusted tolerance 
value of the thickness meter, the adjusting motor of the rolling mill is auto 
matically energized until the foil thickness reaches the required value. Another 
photocell, located at or within the required value, again turns off the adjusting 
motor. Therefore the automatic foil-thickness measurement instrument contains 



Institut Dr. Foerster 

Fig. 4. Noncontacting foil-thickness meter with automatic feedback control 

of rolls. 

four photocells which can be arbitrarily adjusted for upper and lower tolerance 
limit control. 

Metallizing of Condenser Paper. The metallizing of condenser paper in 
vacuum is controlled in a similar manner. The paper with the metal layer passes 
at high speed in vacuum through the fork coil arrangement without direct con- 

Institut Dr. Foerster 

Fig. 5. Vacuum metallizing unit with control of metal-layer thickness by means 
of the square resistance meter (shown by arrow) . 



tact. The resistance per unit square of the condenser paper is indicated directly 
in ohms on the "square resistance" meter. The nondestructive^- measured square 
resistance value controls the metallizing process. This is one example of how a 
nondestructive test instrument takes over control of production processes in large 
industrial plants. Fig. 5 shows a large vacuum metallizing unit in an American 
plant in Europe. Here the square resistance meter ( indicated by arrow) is built 
into the instrument panel. 

Q test method, the range with the best possibility of separation of the interesting 
effect (for example, tensile strength, depth of case, or alloy variation) can be 
moved to the screen center where the reading slit is located. In the prior discus 
sion of the linear time-base method, it was shown that the slit value represents 
the measurement effect perpendicular to the base voltage. If one selects the 
direction of the undesired effect (for example, diameter or permeability effect ) in 
the complex voltage plane as the base voltage, the undesired effect is eliminated 
in the slit value. 

With a new electronic method utilizing a special electron-beam-deflection 
tube, only the slit value is used in automation of the Magnatest Q instrument. 
This signal controls the electrical-sorting mechanism and a statistical evaluation 
instrument. Only the instantaneous value of the Magnatest Q curve, exactly in 
the screen center, is used. 

Sorting Heat-treated Forgings. Fig. 6 shows three Magnatest Q curves for 
a forged part in the conditions: A, hard-forged; B, annealed and quenched; 
C tempered to 65 to 80 kg./mm. 2 . The CURT B is adjusted to zero (horizontal 




STRENGTH-' S5-8Q kg/mm 2 

6 Three Magnatest Q curves of forged parts. Conditions: A, hard-forged; B, 
annealed and quenched; C, tempered for 65 to 80 kg, mm. 2 tensile strength. 

41 6 


straight line) by means of the compensators in the instrument. If, for example, 
3000 forged parts are tested in the conditions A, B, and C, the entire curves of 
one condition result in a spread band as shown in Fig. 7. 

Selecting Test Conditions. The main problem of Magnatest Q testing is 
selecting the test conditions such as field strength, frequency, and number of the 
harmonics, to avoid overlapping of the spread bands of conditions which have to 
be separated from each other. By selecting the phase direction in the voltage or 
impedance plane, the screen picture can be moved so that the point with the 
largest vertical separation of the individual bands can be moved into the reading 
slit in the screen center. 







STRENGTH 65-50 kg/mm* 

Fig. 7. Bands formed by a large number of Magnatest Q curves resulting from 

forged parts in the conditions A, B, and C of Fig. 6. Sorting limits are indicated for 

optimum separation of the three conditions, A, B, and C. 

Establishing Sorting Limits. The sorting limits plotted in Fig. 7 are 
arbitrarily adjustable in vertical direction. In each case they are adjusted in the 
center of the space between the bands which are separated from each other. 
The slit values of the three bands, between which the sorting limits are placed, are 
shown to the right in Fig. 7. The slit is used for electronic control of the 
sorting gates. All slit values which fall in ranges I, II, or III open the corre 
sponding sorting gates I, II, or III in the equipment sketched in Fig. 9. 

Speed of Testing. One cycle of the test frequency is sufficient to open the 
corresponding sorting gate while the test part passes through the test coil on the 
conveyor belt. This amounts to 0.2, 0.017, or 0.0025 sec., depending on the fre 
quency of the Magnatest Q instrument used (5, 60, or 400 c.p.s.). Since only one 
cycle is needed to control the automatic sorter, this slit method provides the 
highest test speed that can be obtained with a 5-, 60-, or 400-c.p,s. method. 



Adjusting Sorting Limits. By arbitrarily adjusting the sorting limits, this 
automatic sorter can be very quickly adapted to specific test problems. For 
example, the band C (tempered to 65 to SO kg./mm. 2 ) in Fig. 7 can be moved to 
the screen center of the Magnatest Q instrument by means of the compensators. 



Fig. 8. Magnatest Q picture of group C from Fig. 7 at four times higher sensitiv 
ity. The nodal point of group C has been moved to the screen center to permit 
sorting by tensile strength values. 

By turning the sensitivity knob, for example, to four times higher sensitivity, the 
spread of band C will become four times as wide. By a suitable adjustment of 
the sorting limits, one can now sort the group C into the following three tensile 
strength groups: 

Group I (tensile strength lower than 70 kg., mm. 2 ). 

Group II (tensile strength between 70 and 80 kg./mm. 2 ), and 

Group III (tensile strength greater than 80 kg./mm. 2 ). 

Fig. 8 shows the Magnatest Q screen picture with a sensitivity four times higher 
than that shown in Fig. 7. The group C from Fig. 7 is moved to the screen center 
by means of the compensators. 


Fig. 9 shows schematically the automation device for the Magnatest Q instrument. 
The test part passes through the test coil on the conveyor belt. When the part 
enters the coil, the position of the sorting gate is reset from the previous position. 
The visual indicator A shows the test-part curve, while the electronic slit B 
utilizes only the instantaneous value in the center of the reading slit. The 



electronic slit controls the sorting gates within one cycle, triggers at the same 
time the colored signal lamp which corresponds to the sorting group, counts the 
test part, and steps up the counter for the specific sorting group. Thus the elec 
tronic slit not only carries out automatic sorting; it also gives numerical data for 
quality control, including the entire number of parts tested and the number of 
test parts in groups I, II and III. Since the counters can be returned to zero 
only by means of special keys, one is able to retain, for example, the quality test 
results obtained during a night shift. 



Institut Dr. Poerster 
Fig. 9. Block diagram of the automatic Magnatest Q instrument. 

Handling of Large Parts. For larger test parts which are difficult to transport 
on a conveyor belt, the electronic slit method can be used to reduce human error. 
Lamps are clamped to the corresponding containers for the three or more groups 
of test parts. These light up, according to the test group, so that the operator 
merely places the test part in the corresponding container indicated by the lamp. 

Conveyorized Systems. Fig. 10 shows an automatic Magnatest Q unit with a 
conveyor belt. The hourly rate of testing lies between 3000 and SOOO parts, de 
pending on the size of the test part. The conveyor belt is suitable for test parts 
whose maximum value during passage through the coil is representative of the 
test part. However, some parts such as drills with soft shafts, complicated forged 
parts, and certain screws require testing at a specific point or in a specific position 
in the coil. Use is then made of automatic tilting or moving coils in which the 
test parts are maintained in the coil in a specific position during testing. After 
the test the coil is tilted by a motor, and the test part slides through the sorting 
gate into the container which corresponds to its quality. 

ments utilizing the Magnatest Q test process will sort, for example, long spring 



Institut Dr. Foerster 
Fig. 10. Automatic Magnatest Q unit with conveyor belt. 



rods into three hardness groups, or in the testing of spheres and rollers, will sort 
out parts which are too soft after heat treatment. In sorting production parts, 
it is always advisable to adjust the automatic equipment to the test part. An 
automatic sorting instrument which must sort large forged parts as well as small 
rollers, etc., necessarily has low efficiency. The significance of production sorting 
of steel parts with eddy current methods, in the steel industry, can be seen in 
Fig. 11, which shows a number of Magnatest Q instruments (indicated by arrows) 
for 100 percent quality control, in operation at the same time in a large European 
steel plant. The Magnatest Q instrument is the most widely used eddy 
current test instrument in European industry, evidencing the great desire for 
automation of the test process. 

Institut Dr. Foerster 

Fig. 11. A number of Magnatest Q units in operation for 100 percent quality 
control in a large European steel plant. 

Automation of Crack Testing. The Magnatest D instrument for steel crack 
testing is automated in much the same manner as the Magnatest Q instrument. 
However, it is equipped with difference coils instead of the absolute coils which 
are used in the Magnatest Q instrument. The defects in the rods are marked with 
color-spray devices. A time delay, which is adjusted to the speed of the semi 
finished parts, assures that the rod is only marked with paint when the defective 
location has reached the point of the paint marker. The beginning and the end 
of the defect are marked in different colors. 

European industrial users of this crack-test instrument state that cracks with 
depths even as small as 1 percent of the rod diameter can be found. 80 However, 
steel types with strong magnetic-permeability variations (heavy straightening 
stresses) are less suited for eddy current testing. Fig. 23 in the section on Eddy 



Current Test Equipment indicate,- how pcrme'itbiHty ptiVet- imirhitnirai -.tri^c-i 
can be separated in the indication from crack effect,-. However, when piTiwaiiiiiTv 
effects caused by inhomogeneous mechanical stresses greatly cmtwi-inh the 
crack effect, separation of crack signals from tin much greater ptTmf\ihility effect 
is very difficult. By mean* of a special te>t-coil dr>kn which \va.- develojed rnun 
exact field-distribution calculations, th< defect r-ui-itivity war- coiiMthrably in 
creased, while the so-called fluttering effect of the tost rod wa- brought to 7<*ni. 

Magnetic Analysis Or]. 

Fig. 12. Zuschlag magnetic-analysis crack test instrument with conveyor for 

semi-finished parts. 

Magnetic-Analysis Equipment. Fig. 12 shows the American crack-test instru 
ment which was developed by Zuschlag of the Magnetic Analysis Corporation.* 1 * 
Zuschlag's work in the development of the basic apparatus for the slit methods 
was of great merit. The instrument shown in Fig. 12 serves for .simultaneous test 
ing for cracks and mixed alloys. The conveyor can be seen behind the instrument. 
Defective rods are pushed out on the "waste" side. 



Automation of Statistical Quality Control 

sorting methods are outstanding because of their test speed, they are especially 
suited to production parts. In this application statistical evaluation of the test 
results gains steadily in importance because the quality of a product is character 
ized by the statistical frequency distribution of measured values such as hardness, 
depth of case, or type of alloy. A large spread in hardness values for a product, 
for example, indicates an error in the hardening process. An especially large 
spread in the eddy current test indications for the same parts before hardening 
indicates alloy differences or heat variation. 

trates the Statimat which can be connected to the various automatic sorters to 

* * <* .' > <" ' ' v f^ ?ij ' V 'i** jf'v **4i ^'^'iff^ ^fmi^^4 
v ! ,t ]i'^ < ^ / 4 /ffi, ' . ' ".feiy 1 ^" ' ' '" t ' fti ^ J( " M * :_!:Z ^ZI--jiaga^ 

Institut Dr. Foerster 

Fig. 13. Statimat used for automatic recording of statistical distribution during 

nondestructive testing. 


indicate statistical distributions during sorting; for example, of carbon content, 
depth of case, tensile strength, hardness, magnitude of cracks, and other charac 
teristics. As soon as the indicating line of a sorting channel reaches the upper limit 
(selected, for example, as 100 parts or 1000 parts), the statistical distribution can 
be automatical!}' determined, after which all indicator lines are returned to zero 
for the next test run. The Statimat also contain:? a number of preselector knobs 
to provide a signal (for example, for turning off a production machine I. For 
instance these operate if the waste at specific sorting channels exceeds a specific 
percentage of the total number of test parts. 

Statistical Records of Alloy or Carbon Content. Fig. 14 illustrates the 
statistical records which can be obtained. Fig. 14(a) shows the distribution of the 
carbon content in one heat of a carbon steel. Fig. 14(b) illustrates the statistical 
distribution of the carbon content for two mixed heats of the same allov. Two 








Fig. 14. Statistical distributions obtained for alloys, (a) Statistical distribution of 
the carbon content in one heat of a carbon steel (horizontal axis, C content ; vertical 
axis, number of test parts with a specific C content), (b) Statistical distribution of 
carbon contents for two mixed heats of the same alloy, (c) Statistical distributions 
for two different alloys with satisfactory separation, (d) Statistical distributions in 
the eddy current sorting of three alloys with satisfactory separations. 

clearly indicated, maximum Gaussian curves correspond to the slightly shifted 
carbon content in the two heats. Fig. 14(c) shows the statistical distribution for 
two different alloys. The Gaussian curves do not indicate any overlapping. Thus 
Fig. 14(c) clearly illustrates the reliable sorting possibility of these two alloys. 
Fig. 14(d) illustrates the corresponding statistical distribution in eddy current 
sorting of three alloys which can be reliably separated. 

Automatic Sorting Systems. Fig. 15 illustrates an automatic instrument for 
sorting into 13 groups. It is used increasingly for nondestructive-sorting processes, 
for dimensional sorting, and for combinations of both methods. The measurement 
factor (depth of case, hardness, carbon content, or dimension) is converted into a 
light deflection whose position can be read on the large scale in Fig. 15. New 



microphotocells whose light-sensitive area is less than 10-' J sq. in. are located 
below the scale. 

During automatic sorting, this light indicator passes a specific number of 
microphotocells while moving from off-scale to the measured value. The number 
of cells passed determines the sorting group. While the light indicator passes a 
microphotocell, an electric impulse is created by the cell. Thus the corresponding 
sorting group is clearly indicated by the number of impulses obtained (i.e., photo- 

Institut Dr. Po 
Fig. 15. Automatic unit for sorting into 13 groups during nondestructive tests. 

cells passed by the light indicator). The number of photocell impulses elec 
tronically counted opens the corresponding sorting gate. The 13 sorting limits 
can be easily adjusted by hand to any desired value. For example, in automatic 
eddy current sorting of sheet materials according to thickness, all sorting- 
limits can be moved so close together that an eddy current sheet-thickness sorting 
with 13 tolerance groups, each of 0.0004-in. width, is obtained. This has proved to 
be very satisfactory in industry. 

Bali-Bearing Tests. The largest part of the balls produced by a well-known 
European ball-bearing plant pass through the two eddy current test arrangements 
shown in Fig. 16 to be tested for cracks. The test speed lies between one and five 
balls per second, depending on the size. The balls roll freely through the test coil 
and are tested for cracks in 1 msec. 

Fig, 17 shows the Magnatest Q picture of test parts made of ball-bearing steel. 
Several specific screen patterns are shown on the left. The mechanically meas 
ured hardness values corresponding to a specific value are indicated to the right 
of the Magnatest Q patterns. The Statimat picture is shown on the right side of 


Institut Dr. 

Fig. 16. Two automatic instruments for rapid production-sorting of balls for 



Fig 17 Magnatest g> picture of test parts of ball-bearing steel. Hardness values 

eiven for corresponding slit value? (left). Frequency distribution of hardness values 

shown by Statimat instrument (right). 







Fig. 18. Magnates! Q spread bands for chain bushings with various depths of case, 
together with frequency distribution of case depths. 

Fig. 19. Multitest point picture with Statimat picture, (a) Actual Multitest point 
picture of pressed parts of AlCuMg. with plotted hardness values, (b) Multitest pic 
ture of (a), turned so that the hardness direction is in the horizontal sorting direction. 
Statimat picture of frequency distribution of hardness values is shown below. 


lust hut Dr. Foerster 

Fig. 19. (Continued.) 

the figure. The frequency distribution of the test parts with specific hardness 
values is also illustrated. The hardness value, 53 Re.., appears most frequently as 
shown by the Statimat picture. 

Other Applications. Fig. 18 shows the Magnates! Q band for chain bushings 
with various depths of case hardening. The depth-of-case frequency distribution 
(Statimat picture), obtained automatically, is illustrated to the right. The largest 
number of test parts has a depth of case in the vicinity of 1 mm. 

In testing with the Multitest instrument, the Statimat picture gives all impor 
tant information concerning the quality of a product. Fig. 19 1. a) shows the Multi- 
test point picture of a number of AlCuMg pressed parts. In Fig. 191 bi the point 
picture is turned, by means of the phase shifter 5 shown in Fig. 13 of the section on 
Eddy Current Test Equipment, so that the hardness direction is m the horizontal 
sorting direction. The frequency distribution of the hardness values obtained with 
the Statimat instrument is shown below the Multitest picture. 



Evaluation of Statistical Information. In general, one can see that the 
Statimat picture obtained automatically in nondestructive quality-sorting of 
production parts gives important information concerning the quality of the pro 
duction. Trouble in a production run is indicated directly in the Gaussian curve 
on the Statimat instrument. One can frequently draw conclusions as to the cause 
from the deviation from the "normal Gaussian curve." Depending on whether (1) 
a displacement of the maximum of the Gaussian curve appears, (2) the Gaussian 
curve, at a constant position of the maximum, increases in width, or (3) several 
maximum points appear, one obtains important information concerning the type 
of trouble in the production process. 

Quality Figure for Continuous Test Materials. A different statistical quality 
number is used in testing for defects such semi-finished parts as rods, tubes, and 

Institut Dr. Foerster 

Fig. 20. Automatic tester for sorting defective ball-shaped test bodies and 
simultaneous sorting into 12 dimensional groups which deviate from each other 

by 40 jidn. 


especially wires. By indicating the number of meters of semi-finished parts tested 
and by electronically counting the defects which have appeared in this length, the 
quality number Q of the tested material can be defined as 

* __ number of meters ,-. 

number of defects 

Thus the quality number gives the average number of meters per defect which 
have appeared. The characterization of the quality by means of a number is 
important, especially for wires in which defective sections cannot be cut out. 

COMBINED AUTOMATIC TESTS. It can be expected that the automa 
tion of nondestructive test processes will expand in the future. The test parts 
must be brought to the instrument by means of a transport mechanism in all 
automatic-sorting operations. In this case it is desirable to combine into a chain 
of tests the various quality tests such as crack testing, dimension testing, rough 
ness testing, form testing, testing for eccentricity, and testing of the depth of a 
case layer. Fig. 20 shows a combination automatic tester in which ball-shaped 
bodies are tested simultaneously for cracks and dimension. After the crack-testing 
operation, in which defective parts are sorted out, the test parts pass the dimen 
sion-testing station and are sorted automatically into 12 groups which deviate 
from each other by lu = 40 uin. 

The Gaussian curve of the dimensions of the test parts is obtained automatically 
during the sorting process. The form of the Gaussian curve permits one to draw 
important conclusions concerning the production process. The automatic instru 
ment shown in Fig. 20 can be used to obtain the statistical distribution of the 
degree of sphericity of the test parts. For this purpose, the test parts which 
were sorted into a specific sorting channel and which may still exhibit diameter 
variations of Ijx (40 jiin.) are sorted once more. One now obtains a Gaussian 
curve with a steep slope within the limits of three adjacent sorting channels. This 
is a result of a certain lack of sphericity of the test parts, since their diameters 
can vary, depending upon the specific point at which they \vere measured. By 
repeating the sorting of test parts from one diameter group, a certain number of 
test parts will fall in the neighboring channels as a consequence of their lack of 

This principle of the determination of quality variations around the circum 
ference of a test body can also be applied advantageously in nondestructive-sort 
ing processes (for example, for detecting soft spots in spheres). Methods such 
as these are especially helpful in meeting the increasing industrial demand for 100 
percent, nondestructive, quality testing and sorting. 


Reference numbers in this section refer to the references at the end of the 
section on Eddy Current Test Indications. 






Range of Test Applications 

Significance of test method ................. 1 

Significance of electrical conductivity ...... 1 

Significance of hysteresis loop characteristics 1 

Significance of impedance characteristics .. 2 

Xoncontacting dimensional controls .......... 2 

Diameter variations of a rod of nonfer- 
rous metal, recorded while the rod is 
passed without contact through the 
Foerster Sigmaflux instrument (Ameri 
can designation, Magnatest FW-300) 
(/. 1) ................................... 2 

Measurement of electrical conductivity with 
a feed-through coil ....................... 3 

Light alloys ................................ 3 

Tungsten wires ............................. 6 

Wavy fiber texture of a tungsten wire, 
caused by excessively large steps in the 
swedging process (f. 2) ................. 3 

Crack testing with feed-through coils ....... 6 

Indications of the ellipse method, includ 
ing test samples with decreasing diam 
eter and surface and subsurface cracks 
with increasing depth (/. 3) ............ 4 

Defects in tung&ten and molybdenum wire .. 6 
Schematic representation of typical de 
fects in tungsten rods, as> detected with 
the Foerster wire -crack test instrument 
(American designation, Magnatest FW- 
200) (/. 4) .......................... v .. 7 

Microsections of tungsten rods and wires 
corresponding to the schematic drawings 
of defects in Fig. 4 (f. 5) .............. 8 

Microsections from the tungsten and 
molybdenum industry (/. 6) ............ 9 

Conductivity Measurements 

Application of the probe coil method ........ 

Probe coil test procedure .................... 

Testing rods of mixed alloys by applying 
the probe coil of Foerster Sigmatest to 
the rod ends (/. 7) .................... 

Control of copper castings ................... 

Decrease in conductivity of copper, cuu,*ed 
by various impurities (/. 8) ............ 

Phosphors content ........................ 

Oxygen content ............................ 

Variation of the conductivity during de- 
oxidation of a copper melt ^ (/. 9) .... 

Electrical conductivity ..................... 

Control of aluminum cas-tings 14 

Influence of metallic additive* on the con 
ductivity of aluminum </. ID) 14 

Silicon content 1* 

Determination of ^eizregation 14 

Conductivity variation WL\MX\ by ^KT^H- 
tion in a cast Mock of AlCuMg allr> 

(f. llj 15 

Testing of mixed alloy* 15 

Electrical conduct mtv values fnr vanuu.- 

Amencan allays </. 12} 1" 

Austemtic stainless steels. 15 

Indirect hardness measurement IS 

Relationship between conductivity and 
hardness in cold age- hardener! alloys 

(/. 13) 1* 

Control of heat treating 19 

Conductivity and hardness of sheets of an 

AlCuMg alloy (/. 141 19 

Quench -hardened parts 19 

Indications of a soft s-pot in s-heet f/. 151 20 

Titanium alloys 21 

Hysteresis-Loop Tests 

Quality testing of ferromagnetic material* ... 21 

Correlation problems 21 

Precision detennination of hycteresis loops .. 21 
Standard teat object for investigating the 
ferromagnetic properties of a material 
on the Foerster Ferrograph f/. 16) .... 21 
Device for magnetizing standard ring 
samples with axial current and measure 
ment of the field strength with the air 

ring coil </. 17) 22 

Screen picture of the Ferrograph with 
calibration of B and H wales f/. IS 1 ) .. 23 

Determination of static hysteresis loop 23 

Supenrnpo.serl hyjateww loup of standairl 
ring sample of Fig. 16 and Cample from 
the j-onie niatenal but with a \ 2" larger 

wall thickne </. 19) 24 

Correction for eddy current effect- 24 

Analyst of li>Mere-i Iw^i jiattenis 25 

Effects of internal "ire-!' -" 

Family of hy-ieie-.* I(MI|.* of a noimal 
-wiiii'le nf P1UI5 re ili 1 -drawn i-jiilmn -it*! 

</. 201 2C 

Family oi hy-teies- l<*l* ul ihe nmiiial 
Cample from Fip. 20 but Mie^-ieliei 

annealed (/. 21 j 26 

Kffects of annealing 25 



CONTENTS (Continued) 


Superimposed loops of the sample from 
Fig. 21, but carburized for 1 hr. at 
900 C. in a C- atmosphere (loop with 
higher magnetization values) and 
quenched (loop with smaller magnetiza 
tion values and higher coercive force) 

(/. 22) 27 

Effects of carburization 28 

Superimposed magnetization loops of two 
normal samples with various hardness 

(f. 23) 27 

Predicting mechanical and structural char 
acteristics 28 

Internal stress in rnangetostrictive materials 28 
Influence of tensile stress on the mag 
netization loop of a hard nickel wire 
(/. 24) 28 

Comparator Bridge Tests 

Principle of operation 29 

Magnates! Q pictures derived from the 
difference in the magnetization loops (/. 

25) 30 

Magnatest Q sorting 31 

Recording test results 32 

Magnatest Q picture of cobalt alloys hav 
ing similar alloy contents (/. 26) 32 

Range of test applications 33 

Setting up Magnatest Q instrument 33 

Adjustment of the Magnatest Q screen 
picture for optimum test conditions (/. 

27) 34 

Sensitivity calibration 33 

Compensation adjustments 33 

Spread band characteristics 33 

Magnatest Q curve and spread band of 
C1015 carbon steel and a free-cutting 

hteel (/. 28) 36 

Selecting optimum test conditions for sorting 

alloyb 33 

Separation of three different alloys 34 

Magnatest Q picture of three alloys, 3415, 
5115, and free-cutting steel with 0.22C, 
with an optimum screen utilization (/. 

29) 36 

(a) Spread bands of a large number of 

rods of C1034, C1045, and 5140 steels. 

(b) Spread bands of unfinished piston 

rods of C1015 and 5115 steels (/. 30) .... 37 

Use of harmonica in separating similar 

alloys 34 

.Sort ing mixed lots 35 

Separation of C1034, C1045, and C1075 
carbon steels and free-cutting steel with 

0.1 percent C (/. 31) 38 

Testing billet materials 35 

Separation of hot-rolled billets of 5132, 

CI075, 5140, and 6150 steels (/. 32) 39 

Spread bands of hot-rolled billets of 
60 X 60 mm. cross -sect ion, of E52 100 
steel, a wrought heat-resisting steel for 
valves, and C1034 and C1015 carbon 

steels (/. 33) 40 

Use of low field strengths 3S 

Spread bandh of rolled billets of the two 
t^prinff steel alloys A anrl B (see anal 
ysis) (/. 34) 41 

Spread bands of the alloys of Fig. 34, but 
with an eight times lower field strength 
(/. 35; 42 

Optimum utilization of screen area 3fl 

Spread bands of 5140, C1045, and 3115 
steels with optimum screen utilization 
(f. 36) 43 

Spread bands of the two alloys E4132H 
and 5132, which differ by only 0.23 per 
cent Mo (/. 37) 44 

Influence of test frequency 40 

Spread bands of the five alloys, C1015, 
C1034, 5115, free-cutting steel with 0.1 
percent C, and free-cutting steel with 
Mn, at a test frequency of 60 c.p.s. 
(/. 38) 45 

At a test frequency of 5 c.p.s., a separa 
tion of the five alloys from Fig. 38 is 
possible at the dashed vertical line 

(/. 39) 45 

Tests of wire samples 41 

Spread bands of C1045, C1034, and 5140 
steels at 60 c.p.s. (/. 40) 46 

Satisfactory separation of the three steels 
with strong inhomogeneous stresses 
from Fig. 40, at a test frequency of 

5 c.p.s. (/. 41) 46 

Advantages of low frequency ratios 42 

Complex voltage plane of the Magnatest 
Q coil which contains a test body with 
the diameter d, the electrical conductiv 
ity ff, and the permeability /irei. (f. 42) 47 

Angle between the conductivity direction 
and the diameter- permeability direction, 
as a function of the f/f p ratio (/. 43).. 48 
Permeability effects at high frequency ratios 42 
Effects of inhomogeneous internal stresses . 43 
Influence of field strength 44 

Spread bands with a test frequency of 
60 c.p.s., corresponding to Fig. 40 but 
taken with a higher test field strength 
(/. 44) "... 48 

Spread bands corresponding to Fig. 44 
but with a test frequency of 5 c.p.s. 

(/. 45) 49 

Effects of straightening 47 

Spread bands of the 4140 and 5140 alloy 
steels at a test frequency of 60 c.p.s. 
(/. 46) 49 

Spread bands of 4140 and 5140 alloy steels 
of Fig. 46 at a test frequency of 5 c.p.s. 

(/. 47) 50 

Effects of stress relief by annealing 50 

Spread bands of 4140 and 5140 alloy steels 
at a test frequency of 60 c.p.s. after 
stress-relief annealing (/. 48) 51 

Spread bands of annealed 4140 and 5140 
alloy steels of Fig. 48, but with a test 

frequency of 5 c.p.s. (/. 49) 51 

Separation of drawn, annealed, and tem 
pered carbon steel 52 

Spread bands of C1034 carbon steel for 
three conditions: cold-drawn, annealed, 
and tempered (/. 50) 52 

Three spread bands of C1034 steel for 
three tensile strength groups: 74,500 to 
79,000, 83,000 to 87,500, and 90,000 to 
100,000 p.s.i. (/. 51) 53 

Evaluation of Fig. 51, showing measured 
tensile strength values as a function of 

the Magnatest Q slit value (/. 52) 54 

Sht indications for quantitative measure 
ments 53 



CONTEXTS (Continued] 


Porting for tensile strengths 54 

(a) Spread hand of C1036 carbon steel at 
various tensile btrength step?, (b) Re 
sults of Magnates! Q sorting (f. 53) ... 55 
Forged parts 54 

Sorting for tensile strength of support 
levers of tempered C1040 carbon steel 
(/. 54) 56 

Sorting of the parts from Fig. 54 for ten 
sile strength, where the small journal of 
the support lever is in the test coil 
(f. 55) 56 

Testing of large crankshafts for correct 
metallurgical condition of the journals 
(/. 56) .' 57 

Screen picture of drop- forged parts of 
C1034 carbon steel hardened (bottom 
curve) and annealed for various tensile 

strengths (/. 57) 57 

Hardness testing of ball and roller bearings 58 

Spread bands of ball-bearing steel (ball 
bearing rings) hardened and annealed at 

various temperatures (/. 58) 58 

Measurement of case depth on case-hardened 

production parts 58 

Magnatest Q spread bands of case- 
hardened valve stems, which were case- 
hardened for 8, 12, and 16 hr. to obtain 

various depths of case (/. 59) 59 

Use of probe and yoke coils 59 

Valve stems 59 

Magnatest Q spread hands? of valves with 

various depths of ca^e (/. 60) 60 

Applications in materials research 60 

Aging of Thomas steel 60 

Results of age-hardening experiment* on 

Thomas steel (/. 61) 61 

Pla>tic flow of teel during tensile te^T- , .. 60 

Crack Testing of Steel 

Principle of operation 61 

Separation of crack and stre-s effect^ 61 

Analysis of screen patterns 62 

Multiple-exposure pictuie of cuives ap 
pearing on screen when a rod with 
inhomogeneous stresses passe* through 
the test coil of the Magnatest D instru 
ment (/. 62) 62 

Multiple-exposure picture of a rod with 

cracks (/. 63) 63 

Interpretation of crack indications 62 

Indications of sound rod 62 

Analysis of sound test rod (/. 64 > . .64 

Indications of defective rod 63 

Analysis of cracked rod (/. 65) 65 

Indications of fine crack 68 

Analysis of fine crack (/. 66) 66 

Indications of short defects 68 

Analysis of short discontinuities (/. 67).. 67 

Crack-test equipment installation 69 

Special applications of eddy current methods 69 

References 69 




Range of Test Applications 

SIGNIFICANCE OF TEST METHOD. The basic theory of eddy current 
testing and design measures used in the test instruments have been discussed in 
the preceding sections on eddy current tests. By the means described in these 
sections the eddy current method was developed to a quantitative measurement 
method which permits the magnitude of an effect to be determined from the test 
indication. In the same manner in which the electrical conductivity of a material 
can be obtained from the indication of a Thomson bridge and the dimensions of a 
test rod, the value of the specific conductivity of a material can be determined 
from the indication of an eddy current instrument, from the test frequency and 
the diameter of the test cylinder. (See sections on Eddy Current Test Principles 
and on Eddy Current Cylinder Tests for a more complete discussion of theory 
and methods of measurement.) 

If slight variations of the conductivity or the diameter of the test body have to 
be detected, the sensitivity and accuracy of the eddy current method is far 
superior to the normal, classical bridge methods. Even a variation of 10~ 5 in 
the diameter or electrical conductivity results in a noticeable deflection. The 
significance of the eddy current method is especially obvious if one realizes that 
these measurement values are obtained without direct contact while the test object 
passes through the test coil at high speed. The physical material properties, such 
as electrical conductivity, permeability, coercive force, and magnetic saturation, 
which are determined by the eddy current method, characterize the test part in 
many ways. 

Significance of Electrical Conductivity. The electrical conductivity is an im 
portant indication of the condition of nonferrous metals; for example, of the 
purity of unalloyed metals. The magnitude of the electrical conductivity is of 
significance especially for the electrical industry. For age-hardenable aluminum 
and titanium alloys, the electrical conductivity is a measure of hardness. Mixed 
alloys can often be sorted by conductivity measurements. Segregation effects 
can frequently be indicated by the curve of the conductivity over an entire metal 

Significance of Hysteresis Loop Characteristics. For ferromagnetic alloys 
the characteristic values of the hysteresis loop give important information con 
cerning the composition and the structural condition. Relationships between the 
magnetic and the mechanical properties of materials were discovered long ago. By 
1889 Hughes had applied a magnetic method for separating hard and soft steel 
parts from each other. 57 Three years earlier, Ryder applied for an American 
patent for the determination of carbon content in iron by measuring the perme- 

42 1 



ability. 70 The American Society for Testing Materials established a Committee 
on Magnetic Analysis in 1917. The work of this committee was concerned with: 

1. Testing steel qualities. 10 

2. Testing of ball-bearing races."' 7 

3. Testing of the relationship between mechanical and magnetic behavior of cast 
iron. 86 

Significance of Impedance Characteristics. However, only modern electronic 
techniques can subject electrical and magnetic properties of materials to 100 
percent, nondestructive quality sorting. Only recently has it been possible to re 
solve the influence of the electrical and magnetic material properties on the im 
pedance characteristics of a test coil whose electromagnetic field reacts on the 
test object. The eddy current methods permit one to obtain, for an unknown test 
object and an unknown test coil, the frequency ratio }/f ff and the number A T , the 
percentage of the "absolute value" per inch of instrument deflection. If the instru 
ment frequency is known, then one obtains the value of the electrical conductivity 
directly from f/f ff = /o~f/ 2 /5066. Furthermore, as a result of the basic analyses 
given in the preceding sections on eddy current tests, one can state (for each 
sensitivity step of the instrument) how large the instrument deflection is for a 
conductivity variation of 1 percent, a diameter variation of 1 percent, and for a 
crack of a given depth. These characteristics make the instruments discussed in 
the section on Eddy Current Test Equipment a quantitative measurement method. 

eter variations of a rod of an AlCuMg alloy, which were recorded with the 
Sigmaflux instrument (American designation, Magnatest FW-300). Diameters of 
the rod measured with a micrometer are plotted below the recorded curve. The 


IQ ftA 






1 ] 










Institut Dr. Foerster 

Fig. 1. Diameter variations of a rod of nonferrous metal, recorded while the rod is 

passed without contact through the Foerster Sigmaflux instrument (American 

designation, Magnatest FW-300). 



measurement results obtained with the eddy current method con be checked 
by means of micrometer measurements in each case. 

The nondestructive^ obtained recorder curves of the diameter of semi-finished 
parts which originated from various metal plants permit an accurate identification 
of the producer. Each producer has its special dimensional characteristics, which 
permit one to draw conclusions concerning the production process. The recording; 
of the dimension of a wire during drawing continuously and quantitatively indi 
cates the wear of drawing tools. For tungsten and molybdenum wires with 
ground surfaces, periodic diameter variations resulting from incomplete grinding 
processes are clearly indicated by the eddy current method. In tensile testing of 
metal rods, the cross-sectional variation during elongation of the rod can be meas 
ured quantitatively with high sensitivity by the eddy current method. The 
Poisson ratio is obtained from the elongation and decrease in cross-section. Pro 
duction parts can be sorted according to dimension with the Multitest instrument, 
independent of the material. 

With the inside coil one can quantitatively determine the inside diameter of the 
tube without dependence on the test material. The detection of corrosion on the 
inside diameter of a tube using an inside coil gives the same indication as an 
increase in tube inside diameter. 

FEED-THROUGH COIL. The Sigmaflux instrument (American designation, 
Magnatest FW-300) is used for the sorting of semi-finished parts for electrical 
conductivity and thus for alloy. There are relatively few nonferrous alloys whose 
conductivities overlap, so that in most cases the conductivity is representative for 
the alloy (see Fig. 12). 

Light Alloys. In the continuously quenched aluminum and magnesium alloys, 
soft spots can appear in the semi-finished parts as a result of faulty water flow, 
steam bubbles, etc. These areas can be detected satisfactorily because of their 
conductivity variations. [Measurement of electrical conductivity while wire or rod 

Institut Dr. Foerster 

Fig. 2. Wavy fiber texture of a tungsten wire, caused by excessively large steps 

in the swedging process. 

f - frequency 
6 - conductivity 
Mrel 'Permeability 

Fig. 3. Indications of the ellipse method, including test samples with decreasing 

42 4 

Institut Dr. Foerster 
diameter and surface and subsurface cracks with increasing depth. 


passes through the test coil often gives valuable clues concerning the production 
process. For example, rods made of cold age-hardened aluminum alloys fre 
quently show a characteristic conductivity curve along the rod. This was found 
to be due to a temperature decrease along the rod during age-hardening, resulting 
in an undesirable hardness curve, 

Tungsten Wires. In 100 percent testing of tungsten wire production, periodic 
conductivity variations appeared along the wire from time to time, as was deter 
mined by Keil 63 . These were indicated by a periodic opening of the straight line 
into an ellipse on the screen of the wire-crack test instrument. Fig. 2 shows a 
microsection of such a wire with a periodic conductivity variation. This was 
caused by the wavy fiber texture which resulted from excessively large steps in 
the s wedging process or by unsuitably formed swedging jaws. The material was 
not uniformly distributed along the wire shown in Fig. 2. This is not an actual 
physical discontinuity; however, such material with a wavy fiber texture is un 
desirable for electrical contacts, since burning off under the action of sparking 
is influenced by the condition of the fiber texture with reference to the contact 
surfaces. 62 Fig. 2 is an example of how nonunif ormities in the production process, 
which normally would not be discovered, are easily determined through non 
destructive measurement of an important material factor. 

cylinder with a local diameter decrease of 5 percent and a surface crack which 
increases in depth from to 30 percent (percentage of the cylinder diameter). 
This crack disappears beneath the surface so that it finally lies 10 percent below 
the surface. The photographs of the ellipse test screen picture show the diameter 
decrease and the cracks with various depths and locations; below these the 
various effects are indicated in the ellipse method at f/f g = 5, 50, and 150. The 
horizontal voltage is selected in the direction of the diameter so that diameter 
variations are represented by an inclined straight line. Fig. 3 shows that cracks 
below the surface can no longer be indicated at high f/f ff values. Furthermore 
it illustrates that subsurface cracks with increasing depth below the surface result 
in a horizontally lying ellipse, which can even tip in the opposite direction (for 
example, a crack of 30 percent depth, 5 percent below the surface) . 

The crack representation of Fig. 3 is used in the ellipse method for a given /// 
value, along with the instrument scale deflection, to determine the depth and 
location of the discontinuity. It should be pointed out that the adjusted sensitivity 
is only 1/100 of the maximum sensitivity of the test instrument in Fig. 3. Even 
cracks having a depth of 1 percent of the diameter will result in an ellipse which 
can be readily evaluated at maximum sensitivity. 

allows schematically the types of gross defects in W and Mo wire which were 
detected over a period of years by Keil and Meyer 3 with the wire-crack test 
instrument. The defects (a), (b), and (c) are compacting defects. During the 
compacting of the metal powder before sintering, the compacting form sometimes 
yields and springs back after the compacting die is released. This causes the 
formation of hairline cracks in the compacted body. These cracks cannot be 
welded in the subsequent finishing process. Such cracks can appear along the 
lower edge of the die or can proceed diagonally from one edge to the other of the 
compacted part. In the latter case, the cracks are distorted into S-shapes during 
the swedging process [Fig. 4(b)]. Fig. 4(d) shows a lap on the rod surface. An 
erroneous sequence in dimensions of swedging jaws and too low swedging tempera- 

Fig 4 Schematic representation of typical defects in tungsten rods, as detected 
with lie Foerster wire-crack test instrument (American designation, Magnatest 
FW-200) (a b, c) Compacting defects, (d) Surface lap. (e) Core splitting caused 
bv low swedging temperature or excessive steps in the dimensions of the swedgmg 
jaws (f) Cavities caused by incorrect swedging procedure, (g) Cavities caused In- 
melting channels, (h) Grinding cracks. 

42 7 

Institut Dr. Foerster 

Fig 5. Microsections of tungsten rods and wires corresponding to the schematic 

drawings of defects in Fig. 4. (a, b, c) Compacting defects, (d, e, f) Defects caused 

by too heavy swedging. (g) Defect originating from a central flaw, caused by too 

heavy swedging. (h) Internal melting channel. 

42 8 


Institut Dr. Foerster 

Fig. 6. Microsections from the tungsten and molybdenum industry, (a) Defective 
structure with holes, (b) Weak spot where a crack originates which appears during 
finishing of the wire in the melting furnace, (c) Cracks caused by compacting defects 

before sintering. 



tureri lead to splitting inside the rod, as shown schematically in Fig. 4(e). In 
extreme cases, cavities can appear according to Fig. 4(f.) Sometimes channels of 
local melting appear during final sintering and are deformed during swedging 
into a string of cavities, as in Fig. 4fg). The typical characteristic discontinuity 
caused by the swedging process, shown in Fig. 4(f), is a smooth (melted) inside 
surface of the cavity. 

Small netlike cracks sometimes appear on finished electrical contacts made 
from material which was found to be free of defects when the rod was tested with 
the wire-crack test instrument. These cracks can be detected on the finished con 
tacts by the Defectometer with probe coil. These defects are caused in the final 
finishing steps, either during cutting off the contact buttons or during grinding of 
their surfaces with insufficient cooling [see Fig. 4(h)]. 

Fig. 5 shows several microsections of tungsten rods and wires as proof of the 
defects shown schematically in Fig. 4. The natural defects in Figs. 5 (a), (b), and 
(c) correspond to the schematic drawings of Figs. 4(a), (b), and (c). Figs. 5(d), 
(e), and (f) illustrate defects caused by too heavy swedging; these correspond to 
Fig. 4(e). Fig. 5(g) shows a crack which proceeds radially from a flaw located in 
the core to the outside surface. The crack itself extends along the rod even though 
its cause is localized in the core. Fig. 5(h) shows a defect with smooth inside sur 
face which resulted from a melted channel and was caused during final sintering 
of the tungsten rod [compare with Fig. 4fg)]. Fig. 6 shows microsections of 
tungsten rods taken from a paper by Gromodka 53 , who has employed the Foerster 
wiie-crack test instrument in the production of tungsten and molybdenum wire. 
Fig. 6 (a) shows a defective structure with holes through which air can enter into 
the vacuum envelopes for which these tungsten wires are used as sealed-in lead 
wires. Fig. 6(b) shows a defect which does not yet appear as an actual material 
separation but which, however, will burst open in the melting furnace during 
finishing of the wire. Fig. 6(c) shows the usual splits which appear as a result of 
compacting defects before sintering. The preceding Figs. 3 to 6 are examples of 
how, by means of 100 percent nondestructive crack testing, defects are not only 
sorted out but also how, by means of a systematic evaluation of the defect 
phenomena, the defect causes can be discovered and thus be largely eliminated 
from the production process. 

The Sigmaflux eddy current instruments and the rod-crack test instrument have 
been used in various branches of the European metal industry for systematic 
defect analyses in other materials. Rods, wires, tubes, and shapes of materials such 
as copper, aluminum, light metal alloys, brass, bronze, German silver, and stain 
less steel have been tested with these instruments. Important shipments of 
semi-finished parts are frequently given 100 percent tests in the European metal 

Conductivity Measurements 

method for conductivity measurement of metals of all shapes was introduced to 
industry in 1939. Before the suppression of the lift-off effect of the test coil in 
the instrument indication, this method could be used for smooth metal surfaces 
only in rare cases. After the suppression of the lift-off effect in 1941, 44 45 the 
application range of the Sigma test instrument increased rapidly. By 1944 many 
hundreds of these probe coil instruments, which contained only one electronic tube, 
were used in industry. Today, the Siginatest instrument in its improved design is 
the most widelv used eddy current method in industrv. 



The Sigmatest probe coil instrument is used in industry for the following tests: 

1. Determination of the electrical conductivity of parts, made of copper, alu 
minum and their alloys, for the electrical industry and determination of the 
degree of purity of metals. 

2. Sorting of mixed metallic alloys. 

3. Hardness measurement on alloys whose hardm s> exhibits an unequivocal rela 
tionship with the electrical conductivity. 

PROBE COIL TEST PROCEDURE. For a conductivity KM the probe 
coil is placed on the part to be tested, as shown in FIG;. 7, which result.- in a deiif c- 
tion of the pointer on the meter. The large conductivity scale of the te-t inurn 
ment is calibrated in conductivity units, i.e., either in meter per uhm-mm.- or in 

Institut Dr. Foerster 

Fig. 7. Testing rods of mixed alloys by applying the probe coil of Foerster 
Sigmatest to the rod ends. 

percentage IACS (International Annealed Copper Standard). The control knob 
is adjusted until the pointer of the indicating instrument rests at the zero point 
in the center of the scale. Now the absolute value of the electrical conductivity 
can be read directly on the large conductivity scale. Thus the electrical conduc 
tivity can be measured within a few seconds. Only a flat surface on the part, of 
0.4-in. maximum diameter, is required to be accessible. 

In addition to determination of the electrical conductivity, the deflection method 
is used for the rapid sorting by conductivity (i.e., for hardness, purity, alloy). 
For this the pointer of the meter is adjusted to zero for the average value of the 



results in a zero deflection. By means of the sensitivity control, one can select the 
sensitivity so that a specific deflection (for example, ten scale divisions) is ob 
tained, in case of an electrical conductivity variation of 1, 5, or 10 percent IACS. 
Sorting can be carried out rapidly because one will observe only the deflection of 
the meter when the probe coil is placed on the part, which requires approximately 
1 sec. Thus it is possible to sort production parts according to conductivity with a 
speed of about 1500 to 3000 measurements per hour, obtaining the conductivity 
value for each individual part. 

If one selects a low sensitivity, the entire conductivity range can be contracted 
on the scale of the zero meter. This is of significance, for example, for a rapid n 
scrap sorting when alloys which cover a very large conductivity range must be 
separated from each other and sorted according to alloy groups. 

In hardness testing the sensitivity can be adjusted so that a positive or nega 
tive deflection of one scale division corresponds to a hardness variation of one 
Brinell unit, for example. 

CONTROL OF COPPER CASTINGS. The determination of the electrical 
conductivity is of special significance to the producers of copper or aluminum wire 
conductors. The minimum permissible conductivity of copper wires in the an 
nealed and hard-drawn conditions is specified by the electrical industry. To main 
tain these values, the copper must be free of additives. The great influence of 

ffm/flmm 2 

002 004 006 0,06 0,1 

Fig. 8. Decrease in conductivity of copper, caused by various impurities. 

small amounts of other metals on the conductivity of copper was investigated and 
has been reported in numerous papers. 58 Fig. 8 shows the decrease in conductivity 
of copper as a result of various impurities. The Sigmatest instrument saves time 
and money in testing the degree of purity of metals, since the conductivity can 
be determined on small solidified samples of the melt, while the melt itself is still 
in the refining furnace. 

Phosphorus Content. A specific phosphorus content is desirable in order 
to obtain a satisfactory pourability of copper for casting. Soft copper, before the 



addition of phosphorus, ha?-- a conductivity to'TwrT-n 51 and 55 meter- per ohm- 
mm. 2 , while the conductivity decreases to l*t\vrf n 40 and 41 merer-' per ohm-mm.- 
after a phosphorus addition of between 0.01 S TO 0.03S percent. Thfc ran } 
determined within a short time with the test coil instrument on >mill melt 
samples. The testing, including taking of the liquid sample, require,- only 2 to % 
min. Thus phosphorus can be added to the liquid melt until the denrf'il con 
ductivity is obtained. This method is not limited to copper casting only hut can 
also be used for alloys which contain a high amount of copper; for example, con 
ductive bronze with tin, silicon, or cadmium, and lead-bronze. 

Oxygen Content. Furthermore the probe coil instrument of Fig. 7 i* used for 
the purely operational determination of the oxygen content in oxidizing melting 
of copper, since the conductivity of copper decreases with increasing oxygen 
content. 9 After completion of the oxidizing melting period, the melter ha,- an 
approximate idea of the oxygen content, and he can measure the amount of redw 
ing agents to be added from the result of the conductivity measurement of the 

S m> 




Xlmm 2 














,4 0,3 0.2 QJ ai 02 0, 

Fig. 9. Variation of the conductivity during deoxidation of a copper meh.* 

sample drawn out from the melt. Fig. 9 shows the conductivity increase after 
deoxidation and the conductivity decrease following further phosphorus-copper 
additions. The conductivity values (ordinate of Fig. 9 } give indications of possible 
impurities, so that corrections can be made before casting. The progress of the 
melting-loss of the additives which are to be removed can be controlled by drawing 
samples continuously during the refining process. 

Electrical Conductivity. The Sigmatest instrument is also used for the testing 
of highly conductive copper rings for squirrel-cage induction motors. For these 
rings the customer requires a conduct ivity of 9S percent I ACS (56.8 meters per 
ohm-mm. 2 ). The fulfilment of the requirement can be determined nondestnic- 
tively within seconds on finished squirrel-cage motors. 

In British plants, copper blocks (melted in their own plants as well as in other 
plants) are tested for conductivity with the Sigmatest in order to eliminate those 



which do not meet the British specifications for copper conductivity . 51 It is of 
significance that the test speed of 100 blocks in 3 min. is limited only by the 
transport of the block sections. 

CONTROL OF ALUMINUM CASTINGS. The degree of purity of alu 
minum can be tested analogously. Fig. 10 shows the influence of metallic additives 
on the conductivity of aluminum. Even in aluminum foundries there exists a 
desire to test the cast block for its suitability for use in wires made according to 
the technical requirements for minimum conductivity. The Sigmatest permits the 
determination of the conductivity instantly on drawn samples of very small dimen 
sions. In this manner the conductivity can be determined on analysis samples 
drawn during casting. In order to produce a uniform reference structure, these 
materials are precipitation-hardened for 4 hr. at 350 C. 

ai 0.2 03 o.4 as 0.5 

Fig. 10. Influence of metallic additives on the conductivity of aluminum. 

Such conductivity measurements were carried out in a plant on a large number 
of aluminum samples drawn from the melts, and the conductivity values so 
obtained were compared with the conductivity of both pressed wires and hard- 
drawn wires. It was thus determined that the conductivity of the cast metal in 
the heterogeneous state was lower than that for pressed and precipitation- 
hardened wires from this block, by only 0.2 meter per ohm-mm. 2 maximum. By 
means of these results, one can draw conclusions from the measurements on cast 
analysis samples concerning the electrical conductivity of the wires produced 

Silicon Content. Impurities in aluminum often include iron and silicon. The 
conductivity of aluminum is influenced only slightly by iron because of its low 
solubility. 72 Silicon, however, which is soluble to 1.65 percent in aluminum at the 
eutectic temperature (578 C.), results in a large decrease in conductivity from 
pure aluminum. In the range above 0.15 percent silicon, an approximately linear 
relationship exists between concentration and conductivity. Therefore the' Sigma 
test is used in industry continuously for the testing of the degree of purity of 
aluminum. 83 

Determination of Segregation. Fig. 11 (a) shows the variation in conductivity 
as a result of segregation in a cast block of AlCuMg alloy. The test was carried 



out after the rough surface had been removed, in order to have a .sufficiently flat 
test surface for the Sigmatest coil. During scanning of the cross-section, a picture 
of the segregation zone was obtained. A subsequent chemical analysis showed that 
the conductivity curve was influenced essentially by the copper segregation, Fijr. 
11 (b). In any case, the Sigmatest instrument is a simple means of determining the 
extent of segregation by scanning a block only once. 





! 1 





\ \ 
I \ 

15 K) 5 5 10 15 


Fig. 11. Conductivity variation caused by segregation in a cast block of AlCuMg 
alloy, (a) Variation in conductivity due to segregation, (b) Percentage* of roppor 


TESTING OF MIXED ALLOYS. Various alloys can be separated reliably 
with the Sigmatest when the mixed materials exhibit differences in conductivities. 
The conductivities of numerous alloys are given in Fig. 12. For each metal, and 
especially for each alloy, a specific spread range exists which is caused by im 
purities in pure metals; in alloys, by the permissible variation in the percentage 
content of the alloy elements. 

Because of the large number of different alloys, the same conductivity values 
sometimes appear for different materials. In such cases' other characteristics can 
be used for separation. In scrap sorting, the color, tyjH? of finish, and appearance 
of fractured pieces are additional characteristics. In doubtful cases, the sj)c j cifie 
gravity and the corrosion test, or the hardness, give an indication. For heavy 
metals, only special brass and bronzes may exhibit the same conductivity, while 
there are several overlaps for light metal alloys. 

Austenitic Stainless Steels. Recently, Sigmatest instruments have been in 
creasingly used for the sorting of austenitic steels (stainless steels). For this pur 
pose a Sigmatest was developed with a conductivity range of from 0.5 to H meters 
per ohm-mm.-. The absolute value of the conductivity can be determined for 
stainless steel onlv when the material i^ not magnetic; i.e., when the permeability 







equals one or very nearly one. As was discussed in the theory of the probe coil 
method, the product of conductivity and permeability plays a decisive role. There 
fore an exact measurement of the conductivity at high permeability values is no 
longer possible. However, since the absolute value of the conductivity is of little 
interest in sorting, the method has become increasingly popular, especially since 
it separates the strong magnetic alloys from each other very sensitively. 

A large American plant uses this instrument to separate various types of high- 
temperature and austenitic steels from each other. Testing of mixed lots is 
carried out only in rare cases with the absolute method, i.e., by reading the actual 
conductivity value on the calibrated scale. Most testing of mixed materials is 
done with the much faster relative method, by observing the deflections on the 
zero meter. In these tests the sensitivity is decreased so that the measurement 
values of all interesting materials are still on the scale of the instrument. For 
low- conductivity materials such as graphite or carbon, a special Sigmatest instru 
ment has been developed with a correspondingly low conductivity range. 

hardened aluminum alloys has, according to theoretical principles, an unequivocal 
relationship to the hardness. 66 Thus indirect hardness measurement with the 
Sigmatest has become a widely used method. For the age-hardenable alloys of 
the types AlCuMg, AlMgSi, and AlMgZn in the technically interesting range of 
the cold age-hardenable state, there is a proportionality between hardness and 
conductivity. Consequently mechanical hardness tests can be replaced by a con- 


13 20 22 




Fig. 13. Relationship between conductivity and hardness in cold age-hardened 


ductivity measurement. Vosskuehler 84 has clarified these relationships with the 
Sigmatest for the seawater-proof AlMg alloys. Fig. 13 shows how the conductivity 
of AlMgSi, AlCuMg, and AlMgZn alloys decreases with increasing hardness. The 
hardness measurement with the Sigmatest instrument has gained in significance 
especially in the testing of light -metal pressed parts. The test speed is many times 
higher than that of the Brinell test, even when more modern rapid measurement 
methods are used. In one metal plant, thousands of man-hours wore saved by the 
use of only one Sigmatest instrument. 



Hardness testing is carried out with the Sigmatest instrument exclusively by 
the relative method. The tolerance range is adjusted on the zero meter of the 
instrument by means of two test parts whose hardness has been mechanically 
measured. The calibration is carried out for each alloy by means of two test parts 
with various hardnesses of each alloy to be tested. 

Control of Heat Treating. Indirect hardness measurement with the eddy 
current instrument is used also with overheated parts, or parts quenched at 
high temperature. These can be eliminated as a result of their low conductivity. 
Fig. 14 shows how the conductivity of thin sheets steadily decreases with increasing 

490 500 510 520 530 

Fig. 14. Conductivity and hardness of sheets of an AlCuMg alloy. 

quenching temperature, whereas the Brinell hardness (as well as the tensile 
strength and the yield point) peaks at some temperature and decreases for higher 
quenching temperature, so that an unequivocal relationship between the hardness 
and structure condition is no longer possible. Furthermore one can see in Fig. 14 
that the Brinell hardness values spread considerably more than the conductivity 
values on the Sigmatest instrument. The spread range is indicated by the shaded 
area around each curve. The conductivity value measured over a specific area is 
more representative of the condition of the alloy than the hardness value obtained 
on a very small area. With Sigmatest testing, one can maintain the range of the 
specified quenching temperature (505 C. to 520 C.) and the specified hardness 
(125 to 145 kg. per mm. 2 ) much more accurately than by measuring the 
Brinell hardness. 

Quench-hardened Parts. The Sigmatest is also used for controlling the uni 
formity of hardness on semi-finished parts (rods, sheets, extruded shapes, and 



tubes). When numerous semi-finished parts are quenched simultaneously, so- 
called soft areas appear occasionally if the parts are not submerged uniformly 
in the bath. The same defect can appear on extruded shapes which are quenched 
immediately after extrusion. For the separation of the extruded parts, the press- 
feed is turned off, and a heterogeneous area forms between the quenched portion 
and the portion remaining still hot. Also, an area with strongly varying hardness, 
as well as formation of steam bubbles, can appear because of a variable water 

For testing of extruded shapes, a search coil is moved along the extrusion. 
Testing the entire length of 15 meters requires less time than a single Brinell im 
pression. A special probe coil was developed to detect soft spots quickly on large 
metal surfaces such as sheets. Fig. 15 (a) shows the diagram which results when 
the probe coil moves over the sheet, where no special steps were taken for the 



































Fig. 15. Indications of a soft spot in sheet, (a) Without suppression of the lift-off 
effect. The periodic deflections are caused by irregular movement of the wheels of 
Hi- traveling coil, (b) With suppression of the lift-off effect. 


suppression of the lift-off effect. On the other hand, Fig. 15 (b) shows how the 
large deflections are suppressed by the lift-off effect, so that only the alloy condi 
tions such as structure or soft spots are indicated. 

Titanium Alloys. Recently the Sigmatest instrument has proved successful in 
the hardness testing of titanium alloys, since an unequivocal relationship also 
exists between the electrical conductivity and the Brinell hardness. 

Hysteresis-Loop Tests 


magnetic behavior of a ferromagnetic material is characterized by its magnetiza 
tion curve. It is of significance that the various factors of the magnetization loop, 
such as saturation, remanence, coercive force, and permeability, are influenced in 
a different manner by the material properties such as alloy composition, structure, 
and internal stress. 

The magnetic saturation is mainly a function of the chemical composition and 
the crystal structure, while the coercive force, the most important magnetic 
characteristic factor in eddy current testing of materials, is influenced by the 
following material properties: 

1. Internal stress and magnetostriction. 

2. Number, magnitude, and distribution (degree of dispersion) of foreign bodies 
embedded in or between ferromagnetic crystals. 

3. Crystal energy and size of the ferromagnetic crystals. 

The magnetostriction, crystal energy, saturation magnetization, and Curie 
point are independent of the structural condition; i.e., these factors are among the 
structure-insensitive factors. However, the coercive force, the permeability, and 
the Rayleigh constant are influenced by internal stress and the structural condi 
tion. Therefore, these factors belong to the structure-sensitive magnetic char 
acteristic factors. The magnetic behavior depends to a larger degree on the 
selection of the initial materials, and the melting, foundry, rolling, and annealing 
processes, than any other characteristic. 

Correlation Problems. From the many effects which influence the magnetic 
characteristic values, those which are of interest (for example, hardness or carbon 
content) have to be filtered out during testing by means of suitable devices. The 
problem of nondestructive testing with eddy current methods consists in drawing 
conclusions concerning properties, such as hardness, tensile strength, and the alloy 
content, from the magnetic and electrical values of the test body. Thus, it is 
necessary to investigate the following questions: 

1. Which relationships exist between the mechanical properties (hardness, tensile 
strength, etc.), the metallurgical properties (for example, carbon content or 
structure), and the magnetic properties of the material? 

2. How are the magnetic material values indicated in the eddy current sorting 

3. How can test conditions be created so that specific desired properties have an 
optimum indication, while undesired properties are suppressed? 


systematic investigation of the relationships of metallurgical and technological 
material properties, a method has been developed for the representation without 
distortion of the hysteresis loop of steel samples which one can easily produce 
from standard semi-finished parts. As is well known, only the infinitely long 



sample or the ring sample without magnetic poles permits representation of the 
hysteresis loop without distortion. After many experiments, the tube-ring 
sample shown in Pig. 16 was used as a standard sample (normal sample) for 
magnetic investigations. The sample of Fig. 16 can be produced easily on a lathe 


1=50 mm - 

Fig. 16. Standard test object for investigating the ferromagnetic properties of a 
material on the Foerster Ferrograph. 

from a short piece of bar stock. For magnetizing, this sample is placed on a con 
ducting rod (Fig. 17) which in turn is placed in a high-current fixture and 
through which currents flow with amplitude adjustable up to 5000 amp. at 60 
c.p.s. Thus a circular magnetization without free poles is produced in the sample. 
The sample contains a winding for measurement of the induction. At the same 
time a ring coil is placed over the high-current rod for the measurement of the 
field strength. 
The entire distortion-free magnetization loop appears on the screen of the 

Institut Dr. Foerster 

Fig. 17. Device for magnetizing standard ring samples with axial current and 
measurement of the field strength with the air ring coil. 



Ferrograph (Fig. 18). The Ferrograph was developed especially for magnetic 
material investigations. 22 It possesses ten sensitivity steps on the B and H chan 
nels. For each sensitivity step, the calibration factor appears for the B channel, 

Institut Dr. Foerster 
Fig. 18. Screen picture of the Ferrograph with calibration of B and H scales. 

o-auss per centimeter in the vertical, and for the H channel, oersted per centimeter 
in the horizontal. Thus, as shown in Fig. IS, all interesting data necessary ior the 
evaluation appear in the photograph of the screen picture. 

hv^tere^ loop at c.p.s. can be obtained from the 60-c.p.s. dynamic loop in the 
following manner: Two ring samples, one right next TO the other, are cut in>m 
the .*ame rod The -dimensions of .the standard sample (diameter D M , length 1, 
and "wall thickness Tf n ) .aje .takeji according .to Fig. 16, while tor. the -second 



sample the following dimensions are selected: L = L n /-\/2:W W tl \/2. The 
portion of the area of the hysteresis loop which results from eddy currents (the 
dynamic component) increases, according to the theory of eddy currents, with the 
square of the wall thickness W. Thus the sample whose wall thickness is larger 
by \/2 results in a doubling of the eddy current component of the hysteresis loop 
with reference to the standard sample. Fig. 19 shows superimposed screen picture 
of the standard sample and the sample having a \/2 times larger wall thickness. 

Institut Dr. Foerster 

Fig. 19. Superimposed hysteresis loop of standard ring sample of Fig. 16 and 
sample from the same material but with a VFlarger wall thickness. 

Correction for Eddy Current Effects. The portion of the area of the 
hysteresis loop due to eddy currents in the sample, with the yT times larger wall 
thickness, is twice as large as that for the standard sample. Thus the difference 


between the two hysteresis areas, i.e., the area between the inside and the outside 
loop, represents the portion of the area due to eddy currents in the standard 
sample (the inside loop). 
The area A n of the hysteresis loop of the standard sample is given by 

where H L is the hysteresis loss of the static loop at / = c.p.s. and E L is the eddy 
current loss. For the sample with the \/2 times greater wall thickness, a doubling 
of the portion of the area of the hysteresis loop results, due to eddy currents. 
Thus, the area of the hysteresis loop of this sample results in 

A = H L 4- 2E L (2) 

By subtracting Eq. (1) from Eq. (2), Eq. (3) is obtained: 

A-A* = EL (3) 

i.e., the strip between the inside loop and the outside loop represents the eddy 
current component of the area of the inside loop. Thus, if the strip between 
the inside and outside loop is subtracted from the inside loop, so that the hori 
zontal distance of the outside loop from the inside loop is always decreased toward 
the inside, the static magnetization loop at / = 0c.p.s. is obtained. Thus the 
influence of the electrical conductivity of the material on the hysteresis loop can 
be determined separately. 

standard sample, the effect of the various influential factors of the hysteresis loop 
is illustrated by a large number of screen pictures. In the subsequent text the 
influence of the carbon content, the hardening process, and the internal stresses 
on the magnetization loop of steel will be explained by means of a qualitative 
description. At the same time the field strengths which permit an especially good 
separation of the individual material properties will be indicated. 

Effects of Internal Stress. Fig. 20 shows a family of hysteresis loops for a 
standard sample of cold-drawn C-1015 carbon steel, while Fig. 21 illustrates the 
family of hysteresis loops after annealing to remove internal mechanical stresses. 
In addition to the decrease of the coercive force, it is clear that the greatest 
variations in the degree of magnetization of the sample, with and without stresses, 
appear at low field strengths. This behavior is typical, so that stresses (for 
example, caused by the straightening process) always result in an especially large 
influence on the magnetization values at low field strengths; i.e., below one-half 
the coercive force. At a field strength corresponding to % horizontal unit, 
the ratio of the magnetization of the stress-free sample to that of a sample with 
internal stresses is 5:1. Therefore, in the quality sorting of materials which also 
exhibit inhomogeneous stresses fas caused by straightening), it is necessary to 
work at field strengths which are higher than the coercive force. 

Effects of Annealing. Fig. 22 shows the loops, photographed one on top of the 
other, of the standard sample of C1015 in the following conditions: 

1. Annealed in C-atmosphere at 900 C. for 1 hr. and slowly cooled (upper curve). 

2. Annealed in C-atmosphere at 900 C. for 1 hr. and quenched. 

The recarburized and slowly-cooled sample already shows a large increase in the 
coercive force, as compared with Fig, 21. For the quenched sample, a further 
increase in coercive force is observed, plus the great permeability decrease. 

Institut Dr. Foerster 

Fig. 20. Family of hysteresis loops of a normal sample of C1015 cold-drawn 

carbon steel. 

Institut Dr. Foerster 

Fig. 21. Family of hysteresis loops of the normal sample from Fig. 20 but stress- 
relief annealed. 


Institut Dr. Foerster 

Fig. 22. Superimposed loops of the sample from Fig. 21, but carburized for 1 hr. 

at 900 C. in a C-atmosphere (loop with higher magnetization values) and 

quenched (loop with smaller magnetization values and higher coercive force). 

Institut Dr. Foerster 

Fig. 23. Superimposed magnetization loops of two normal samples with various 

hardness. The samples were carburized 1 and 3 hr. at 900 C. and then quenched. 

The field strength was selected so that the hardness variations were represented 

optimally by means of the magnetization amplitude. 




Effects of Carburization. Fig. 23 shows the standard sample at the field 
strength for optimum separation of hardness variations in the following con 
ditions : 

1. Carburized at 900 C. for 1 hr. and quenched (upper curve). 

2. Carburized at 900 C. for 3 hr. and quenched. 

The lower curve shows a greater hardness because of the higher carbon concen 

Predicting Mechanical and Structural Characteristics. The preceding four 
examples illustrate how cold-forming, variation of the carbon content, and hard 
ness variations result in a considerable variation in magnetic properties. These 
great influences on the magnetic properties caused by mechanical and metal 
lurgical effects meet the necessary requirements for eddy current quality sorting. 
This eddy current quality control is useful because it is possible to draw conclu 
sions concerning the mechanical and metallurgical properties of the material from 
the variations of the electromagnetic properties. 

Internal Stress in Magneto strictive Materials. Internal mechanical stresses 
have a pronounced influence on the magnetization loop. This influence becomes 

Institut Dr. Foerster 

Fig. 24. Influence of tensile stress on the magnetization loop of a hard nickel 
wire. The loop with the highest magnetization value corresponds to the wire without 
tensile stress. The loops shown below correspond to a tensile stress of 2, 4, 6, and 

10 kg. per mm. 2 . 


larger with higher values of magnetostriction. Fig. 24 shows the effect of tensile 
stress on the magnetization loops of a hard nickel wire which is loaded to 0, 2, 4, 
6, and 10 kg. per mm. 2 . Under tension the magnetization values of nickel decrease 
greatly (negative magnetostriction). The investigation of the variation of the 
magnetic values as a result of tension has led to a quantitative method of 
measuring the internal stresses. The build-up of internal stresses has been 
observed quantitatively by means of the stretch and drawing processes, and the 
decrease of internal stresses by means of the annealing process. 4 ' 49 

Comparator Bridge Tests 

PRINCIPLE OF OPERATION. The use of the hysteresis difference loop 
for nondestructive quality testing of steel was suggested in 1939 , 22 Today, this 
difference method is used in various comparator bridges. However, it is evident 
that the hysteresis loop, or the difference loop, is not an optimum representation 
of the magnetic material values for the purpose of quality sorting. In the sub 
sequent test the Magnatest Q picture will be discussed as an improved representa 
tion of magnetic and electrical material characteristic values. 

Fig. 25 (a) shows the two magnetization loops of the carbon steels C1015 
(inside loop) and C1034 (outside loop), photographically superimposed. The 
vertical magnetization M is measured by the integration of the measurement 
voltage E at the secondary difference coil. 

M = lEdt (4) 

is plotted above the field strength H and is indicated in Fig. 25 by }(H) . 

Fig. 25 (a) was obtained by inserting two steel samples (C1015 and C1034) in 
the secondary coil S l} one after the other, as indicated in the sketch of the 
secondary coil. In Fig. 25 (b) the rod of the steel C1015 is inserted in Si, and 
after the sample is removed from Si, the steel sample C1034 is inserted in S 2 . 
This results in opposite directions for the two hysteresis loops. Fig. 25fc) shows 
the difference of the two loops of Fig. 25 (b), obtained by placing the two steel 
samples in 5 t and S 2 at the same time. Increasing the amplification by a factor of 
4 results in Fig. 25 (d). Figs. 25 (c) and (d) represent an important characteristic 
step of the eddy current sorting method: By using two equal or similar steel 
samples in the two secondary coils, only a slight difference appears; this is then 
amplified to attain the desired resolution. 

A time-dependent representation of the magnetization difference is selected 
instead of the field-dependent representation in Fig. 25(e). Thus the repre 
sentation becomes unequivocal because only one magnetization value belongs to 
each abscissa value, in contrast to the previous pictures. In Fig. 25 (f), the time 
derivative of the magnetization difference is represented, instead of the time- 
dependent magnetization, by means of elimination of the integration. 

Figs. 25 (e) and (f) describe entirely different properties of the magnetization 
loop* Fig. 25 (e) represents the difference of the magnetization value of the two 
test bodies placed in S l and S 2 as a function of time, while Fig. 25 (f) represents 

f^l-^ = /(t) (5) 

at at 

This can be expressed as follows: 

dMi (dH\ dM* (dH\ _ . (6) 

~dH\~dt) dH \dt) ;u; 




Fig. 25. Magnates! Q pictures derived from the difference in the magnetization 
loops, (a) Superimposed magnetization loops of C1015 and C1034 carbon steels. 
(b) Sample C1015 in coil 1, and after removal of sample C1015, sample C1034 in 
coil 2. (c) Difference of magnetization loops of the C1015 and C1034 samples as a 

The two expressions dM^/dH and cBL 2 /dH represent the differential per 
meability of the magnetization loops. Thus Fig. 25 (f) represents the difference 
of the permeability value of the two test bodies Si and &, multiplied by the factor 
dH/dt. By multiplying by dH/dt, the difference of the" permeability" of the two 
samples is distorted according to a cosine curve. At small angles, the cosine is 
nearly unity, so that the permeability difference is represented undistorted. 
However, the distortion of the permeability difference according to a cosine curve 
is meaningless in quality sorting. 

If the curve of Fig. 25 (f) is electrically differentiated with respect to time, so 

then essentially only the variations in the curvature (second derivative) of the 
magnetization loops of the two samples in S I and S 2 are represented, merely 
multiplied by the factor d 2 H/dt 2 . 



4 AM'f(H) 


65 AM*f(t) 



f (t) 


Iristitut Dr. Foerster 

function of the field strength, (d) Same as in (c) but at a four times greater ampli 

fication of the magnetization difference, (e) Unequivocal representation of the 

magnetization difference from (d) above a linear time scale, (f) Time differential 

quotient of the magnetization difference from (e) above a linear time scale. 

MAGNATEST Q SORTING. With Magnatest Q curves, either the mag 
netization variations, the permeability variations, or the variations of the 
curvature of the hysteresis loop of the test body with reference to a standard are 
indicated depending on the selection of the representation. 

The influence of the test field strength must also be considered. Test bodies 
with homogeneous mechanical deformation can be tested at low field strengths. 
Because of the low initial permeability, the penetration depth is then greater so 
that edge decarburization or soft spots have little influence. Large permeability 
variations can be evident for mechanically deformed parts, such as rods which 
have passed through a straightening machine. Therefore it is practical to use a 
high field strength with integrated measurement voltage (magnetization difter- 

n addition to the field strength and the type of representation, the frequency 
can also be adapted to the test problem. For many problems, a frequency o 
5 c p s is superior to 60 c.p.s. It provides at the same tune a flexible theoretical 


interpretation of the frequency influence on the analysis possibility of the eddy 
current quality testing. 

Recording Test Results. Considerable numbers of test results were gathered 
in recent years by distributing among many hundreds of users the printed trans 
parent paper charts for the screen of the Magnatest Q instrument. Permanent 
records of the test results can be obtained quickly with the aid of these trans- 

Institut Dr. Foerster 

Fig. 26. Magnatest Q picture of cobalt alloys having similar alloy contents. One 

alloy is identified in the picture by a grid. 

parent paper charts by tracing screen patterns with various colors, so that the 
suitability of the test method for a specific test problem can be observed at a 

Another way of recording test results in easily observable manner is by photo 
graphing the various screen patterns. For example, in investigating how two 
alloys, A and B, can be separated in the Magnatest spread band, a transparent 
disc with a grid is placed in front of the screen while the curves of the alloy A 


are taken, so that the curves in the photograph are interrupted in a characteristic 
pattern. With the aid of discs with various grids, one can easily differentiate 
between the curves of the various alloys. Fig. 26 shows the curves of two cobalt 
alloys which are very similar in their alloy contents and which are clearly distin 
guished from each other through the grid pattern present in one set of curves. 

Range of Test Applications. The Magnatest Q pictures given in the subse 
quent discussions represent a small selection from large numbers of test results 
obtained from industry. The following test problems are treated: 

1. Sorting of rod material according to alloy. 

2. Sorting of rolled billets according to alloy. 

3. Sorting of small parts according to alloy and condition. 

4. Sorting for heat treatment and structural conditions, and thus sorting accord 
ing to hardness, tensile strength, etc. 

5. Sorting for depth of case and edge decarburization. 

6. Sorting with special coils. 

schematically the optimum adjustment of the test instrument. Suppose that two 
alloys, A and B, are to be separated from each other. First, a sample of the 
alloy A is inserted in the test coil. The pattern of Fig. 27 (a) will appear on the 
screen. The sensitivity is selected so low that the entire curve of AI of sample 
appears on the screen. 

Sensitivity Calibration. The maximum of this curve AI is brought, by means 
of an adjustment on the instrument, to a specific indication height called the 
absolute value. In this way, for all sensitivity positions of the sensitivity knob, 
the sensitivity is indicated in percentage of this absolute value. This makes it 
possible, independent of the specific instrument, the specific coil, and the specific 
material, to relate the information to the absolute value of the material. Thus 
results obtained at various locations can be easily compared with each other. 

Compensation Adjustments. Next, a sample of the alloy A is inserted in each 
of the two difference coils. The solid curve AI, appears on the screen at a higher 
sensitivity [see Fig. 27 (b)]. If sample A^ is now replaced by a sample B l of the 
alloy B, then the interrupted curve B l appear* as in Fig. 27lbL Now, after the 
sample B is again replaced by the sample AI, the compensator is? adjusted so 
that curve A l becomes a horizontal straight line, as shown in Fig. 27(c). In the 
next step, Fig. 27 (d), the curve B l is displaced by operating the phase shifter, so 
that the maximum lies in the center of the screen. In Fig. 27 (e) the curve AI is 
symmetrically displaced toward the bottom, with reference to the curve B^ by 
means of the compensator. Now, in Fig. 27(f), the amplifier sensitivity is in 
creased so that the screen is fully utilized. 

Spread Band Characteristics. Fig. 28 shows curves for rods of C1015 carbon 
steel and for free-cutting steel. The spread band around the two curves is 
plotted from the results when 1000 rods of each material are fed through the coil. 
Thus the spread band gives the variations which can result from segregation 
effects, internal stresses, inhomogeneous deformations, and edge decarburization 
effects within one heat. 

ALLOYS. The main problem of the Magnatest Q testing is to select the follow 
ing test conditions: sensitivity, type of indication, frequency, field strength, and 
phase position, as well as the compensation adjustment so that no overlap appears 








Fig. 27. Adjustment of the Magnatest Q screen picture for optimum test condi 
tions, (a) Adjustment of the absolute value, (b) Curve A-L and curve E^ (inter 
rupted) at a sensitivity eight times higher than in (a), (c) Curve Ai adjusted to a 
horizontal line by means of the compensators, (d) Maximum of curve Bi (inter 
ior alloys which vary only slightly in their contents. The optimum adjustment 
was selected for Fig. 28 so that the spread bands are symmetric. The largest 
variations between the two spread bands are on the vertical center line of the 
screen (reading slit of the Magnatest Q instrument). 

Separation of Three Different Alloys. Fig. 29 shows the optimum representa 
tion of the separation possibility of three different alloys. The alloy spread bands 
are moved so that the maximum separation between bands is at the" vertical center 
line of the screen, and by means of the compensation controls, the curves are 
moved so that the screen is fully utilized. Fig. 29 shows the optimum adjustment 
for the three alloys 3415, 5115, and free-cutting steel with 0.22 percent' C. The 
Magnatest Q picture in Fig. 30(a) shows the three alloys C1034, C1045, and 
5140. The spread bands of the alloys C1034 and Cl045*are very close at the 
point of the best possible separation, since both alloys are close with reference 
to their carbon content. 

Use of Harmonics in Separating Similar Alloys. One can see, even in Fig. 
30(a), that related alloys have similar characteristic curves. The alloy 5140 dis 
tinguishes itself in a characteristic manner from the curves of the' two other 
carbon steels. Fig. 30(b) ; shows the separation possibilities 'for- the alloys 1015 
and 5115.. The curves, wiiich-were obtained on unfinished-- piston rods'," --clearly 






















Institut Dr. Foerster 

rupted line) moved to the screen center with the aid of the phase shifter, (e) Ai and 
Bi symmetrically displaced toward the top and the bottom by means of the com 
pensators, (f) Four times increase of sensitivity, providing full utilization of screen 
picture with optimum separation in the screen center. 

show that the two alloys distinguish themselves essentially in the harmonics, 
while the amplitude of the fundamental wave is almost equal. This is a typical 
case in which a better separation possibility for the alloys is given by favoring the 

Sorting Mixed Lots. In general, the problem of sorting mixed alloys is to 
detect, in a mass of test parts of one alloy, a part of the mixed alloy. Occasionally, 
however, it happens that a series of mix-ups occurs at the same time, and these 
have to be sorted with the Magnatest Q into their correct alloy groups. Fig. 31 
is an example for the separation of four different alloys, C1034, C1045, C1075, 
and free-cutting steel with 0.1 percent C. It is shown that the separation in the 
center vertical "(reading slit) is incomplete because an overlapping between the 
alloys C1045 and C1075 is evident. If, however, the Magnatest Q picture is 
adjusted by means of the phase shifter so that the dashed line is in the screen 
center, then all four alloys are satisfactorily separated from each other. 

TESTING BILLET MATERIALS. The mixed material previously dis 
cussed pertained to semi-finished rods. In the subsequent text, mix-ups of rolled 
billet material with square cross-sections of 60 X 60mm. are discussed. The hot- 
rolled, semi-finished parts are generally considered homogeneous and uniform 
with reference to their structure. Thus, difficulties which appear because of the 

Fig. 28. Magnatest Q curve and spread band of C1015 carbon steel and a free- 
cutting steel. 

Fig. 29. Magnatest Q picture of three alloys, 3415, 5115, and free-cutting steel 
with 0.22 C, with an optimum screen utilization. 



Fig. 30. (a) Spread bands of a large number of rods of C1034, C1045, and 5140 

steels, (b) Spread bands of unfinished piston rods of C1015 and 5115 steels. A 

better separation is obtained by favoring the harmonics. 




Fig. 31. Separation of C1034, C1045, and C1075 carbon steels and free-cutting steel 
with 0.1 percent C. Separation of all four alloys is possible at the dashed vertical 


permeability variations due to cold working are not to be expected. Fig. 32 
shows the Magnatest Q spread bands of the alloys 5132, C1075, 5140, and 6150. 
As can be seen, the four alloys can be satisfactorily separated with the selected 
optimum adjustment which was carried out according to the schematic shown 
previously. All spread bands are separated far from each other, and the spread 
ranges are relatively small because of the homogeneous condition. It is of 
significance that the two chromium steels, 5132 and 5140, which are very similar 
in their alloy content, can be separated from each other without difficulty. A 
billet, of the C1075 steel is used as a compensation sample. 

Fig. 33 shows the test results for the four steels C1015, C1034, E52 100, and a 
wrought heat-resisting steel, which can be satisfactorily separated. For optimum 
adjustment, the spread band of the compensation sample is moved far enough to 
the top so that the screen is fully utilized. 

Use of Low Field Strengths. Fig. 34 shows the spread bands of two spring 
steel alloys, A (55 Si 7} and B (65 Si 7) which lie close together in their alloy 
content (see analysis). In this figure the two alloys are illustrated in two different 
heats each, so that relatively wide spread bands are obtained. However these can 
be clearly distinguished, especially if worked at very low field strengths, as in 
Fig. 3o. In Fig, 35 the field strength is only one-eighth of that of the preceding 

The influence of the field strength on the possibility of separation is of special 
importance in complicated cases. A very low field strength acts like a lower fre- 



Analysis A 

































Fig. 32. Separation of hot-rolled billets of 5132, C107S, 5140, and 6150 steels. 

quencv since at low field strengths the material appears only in it? very low initial 
permeability. Thus, according to the theory of eddy currents, the /-/,, value 
becomes smaller. However, at small /'/, values, the separation between con- 
ductivitv, permeability, and diameter effects is consideraby better than at higher 
/// values! In the range of small ///, values, the conductivity direction is almorf 
perpendicular to the diameter and permeability direction on the impedance plane 
P ffince the billets are not very exact in their dimensions, ,t is pract.ca to select 
the field strength so that the ///, value is kept small by means of the o ^ initial 
permeability. In this wav the conductivity effects, which are essentialh re.pon- 
sSTfor an alloy separation, can be separated satisfactorily trom permeability 
and diameter effects. 

Optimum Utilization of Screen Area. Fig. 36 shows another example with 
the steels C1045, 5140, and 3115. The C1045 steel serve, as a compensation 



Analysis B 























E52 100 







Fig. 33. Spread bands of hot-rolled billets of 60 X 60 mm. cross-section, of E52 
100 steel, a wrought heat-resisting steel for valves, and C1034 and C1015 carbon 


sample. Again the curve of this compensation sample is moved to the bottom 
until the three spread bands are distributed nearly symmetrically over the screen 
of the Magnatest Q instrument, so that an optimum utilization of the screen is 
assured. If one were to adjust the spread band of the C1045 steel in the hori 
zontal, then the spread band of the alloy 5140 would fall outside the screen. 
Fig. 37 shows the separation of two very close alloys, E4132H and 5132, which are 
distinguished from each other by only 0.23 percent Mo. 

INFLUENCE OF TEST FREQUENCY. For a number of applications, a 
frequency of 5 c.p.s. gives better results than a higher frequency in Magnatest 
Q tests. Fig. 38 shows that the separation of these similar alloys is not possible 
with 60 c.p.s., since no range is available at which the spread bands of these alloys 



Analysis C 










Fig. 34. Spread bands of rolled billets of the two spring steel alloys A and B 

(see analysis). 

are distinguishable from each other. Wherever one alloy distinguishes itself from 
the other, still other alloys fall together at the same phase point. However, 
Fig. 39 shows the same alloys sorted at 5 c.pjs. Here, it can be clearly seen that 
the better resolution is at the lower test frequency, which is especially obvious in 
this difficult borderline case. At the location of the dashed line, all five alloys can 
be reliably separated from each other. 

Tests of Wire Samples. The following example shows an especially interesting 
case: In the continuous testing of mixed wires, one piece was cut from each 
roll, then straightened by hammering, and tested with the Magnatest Q for 
mix-up. Fig. 40 shows the three steels, C1045, C1034, and 5140, tested with a 
frequency of 60 c.p.s. It is clearly seen that there is no point at which the three 
alloys can be separated from each other. The spread bands of the individual 
alloys always overlap in certain phase ranges. However, when tested with 5 c.p.s.. 
as in Fig. 41, a range appears in which a satisfactory separation of the spread 
bands of the three steels is passible. 



Fig. 35. Spread bands of the alloys of Fig. 34, but with an eight times lower field 


Advantages of Low Frequency Ratios. It is of interest to note why, in the 
aforementioned cases of similar alloys and of alloys in which internal stresses were 
induced by means of straightening, a separation can be carried out considerably 
better with 5 c.p.s. than with 60 c.p.s. 

Fig. 42 represents the complex voltage plane of the Magnatest Q coils con 
taining a sample with a specific conductivity, specific diameter, and specific per 
meability. From the formula given in Fig. 42, one can calculate the ratio of the 
measurement frequency / to the limit frequency f ff . It can be seen that at small 
f/fg values (/// < 10 1, the conductivity direction is almost perpendicular to the 
diameter and permeability direction. When permeability and conductivity effects 
are to be separated from each other, one must work at small f/f g values. The 
larger the f/f g value, the smaller the angle between the conductivity, the perme 
ability, and the diameter direction. At higher /// values (for example, above 10), 
a separation of the effects becomes more and more difficult. 

Fig. 43 shows the angle between the direction of diameter and permeability 
changes, and the conductivity direction, as a function of the ratio ///,. The more 
nearly the angle between a and i\.i wl ,,d] approaches 90 deg., the better the 
analysis possibility. In the vicinity of 180-deg. angles between <j and (\L Ki ,d), 
the influence of internal stresses can no longer be separated from conductivity 

Permeability Effects at High Frequency Ratios. At high /// values, the 
effect of conductivity variations on the Magnatest Q indication can be canceled 
by means of corresponding permeability variations because. both effects have 




























Fig. 36. Spread bands of 5140, C1045, and 3115 steels with optimum screen 


approximately the same direction. However, at small / / values !///<, < 101, an 
unequivocal separation of the effects is possible. Fig. oS illustrates a difficult case 
in which similar alloys cannot be separated at 60 c.p.s. but can be separated at 
5 c.p.s. This results from the fact that the various alloys are distinguished from 
each other in their electrical conductivity. However, at 60 c.p.s. the conductivity 
effects are overlapped by simultaneous permeability effects, so that a separation 
is no longer possible. On the contrary, at 5 c.p.s. the f f a value of the sample is 
lowered so that the conductivity forms a large angle with the permeability direc 
tion. Here the spread bands of the individual alloys can be separated in a wide 
phase range. 

Effects of Inhomogeneous Internal Stresses. Fig. 40 showed test results on 
wire ends cut from wire rolls and straightened by means of hammering. This 
resulted in a strong inhomogeneous deformation. Heavy internal stresses were 
present, causing relatively large permeability variations. These permeability 

















Fig. 37. Spread bands of the two alloys E4132H and 5132, which differ by only 

0.23 percent Mo. 

variations prevented a separation of the three alloys at 60 c.p.s., since these 
permeability variations are very nearly in the conductivity direction at 60 c.p.s. 
Under these conditions the conductivity effects, in which the three alloys are 
easily distinguishable from each other, cannot be separated from the permeability 
variations resulting from internal stresses. At 5 c.p.s., however, the conductivity 
direction is different from the permeability direction. Therefore a phase range is 
available in which the three alloys can be satisfactorily separated from each other 
because of the conductivity variations. These two examples are typical of a 
number of applications of the 5 c.p.s. Magnatest Q. It is significant that the 
reasons for a better analysis possibility with the low frequency are given by the 
theory of eddy currents. 

INFLUENCE OF FIELD STRENGTH. Figs. 44 and 45 show the spread 
bands from the samples of Figs. 40 and 41 but taken, however, at a higher field 
strength. As is well known, permeability variations caused by inhomogeneous 
internal stresses have a greater effect at low field strengths. At higher field 
strengths, the stress influence disappears more and more in the harmonics. At a 

Fig. 38. Spread bands of the five alloys, C1015, C1034, 5115, free-cutting steel 
with 0.1 percent C, and free-cutting steel with Mn, at a test frequency of 60 c.p.s. 

There is no phase point available at which all five alloys can be separated. 

Fig. 39. At a test frequency of 5 c.p.s., a separation of the five alloys from Fig. 38 
is possible at the dashed vertical line. 



Fig. 40. Spread bands of C1045, C1034, and 5140 steels at 60 c.p.s. For the test, 
the test parts are cut from a wire roll and straightened by hammering. Because of 
strong inhomogeneous stresses, separation of the three alloys is not possible at any 

phase point. 

Fig. 41. Satisfactory separation of the three steels with strong inhomogeneous 
stresses from Fig. 40, at a test frequency of 5 c.p.s. 





? medium 
I Resolution 


Institut Dr. Foereter 

Fig. 42. Complex voltage plane of the Magnatest Q coil which contains a test 
body with the diameter d, the electrical conductivity a, and the permeability Hrei.. 

low field strength, according to Fig. 40, no phase range is available in which the 
three alloys can be separated because of the strong inhomogeneous deformation of 
the test parts. At higher field strengths (Fig. 44) a phase range, although small, 
appears around the vertical dashed line, making possible a separation of the three 
alloys. If both a high field strength and a low frequency are used (Fig. 45), then 
a wide phase range with a good separation possibility is obtained. 

EFFECTS OF STRAIGHTENING. The following four Magnatest Q 
picture? represent an especially interesting example. In all four pictures the 
spread bands of the two alloys 5140 and 4140 are present. The test objects were 


f between Gondji re i,d 

Fig. 43. Angle between the conductivity direction and the diameter-permeability 
direction, as a function of the ///, ratio. 

Fig. 44. Spread bands with a test frequency of 60 c.p.s., corresponding to Fig. 40 
but taken with a higher test field strength. 



Fig. 45. Spread bands corresponding to Fig. 44 but with a test frequency of 5 c.p.s. 

Fig. 46. Spread bands of the 4140 and 5140 alloy steels at a test frequency of 
60 c.p.s. The test objects were straightened by means of hammering. 



again cut from wire rolls and straightened by means of hammering in order to 
induce strong internal stresses. As can be seen in Fig. 46, the test is almost 
impossible at 60 c.p.s., since only a very small phase range is available in which 
the spread bands of the two alloys separate. At 5 c.p.s. (Fig. 47), however, a wide 
phase range with a good separation possibility is available. The reason is a 
decrease of the /// value of the test object by using the lower frequency of 
5 c.p.s., where the conductivity direction is nearly perpendicular to the diameter 
and permeability direction, so that conductivity effects can be separated from 
inhomogeneous permeability effects caused by straightening stresses. 

Fig. 47. Spread bands of 4140 and 5140 alloy steels of Fig. 46 at a test frequency 

of 5 c.p.s. 

Of interest is the single point P of the spread band from the alloy 4140, at 
which all Magnatest curves intersect. This point is an additional characteristic 
of a specific alloy. 

of samples shown previously in Figs. 46 and 47 are again illustrated in the two 
Figs. 48 and 49, but after annealing in order to stress-relieve the material. Thus 
it is possible that a good separation will occur between 4140 and 5140 even at 
60 c.p.s. However, the spread band of the 4140 alloy is relatively wide. This 
picture clearly shows that a separation of the two alloys at 60 c.p.s. in Fig. 46 is 
impossible only because of the strong inhomogeneous stresses; i.e., because of 
permeability variations. After annealing, the permeability values are approxi 
mately equal, so that a separation is also possible at 60 c.p.s. 

However, Fig. 49 shows that a separation at 5 c.p.s., also in the annealed condi 
tion, is considerably more favorable. The spread band of the alloys to be 

Fig. 48. Spread bands of 4140 and 5140 alloy steels at a test frequency of 60 c.p.s. 
after stress-relief annealing. 

Fig. 49. Spread bands of annealed 4140 and 5140 alloy steels of Fig. 48, but with 
a test frequency of 5 c.p.s. 




separated is generally smaller at 5 c.p.s. because the surface of the material has 
more influence at 60 c.p.s. than at 5 c.p.s. Thus, slight edge decarburization 
in the wire increases the spread band width, whereas at 5-c.p.s. test frequency, a 
relatively higher penetration depth exists. At 5 c.p.s. a measurement value is 
obtained which is representative of a large portion of the test body volume. 
Therefore, variations in the surface, such as edge decarburization and soft spots, 
are less noticeable at 5 c.p.s. than at 60 c.p.s. 

Separation of Drawn, Annealed, and Tempered Carbon Steel. It is also of 
interest to investigate the influence of the thermal and mechanical treatment of 
ferromagnetic material on the Magnatest Q picture. Fig. 50 shows the spread 

Fig. 50. Spread bands of C1034 carbon steel for three conditions: cold drawn, 
annealed, and tempered. 

bands of the steel C1034 in three conditions: drawn, annealed, and tempered. 
The picture clearly shows the great influence of the type of treatment on the 
magnetic properties. The electrical values vary only slightly between the three 

A laboratory investigation was made in a large automobile plant for the purpose 
of determining the influence of the heat treatment and thus the tensile strength 
of automotive parts on the picture of the Magnatest Q instrument. 4 A large num 
ber of guiding sleeves of the material C1034 served as experimental specimens 
(these guiding sleeves consist of tube sections; length, 76 mm.; outside diameter 
42 mm.; and inside diameter, 30 mm.). The test parts were subjected to various 
heat treatments so that three different ranges of tensile strength were obtained. 



For the test parts of the first group, which after a normal annealing (% hr. at 
S60 C.) were uniformly cooled in the furnace for 12 hr., the tensile strength 
values were between 74,500 and and 79,000 p.sl Group.- 2 and 3 were cooled 
after the same annealing in still air or by meant 1 of air circulation ,-o that the test 
parts of group 2 had a tensile strength range between 83,000 and S7,500 p>.i. y 
and those of group 3 had a spread range between 90,000 and 100,000 p.s.i. 

Fig. 51 shows the three spread ranges for the three tensile strength group.-? of 
the test parts, whose heat treatment was described above. In Fig. 52 the meas 
ured tensile strength values for the three groups, with various cooling periods, are 
plotted as a function of the slit value on the screen of the Magnate^ Q. The three 

Fig. 51. Three spread bands of C1034 steel for three tensile strength groups: 

74,500 to 79,000, 83,000 to 87,500, and 90,000 to 100,000 p.s.i. The three tensile 

strength groups were obtained by cooling periods of different duration. 

tensile strength groups are clearly separated from each other, and it is shown that 
the slit values give an even better differentiation of the heat treatment condition 
than the tensile strengths which are obtained by means of ball impression at only 
one point, while the magnetic values are obtained as an average over a larger 

Figs. 51 and 52 are an example of an essential characteristic of the Magnatest Q 
instrument. On one hand, it gives an entire curve with its specific characteristic 
for a specific material or material condition. On the other hand, quantitative 
measurement values from the entire curve can be read as a result of the introduc 
tion of the slit value. Thus the picture screen fulfills two functions: First, it 



represents the entire curve with all the characteristics of a two-dimensional 
picture. Second, the screen of the Magnatest Q instrument fulfills at the same 
time the functions of a measurement instrument with pointer indication. In 
general, for an optimum analysis possibility, two characteristics of the test 
instruments are required: two-dimensional representation of a characteristic 
curve, as well as the possibility of indicating specific measurement values. 

f/vi nfin 









t * 



a/\ /W) 







(5 -< 

r -4 

r _ / 

> c 

? +; 

? +x 

? ^d 

> +a 


Fig. 52. Evaluation of Fig. 51, showing measured tensile strength values as a 

function of the Magnatest Q slit value. The values obtained electromagnetically 

give a higher resolution of structure variations than hardness measurement. 

SORTING FOR TENSILE STRENGTHS. Fig. 53 (a) shows another ex 
ample of the separation of automotive parts according to tensile strength values. 
From a number of tubes of C1036 material, all parts were to be rejected, for 
reasons of machinability, whose tensile strength was above 115,000 p.s.i. Test 
parts having a tensile strength higher than 115,000 p.s,i. have caused excessive 
wear of the finishing machines. The Magnatest Q instrument sorted the test parts 
into four groups, which are given in Fig. 53 (a). 

An annealed part with a tensile strength of 100,000 p.s.i. served as compensa 
tion test body. The Magnatest sorted four groups with the following tensile 
strengths: 107,000 to 113,000 p.sl, 113,000 to 115,000 p.sl, 116,000 to 120,000 
p.sl, and 122,000 to 129,000 p.s.i. Fig. 53 fb) shows the result, which is typical for 
the application of the test instrument for the sorting of parts with excessive 
hardness. Such experiments with technological proof of the eddy current test 
results serve for the calibration of the test instrument for 100 percent sorting. 

Forged Parts. The following example (Fig. 54) was obtained for forged parts. 
Testing for tensile strength is especially successful here. In the finishing of sup 
port levers, difficulties were encountered in a plant because of excessive tool 
wear, since the tensile strength limits of 115,000 p.s.i. were occasionally exceeded. 
The support lever tested has two journals which were separated from each other 
by a distance of 200 mm. With a suitable Magnatest Q coil, it was possible to 
determine the tensile strength values of the two journals separately. Figs. 54 and 
55 give the tensile strength values from the same test part, measured at the large 







106 60 40 1 

107,000-113,000 113.000-115,000 116,000-120,000 122.000-129,000 

Number tested 

Measured tensile 
strength in p.s.i. 




Too hard 

Too hard 


Fig. 53. (a) Spread band of C1036 carbon steel at various tensile strength steps. 

Rejection of all parts having a tensile strength of 115,000 p,s.i. (above the dashed 
horizontal line) is possible, (b) Results of Magnatest Q sorting. 

journal (Fig. 54) and at the small journal (Fig. 55). Figs. 54 and 55 are an 
example of the possibility of testing specific zones separately, which is of special 
importance on complicated test parts. 

Fig. 56 shows the testing of large crankshafts. Because of the considerable 
weight of these test parts, the test coil is placed on the journals, which are thus 
tested for the correct metallurgical condition. 

Fi- 57 also shows results of tests on drop forcings of the alloy C10b4 at vari 
ous heat treatment conditions. The curve with the largest peak at the bottom 
corresponds to the hardened condition, while the other spread bands correspond 
to conditions of the forced parts which were annealed for various tensile strength 
ranges. The spread bands in Fig. 57 cover the tensile strength range of from 
86,000 to 130,000 p.s.i. 

Fig. 54. Sorting for tensile strength of support levers of tempered C1040 carbon 
steel. The test coil contains the large journal of the support lever. 

Fig. 55. Sorting of the parts from Fig. 54 for tensile strength, where the small 
journal of the support lever is in the test coil. Example of testing only a portion of 

a production part. 


Iiistitut Dr. Foerster 

Fig. 56. Testing of large crankshafts for correct metallurgical condition of the 


Fig. 57. Screen picture of drop-forged parts of C1034 carbon steel hardened 
(bottom curve) and annealed for various tensile strengths. 





Magnates! Q instrument is also used in the ball-bearing industry for the sorting 
of ball-bearing rings, rollers, and balls according to heat treatment condition. 
Fig. 58 shows the spread bands of ball-bearing rings which were tempered and 
annealed for various hardnesses. The curve with the peak at the bottom corre 
sponds to the annealed condition. Ball-bearing steels in general result in small 
spread bands for the various conditions, so that the eddy current sorter provides 
a higher accuracy than hardness test methods. The speed of hardness testing of 
items such as ball bearing rings is between 4000 and 8000 pieces per hour. The 
test parts are transported through the test coil at a specific distance from each 
other by means of a transport belt. 

Fig. 58. Spread bands of ball-bearing steel (ball-bearing rings) hardened and 
annealed at various temperatures. Bottom curve corresponds to the annealed con 

The Magnatest Q is also used for the sorting of hard and soft balls which pass 
through the coil in close succession in an endless chain. If a soft ball appears 
between the hard balls, the sorting gate is triggered, and in addition to the soft 
ball, several hard balls will fall into the rejection group. At the end the soft balls 
are retested and separated from the hard balls. Here, of course, an extremely high 
test speed is obtained. 

DUCTION PARTS. The Magnatest Q is used in many plants for the measure 
ment of the depth of case. Fig. 59 shows the Magnatest Q spread bands of 
case-hardened valve stems which were annealed in a carbon atmosphere for 8, 12, 



and 16 hr. Case depths of 0.6, 0.9, and 1.6 mm. were obtained. In sorting for case 
depths, it is relatively simple to adjust the sorting limits >o that all te&t parts 
with too little or too much case depth will be rejected. For case depths up to 
approximately 1.5 mm., the frequency of 60 c.p.s. can be used. For greater case 
depths (which, for example, are common in crankshafts), a frequency of 5 c.p.s. 
is necessary. 

Use of Probe and Yoke Coils. Besides the test coil into which the te*t part is 
inserted for testing, yoke and probe coil arrangements are used more and more in 
connection with the Magnates! Q instrument. In this case the test coil is placed 
on the test part. Thus it is possible to measure the magnetic and thereby the 

Fig. 59. Magnates! Q spread bands of case-hardened valve stems, which were 
case-hardened for 8, 12, and 16 hr. to obtain various depths of case. 

mechanical and metallurgical properties at a specific point. In large European 
automobile plants, the case depth at specific points on the intake valves for auto 
mobiles is tested with the aid of yoke coils. Another application of the point 
measurement is the testing for constant case depths on large machine parts. 

Valve Stems. Finally, Fig. 60 shows the satisfactory differentiation of the case 
depth on valve stems, obtained with the Magnatest Q method. It can be seen in 
Fig. 60 that case-depth variations of 0.1 mm 4/1000 in. have a considerable 
influence on the indication, so that sorting in increments of 1 '10-mm. case depth 
is entirely possible. If one considers that this test is possible on fully automatic 
Magnatest Q equipment within a fraction of a second, then one realizes the saving 
of time which is connected with such a nondestructive test of an important factor 
of the test object. 



Fig. 60. Magnatest Q spread bands of valves with various depths of case. Resolu 
tion better than 4/1000-in. depth of case. 

beginning of this section, the examples discussed previously represent only a small 
selection of numerous test results obtained in many hundreds of European plants 
with this widely used eddy current method. In addition to being used in non 
destructive material testing, the Magnatest Q is also used for many problems in 
nondestructive material research. 

of Thomas Steel. Because of the unusually high sensitivity of the 
electromagnetic difference method represented by the Magnatest Q, it is possible 
to notice slight variations during tensile tests or during aging with a sensitivity 
which far exceeds that of any other method. Fig. 61 shows the results of 
aging experiments on Thomas steel " In the curve 4, the Thomas steel was 
normalized with 0.06 percent C and 0.01S percent N, and quenched in water from 
680 C. and precipitated at 110 C. It can be seen that even after 2 hr., a char 
acteristic beginning of the precipitation is noticeable by means of the variation of 
the Magnatest Q values. In Fig. 61 the slit values of the Magnatest curve of the 
test body are plotted as a function of time. In curve B the same Thomas steel was 
stretched by 5 percent and precipitated at 100 C. It is clearly shown how the 
nitrogen segregation is sensitively indicated by the magnetic measurement 
method. In Fig. 61 the selected test sensitivity is only one-twentieth of the maxi 
mum sensitivity of the instrument. 

Plastic Flow of Steel During Tensile Tests. Additional experiments were 
earned out to clarify the behavior of steel during tensile tests, where even far 
below the yield point an irreversible deformation must appear, which is indi- 



100 200 300 

Fig. 61. Results of age-hardening experiments on Thomas steel. Curve ^1 .shows 
Thomas steel with 0.06 percent C and 0.018 percent N, normalized and quenched in 
water from 680 CL then aged at 110 C. The slit values of the Magnatest Q are 
plotted against time. Curve B shows Thomas steel stretched by 5 percent and aged 
at 100 C. Nitrogen segregation during aging causes the variation of the Magnatest Q 


cated by the magnetic values. It was demonstrated early that the magnetic differ 
ence method is the most sensitive method with which to determine the beginning 
of the first plastic flow. 47 

Crack Testing of Steel 

PRINCIPLE OF OPERATION. The Magnatest D has also been used in 
crack testing of semi-finished steel parts. The Magnatest D operates on a prin 
ciple similar to that of the Magnatest Q. However, it uses a difference coil for 
electromagnetic self-comparison of two different portions of the same test object. 
For crack testing, a considerably higher field strength is used than for quality 
tests, which were discussed in the preceding text. The analysis of the Magnatest 
D curve picture on the screen of the cathode-ray tube shows that the internal 
stress effects which appear in rods after a straightening process are indicated in 
a specific phase range. Cracks, however, appear in a different phase range, so 
that the two effects can be separated from each other. In general, the amplitude 
of screen patterns caused by stress effects, i.e., variations of the permeability, 
are not noticeably greater than effects which are caused by cracks. 

of crack testing (see preceding sections on eddy current tests) , it was explained 
why crack effects and stress effects appear at different points on the screen. By 
means of the phase shifter, the curve on the cathode-ray tube is displaced so that 
the stress effect has a zero point at the slit in the center of the screen. Thus effects 
which appear in this slit are caused by cracks. By electronic means the deflections 
in the slit in the screen center are converted into a signal in which the beginning 
and the end of the crack are indicated differently. As is well known, the differ- 



ence method indicates only variations of the crack depth. In other words, no 
quantitative data concerning the crack depth are available. However, since a 
crack, with the exception of very rare cases, is subject to continuous depth varia 
tions, an indication of this variation effect is usually given. The beginning of a 
crack results in a deflection on the screen in the opposite direction from that for 
the end of the crack because the difference coil measures first the increasing crack 
and then the decreasing crack. 

ANALYSIS OF SCREEN PATTERNS. In the exact investigation of the 
effects of cracks and stresses on the screen picture of the Magnatest D instrument, 
individual photographs of specific characteristic effects are made first. Then the 
technique of multiple exposures is used. Here the shutter of the camera 
remains open as the rod proceeds through the test coil, so that all curves which 

Institut Dr. Foerster 

Fig. 62. Multiple-exposure picture of curves appearing on screen when a rod with 
inhomogeneous stresses passes through the test coil of the Magnatest D instru 

appear on the screen are recorded by the camera. Fig. 62 shows the entirety of 
all curves which appear on the screen when internal stresses are present, while 
Fig. 63 shows the multiple-exposure picture of a cracked rod. It is clearly visible 
that maxima of the indicating effect appear at several points. (The theory for 
these tests was given in the preceding sections on eddy current tests.) 

istic examples of cracks in steel are illustrated by means of curve pictures on the 
screen of the Magnatest D, as well as fracture pictures and microsections of the 
test objects. 

Indications of Sound Rod. Fig. 64 (a) shows the multiple-exposure picture of 
a sound rod. In this picture the camera shutter remained open over the entire 


Institut Dr. Foerster 
Fig. 63. Multiple-exposure picture of a rod with cracks. 

length of the rod. Fluctuations appear only at the peaks, which are caused by 
internal stresses. Fig. 64 (b) shows a screen picture of a specific point on the 
rod (single exposure). It shows that the spread band in the center portion of the 
curve, which normally indicates the crack effect, is very narrow. Fig. 64 (c) illus 
trates the confirmation for this picture. The fracture shows no defect indications, 
even under the microscope. Fig. 64 (d) shows the microsection etched according 
to Oberhofer. The segregation structure appears clearly in the cross-section of 
the test body. Fig. 64 (e) shows an edge section of the unetched microsection 
where one can see an accumulation of nonmetallic inclusions; these, however, 
do not influence the screen picture. Fig. 64(f) shows the microsection etched 
with HN0 3 , which indicates no interruptions. 

Indications of Defective Rod. Fig. 65(a), however, shows a multiple- 
exposure picture of a rod with a defect. The multiple-exposure curves in the 
center of the field, which were obtained while the rod passed through, clearly 
indicate a defect. Fig. 65 (b) shows the single-exposure picture at the largest 
defect variation, i.e., the largest deflection of the curves in Fig. 65(a). Fig. 65(c) 
shows the fracture in the plane of the largest defect variation. One can clearly 
see the crack which was opened by the fracturing; this, however and it is 
essential to point this out is not visible on the surface in its continuation. The 



Institut Dr. Foerster 

Fig. 64. Analysis of sound test rod. (a) Multiple-exposure picture of a defect-free 
rod. (b) Single exposure of the Magnatest D curve at a specific point of the rod. 
(c) Picture of the rod fracture. No defects are detectible. (d) Etching picture accord 
ing to Oberhofer for a clear illustration of the P and S segregation in the rod cross- 
section, (e) Microsection, unetched (100X). (f) Microsection, etched (100X). 

short portion, in which the crack came to the surface, caused the indication of 
the maximum defect variation. Fig. 65 (d) shows the mierosection etched accord 
ing to Oberhofer. The upper portion of the picture shows ray-shaped subsurface 
voids, of which the one in the center nearly reaches the rod surface. Fig. 65 (f) 
shows the defect at 100X magnification. Fig. 65 (h) shows the etched mierosection 
at a 100X magnification. Even at a 30 X magnification, the full size of the void is 
not visible. However, the 100X magnification shows that the crack, which is 
continuous to the rod surface, also passes through somewhat larger nonmetallic 

liistitut Dr. 

Fig. 65. Analysis of cracked rod. (a) Multiple-exposure picture of a >teel rod with 
one defect, (b) Single exposure at the largest defect variation, (c) Fracture picture 

of the point of the rod which corresponds to the curve in picture (b). Clear repre 
sentation of the crack which was opened by fracturing, (d) Etching picture according 
to Oberhofer for the representation of the P and S> segregation, (el U net died fine- 
section (SOX), (f) Unetched fine-section (100X). (g) Etched -ection (40X ;. fh) 
Etched section (100X). Crack runs along several nonmetallic inclusions to the 







>,&< -'?/> * 
% ,*, , ; " 

yv* ; , : 



Institut Dr. Foerster 

Fig. 66. Analysis of fine crack, (a) Multiple-exposure picture of a crack, taken with 
the camera shutter open, (b) Largest defect variation as single exposure, (c) Frac 
ture picture m the plane of the rod which produced the single exposure (b) 
(d) itching picture according to Oberhofer for the representation of the P and S 
segregation m the cross-section, (e, f) Unetched fine-section (30X and 100X). (g, h) 
Etched fine-section (SOX and 100X). 



Ji8S*'' ; H ; ,...i, ilk 




Institut Dr. Foerster 

Fig. 67. Analysis of short discontinuities, (a) Multiple-exposure picture of a steel 

rod with several short defects, (b) Single exposure of the largest defect variation 
from (a), (c) Fracture picture, showing inclined, incoming crack on the left side, 
(d) Etching picture according to Oberhofer. Because of the low magnification, the 
crack is not visible, (e, f) Unetched fine-section (30 X and 100 X). Cg, h) Etched 
fine-section (SOX and 100X). 



Indications of Fine Crack. Fig. 66 (a) shows a multiple-exposure picture 
which again points unequivocally to a defect. Fig. 66 (c.) shows the fracture in 
the plane of the largest defect variation and clearly shows the indicated defect 
in the upper portion. Fig. 66 (d) shows the microsection etched according to 
Oberhofer. However, the defect indicated in Fig. 66(b) is not visible with this 
magnification. Fig. 66(e) shows the unetched fine-section at a 30x magnification. 
Here, also, it is a case of an elongated void which was opened up to the external 
skin of the rod by the straightening operation. Fig. 66 (f) shows the same defect 
at a 100X magnification. Fig. 66(g) shows the fine-section etched with NHO ?> 
at a 30X magnification, while Fig. 66(h) shows the same section at a 10QX 

InstitufcDr. Foerster 

Fig. 68. Automatic unit in a steel plant for simultaneous quality testing of semi 
finished steel parts for mixed alloys and defects. 

Indications of Short Defects. Fig. 67 (a) shows the multiple-exposure picture 
of a rod with several defect variations. It can be concluded from the structure 
of the multiple-exposure pattern that several short defects exist in the rod. The 
beginning of an individual defect is indicated by an upward turn of the curve, and 
the end of the defect by a downward turn of the curve. Fig. 67 (b) shows the 
largest defect variation, i.e., the greatest amplitude in the crack-sensitive phase 
range, as a stationary picture. Fig. 67 (c) shows the fracture which is placed in 
the plane of the greatest defect variation. An inclined, incoming crack is visible 
on the left side of the fracture. Fig. 67 (d) again shows the microsection etched 
according to Oberhofer. In this picture the defect is not visible because of the low 
magnification. Fig. 67 (e) shows the unetched fine-section at a 30 X magnifica 
tion. The inclined, incoming crack is clearly visible. Fig. 67 (f) shows the same 



section at a 100 X magnification. Fig. 67 (g) show? the etched fine-action at a 
SOX magnification, and Fig. 67fh) at a 100X magnification. The defect curve in 
all microsections is as clear as the indication form of the Magnatest D curve 

Crack-Test Equipment Installation. Figs. 63 through 67 illustrate the com 
parison of crack indications shown on the screen of the Magnates* D instrument 
with the results obtained metaUographically on the defects'. These examples 
represent a small selection from extensive research material collected during recent 
years. By systematic analysis of defects found with the Magnates! D~ inurn 
ment, the cause from which these defects originated could be determined, as for 
the tungsten wires previously discussed. The pictures are taken from a paper 
by Sprungmann S1 describing industrial experience with crack testing in a rolling 
mill, using the Magnatest D instrument. Fig. 68 shows the units used there. 

discussed, the various eddy current methods for nondestructive testing have found 
wide application for test and research purposes. A few practical applications of 
scientific and technical investigations are listed here. The following methods, 
which are carried out with the instruments described in the previous sections, are 
of practical significance: 

1. Xoncontacting measurement of the electrical conductivity at high temperatures 
(taking of temperature-conductivity-diagrams of alloys; control of the sinter 
ing process at high temperatures) . 

2. Measurement of the thickness of an intercrystallme corrosion layer. 

3. Determination of the diffusion speed of two metals as a function of tempera 

4. The nondestructive measurement of the corrosion rate under the action of the 
corroding medium. 

a. For rod-shaped samples. 

b. For flat samples. 

5. Electronic indication of a metallurgical condition diagram on the screen of a 
cathode-ray tube. 

6. Noncontacting measurement of the magnetic property variations at high tem 

7. Noncontacting measurement of the eccentric position of a conductor in an in 
sulating coating. 

8. Control of the metallizing process of metal paper. 

9. Method for noncontacting measurement of heating of sewing needles during 

10. Quantitative determination of the magnitude and type of internal stress in 
ferromagnetic materials. 

The methods listed here are discussed more thoroughly in the literature. 42 

The application and interpretation examples given in this section represent only 
a small portion of the entire application range of the eddy current method. If it 
is considered that these methods exceed most other nondestructive te?t methods 
in test speed, then it is realized that these test methods are excellently suited to 
control the quality of metal products by means of 100 percent tests. 


1. ALLEN, J. \V., and R. B. OLIVER. ''Inspection of Small-Diameter Tubing by Eddy- 

Current Method," Nondestructive Tenting, 15, Xo. 2 (1957) : 104. 

2. BATES, L. F., and X. UNDERWOOD. BJSJRA., Rept. No. MG EG 82 (194S). 


3. BEUSE, H., and H. KOELZER. "Erfahrungen mit der magnetinduktiven Pruefung 

von Stabstahl auf Risse (Experience with the Magnetic Testing of Steel Rods 
for Cracks)," Arch. Eisenhuetenw., 23 (1952) : 363. 

4. . "Industrielle Erfahrungen mit dem Magnatest-Q-Geraet (Industrial Ex 
periences with the Magnatest Q Instrument)," Z. Metallk., 45 (1954) : 677. 

5. BLANDERER, G. "Erkennung und Sortierung von Metallproben (Identification and 

Sorting of Metal Samples)," Z. Erzbergbau Metallhuetenw., 5 (1952) : 257. 

6. BOZORTH, R. M. Ferromagnetism. 2d ed. New York: D. Van Nostrand Co., 

Inc., 1953. P. 847. 

7. BREITFELD, H. "Die zerstoerungsfreie Pruefung von Metallen mit dem magnet 

induktiven Tastspul-Geraet (Nondestructive Testing of Metals with the 
Magneto-Inductive Probe Coil Instrument)," Metall, 9 (1955) : 14. [Sigmatest] 

8. BROWN, R. J., and J. H. Bridle. "A New Method of Sorting Steels," Engineer, 

176 (1943): 442. 

9. BUNGE. G. "Betriebliche Anwendung eines Tastspulgeraetes bei Nichteisen- 

Metallen (Industrial Application of the Probe Coil Instrument on Nonferrous 
Metals)," Z. Metallk. , 45 (1954) : 205. [Sigmatest] 

10. BURROWS, C. W., and F. P. FAHY. "Magnetic Analysis as a Criterion of the Qual 

ity of Steel and Steel Products," Am. Soc. Testing Materials, Proc., 19, Pt. II 
(1919): 5. 

11. CANNON, W. A., JR. "'Industrial Application of Eddy Current Testing," Nonde 

structive Testing, 11, No. 5 (1953): 30. [Sigmatest, Sigmaflux] 

12. ."Practical Conductivity Measurement for the Electrical Industry," 

Nondestructive Testing, 13, No. 6 (1955): 32. [Sigmatest] 

13. CAVANAGH, P. E., and T. WLODEK. "Magnet Stress Analysis," Am. Soc. Testing 

Materials, Spec. Tech. Publ., 85 (1949) : 123. 

14. CORNELIUS, J. R. "Electronic Inspection," Aircraft Production, 10 (1948) : 52. 

15. COSGROVE. L. A. "Quality Control Through Nondestructive Testing with Eddy 

Currents/' Nondestructive Testing, 13, No. 5 (1955): 13. 

16. "Electronic Comparators, An Important Development in Testing Equipment/' 

Automobile Engr.,3? (1947): 206; ''Electronic Comparators, Some Applications 
of Cornelius Equipment," ibid., 271. 

17. FOERSTER, F. "Die automatische Sortierung von Massenteilen nach Rissen mit 

einem werkstoffunabhaengigen Verfahren (The Automatic Sorting of Production 
Parts for Cracks with a Material-Independent Method)," Z. Metallk. (in 

IS. . <k Die Automatisierung der zerstoerungsfreien Qualitaetspruefung mit dem 

Magnatest-Q (The Automation of Nondestructive Quality Testing with the 
Magnatest Q)," Z. Metallk. (in press) 

19. . "Einfuehrung in die Gnmdlagen der Wirbelstromrisspraefung von Halb- 

zeug (Introduction to the Foundation of Eddy Current Crack Testing of Semi- 
Finished Parts)," Z. Metallk. (in press) 

20. ."Grundlagen und Praxis der Dickenmessung duenner Metallischichten 

(Principle and Practice of the Thickness Measurement of Metal Layers)," Z. 
Metallk. (in press) 

21- ."Die magnetische und elektromagnetische Sortentrenmmg von Stahlhalb- 

zeug und Massenteilen (The Magnetic and Electromagnetic Sorting of Semi 
finished Steel Production Parts)," Arch. Eisenhuettenw., 25 (1954): 383. 
[Magnatest Q and Multitest] 

22. ."Ein Mes.geraet zur schnellen Bestimmung magnetischer Groessen (An 

Instrument for the Rapid Determination of Magnetic Quantities)," Z. Metallk , 
32 (1940): 184. 

23. ."Neuartige Geraete zur zerstoerungsfreien Werkstoffpruefung (New In 
strument? for Nondestructive Material Testing)," Z. wirtsch. Fertigung 4 
(1941). [Metallotest] 

24. . ''Neuere Verfahren zur zerstoerungsfreien Werkstoffpniefung (Newer 

Methods for Nondestructive Material Testing)," Berg- u. Huettenmaennische 
Munats., 95 (1950): 284. 


25. FOERSTER, F. "Ein neucs Verfahren zur Automatisienmg dor sralisti>rlnn Quali- 

taetskontrolle in der zerstoerungsfreien Werkistoffpruefung (A New Method for 
the Automation of the Statistical Quality Control in Nondestructive Material 
Testing)/' Z. Metallk. (in press) [Statimat] 

26. . "Non-Destructive Electronic Sorting of Metals for Physical Properties,'' 

.4w. Soc. Testing Materials, Spec. Tech. PnbL 145 (1953) :*S9. [Multitest] 

27. ."Nouveaux precedes d'essai electronique non destmctif des materiaux 

(New Procedures for the Electronic Nondestructive' Testing of Materials)," 
Metaux (Corro$ion-Inds.) t 2b (1951): 497. 

28. ."Theoretische und experimentelle Grtmdlagen der elektromagnetisrlipn 

Qualitaetssortierung von Stahlhalbzeug und Stalilteilen, I. Die mugnetinduk- 
tiven Verfahren bei alleiniger Beruecksichtigung der Grunchvelle (The Theo 
retical and Experimental Foundation of the Electromagnetic Quality Sorting of 
of Semi-Finished Steel Production Parts. I. The Magneto-Inductive Method 
with the Exclusive Consideration of the Fundamental Wave)," Z. Metallk., 45 
(1954) : 206. [Magnates* Q] 

29. ."Theoretische und experimentelle Grundlagen der elektromagnetischen 

Qualitaetssortierung von Stahlteilen, IV. Das Restfeldverfahren (Theoretical 
and Experimental Foundation of the Electromagnetic Quality Sorting of Steel 
Parts, IV. The Residual Field Method)/' Z. Metallk., 45 (1954) : 233. 

30. ."Theoretische und experimentelle Grundlagen der zerstoerungsfreien 

Werkstoffpruefung mit Wirbelstromverfahren, I. Das Tastspulverfahren (Theo 
retical and Experimental Basis for Nondestructive Material Testing with Eddy 
Current Methods, I. The Pickup Coil Method),' 1 Z. Metallk. , 43 (1952): 163. 

31. /'Theoretische und experimentelle Grundlagen der zerstoerungsfreien 

Werkstoffpruefung mit Wirbelstromverfahren, VII. Die magnetinduktivr 
Risspruefung von Stahl (Theoretical and Experimental Basis for the Nonde 
structive Material Testing with Eddy Current Methods. VII. The Magneto- 
Inductive Crack Testing of Steel)," Z. Metallk. t 45 (1954) : 221. [Magnatest Dl 

32. . ''Theoretische und experimentelle Grundlagen der zerstoerungsfreien 

Werkstoffpruefung mit Wirbelstromverfahren, IV. Praktische Wirbelstrom- 
geraete mit Durchlaufspule zur quantitativen zerstoerungsfreien Werkstoff 
pruefung (Theoretical and Experimental Basis for the Nondestructive Material 
Testing with Eddy Current Methods, IV. Practical Eddy Current Instrument? 
with Feed-Through Coil for the Quantitative Nondestructive Material Test 
ing)," Z. Metallk., 45 (1954): ISO. [Sigmaflux, Wire-Crack Test Instrument. 

33. . "Theoretische und experimentelle Grundlagon der zerstoerungsfreien 

Werkstoffpruefung mit Wirbelstromverfahren, VI. Die beruehrungsfreie 
Messung der Dicke und Leitfaehigkeit von metal Hschen Oberflaechenschichten. 
Folien und Blechen, 1. Teil: Theoretische Grundlagen (Theoretical and Experi 
mental Basis for the Nondestructive Material Testing with Eddy Current 
Methods, VI. The Contact-free Measurement of the Thickness and Con 
ductivity of Metallic Surface Layers, Foils and Sheets, Pt. 1. Theoretical 
Basis) ," Z. Metallk., 45 (1954) : 197. 

34. . "Theoretische und experimentelle Grundlagen des Magnatest Q Geraetes 

(Theoretical and Experimental Foundation of the Magnatest Q Instrument)," 
Z. Metallk. (in press) 

35. . "Das Wirbelstromverfahren als quant itatives Pruef- und Messverfahren 

der zerstoerungsfreien Werkstoffpruefung (The Eddy Current Method as a 
Quantitative Test and Measurement Method for Nondestructive Material 
Testing)," Z. Metallk. (in press) 

36. /'Wirbelstromverfahren zur zerstoerungsfreien Werkstoffpruefung von 

Metallen (Eddy Current Methods for Nondestructive Material Testing of 
Metals)," Aluminium, 30 (1954): 511. [The methods discussed in this article 
include Sigmatest, Sigmaflux, Crack Tester, Layer Thickness Meter, Wall 
Thickness Meter.] 


37. FOERSTER, F. "Wirtsehaftliche Gesichtspunkte bei der zerstoerungsfreien Werk 

stoffpruefung (Economic Considerations in Nondestructive Material Testing)," 
Z. Metallk., 44 (1953) : 346. 

38. . "Wirtschaftliche Gesichtspunkte der zerstoerungsfreien Werkstoffpruefung 

(Economical Considerations in Nondestructive Material Testing)," Indu&tne- 
b/fltt,54,No.9 (1954). 

39. . "Die zerstoerungsfreie elektronische Sortierung von Metallen nach physika- 

lischen Eigenschaften (The Nondestructive Electronic Sorting of Metals Ac 
cording to Physical Properties)," Schweiz. Arch, angew. Wiss. u. Tech., 19 
(1953): 57. [Multitestl 

40. . "Die zerstoerungsfreie Messung der Dicke von nichtmetallischen und 

metallischen Oberflaechenschichten (Nondestructive Measurement of the 
Thickness of Nonmetallic and Metallic Surface Layers)," Metall, 7 (1953): 

41. ."Die zerstoerungsfreie Messung der Dicke von nichtmetallischen und 

metallischen Oberflaechenschichten (The Nondestructive Measurement of the 
Thickness of Nonmetallic and Metallic Surface Layers)," Jahrbuch Ober- 
flaechentech. (1954). 

42. .Zer&toentngsfreie Werkstoffjmiefung mil elektrischen und magnetischen 

Verjahren (Nondestructive Material Testing with Electrical and Magnetic 
Methods). Berlin, Goettingen, Heidelberg: Springer-Verlag. (in press) 

43. FOERSTER, F.. and H. BREITFELD. "Theoretische und experimentelle Grundlagen 

der zerstoerungsfreien Werkstoffpruefung mit Wirbelstromverf ahren : V. Die 
quantitative Risspruefung von metallischen Werkstoffen (Theoretical and Ex 
perimental Basis for the Nondestructive Material Testing with Eddy Current 
Methods, V. Quantitative Crack Testing of Metallic Materials)," Z, Metallk., 
45 ( 1954) : 188. 

44. . ''Die zerstoerungsfreie Pruefung von Leichtmetall mit Hilfe einer Tast- 

t=pule (The Nondestructive Testing of Light Metal by Means of a Pickup 
Coil) ," Aluminium, 25 (1943) : 253. [Sigmatest] 

45. ."Zerstoerungsfreie Pruefverfahren auf elektriseher Grundlage (Nonde 
structive Material Testing on an Electrical Basis)," Aluminium, 25 (1943) : 130. 

46. . Zerstoerungsfreie Werkstoffpruefung mit Wirbelstromverf ahren, II. 

Praktische Ergebnisse und industrielle Anwendungen des Tastspulverfahrens 
(Nondestructive Testing of Materials with Eddy Current Methods, II. Prac 
tical Results and Industrial Application of the Pickup Coil Method)/' Z. 
Metallk., (1952): 172. [Sigmatest] 

47. FOERSTER, F., and K. STAMBKE. ''Magnetische Untersuchungen innerer Span- 

nungen, I. Eigenspannungen beim Recken von Nickeldraht (Magnetic Investi 
gations of Internal Strain, I. Internal Stress During the Stretching of Nickel 
Wire)." Z. Metallk., 33 (1941) : 97. '*IL Eigenspannungen bei duesengezogenem 
Nickeldraht (Internal Stress in Nickel Wire Extruded Through a Nozzle), 1 ' 
ibid.. 104. 

48. . "Theoretische und experimentelle Grundlagen der zerstoerungsfreien 

Werkstoffpruefung mit Wirbelstromverfahren, III. Verfahren mit Durchlauf- 
spule zur quantitativen zerstoerungsfreien Werkstoffpruefung (Theoretical and 
Experimental Basis for the Nondestructive Material Testing with Eddy Cur 
rent Methods, III. Feed-Through Coil Method for Quantitative Nondestruc 
tive Material Testing)," Z. Metallk., 45 (1954) : 166. 

49. FOERSTER, F.. and H. WETZEL. "Zur Frage der magnetischen Umklappvorgaenge in 

Eisen und Nickel (On the Question of Magnetic Reversals in Fe and Ni) " Z 
Metallk., 33 (1941): 115. 

50. GERLACH, W. "Magnetische Verfahren zur Werkstoffpruefung (Magnetic Method 

for Material Testing)/ 1 Z. Tech. Physik, 15 (1934) : 467. 

51. GILMOUR. C. M. ''Non-Destructive Testing," Metal Ind. (London), 85 (1954)- 


52. GRAEP, H. "Die zerstoerungsfreie Qualitaetspruefung von grossen Schmiedeteilen 


mit dem Magnatest-Q (Nondestructive Quality Trying of Large Forged Parts 
with the Magnatest Q). v Z. Metallk. 'in press) 

53. GROMODKA, E. ''Zerstoerungsfreie Pruefung von Wolfram- und Molybdaen- 

Draehten mit Hilfe des Drahtrisskuwimetcrs (Nondestructive Testing of 
Tungsten and Molybdenum Wires with the Aid of the Wire Crack Kiiwimeter)," 
Metall, 5 (1951) : 335. [Wire-Crack Test Instrument] 

54. HEN NIG. R. "Ueber die Anwendung magnetischer Differenzverfahren, insSi?- 

sondere bei der Untersuchung von Aushaertungs- und Reckulterungsvorgaengen 
sowie waehrend elastischer und plastischer Verforniung (Concerning the Appli 
cation of the Magnetic Difference Method. Especially in the Investigation of 
Age-Hardening and Recrystallization Precede? as Well as During Elastic ami 
Plastic Deformation)," Diplomarbeit der Tt'chai^ch'u Hochxchnh- Aachen, 

55. HOCHSCHILD, R. "Eddy Current Testing by Impedance Analysis," XondMtnicttw 

Testing, 12, No. 3 (1954): 35, 51. [Multitest, Sigmatest 1 

56. . "The Theory of Eddy-Current Testing in One (Not-So-Easy) Le^on." 

Nondestructive Testing, 12, No. 5 (1954) : 31. 

57. HUGHES, D. E. l 0n an Induction Balance," Phil. Mag., [5] 8 (1879): 50. 

58. KAYSER, 0., F. PAWLEK, and K. REICHEL. "Die Beeinflussung der Leitfaehigkeit 

reinsten Kupfers durch Beimengungen (The Influence of Impurities on the 
Conductivity of Purest Copper)," Met all 8 (1954) : 532. 

59. KEIL, A. "Anwendung von Wirbelstromverfahren in einem Metallwerk (Applica 

tion of the Eddy Current Methods in a Metal Plant)," Z. Metallk. (in press) 
[Sigmatest, Wire-Crack Test Instrument, Equivalent Layer Thickness Meter] 

60. . "Leitfaehigkeitsmessungen an galvanisch erzeugten Metallfolien (Con 
ductivity Measurement on Galvanically Produced Metal Foils)/' Mttall- 
oberflaeche, 9 (1955): 81(A). [Equivalent Layer Thickness Meter 1 

61. KEiL 3 A., and C. L. MEYER. "Die Anisotropie der elektrischen Leitfaehigkeit 

einiger gesinterter Kontaktwerkstoffe (The Anisotropy of Electrical Conduc 
tivity of Some Sintered Contact Materials)/' Z. Metallk., 45 (1954): 119. 

62. ."Der Einfluss des Faserverlaufes auf die elektrische Verschleissfestigkeit 

von Wolfram-Kontakten (The Influence of the Direction of the Fibers on the 
Electrical Corrosion-Resistance of Tungsten Contacts)," Elektrotech. Z., 72 
(1951): 343. 

63. . "Die zerstoerungsfreie Risspruefung von Wolframstaeben nach dem 

Wirbelstromverfahren (Nondestructive Crack Testing of Tungsten Rods Ac 
cording to the Eddy Current Method)," Z. Metallk., 45 (1954): 194. [Wire- 
Crack Test Instrument] 

64. KEIL, A., and G. OFFNER. "Ueber die Pruefung von Metallbelaegen auf Isolier- 

stoffen mit einem Wirbelstromverfahren (Concerning the Testing of Metal 
Layers on Insulating Materials with an Eddy Current Method)," Z. Metallk., 
45 (1954) : 200. [Equivalent Layer Thickness Test] 

65. KOCH, W. u lndustrielle Anwendungen des elektromagnetischen Qualitaete- 

pruefgeraetes in der stahlerzeugenden und stahlverbrauchenden Industrie (In 
dustrial Application of the Electromagnetic Quality Test Instrument in the 
Steel-Producing and Steel-Consuming Industry)/' Z. Metallk. (in press) 
[Magnatest Q] 

66. KOESTER, W T ., and H. BREITFELD. "Haerte und Leitfaehigkeit kalt ausgehuerteter 

Aluminium-Legierungen (Hardness and Conductivity of Cold-Hardened Alu 
minum Alloys)," Z. Metallk., 35 (1943) : 163. 

67. MATTHAES, K. "Abnahmepruefung von Aluminium-Legierimgen mit den elektro- 

induktiven Pruefgeraeten von Dr. Schirp (Receiving Testing of Aluminum 
Alloys with the Electroinductive Test Instrument^ by iSchirpJ/' Aluminium, 
25 U943): 106. 

68. . "Erfahrungen bei der zerstoenmgsfreien AVerkstoffpmefung (Experience 

with Nondestructive Material Testing)/' Metdl, 5 ( 1951) : 544. 


69. MATTHAES, K. "Magnetinduktive Stahlpruefung (Magnetoinductive Testing of 

Steel) J'Z.Metallk., 39 (1948): 257. 

70. McCLrRG, G. 0. "Theory and Application of Coil Magnetization," Noiidestruc- 

tive Testing, 13, No. 1 (1955) : 23. 

71. NACHTIGALL, E. "Die elektrische Leitfaehigkeit von Aluminium (The Electrical 

Conductivity of Aluminum)," Aluminium, 30 (1954) : 529. [Sigmatest] 

72. NOWOTNY, H., W. THURY, and H. LANDERL, ''Die Bestimmung von Silicium in 

Reinaluminium mittels elektrischer Leitfaehigkeitsmessung (The Determina 
tion of Si in Pure Aluminum by Means of Conductivity Measurement)," Z. 
anal. Chem., 134 (1951) : 241. [Sigmatest] 

73. O'DELL, D. T. "Apparatus for the Detection of Splits in Tungsten Wire," J. Sci. 

7wj5/r.,20(1943): 147. 

74. PLANT, W. R., and C. MANUAL. "Eddy Current Testing of Capillary Tubing," 

Am. Soc. Testing Materials, Spec. Tech. Publ, 223 (1958). 

75. ROBINSON, I. R. "Magnetic and Inductive Nondestructive Testing of Material," 

Metal Treatment, 16 (1949) : 12. 

76. RYDER, C. M. "Devices for Testing Carbonization of Metals," U.S. Patent No. 

185,647 (1876). 

77. SANFORD, R. S M and M. F. FISCHER. "Application of Magnetic Analysis to the 

Testing of Ball Bearing Races," Am. Soc. Testing Materials, Proc., 19, Pt. II 
(1919): 68. 

78. SCHIRP, W. "Die magnetinduktive Pruefung von Rohren (The Magneto-Induc 

tive Testing of Tubes)," Elektwtech. Z., 60 (1939): 857. "Neue magnetinduk 
tive Pruefgeraete fuer Halbzeuge aus Nichteisenmetallen (New Magnetoinduc 
tive Testing Apparatus for Semi-Finished Material of Nonferrous Metals)," 
ibid., 64 (1943) : 413. 

79. . "Zerstoerungsfreie Werkstoffpruefung mit Hilfe elektronischer Verfahren 

(Nondestructive Material Testing with the Aid of Electronic Methods)," Elek- 
tropost, 8/9 (1954) : 234. 

80. SPRUNGMANN, K. a Betriebliche Erfahrung der Wirbelstrom-Risspruefung bei 

Stahlhalbzeug (Industrial Experience with Eddy Current Testing of Semi 
finished Steel)," Z. Metallk., 45 (1954) : 227. [Magnatest D] 

81. . tk lndustrielle Erfahrungen der Risspruefung von Stahlhalbzeug mit dem 

Magnatest D (Industrial Experience with the Magnatest D in the Crack Test 
ing of Semi-finished Steel Parts)," Z. Metallk., (in press) 

82. . ''Die Risspruefung von Stahlhalbzeug mit dem Magnatest D (The Crack 

Testing of Semi-finished Steel Parts with the Magnatest D)," Z. Metallk. 
(in press) 

83. STAATS, H. N. "Conductivity Now a Useful Nondestructive Testing Tool," 

Materials and Methods, 38, No. 4 (1953): 124. [Sigmatest] 

84. VOSSKUEHLER, H. "Zerstoerungsfreie Pruefung der Al-Mg-Zn-Legierung Hy 43 

auf magnetinduktivem Wege (Nondestructive Testing of an Al-Mg-Zn Alloy 
Hy 43 with a Magnetoinductive Method)," Metall, 3 (1949): 247, 292. 

85. WIELAND, F., and F. ROSCHE. fk Elektromsche Fehlersortierung mit dem Multitest- 

Geraet (Electronic Defect Sorting with the Multitest Instrument)," Z. Metallk., 
45 (1954): 231. [Multitest] 

86. WILLIAMS, S. R. "The Magnetic-Mechanical Analysis of Cast Iron," Am. Soc. 

Testing Materials, Proc., 19, Pt. II (1919) : 130. 

87. ZEPPELIN, H. VON. "Betriebliche Schnellbestimmung und Aenderung des 

Phosphorgehaltes in Schmelzen von Kupfer und Kupferlegierungen (Rapid 
Industrial Determination and Change of Phosphorus Content in Metals of 
Copper and Copper Alloys)," Giesserei, 38 (1951) : 51. [Sigmatest] 

SS. ZIJLSTRA, P. "Die Pruefung von Draehten aus Wolfram und Molybdaen auf 
Risse (The Testing of Tungsten and Molybdenum Wires for Cracks)," Philips 
Tech. Rev., 11 (1949): 12. 

89. ZI/SCHLAG, T. "Magnet Analysis Inspection in the Steel Industry," Am. Soc. Test 
ing Materials, Spec. Tech. PubL, 85 (1949) : 113. 





Basic Methods of Ultrasonic 

Transmission and reflection techniques 1 

Basic testing methods (/. 1) 2 

Ultrasonic test systems 1 

Ultrasonic frequency ranges 1 

Ultrasonic stress ranges I 

Applications of ultrasonic techniques 2 

Advantages of ultrasonic tests 2 

Limitations of ultrasonic tests 2 

Development of tests 3 

Early industrial applications 3 

Recent improvements 3 

Generation of Ultrasonic Vibrations 

Ultrasonic sources 4 

Ultrasonic transducer types 4 

Typical transducer characteristics 4 

Properties of thickness-mode transducer 

elements (/. 2) 4 

Activity constant 4 

Resolving power and sensitivity 5 

Additional design requirements 5 

Typical search units (/. 3} 6 

Straight-beam contact units 5 

Straight-beam faced units 5 

Angle-beam contact units 5 

Immersion units 7 

Contoured or focused units, with acoustic 

lenses 7 

Standard ratings 7 

Search unit frequencies and sizes most 

commonly used (/. 4) 7 

Special units 7 

Ultrasonic Wave Propagation 



Acoustic properties of materials (/. 5) ... 


Principal wave velocities in solids (/. 6). . 

Calculation of acoustic velocities 

Relationships between ultrasonic velocities 
and the elastic constants of isotropic 

solids (/. 7) : - 

Relationship of velocity ratio* to Poi 
son's ratio (/. 8) 


Effects at plane interfaces 12 

Reflection 12 

Percentage ultrasonic reflation coefficient^ 
for normally incident plane wave* flt 
interfaces between \rmons material 1 I/. 

9) 13 

Transmission through thin plate? 14 

Transmission through thin plate (/. 10) .. 14 

Resonance thickne** measurements 15 

Refraction lo 

Refraction and mode conversion at plane 

interface (angular incidence) (/, 11* 16 

Mode conversion 17 

Mode conversion by reflection ;tt plane 

interface (/. 12) 16 

Angular and amplitude relation.4np^ IS 

Amplitude ratio of reflected -to-mc-iilerit 

longitudinal waves (/. 13) 17 

Amplitude ratio of reflerted-to-im.'iilent 

shear waves (/. 14) IS 

General cases of angular incidence 19 

Longitudinal waves in water incident on 

steel (/. 15) 10 

Longitudinal wave** in stivl incident on 

water (/. 16) 20 

Shear waves in steel, vibrating in plane 

of incidence, incident on water (/. 17) .. 21 

Refraction analyzer </. 18) 22 

Refraction data (/, 19) 23 

Beam spread (/. 20) 24 

Diffraction 21 

Attenuation 22 

Attenuation sources and relationships (/. 

21) B 

a', attenuation (db. cm.) 2 (/, 22) 26 

Disturbing influences 24 

Basic Instrumentation and Techniques 

Classification of equipment 26 

Through- transmission techniques 27 

Equipment 27 

Types of throiigh-tianMiiiasicin 'ysttni-* 

U 23) 27 

Defect detectivity 27 

Limitations 28 

Advantages 2M 

Applications - 2S 

Techniques using amplitude and transit -time 

indication 28 

Tune miMirenwnt s.ysln* 2 

Equipment 2s 



CONTENTS (Contort) 


Basic pulse-reflection system (/, 24) 29 

A-scan presentation 29 

Functional diagram of pulse-reflection A- 

scan equipment (/, 25) 30 

Techniques 30 

Basic testing methods (/, 26) 31 

Distance or time scale markers 30 

Limitations 31 

Testing distance obscured by pulse (inter 
face signal or initial pulse) (/. 27) 32 

Sensitivity 31 

Alternative arrangements 32 

Special features 33 

B-scan presentation 33 

B-scan presentation (/. 28) 34 

Equipment 33 

Defect detectibility 33 

Alternative arrangements 34 

C-scan presentation 34 

Equipment 34 

C-scan presentation (/, 29) 35 

Alternative arrangements 34 

Gated systems 35 

Techniques using transducer loading effects ,, 35 

Equipment 36 

Manual resonance system (/. 30) 36 

Automatic resonance system (/, 31) 36 

Resonant frequencies 36 

Thickness measurements 37 

Alternative arrangements 37 

Calibration Problems 

Xeed for calibration 37 

Basic considerations 37 

Major parameters 38 

Parameters affecting signals in ultrasonic 

pulse-reflection technique i/. 32) 38 

Operator-controlled parameters 39 

Parameters controlled by inspection prob 
lems 39 

Evaluation of parameters 39 

Basic types of test blocks used for cali 
bration (/. 33) 40 

Amplitude- distance data for standard 

contact search units (/. 34) 41 

Beam profile variation with distance (/, 

35) 41 

Comparative amplitude-distance data, 

contact test (/. 36) 42 

Comparative amplitude-distance data, 

contact tests (/. 37) 43 

Comparative amplitude-distance data ; 

immersed tests for four materials (/, 38) 44 
Reflections from immersed steel spheres 

(/. 39) 45 

Amplitude versus area calibration data 

(/. 40) 46 

Calculated signal-amplitude ratios in im 
mersed testing of various metals (/, 41) 46 
Test conditions for Fig. 41 (/. 42) 47 

Ultrasonic Test Features 

Testing applicability 47 

Advantages 47 

Mechanical testing 47 

Instantaneous indication 47 

Volume scanning 47 

Versatility 47 

Safety and convenience 48 

Wide applicability 48 

Sensitivity and directivity 48 

Portability 48 

Ease of operation 48 

High development 48 

Wide acceptability 48 

Capabilities 48 

Established applications 48 

Applications proved feasible 49 

Potential applications 49 

Limitations 49 

Defect detectibility 49 

Coupling and scanning 50 

Economics 50 

Criteria for successful inspection 50 

References 50 




Basic Methods of Ultrasonic Inspection 

spection is usually performed by one of two basic methods. A beam of ultrasonic 
energy is directed into the specimen and (1) the energy transmitted through it is 
indicated or (2) the energy reflected from areas within it is indicated. Inspection 
is accomplished because the ultrasonic beam travels with little loss through 
homogeneous material except when it is intercepted and reflected by discontinu 
ities in the elastic continuum. Fig. 1 illustrates two basic techniques applied to 
internal flaw detection. In Fig. Ha), the flaw is detected by the decrease of trans 
mitted energy at the receiver; in Fig. Ifb) it is detected by energy reflected to the 

ULTRASONIC TEST SYSTEMS. A complete ultrasonic inspection system 
consists of the basic components shown in Fig. 1, including: 

1. Electrical signal generator, G. 

2. Transmitting transducer, or "search unit." 

3. Couplant to transfer acoustic energy to specimen. 

4. Test specimen. 

5. Couplant to transfer acoustic energy to receiver. 

6. Receiving transducer, or "search unit." 

7. Electrical indicator, /. 

Component design and arrangement will depend primarily upon which 
specific characteristics of ultrasonic wave propagation are utilized for detection 
and measurement of specimen properties. The phenomena involved may include: 

1. Velocity of wave propagation. 

2. Beam geometry (focusing field pattern or dual-transducer systems). 

3. Energy transfer (reflection, refraction, or mode conversion). 

4. Energy losses (scattering, absorption). 

Ultrasonic Frequency Ranges. Ultrasonic inspection utilizes high-frequency 
mechanical vibrations for nondestructive testing of materials. Most commercial 
ultrasonic testing is done at frequencies between 1 and 25 megacycles per second 
(Me.). However, applications exist for frequencies as low as 25 kilocycles per 
second (kc.) and as high as 200 Me. Various ultrasonic inspection techniques and 
instruments have been developed to beam ultrasonic energy directly through test 
objects. Low-frequency resonance methods (see section on Natural Frequency 
Vibration Tests) , in which the entire specimen is caused to vibrate at sonic fre 
quency, should not be confused with ultrasonic methods whose probing beam is 
usually restricted to a small fraction of the specimen. 

Ultrasonic Stress Ranges. All mechanical testing methods involve similar 
phenomena described by the fundamental laws of mechanics and acoustics. The 




various methods differ primarily in the frequency and magnitude of the stresses 
developed in the test material. Sonic and ultrasonic nondestructive tests employ 
low-amplitude stresses which do not permanently affect the specimen. Destruc 
tive mechanical tests, such as static physical tests and forced-vibration fatigue 
testing, usually involve high-amplitude stresses. These may cause heating, non 
linear effects, permanent deformation, and eventual rupture of the sample. 





Fig. 1. Basic testing methods, (a) Flaw detected by decrease of energy at receiver, 
(b) Flaw detected by energy reflected to receiver. 

Applications of Ultrasonic Techniques. Because ultrasonic techniques are 
basically mechanical phenomena, they are particularly adaptable to the determina 
tion of structural integrity of engineering materials. Their principal applications 
consist of: 

1. Flaw detection. 

2. Thickness measurement. 

3. Determination of elastic moduli. 

4. Study of metallurgical structure. 

5. Evaluation of the influence of processing variables on the specimen. 

Advantages of Ultrasonic Tests. The desirable features of ultrasonic tests 
include : 

1. High sensitivity, permitting detection of minute defects. 

2. Great penetrating power, allowing examination of extremely thick sections. 

3. Accuracy in the measurement of flaw position and estimation of flaw size. 

4. Fast response, permitting rapid and automated inspection. 

5. Need for access to only one surface of the specimen. 

Limitations of Ultrasonic Tests. Test conditions which may limit the appli 
cation of ultrasonic methods usually relate to one of the following factors: 

1. Unfavorable sample geometry; for example, size, contour, complexity, and 
defect orientation. 

2. Undesirable internal structure; for example, grain size, structure porosity, inclu 
sion content, or fine, dispersed precipitates. 


Development of Tests. The possibility of utilizing ultrasonic' waves for non 
destructive testing was recognized in the 1R30V in Germany by Mnlhau.w, 1 
Trost, 2 Pohlman, 3 and in Russia by Sokoloff, 4 all of whom investigated variou^ 
continuous wave techniques. Flaw detection equipment was eventually devel 
oped, based on the principle of ultrasonic energy interception by n gro?s flaw in 
the path of the beam. This technique later became known MH the through-trans 
mission method. An ingenious transmission system developed by Pohlman pro 
duced shadow-like images of internal flaw?/" Later, .several transmission-type fhvr 
detectors were marketed. 

During this early period, efforts were also made to employ reflected as well it> 
transmitted ultrasonic waves. These were intended to overcome certain limitations 
of the earlier methods, especially the necessity of requirine access to both speci 
men surfaces. Xo practical method was found, however, until Firestone invented 
apparatus utilizing pulsed ultrasonic wave trains to obtain reflections from 
minute defects. 6 ' 7 This development, which he called the "Supersonic Reflecto- 
scope," was aided by the rapid growth of electronic instrumentation techniques. 
It led during the 1940's to the marketing of practical ultrasonic flaw detectors in 
the United States and abroad. In the same period, ultrasonic test equipment was 
developed independently by Sproule in England s (see the section on Double- 
Transducer Ultrasonic Tests). As with early industrial X-ray equipment, the first 
instruments were for the most part considered to be laboratory tools and were 
installed in metallurgical research departments. 

Early Industrial Applications. Production applications were soon found, and 
ultrasonic inspection was applied to critical quality control problems. Among 
the outstanding early applications was the inspection of the first jet-engine rotor 
forgings for internal flaws 9 (see section on Ultrasonic Immersion Test Indica 
tions). In the meantime, fundamental and applied research continued and many 
significant contributions were made. Firestone and his associates at the University 
of Michigan investigated transducer mechanisms, 10 polarized sound using shear 
waves, 11 applications of Rayleigh or surface waves, 12 the Raybender for variable- 
angle inspection, 13 the delay column or "schnozzle" for close-to-surface inspec 
tion, 14 a pulsed-resonance method for thickness measurement, 15 and a variety of 
Lamb or~"plate" wave techniques. 10 Other developments included a frequency- 
modulated resonance thickness gage by Erwin, 17 - 1S improved immersed inspection 
*vstems by Erdman, 19 ' 20 and several ultrasonic visualization or flaw-plotting 
techniques by Sproule, 21 Erdman, 2 -' Wild and Reid, 2 * Howry, 24 and others. (See 
Pringle 25 and Smack 20 ) 

Recent Improvements. Recent developments relate primarily to one or more 
of the following: 

1 High speed, automated inspection systems. 

2. Improved instrumentation for greater resolution of flaw indications. 

3. Better data presentation. 

4. Simpler interpretation. 

5. Studies of fine changes in metallurgical condition. 

6. More detailed analyses of the acoustic phenomena involved. 

Burin* the same period those concerned directly with application of ultrasonic 
inspection techniques made contributions toward its utilization and to the estab 
lishment of procedures and standards, particularly in the aircraft, electrical, and 
nuclear energy fields. 



Generation of Ultrasonic Vibrations 

ULTRASONIC SOURCES. Mechanical vibrations for measurement, anal 
ysis, or test purposes are generated by electromechanical transducers, i.e., ele 
ments having the ability to transform electrical into mechanical energy, and vice 
versa. For ultrasonic inspection at frequencies above 200 kc., piezoelectric 
transducers are used. These employ materials which generate electric charges 
when mechanically stressed, and conversely, become stressed when electrically 
excited. Transducer elements suitably mounted for inspection work are commonly 
called search units, crystals, or probes. (See section on Ultrasonic Transducers 
for details of transducer materials, construction, and characteristics.) 

ULTRASONIC TRANSDUCER TYPES. Transducer materials 27 > 2S - 29 
having the best characteristics for search units are (1) natural quartz crj T stals, (2) 
lithium-sulfate monohydrate crystals, and (3) polarized polycrystalline ceramics 
such as fired barium titanate. Transducer elements which operate as thickness 
expanders are widely used. These produce motion similar to that of an oscillating 
piston and generate compressional waves in the specimen. For special studies re 
quiring transverse waves, crystal cuts can be used which produce shear motion. 
If maximum sensitivity is required, the element is electrically driven at its funda 
mental resonant frequency. 

typical 1-Mc. transducer elements are given in Fig. 2. 30 Values for other fre 
quencies, /, or areas, A, can be found by using the relationship : 

Thickness : 
Capacitance : 

t = U/j 
c = ci X / X A 


where t = thickness, in. 

ti = 1-Mc. thickness, in. 
/ = frequencj', Me. 
c = capacitance, mxf . 
Ci = unit capacitance, M|if/in. 2 . 
A = area, in. 2 . 



Thickness, fc Capacitance, ci Activity Constants* 
(/=lMc.) (Area = 1 in. 2 ) ~ T 

X-cut quartz 


0.1126 in. 9 ji^if 



Y-cut quartz 


0.084 in. Hm-if 

Lithium sulfate 


0,100 in. 22 mif 



Ceramic titanate 


0.1 in. 2000 nuf 



Determined experimentally for typical search unit assemblies. 

Fig. 2. Properties of thickness-mode transducer elements. 

Activity Constant. Activity constants indicate approximate sensitivities of 
elements as acoustic transmitters, S T , and receivers, S R , as compared with X-cut 
quartz. The absolute magnitudes are dependent upon the mechanical loading on 
front and back faces of the transducer, the nature of the applied electric signal, 
and the effective electrical impedances involved. 


Resolving Power and Sensitivity. The resolving power of a search unit is 
directly proportional to its band width (A/), which is primarily a function of the 
damping produced by the mechanical loading on its faces. Conversely, the number 
of cycles required for crystal vibration to reach full amplitude when driven by 
constant a.-c. supply voltage is given by its mechanical Q (the reciprocal of 
band width expressed in percentage) . Typical measured values of A/ and Q deter 
mined experimentally for 1-Mc. ultrasonic immersion search units (i.e., with a 
water load on front face) are given in the accompanying table. 27 - 30 

Transducer Backing A/(Mc.) Q 






Loaded epoxy 




Barium titanate 




Barium titanate 

Loaded epoxy 



Lithium sulfate 

Loaded epoxy 



Since the sensitivity of a given system increases directly with the Q of its com 
ponents (search unit, pulser, and electrical amplifier), a compromise to achieve 
the optimum sensitivity-resolution product is desirable. System Q's of 3 to 10 are 
typical hi commercial pulsed-wave-train flaw detectors. For a given Q, resolution 
increases directly with system frequency. 

Additional Design Requirements. In the design of practical search units for 
various applications, additional requirements to be considered include: 

1. Mechanical: contact area, wear resistance, waterproofing, and connectors. 

2. Electrical: voltages, wave shapes, capacity, and grounding. 

3. Acoustic: noise level, beam divergence, and face plates. 

Construction of four principal types of search unit assemblies is shown in Fig. 3. 

Straight-Beam Contact Units. Straight-beam contact search units, whose 
construction is shown in Fig. 3 (a), have one face of the crystal exposed to contact 
the work These units are widely used in the 0.5- to 10-Mc. frequency range and 
can generally be used on any reasonably flat surface which is electrically conduc 
tive. Typical variations in construction include spring mounting and curved 

Straight-Beam Faced Units. Fig. 3(b) shows the construction of straight- 
beam faced units. These employ thin wear plates to prevent crystal breakage 
and to protect the front electrode which provides internal grounding Facings ol 
quartz, metal, plastic, and rubber have been used. Applications include testing of 
rough surfaces and electrical nonconductors. For rapid testing of large plate, 
a special unit employing a moving plastic belt between the crystal and specimen 
has been developed. 

Angle-Beam Contact Units. Angle-beam search units, which direct the beam 
awav from normal incidence toward selected areas within a specimen, use a wedge 
between the crystal and sample [Fig. 3(c)]. Various wedge materials have been 
used the most successful being methacrylate resin Particularly useful units are 
available which utilize shear or surface waves produced by refraction and mode 
conversion. 32 Most of the commercial angle-beam units operate in the 1- to 5-Alc. 




range. Special types include curved shoes for pipe inspection 33 and variable- 
angle units for axle testing. 34 

Immersion Units. Search units used in immersion testing are separated from 
the test object by a couplant of considerable thickness [Fig. 3(d)]. (See section 
on Ultrasonic Immersion Tests for details of ultrasonic immersion techniques.) 
The crystal mounting must be thoroughly waterproofed and a grounding electrode 
must be provided on the front face. Search units are available for all standard 
test frequencies within the range from 200 kc. to 25 Me. 

Contoured or Focused Units with Acoustic Lenses. The addition of acoustic 
lens elements to the front face of immersion search units makes possible the 
focusing of ultrasonic beams. Cylindrical curvatures permit focusing the sound 
energy to enter cylindrical surfaces normally or along a line focus. Spherical lens 
curvatures focus the sound at a point. (See section on Ultrasonic Transducers for 
further characteristics and applications of acoustic lens systems.) 

Standard Ratings. The frequencies and sizes of search units most commonly 
employed for industrial inspection are shown in Fig. 4. Off-frequency operation 
is sometimes used, particularly at very high frequencies where thick crystals are 
frequently driven at multiple odd harmonics. 

f^TWCJ-J-rt 1 * 

Diameter ** of Active Area (in.) 


Urystai * 

(I"TI ^ 



Contact Faced Angle Beam Immersed 



- 3 



iy Sj 2 - ixi iy s 



%,!% iVs 1X1 %,!% 



%,!% %,!% %XV2,1X1 %,!% 



%,!% % %X%,1X1 %,%,!% 



% - %,% 



- % 



- % 




* Approximately correct for X-cut quartz, lithium sulf ate, and titanate ceramics. 
** Round crystals are generally accepted as industry standards except in angle-beam 
units, for which square or rectangular elements are preferred. 

Fig. 4. Search unit frequencies and sizes most commonly used. 

SPECIAL UNITS. For applications requiring nonstandard units, various 
special styles, sizes, and frequencies have been developed, including: 

1. Dual crystals, common holder. 

2. Large crystals, 1 X 4 in. and larger. 

3. Mosaics, three or more crystals. 

4. Small crystals, %-in. diam, and less. 

5. High frequency, up to 50 Me. (fundamental). 

6. Alternate crystal materials: lithium sulfate, fired titanate ceramics. 

7. Sandwich and "tandem" arrangements. 

8. Y-cut crystals for shear wave generation. 




O -j 





i-iO(N CO O-^OOS 


O G> T-J (N i-I C<l C^' N rH 

O OO lOi If li < 



? *": 1-1 ^ **-. <p 

03 CO T^ 
CO rH i-J 


'I '-I **: '-I CS w t>; oo H . cq oj p i> eo cq i-j cs T-H oq 
co co" co o$ c<i o o co * co c<J co c<i <N co" co csj co cq 

" rH " ' CO CO CO i-l 


co co o i> ^ en 10 j> os 


3 w csi co co co IH" r-i i 

3 CO 00 C 


* 1O ^' ui Ttf' CO "3 Tji l to Tji 


' CO 


S S S d 
"3*3 ^B 

==3 g) 



.8 2 



> s 

c o 







Ultrasonic Wave Propagation Characteristics 

ASSUMPTIONS. The major effects encountered in practical ultrasonic 
inspection can be predicted with reasonable accuracy by making the following 
simplifying assumptions : 

1. The ultrasonic beam consists of plane wave fronts. 

2. Elastic moduli are independent of stress amplitude, frequency, and direction. 

3. When frequency is a parameter, particle motion is continuous and sinusoidal. 

These conditions are generally valid for ultrasonic materials inspection unless 
extremely short or very large amplitude pulses are employed. 

WAVELENGTH. Wavelength, ?., a parameter useful for describing certain 
characteristics of transducers, beam geometries,- and modes of propagation, is re 
lated to oscillation frequency and velocity of propagation as follows: 

X = v/f (3) 

where X = wavelength, cm. (in., ft., etc.). 

v = velocity, cm./sec. (in./sec., ft./sec., etc.). 
/= frequency, c.pjs. 

Values of Jl in several common materials at 1 Me. are given in Fig. 5 for the 
longitudinal mode. 

VELOCITY. Several possible modes of vibration can propagate in solids.^ In 
general, pulse-time techniques indicate the wave packet or "group" velocity. 
This velocity is independent of sample geometry when the cross-section is very 
large compared with beam area and wavelength. The principal wave velocities 
encountered in solids at ultrasonic frequencies are given in Fig. 6. 







Thin rod 

Small diameter bar 





Extended media 
Extended media 



V R 


Semi-infinite free surface 




V P 

Lamb or 

Thin sheet 




Fig. 6. Principal wave velocities in solids. 

Longitudinal waves are dilational or nondistortional, while shear waves involve 
only distortion and are polarized. V L is sometimes called the congressional, plate, 
or bulk wave velocity (not to be confused with plate waves in thin sheets or bulk 
waves in fluids). By definition, V L , V T , and V R are independent of frequency. _ In 
objects where one or more dimensions approach a wavelength, various phasing 
effects can occur. In thin sheet and small tubing, for example, various Lamb or 
'"plate" waves are encountered which have intermediate values of Vp n , varying 
with the test conditions. 12 


Calculation of Acoustic Velocities. To calculate absolute values, a consistent 
set of units must be used. For example, in c.g.s. units: moduli (Y } \i), dynes/ 
cm. 2 ; density p, gram/cm. 3 ; velocity V, cm./sec. Formulas for computing these 
velocities from the density and appropriate elastic "constants," as well as their 
interrelationships, are given in Fig. 7. Like elastic moduli, velocities can be inter 
related as functions of Poisson's ratio (a) only. Their ratios are shown in con 
venient form in Fig. 8. 

Velocity Formulas 
(a) V = ^- 

Thin-rod velocity 

< b) F - = - = Longitudinal velocity 

= Shear 

Velocity Ratio Equations 

M , V _ /(l + o)(l-2g) 

(d) FI~V (1-a) 

t ^ VT 

(e) -TJT = ' 

(f ) ^ = 'A. Bergmann approximation 

Poisson's Ratio Formulas 

2 _ 

Elastic Constants Formulas 
(i) = V T z f> 


Key to Symbols Used: 

Vo = thin rod velocity, cm./sec. 

VL longitudinal wave velocity, cm./sec. 

V T transverse wave velocity, cm./sec. 

Vs = Rayleigh or surface wave velocity, cm./sec. 

Y = Young's modulus of elasticity, dynes/cm. 2 . 

\JL = shear modulus of elasticity, dynes/cm. 2 . 

o = Poisson's ratio. 

p = density, grams/cm. 3 . 
K = bulk modulus, dynes/ cm. 2 

Fig. 7. Relationships between ultrasonic velocities and the elastic constants of 

isotropic solids. 












I I 



I I 





0.2 0.3 


Fig. 8. Relationships of velocity ratios to Poisson's ratio. 



While precise measurement of ultrasonic velocities is possible with present in 
strumentation, values within a few percent are usually adequate for inspection 
problems. One exception is the determination of elastic moduli by velocity meas 
urements, which may require high accuracy for both velocity and density data to 
achieve the precision desired. 

IMPEDANCE. Another useful parameter of a material is its impedance, or 
more accurately, its characteristic acoustic impedance, Z, which is defined as 'the 
product of density p and wave velocity v (usually longitudinal) . Impedance is 
then given by 

Z = pv gram/cm . 2 -sec. (4) 

where p = density, gram/cm. 3 . 
v = velocity, cm. /sec. 

Fig. 5 gives representative velocity, wavelength, and acoustic impedance data for 
typical materials encountered in ultrasonic inspection work. 

EFFECTS AT PLANE INTERFACES. Ultrasonic inspection techniques 
depend upon the principle that the propagation of an ultrasonic beam is influenced 
by the acoustic properties of the media in which it is propagating. Variations 
can produce (1) reflection, (2) refraction, (3) mode conversion, and (4) diffrac 
tion of the beam, or various combinations of these effects. 35 * 36 With extended 
plane interfaces and the assumptions given previously, the beam intensity relation 
ships can be calculated. In addition, attenuation of the beam with distance results 
from losses in common materials, the magnitude being determined experimentally. 
The values obtained for beam intensity are stated in units of either energy or 
power. Relative amplitudes of the particle velocity, displacement, and pressure 
in a given medium can be obtained by taking the square root of the energy ratios. 
These can be related to indications on ultrasonic flaw detectors. Since these em 
ploy piezoelectric transducers, linear amplifiers, and proportional display systems, 
their indications are proportional to amplitude or to square root of intensity. For 
nonidealized conditions, complete analysis of all the effects involved is difficult 
and is all too often neglected, with consequent errors in ultrasonic test interpre 
tation. Typical of such factors are surface roughness, specimen curvature, struc 
ture variations, irregular defect shapes, and nonuniform beam characteristics. 
(See section on Ultrasonic Fields for analyses of -these factors.) 

Reflection. The percentage of incident energy reflected when a beam encounters 
an abrupt change of medium depends upon (1) impedance "mismatch" Z^/Z* and 
(2) angle of incidence. At normal incidence (the simplest case), the ratio be 
tween reflected beam intensity W r and incident intensity W i is given by the reflec 
tion coefficient R, where 


=.(7+1) ) 

The ratio between transmitted beam intensity W t and incident intensity W- is 
given by the transmission coefficient T as 

y-jjj._ 4Z*Zi __T__ 

-Wt' (Z a + Zz) a "-(r+l)* "' - - (6) 

The sum of the reflection and transmission coefficients is unity; i.e., 


In the preceding relations we define 

Zi = pj'i (medium 1) 
Z = p 2 V 2 (medium 2) 

The impedance ratio #2/^1, or mismatch factor r, is a convenient base for plot 
ting the intensity relationships. This ratio is of the order of magnitude of 20 for 
liquids to metals (possibly 80 percent reflection) and about 100,000 for gases to 
metals (virtually 100 percent reflection) . 

Medium 2 





* ~. >* % 





Medium 1 


J 3 3 M 1 6 3 

<53 Z U pq J 2 O 










21 24 18 14 3 1 2 








0.2 0.3 1 9 16 31 








0.8 2 12 19 34 








0.2 7 13 19 








5 10 23 








1 9 
























r xlOO% 



















Oil (xformer) 


* (gm./cm. 2 -sec.) X 10 6 . 

Fig. 9. Percentage ultrasonic reflection coefficients for normally incident plane 
waves at interfaces between various materials. 

Reflection coefficients (times 100 percent) for typical conditions which might 
be encountered in practice are given in Fig. 9. 35 The proportion of transmitted 
energy can then be easily found by Eq. (7). Intimate molecular contact or 
"wetting" of interface surfaces is assumed. 

Eqs. (5), (6), and (7) can be used to determine the results of passing the beam 
through any number of interfaces, provided the layers are thick compared with 
the length of the wave trains employed so that interferences between the effects 
of successive interfaces are avoided. Possible applications include comparison^ of 
test blocks of different materials, evaluation of bonding conditions, and calibration 
of instruments. To avoid errors, it is preferable to utilize energy relationships only 



and to determine amplitude ratios at the transducer as a final step. Angular inci 
dence introduces additional effects described later; however, for very small angles 
of incidence as sometimes used in flaw size or orientation studies, the same gen 
eral results are obtained. For two media having unequal impedances but equal 
velocities, the angles of reflection and of refraction would equal the angle of inci 
dence, with all angles from the normal being equal. Any differences in acoustic 
velocities between adjacent media produce refraction and possible mode conver 
sion, as described subsequently. 

Transmission Through Thin Plates. When the thickness of a layer becomes 
less than the length of the wave train propagating through it, interferences occur 
which violently affect the various amplitudes involved. Long wave trains can 
approximate a steady state or continuous wave condition, producing confusing 
results. Conversely, such resonance effects can be used effectively to measure 




5-^-- 10.6 

O.I 0.2 0.3 0.4 0.5 0.6 0.7 0.8 


Fig. 10. Transmission through thin plate, (a) Schematic of testing set-up. Item 1 
is the water medium. Item 2 is the plate, either aluminum or Plexiglas. Other items 
are: t a , sample thickness; T, transmitter; R, receiver, (b) Relative transmission 
curves. For curve A, 2 /% = 10.6; for curve B, Z 3 /Z l =22, where T P is thickness 

of plate. 


sample thickness, detect unbending, or determine velocity. Several commercial 
instruments have been developed for such purposes. Interferences occur when 
ever multiple reflections within a layer cause superposition of waves having dif 
ferent time relationships. The magnitude and direction of the interference effects 
depend upon the phase angles. The effects are largest when : 

1. Attenuation losses are low. 

2. Layer thicknesses approach a few wavelengths or less. 

3. Surfaces are impedance ^mismatched'' to give poor coupling. 

4. Provision is made for adjusting or "tuning" the frequency of the transmitted 

The effect is illustrated by Fig. 10, which shows the transmission through a 
thin solid sheet, bounded on either side by a liquid, as a function of sample 
thickness. In terms of sample thickness t<> and wavelength k within it, and 
neglecting losses, the relative transmission is given by 

where r = Z s /Zi 

Transmission maxima are repeated at thicknesses of integral half -wavelengths; 
and minima, at odd quarter- wavelengths. A simple adaptation of this technique 
can be used to measure velocity and attenuation in plastic sheet. 37 ' 38 

Resonance Thickness Measurements. A technique for measuring thickness 
from one side of an object, utilizing the resonance effect with pulsed wave trains, 
has been described. 12 The search unit can contact the work piece or be separated 
by a water column. Either constructive or destructive interference can be used, 
depending upon the sharpness of the tuning desired. The method can be used 
effectively for detection of unbonded areas in adhesive-cemented structures. 

Resonance instruments designed primarily for thickness determination are 
usually tunable continuous-wave systems which indicate the loading of the trans 
ducer by the sample. Instrumentation details are given in the section on Ultrasonic 
Resonance Tests. 

REFRACTION. When an ultrasonic beam is incident other than normally 
upon an interface between two materials having unequal sound velocities, the 
transmitted beam will assume a new direction of propagation, given by the angle 
of refraction. In the most general case, additional beams may be generated by 
mode conversion (see Fig. 11). 

The angular relationships for plane-wave propagation are given by an equa 
tion known in optics as Snell's law : 

sin 0i __ sin fa _ sin S / Q x 


The equation gives a necessary, although not sufficient, condition for the existence 
of transmitted or reflected beams at any angle. Critical angles can be deter 
mined by Eq. (10), beyond which no transmitted or reflected beam of a stated 
velocity can exist. 

fc^sin-'CVi/Fs) (10) 

43 -IB 





Pig. II. Refraction and mode conversion at plane interface (angular incidence). 





Fig. 12. Mode conversion by reflection at plane interface. 



Since no energy values are given, the intensities of the refracted beams or of those 
generated by mode conversion must be determined by other relationships. 

MODE CONVERSION. In an extended medium, acoustic energy propagates 
in three principal modes : as longitudinal, shear, or surface waves. Each propa 
gates at a characteristic velocity for a specific material. When a beam strikes an 
interface between materials of different acoustic velocity or impedance properties, 
at other than normal incidence, some energy may be converted to other modes of 
vibration during reflection or refraction. A simple example is reflection of an inci 
dent longitudinal beam on a free surface within a solid, such as the steel block of 
Fig. 12. Since a negligible amount of energy leaves the steel because of the extreme 
mismatch to air, the reflection must be total. However, to satisfy all continuity 
conditions, two reflected beams exist. One beam consists of longitudinal waves 
and the second of shear (transverse) waves, polarized as shown in Fig. 12. At 

+ 1.00 



10 20 30 40 50 60 70 



Fig. 13. Amplitude ratio of reflected-to-incident longitudinal waves. 



large or glancing angles of incidence, the conversion to shear waves is almost total. 
In practical testing, this condition frequently occurs as a result of beam spread 
and produces so-called ghost and subsidiary indications. 12 

Angular and Amplitude Relationships. Angular relationships are uniquely 
given by the reflection equivalent of SnelTs law. However, computation of the 





5 10 15 20 25 30 35 


Fig. 14. Amplitude ratio of reflected-to-incident shear waves. 



energy values requires a set of continuity equations known as Knott's equa 
tions. 39 - 40 Tor a solid-to-air (free surface) condition, these reduce to functions 
of Poisson's ratio only and can be conveniently shown graphically. The amplitude 
ratio between incident and reflected beams as a function of angle for various 
Poisson's ratios is shown in Figs. 13 and 14. 41 Critical angles for minimum, maxi 
mum, or total conversion can be found from these curves of angular incidence. 



General Cases of Angular Incidence. In addition to the simpler interface 
conditions of normal incidence between two media and the angular reflection at a 
free surface, several more complicated but important cases of angular incidence 
arise in practice as follows : 

1. Liquid to solid: incident longitudinal wave only. 

2. Solid to liquid : 

a. Incident longitudinal wave. 

b. Incident shear wave. 

c. Incident surface wave (rare). 

3. Solid to solid: 

a. Incident longitudinal wave. 

b. Incident shear wave. 

c. Incident surface wave. 







Fig. 15. Longitudinal waves in water incident on steel. Power in entire reflected 

and transmitted beams. 



In each case the reflected and transmitted beams may include both longitudinal 
and shear types unless (1) critical angles are exceeded or (2) the medium is a 
nonviscous liquid which can support only longitudinal waves. 

In ultrasonic testing, two relations are usually of interest: first, that giving 
intensity of the beam entering the specimen; and second, that describing the 





Fig. 16. Longitudinal waves in steel incident on water. Power in entire reflected 

and transmitted beams. 

beam returning to the search unit. Any case can be solved by the use of Knott's 
equations, provided all the parameters, including the densities and elastic con 
stants of the media involved, are accurately known. A few specific cases, including 
water-steel, have been completely worked out. 13 The results are shown graph 
ically in Figs. 15, 16, and 17. 

The equations for any specific system can also be solved experimentally with 
a device similar to the optical refractometer. Refraction analysis equipment 42 



developed especially for investigating plastic wedges is shown in Fig. 18. Reason 
ably close correlation with theory has been obtained, even though additional 
effects such as couplant and beam spread are introduced by test conditions. 
Amplitude and angle data for Lucite and steel are shown in Figs. 19(a) and (b), 
respectively. Optimum wedge angles for various applications can be determined 






\ 20 



Fig 17 Shear waves in steel, vibrating in plane of incidence, incident on water. 
*ig. i/. onear reflected and transmitted beams. 

from such data. The experimental approach has certain advantages. Actual test 

the beam spread can be measured, as sh 

DIFFRACTION. When a plane wave of infinite cross-section passes through 
an e, the emerging beam diverge, and also has nonplanar wave fronts. 





Sperry Products, Inc. 

Fig. 18. Refraction analyzer. 

These effects result from wave interference or diffraction, as in optics. Similarly 
the beam pattern from a transducer of finite area or the reflection from a small 
discontinuity will have a diffraction pattern. (The theory of this phenomenon and 
some of its practical implications are discussed in the section on Ultrasonic 
Fields.) A complete analytical solution of a typical flaw detection situation is 
usually impractical because of complex geometry and phasing conditions encoun 
tered. Three important effects can be described qualitatively: 

1. Transducer beam spread. 

2. Beam profiles. 

3. Defect reflection factors. 

Their importance in practical ultrasonic inspection cannot be overestimated. 
They relate to instrument calibration, flaw evaluation, optimizing scanning paths, 
and explanation of spurious reflections. In particular it is necessary to understand 
the characteristics of the ultrasonic beam referred to as the "near field" and 
"beam profile" effects (see section on Ultrasonic Transducers) . 

ATTENUATION. Elementary plane-wave theory usually assumes no trans 
mission losses other than interface effects. However, energy losses do occur in 
all materials to some extent and must be considered in certain aspects of ultra 
sonic inspection. 27 * 29 35 > 43 For plane-wave propagation, the attenuation constant 
a is given by 


where a attenuation constant, nepers per unit distance (e.g., cm.). 
Y = ratio of intensities at two points a unit distance apart. 



Alternate forms sometimes used and their conversions are: 
a' = 8.68a (db./cm.) 
Ai = }.a (nepers/ wavelength) 
A = 8.68Xa (db./wavelength ) 

The attenuation a is a difficult quantity to measure accurately, particularly in 
solid materials, such as the common structural metals, at the test frequencies 
usually employed, i.e., 1 to 10 Me. The over-all result usually observed includes 
other loss mechanisms such as beam spread, field effects, couplant mismatch. 
transducer loading, or sample geometry. These factors cannot be isolated without 
special techniques and equipment. 28 ' 29 - 44 



EES (0 2 ) 



2.25 Me 


5 10 

20 30 1 40 50 

60 70 


51020 30 40 50 60 70 


Sperry Products, Inc. 

Fig. 19. Refraction data, (a) Amplitude data, (b) Angle data. 



For most flaw detection work it is helpful to know the approximate "penetra 
bility" of the specimen in terms of a similar reference sample of a known material. 
Large differences in attenuation are indicated as sizable changes in the reflection 
obtained from the far face of a flat specimen, i.e., its fl back reflection." 7 ' 45 The 
technique is therefore commonly known as the loss-of-back-reflection method 
and is sometimes used to show unrefined grain structure in forgings. In such 
materials an attenuation increase is usually accompanied by a noticeable rise in 
random reflections or noise from scattering surfaces, probably caused by larger 
grain faces, excessive porosity, or numerous inclusions. These interfering reflec 
tions usually limit the upper frequency that can be used practically for testing 
a -given material. Quantitative studies of attenuation phenomena offer many 
potentialities both in theoretical and applied fields. Some of the interrelationships 
of interest are shown in Fig. 21 , 46 



Sperry Products, Inc. 

Fig. 20. Beam spread. Amplitude of refracted shear wave as a function of receiver 


Measured values of attenuation in some typical materials, extracted from 
several publications, are shown in Fig. 22 with their probable frequency relation 
ships. 29 

DISTURBING INFLUENCES. The conditions encountered in practical 
inspection may depart from those of plane-wave theory in several aspects. This 
makes it quite difficult to predict quantitatively the amplitudes of the various 
signals obtained from given specimens. The major effects relate to one or more 
of the following: 

1. Beam spread due to finite transducer size. 

2. Attenuation of beam due to such causes as scattering and damping. 

3. Beam nonuniformity due to field effects. 



M J 























stalline Mater 

rystals with p 


i ?" v 




deformation ir 



in steel embri 




o -S 

II 1 














4. Transducer loading by couplant or specimen. 

5. Defect reflection factor related to such factors as shape, size, surface, and 

6. Masking "noise" from normal discontinuities. 

7. Instrumentation variables. 

Considerable experimental work and some analytical study have been applied 
to isolate and evaluate the parameters involved. (Some of the effects are dis 
cussed in the sections on Ultrasonic Transducers and Ultrasonic Fields.) 

Frequency (Me.) 

1.0 2.5 5 10 25 

Aluminum, 17ST (0.13 

mm. grain) - <0.01 0.02 0.12 1.06 

Aluminum, 17ST (0.23 

Glass, Corning 012 













Magnesium (0.21 mm. 







Magnesium (2.0 mm. grain) 






Neoprene, GN 












Quartz, fused 






Carbon tetrachloride 












Water, distilled , 






Air (dried) 






Fig. 22. a', Attenuation (db./cm.). 29 

In general, existing theory is not directly applicable to solution of the condi 
tions encountered in practice. Its refinement to provide for all the variables 
involved is probably unwarranted. Beyond the effects which can be correlated 
with plane-wave and elementary transducer theory, an empirical approach will 
probably be necessary for some time, particularly with respect to those variables 
affected by the frequency of the ultrasonic waves involved. For these reasons, 
calibration of inspection equipment for critical applications is usually per 
formed experimentally with a set of test blocks simulating the range of conditions 
expected (material, distance, defect type and size, etc.), as described in this section 
and in the section on Ultrasonic Immersion Tests. 

Basic Instrumentation and Techniques 

CLASSIFICATION OF EQUIPMENT. Equipment for ultrasonic inspec 
tion of solid materials usually falls within one of three categories, based on the 
parameter indicated: 

1. Amplitude of transmitted energy only. 

2. Amplitude and transit time of transmitted or reflected energy. 

3. Loading of transducer by specimen. 

Other arbitrary subdivisions are possible, and in some techniques, combination 
effects must be considered. 



using only amplitude information depend on the principle that certain specific 
changes in the sample will produce significant changes in the intensity of an 
ultrasonic beam passing through it. 35 













1 1 



1 i 








I 1 

L J 






H h 









1 RATE 1 

1 GENERATOR | *"" 


Fig. 23. Types of through-transmission systems. 

Equipment. Structurally, the equipment need consist only of a sound source, 
receiver, sample, and suitable couplant, as shown in Fig. 23. Usually, however, 
a scanning mechanism and recording or alarm device are required, since the 
ultrasonic beam area is likely to be much smaller than that of the sample cross- 

Defect Detectibility. As in any "shadow-casting" situation, 47 the defect 
detectibility of this system depends principally on: 

1. The ratio of defect area to beam size. 

2. The separation between defect and the transducers. 


Limitations. In addition to these limitations, other problems may be involved, 
such as spurious signals from multi-path reflections, amplitude variations due 
to minor geometry changes, undesirable resonances of sample or couplant, and 
direct electrical cross-talk between transducers: If contact coupling is used, 
pressure effects are large. If sample immersion is used, standing waves can 
sometimes occur when continuous waves interfere after reflection. To minimize 
such variables, the electric signal applied to the transducer is usually frequency- 
modulated (wobbulated) or else pulsed (chopped). In this way resonances are 
averaged out and standing waves are reduced. Several possible configurations of 
such equipment are shown diagrammatically in Fig. 23 (a) through (e). 

Advantages. Advantages of the method include simplicity of equipment, ease 
of interpretation, and noncritical sample alignment. Although this was the 
earliest known method, only a few of the several thousand ultrasonic inspection 
installations now in world use are known to employ the transmission technique. 
A large percentage of these are modified pulse-time instruments. Where opti 
mum results are desired, it is usually preferable to employ a gated pulse-time 
system adapted to transmission indication. This retains the basic simplicity of a 
transmission test without the disadvantages of nongated systems. 

Applications. The principal applications described in the literature relate 
to detection of the following flaws : 

1. Lack of bonding in clad fuel elements. 48 ' 49 

2. "Cupping" in drawn stock. 50 

3. Flaws in grinding wheels. 

4. Laminations in rubber tires. 51 

DICATION. The most versatile techniques for ultrasonic inspection of solid 
materials involve measurement of two parameters simultaneously: 

1. The amplitude of signals obtained from any internal discontinuities. 

2. The time required for the beam to travel between specific surfaces and these 

Time Measurement Systems. Three basic time-modulated measurement sys 
tems can be used: (1) pulse time, (2) frequency modulated (FM), and (3) res 
onance or phasing. Each has certain advantages and limitations. However, at 
the present time almost all instruments used for the detection of' small defects 
are pulse-time devices, while most thickness-measurement units use 'resonance. 
FM techniques have been limited to special applications. 20 Most resonance 
equipment also employs the technique of indicating the loading of the transducer 
by the work piece. 

Equipment. The basic components of a pulse^time. flaw detector 7 are shown 
in Fig. 24. Similar techniques are employed in some radar and depth-sounding 
equipment, although the problems relating to interpretation, resolution, and fre 
quency range may be very different. A short electric pulse is generated and 
applied to the electrodes of the search unit. This produces a short, train of ultra 
sonic waves which are coupled into the specimen. Timing circuits then measure 
the intervals between the transmittal of the initial pulse and the reception of 
signals from within the specimen. This cycle is repeated at regular periods so 
that an essentially continuous indication is obtained. The pulse repetition rate 
is made low enough so that reverberations within the specimen decay completely 
between pulses. Several types of instrumentation -are possible, depending prin- 



cipally on the methods of timing and indicating. Many application techniques 
can also be used, depending on the number and style of search units employed, 
the frequency range covered, and the means for coupling or scanning. In addition, 1 
numerous variations and combinations of circuit components can be incorporated 

Fig. 24. Basic pulse-reflection system. 

into the equipment. However, only those in common use will be described. A 
convenient breakdown of basic types relates primarily to the method of present 
ing flaw information, 25 as shown in the table here. 


Means of Presentation 

Indicates Principally 

A scan 

CRO screen 

Flaw depth and amplitude of 

flaw signal. 


CRO screen 

Flaw depth and flaw distribu 

tion in cross-sectional view. 


CRO screen 

Flaw distribution in plan view. 

Gated systems 

Electrical signal for alarm, 

Determined by technique. 

marker, facsimile, or 

chart recorder 

A-Scan Presentation. Most instruments in use have a basic A-scan presenta 
tion of the type shown in Fig. 25. The horizontal base line on the cathode-ray 
(CRO) tube screen indicates elapsed time (from left to right) and the vertical 
deflection shows signal amplitudes. For a given ultrasonic velocity in the speci 
men, the sweep can be calibrated directly in terms of distance or depth. Con 
versely, when the dimensions of the sample are known, the sweep time can be 
used to determine ultrasonic velocities from which elastic moduli can be calculated. 
The signal amplitudes represent the intensities of transmitted or reflected beams. 
These may be related to flaw size, sample attenuation, beam spread, and other 








"" RATE 1 

1 F 


Sperry Products, Inc. 
Fig. 25. Functional diagram of pulse-reflection A-scan equipment. 

Techniques. As shown in Fig. 26, several techniques can be used. These 
include through-transmission or reflection, single or double search-unit systems, 
and contact or immersion coupling. Other possibilities include angle-beam and 
surface-wave techniques using single or dual search units, and various combina 
tions of methods. In the United States, single search-unit operation is used for 
most reflection-testing applications, as represented by Fig. 26(c), (d), and (e). 
The two-transducer technique is more widely used in Great Britain and Europe! 
(See section on Double-Transducer Ultrasonic Tests.) 

Distance or Time Scale Markers. To permit easy reading of flaw depth scale 
markers are usually provided. These can be adjusted, using test blocks, so that 
each space represents some desired distance in the material (such as 1 in ) The 

BCD A r*A E 






Fig. 26. Basic testing methods. 

time intervals for markers or sweep length can also be computed from the known 
velocities in the materials of interest. The values in the table here are approxi 
mately correct for the total reflection time in typical materials. 


Longitudinal Waves Shear Waves 


40 imsec /in 


20 jisec./in. 

40 jisee./in. 


10 jisec /in 

20 jisec /in. 

Limitations. A wave train consisting of 10 cycles at 1 Me. in a metal occupies 
1 in. of sweep time. It will partially obscure any flaw-reflection indications 
occurring at the same instant. Similarly, if the amplifier circuits have a recovery 
period of 10 [xsec. after a large signal, as much as 1 in. of material may be hidden. 

The transducers and amplifiers used in typical systems tend to "ring" or to 
"stretch" the signals even when the initial pulse contains only a single cycle. The 
resolution of such systems, i.e., the ability to detect a small signal immediately 
after a large one, is limited to about 5 to 10 cycles, depending upon the absolute 
signal amplitudes involved. Values of wave-train length in aluminum for stand 
ard test frequencies and several pulse periods are given in Fig. 27. 

Sensitivity. The absolute sensitivity required varies greatly with the applica 
tion. In practical testing, the sensitivity of commercial instruments is usually 
adequate to detect the smallest defects of structural concern such as cracks, inclu- 


sions and porosity, although other limitations such as resolution may exist. Rea 
sonable amplifier linearity is desirable for calibration and flaw comparison 
purposes. Reading accuracies- of 1 part in 20 are usually sufficient, since dis 
crepancies due to other variables such as coupling: and alignment predominate. 
Precision markers or high sweep linearity are ordinarily not required for flaw 
detection, although one or the other may be necessary for accurate thickness or 
velocity measurements. In some cases precision markers are provided as special 

Pulse Frequency 

Pulse Length in Cycles 






0.59 in. 

2.95 in. 

5.90 in. 





































Fig. 27. Testing distance obscured by pulse (interface signal or initial pulse). 

Thickness of a material in inches having a velocity of 2.36 X 10 5 in./sec. (6 X 10 3 

cm./sec.) equivalent to pulses of various lengths and frequencies in echo testing. 

Alternative Arrangements. Among the possible alternative constructions for 
circuit components are the following: 

1. Synchronizer (rate generator) 

a. Pulse repetition rate: fixed at 60 c.pjs., 400 c.p.s., etc. Adjustable (fixed 
range). Stepped (with sweep range). Variable (50 c.p.s. to 1000 c.pjs.). 

b. Locking signal: line voltage (or harmonic) ; internal. 

2. Pulser 

a. Wave shape: impulse, spike, gated sine wave, damped wave train, etc. 

b. Type: tunable, impedance matched, variable amplitude, etc. 

c. Circuit : thyratron, hard tube, pulsed oscillator. 

3. Amplifiers 

a. Type: tuned radio-frequency, wide band, superheterodyne. 

b. Response : linear, sharp cut-off, logarithmic, etc. 

c. Sensitivity: time variable (TVG) or constant. 

d. Controls: gain, input attenuation, reject, variable band width, etc. 

4. Signal Display 

a. Type: RF wave train, video. 

b. Source: RF output, IF output, rectified (envelope), differentiated video, 

5. Sweep 

a. Type: logarithmic, high linearity, conventional. 

b. Delay: adjustable, automatic, none. 

c. Expansion: fixed, adjustable, related to sweep. 


6. Markers 

a. Type: fixed scale on CRO, precision electronic, adjustable square wave, 
movable step mark, etc. 

b. Source: crystal oscillator, adjustable multi-vibrator, precision integrator, 

c. Display: superimposed on signals, alternate sweep, separate trace, intensity 
modulated, etc. 

7. Signal Gates 

a. Types: amplitude proportional, fixed level alarm. 

b. Output: D-C level, modulated, rectangular wave at PRR, pulse stretched, 

8. System 

a. Frequency range: single, selectible, continuous tuning; low (50-200 kc.) ; 
intermediate (200 kc.-5 Me.) ; high (5-25 Me.). 

Special Features. Special instrument features required for some applications 
may include the f ollowing : 

1. Provision for two-crystal operation. 

2. Interference elimination circuitry. 

3. Compensation for long search-unit cables. 

4. Stabilization for extreme line-voltage changes. 

5. Exponential calibrator for attenuation measurements. 

6. Exceptional portability for field use. 

7. Remote indicators. 

8. Extended frequency range (for example, above 100 Me.). 

9. Ultra-high resolution. 

B-SCAN PRESENTATION. When the shape of large flaws or their dis 
tribution within a sample cross-section is of interest, the B-scan display can be 
useful. In addition to the basic components of the A-scan unit, provision must be 
made for these additional functions: 

1. Intensity modulation or brightening of the CRO spot in proportion to the 
amplitude of the flaw signal. 

2. Deflection of the CRO trace in synchronism with the motion of the search unit 
along the sample. 

3. Retention of the CRO image by use of a long-persistence phosphor. 

A system for producing a section view in rectangular coordinates is shown in 
Fig. 28. Other possible scanning procedures for specific applications are described 
in the literature. 22 * 23 > 24 52 

Equipment. Often a B-scan display is used in conjunction with A-scan inspec 
tion or as an attachment to standard A-scan equipment. Therefore the design 
criteria depend greatly on that equipment and the application. Where high 
speed scanning is required, the longer persistence of the B-scan display may be 
an advantage to the operator. 

Defect Detectibility. The effectiveness of the B scan in showing flaw detail 
depends upon the relationship of flaw size, beam area, and wavelength. Optimum 
results are obtained with larger flaws, smaller crystals, and higher frequencies. 
For other conditions, beam-sharpening techniques such as focusing and elec 
tronic contrast enhancement may be required. 



Alternative Arrangements. Presentations similar to the B scan, but not em 
ploying CRO displays, have also been used. One form uses a matrix of small neon 
lamps somewhat in the manner of an animated billboard. 30 It has high brightness 
and permanent memory but very limited resolution. To provide reasonable detail 
for display of flaw distribution, several hundred lamps were used. In certain 
applications, photographic film or chart recorders have been used for image 
formation. 21 



Fig. 28. B-scan presentation. 

C-SCAN PRESENTATION. By synchronizing the position of the CRO 
spot with the search-unit scanning motion along two coordinates, a plan view of 
the sample can be developed similar to the common PPI (plan position indi 
cator) radar display. 25 - 52 

Equipment. In addition to the circuitry required for a B scan, provision must 
be made for eliminating unwanted signals such as the initial pulse, interface echo, 
or back reflection, which would obscure internal flaw signals. An electronic gate 
is used to render the display circuits sensitive only for the short intervals of sweep 
time when signals from the desired depth range occur, as shown in Fig. 29. 
Because of the relatively long period required to scan a large area, conventional 
CRO tubes do not have adequate persistence for practical ultrasonic C scans. 
Chart recorders, facsimile techniques, and the newer memory tubes have been 
found more satisfactory. Each has certain limitations in the detail that can be 
displayed, the maximum scanning speed, size of plot, and other factors. 

Alternative Arrangements. In certain cases hybrid systems which present 
some information both as to flaw size and location have been used with some 
sacrifice in flaw shape and position detail. 53 The C-scan display can also be 



developed with through-transmission techniques and appropriate scanning and 
indicating methods. 48 - 54 

Gated Systems. The necessity of gating or selecting certain portions of the 
sweep period for a C-scan display was mentioned previously. In general, gating 
provisions are required for all automatic systems which alarm, mark, record, 
chart, or otherwise replace visual interpretation. Such gating circuits may be 
built into the flaw detector or supplied as separate attachments. 

H I 1 1 'I ' 


Fig. 29. C-scan presentation. 

Commercial recording attachments, or monitors, usually provide at least two 
gates, one to indicate the presence of flaws within the sample and the second 
to show a decrease of back reflection. Some units provide additional flaw gates 
tfo that two or more alarm levels can be set or so that different depth increments 
can be examined, 25 ' 26 55 

If the cross-section of the specimen varies during the scanning cycle in an 
automatic test, the gating periods must be simultaneously adjusted, a technique 
known as programming. In addition, other functions such as sensitivity, search- 
unit angle, and recorder speed may have to be controlled. One system intended 
for the inspection of large gun tubes having a variable cross-section uses a 
pre-set program controlled by stepping relays? which are in turn set by the 
location of the search unit in the gun tube. r>f{ One installation developed for the 
automatic inspection of contoured jet engine forgings employs a punched tape to 
control each of the six variables involved. na 


most common instruments utilizing transducer-loading effects are those which 
employ resonance effects to indicate sample thickness. 



Equipment. The basic system is shown in Fig. 30. By the proper selection of 
components, it is possible to tune through one or more resonances from which thd 
sample thickness can be determined. 17 ' 27> 57 



Fig. 30. Manual resonance system. 

A more elaborate system which automatically sweeps the frequency and dis 
plays thickness indications on a CRO screen is shown in Fig. 31. (Typical 
instruments and their uses are described in the sectio'n on Ultrasonic Resonance 
Tests.) Calibration is usually performed experimentally with known samples, 
although appropriate frequency-thickness scales can be calculated. The crystal, 
couplant, and sample comprise a composite transducer whose electrical impedance 
can be calculated as a function of frequency. 58 At resonance the impedance be 
comes minimum, causing an increase of loading on the source. 



Fig. 31. Automatic Resonance system. 

Resonant Frequencies. The lowest resonant frequency of the system, / 5 , is 
given approximately by the following conditions, provided the crystal is operated 
near its 'fundamental frequency: 

i^T (12) 



where t = thickness, 
fc = tV/i. 

/i = lowest resonant frequency 
v = longitudinal velocity, 

Certain conditions may produce spurious signals giving irrelevant indications, 


An infinite series of higher-order harmonic resonances is also possible, as 
given by 

= (15) 

where tai = ki/n. 

n 1,2, 3, etc. 
j n = successive resonances. 
v = longitudinal velocity. 
t = thickness. 

Conversely, by measuring the frequencies of two successive harmonics, that of the 
fundamental can be obtained as 

A = /(n + u - in = A/ (16) 

Thickness Measurements. Test-object thickness is given by 

v _ nv _ v 
" 2t* " 2/n " 2 (A/) U7; 

If the scale is calibrated in thickness instead of frequency (based on ^ = 
successive thickness-scale readings can be used, as follows: 


where t a and 4 6 are scale readings for successive thickness indications, t ft > t fl . 

Alternative Arrangements. Modifications of the technique have been devel 
oped which display the impedance characteristic of the specimen, 27 - fit) one applica 
tion being for the inspection of adhesive-bonded aircraft components. 

Calibration Problems 

NEED FOR CALIBRATION. Interpretation of indications obtained from 
metallurgical discontinuities by various ultrasonic inspection methods is discussed 
in several subsequent sections. Determination of their seriousness is usually out 
side the scope of nondestructive testing. To permit such an evaluation, however, 
appropriate calibration techniques and reference standards are necessary. 
This is particularly true in the application of pulse-reflection instrument types 
because of the large number of variables possible. 

BASIC CONSIDERATIONS. There is no equivalent of the X-ray pene- 
trameter in ultrasonic testing. In most ultrasonic testing, only an indirect 
calibration procedure is practical. This usually involves comparison with indi 
cations from reference blocks containing drilled test holes, or in a limited 
number of cases, from reference surfaces within the specimen. Neither method, 
however, provides a direct calibration of system response to flaws which may be 
located anywhere within the specimen cross-section. Except for certain types of 
curved specimens, the signals obtained from the front face (interface reflection) 
or back surface (first back reflection) are usually of far too great amplitude for 
use as a reference during testing at high sensitivity for small defects. The use of 
small reflecting surfaces placed at some point in the beam path, but external to 
the sample in the manner of X-ray penetrameters, has only limited advantages 
and is usually impractical except for certain conditions in immersion testing. At 



the present time, therefore, calibration of instruments is done principally with 
sets of special reference blocks intended for specific applications. 30 - fll 62 

Several investigations of calibration factors have been undertaken to provide 
correlation factors for interrelating various parameters. An interesting and useful 
result has been a proposal to use small steel spheres of % to 1 in. diam. as a basic 
sensitivity reference for immersed testing and to check secondary standards such 
as test blocks. . 

The required accuracy of calibration can vary greatly with the application. 
In some cases only a crude check on instrument performance is necessary; in 
others, only reference settings, so that results can be duplicated later. When 
acceptance of material is based on rigid ultrasonic test specifications, consider 
able attention must be given to calibration techniques. Fabrication of duplicate 
sets of reference blocks required for such applications can prove to be a serious 

MAJOR PARAMETERS. The components of an ultrasonic-testing system 
and their principal variables are shown in Fig. 32. Some of the characteristics 















Fig. 32. Parameters affecting signals in ultrasonic pulse -reflection technique. 


of such a system can be anticipated from fundamental principles of ultrasonic 
testing. Others may be related primarily to: the specific components used in 
the instrument or search units; wave shape; amplifier recovery time; and trans 
ducer field patterns (see section on Ultrasonic Transducers). 

Operator-controlled Parameters. The major parameters controlled by the 
operator relate to: 

1. Equipment selection 

a. Instrument type. 

b. Search unit-type and size. 

2. Equipment operation 

a. Technique. 

(1) Coupling method. 

(2) Scanning sequence. 

(3) Peaking procedure. 

b. Control settings. 

(1) Frequency. 

(2) Pulse length. 

(3) Linearity. 

(4) Distance/time-varied gain (TVG) or sensitivity-time gain (STG). 

Parameters Controlled by Inspection Problems. Those parameters deter 
mined by the specific inspection problem involve: 

1. Specimen properties 

a. Velocity. 

b. Impedance. 

c. Geometry. 

d. Surface. 

e. Attenuation. 

f. Noise level. 

2. Flaw conditions 

a. Depth. 

b. Size. 

c. Shape. 

d. Impedance. 

e. Orientation. 

EVALUATION OF PARAMETERS. Numerous investigations have been 
undertaken to isolate and evaluate one or more of the many factors involved in 
calibration. Early work at the Aluminum Company Laboratories, relating to 
distance-amplitude factors and to the fabrication of reference blocks, is note 
worthy. 63 Projects under U.S. Air Force sponsorship have yielded considerable 
data applicable to many inspection problems. 30 * 61 

Several typical examples from work sponsored by the Air Materiel Command, 
U. S. Air Force, illustrate these results. Fig. 33 describes the several types 
of test blocks used during the project. Fig. 34 presents normalized ampli 
tude-distance data (semi-log plot) in aluminum for several standard contact 
search units and test frequencies on the Type A-l test blocks of Fig. 33. Fig. 35 
represents the beam profile for one of these curves, i.e., the relative amplitudes 
obtained as the search unit traverses a flaw at different depths. 

Fig. 36 shows amplitude-distance data (log-log plot) for several test hole 
diameters. Fig. 37 shows amplitude-distance data (log-log plot) for several 
different test hole shapes as shown in Fig. 33. 















Fig. 33. Basic types of test blocks used for calibration. 


43 ' 



.4 .6 .8 1 

2 4 6 8 10 20 40 60 80 100 200 


Fig. 34. Amplitude- distance data for standard contact search units. 


! i ' : i . 

l'/2 1 3 /4 2 2'/2 3 3'Xg 4 


5 6 8 12 


I I I I I 

1/4 1/2 3/4 1 

Fig. 35. Beam profile variation with distance. 5 Me., ! 3 /s-in. diam. contact search 











i 3 ' 

j 20 







/> 4 ^ 


V \ 












\ s 







o * 



i j 

i < 

8 5 





- X^* 









\ \ 


r" y 













.2 .3 .4 .5 . 2 345678 

910 20 30 4051 



Fig. 36. Comparative amplitude-distance data, contact test, 5 Me , %-m diam 
search unit data for &-, % 2 -, % 2 -, %a-, and % 2 -in. diam. test holes. 






































i > 



, ** s 

















2 ,3 . 

4 1 .5 .6,7. '.9' .0 2 3 4 5 6 78910 20 30- 40 DO o 

Fig 37 Comparative amplitude-distance data, contact tests, 5 Me, P/s-in, diam. 
search unit, for various hole shapes. (Flat-bottom, F; cylindrical, T; conical, C; 

spherical, S,) 


















^ A > 



& ft 
3 7 

i , 





? 6 








f, * 





55 4 


^ A 





2: 3 

K . 

















234.. .8.91.0 2 3 4 5 6789 10 20 30 40 50 60 

Fig. 38. Comparative amplitude-distance data; immersed tests for four materials. 
Magnesium, M; aluminum. A; titanium, T; steel, S. 





5 10 



3 - 


2 1/4 MC, 1 1/8, 12 1/4 IN. WATER PATH 






o 2 - 











CL j , 



* 5 MC 3/4, 12 IN. WATER PATH 







* >. 

- 10 MC 3/8, 6 IN. WATER PATH 


Fig. 39. Reflections from immersed steel spheres. 



25 MC 
20 MC 

S 15 MC 





i 10 MC 



I 4 9 16 25 36 49 
123456 7 8 

5 MC 


3/16" 3 '8" 3/4" 


Fig. 40. Amplitude versus area calibration data. Immersion test on Alcoa reference 

blocks, Series A. 

Fig. 38 shows amplitude-distance data (log-log plot) for several typical aircraft 
materials. Fig. 39 shows the amplitude of a calibrating signal obtained from an 
immersed steel sphere. Fig. 39 fa) gives the amplitude-distance relationship and 
Fig. 39 (b) gives amplitude-ball-diameter data for several frequencies at the close 
field limit. Fig. 40 presents amplitude-area data (linear plot) for one set of 
Type A-l blocks at several test frequencies. Fig. 41 gives the calculated ratios 
between various reflection amplitudes for various materials immersed in water. 
Test conditions are shown in Fig. 42. 


Case 1 (Water Backed) 


2 (Air Backed) 





















(1 - R) 










.12 ! .94 













.28 .85 










_ /z.-z. 

Fig. 41. Calculated signal-amplitude ratios in immersed testing of various metals. 







* t * 

A 1 * 

t t 















1 t 

A \ 





Fig. 42. Test conditions for Fig. 41. 

Ultrasonic Test Features 

TESTING APPLICABILITY. The features of ultrasonic technique? with 
respect to their applicability to nondestructive testing can be evaluated in terms 
of their advantages, capabilities, and limitations. 

ADVANTAGES. Since only low-amplitude, inaudible sonic waves are util 
ized, the specimen is not damaged by the test. 

Mechanical Testing. Ultrasonic waves are mechanical vibrations, and there 
fore ultrasonic inspection is especially suited to detection of elastic discontinuities 
and measurement of physical properties such as flaws, porosity, structure, and 
elastic constants. Optical, magnetic, chemical, and other properties are not 
ordinarily indicated. In order of detectibility, the discontinuities usually indi 
cated are geometrical, gross flaws, minute defects, and finally structure. 

Instantaneous Indication. Instrumentation is basically electronic, with indi 
cations being obtained almost instantaneously. This characteristic permits rapid 
scanning of parts, with automatic positioning, plotting, and alarming. 

Volume Scanning. The ultrasonic beam almost instantaneously searches a 
complete volume of material extending from the front to the back surface of the 
specimen. Each incremental scan of volume requires only a fraction of a second. 
With a 1-in. square search unit and a pulse repetition rate of 60 c.p.s., a steel roll 
10 ft. long could be tested at the rate of a quarter-ton per second. Since the 
instrument response time is negligible, practical inspection speeds are determined 
by such factors as the scanning mechanism, handling equipment, human reaction 
time, and pulse repetition rate. 

Versatility. The ultrasonic method permits inspection of a wide range of 
samples as to size, geometry, area of flaw, and other requirements. Basically it 
detects internal, hidden defects which sometimes may be deep below the surfaces. 
Search units are available to generate waves of several types, including longitu- 


dinal, shear, and surface. Applications range from thickness measurements of thin 
steel plate to internal inspection of 100-ton turbine rotors. 

Safety and Convenience, There is no hazard to the operator or to nearby 
personnel. Access to only one surface is necessary in most cases. 

Wide Applicability. Most nonporous, resilient materials used for structural 
purposes (such as steel, aluminum, titanium, magnesium, and ceramics) can be 
penetrated. Even large cross-sections several feet thick can be tested successfully 
for minute flaws. 

Sensitivity and Directivity. The use of a high frequency, well-defined beam of 
sound permits detection and location of the smallest flaws of structural im 
portance. Disc-like defects or cracks of almost zero thickness are readily detected. 
Depth can be indicated to within a small fraction of an inch, even in thick speci 
mens. The beams used have searchlight directivity, being confined to cones of 
low divergence by their short wavelengths. Angle-beam techniques permit direc 
tion of the beam toward any area of the object. 

Portability. Some equipment can be used in shop, laboratory, warehouse, or 
field, thus permitting in-place inspection and requiring only moderate power from 
the a.-c. line or a small generator. 

Ease of Operation. Operator training can be simple and brief for many 
specific applications where the problem is clear-cut and the technique is estab 
lished. (Yet the method remains versatile and has flexibility for the solution of 
new and more advanced problems. Most operators soon find the technique inter 
esting and challenging.) 

High Development. Several years of development have resulted in the avail 
ability of several basic instruments, many application techniques, considerable 
knowledge, and a wide assortment of accessories, search units, fixtures, and other 

Wide Acceptability. Ultrasonic inspection techniques are widely accepted 
throughout industry. Ultrasonic tests are used for quality control and materials 
inspection in the major heavy industries, including electrical manufacturing, pro 
duction of steel, aluminum, and titanium, in the fabrication of airframes, and in 
jet engine manufacture. Ultrasonic testing for preventive maintenance is used for 
in-place inspection to prevent failure of rolling stock axles, press columns, earth- 
moving gear, mill rolls, mining equipment, and other machines and components. 
Various technical societies and standards committees have recognized the necessity 
for ultrasonic inspection and are preparing test procedures, acceptance standards, 
and other essential data. 

CAPABILITIES. A variety of classifications could be utilized to tabulate the 
principal applications. The division outlined here is intended to suggest rather 
than to present details. 

Established Applications. Equipment now in everyday use is mainly con 
cerned with (1) flaw detection and (2) thickness measurement. Examples of flaw 
detection, listed according to end use, are: 

1. Mill components: rolls, shafts, drives, and press columns. 

2. Power equipment: turbine forgings, generator rotors, and pressure piping. 

3. Jet engine parts: turbine blanks and compressor rotors. 

4. Airframe components: forging stock and frame sections. 

5. Machinery materials: die blocks, tool steels, and drill pipe. 


6. Rolling stock: axles, wheels, and crankpins. 

7. Track maintenance: rail joint areas. 

8. Nuclear reactors: clad fuel elements and heat exchanger Tubing. 

The flaws to be detected may be voids, cracks, inclusions, segregations, lamina 
tions, bursts, flakes, or other types. They may be inherent in the raw material, 
result from fabrication and heat treatment, or occur in service from fatigue, 
corrosion, and other causes. Testing may be performed manually or automatically, 
by immersion or by contact, depending upon the conditions. 

Thickness measurements, the second general application, are made principally 
on aircraft propellers, wing sections, armor plate, steel castings, submarine hulls, 
and other structural components. Numerous special instruments available are 
applicable for thickness ranges of from V in. to several feet. 

Applications Proved Feasible. Instrumentation already developed can be 
applied to many different problems, provided special techniques are worked out. 
Among such applications are: 

1. Rate of growth of fatigue cracks. 

2. Detection of bore-hole eccentricity. 

3. Measurement of elastic moduli. 

4. Study of press fits. 

5. Investigation of corrosion rates. 

6. Control of weld quality. 

7. Metallurgical research relating to structure. 

8. Inspection of grinding wheels, ceramics, and concrete. 

Potential Applications. Investigations indicate the possibility of applying 
ultrasonic techniques to a variety of unsolved problems such as: 

1. Metallurgical analysis and control of case depth and hardness, precipitation of 
alloy constituents, and grain refinement. 

2. Detection of incipient metal fracture. 

3. Theoretical crystallography and metallurgy. 

4. Mechanical properties of liquids and plastics. 

LIMITATIONS. Conditions may exist which prevent the application of 
ultrasonics to specific inspection problems. Usually these difficulties are related 
to one or more of these limiting factors: 

1. Defect detectibility. 

2. Coupling and scanning problems. 

3. Economics of inspection. 

Defect Detectibility. Detectibility is related to sensitivity, resolution, and 
noise discrimination. Briefly, sensitivity refers to the ability of the instrument to 
detect the minute amount of sound energy reflected from a discontinuity. Resolu 
tion measures the ability to separate indications occurring close together from 
two nearby points in an object. Noise discrimination is the capacity for indicat 
ing desired signals from defects in the presence of simultaneous, unwanted signals 
of either electrical or acoustical nature. 

These factors of detectibility are not necessarily independent, so it is difficult 
to state the practical limits quantitatively. However, testing may be marginal or 
unsatisfactory if the specimen is extremely thin, geometrically complicated, or has 
exceedingly coarse grain structure. 

Ordinarily the size or type of flaw is not an actual limit in practical use. For 
example, in the case of fatigue cracks in axles near press-fitted wheels, the smallest 


and shallowest crack which can be detected is limited less by the instrumentation 
or material than by the acoustic "noise" generated by the press fits. Similarly, 
testing of threaded parts for shallow root cracks may not be feasible. 

Coupling and Scanning. Problems of coupling and scanning occur to some 
extent in every application. A liquid or paste couplant which "wets" the surface 
must normally be used between the search unit and specimen. (Certain soft solids 
can occasionally be used as couplants.) Total immersion of the piece sometimes 
can be used advantageously. Extremely rough surfaces may preclude any test 
by preventing effective sound coupling. Scanning of rough, complicated, or small 
parts may be difficult or impossible for practical testing. 

Economics. The economic factors of ultrasonic testing are evaluated more 
readily with an understanding of the basic features of the technique. Economic 
factors vary in importance. For instance, safety aspects, customer requirements, 
or other factors may be the predominant considerations. 

specific industrial applications in which ultrasonic inspection has been particularly 
successful indicates certain keys to success with the technique. Among factors 
common to its successful use are: 

1. Clear definition of test problems. 

2. Good supervision of operators. 

3. Adequate reference specimens. 

4. Practical test specifications. 

5. Realistic acceptance standards. 

6. Detailed test records. 

7. Frequent inspection where required. 

These considerations actually are common to the success of all types of non 
destructive test methods. Since testing and inspection represent considerable 
investment in equipment and manpower and possible delays in production, careful 
attention to these factors of success is always warranted. Where safe usage of the 
product is contingent upon careful testing, rigorous adherence to these rules is 


1. MTTHLHAUSER, 0- "Verfahren zur Zustandsbestimmung von Werkstoffen, Be- 

sonders zur Ermittlung von Fehlern darin (Method of Determining the State 
of Work Materials, Particularly for the Investigations of Flaws in Them)." 
German Patent No.' 569, 598 (1931). 

2. TROST, A. "Nachweis von Werkstofftrennungen in Blechen mit Ultraschall (De 

tection of Material Separation in Metal Sheets by Ultrasonics)," Z. Ver. deut. 
Ing., 87 (1943) : 352. 

3. POHLMAN, R. "Ueber die Moeglichkeit einer akustischen Abbildung in Analogic 

zur Optischen (On the Possibility of an Acoustic Image in Analogy to an 
Optical One)," Z, Physik, 113 (1939) : 697. 

4. SOKOLOFF, S. "Zur Frage der Fortpflanzung Ultraakustischer Schwingungen in 

Verschiedenen Koerpern (On the Problem of Transmission of Ultrasonic Vibra 
tions in Various Materials)," Elek. Nachr.-Tech., 6 (1929) : 454. 

5. VAN VALKENBURG, H. E., F. H. EARDIE, and R. C. GOODMAN. "Acoustic Image 

Inspection," Final Kept., UJ5. Navy, BuAer, Contract No. N0a(s)8868 (1950). 

6. FIRESTONE, F. A. "Flaw Detecting Device and Measuring Instrument." U.S. 

Patent No. 2, 280, 226 (Apr. 21, 1942) . 


7 FIRESTONE, F. A. "The Supersonic Reflectoscope for Interior Inspection." Metal 
Progr., 48 (1945) : 505. 

8. DESCH, C. H., W. J. D.uvsox, and D. 0. SPROULE. "Detection of Cracks in Steel 

by Means of Supersonic Waves/' Engineer, 181 (1946): 467. 

9. ''Ultrasonic Testing of Gas Turbine Disks," Metal Progr., 57 (1950) : 468. 

10. FIRESTONE, F. A., and J. R. FREDERICK. "A Theoretical and Experimental Study 

of the Quartz Crystal as Applied to the Supersonic Reflectoscope." Sperry 
Products Rept. TR-004, Univ. Michigan, Dept. Phys. and Eng.. Research 
Project M-530 (1945). 

11. . "Refinements in Supersonic Reflectoscopy. Polarized Sound," J. Acoust. 

Soc.Am.,lS (1946): 200. 

12. FIRESTONE, F. A. "Tricks With the Supersonic Reflect oscope, 1 ' Nondestructive 

Testing, 7, No. 2 (1948) : 5. 

13. FIRESTONE, F. A., and D. S. LING, JR. ''Reflection and Refraction of Supersonic 

Waves The Raybender," Sperry Products Rept. TR-001, Univ. Michigan, 

Dept, Phys. and Eng., Research Project M-530 (1945). 
14 FIRESTONE, F. A. (To Sperry Products, Inc.) "Supersonic Inspection for Flaws 

Lying Near the Surface of k Part," U.S. Patent No. 2, 467, 301 (Apr. 12, 1949). 
15 t . (To United Aircraft Corp.) "Resonance Inspection Method." US. 

Patent No. Re. 23,090 (Mar, 15, 1949). 

16. WORLTON, D. C. "Ultrasonic Testing with Lamb Waves," Nondestructive Test 

ing, 15, No. 4 (1957): 218. 

17. ERWIN, W. S. "Supersonic Measurement," Steel, 116, No. 10 (1945) : 131, 188. 

18. ERWIN, W. S., and G. M. RASSWEILER. "The Automatic Sonigage," Iron Age, 

160, No. 4 (1947) : 48. 

19. EBDMAN, D. C. "Design and Application of Supersonic Flaw Detectors," Elec. 

Eng., 67 (1948) : 181. 

20. "Ultrasonic Flaw Detector," ILS. Patent No. 2, 593, 865 (Apr. 22, 1952). 

21. "Ultrasonic Flaw Detector," Bull Glass Developments, Ltd., Sudboume Road, 

Brixton Hill S.W. 2, England (1951). 

22. ERDMAN, D. C. "Weld Inspection Tool," Final Rept., U. Navy, BuShips, Con 

tract No. NOBS-50564 (1951). 

23. REID, J. M., and J. J. WILD. "Ultrasonic Ranging for Cancer Diagnosis," Elec 

tronics, 25, No. 7 (1952) : 136. 

24. HOWRY, D. H. "Techniques Used in Ultrasonic Visualization of Soft Tissue Struc 

tures of the Body," IRE Conv. Record 1955, Pt. 9: 75. 

25. PRINGLE, F. E., JR. "SIMAC Sonic Inspection. Measurement, and Control," 

Nondestructive Testing, 14, No. 3 (1956): 22. 

26. SMACK, J. C. "Immersed Ultrasonic Inspection with Automatic Scanning and 

Recording of Warning Signal," Nondestructive Testing, 12, No. 3 (1954): 29. 

27. HUETER, T. F., and R. H. BOLT. Sonics. New York: John Wiley & Sons, Inc., 


28. MASON, W. P. Electromechanical Transducers and Wave Filters. 2d ed. New 

York: D. Van Nostrand Co., Inc., 1948. 

29. . Piezoelectric Crystals and Their Applications to Ultrasonics. New York: 

D. Van Nostrand Co., Inc., 1950. 

30. VAN VALKENBURG, H. E., R. A. WILSON, and W. R. OLHA. "Ultrasonic Inspec 

tion Equipment Program," UJS. Air Farce, Air Materiel Command, Final Rept. 
TR-065, Contract AF 33(600)-29879 (1956). 

31. CARLIN, B. (To Sperry Products, Inc.) ''Supersonic Testing." U.S. Patent No. 

2,527,986 (Oct. 31, 1950). 

32. COOK, E. G., <and H. E. VAN VALKENBHRG. "Surface Waves at Ultrasonic Fre 

quencies," ASTM BulL, No. 198 (1954) : TP 127. 

33. MORIARTT, C. D. "Ultrasonic Flaw Detection in Pipes by Means of Shear 

Waives," Trans. Am. Soc. Mech. Engrs., 73 (1951) : 225. 

31 WEBB, R. H. (To Sperry Products, Inc.) "Variable Angle Ultrasonic Trans 
ducer/' US. Patent No. 2. 602, 102 (July 1. 1952). 


35. BERGMANN, L. Der Ultraschall und seine Anwendung in Wissenschaft und Tech- 

nik (Ultrasonics and Their Applications in Science and Industry). Zurich: 
S. Hiral Verlag, 1954. 

36. KINSLER, L. E., and A. R. FRET. Fundamentals of Acoustics. New York: John 

Wiley & Sons, Inc., 1950. 

37. COCHRAN, D., and R. W. SAMSEL, "Ultrasonics," Gen. Elec. Rev., 47, No. 8 

(1944): 39. 

38. GIACOMINI, A., and A. BERTINI. "Perfezionamento del metodo per accertare 

Fomegeneita dei solidi mediante ultrasuoni (Improvement of the Method of 
Testing the Homogeneity of Solids by Ultrasonic Waves)," Ricerca sci., 10 
(1939): 921. 

39. KOLSKY, H. Stress Waves in Solids. London: Oxford University Press, 1953. 

40. SCHOCH, A. "Schallreflexior, Schallbrechung und Schallbeugung (Sound Reflec 

tion, Transmission, and Refraction)," Ergeb. exak. Naturw., 23 (1950) : 127. 

41. ARENBERG, D. L. "Ultrasonic Solid Delay Lines," J. Acoust. Soc. Am., 20 (1948) : 


42. McGAUGHBY, W. C. "Angle Beam Transmission Through Solid Interfaces," 

Sperry Products, Inc., Eng. Rept. 1050-8 (1953), 

43. FIRESTONE, F. A. "Acoustics," Dwight E. Gray, ed., American Institute of Physics 

Handbook, Sect. 3. New York: McGraw-Hill Book Co., Inc., 1957. 

44. RODERICK, R. L., and R. TRUELL. "The Measurement of Ultrasonic Attenuation 

in Solids by the Pulse Technique and Some Results in Steel," J. Appl. Phys., 
23 (1952) : 267. 

45. GROSSMAN, N. "New Methods Determine Grain Size Ultrasonically," Iron Age. 

172, No. 27 (1953): 72. 

46. TRUELL, R. "Ultrasonic Attenuation Measurements for Study of the Engineering 

Properties of Materials," M ech. Eng , 77 (1955) : 585. 

47. KRUSE, F. "Zur Werkstuckpruefung mittels Ultraschalls (Materials Testing with 

Ultrasonics)," Akust. Z., 4 (1939) : 153. 

48. BECK, W. N. "An Ultrasonic Scanner and Recording System," Nondestructive 

Testing, 15, No. 1 (1957) : 42. 

49. McGoNNAGLE, W. J., J. H. MONAWECK, and W. G. MARBURGER. "Methods of 

Bond Testing," Nondestructive Testing, 13, No. 2 (1955) : 17. 

50. WELLER, E. F., JR. "ABC's of Ultrasonic Inspection," SAE Trans., 63 (1955) : 177. 

51. MORRIS, W. E., R. B. STAMBAUGH, and S. D. GEHMANN. "Ultrasonic Method of 

Tire Inspection," Rev. Sci. Instr., 23 (1952) : 729. 

52. BUCHANAN, R. W., and C. H. HASTINGS, "Ultrasonic Flaw Plotting Equipment. 

A New Concept for Industrial Inspection," Nondestructive Testing, 13, No. 5 
(1955): 17. 

53. BENDZ, W. I. "Discussion Describes Equipment for Making Permanent Acoustic 

Test Records," SAE Trans., 63 (1955) : 188. 

54. KRAUTKRAMER, J., H. KRAUTKRAMER, and 0. RUDIGER. "Ein Ueberschallgeraet zur 

zerstoerungsfreien Werkstoffpruefung (Ultrasonic Apparatus for Nondestruc 
tive Testing)," Arch. Eisenhuet