Full text of "Design of dynamos"

See other formats

Google

This is a digital copy of a book that was preserved for generations on library shelves before it was carefully scanned by Google as part of a project

to make the world's books discoverable online.

It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject

to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books

are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover.

Marks, notations and other maiginalia present in the original volume will appear in this file - a reminder of this book's long journey from the

publisher to a library and finally to you.

Usage guidelines

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing tliis resource, we liave taken steps to
prevent abuse by commercial parties, including placing technical restrictions on automated querying.
We also ask that you:

+ Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for
personal, non-commercial purposes.

+ Refrain fivm automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the
use of public domain materials for these purposes and may be able to help.

+ Maintain attributionTht GoogXt "watermark" you see on each file is essential for in forming people about this project and helping them find
additional materials through Google Book Search. Please do not remove it.

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just
because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other
countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner
anywhere in the world. Copyright infringement liabili^ can be quite severe.

About Google Book Search

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers
discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web

at |http: //books .google .com/I

iLCD, Google

i.sJ,^r .

D, Google

TfT

'TVS'

DESIGN OF DYNAMOS

D, Google

Design OF Dynamos

SILVANUS P^^HOMPSON

Xonbott :

E. & F. N. SPON, Ltd., 125 STRAND

SPON & CHAMBERLAIN, 123 LIBERTY STREET

izecoy Google

iLCD, Google

PREFACE.

These notes on Dynamo Design are not intended to super-
sede the more complete handbooks on the special branch of
electrical engineering of which this is only a part. In the
forthcoming new (seventh) edition of the Author's Dynamo-
Electric Machinery many other examples of design will be
found. The present short work, intended primarily for the
Author's own students, is purposely confined to continuous-
current generators. As it will be used by engineers, chiefly
in Great Britain, in her Colonies, and in the United States,
the calculations and data have been expressed in inch
measures. But the Author has adopted throughout the
decimal subdivision of the inch ; small lengths being given
in mils, and small areas of cross-section in square mils, or,
sometimes also, in circular mils, to suit American practice.

In the section on Armature Winding Schemes special
attention is given to series-parallel windings, and to the
doctrine of the " equivalent ring."

The Author's grateful acknowledgments are hereby given
to various manufacturing firms and engineers who have
supplied him from time_ to time with drawings and infor-
mation that is made use of in this work, and in particular

DigitizecyGOOQle

2193^6

vi Dynamo Design.

he is indebted to the Oerlikon Machine Works, Messrs.
Ernest Scott and Mountain, the Allgemeine Elektrizitats-
Gesellschaft, Messrs. Brown, Boveri & Co., the General
Electric Co. of Schenectady, Messrs. Koltien & Co., the
English Electric Manufacturing Co., the Electric Con-
struction Co., the International Electrical Engineering Co.,
to La Comp^nie de I'lndustrie Electrique, of Geneva, to the
British Thomson- Houston Co., to Mr. H. F. Parshall, and
last, but not least, to Mr. A. C. Eborall.

He also acknowledges the substantial help rendered by
his assistants, Mr. F. I. Hiss and Mr, E. W, Short, in calcu-
lation and tabulation, and in the preparation of cuts.

It is impossible to conclude these acknowledgments
without a reference to the sudden and premature decease,
while this work is passing through the press, of Professor
Sidney H. Short, whose name occurs several times in its
pages. The strong simplicity which characterized the
machinery of his design was a reflex of the personal
qualities which endeared him to many friends in the circle
of electrical engineers on both sides of the Atlantic.

SILVANUS P. THOMPSON.

Technical College, Finsborv, London.

Nmtmber, 190a.

D, Google

CONTENTS.

Preface v

I. Dynamo Design as an Art i

II. Magnetic Calculations as applied to Dvnamo

Design 3

III. Copper Caixulations : Coil Windings . . 40

IV. Insulating Materials, and their Properties . 71

V. Armature Winding Schemes 78

VI, Estimation of Losses, Heating, and Pressure

Drop 113

VII. The Design of Continuous-Current Dynamos . 134

VIII. Examples of Dynamo Design .... 160

I. Wire-Gauge Tables (Copper), British,
II. „ „ „ American,

III. Schedules for Design of Continuous-Current

Dynamos.

Index 341

oy Google

I. Magnetic Curves for Iron and Steel.

II. Continuous-Current Generator of Scott and
Mountain, MP 6 — 150 — 450.

III. Oeelikon Co.'s Generator, MP 4 — a6s — 370.

IV. „ „ „ MP 12— 500— joo.

V, KOLBEH AND Co.'S GENERATOR, MP lO — 350 — 125,

VI. „ „ „ Annature.

VII. „ „ „ Field Magnet Details.

VIII. International El, Eng- Co.'s, MP 8—450 — 250.

oy Google

T^f c Tr- M OT7 r\\/Ts.T A A/rr^c

Ei-rnta.

The reader is requested kindly to make the following corrections
(1 the pages where they occur :

Page 41, line 6, >^ 115° F. read 135° F.
'I 42. n ^^, f^r 0-0000110812 reaii O'ooooroi8i2
>i 44i I. S' fi'' o'ooioszz >r«(/ o'oooiosza
« 44i ., (>, for 0-009710 rfiT*^ 0-000971
" 44. 7. i3i /^ 0002936 rfad 0-002943
11 44. .. 14. yi"" 46'65 rifffi/ 4677

„ 75, in the Table, for mica 800-5000 read mica 800-8000
„ 142, in the two formulK at the bottom, for kw. read watts.
„ 234, in the Table—
for

r^F I"! 3^|'°°| 6004-2 100 439 3640 36337400 63

MP 1 12, soolioo 900II1-8 iisa«53ia'54oss2|3740o| 96

exist ; and a perception ol the reasons wny tney are
successful ; — these and many other things are requisite in the
designer who is to produce machines that will hold their own
in the competition of to-day.

In his treatise on Dynamo-Electric Machines the author
has treated the subject broadly, and with some reference not
only to the historical evolution of the various types of
machine, but also to the abstract theory which must be
acquired if a thorough grasp of the subject is to be attained.
But there are many engineers who have followed some course
of instruction in the theoretical part of the sciences of
magnetism and electricity, who yet have no knowledge of the

::,y Google

D, Google

DESIGN OF DYNAMOS.

CHAPTER I.

DYNAMO DESIGN AS AN ART.

Dynamo Design is an art not to be acquired without
practice and experience ; and like other branches of engineer-
ing design it reposes upon certain fundamental scientific
principles. These can be laid down definitely, and taught
with precision. But in the application of them in design to meet
the varied needs of an ever expanding industry there is wide
scope for choice and for individual preference. Time and
experience have indeed taught the general lines along which
dynamo design must proceed. But no one yet ever designed
a successful dynamo by mere rules. A grasp of principles,
electrical and mechanical ; a knowledge of machinery and its
construction ; an acquaintance with the successful forms that
exist ; and a perception of the reasons why they are
successful ; — these and many other things are requisite in the
designer who is to produce machines that will hold their own
in the competition of to-day.

, C-'OOgIc

2 Dynamo Design.

way in which that theory is applied in dynamo design. The
immediate end and aim, therefore, of the present book is to
give to such a working insight into the procedure of dynamo
design as carried out in recent years for the construction of
continuous current dynamos of modern type. Considerations
of space, and the desire not to enter too far upon the topics
treated of in the author's other works, Dynamo-Electric
Machinery, Polyphase Electric Currents, The Electromagnet
and Elementary Lessons in Electricity and Magnetism, have
determined him to confine the present publication strictly to
the design of Continuous Current Generators, and of these to
treat only of the principal kind, leaving aside small machines,
and special types such as arc-lighting machines. These
are treated of in his larger work on Dynamo-Electric
Machinery. To that work those readers are referred who, for
want of previous general acquaintance with the subject, find
the considerations laid down in the following chapters to
assume points that are not familiar to them. The author
assumes, indeed, that his reader has some acquaintance with
such matters as elementary magnetism and the magnetic
properties of iron, permeability and hysteresis. He also
assumes a general knowledge of electric conduction and
insulation, and of the elements of electrical measurement.

The present work does not go into the theories of armature
winding, nor into the practical modes of carrying it out in the
shop. For these also he refers the reader to his larger
treatise.

After all, however fundamental the necessity of scientific
principles, sound theories, and rules derived from the experi-
ence and practice of others, dynamo design remains an art.
It needs the eye to see, as well as the mind to understand.

izecoy Google

CHAPTER 11.

MAGNETIC DATA AND CALCULATIONS.'

All dynamo design is based upon a knowledge of the
magnetic properties of iron and steel. During the past twenty
years thousands of brands of various qualities have been sub-
jected to test as to their magnetic properties by scientific
authorities, and there exists an extensive literature on the sub-
ject. The principal thing to know is the appropriate density
of the magnetic flux, and the amount of excitation required to
produce it, in any given specimen. In this book the letter N
is used to denote tke magnetic flux, that is to say, the total
number of magnetic lines, carried by any iron core. If the
area of section of this core is denoted by the letter A, the
density of the magnetic flux will be equal to N -^ A. When
the' sectional area is given in square inches the letter used
to denote the flux-density (i.e. the number of magnetic lines
per square inch, is B. In cases where the area is given in
square centimetres, the letter used for the flux-density will
be Si' The magnetizing forces required to excite any re-
quired flux-density in the magnetic circuit of a dynamo are
obtained by causing an electric current to circulate around
the iron core. It is found that the magnetizing force thus,
produced is proportional both to the amount of current (i,e.
the number of amperes) so flowing, and to the number of
times it circulates around the core (i.e. the number of turns in
the magnetizing coil). In other words, the magnetizing force
is proportional to the number of ampere-turns. For brevity
we sometimes describe the total number of ampere-turns of
circulation of current around a core as " the excitation" It

B 2

oy Google

4 Dynamo Design.

goes without saying that the higher the flux-density required,
and the greater the length of the iron through which the
magnetic fliix is to be driven, the greater is the amount of
excitation needed. To drive a magnetic flux through air
requires a much greater amount of excitation than is required
for an equal flux-density through an equal length of iron.
The coefficients used in calculating air-gaps are mentioned on
p. 28. In treatises on theoretical magnetism it is usual to
describe magnetizing forces in terms of a theoretical unit
(derived from the Centimetre-gramme-second system), which
is such that if applied to an air-core one centimetre in length
it would produce a flux-density of one Hne per square centi-
metre. The usual symbol for denoting the theoretical value
of magnetizing forces in terms of this unit is Jf. For example,
suppose it was stated that the magnetizing forces were
such in some case that K = 50, this would mean that they
were so stroog that if applied to a layer of air I centimetre
thick they would produce in the air a flux-density of 50 lines
per square centimetre. If applied to an equal length of iron,
the resulting flux-density ^ would be immensely greater, since
iron is much more permeable magnetically. The ratio of
oJ to K in any material is called its permeability. The
magnetic properties of iron, and the variations of its per-
meability may be described in various ways by statistical
tables, derived from experiments. But for practical pur-
poses it is far more convenient to exhibit them by means
of magnetization curves, that is, curves connecting the amount
of flux-density produced in the material with the magnetizing
force necessary to produce it. Moreover, as the ratio of the
former to the latter is a measure of the permeability of the
material with which we have to deal, we might use this
ratio to estimate the amount of magnetism that would
be produced in a given material by the action of a definite
magnetizing force or vica versa. In commercial work, how-
ever, it is found more convenient to dispense altogether with
the permeability in magnetic calculations, and to work
directly with the ampere-turns required per unit length of
material in order to produce a definite flux-density in it

D,j,i,:«,.,,Googli:

Magnetic Calculations. 5

For instance, we know' that the magnetizing force J£ is
related to the ampere-turns per centimetre of length by the

equation :

~f _ 4xC S.
lo"/ '

where C is the current in amperes, S the number of turns and
/ the length in centimetres. Transposing, we find that the
number of ampere-turns per centimetre of length will have
the value

/ 47r

And, as l inch = 2-54 centimetres, we shall have for the
number oi ampere-turns per inch length of material, the value

I 4x

Hence, if we have already got, for any specimen of iron, the
curve connecting ^ and %, we can at once change the K
values for the more convenient ampere-turns per inch by
changing the scale of the abscissae ; the point, for example,
marked 50 on the % axis will now read lOi on the scale of
ampere- turns per inch.

Such a set of curves, connecting the flux-density in lines
per square inch with the ampere-turns required per inch of
magnetic path, in different materials is given in Plate I. The
curve marked I, is for armature sheet, and represents this
material as supplied by Messrs. Shaw, of Middlesboro', Curve
II. represents the cast steel for dynamo purposes made by
Messrs. Edgar Allen, of Sheffield. Curves III., IV. and V.
are for good wrought iron, malleable cast iron and good cast
iron, respectively. In the drawing office each dynamo designer

' Readers who desire further informalion about magnetic units and their
measurement should refer to (he Author's Etemaitiay Lessons in Electricity and
Magnetism, or to his treatise on J'Ae Electromagnet. In the latter will also be
found an account of the various methods of measuring the magnetic qualities of

D, Google

6 Dynamo Design.

ought to provide himself with similar curves for the particular
brands of iron and steel which he uses. All the good makers
of iron and steel for dynamo purposes will furnish curves for
the materials which they produce. The additional curves given
in Fig. I relate to wrought iron when worked at very high
flux-densities ; one is due to Mr. Parshall, the other relates
to Messrs. Sankey's special quality of armature stampings.
Such curves find their principal application in calculating the
ampere-turns required forthe teeth of slotted armatures, which
are frequently worked at very high flux-densities ; but they
must be used with caution, on account of the limited amount
of knowledge at present available on this subject. Mr. H. S.
Meyer gives measurements' on a sample of still higher quality.
Barrett ^ has recently found that a particular steel contain-
ing 2\ per cent, of aluminium, made by Hadfield, of Sheffield,
has a higher permeability than any known brand of wrought

Looking at the curves of Plate I., one sees that if one
wishes to know, for example, how much magnetizing force is
required to produce a flux-density of, say, B = 100,000 lines
per square inch, in wrought iron, one follows out the curve
of wrought iron up to the level of 100,000, and then
dropping perpendicularly on to the horizontal scale one
observes that it will require 73 ampere-turns per inch length
of the iron.

Example. — Find the number of ampere-turns of excitation neces-
sary to drive a flux of 12,000,000 lines through a cast-iron yoke, the
cross-section of which is 300 square inches, and the length 17 inches.
Since 300 square inches carry 12,000,000 lines, the flux-density B
will be 40,000 lines per square inch. Reference to the curve for
cast iron on Plate I. will show that this will require 77 ampere-
turns per inch ; and as the iron is 1 7 inches long the answer is
17 X 77 = 1309 ampere-turns.

Further examination will show that though for flux-
densities below 85,000, mild cast steel is less magnetizable

' Ekklrol. Zatschti/I, xxiv. 769, September u, 1901.
' See Jearti. Inst. Elec. Enginiers, xxxi. 709, 190Z.

iLCD, Google

Magnetic Calculations.

D, Google

8 Dynamo Design.

than wrought iron, yet at higher flux-densities the mild steel
is equally good or even better. Being cheaper, it has come
very largely into use for dynamo magnets and pole-cores.
Though called " mild steel," it is in reality very nearly pure
iron, as it contains only about 0'2 per cent of carbon, and is
incapable of taking a temper.

For an account of the various methods of measuring' the
magnetic qualities of iron, and for detailed information as to
the newest kinds of iron and magnet steel, see the author's
treatise on The Electromagnet.

Sheet iron often shows very different qualities in different
parts of the same sheet. According to Rohr,' parts near the
edges of the sheet are often better annealed than parts from
the middle. Repeated annealing tends to make any sample
more homogeneous magnetically. The most homogeneous
magnetic material hitherto produced is annealed cast steeL

Heat wasted in Cycles of Magnetization. — It has long been
known that when iron is subjected to rapidly recurring
magnetization , and demagnetization, or to an alternating
magnetization, it becomes hot. This heat so wasted in the
iron is due to two causes, (i) hysteresis, (2) eddy-currents.

Hysteresis is a species of magnetic fatigue which wastes
energy at every reversal of the magnetization, particularly in
all hard kinds of iron and steel. To minimize this source of
loss the cores of armatures must be made of material which
has as low a hysteresis as possible.

' Consull also the following works : —

Ewing J. A., various papers in the Philosophical Transactions of the Royal
Sociel; in the years 1885 lo 1894. A full r^ume is given in his book Magnate
Induction in Iron and other Metals. London, 1S94.

Hopkinson, Dr. J., papers in the FhHosofhical Transactions of the Royal
Society, 1885 lo 1895. Those of chief importance are reprinted in his Original
Papers (1901), vol. i.

Du Bois, H. J. G., Magnelische Kreise, deren Theorie vnd Anviendungtit.
Berlin, 1894.

Jackson, Dugold C, Eleciromagnetism and the Constructian of Dynamet
(Macmillan).

Paiahall, H, Y., Electric Generators, London, 1900; also />w. Inst. Civil
Engineers, cjxvi. May 19, 1896.

Schmidi, Dr. E., Die Magnelische Untersuchung des Eisens. Halle, 1900.

• Eleklrot. Zeitsckr., lix. 712, 1898.

izecy Google

Magnetic Calculations. 9

Eddy-currents are currents induced in the substance of the
iron core itself by the magnetic changes. They can be
reduced indefinitely by laminating the cores, which are built
up of thin sheets of iron or of a special soft steel ; the usual
thickness being from 25 to 40 mils thick, though in some cases
core-sheets as thin as 15 mils (= about 0'6 millimetre) are
employed.

Calatlalum of Heat- Waste in Iron Cores. — The energy lost
per cycle depends not only upon the nature of the material
but also upon the degree to which the magnetization is carried
in each cycle — in fact upon the amplitude of the cycle. The
loss of energy per cycle is more than proportionally great ;
doubling ^ more than doubles the energy loss per cycle.

Mr. C. P. Steinmetz ' has given the following law connect-
ing the hysteresis loss k in ergs per cubic centimetre of iron
per cycle and the flux-density Si. He iinds that

where ij is a constant, called the hysteretic constant, depending
upon the kind of iron. This law is true for cycles performed
either slowly, or as rapidly as 200 per second. The following
table gives the hysteretic constant 1; for difiTerent materials *
when ordinary frequencies are employed.

From experiments with actual transformer plates, at n
cycles per second, the hysteretic loss, in watts per cubic inch
of iron, was found to be

W* = 0-83 X 1; X « X B' * X 10""'.

In Fig. 2 there have been plotted the hysteresis losses
in watts per pound of iron, at a frequency of 30 cycles per

' Ama: Jnst. Eke. Engimers, Jan. 19, 189a ; Eltelridan, Feb. 12, 19 ftad
36,1892. The exponent is not always exactly 1 ' 6 ; it vaiiesbetween 1*5 and I '9.

■ For parlicolaia of Ewing's Magnetic Tester for measniine H^rsteieds in sheet
iron, see Intl. EUc. Engintert, April 25, 1S95, also EUclrteian, xxxiv. 7S6. To
reduce these values from ergs per cabic centimetic percycle to the more ordinary
value in watts per pound of iron at loo cycles per second, multiply by the factor
0*000589. Barrett Iinds the values for Swedish charcoal iron, for Sankey's "Lohys"
iron, and for aluminiuin-iron to be respectively 038, 0-32 and 0-23 watts per
potmd, with a maximum flux-density of 4000 lines per square cc

iLCD, Google

lo Dynamo Design.

second, for different values of the flux-density (in British
measure). The iron is here taken to be of an ordinary good
quality.

;

■

^'^

iLCD, Google

Magnetic Calculations.

TABLE I,— HvsTBRKTic Constants for Different Materials.

Maleiial.

c".=ft

.u..,,.,.

HyiiMcrie
Coiuunt D-

Very soft iton wire . .
Very thin soft sheet iron .
Thin good sheet iron . .
Thick sheet iron . . .
Most ordinary sheet iron .
Sanlcey's Lohys iron . .

■0024
■003
'0033
-oat

Soft annealed fast iron .
Soft machine steel . . .

Cast steel

Cast iron

Hardened cast steel . .

■008
■0094

■016
■025
■00068

Similarly, Fig. 3 gives in the form of a graph the
losse?, b}- hysteresis, in watts per cubic inck of laminated iron,
such as is used in armature cores ; the curve in this case being
plotted for one cycle per second, the corresponding flux-
densities being given per square inch.

Example.^-Y'mA the power wasted by hysteresis in the core-body
of the 8-pole dynamo, page 146, assuming B = 65,000, the total
volume of the iron being 16,320 cub. inches, and the frequency of the
reversals 10 cycles per second.

TABLE II.— Waste of Powbr by Hystbrbsis.

p^?riri:^«

[■rci.bLrf».«

W.lB_wa.l*d

^" 'mlj^ """'"

""■'■""""'■■

'ZtS'

t«cjcl«p«
second.

4,000

25,800

1 0-0023

400

5.000

32.250

00033

S7-S

57S

6,000

38.700

O-0043

750

7,000

45.150

0-0053

92- S

9*5

g.OOO

51,600

i 0-0064

lit

10,000

64,500

0-0090

156

1560

ia,ooo

77.400

00119

206

3060

14,000

90. 30°

0-0151

262

26Z0

16,000

103,200

00186

324

3240

17,000

109,650

o-ozzS

394

3940

18,000

116,100

0-0282

487

4870

D, Google

12 Dynamo Design.

Table II. gives the number of watts wasted by hysteresis
in well-laminated soft wrought iron when subjected to a
succession of rapid cycles of magnetization.

To facilitate calculations, Table III. gives the number of
watts wasted per cubic inch per cycle for three different values
of the hysteretic constant.

TABLE III.

Li«.|».

limiBch. Wuumnedpcr

cubic imli of bo» per cycle per •ccQDd.

3 , = 0-002.

1, = 0-003.

)i = 0-004.

000 o- 0038*9

0-005767

0-007658

St*

000 0-005478

0-008280

0-010956

000 0-007330

o-oiioSo

0014640

000 000934s

0-014090

0-018690

000 0-011586

0-01745S

0-023172

000 0-013993

0-021082

o' 027986

100

000 0-016666

0-025000

0033333

000 0-019322

0-029100

0038644

120

000 o-oazo94

0033425

0-044188

130

OOD 0-025248

0-038025

0-050496

140

000 0-028386

0-012750

0-056772

These values are plotted in the curves of Fig, 3. For
other frequencies the values must be multiplied simply by tke
frequency.

Besides the hysteretic loss in iron plates, there is
also a loss due to eddy-currents in the iron. This loss varies
as the square of the thickness of the iron, the square of the
frequency, and the square of the flux-density. There has been
obtained by calculation the formula

: 40-64 X ^ B'

W, being the loss in watts per cubic inch of core made up
of the strip, and t being the thickness of the strip in inches.

izecoy Google

Magnetic Calculations.

■ 1 1 I

HYSTERESIS LOSSES
IN IRON.

ooes

ry, IN

PERS

au*R

EINC

ffi

' '

Fig. 3.— Hysteresis Losses in Watts per Cubic Inch.

izecy Google

14 Dynamo Design.

Thus we have the total loss in watts per cubic inch, due tq
both hysteresis and eddy currents —

W = 083 -ei n B" X 10"' + 40-64^ B'' "* X 10"".

Tliis has been found to agree very closely with practice.
In order to adapt the data to British measures, and to
facilitate computation, Table IV. has been prepared, showing
the values of the eddy-current losses in sheet iron of four
different thicknesses, with a frequency of one cycle per second.
The values for other frequencies will require to be multiplied
by the square of the frequetuy.

Eddy-currE
R 1

..Win„„.pe,

cubic iiicl.,>t.C¥

kp«l«.

, ( = .on.ils

.=»™l.

/ = ,on,ili

( = 60 mils

000 '

000006 ,

0000026

000234

000

000010 1

0^000040

000162

000366

000

oooois

0-000058

000234

000527

000 .

000020 ;

0-000079

000318

000717

000

000025 !

0-000104

000416

000935

000

oa»33 j

0-000132

000526

001 185

100

000 1

000041

0-000168

000650

001463

000 1

000049

0000196

000787

001770

I»

000

000058

o-oooa34

000936

002107

130

000

000068 1

0-000275

001099

002490

140

000 o-ooooSo

0-000318

O-00127S

0-002867 1

These figures are graphically plotted for reference in
Fig. 4

Example.— YmA the number of watts wasted by eddy-currents in
the armature core-body (not including teeth) of the 8-pole dynamo
described on p. 146. Taking B = 65,000 lines per square inch, the
frequency 10 cycles per second, the thickness of the core-disks as
40 mils, and the number of nett cubic inches of iron as r6,32o, we
proceed as follows. Referring to the curves of Fig. 4 we pick out
the curve for 40-mil iron, and follow it up to opposite the value of

:.y Google

'ic Calculations. ^ 15

65,000, at which point we read off on the other scale the value
0-000275 ^s the numher of watts per cuhic inch at a frequency of
I cycle per second. Then multiplying up by the square of the

—

— 1 1 — \ y- — -]— "

—

EDDY CURRENT LOSSES
IN IRON.

ooore

O'OOIO

»/

O'Oooe
5

:S^

1
%

Fig. 4. — Eddy-Current Losses in Sheet Iron,

frequency, and by the number of cubic inches, we find as the total
eddy-current waste, 449 watts. Had we taken core-disks of half the
thickness this waste would have been reduced to 112 watts.

Examples of calculation of the waste of power by eddy-

, ,.« .„Googlc

1 6 Dynamo Design.

currents and hysteresis in the iron — usually called, for brevity,
the iron-losses — in the armatures of continuous current dynamos
will be found in Chapter VIII. pages i68 and 182.

Ewing has shown that vibration tends to destroy residual
effects. There is also some evidence that with very rapid
frequencies there is less work wasted per cycle than there
would be in the same cycle performed slowly.

Rotational Hysteresis. — When an armature core is rotated
in a strong magnetic field the magnetization of the iron is
being continually carried through a cycle, but in a manner
quite different from that in which it is carried when the
magnetizing force is periodically reversed, as in the core of a
transformer. Mordey has found ' the losses by hysteresis
to be somewhat smaller in the former case than in the latter.
Baily ^ found the losses, for a rotating density lower than ^ =
1 5,000, to be slightly lower than is the case for alternating
fields ; but in stronger fields the rotational losses diminished
after that point, and became nearly zero when ^ ^ 20,000.
Dina ^ has, however, failed to confirm the latter result.
'^ Retardation of Magnetisation. — It has long been known
that in solid cores of electromagnets the rise and fall of the
m^netism is retarded by eddy-currents in the iron, the outside
part of the iron becoming magnetized first when the current is
turned on ; whilst the magnetism of the inner parts grows
up later as the eddy-currents in the outer part die away.
There is thus a regular penetration or propagation of the
magnetism from the outer to the inner parts of the core.
When the magnetizing-current is cut off, the inner part is the
last to lose its magnetism. In large dynamos many minutes
may elapse before the magnetism attains its maximum. For
this reason the author pronounced it useless to put a compound
winding upon certain dynamos with large solid bipolar electro-
magnets for use as electric railway generators. Hopkinson *
showed that the retardation varies as the square of the linear
dimensions.

' See also Ewing in Electrkian, xxvii. 6oz, l8gi.

' Phil. Trans., cbixvii. 715, 1896.

' Elektrot. Zattchrift, xixv. 43, 1902.

' Joum. Init Eta. Engineirs, Feb. 1895, and Phil. Trans., 1895.

.oogle

Magnetic Calculations. 17

Magnetic Dampers.— \{ a magnetic flux, whether in air or
in iron, be surrounded by a closed conductor such as a copper
ring or tube, or a copper wire coil the ends of which are
united together to form a closed circuit, it is impossible either
to increase or to diminish this magnetic flux without setting
up induced currents in the surrounding conductor ; and these
induced currents always tend to oppose, and therefore to delay
the change of the flux. Hence, it is possible to protect any
m^net pole against sudden changes in its magnetism by the
simple device of surrounding it with a solid coil or circuit of
copper to act as a magnetic damper. For this end Brush, in
1878, surrounded the limbs of his field-magnets with a copper
tube. In recent times Hutin and Leblanc have proposed a
device called an anwrtisseur (i.e. a damper) to prevent distor-
tions of the magnetic field under the poles. For similar
reasons the Westinghouse Co. inserts copper dampers between
the pole-tips of its alternators and converters. A copper ring
round a pole may thus prevent rapid changes in the enclosed
flux, but cannot prevent distortion within the enclosed space
from one part of the pole-face to another. To prevent this,
or lessen it, one or more copper bars must be inserted across
the pole-face and short-circuited together by outer bands. As
induced currents can be set up in solid iron or steel as well
as in copper it follows that solid steel pole-shoes to some
extent serve the same purpose as magnetic dampers, though
less effectively than an amortisseur.

Coefficient of Dispersion.

To produce a definite electromotive-force with a given
number of conductors rotating at a given speed, a certain
magnetic flux must be cut by them. The function of the field
system of a dynamo is to provide this flux, which may be
called the useful flux because it is the actual flux being cut
by the conductors and producing the electromotive-force.
Now look at Fig. 5, which shows a typical bipolar magnetic
circuit. In addition to the useful flux in the air-gap, there
is a stray flux from all parts of the field system, and' both

oy Google

1 8 Dynamo Design.

the useful and stray fluxes have to be produced by the excit-
ing ampere-turns wound on the magnet limbs.

If we call the total flux (per pole) produced by the
exciting coils in the magnet core N„, the useful flux which
actually enters the armature N„ ; and the stray flux which is
dispersed N, ; then obviously

The ratio of the stray flux to the useful flux is sometimes
called the dispersion. The ratio of the total flux to the

Fig. 5. — Stray Field of k Bipolar Dynamo.

useful flux is called the coefficient of dispersion, or coefficient of
allowance for leakage, or, less correctly, the leakage coefficiefit,
and is denoted by the symbol v. Thus we have

This coefficient of dispersion is therefore a number, greater
than unity. It varies between 1*15 and 1*7 in the usual

types of machinci

Example. — In a certain 8-pole tramway generator the total flux
per pole in the magnet core was 18,673,000 lines. Of these only

oy Google

Magnetic Calculations. 1 9

15,400,000 entered the armature, while 2,772,000 were dispersed.
The coefficient of dispersion was therefore v= 1 ■ 18, the stray flux
being 18 per cent, of the useful flux.

Although the stray field does not cause a waste of energy,
yet it is objectionable in any class of machine on account of
the extra material that must be put into the field system to
make up for it — that is, more iron is necessary in the yoke
and pole cores to carry the extra flux, and the length of each
turn of the copper winding round them is increased. It be-
comes of importance, therefore, to design magnetic circuits to
have a minimum amount of magnetic dispersion in order to save
expense. The magnitude of the stray field depends chiefly
upon (a) shape of the magnet limbs — thus circular cores will
have less leakage than those of rectangular shape, on account
of the smaller area of the side flanks ; (*) upon the length of
the air-gap, because the greater the reluctance of the latter
the greater the tendency for the flux to take other alternative
paths ; and (c) upon the degree of saturation to which the field
system is pushed, because the magnetic conductivity of the
leakage paths in the air is constant while that of the iron cores
decreases as the degree of saturation is raised. It is evident,
therefore also, that not only will the coefficient of dispersion
vary with different types of machine, but it cannot, as a rule
be constant with a given machine but must vary with the exci-
tation. Moreover, when a large current is being taken from
the armature, the demagnetizing action of the armature due to
the forward lead of the brushes, directly promotes dispersion, as
it raises an opposing magnetomotive-force in the direct path
of the magnetic lines, tending to blow them aside, as it were.

The most accurate way of finding the dispersion coefficient
of a machine is by experiment. If in the case of a bipolar
dynamo we wind around the armature a search coil with its
plane at right angles to the magnetic flux, and connect it up
to a ballistic galvanometer, we shall, upon making or breaking
the exciting current, obtain a throw Di proportional to the
flux passing from pole to pole. If the search coil be wound
upon the limbs just at the neck of an exciting coil, a second
throw Dj may be obtained in the same way, and which will

c 2

20 Dynamo Design.

be proportional to the total flux produced bj' the windings.
Then we have

" Di

for the particular excitation used.

Before each reading, the current in the fields should be
reversed several times in order to wipe out any residual
magnetism. In order to allow for the effect of the armature
current, a few accumulators in series with an adjustable
resistance may be connected to the brushes, and an appro-
priate lead given to them, the direction and amount of the
current being such that the armature demonetizes the field-
magnets to the same extent as would be the case with the
machine running on full load. The ratio of the maximum
throw to the throw given by the armature search coil under
these conditions will then approximately give the full-load dis-
persion coefficient. The principal objection to this method is
the great strain it imposes upon the insulation of the magnet
coils. As a general rule it cannot be employed with shunt
windings, and for such machines a test-winding of fewer turns
(and correspondingly larger section) must be wound on.

This method is due to the late Dr. J. Hopkinson, who found
the dispersion coefficient of a bipolar dynamo of the " under "
type to be I ■ 32,

Another method is by means of an alternate current.
Wind on a search coil of Sj turns as before, and connect it to
a voltmeter. Now send an alternating current (preferably of
low frequency) of known electromotive-force E, round the
field coils, whose number of turns Si is known. Let the
voltmeter reading be Ej, Then

N„«|^ andN. «|».

Hence

_ El X Sa
" Ej X S,"

The eddy-currents produced by the alternating flux in the

ii.w .„Googk'

Magnetic Calculations. 21

solid cores would not affect the results, but might be incon-
venient if the current was left on too long.

The third way of determining the dispersion coefficient is by-
experiment upon the finished machine, whose dimensions and
winding data are known. The method applied to a shunt
machine, is as follows.

Run the machine at its normal speed of n revolutions per
second, with its full-load of Ca amperes. Measure the shunt
current C„, and electromotive-force at terminals V ; and also
note the lead of the brushes. From the resistance of the
armature (brush to brush) and the load current, we calculate
the ohmic drop as R^ x Ca. This added to V gives E, the
volts actually generated at full-load. The full-load useful flux
must hence be

N„ =

E X 10"
« X Z •

Now calculate the ampere-turns required to drive a flux of
Na lines through the whole magnetic circuit. Call them Xj.
From the known lead of the brushes and the armature current

Fig, 6.— Estimation of Leakage by Exploring-Coii.s.

calculate the value of the demagnetizing ampere-turns as
indicated on page 1 27. Subtract them from the observed full-
load ampere-turns S„ C„, obtaining a value Xa- Then the
dispersion coefficient is approximately given by
^Xj
" Xi"

izecoy Google

2 2 Dynamo Design.

To obtain a nearer approximation, take the value of de-
magnetizing ampere-turns as calculated, multiplyby this value
of V, and then subtract from S„ C„, obtaining a new value
for Xj to be used as above. But this method as a whole is
not capable of giving very accurate results.

A highly detailed examination was made at Schenectady
upon a multipolar dynamo, to ascertain the fluxes through
the various parts. A large number of exploring coils were
wound over the machine as indicated in the accompanying
Fig. 6. The results are given in Table V.

TABLE v.— Fluxes in Various Parts of a Dynamo.
MP— 6—400— soo (G. E. Co.).

FLui Tbiough ihiu Pan

C«l

Fig. 6,

^'^iL-t^^

.npeiH Wi<hig-.i,

.p™

Wi.h^..,^^p.™

2,220

000 2.670,

000

3,060,000

2,284

000 2,770,

000

3,111,000

3,660

c»o S.465,

000

6,180,000

3,880

000 5.840,

000

6,615,000

4,620

000 6,120,

000

6,950,000

4.89s

000 6,350

000

7,400,000

4,900

000 6,480

000

7.52S.«»

5"^

4 750

000 6,290

000

7.333.<»o

6
7

4,830
2,356

000 6,470

000 3.'20

000

7,38S.ooo
3.575.000

2,480

000 3.100

000

3,470.000

2,500

000 3,120

000

3.500,000

000 44

200

S4,ooo

lla

3,9">

OCX) 5,140

000

6,(80.000

tij

4,000

000 5,200

000

6,iiS,otx.

12a

498

000 731

000

985. 000

lib

473

000 728

000

934.000

Looking at the figures in the last column, with the excita-
tion at its highest, we see that the maximum flux in the pole-
core was about 7,400,000 lines. Of these about 6,180,000

izecoy Google

Magnetic Calculations. 25

actually entered the armature, the sum of the fluxes measured
by exploring coils Nos. \a and \b agreeing closely with that
of No. 2. The yoke appears to be of insufficient section, as
the flux passing through No. 8 is less than half of that through
Nos. s or 6. Taking the figures above we see that the co-
efficient of dispersion at the highest excitation is v = r ' 19,

TABLE VI.— Dispersion Coefficients,

, Ovn-
Outputm ''?•"
Kilowi„ts ' ""■ '3'

IIXE.M

p.ife.

^1?

Multi-

no 5 . 1-4

1-6 . rss

»'7

1-30

I -65

'■75

Sto 25 '-as

•■45 "'4

1-55

...

•■s

i-SS

i-3a

25 to 100 i-az

I 35 ; 1-32

I -45

I '16

1-4

»-45

i-»8

100 to 300 ..

i-aS

i'7

1*23

300 to 1000 ..

116

Table VI. given below gives the value of the dispersion co-
efficient for various types and sizes of machines, the values in
every case being if anything, in excess of the true amount, as
they have been obtained from calculation and experiment
under the most unfavourable conditions.* Magnetic dispersion
is always greater with the smaller sizes of machine, on account

' See DynamB-ma^iinis, by A. Wiener, 1902. For earlier data on the stray
fields of dynamos, see Heriog in El. Rev., xxi. 186 and 205, 1887: Cathart.
Elatrkian, xxiii. 644, l88g ; Wedding, Eleclrot. Zeitschrift, xiii. 67, 1891 ;
MaTor, EUctrkal Enginitr, rii, 428, 1894: W. B. Esson in Joum. Inst. Elee.
trieal Engineers, Feb. 1890 ; and W. L. Pnffei in Technology Quarterly, iv. MS,
Oct. iSoi. Some recent researches on magnetic dispeision are those of Rother
in the Eleklrotechnisehe Zeitschrift for May 36, 1898 ; and Picoa, Bulletin Soc.
Internat. des Sliclridens, June 1902, p, 425. Attempts to Teduce dispersion are
disctissed by Kelly in Electrical World, mii. 161, 1898, and by Guilbert in
L'Eclairage Eleclrigue, xviii. Z98, 1899.

D, Google

24 Dynamo Design.

of the difficulty of properly dimensioning the field-system.
It is also greater with cast-iron magnets and pole-pieces, and
as we have seen already, with smooth core armatures. The
values given below may consequently be looked upon as being
high for slotted armatures and wrought-iron or cast-steel
fields.

Calculation of Dispersion.

It is possible to predetermine, from the working-drawings

of a dynamo before it is built, the probable amount of dispersion.
Calculations of the dispersion are based upon the principle that
where a circuit offers alternative paths, the magnetic flux will
divide itself between the paths in the proportion of their
relative facility for flow, exactly as an electric current divides
where there are alternative conducting paths. In fact, the law
of shunts has been found to hold good for magnetic lines.
The reader should consult the researches of Ayrton and
Perry^ on this point. It follows that along any branched part
the joint permeance^ (or magnetic conductance) will be the
sum of the permeances of the separate paths. Hence, if the
permeances of the separate paths of the useful and waste
magnetic fluxes of a dynamo are known, the coefficient of
dispersion, v, can be calculated, it being the ratio of the
total flux to the useful flux. Call the useful flux u and the
waste flux w ; then

_ n + w

But each of these is a complicated quantity ; therefore the
more complete formula is

In order to determine the separate permeances along the
various leakage paths, we must resort to some useful rules or

■ youm-Soc. Teleg. Engineers and Ehch-idans, ^t,q, i886.
' Permeance is of course the reciprocal of magnetic reluctance ; just as
electric conductance is the reciprocal of electric ri

,1.0, Google

Magnetic Calculations. 25

lemmas originally suggested by Professor Forbes/ which con-
sist in certain approximate integrations. For the convenience
of British engineers the values have been recalculated into
inch measures instead of centimetre measures.

The unit reluctance and unit permeance are so chosen as
to obviate the subsequent necessity of multiplying the ampere-
turns by 4 TT -T- 10. This will make the reluctance of the inch
cube of air equal to ro -i-47r divided by2"54 = 0'3I33 ; and
its permeance to 3 • 1918.'

Rule I. — Permeance between two parallel areas facing one
another. Assume that the magnetic lines are straight and
equally distributed over the surface : then.

Permeance = 3* I918 X mean area (square inches) -r- distance
(inches) between them

= 1-596 X (A," + A,")-;-^;".

Rule II. — Permeance between two equal adjacent rectangular
areas lying in one plane. Assuming the lines of flux to
be semicircles, and that distances d" and d^ between their
nearest and furthest edges respectively are given ; also d' their
width along the parallel edge : —

Permeance = 2 "274 x a" X Ic^w-p,-

Rule III. — Permeance between Iws equal parallel rectangu-
lar areas lying in one plane at some distance apart. Assume
the lines of flux to be quadrants joined by straight lines.

Permeance = 2 '274 X a" X logio-( I +

^w-^n

Rule IV. — Permeance between two equal areas at right-
angles to one another.

Permeance = double the respective values calculated
by Rule II.

■ Journ. Soc. Teleg. Enginari, itv. 551, 1ES6.
- See the Author's work TJu Eltetrsmaitict.

D, Google

26 Dynamo Design.

If measures are given in centimetres these rules become
the following :—

I. i(A, + Aa)^^
n. ^ hyp. log ^?.

ni.?h,p,iog(,4---<\-'''>).

Using these rules to predetermine the stray field to fly-
wheels, pedestals and shafts, it is possible from the working
drawings to predict the performance of a machine to within
2 per cent.

The author has given (in his work on The Electromagnet)
some further rules, including the case of permeance between
two parallel cylindrical limbs. The reader should also consult
the writings of Kapp,' Wiener' and Arnold * for the predeter-
mination of the dispersion coefficient, the last named author
going into the question at great length.

Goldsborough * has laid down the theorem that, assum-
ing a fixed difference of magnetic potential between the
surface of a pole-piece and that of an armature core (the
latter surface supposed to be smooth), the intensity of the
magnetic field at any point at the surface of the armature will
be proportional to the sum of the reciprocals of the distances of
that point from all the points on the perimeter of the pole-piece
made by a section-plane passing through that point. On this
principle he has calculated the distribution of the flux in the
gaps in certain cases.

By definition the dispersion coefficient v = (N„ + N,) -j-
Nfl ; and as the useful and stray fluxes are respectively propor-
tional to the permeances of the useful and stray paths, if we
write P»for the permeance through the gap and teeth, and
P, for the permeance of the stray field, we may write v =

' Etektromichanische Konstructionin, by G. Kapp, p. 9.

' Dynamo-machines, by A. Wiener, 190a, pp. 217-265, end 614-628.

• Die Gltiehitrom-Maschinen, 1902.

* Tratu. Amer. last. El. Engineers, June 30, 1898, p. 515.

izecoy Google

Magnetic Calculations. 27

(P„ + Pi ) -:- Pb . Now P, is a constant, being through air,
whereas P„ being partly through air and partly through iron
will diminish as the saturation of the teeth increases towards
full-load. Hence v will rise as the excitation is increased.

Determination of Exciting Ampere-Turns.

The calculation of the ampere-turns necessary to drive a
certain useful flux N^ across the air-gap of a dynamo is a
straight- forward matter if we know the dispersion coefficient of
the machine, and the magnetic properties of the materials
used to carry the flux, as laid down in curves such as those
in Plate I.

The method of using these curves for the purposes of
dynamo calculation is as follows. We are given : —

Na useful flux per pole ;

Ai magnetic area of yoke ;

A, „ of field cores ;

A3 „ of air^ap ;

At „ of teeth under one pole ;

As „ of armature core ;

Li length of magnetic path in yoke ;

La „ „ in two field cores ;

L3 „ „ in two air-gaps ;

Li „ „ in twQ teeth ;

Lfi „ „ through armature core ;

and the question is to find the ampere-turns per pair of poles
necessary to produce the flux of N^ magnetic lines in the
air-gap.

Now the total flux is

Consequently, by dividing this by the magnetic area of yoke
and magnet cores we obtain

Bi flux-density in the yoke.

Ba „ in the magnet cores.

oy Google

28 Dynamo Design,

Also, by dividing the useful flux N^ by the magnetic area of
air-gap, teeth and armature-core we obtain respectively

Bg flux-density in air-gap.

B^ „ ' in teeth.

Bj „ in armature-core.

To find the ampere-turns necessary for yoke, magnet-cores,
teeth, and armature-core, all we have to do is to take the
magnetizing curve for the respective material, and see how
many ampere-turns per unit length are required when these
parts are worked at flux-densities of Bi, Bj, B4 and Bg, re-
spectively. Let the numbers so found be noted by Si, h^, S*
and 3f Then

Ampere-turns for yoke = Li X Si

„ for two field-cores = Lj X Sa

„ for two sets of teeth = L4 X 5»

for armature-core = Lj X Sj

With regard to the ampere-turns ■ for the air-gaps, we
have to resort to the use of gap-coefficients. The coefficient is
10^4^7 if centimetre units are used ; or if the flux -density is
in lines per square inch, and the length of magnetic path is
reckoned in inches, the coefficient becomes 0'3I33. These
coefficients are used as follows ; —

Ampere-tums for gaps = o-8 X oS's X L'3 in centimetre units.
„ „ = 0"3I33 X B3 X L3 in inch units.

The sum of the ampere-turns required for the diflferent
parts will then give us the total ampere-tums required per
pair of poles.

Some little discretion must be used in reckoning out what
may be termed the " magnetic dimensions " of the machine,
that is, the mean magnetic path, and the effective iron area
traversed by the flux. It may not be out of place to take up
this question a little more fully.

{a) Yoke. — The only point to be remembered here is that

izecoy Google

Magnetic Calcu'ations. 29

in multipolar machines the yokes will only carry half the
total flux, as it will divide each way. The magnetic length
is the mean length of path.

(b) Magnet-Cores. — The magnetic section is simply the
section of one core. The magnetic length L^ is that of two
pole-cores.

{c) Air-gap, — The length is twice the distance from iron
to iron. With regard to the magnetic section to be taken, it
is always more or less a matter of judgment and experience,
on account of the spreading of the flux from the pole-piece,
or fringing as it is frequently called. For machines having
smooth core armatures, and where the length of pole-piece
is equal to the gross length of the armature core, the magnetic
area of the air-^ap may be taken as the area of one pole-piece
plus a small area equal to the length of one air-gap multiplied
by the periphery of one pole-piece. For machines with slotted
armatures, the air-gap area may be taken as the mean of the
pole-face area and of the iron area at the face of the teeth
under one pole. But the number of teeth so reckoned should
be increased by one or two over the actual number under one
pole, to allow for the fringing ; such allowance depending
upon the length of the gap, shape of the teeth at the armature
periphery, and flux-density at which they are worked. On
account of distortion of the field, the magnetic area of the air-
gap may be different at full-load from what it is at no-load,
but the two rules above will generally be found good enough.
{d) Armature Core. — Here again if the machine is multi-
polar, the core will only have to carry half the useful flux. The
magnetic length is the length of the mean path lying between
the roots of the teeth and the periphery of the internal hole.
The magnetic section is less than the gross section by 10 to
25 per cent., on account of the insulation of the core-disks and
the presence of ventilating ducts. If these latter are absent, as
is usually the case with small armatures, allow lo per cent, as
space-loss if the disks are varnished, and 15 per cent, if paper
insulation is employed. If air-ducts are present, their width
must be subtracted from the gross length when computing the
area. For paper insulated armatures with the usual allow-

.„Googk'

30 Dynamo Design.

ance of ventilating ducts, the nett length of core (parallel to
shaft) is generally 75 per cent, of the gross length.

ie) Teeth. — The total length of tooth traversed by the flux
is equal to the depth of a slot multiplied by 2. The width of
one tooth to be taken as the mean width. The number lying
under one pole may be taken as the number of teeth in the
polar angle plus one or two, depending on the length of the
air-gap, in order to allow for spreading. The magnetic area
of one tooth will therefore be the mean width of tooth multi-
plied by the nett length of armature (that is, gross length
minus total width of air^ducts minus 10 to 15 per cent, space
lost through insulation). But there is yet an important point.
If the teeth are worked at densities of 100,000 lines, or more,
per square inch, part of the useful flux will pass into the core
by way of the slots, because these offer a path in parallel whose
magnetic conductivity is comparable with that of the teeth
themselves.

It follows, therefore, that the ampere-tums for the teeth
calculated out on the basis that they carry the whole of the
flux, will be in excess of the right amount at high values of
tooth flux-density. We will now proceed to show how the
true value of tooth flux-density B* may be estimated if we
know the apparent flux-density in the teeth which we will
call Ba. Further, we will denote by

by mean width of tooth ;

b% width of slot ;

/ nett length of armature, that is, the iron length parallel

to shaft ;
k height or depth of slot ;

f ratio of nett length to gross length of armature core ;
Na flux from one pole, as before ;
N^ flux actually carried by teeth.
Then we have

Iron section of one tooth = ^1 x ^

izecoy Google

Magnetic Calculations. 31

The actual section of air-space per slot forming an alter-
native path in parallel for the flux is given by the area of one
Aatplus the area of the space lost along one tooth by insula-
tion and ventilating ducts, or

Section of air-space \_^aX/,/ _j-, d ^ x I
per slot / - J + ^ > /J -J^

Now the flux N„ coming out of the pole-piece will divide
itself between tooth and air-space in inverse proportion to the
reluctance of these two. The flux in the air-space is (No — N4).

Hence we have

where p. is the pernieability of the tooth when transmitting
the actual flux N,,
Also

and by division

{K - N.) h + di- ii/
N* (h + h- Kf + hfp) = wjh ^

As stated above, a common ratio of iron length to gross
length for slotted armatures with air-ducts and paper insu-
lation is 0*75. Putting in this value of/ in the above equation
we have

B« _ 0-75 y-htt.

6« *a + 0-25 *i -I- 0-75 *i y-

B4 _

Ba I ■ 34 *i + o- 33 by-¥ hiL
To put this into practical shape, take ratio of bi to l

oy Google

Dynamo Design.

HJ.3aj. 3HX Ni ^g AJ.i9Naa -xn-id -ivr>j.3v

izecoy Google

Magnetic Calculations. 33

assume values for B,, find the corresponding permeabilities
from such a curve as that of Fig. i aqd calculate B,. Then
a curve connecting B^ and 84 for this particular ratio of b\
to b^ can be plotted, showing what the true flux-density in the
teeth is when the apparent flux-deniity (that is, total flux per
pole divided by iron area of teeth under one pole, or N^ -J- A,
= Ba) has any particular value. This has been done in the
three curves shown in Fig. 7 for three usual ratios of b\ to b^
using the above equation. If/ has a value different from o ■ 75,
the equation should be correspondingly altered and new
curves plotted when great accuracy is desired-

ExAMPLE OF Calculation.

In order that the foregoing rules may be clearly under-
stood, and to exemplify the use of the curves, etc., we will take
a concrete case for purpose of illustration. In Fig. 8 is
given a dimensioned sketch of part of a modem six pole
200 kilowatt machine. We will proceed to calculate how
many ampere-turns per pair of poles are required in order to
produce a flux of 12,500,000 lines through the air-gap.

A reference to the table of dispersion coeflicients on page
23 gives us an approximate figure,

v= I- 18,
and hence,

N„ = Na X I ■ 18 = 14,750,000.

The next thii^ to do is to make an estimate of the
magnetic lengths and areas. We have
Yoke. Area= 17-5 x 5

Ai = 87 ■ 5 square inches.

For the length of mean path, we can either scale it off from
the drawing, which is, as a rule, more convenient, or estimate
it from

L, = 5 + i (5 9-9+ 33'5 + 5) X 314 1

= 56-5 inches.

izecoy Google

Dynamo Design.

izecoy Google

'.ic Calculations. 35

As no allowance has here been made for the rounding off
of comers along the mean path taken by the flux, we may
say

L. = 55-

Magnet Cores. — As these are circular, we have

A3 = 14-25 X I4"25 X 0-785
= 159" 5 square inches.

The magnetic length of the two cores and pole-pieces is
L, = 2 X 16-75 = 33-5-

Air-gaps.

= 59-9- 59'35
= o"65 inch.

For the area, we take, as stated above, the mean between
pole-piece area and the area of the teeth under one pole at
their tops. As the air-space in this machine is short, we take
for the number of teeth acted on by one pole the actual
number lying in the polar angle, plus one. Had the air-space
been longer we should have added two}

From the sketch we see that the polar angle is 44* 3°. As
there are altogether 220 teeth, the number in the actual polar
angle is

220 X ^4^ = 27.

360

Adding one to this, we have 28 as the number transmitting
the flux. Now the iron area of a single tooth at the top is

1 14-25 - (3 X 0-375) I X 0-9 X 0-439
= 5 -06 square inches.

' This allowaoce for iae fringing of the magnetic field, which increases the
useful flux entering th<! armature frum one pole, is a matter of judgment and
experien>:e. Fischer- Hianen has given elaborate rules. - For smooth-cored a.rma-
tures it is usual to estimate the width of the fringe as equal to the gap from iron
to iron. See a paper ilso bj SaDiler in the Zaltchri/I fUr Etdtiroucknik, xviii.
563, Nov. 1900.

D 2

izecoy Google

36 Dynamo Design.

The iron area at the top of the teeth under one pole is
hence

28 X 5 '06 = 142 square inches.

And the area of the pole-face is

13 x(59'9 X 3-1416 X^)
= 302 square inches.
Hence, we have for the magnetic area of the air-space,
Aj = -5^"tJ_ = 222 square inches.

Teeth. — For the magnetic length we have
Li= 2 X 1-625 = 3*25 inches.

And the mean iron area of the 28 teeth acted upon by one
pole is

{14-25 - (3 X 0-375)} X 0-9 X 0-406 X 28,
A, = 134" S square inches.

Armature Core. — The mean leng^th of the path taken by
the flux is best obtained from the drawing ; otherwise we have

8-75+{('"^3^5)x3-.46xl}
= 33*55 inches.
Or say L( = 33 inches.
The magnetic area is

{14-25 -(3 Xo-375)} X 0-9x8-75,
A> = 103*5 square inches.

Having now found the magnetic dimensions, we can con-
struct the table given below. The flux-densities have been ob-
tained by dividing the flux in each part by the corresponding
magnetic area ; as the density in the teeth is in this case
below 100,000, we may assume that no correction is necessary
— that is, we may consider B^ = Bj, the entire flux being

izecoy Google

Magnetic Calculations. 37

carried by the teeth. The rest of the working is sufBciently
obvious, the final result being that to force the 1 2 ■ 5 megalines
through the iron and across the gaps, something over 14,000
ampere-turns per pair of poles are required. The actual
number would be taken in practice as 15,000 at least, in order
to allow for differences in the iron, etc.

Na= ia,soo,oc»; Nk= 14,750,000; >■= im8.

FmofMaduK.

Material.

■sse

^S^

c^ii.

1 fm"

Required.

Yoke

Cast steel

33-5

87-5

84100

34-a

1330

3 Magnet com

Ditto

1
159 -S 92500

34-5

1090

a Air-gaps

Air ! 0-6S

134-5

56300
93000

XO-3I33

11500

2 Teeth

Sbeet iron

3-25
33

Armature core

Ditto

103 s

60400

133

Total ampere-tums per pair of poles = 141 14.

By similar calculations we can find the ampere-turns
required to force other values of N^ across the air-gap and
through the iron parts of the machine. By plotting the
values of excitation so obtained against the corresponding
values of N^, we obtain what is known as the saturation
curve, of the magnetic circuit in question ; the ordinates of the
curve representing also the corresponding values of the induced
electromotive-force to a different scale. Examples of such cal-

• This number o'3i33 is the gap-coeffidtnt and represents the number of
ajnpere-tnms per inch length of path requisite for a flui-densily of I line per square
inch. Multiplying the llux-density of the preceding column by this coeSicieDt gives
the numher of ampeie-tums needed for that density, per inch of path in air ; and
multiplying this number by the magnetic length of the a air-gaps, in column 3,
gives finally the number of ampere- turns needed for the 2 gaps.
56300 XO'3i33XO'6s= 11500. ■

oy Google

Dynamo Design.

culated curves will be found in Chapter VIII. on Examples of
Dynamo Design.

We have here calculated the ampere-turns needed for a
pair of poles ; but as the two halves of the magnetic circuit so
considered are alike, one may, if preferred, calculate simply
the ampere-turns per pole, taking only one-half of a magnetic
circuit, including, of course, one gap, one pole-core, etc., and
taking yoke and armature core at half the lengths estimated
as above. A convenient form of schedule for such calculations
will be found in Appendix I,

, FEG 9 — CORK-DlSTttlBUTION OF FLUX

In a careful study, in part theoretical, but confirmed by
experiments, Goldsborough ' has shown that in the armature
of a multipolar dynamo the paths of the magnetic lines
through the armature core not symmetrical, and that they
are not distributed with equal density through the cross-
section of the core (Fig. 2), being denser in the region
immediately below the roots of the teeth, and less dense
near the intenal circumference of the core. At full load these
inequalities are more marked. As a consequence any calcu-
lations as to hysteretic losses in the core made on the assump-
tion of uniform distribution will understate the actual waste of
energy.

' Air-gap and Cote Distribution Studies, Trans. Amir. Init. El. Engineers,

izecoy Google

Magnetic Calculations. 39

Similarly, Mr. Dettmar has shown in the Eiektrotechniscke
Zeitsckrift for igoo, vol. xxi. 944, that the density of the
flux in the core-body diminishes, not quite in arithmetical
proportion, from a maximum below the roots of the teeth to a
minimum at the internal periphery.

If the pole-pieces are not laminated, the width of the gap
should not be much less than about ij times the width of the
slot, otherwise the unequal distribution of the flux at the pole-
face will set up harmful eddy-currents.

Professors H. Hele-Shaw and A. Hay have published in
the Philosopkical Transactions, cxcv. 303, 1900, a very remark-
able paper on lines of induction in a magnetic field, the dis-
tribution of which they have studied by the aid of a beauti-
ful hydraulic model in which the stream-lines in glycerine
imitate the forms of the magnetic lines under varying con-
ditions. Amongst these they show the distribution in the case
of a toothed armature with a gap approximately equal to the
breadth of a tooth and with slot slightly wider. In the gap
the density of the lines shows alternate maxima and minima,
the lines being very slightly curved at the level of the teeth ;
but below this level those that enter the slot swerve sharply
round to enter the flanks of the teeth.

Except in the case of very highly saturated teeth, there is
no field in the slot at any greater depth than about equal to
the slot width. The ratio of the density of the field in the
slot to the density of held in the tooth is roughly the same as
the ratio of the gap-length (from iron to iron) to the sum
of gap-length and tooth length.

Herr Dick has shown, in the EUctrotecknische Zeitsckrift
for July 1901, that if account is taken of the flux-densities
along the tooth, the ampere-turns actually needed will be con-
siderably less than the number (only about ^) calculated from
the mean between the maximum value at the roots and the
minimum value at the tops of the teeth. In the same journal
for November 1901, Dr. Corsepius has shown how the design of
armatures is dependent on the ratio between the width of the
tooth and the width of the slot.

oy Google

Dynamo Design.

CHAPTER III.
COPPER CALCULATIONS: COIL WINDINGS.

Weight of Copper Wire, — Pure copper has a specific
gravity of 8'9 at the ordinary atmospheric temperature of
15° C. Hence

I cubic centimetre weighs 8 -9 grammes ;
I cubic foot „ 555 pounds;

I cubic inch „ 0*3213 pounds.

A rod of copper i inch in diameter, and I foot in length
weighs 3 ■ 028 lbs. Hence the weight of a copper wire can be
found by multiplying tt^ether its length in feet, its sectional
area in square inches, and the coef^cient 3*028. A wire
I mil in diameter and i foot long weighs 0*000003028 lb.
Hence if rfbe the diameter in mils, and / the length in feet,
the weight of the wire will be

weight in lbs. = (? X / X 0-000003028;

or a wire d mils in diameter weighs lbs. per foot

330250

Example. — 30 feet of a No. i S.W.G, copper wire, which is 300
roils in diameter, weighs 8 ■ 17 lbs.

In the case of copper strip of rectangular section, if the
width and depth of the strip are given in mils, and the length
in feet, the weight in pounds can be found from the rule that
. weight in lbs. = sectional area in sq. mils X / X 0*000003855.

Electric Resistance of Copper. — Pure copper has a specific
resistance that increases slightly with temperature.

itizecy Google

Copper Calculations. 41

The resistance of a centimetre cube of pure copper, in
ohms, has the following values : —

At eye.

.5-C.

At 30° C.

At6o°C

AnnaOed

o' 00000159039

o-oo<

300169159

000000179559

0-00000200401

o-o<»ooi62a46

000.

300172676

o- 00000183 180

0-00000204442

At 3!" F.

At6cPF.

At 115= F.

o" 00000159039

0-00000169639

0-00000.98847

Hmid-dr»wii

□'00000163246

0-00000173054

' 00000202556

The rise of resistance of copper with temperature is
approximately -^ of one per cent., per Centigrade degree, or
I of one per cent, per Fahrenheit degree.

If the resistance Rooo. at freezing-point, of any copper con-
ductor be known, its resistance R^on. at any temperature Q on
the Centigrade scale, can be accurately calculated by the
formula of Clarke, Forde and Taylor :

R«!. = R(i»o. {I +o-oo426;44^+ o-oooooiii93(9"};
or on the Fahrenheit scale,
R*S.. = R3»»F.{'+°'°°^37o8(^-32)+o-oooooo34S48(^-32)'}.

The following are some useful rules for calculating the
resistances of copper as used in construction of electric
machines. In all cases it is assumed that the material is
pure annealed copper, commonly called "high conductivity"
copper. If the copper is " hard drawn " instead of " annealed "
the resistance may be some 2 per cent greater for an equal
cross-section and equal length. Resistances are given in ohms.

British Units. — Resistance of i inch cube is

o-ocx3O0o626i5 at o°C. or32''F.

o ■ 00000066639 at \ 5° C.

000000066788 at 60° F.

0-00000070694 at 30° C.

0-00000075085 at 115°^

0-00000078899 at 60° C.

izecoy Google

42 Dynamo Design.

A rod of copper, i foot long and i square inch in cross-
section, has the following resistance : —

0-0000075138 at o'C.

00000079966 at IS^C.

0-0000084833 at 30° C.

00000094679 at 60" C.

A rod of round copper, i foot long and i inch in diameter
(having therefore a sectional area of i circular inch), has the
following resistance : —

0-0000095664

o^C.

O'O000II08l2

15° C.

"0000108007

30" c.

0*0000120545

60° C.

A wire of copper, i foot long and having a sectional area
of I square mil, has the following resistance : —

7-5138 ohms

at 0" C.

7-9966 „

al IS-C.

8-4833 „

at 30° C

9-4679 „

at 6o°C.

A round wire of copper, i foot long, having a diameter of
I mil (and therefore having a sectional area of i circular mil),
has the following resistance : —

9-5664 ohms at

o°C.

IO-I8I2 „ at

.5°C.

10-8007 » at

30" C.

12-0545 „ at

60° C.

The resistance of a copper strip, the length of which is
given in feet and the sectional area in square mils, may
therefore be calculated by the rule : —

At 0° C. ohms per foot = 7*5138 \ ■■ ., , .

A . o ^ ^^\ divided by

At 15° C. „ = 7-9966 . ■'

A^ .„o <- o. o.. S area in

]. mils.

.coy Google

At 30= C. „ = 8-4833 I

A. i o ^ \ sq. mils.

At 60° C. „ = 9-4679 I ^

Copper Caknlaiions. 43

The reaistance of a round copper wire, the length of which
is given in feet and the diameter in mils (which diameter, if
squared, gives the sectional area in circular mils) may
therefore be calculated by the rule r —

At 0° C. ohms per foot = 9- 5664^1

At is=C. „ = 101812 I

At 30° C. „ = 10-8007 [

At 60° C. „ = 12 -0545 J

divided by
diameter
squared.

Metric Units. — A rod of copper, i metre long and of
I square millimetre cross section, has the following resist-
ance : —

0-0159039 ohms at 0° C.

0'0i6c2S9 „ at 15° C.

00179559 11 at 30° C. I

0-0200401 „ at 60° C.

A round wire of copper, l metre long and i millimetre in
diameter, has the following resistance : —

0-0202487 ohms at 0° C.

0-0215499 It at 15" c.
0-0328614 " at 30° c.

0-0255148 „ at 60° C.

The resistance of a copper strip, the length of which is given
in metres and the sectional area in square millimetres, may
therefore be calculated by the rule : —

At 0° C. ohms per metre = 0-0159039] dj^yed by
At IS C. „ = 0-01692591 .,,/

At 30° C. „ = 0'0I795Sq( ^ ^

A.i = r- /i'^jy metres.

At 60' C. „ = 0'020040i J

The resistance of a round wire Xh^ length of which is given
in metres and the diameter in millimetres may be calculated
by the rule : —

At 0° C. ohms per metre = 0-0202487]
At 15° C. „ =00215499 1

At 30° C. „ = 0*0228614 ]

At 60" C. „ = o'0255i48'

I divided by
diameter
squared.

izecy Google

44 Dynamo Design.

Example I. — To find the resistance at 60° C. (warm) of a copper
strip 9'5 feet long, the rectangular section of which measured bare
is iiS mils by 785 mils. The product of 118 and 785 gives as the
sectional area 92630 sq. mils. Hence by the rule given above the
resistance of one foot length is 9-4679-;- 92630 = 0*0010322 ohm.
Therefore the resistance of 9-5 feet at this temperature is 0-00971
ohm.

Example II. — To find the resistance at 60° C. (warm) of a shunt
coil of 3050 turns of a round copper wire. No. 16 S.W.G., the mean
length of one turn being 5-21 feet. No. 16 S.W.G. has a diameter
of 64 mils, therefore a sectional area of 64 x 64 (= 4096) circular
mils. Hence by the rule the ohms per foot will be 12 '0545 divided
by 4096 =! 0-002936 ohm. So the total length being 15890 feet, the
total resistance will be 46 "65 ohms.

The following rules are useful for copper wires at 30° C. :

Section in sq. mils = 10 -8 X length in feet -j-

resistance in ohms.

Length in feet = section in sq. mils X resist-

ance in ohms -r- I0"8.

8483 -~ section in sq. mils = resistance per rooo feet of
length.

Electrical Measurement of Temperature. — If the rise of
temperature of an armature or of a field-magnet coil is
measured at the surface by the common process of laying
upon it the bulb of a thermometer covered with a pad of
cotton wool, the temperature so measured will not be the true
temperature of the interior, but considerably below the true
average temperature of the armature or coii. If the resistance
of the coil is measured, then the true internal temperature can
be ascertained, provided the resistance of the coil at 0° C, or
at 15° C. is known. For practical purposes a near enough
approximation can be found by the formula : —

where R is the resistance as measured when cold, and R' the
resistance as measured when hot

oy Google

[Copper Calculations. 45

Stranded Copper Conductors. — Stranded copper wires are
seldom now used in dynamo construction ; but compressed
stranded conductors are still occasionally found in smooth-
cored armatures. An example is furnished by the toothed-
cored armatures of the motors of the Central London Railway
in which are employed conductors made of 49 strands of
No. 19 B. and S. gauge, having an apparent cross section of
0"o6o sq. inch. Now i such wire has a section of O'ooioi
sq. inch, and 49 would therefore have O' 049s sq. inches in total.
But allowance must be made for the increased length due to
twisting of the strands, and experiment shows the conductor
to have such a resistance that its equivalent section would be
only o ■ 046.

Space-Factor.

In all cases where insulated windings, whether of wire or
strip, are used, it is obvious that the copper section in any
slot or tunnel through the core disks does not occupy the
whole of the space, and the fraction of the space occupied
obviously depends upon the thickness of insulation, and upon
the shapes of the slots and of the conductors. The ratio of
the nett cross-sectional area of the copper in a slot to the
gross cross-sectional area of the slot is called the space-factor.

Space-Factor in Armatures. — The insulation within a slot
consists of two parts : that which is employed as a lining to
the slot to protect the iron from contact with the conductors,
and that which surrounds the individual conductors to protect
them from contact amongst themselves. The slot- lining
must be relatively thick, because the iron core must be insu-
lated from the full voltage of the machine, while the insulation
around the individual conductors may be much thinner, as
the difference of potential between any conductor and its
neighbours will only be a small fraction of the full voltage.
If every conductor had a slot to itself each slot must be lined
with the thicker insulation ; whereas if several conductors
are placed in one slot, one stout lining will surround them all,
and a larger fraction of the area of the slot will be filled with
copper. The space-factor is therefore higher if the conductors

oy Google

46 Dynamo Design.

are so grouped. In the lighting generator of Scott and
Mountain, p. 160, working at 250 volts, the area of the slot is
©■65 square inch. The total area of section of the four copper
conductors in the slot is 0*308, so that the space-factor is
0'473. Suppose the case of a conductor, the section of which
was 500 X 150 mils, having therefore a sectional area of 75,000
square mils. Let this be overwound with thin insulating tape
to a thickness of 15 mils, making the dimensions, covered,
530 and iSo mils : its gross sectional area will be 95,400
square mils. Now suppose that to insulate it properly from
the iron core a paper and mica insulation 60 mils thick all
round is necessary, the slot area for one such conductor
must be obviously increased to 650 X 300 = 195,000 square
mils at the very least In practice it will be more, as there is
usually a little extra space allowed for packing, and for a
wedge under the binding wires. The space-factor cannot
possibly exceed o'395. But if four such conductors are put
together, and the thicker insulation simply surrounds the
group, the area of slot will have to be at least 1 180 X 480 =
566,400 square mils, and the maximum space-factor will be
raised to 0'529.

The space-factors in the armatures of some 500-voIt machines
are as follows: — Oerlikon Co.'s machines o-6, 0*727, and o'8;
Parshall's lo-pole 06; Kolben's 6-pole and lo-pole 0*417 and
0*516; Ganz's 6-pole 0*413; Hobart's large generators 0-46,
o'49, and o'5i.

In a series of sso-volt generators designed by Mr. S. H. Short,
ranging from 200 to iioo kilowatts, the space-factor ranged from
0433 to o'533, with a mean value of 045.

Of machines at other voltages : —
Brown, Boveri and Go's. 350 volt, 032 : 1000 volt (page 204),
0-214; 120 volt 0-505.

Thury's metallurgical machines (page 224), 0*53 and o'57,
Kolben's 4-pole at 260 volts 0-506; i8-pole at 115 volts, 0-53.

Mr. Rothert states that in a series of 240-volt machines of
all sizes, the space-factor of the magnet coils varied from o'~5
to 0-7.

izecoy Google

Copper Calculations. 47

In the case of high-voltage machines, particularly alter-
nators, the space-factor is greatly reduced. The following
example, due to Herr Kando, illustrates modem practice.

In Fig. 10 are shown three separate cases where slots of
the same size are filled. In a, for a 500 volt machine there
are 4 conductors each of 103 -5 square mm, section; or in
total 414 square mm. of copper. Each is separately sur-
rounded with its own covering i mm. thick. The area of the
slot is 668 square mm. So the space-factor is 0'63. In b,
for 3000 volts there are 24 round wires, each 3 mm. in
diameter, each also covered to a thickness of i mm. The
slot lining must be about 4 mm. thick. The total section of

Fig. 10. — Slot-Sface at Different Voltages.

copper is 170 square mm., and the space-factor has fallen to
0'25. In c, for 10,000 volts there are 80 wires, each 0'8 mm.
diameter, with individual coverings o*8 mm. thick. The slot-
lining must be increased in thickness to about 6 mm. The
total section of copper has fallen to 40 square mm. ; and
the space-factor to o ■ 06.

Space-Factor in Field-Magnets. — In winding bobbins for
field-magnets the space-factor is determined largely by the
question whether round wires, or wires of square or rectangular
section are used. In cases where rectangular wires are em-
ployed there is less waste space : and moreover there is a great
gain in avoiding such waste space as that which fills the inter-
stices between the wires, whether air or insulating material.
Insulation is always a bad conductor of heat, and prevents
the internally generated heat from escaping as quickly as it
should. Of all non-conductors of heat, entangled air is the

48 Dynamo Design.

most perfect, witness the non-conducting properties of felt,
eider-down, etc. Therefore square or rectangular wire should
always be used if possible.

If round wires are used, the space-factor will be determined
chiefly by the relative thickness of the wires and of their in-
sulating covering; but it will also be affected by the question
of the partial bedding of the wires of one layer between those
of the layer beneath. Suppose the wires to lie in precisely
square order, without bedding, as in Fig. 1 1 ; then if the

Square Okder of Bedding. Hexagonal Ordei! of Bedding.

diameter of the bare wire is d, and that of the wire covered is
d^, then since the area of each small circle is O" 78541^, and as
the area of the small square enclosing the outer circle is rf,',
the ideal space-factor would be

Or, with an infinitely thin insulation it could never exceed
0-7854.

Suppose however an extreme case of bedding, as in Fig.
12, so that the wires lay in hexagonal order like the cells of
a honeycomb, the space-factor then would be

or with infinitely thin insulation would be 0-906.

If rectangular strip is used, uniformly covered, there is no
bedding and no idle space save at the ends of a layer where
the coil ascends to the next layer. If the breadth and thick-
ness of the bare strip are called a and b, and when covered
fli and bx, the space-factor is simply ab~-ax h- Edge-wound

izecoy Google

Copper Calculaitons. 49

strip has the highest space-factor of any winding. Messrs.
Ferranti find it to range from o'83 to 0*93.

Now, in practice, there is with round wires very little
bedding. Some writers have assumed 10 per cent., others
15 per cent. But this is far beyond the facts. Especially in
the case of bobbins of small diameter the wire refuses to bed ;
since, as the successive layers are wound from right to left,
and then left to right, each turn must at some point ride over
a turn in the layer below it Bedding, even in the hand of an
experienced winder, seldom exceeds 3 per cent. The safest
course is to assume that there will be no bedding at all and to
take the space-factor, if not known from actual experience, as
given by the formula above.

For the shunt- windings of dynamos of standard types at
say 500 volts, the space-factor has values seldom below 0'45-
Tbis is the figure for the Scott and Mountain 6-pole machine,
on p. 160; the Kolben lo-pole machine, p. 216, has 0"6o.
Mr. Mavor gives values from 0-43 to O'jos for the magnets
of M avor and Coulson dynamos.

Some actual figures are given by Dr. S. S. Wheeler for a
number of different wires insulated to different thicknesses.

These are exhibited graphically in Fig. 13, the full curves
representing the observed values, and the dotted curves the
values by the formula assuming square order. It is seen that
the larger sizes of wire do actually bed a little, giving a
space-factor slightly higher than the calculated value.

Calculations of Bobbin- Winding. — The space-factor is
closely connected with another important quantity, namely
the resistance per cubic inch of the winding. By this expression
is meant the total resistance of the winding divided by the
number of cubic inches of volume which it fills. Suppose a
wire covered to the thickness of 100 mils to be wound on a
bobbin. There will be 10 wires side by side per inch length
of the bobbin, and (if no bedding is assumed) there will be 10
layers per inch thickness, therefore, 100 wires through the
square inch of cross-section. A cubic inch taken orthogon-
ally (neglecting curvature) would therefore contain 100 wires
each one inch long, and if these were joined in series with one

oy Google

50 Dynamo Design.

another, the total amount of resistance within that cubic inch
would clearly be icxj times the resistance per inch. If Si be
the diameter of the insulated wire, then l/V will be the
number that go to a square inch of the winding section.
Hence if we know the resistance per inch of the copper wire
used, the resistance per cubic inch of the winding can be found

J^^Si.

jlMt.

—

„.

''''

■/■

VALUES OF THE

SPACE FACTOR FOR

ROUND WIRES OF

VARIOUS GAUGES

M 40 CO M 100 IZO .

K. .

O 2.

JO 2Z0 MILS.

3o» «>«» .V le .i, - i i ; B«.s-

Fig. 13.— Space-Factor f

? Different Thicknesses,

by dividing it by the square of the diameter of the covered

wire. If thCj diameter of the bare wire is given in mils, we

have : —

Resistance per inch at 1 5° C = o ■ 8484 -i- diameter squared.

„ „ id'C = 0"900l -r- diameter squared.

„ „ 60° C = I '045 -j- diameter squared.

Or, if the diameter bare, d, and the diameter covered, ^u be

oy Google

Copper Calculations. 5 1

expressed in mils, the resistance of one cubic inch of the
winding will be given by the formula : —

0-8484 . ,ar-

o-gooi ^ ,„or-

p,^ Al^l. tea" C.
d* X di'

This assumes, of course, square order in the winding, the

space-factor in this case being — . — , .
4 «i

If the number of turns of wire in the coil be S, and the
area of the winding space be L xT, the number of wires
through a square inch of the winding space will be S / LT.
If we multiply the resistance per inch cube of copper by the
square of this number, and divide by the space-factor we shall
obtain the resistance of one cubic inch of the winding : or

pi = O'ooo,ooo,666 i-5-~j- ohms at 15° C,

pi = 0-000,000,789 j ohms at 60° C.

It will be noted HaaXfor a given number of turns of wire
in a bobbin of given winding space, the resistance per cubic
inch, as well as the total resistance, will vary inversely as the
space-factor.

Example, — A bobbin, of which the nett winding-space is 10 inches
long and ij inches deep, is to be wound with 540 turns of wire.
Assuming a space-factor of o'6, the resistance per cubic inch, at
60° C, will be 0-000000789 X 540 X 540-^(15 X 15 X 0-6) =
0-001704 ohm. And if the mean length of one turn is 44 inches,
the total volume will be 10 X i^ X 44 = 660 cubic inches, making
the total resistance 1-135 <>hms.

izecoy Google

$2 Dynamo Design.

To find the proper gauge of wire to fill a given bobbin to a
prescribed resistance. — Dividing the prescribed resistance by
the volume of the winding space, one obtains the number of
ohms per cubic inch. Reference to a table of wires with
various thicknesses of covering for which the values are known,
will enable the proper gauge to be picked out. For such
Wire-Gauge Tables see the Appendix.

To find the proper gauge of wire to carry a given current. —
Suppose, as in the case of a shunt machine, one can estimate
beforehand the permissible current, by dividing the permis-
sible number of watts wasted on excitation by the voltage of
the dynamo, one may then estimate the gauge of the wire re-
quired by knowing what is a suitable ampere-density. In
stationary coils a density of 600 to 900 amperes per square
inch is customary, (This is roughly from l to r^ amperes per
square millimetre.) Otherwise stated, one allows from 1 100 to
1666 square mils per ampere, or 1400 to 2100 circular mils per
ampere.

Example. — Estimate the gauge of wire required for the magnets
of a 300 kilowatt shunt dynamo at 500 volts. Assume that one can
afford a 1 per cent, waste of energy, or 3000 watts. At 500 volts
this is 6 amperes. The wire, at 600 amperes per square inch, will
require W^ of a square inch or 12,566 circular mils section. Refer-
ence to wire gauge tables shows that the nearest size is a No. 11
S.W.G., which has a section of o'oios square inch, or 13,200
circular mils. This has (at 60° C.) a resistance of o ■ 947 ohms per
1000 feet, and a weight of 41 lb. per 1000 feet. Now that 500
volts shall send 6 amperes implies a total resistance of 500-^6 =
83 '3 ohms. The total length of shunt wire needed will therefore be
83 ■ 3 -i-o' 947 or about 87 times 1000 feet, or 87,000 feet, webbing
about 3567 lb.

In small machines the current density in the shunt-winding
may safely exceed looo amperes per square inch, and even
attain 1400. In the Scott and Mountain generators, p. 198, it
varies from 660 to 1360.

Given the ampere-turns, the volts, and the mean length of
one turn, to find the gauge of the wire, the resistance, the number
of turns and the volume.

izecoy Google

Copper Calculations. 53

This case derives its importance from its use in calculating
shunt windings. Let the prescribed number of ampere-turns
be called CS, neither C nor S being separately known. Let
the volts applied to the terminals of the bobbin be V, and let
M denote the mean length of one turn. (In many cases this
will be only approximately known at first.) Let tx stand for
the resistance per inch length, and p, the resistance (of the
Covered wire) per cubic inch. These last are supposed to be
tabulated for various gauges. Now the resistance R of the
coil may be expressed in two different ways :

. . (I)

R = ^Mr,S

and

M r, CS = V,

whence

(CSJM - '' '

which fixes the gauge.

Example.— Py shunt dynamo with 8 coils in series, working at
200 volts, requires 5200 ampere-tums of excitation per pole. The
pole cores are circular of 10 inches diameter : the winding is ex-
pected to lie about 3 inches deep; whence internal diameter of
windings will be about 11 inches, external about 16, so that the
mean length of i turn will be about 43 -5 inches. Then V = 25 ;
CS = 5200; M = 42'5 ; whence r^ = o'oooii3i ohms per inch.
This is equal to o'ooi3572 ohms per foot or i '357 ohms per 1000
feet. Referring to the Wire-Gauge Table in the Appendix we
observe that the nearest laigei size is a No. 12 S.W.G., which (at
60° C.) has a resistance of i' 114 ohms per 1000 feet

If a square wire is to be used, the area in square mils may
be found by dividing 0*79 by the number of ohms per inch
calculated as above.

The gauge having been found, let a suitable thickness of
insulation be fixed upon and the resistance per cubic inch p^
be ascertained. It may be noted that if square order of wind-
ing be assumed then px = ^ -^ d^ \ where d^ is the diameter of

Digil.ze.:,, Google

54 Dynamo Design.

the covered wire in mils : but pi is best taken from actual
tables of windings that have been carried out If pi is thus
known, pi X volume = R. If the volume of the coil is given
this settles the resistance,' and if R is thus ascertained dividing
it by M^i gives the number of turns S. Or dividing R by r^
gives the total length required for the coil.

If the volume of the coil is not prescribed beforehand we
must work from other data. Suppose the number of watts
that may be wasted in keating'\s,%\\(m. (This may be estimated
{a) as a percentage of the whole output, see p. 117; or {b) from
the estimated available cooling surface and the permissible
rise of temperature, see p, 66.) Call the watts that will be
wasted in heating the coil W. Then

W = VC = C'R = V7R . . (3)

and as R = pi X volume
it follows that

volume = — ^ = — p, , . . 4

From this we see that the volume can be calculated if either
the watts or the current are prescribed.

If the permissible temperature is also presc ribed, this (see
p. 66) fixes the permissible number of watts per square inch
of cooling surface ; and this latter being settled, determines
the number of square inches of surface that the coil must have.
If the coil as designed proves to have an insufficient surface,
then it must be re-designed so as to have a longer length, or
else the volume of the whole must be increased, and a greater
weight of copper used ; and if new dimensions are thus chosen
a new value must be taken of the mean length of one turn
and the computation repeated.

It must be ever borne in mind that in shunt windings, if
the mean length of one turn is prescribed, and a given number
of ampere-turns is prescribed, everything depends upon the
resistance per turn, and therefore on the gauge. Suppose a
shunt winding to have 1000 turns, it will have a certain resist-

' See also foimulte by Lowit in Eliklrol. Zeitschr,, xxi. 8St, 190a.

D, Google

Copper Caltmlations. 55

ance, therefore at the prescribed voltage takes a certain current
producing a definite number of ampere-tums. Now suppose
that half the windings are cut out, while preserving the same
mean length per turn. The remaining 500 turns will offer
half the resistance, and will therefore receive twice as much
current as before, bringing up the ampere-turns to the previous
value ; but the C*R loss will have been doubled. Increasing
the length of a shunt bobbin while preserving the same depth
of winding and same gauge of wire, will therefore enable the
required excitation to be obtained with a lessened waste of
energy.

If the volts, heat loss, the watts per square inch, mean
length of one turn, and resistance per cubic inch are all known
the length L of the bobbin (in inches) can be found from the
formula

= (-f-S)-"'^ ■ «

where W is the number of watts of permissible heat loss, and
f the permissible number of watts per square inch, dependent
on the permissible temperature rise.

Another way of calculating the gauge of the wire from the
ampere-tums, the volts, and the mean length of one turn is as
follows I — Let k be the resistance of a wire i inch long and
I mil in diameter ( = 0"9 ohms at 30° C), then

where d is the diameter in mils. We may deduce : —

,,^^. . . (5,

Example. — Takii^, as in the former example, V = 35 ; CS =
Szoo; M = 42"S ; and taking i = o'g, the fomiula gives d = 89 '2
mils, which is between Nos. 13 and 14 S.W.G. If we had taken
the temperature as 50° C-, we should have had k = 0-973, and
rf = 93 mils, which is almost exactiy No. 13 S.W.G.

If square order in the winding be assumed the number of

Digitizecoy Google

56 Dynamo Design.

turns in one layer and number of layers, occupied by a coil of
S turns of external diameter d^ mils, having a nett length of
winding space L (inches), can be calculated from the following
formulae :

No. of turns in i layer = looo L -i- d^.

No. of layers „ = Srfi -=- lOOO L.

Hence the radial depth or thickness T of the coil would
be

T = S(/i -j- 1,000,000 L ;

but owing to bedding, T will probably come a little less than
this. Further, it is not safe to assume without trial that the
number of turns in one layer can be found by the formula
from a measurement of dx made with callipers. The right
way is to try by winding a piece of coil with the wire in ques-
tion. A few turns may be wound on a wooden core and the
length occupied by 10 turns should be accurately measured,
and divided by 10 to find the working value of d-^.

Curves to facilitate calculations for magnet winding have
been given by Mr. H. H. Wood in the Electrical World, xxv.
pp. 503 and 529, April 1895.

The following rules, due to Mr. Kapp, give the weights of
copper in coils, W standing for the permissible number of watts
wasted, D the mean diameter in inches, M the mean length of
one turn in inches (for coils not circular in shape), and CS
stands for the prescribed number of ampere-turns of excitation.

Weightinlbs. = 2-4 x io-« x ^^^^' ; . (7)
weight in lbs. = 0-245 X lO"' X i^^^i'. . (8)

Coil Winding.

Coils for field -magnets may be classified as (a) bobbin-wound,
(b) former-wound. In those wound on bobbins no special
instructions are needed, except as to modes of fixing and
bringing out the ends. Square wire is preferable in every case

izecoy Google

Copper Calculations, 57

where the wire is to be wound to a radial depth exceeding
one inch, as it gives a much better space-factor than round
wire. Better still is edge-strip winding where it can be used.

Field'fnagnet Bobbins. — These are made variously of brass
with brass flanges, of sheet iron with brass flanges, of very thin
cast iron, sometimes even of zinc Some makers use sheet
metal with a flange of hardwood, such as teak. The reader
should examine the examples given in the following pages : in
particular the Scott and Mountain machine, page 160,
Plate II.; the Kolben machine, page 216, Plate VII.; and
the English Electric Manufacturing Co.'s machine, page 222.
Ample pains must be taken to line the bobbin with adequate
insulating materials such as layers of press-spahn, vulcanized
fibre, or varnished mill-board. Great attention must be
paid to the manner of bringing out and securing the
inner end of the coil. If a bobbin b simply put upon a
lathe to be wound, the inner end of the wire, which must
be properly secured, requires to be brought out in such a
way that it cannot possibly make a short-circuit with any of
the wires in the upper layers as they cross it. A method of
winding which obviates all difficulty on this score is to wind
the coil in two separate halves, the two inner ends of which are
united, so that both the working ends of the coil come to the
outside. Fig. 14 shows such a bobbin. The windings are
secured by bindings of tape. This method of construction

Fig. is.--Strif-w

has been used for years in winding the secondaries of in-
duction coils, where it is desirable to keep the ends of the
winding away from the iron core and from the primary coil-

oy Google

Dynamo Design.

Mr. C. E. L. Brown introduced a method (Fig, l6) of piling up
the windings conically, without any end flanges, thus avoiding
some of the risks of break-down ; and for high voltage work
he adopted the plan of similarly piling the windings into two
heaps, so that both the free ends come at the outside. In
winding copper strip for the coils of tramway motors a
similar device has been resorted to ; the union of the two strips
being effected at the interior of the coil as in Fig, l6.

In another form of machine, by Messrs. Alioth, the pole
cores, which are removable, are themselves shaped to serve
as bobbins, and, after being served with a protecting layer of
insulation, are wound in the lathe. This is shown in Fig. 17.

Fig. 17. — Wound Pole-Corz,

Fig. 16.

Former-wound coils are wound upon a block of wood to
which temporary checks are secured to hold the wires together
during winding.' Such coils have pieces of strong tape wound
in between the layers and lapped at intervals over the windings
so as to bind them together. For tramway motors, which are
subject to excessive vibration, very strong tapes must thus be
woven in ; and the completed coil is served with two layers of
tape each separately soaked in insulating varnish. The whole
coil is then soaked with insulating varnish and stove-dried.
Preferably a current is sent through the coils to heat them

' For an illustraled account of Ihe use of former-blocks in windinE, see the
EUttrical Engineer (London) for May 1902.

D, Google

Copper Calculations. 59

iotemally while they are being baked. They thus become
thoroughly mummiBed and hard. For such motors an asbestos
insulation is sometimes prescribed. All field-magnet coils,
whether bobbin-wound or former-wound, ought indeed to be
thoroughly soaked with varnish and. stove-baked.

Bringing out and fixing of Ends. — Figs. iS to 20 illustrate
methods used for bringing out the ends of coils. In Fig. 18
copper strip, laid in behind an
end-sheet of insulating material,
makes connexion to the inner
end, as shown in the upper side
of the figure, while another strip,
shown in the under side simi-
larly inlaid, serves as a mecha-
nical as well as an electrical
attachment for the outer end of
the winding. This device is due
to Mr. Kapp.

Another method, due to
Messrs. Ganz and Co., is illus-
trated in Figs. 19 and 20.

A simple device for securing
the outer end is to fashion a
terminal piece like Fig, 21 so
that it can be laid upon the
windings, the last three or four
turns of which are wound over
its base, and after winding are
bared at the place and soldered
securely upon it.

Insulation of Field-Magnet
Coils. — It is not absolutely

necessary to use any mica preparation for insulation of field-
magnet bobbins, several layers of paper preparations being
more often used. One-tenth of an inch thickness, if made up of
several superposed layers, is generally adequate. Varnished
canvas is useful as an underlay, and press-spahn or vulcanized
fibre for lining the flanges. It is important to protect the Joint

izecoy Google

Dynamo Design.

between the cylindrical part and the flanges. As an example
of careful insulation, may be cited the method adopted at
Schenectady for insulating the magnets of the Edison bipolar
machines, working at lOO to 125 volts, which are insulated as

follows : End-rings of hard rubber are wedged upon the iron
cores with mica. When bits of sheet mica are used, these
are cut to be i J inch wide and at least 3 inches long ; but when
" made mica " sheets are used, long strips 5 inches wide are cut,
and conformed by heating to the curvature of the core. In either
. case the mica projects at least i inch on the inner side of the

Fig. si. — Coil TERMiNAt Piece,

ring. Then over the core is laid one layer of varnished
muslin 24 mils thick, cut to the exact width between the end-
rings. Upon this are placed two layers of plain pressed
board 20 mils thick, cut one inch wider than the width

izecoy Google

Copper Calculations. 6i

between the end-rings, and serrated with V-cuts ^-inch deep
at its edges, so as to allow these edges to make flanges against
the end-rings, the serrations of the two layers breaking joint
one with the other. The total thickness of core insulation is
thus 64 mils, A core-paper is laid between every four layers
of winding. Between series and shunt coils, in compound-
wound machines there is as careful an insulation as on the
cores. When the winding is completed two layers of pressed
board are laid over, and served with an external winding of
hard rope, and varnished.

For machines up to 250 volts, 4 layers of oiled pressed
board are used over the muslin.

For machines up to 500 volts or more, 3 layers of oiled
linen 5 mils thick, not turned up at edges, are placed over the
muslin. Over these come first 4 layers of oiled pressed
board, and then 2 layers of plain pressed board, the latter
with edges serrated to form flanges. This makes a total
thickness of insulation 159 mils. Core-papers are laid between
every 3 layers of winding, and three layers of pressed board
arc served on the outside.

The protective external lagging covering the outer surface
of the completed coils is not altogether a benefit, for it tends
to prevent dissipation of heat.

Ventilation of Field-Magnet Bobbins. — It is not usual to
ventilate the bobbins of field-magnets ; but in some cases, as
for example the magnet-wheels used for alternators, some
makers design the bobbins to be of greater length (parallel to
the shaft) than the length of the pole-core, thus affording
windways. Also the series coils of compound-wound genera-
tors are sometimes supported away from the shunt-coils, with
air-ways between them.

Heating of Coils. — The heat inevitably generated in the
copper coils is dissipated in two ways. It passes by conduc-
tion through the copper and the insulation, either to the
external surface, whence it passes off" by radiation and con-
vexion into the air, or to the magnet core and yoke, which in
turn conduct it away and dissipate it from their surfaces. In
large multipolar machines the masses of metal in the pole-

oy Google

62 Dynamo Design.

cores and magnet frame are more efficacious than the exter-
nal surface of the coil in dissipating the heat

Some considerations in general concerning the heating of
coils may here be discussed. If it be assumed that the
thickness of the insulation is proportional to the thickness of
the wire upon which it is wound, it follows that the weight of
copper in a coil filling a bobbin of even dimensions will be
the same, whether a thick wire or a thin one be used. Further,
for a given volume to be filled with coils, the resistance in ohms
of the coil will vary directly as the square of the number of
turns in the coil. For if a coil wound with lOO turns of a
given gauge be rewound with 200 turns of a wire having half
the sectional area, the resistance of the new winding will
obviously be four times as great as that of the original wind-
ing. Also by a similar argument, it follows that the resistance
of a coil of given volume will vary inversely as ike square of
the sectional area of the wire used. And as the area is propor-
tional to the square of the diameter of the wire, it follows that
the resistance is inversely proportional to the fourth power of
the diameter of the wire used.

The amount of heat developed per second in a coil is the
product of the resistance into the square of the strength of
the current To avoid waste, therefore, no unnecessary resist-
ance should be introduced into any main-circuit coil. It is
easy to show that with a coil of given volume, the heat-waste
is the same for the same magnetizing power, no matter
whether the coil consists of few windings of thick wire or
many windings of thin wire. The heat per second is C* R^,
and the magnetizing power is S C ; C being the current, R^ the
resistance, and S the number of turns. But R^ varies as
the square of S, if the volume occupied by the coils is con-
stant For suppose we double the number of coils, and halve
the cross-sectional area of the wire, each foot of the thinner
wire wilt offer twice as much resistance as before ; and there
are twice as many feet of wire. The resistance is quadrupled
therefore. The heat is then proportional to C* S* : and there-
fore the heat is proportional to the square of the magnetizing
power. If, therefore, we apply the same magnetizing power

izecoy Google"

Copper Calculations, 63

by means of the coil, the heat-waste is the same, however the
coil is wound. To magnetize the field-magnets of a dynamo
to the same degree of intensity requires the same expenditure
of electric energy, whether they are series wound or shunt
wound, provided the volume is the same, and the space factor is
unaltered. Any increase in the space-factor is equivalent to
a lai^er volume, or to the discovery of a wire having a lower
specific resistance. With a higher space-factor the prescribed
excitation can be attained with a lesser waste of energy.
This is the reason for the advantage of using square wire or
strip winding instead of round wire.

A simple way of looking at this matter is to regard the
whole winding as consisting of one turn, there being a current,
equal to the total ampere-turns, going only once round. Then
this current divided by the total cross section of copper gives
the current-density. We then see that for equal-sized bobbins
(containing the same amount of copper) the magnetizing effect
is simply proportional to the current density. Further, the
power wasted per lb, of copper is proportional to the square of
the current-density. The following Table Vil. gives the waste
in watts for different current-densities in both inch and centi-
metre measure. The temperature of the coil is taken at
30' C, at which temperature the resistance of an inch cube of
copper may be taken at O' 7 X 10 ' ohm.

If the volume of the coil (and the weight of copper in it)
may be increased, then the heat-waste for a given magnetizing
force may be proportionally lessened. For example, suppose
a shunt-coil of resistance r has S turns ; if we wind on another
S turns in addition, the magnetizing power will remain nearly
the same, though the current will be cut down to one-half
owing to the doubling of the resistance ; and the heat-loss
will be halved, for 2 R^ x (J C)^ will be J C^ R,.

It is assumed in the foregoing argument that we get double
the number of turns on if we halve the sectional area of the
copper wire. This is not quite true, because the thickness of
the insulating covering bears a greater ratio to the diameter
of the wire for wires of small gauge than for wires of large
gauge. In designing dynamos, moreover, one ought to be

oy Google

Dynamo Design.

guided by the question of economy, not by the accident of
there being only a certain volume left for winding. If there
is insufficient space round the cores to wind on the amount of
wire that economy dictates, new cores should be designed,
having a suMcient length to receive the wire which is economi-
cally appropriate.

TABLE VIL— Loss o

Power in Copper Conductors at Diffekent
C u rkbnt-Densitibs.

CucTcnl

D«i«ty.

w.„

M]. in.

-"S.-

'^^-

Pel cubic cm.

Per lb. of
Cbpi-t.

400

62-

0112

0068

0-548

500

77-5

0-I75

0106

0-S44

600

0-252

0154

0784

700

108-5

0340

0204

I 057

800

124

0-448

0273

'■393

9tx>

139-5

0-S57

0340

i«8

1000

J55

0-7

042

3-17

1500

232

1-57

096

4-88

2000

3'o

2-8

171

8-71

2500
3000

387
465

4-37
6-3

266
384

13-59
19-59

SS"-

542

8-5

Sio

26-43

4000

620

II-2

0-683

34-83

In order then that any coil (whether upon the armature or
iield-system) may not overheat, it must have sufficient surface
relatively to the amount of heat developed in it by the current.
For equal watt loss per unit area of radiating surface, the
amount of heat developed will be entirely different in field-
magnets and armatures, on account of the different conditions
under which the heat is liberated, and consequently we must
consider them separately.

Heating of Field-Magnets and Stationary Bobbins gene-
rally.

izecoy Google

Copper Calculations. 65

Let «/„ be the total watts wasted in the field-coils at
full load, that is, w„ = (C*„ R„ + C% R\).
Ai be the total heat-radiating area of all the
bobbins, in square inches, not counting end
flanges and interna! surfaces (if any).
6„ represent the final temperature rise above the
surrounding air.
Then

The value of the constant h depends upon the depth of
winding, upon the amount of the draught set up by the
fanning action of the armature, and upon the condition of
the air, that is, circulating or still. According to Mr. W. B.
Esson, the value h may be taken as 55 for ordinary field
bobbins. That is to say, an emission of wasted heat at the
rate of i watt per square inch will cause a rise of 55° C. ; or,
if 33° C. be taken as the permissible amount of rise, the coil
must expose i ' 83 square inches per watt wasted in it. This
figure appears to be low for modern machines. The tempera-
ture is here assumed to be measured by thermometer at the
surface of the coil, covered with a pad of cotton-wool. For
the usual shape and dimensions of field bobbins, more par-
ticularly those of multipolar machines other than iron-clad
types, the formula

6„ (in Centigrade degrees) = -^ X 75 . (9)

will be found to give good results. The value of the heating
constant is higher for iron-clad types and enclosed motors.
For shunt bobbins this formula gives directly the maxi-

izecoy Google

66 Dynamo Design.

mum shunt current C„ that may be used if the temperature
rise is prescribed as a limit. Thus

i«, = V X C. = Ci R.

V 75

x-r;- • ■ • <-°'

Or, if the excitation watts and temperature rise are given
we have for the necessary radiating surface of the coils

A, = 7S_x »■. („)

In the case of edge-strip coils being used, the temperature
rise will be much less than that calculated by these formulje,
because in coils of this species the internally-generated heat
is conducted much better to the surface, whence it escapes
without the internal temperature rising so high. Messrs.
Ferranti found the temperature rise after 6 hours in a 1500
kilowatt machine at 150 revs, per min. to be only 16 deg. C.
though the current-density was 920 amps, per sq. inch. In a
150 kilowatt machine at 380 revs, per min., and 1200 amps.
per sq. inch, the temperature rise was only 14' s deg. C. after
6 hours. In one case where an edge-strip winding was in two
layer? with insulation between, though the current-density
was only 800 amps, per sq. inch, the rise was about 28 deg. C
after a 5 hours' run.

At the Oerlikon Works, a limit of 30 deg. C. assigned to
the heating of a stationary bobbin, is found to correspond to
an emission of 0'45i5 watts per square inch : or 2'2 square
inches of radiating surface arc necessary for getting rid of each
watt wasted in heating. This makes the constant h = 66.

If we assume that a limit of temperature rise of 50 deg. C,
above that of the surrounding air is safe, then the largest
current which may be used with a given stationary magnet
coil, is expressed by the formula : —

Maximum permissible current = 0'95*/ - •
Similarly, for shunt coils we have

Maximum permissible voltage = 0"9S^ARi,.

izecy Google

Copper Calculations. 67

Some recent measurements of the rise and distribution of
temperature in field-magnet coils have been made by
E. Brown,' and by Ncu, Levine and HavilL' Brown's obser-
vation made on a bipolar Siemens dynamo led him to note
how efHcacious in promoting cooling was the metal in proxi-
mity. He recommended that the bobbin-heads should be
made as good conductors of heat as possible ; that any gap
between the pole-core and the bobbins should, if possible, be
filled up with good conducting material ; and that, as bobbins
heat most at the mid-length they should be made of less
depth there, that is of an hour-glass form. The Electric Con-
struction Company undulates (see Fig. 92) the profile of its
field-coils for the purpose of better cooling. Messrs. Neu,
Levine and Havill, using a bipolar Crocker- Wheeler motor,
explored the distribution of temperature throughout the cross-
section of the coils, by electrical measurement of the rise of
resistance of the various parts of the winding, and also
measured the apparent rise of temperature with thermometers.
They plotted isothermal curves showing how under varying
conditions the temperature is distributed, when the coil was,
^l) supported in the air, (2) standing on a table, (3) in [^aceon
the machine at rest, (4) in place on the machine running at
full load; in each case the coil being heated for six hours at
the rated voltage. The first case showed the greatest heating,
for, though the table arrested the circulation of air it seemed
to coot the whole coil. The average rise in the four cases was
Z7'S< 33 "9, 22 '7 and 28*3 deg. C. respectively. Incase 3 the
iron core conducted away more heat than the external air, the
point of maximum temperature being nearer to the surface
than to the core. They observed on the machine running at full
load, a rise of 1 10 deg, C. per watt per square inch of exposed
cylindrical coil surf ace ; or on the machine stationary, a rise of
100 deg. C. This makes the formula: —

W„

' fournalofthe Inst, of E!tc. Eti^inars, xxx. 1159, 1901.
* Eieclrical World, ixxviii. 56, July 13, 1901.

F 2

izecoy Google

68 . . Dynamo Design.

leading to the result that if the limit of rise be set at 30 deg. C-
there must be allowed no less than 3'£6 inches of cylindricaJ
surface per watt wasted in the magnet coil. It appeared that a
surface exposed to contact with iron was nearly twice as effica-
cious as a surface exposed to air, leading to the rule : —

6„ (in Centigrade degrees) =:340 ^ 4.^'^^- a ^'^^

where A, is the area exposed to air and A,- that in contact
with iron.

They found the true mean rise of temperature as mea-
sured by increase of resistance to be i'4 to i'6 or more times
as great as the apparent mean rise measured by thermo-
meter.

In the case of enclosed motors, without any resort to arti-
ficial cooling, it is difficult to prevent the internal temperature
from rising by as much as loo deg. C. above that of the sur-
rounding air. Some makers provide their enclosed motors
with external radiating ribs to aid dissipation of the heat.
For these the temperature-rise (according to Niethammer)
may be reckoned as equal to about 95 to 140 times the total
watts lost divided by the /c/a/ surface, in square inches, that i»
exposed to the circulation of air.

Heating of Armatures or Running Coils. — The amount of
heat liberated in a rotating armature depends principally
upon : —

(i) The heat-radiating surface A^. In estimating this, the
number of square inches exposed to the cooling action of the
air are to be taken, but it is a matter of discretion to estimate
what proportion of the internal surfaces contribute to \X.

(2) The peripheral speed v of the winding. For small
armatures and ring winding the average peripheral speed
(feet per minute) as given by the average diameter of the
armature, is to be taken.

(3) The proportion, within limits, of radiating surface to
polar surface. Naturally, an armature nearly covered by the
pole-pieces will not have, as a rule, such a good chance of

izecoy Google

Copper Calculations. 69

getting rid of the developed heat, as one whose radiating
surface is more open to the air.

The heating of an armature in which w^ total watts (iron
and copper losses) are being wasted can be estimated from
the formula ,

^■-X^'<r+(fir.) ■ ■ • <■'*'

where a and b are constants — the values of which are dependent
on the type of machine.

The constant a varies in ordinary well -ventilated machines
of modern design from 50 to go, while constant b appears to
v"dry from 00004 ^° 0x009, if ^ is in feet per minute. The
curves given in Fig. 22 are, however, more convenient to
employ for estimating the temperature rise, as representing
what is usually found in modern practice. The ordinates
represent the rise of temperature per watt per square inch and
the abscissEe the peripheral speeds. Curve A A is to be used
for small unventilated armatures, and is based upon the
average results of Messrs. A. H. and C. E. Timmermann,' and
with tests made upon actual machines.. Curves BB and C C
are to be used for estimating the temperature rise of small
■well-ventilated armatures and large ventilated armatures
more or less of a fly-wheel nature, respectively.

Hence to find the temperature rise Qa of any armature
running at a peripheral speed of v feet per minute : Divide
the number of watts wasted by the number of square inches,
and then from one or other of the curves find the tempera-
ture rise corresponding at the peripheral speed in question.

At the Oerlikon Works, it was found that, taking a
surface speed of about 2000 feet per second, and a permissible
temperature rise of 30° C, each square inch of armature sur-
face (external) could dissipate from i'29 to i*6i watts; or
each watt requires from 0"6 too- 8 square inch. Assuming
4 per cent of the output to be wasted in armature heat, or
40 watts per kilowatt, the necessary armature surface must
therefore be about 24 to 32 square inches per kilowatt of
output.

' Traits. Amor. Inst. Elalr. Enginltrs, x. 1893,

oy Google

70 Dynamo Design.

Owing to the cooling effect of the air-currents when the
armature is running it is found that when a dynamo is stopped
at the end of a long run, the surface temperature immediately
rises above what it was when the machine was running, as the
heat which is being conducted outwards from the hotter interior
is not now so rapidly got rid of. Thus we find that in Admi-
ralty specifications it is laid down that after the end of a run of

^ ^

■JO

- \

CURV

E AA

c BB
E CC

ron s

FOR 3j
FOR I

KRCE

IMTIL

IL*TE

ireoa

yT«0(

RC*

RMATt

BBS

1 '

•s

•v^

ir'

"~~-

"--

■—■

—-

PERI

»MEB,

L BPC

CO l»

Fe«

r PC(

M,.

„.

2000 3000 4000 8000

Fig. 22. — Curves for Estimating Temperatuke Rise.

six hours at full-load, no part of the machine shall at the end of
one minute after stopping show a greater rise than 30° F.
( = i6-6° C.) above the surrounding air. This does not by any
means imply the final temperature rise, because the thermo-
meter will invariably continue to rise for a much longer period
than one minute. But in any case this temperature limit is
needlessly low, as a rise of twice as much would be perfectly
safe, even in the hottest engine-room.

izecoy Google

CHAPTER IV.

INSULATING MATERIALS, AND THEIR PROPERTIES.

Insulating materials may be classified under several
heads : —

(i.) Vitreot4s, including glass, " vitrite," and sundry

kinds of slags,
(ii.) Stoity, such as slate, marble, steatite, mica, asbestos,

kieselguhr, stone-ware, porcelain, "petrifite."
(iil) Osseous, such as bone and ivory,
(iv.) Resinous, including shellac, resins of all sorts, copal

and other gums.
(v.) Bitumiftous, as bitumen, asphaltum, pitch,
(vi.) Waxy, including bees-wax, solid paraffin, ozolccrit,

and the like.
(vli.) Elastic, sach as indiarubber, natural and vulcanized,

ebonite, gutta-percha.
(viii.) Oily, including various oils and fats of animal and

vegetable origin, as well as mineral petroleum,
(ix.) Cellulose, including dry wood and paper; many
natural substances, such as bamboo, wood pulp,
and many preparations of paper and of wood
pulp, papier- mclch^, press-spahn, manila-paper,
vegetable parchment, "vulcanized fibre," cellu-
loid, "Willesden paper."
(x.) Silk, and allied animal tissues such as catgut.
(xi.) Sulphur.
From these materials, or some of them, there are now
manufactured a number of artificial preparations known under
trade names, such as "ambroin," "megohmite," "stabilite,"
"micanite," " vulcabeston," oiled-paper, "empire cloth," in-

oy Google

72 Dynamo Design.

sulating tape, and kindred fabrics; also special varnishes
such as "armalac," "japan," "enamelac," "Sterling's varnish,"
and " Scott's rubber varnish."

Dielectric Resistance. — All insulating materials are mecha-
nically bad. They difter enormously in their specific electric
resistance, and in their power of resisting penetration by a
spark. They all share the particular property that as their
temperature is raised their electric resistance decreases enor-
mously, and in general they become fairly good conductors so
soon as any chemical change begins. Even . marble, glass
and porcelain begin to conduct as electrolytes below a red
heat. Some are liable to absorb moisture from the atmo-
sphere and so lose their insulating properties.

The most, important thing to know about such insulating
materials as are used in dynamo construction is their power
to resist being pierced by a spark. It is also important to
know whether they are hygroscopic, whether they are im-
paired when their temperature is raised, and whether they
deteriorate with time.

Porcelain and stoneware are used for insulating bushes,
and as supports for terminals. Dry wood and paper prepara-
tions such as press-spahn, vulcanised fibre, papier-mSch^,
oiled canvas and the like are only used for low voltages, or as
secondary insulator?, that is insulators which, while mechanic-
ally holding the conducting parts apart, form a backing for
some better primary insulator such as mica.

Experimental data as to the dielectric strength of insulat-
ing materials have been made from time to time by various
authorities. From these the following have been collected.
Fig. 23 relates to layers of pure mica and of oiled canvas of
different thicknesses, and to the number of volts required to
break them down. The experiments, which were made at
the Oerlikon works, consisted in putting layers of the sub-
stances between the electrodes, and gradually increasing the
voltage until the substance began to heat up between the
poles.

Fig. 24 relates to micajtite, that is to say to thin laminae
of mica cemented tt^ether with a special gum such as pure'

izecoy Google

Insulating Materials. 73.

shellac. It is found that when a sheet of micanite is placed
between two electrodes, and the voltage is gradually raised, a
point is reached when, with a sheet of given thickness, a
current begins to flow through the micanite, heating it up-
■within, and producing a burning which rapidly destroys the
insulation. Micanite is a good insulator even at 150° C, and
its break-down when the limiting voltage is reached, appears
to be due to the chemical decomposition, not of the mica, but

2 0.000
16,000

10,000

/y^

ritS--

"'""'

CKNE3S IN P

ACTIONS OF If

■M

O'OZS

Fig. 23.— Curves <

0-Ofi 0-OTS

f DiELKCTRIC Stringth.

of the cementing varnish. Pure mica in sheets, whether of
white or of brownish or greenish tint, if clear has an enormous
power of resisting puncture by the spark, son>e samples with-
standing as much as 5000 or more volts per mil thickness.
Mica-canvas consists of mica scrap sheets about 2 mils thick,
and overlapping one another, cemented with shellac varnish
between two sheets of canvas ; the total thickness being
about 50 mils and withstanding 3000 volts (alternating). Mica

itizecy Google

Dynamo Design,

long-cloth consists of mica scraps similarly cemented between
a very thin " linen " fabric ; its thickness being about 25 mils.
In making each of these compositions the sheets are baked
for at least twenty-four hours in a steam-heated oven.

The data given in the following table are only very
approximate, as in most cases the compositions vary some-
what, and differ at different temperatures. Taking these
figures as being true at ordinary temperature of the air, it is

0'I2B

s
s

6
i

OOSO

,„--''

VOLTS

Fig. 24.— Insuu

ZOOO 30OO 4000

I Voltage of Micanite.

probable that in most cases a rise of 30 deg. Centig. would
reduce them to half their value, pure mica being less affected,
and paraffined paper much more affected than the rest of
those mentioned.

A discrepancy appears between the figures for mica in
this table (800 to 8000 volts per mil), and the Oeriikon
figures (about 22 volts per mil). The Oeriikon figures relate
to the production of the lieating which indicates incipient
break-down, while the figures in the table relate to piercing

oy Google

Insulating Maierials. 75

by a sudden spark. Many makers allow, as a safe insulation,
X millimetre per Joo volts, which is about 12 volts per mil.

Hobart using a mixed insulation allows the following-
thicknesses in armature slots : i • 2 mm. for 115 volt machines,
I "3 mm. for 230 volts, i ■$ mm. for 550 volts ; the guaranteed
insulation test from copper to iron at 20° C. being from 2500 to
3SC» alternating volts applied for one minute.

TABLE vm.

Di.1«tric

Mtgoh«.pcr

M.teii»1.

mil thickn^..

Roui^.

(volI»"«'u,U

'•'■-'■•'■■

varying in quality.

Mica, dear flat . . .

800-SCOO

30,000

MlcanJte plate . . .

900-1200

1.000,000

MicacanvM ....

60-70

300,000

Mica long-dolh . . .

60-70

3co,coo

Mica paper, fleiible . .

300-800

300,000

Ptess-spahn ....

300-400

ICO

hygroscopic.

Paraffined pspei . . .

800-1000

10,000,000

softens when warm.

Oiled paper ....

SOO-900

1350

imade with pure boiled
i linked oil.

Shellacked paper . .

50-US

carlridge paper.

Manila paper ....

120

dried, unvarnished.

Double coltOD, shellacked

350-300

»s

asonD.CC.wires.

Hard rubber ....

500-1200

600

varies much with quality.

Vulcanized fibre . . .

120-200

400

hygroscopic.

Thinnest insulaling tape

150

thickness about 7 mils.

Vulcabealon ....

0-5

Slate

0-5

fought to be boiled in
1 paraffin.

Dry wood is used in armature construction as a packing
under windings to prevent abrasion where they turn through
a sudden curve. Slate is only used for terminal boards, and
must be free from metallic veins. Vulcabeston, consisting
essentially of asbestos cemented together with a small
quantity of rubber and vulcanized, is useful as being capable
of being moulded ; it does not lose its insulating properties if

oy Google

Dynxttto Design.

heated even to 300" C. Stabilit, which is manufactured in
sheets and tubes, is said to withstand io,CXXi or I5,CXX) volts-
with a thickness of i millimetre or about 400 to 600 volts
per mil. It is non-hygroscopic and can be moulded. The
measured resistance is about 40,000 ohms per mil thickness.
Megohmite is a mica composition also capable of being
moulded. All paraffined compositions have a very high,
apparent insulation resistance,' but arc quite unsuitable if the
temperature rises so little as 30 deg. Centig. Cardboard
baked and then while hot impregnated with shellac or other
insulating varnish makes an excellent material for lining
armature slots. Another paper preparation is known as
" Carton Lyon." For all voltages over 500, an insulation con-
taining mica is to be preferred, and indeed is indispensable
for all high-voltage work.

There is a serious objection to resin, shellac, and to
varnishes containing shellac and resin, that these substances
when heated give off vegetable acids which in time corrode
the copper. Hence other varnishes have been sought which
lire not open to this objection. Among these are " Sterling
varnish " and " enamel lac."

The Pittsburg Insulating Company manufactures various
fabrics and papers impregnated with " Sterling varnish " and
is responsible for the following particulars : —

TABLE rx.

Matcnat.

Gr»dt

^'iSr

P^=u,„T«

«EM to Punctute.

Bond paper

1 A

4-5

5.000- 9,000

Fibre paper .

; A

6-7

8,000 - 10.000

Red rope paper

9-10

9,000 - 11,000

Paper . . .

1 A

6-7

8,000 - 10,000

V^ ....

1 ^

9-10

14,000 - 16,000

10,000 vjIIs

Paper . . .

\ c

11-14

30,000 - z5,o;o

15,000 „

Linen . . .

[ A

6-7

5.000 - 9,000

Linen . , .

: B

lO-II

13.000- 15,000

10,000 ,.

Linen . . ,

15-16

18,000 - 20,000,

iLCD, Google

Insulaiing Maiermls. 77

Sulphur mixed while melted with powdered glass, or with
kieselguhr, forms a composition that can be poured into
sockets or cavities and is an insulator.

Insulation of Core-Bodies. — After a core-body has been
assembled and properly clamped upon its spider, it must be
protected by insulation, so as to diminish any risk of making
short-circuit with any of the conductors. Although these are
each separately insulated, insulation of the core-body is also
necessary as a double protection. Smooth cores are insulated
over the cylindrical surface as well as at the ends. Toothed
cores arc insulated along the slots, as well as at the ends.
Core-bodies for ring-wound armatures must also be insulated
along the inner periphery of the core. "Empire cloth" is
found very suitable for covering smooth cores.

Insulation of Slots. — It is usual to line the interiors of slots
with a layer of insulating material. " Empire paper " is suitable
as a first lining. For machines working at 500 volts or under,
a lining of varnished paper or cardboard, 20 mils thick, is con-
sidered adequate, provided the individual conductors or the
groups of conductors are themselves strongly insulated with
micanlte of 40 mils or 50 mils thickness. Sometimes a lining
of mica-paper is used. For still higher voltages, micanite
linings made of sheet-micanite are used. In the case of tunnel
slots, or slots that are nearly closed between T-shaped teeth,
micanite tubes are preferred. The^e, indeed, are general in
the case of high-voltage alternating generators and motors.

For further and more detailed information the reader is
referred to the following sources : —

C. P. Stdnmttz. — -Note on the Disruptive Strength of Dielectrics.
Trans. Amer. Inst. Eke. Engineers, x. 85, Feb. 21, 1893.

Oerlikon Maschineti Fabrik. — Sur le Calcul de Machines
^lectriques, June 1900.

Sever, Monell and Perry. — Effect of Temperature on Insulating
Materials. Trans. Amer. Inst. Elec. Engineers, xiii. 225, May 20,
1895.

Parshall and Hobart. — Electric Generators (1900).

Canfield and Robinson. — The Disruptive Strength of Insulating
Materials, Electrical Engineer (N.Y,), xvii. 277, March 28, 1894.

3,3,l,ze.:,, Google

•78 Dynamo Design.

CHAPTER V.

ARMATURE WINDING SCHEMES.

Armature windings for continuous current generators and
motors may be classified under two heads : —

1. Parallel Grouping.

[a) Lap-windings (drum or barrel-winding).
{b) Ring-windings.

2. Series and Series-Parallel Grouping.

(u) Wave-windings (drum or barrel-winding).
{U) Series ring- windings.

A mixed lap and wave-winding is sometimes used for
grouping former-wound coils.

For a machine of prescribed speed and voltage the number
of armature conductors necessary to produce the given
voltage will depend not only on the number of poles and on
the magnetic flux per pole, but on the grouping adopted for
the conductors. The formulae connecting these quantities are
as follows, the symbols used having the following meanings :

E = the prescribed number of volts to be generated.
« = number of revolutions/^ i^iTCMf/.
/ = the number of poles,

c = the number of circuits or paths that are in parallel
through the armature from brush to brush,
Ca = the whole current carried by the armature.
Z = the whole number of conductors carried in the slots

of the armature,
K = the number of segments in the commutator.

izecoy Google

Armature Winding Schemes. 79

N = the magnetic flux per pole, meaning the total number
of magnetic lines that reach the armature from
one pole.

The current in any one conductor will obviously be equal
to Ca 4- c.
^ The general formula then is

E = » X Z X N x^-^-io" . . . (I)

whence

Z ^ ^^^^ X 10' ■ . . . . (2)
« X / X N ^ '

In ordinary parallel groupings (lap-wound drum-armatures
and ring-armatures) c = p,so that for these the formula {2) is

simplified down to

„ Ex 10* . ■.

Z = ^^ .... {2a)

Examples. — (i) In a parallel-wound armature of a 12-poIe
tramway generator of the English Electric Manufacturing Co,
(MP12— iioo— 100); E=S5o; «= 1-666; N = 25,647,000 ;
/= 12; c = 13; hence by formula (2) or {2a) Z = 1 248.

(2) In the series-parallel armature of the lo-pole tramway gene-
ratorof Kolben and Co., p. 216 (MP 10 — 250 — 125); E = 550 ;
« = 3 • 083 ; N = 1 2,1 10,000 ; p ■= \o; c= \; hence by formula (2)
Z = 874.

(3) In the series-parallel armature of the i2-pole tramway gene-
rator of the Oerlikon Co., p. 188 (M P 12 — 500—100) ; E = 550 ;
«=T 1-666; Z = 1326; p = 12 ; c = 6 ; hence by formula (i)
N = 12,445,000.

Radial Diagrams. — Figs. 25 and 26 are radial diagrams
in which the conductors of the armature are represented by
short radial lines, while the end-connectors are represented by
curves or zigzags, those at one end of the armature being
drawn within, those at the other end being drawn without the
periphery. With such diagrams it is easier to follow the
circuits and to distinguish the back and front pitches of
the winding. The arrows show the direction of the induced

iLCD, Google

Dynamo Design.

electromotive-forces. In Figs. 25 and 26 the armatures are
supposed to be rotating in a 4-poIe field.

Fig. 25 is a diagram of a lap-winding and Fig. 26 a wave-
winding. It will be seen that while the lap-winding gives four
circuits in parallel, the wave-winding gives but two circuits.
It is a series-winding and gives with the same number fA
conductors double the electromotive-force ; but, as the maxi-

FiG. as-— Lap-Winding 4-

mum conductance of any one conductor is alike for the two
machines, the series-winding, with only two circuits instead of
four, will only yield half the currents.

Field-Step. — It will be noted that whereas in ordinary ring-
windings and in lap-windings the winding at the completion
of each element comes back to a point close to that from
which it started, and therefore in the same polar region, the
wave-windings all step forward to the next polar region of

izecoy Google

Armature Winding Schemes. 81

the same name. There is no abstract reason why windings
should not be imagined in which the step so made from one
clement of the winding to the next should not be to the region
of a still more distant pole, Let m denote the number of
such complete pole-pitches o\Jer which the step is made.
Then, in general, we have Fot lalp-windings ot = o ; for wave-
windings ■m = i; "and for Hni-vwinding m =0 for parallel

grouping, or m = i for series grouping. The cases where
M( >. I are not practical.

Groups of Conductors. — Suppose an armature to have Z
conductors arrayed in simple " elements " (either lap or wave)
consisting each of two conductors joined together as a loop,
the commutator needed would have, therefore, K segments =
i Z. It is easy to see that, using the same commutator, one
might double the number of conductors (and double the

oy Google

Dynamo Design.

elect roinotive-force of the machine) by substituting for each
" element " one consisting of four conductors wound as a
double loop. Generalizing, we may say that if each "ele-
ment " consists of a group of g conductors, if the number of
such groups or elements be called G, then Z=^G=^K.
In the case of ring- windings, where the simplest element is i
turn, g may be any whole number, odd or even. For lap and
wave-windings g must be an even number.

It is possible to go further, and imagine a mixed wave
and lap-winding. For, beginning with a wave-winding, each
element of which is a mere open loop of 2 conductors such as
shown in Fig. 27, one can easily see that for it one might

substitute a group of, say, six conductors consisting of three
laps. Such groups are, indeed, frequently employed in prac-
tice, for generating high voltages, as for example in tram-car
motors, and in the high-voltage generator of Brown (Fig. 74),
since this arrangement lends itself readily to the winding of
coils upon formers in the shop.

Winding Formul.^.

Terms used in the Theory. — It is essential to understand
the terms and the sense in which they are used.

Any winding is said to be re-entrant which returns on
itself so as to form a closed coil. An armature-winding is
said to be singly re-entrant if it re-enters itself after simply
passing in regular order through all the coils arranged around
the armature core. Thus an ordinary Gramme ring, or a
simple lap-wound drum-armature (Fig. 25J, is singly re-

nGooglc

Armature Winding Schemes. 83

entrant There may, if used in a multipolar field, with
several sets of brushes at Its commutator, be various paths
through it ; but so far as re-entrancy is concerned it is singly
re-entrant. The symbol for a singly re-entrant winding is O,

An armature may be wound with two independent circuits
each of which is singly re-entrant. Fig. 29 shows a ring-
armature wound thus. These two windings might have been
furnished with two independent commutators, one at each end.
But instead, the number of commutator segments is doubled,
the two sets of bars being alternated or imbricated between
one another. The brushes must be made broad enough to
overlap at least 2^ bars of the commutator, so as to collect
from both windings simultaneously. In a two-pole field, with
two sets of broad brushes, this armature would give four paths ■
In parallel from brush to brush. Such a winding is described
as duplex. The odd numbers form one winding, the even
numbers another. Three independent windings with three
sets of commutator bars similarly imbricated would be called
a triplex winding.

An armature is said to be doubly re-entrant if its winding
only re-enters on itself after having made two passages around
the coils of the armature. This term is best elucidated by
the example of Fig. 30. This consists of a ring-winding in
17 groups. They are joined together in a way, precisely
akin to the duplex winding just described ; each coil being
Joined to the next but one, but not to the one immediately
next to it. But as the total number of sections is uneven, the
coils do not form two separate windings- If we begin with
the coil numbered i, we see it is joined to 3, 5, 7, 9, etc.
until we come to number 17, by which time it has com-
pleted one round of the periphery, but is not yet re-entrant,
for now it goes on to the coils 2, 4, 6, etc., to coil 16, from
which it finally re-enters the starting point. The symbol for
suchsi doubly re-entrant winding is, ij_^ • This winding will
also require broad brushes that bridge over more than two sec-
tions of the commutator at one time. Like the duplex winding
of Fig. 29, it doubles the number of paths from brush to brush.
In fact, it is electrically the equivalent of a duplex winding

G 3

Fig. 3a— Doubly Re-entrant Ring Winding.

, Google

Armature Winding Schemes. 85

save for the fact that it requires an odd number of coils. In
armatures of many turns this difference Is quite immaterial.
For example, an armature with 2cx> coils as a duplex winding,
and a doubly re-entrant armature with 201 coils, if revolved
at the same speed in the same field would only differ by \ of
I per cent, in their electromotive-force. Lap-windings may
also be made doubly re-entrant, see Fig. 32, p. 95. A trebly-
re-entrant winding might be made by choosing the number of
sections so as to become re-entrant only after a travel com-
pleting three rounds of the peripherj-. For example, a 20-coil
ring-winding joined according to the following scheme : —

I — 4 — 7 — 10 — 13 — 16 — 19 — 2 — 5 — 8 — II — 14 — 17 — 20 — 3
— 6 — 9 — 12 — 15 — 18 — I. The symbol for treble re-entrancy
is 'q^ , It is the electrical equivalent of a triplex-winding
made of three simplex singly re-entrant windings OOO;
and like the triplex-winding will require brushes broad enough
to cover 3^ adjacent commutator bars at least.

Armature windings are also described in terms of the
number of paths which they afford for the current to follow
from the negative brushes through the windings to the positive
brushes. This number, in closed coil armatures (which are
the only ones here dealt with) is always even. In simplex
parallel- wound armatures (whether ring or drum) the number
of such paths or circuits is always equal to the number of
poles : in duplex parallel-wound armatures to twice the
number, and so forth. In simplex series-wound armatures,
the number of such paths is always two : in duplex series-
wound armatures, four, irrespective of the number of poles.
It is common to refer to windings by the number of circuits
they present : thus, one speaks of a ten-circuit winding,
meaning one in which there are ten paths through the winding
from — to +. The current in one circuit will be equal to
the whole armature current divided by the number of circuits.
There are also methods of winding due to Arnold, which
result in a series-parallel arrangement. Thus it is possible to
have a 6-pole machine, with 4 paths through the armature.
This might be carried out as a doubly re-entrant wave-
winding. (See page 96,)

izecy Google

86 Dynamo Design.

The next term which requires definition is the pitch or
spacing of the winding. This term denotes the distance from
one element of the winding to the next similar element in the
succession ; and it is usual to express ^^ pitch of the winding
in terms of the number of conductors spanned over, or less
usually in terms of the number of elements of winding (loops,
or groups of conductors) passed over, or sometimes in terms
of the number of slots passed over. It is not usual to express
the pitch, either in actual peripheral length, or in terms of
angle subtended, or in terms of the pole-pitch. Suppose all
the conductors to be numbered consecutively around the
periphery of an armature, and that No. I is joined at the front
end to No. l6, thus forming a loop, and that No. i6 is joined
at the back end to No. 31, then the pitch at both ends is 15.
In wave-winding the pitch at both ends is positive, that is to
say the winding goes continually forward. In lap-winding the
pitch at the two ends is different. Thu,';, if at the front end
No. I is joined to No. 18, and if at the back of No. 18 the end
connexion laps back to No. 3, the front pitch is -}- 17, while
the back pitch is - 1 5. In that case the resultant pitch is 2,
and the average pitch is 8. We shall use the symbols y\ and
y^ for the front and back pitches re5pectively,_y for the total pitch
and^ for the average pitch. Since it is obvious that the simplest
element, whether of lap or wave-winding, is a loop of two con-
ductors united together, and since in every such loop one of the
two conductors ought to be passing a south pole at the time
when the other is passing a north pole, it follows that the width
across the loop ought to be approximately equal to the pole-
pitch. In fact the average winding-pitch must be in lap-
windings a little less, and in wave-windings a little less or a
little greater, than the pole-pitch. In lap-windings the larger
of the two pitches may equal the pole-pitch, ought not to
exceed it, but may {and with some advantage) be less ; while
the smaller of the two pitches should not be less than the
width of the pole-face.

Condition of Re-entrancy. — The condition that a winding
siiall return afterafinite number of symmetrically spaced steps
to the conductor from which it starts may be stated thus : —

oy Google

Armature Winding Schemes. 87

The resultant step from element to element, multiplied by the
number of conductors per element, and by the number of such
resultant steps, must equal the whole number of conductors
multiplied by some whole number. For example, let a lap-
winding consist of 80 simple loops, having a resultant pitch
= 2. This will be re-entrant if the whole number of such
conductors is 160. In the case of wave-windings, where the
loops go zig-zagging around the periphery, the number of
elements per round, multiplied by the number of rounds, and
by the number of conductors per element, will obviously give
the whole number of conductors so united. Now the pitch
must hz such as to be approximately equal to the pole-pitch,
but not exactly, otherwise the winding would become re-
entrant at the first round. Or conversely, the whole number
of conductors must be such that with a winding-pitch approxi-
mately equal to the pole-pitch, the winding shall become re-
entrant only after a number of rounds. Thus, for. example,
in an 8-pole machine, with single winding pitch 25> the total
number of conductors must not be 35 X 8 = 200, otherwise
the winding would re-enter after the first round of 4 loops.
It might be either 198 or 202, for then round after round
would be completed before re-entrancy was finally attained.
This example illustrates so well the essential principle of a
simplex wave-winding that it may be further considered.
Suppose we construct a winding table for this winding,
using 202 = Z. Take the even numbers of the conductors
as going down from front to back ; the even numbers being
the return conductors leading up from back to front.
Starting with No. I the connexions run as follows : —
I — 26 — 51 — 76 — lOi — 126— 151 — 176—201, the eighth step
thus completing the first round, and failing to re-enter by 2
places. The second round of eight steps similarly goes from
201 to 199, the third to 197, and so forth. The winding thus
recedes 2 places at each round. By the end of the twenty-fifth
round 200 steps will have been taken, and the winding will
have slipped back 50 conductors from No. I, and the 30Oth
step will therefore end on No. 153. There remain two steps
to be taken, viz. from 153 to 178, and from 178 to No. i, thus

oy Google

Dynamo Design.

Winding Tablx for 8-Pole Drum Armature; loi Conductors;
Srries Grouping ; Brdshes (±) 135° apart.

FIB FBiFjB'FiBF

lOI

126

176

201

124

149 ■ 174

f99

22 .

121

147

172

197

I20

'45

170

19s

118

'43

16S

'93

116

141

166

191

114

139

164

1S9

112

'37

162

187

IIO

'35

160

i8s

83 ,

io3

'33

'58

183

106

'3'

156

181

104

129

154

179

5»

102

127

"Sa

177

202

100

125

'50

17s

. 300

123

148

'73

.98

121

146

171

196

144

169

194

117

142

167

192

140

i6S

190

'38

163

188

III

136

161

186

109

134

159

18+

107

132

'57

182

'OS

130

'SS

180

'03

128

'53

'78

•

finally completing the re-entrancy. Fig. 31 illustrates the
first two rounds of this winding. It is assumed, but not shown
that the winding is in two layers ; the conductors in the upper
layer being those with odd numbers, those in the under layer
with the even numbering. On examining this table it will

oy Google

Armature Winding Schemes. 89

be seen that the conductor which is half-way through the
winding from No. I, is No. 102. These are not at opposite
ends of a diameter, but are f of a circumference apart. As a
matter of fact the brushes — two sets of which only are essen-
tial, as there are only two circuits through the winding — may
be either J, | or f of a circumference apart. In one sense this
winding has a 25-fold re-entrancy, seeing that at every 8 steps

Fig. 31.— Wave-Windino: 8-pole, z-drciut, singly

the periphery is perambulated. But, strictly, the v/inding is
only singly re-entrant Had there been 204 conductors, or
196 conductors, so that at each round of eight steps the
winding receded or advanced by four places (instead of two
places), the result would have been different, giving a doubly
re-entrant winding with four paths. The number of paths
through the armature corresponds to the number 2, 4, or 6, etc.,
by which the first round fails of re-entrancy, In the case

jijiiizecy Google

90 Dynamo Design.

illustrated by the winding table we see that the principle of
re-entrancy enunciated at the beginning of this paragraph is
fulfilled, for the resultant step of 30 multiplied by the number
which is 2, of conductors per element, multiplied by the
number of resultant steps, namely (25 x 4) + I, makes 6060,
which is an integral multiple of the whole number of con-
ductors 202.

Condition of each Conductor being encountered once. — It is
not enough that the winding should be re-entrant. It should
(in a simply re-entrant winding) be such that «//the conductors
should be encountered, and that each should be encountered
once only.

Case I. Lap-Windings. — Let_yi be the forward pitch and
_j'a the backward pitch, its actual value being negative. Then
y\ 4- y^i is the resultant step. If we confine ourselves to the
practical case that each element or section of the winding is a
simple loop, the number s of such sections will be equal to

number of resultant steps if multiplied by the length of each
resultant step) will equal the total travel of the winding.
This will be equal to Z if the winding is singly re-entrant
But if the lap has been such (for example if_yi is 25 and /a is
— 21, then/i -|-/a = 4) that re-entrancy is not effected without
travelling more than once round the periphery, then the total
travel will be equal to U Z, where U is the number of times
the periphery has been travelled round. This gives us as the
first condition that

Now U may be 1, 2, or any whole number ; hence it follows
thatji -V y^ must in every case be an even number. Further,
the condition that no conductor shall be encountered twice is
that for no number of steps whatever shall /i — y^, however

oy Google

Armature Winding Schemes. 91

often repeated, be equal to /,. Or taking m as any whole
number

whence

*^^=-. . . • (2)

yi l — m

It follows from this inequality that y^ and /a cannot possibly
have any common factor, and as their difference must be even,
it follow? that both of them must be odd numbers.

Case IT, Wave- Windings. — The resultant step for an

Z
element being/, +/s, and the number of such steps being -

and the total travel being U times round the periphery, we
have

2 /

f(ji +:>'.) = UZ;

yy^j..^. =

U. . . (3)

In the case of the winding table given above where j'l and
yt are each 25, the total travel is 25 times round the periphery,
. Now, in order that no conductor be encountered twice it is
clear that not by any number of repetitions of the step/i +_ys
shall it be possible to recur to the step_yi beyond any previous
number of the repetitions of the step /, + ^j. Or, if m and «
are any whole numbers it is clear that tn times /i +_j'i must
not equal « times y-^ + y2 steps plus y^. Or in symbols

■m (j, +>i) ^ « {yi +7a) +^1- . • (4)

It follows that in this case also yi and y^ cannot ha-ve any
common factor ; and as U may be any number, odd or even, it
follows from [3] that as their sum must be even, both of them
may be odd. They may be, however, equal to one another,
and this is the common case.

General Formults. — We are now ready to state the general
formulae for winding's. These maybe put either (1) in terms
of the number of segments K of the commutator, and of the

izecoy Google

92 Dynamo Design.

pitch of the winding y^ in terms of the number of commutator
segments over which the element of the winding spans, or (2)
in terms of the number of conductors Z and of the pitches jjii
and y^ as defined above.

These general formulae are as follows : —

If y stands for the complete step, not from conductor
to conductor, but from the first conductor of any group to the
first conductor of the next group, m for the field step, and G
for the total number of groups in the winding, we shall have

\py + c = mg(^ = mL . . (i)

when

2,^&±jc. ... (3)

y=^"^i^ ... (3)

which are the general formulae for symmetrical windings.
For lap-windings m = o, where it follows that

— 2c , y

y=+.-;. . . . (4)

y being dissected into 2 parts /, and j/^, of which y^ is negative,
each of which is either equal to or slightly less than Z//, and
which difier from one another by 2c -i-fi.

For wave-windings »? = I, so that the complete step
becomes

y-i^^. ... (5)

and if this is made up of two equal back and front pitches^
and 7a of equal value, we shall have

y=}"=^- ... (6)

In lap-windings the step of the winding at the commutator
is related to winding pitch by the simple rule : —

yi=y^g- ■ ■ ■ (7)

Thus in a simple lap-winding, where y = 2, and where each

izecoy Google

Armature Winding Schemes. 93

dement of the winding is a simple loop made of two conductors
so that ^ = 2, we haxey/, = i.

It would be easy to write out a number of special formulse
for special cases. Four each of lap and wave must sufhce,

Lap-Windings.

{L) Simplex Singly Re-entrant (Parallel) Lap- Witiding.

c= p\ m = o\ and, \{ g=2; Z = 2G = 2K.

y = ±2\

_ 2
v, < — and must be odd',

-yt '=yi + 2;
yk = I.

(«.) Duplex or Multiplex (Parallel) Lap- Winding, consisting
of X independent windings, each of which is a simplex singly
re-entrant lap-winding.

c = px; m =0 ; and if ^ = 2,Z = 2G = 2K.

y = ± 2x;

" _ Z

Vi < - - and must be o^ ;

,, ,:.::. "

n = ± *.

(Hi.) Stmp/ez(Senes Parallel) Doubly or Multiply Re-entrant
Lap-Winding. (See remark on p. 83 as to meaning of term).
^ = 4, 6, or other even number greater tllan 2 ; ot = i? ; and if

^=2, Z = 2G = 2K.

y~±';

— Z

y\ < - and must be odd;

y^ =yi - *:■,

yk= ±kc.

izecoy Google

94 Dynamo Design.

(iv.) Duplex or Multiplex {Series- Parallel) Doubly or Multi-
ply Re-entrant L^ Winding, consisting of x independent wind-
ings, each of which is a simplex doubly or multiply re-entrant
lap-winding. Here x = 2, 3, 4 or any whole number ; c^ = the
number of circuits through any one of the independent series
parallel windings (may be 4, 6, or other even number higher
than 2) ; m = 0; and if ^ =2, Z=2G = 2K.

y = ±^ei ;
= Z

and must be odd :

yi=y\ ~ ^ei ;

yi= ± i^(^i.

Figs. 32 and 33 afford examples of case ii. and case ui.
above. Fig. 32 is a duplex lap-winding in which p = 4,
Z = 32,_yi = +9 and/j = — S- There are two independent
circuits exactly as in the duplex ring-winding. Fig 29, p. 84.
Symbol OO- F'g- 33 corresponds to the doubly re-entrant
ring-winding. Fig. 30, p. 84. In it / = 4, Z = 34, _yi = -|- 9
and ^i = - 5. Symbol (q) . In both cases c = 8.

Wave Windings.

(('.) Simplex Singly Re-entrant (Series) Wave- Winding.

c = 2; m = I ; and if ^ = 2, Z = 2 G = 2 K.

y is the average of /, and y^,

Z= py ±2;

_ _ Z ^ 2 and must be odd, and must not have

Vi — yt -^ ^^y common factor with Z.

2Kq: 2

(ii.) Duplex or Multiplex (Series) Wave- Winding, consist-
ing oix independent singly reentrant simplex wave windings.

3,3,l,zec:,yGqOglc

Fig. 33.— Simplex Doubly Re-entrant Lap- Win ding.

,Goo!;lc

96 Dynamo Design.

JT = 2, 3, 4, or any whole number ; the number of circuits Ci,
in any one of the simplex windings = 2 ; w = I ; and if^= 2
then Z = 2 G = 2 K. Number of circuits in parallel = Ci x,

Z = xpy ±2x;

V = V — ^ -^ ^^ ■ I" those cases where j is
' xp ' even, ,^1 may =y + l>

and_ya = y — l.
zYiT 2X

{Hi.) Simplex Doubly or Multiply Re-entrant (series-parallel)
Wave-Winding; also called Arnold's winding.
c = 4, 6, or other even number higher than 2 ; m = i; and, if
^=2, Z = 2G= 2K.

Z=pj'±c;

(m) Duplex or Multiplex Doubly or Multiply Re-entrant
(Series-Parallel) Wave- Winding, consisting of x independent
■ wave-windings each of which is doubly or multiply re-entrant
Ci = 2, 4,6, or other even number ; number of circuits in parallel
= c,jjr; m = 1 ; if^ = 2, then Z = 2 G = 2 K.
Z = xpj> ±eix;
Z Te,x

The circumstance that if in a wave-winding _}' and Z have
any common factor there will be a corresponding number of
independent windings, leads to some curious results. Further,
the circumstance that if in any wave-winding the number of
circuits is made equal to the number of poles, leads to the
result that in this case the wave-winding becomes identical to

oy Google

Armature Winding Schemes. 97

a simple lap-winding or ring-winding. In the case of bipolar
machines wave and lap-windings are identical, the only
difference being the question whether _yi = y^ or not. A series
grouping cannot be effected by a lap-winding : it may be
effected by a wave-winding or by a mixture of wave and lap-
winding. In the case of 4-pole, 8-pole and 12-pole machines,
a simplex series winding cannot be made with 4 conductors
per segment of the commutator. Nor. in the caseof 6-pole
and l2-pole machines can a simplex series winding be made
with 6 conductors per segment. In general, for a machine
with «-poles or 2 « or 3 «-poles, it is impossible to make a two-
circuit winding having n conductors per segment of the
commutator. Or, stated another way, to make a two-circuit
wave-winding, the number of conductors must not be a
multiple of the number of poles.

The number of circuits made by any winding can be
calculated by the following formulcE, derived from those
previously given.

Lap- Windings.

'"ify- ...(•)

Wave- Windings.

c = Z~fiyi . . ((9)

if ^, =^2. If not, then take instead of _y, the average pitch.
Ring- Windings {parallel).

c=py. ... (7)

Examples -.—(i.) p =(> ; Z = 374; K=i87; .v, = 47, ^^ =
— 45 ; f = 6. Here/ = 2, the winding being a simple lap-winding.

{it.) p= i>; Z = 434; ji = 73, /a = 71. This is a wave-
winding with average pitch of 72. Hence by formula {j3), there
will be two circuits only, the winding being singly re-entrant.
Symbol O.

{Hi.) / = 6 ; Z = 442 ; ^1 = 71 ; Ji = - 67. This is a lap-
winding, with y =yi -|-^a = 71 — 67 = 4. Hence by formula (a)
there wiU be 12 circuits, the winding being duplex singly re-entrant
Symbol O O.

{iv.) p = 8; Z = S72 ; ^1 = 71, y^, = 71. This is a wave-
winding, giving by formula (,8) 4 circuits, the winding being doubly
re-entrant. Symbol (q) .

oy Google

Dynamo Design.

Symbol O
Simplex, Singly re-entrant.

iLCD, Google

Armature IVinding Schemes.

Six-Circuit. Simplex, Trebly re-entrant.

Wmdmg

L,j,l,„,;,„ Google

loo , Dynamo Design.'. ■ '. .

(i/,) / = 4 ; Z = 246 ; yi = 61, y^ = 59. This is a wave-
winding with average pitch 60, giving 6 circuits, with a trebly
re-entrant winding. Symbol (^go} •

(vi.) p = a,; Z = 438 ; y\ = y^ = iii. This is a wave-winding,
but as Z andj'i contain 3 as a common factor there will be 3 inde-
pendent wave windings, each singly re-entrant; and there will be
d circuits. Symbol O O O.

Figs. 34 to 39 give a set of examples of wave-windings to
elucidate the rules.

Fig. 34 is a 6-pole, two-circuit winding (sometimes called
"multipolar series"), with 32 conductors. The winding is
singly re-entrant. The winding pitch _>>, =_j', = 5. Hence by
the rule c = Z — py, there will be two circuits. Below the
figure is shown the equivalent ring, having the 32 conductors
rearranged in the order of their occurrence (see Arnold's
"reduced scheme," p. 112), the two circuits implying a two-
pole field. The advantage of this mode of representation is that
in the equivalent ring the windings do not overlap one another.
Fig. 35 depicts a 4-pole, six-circuit winding, with 34 con-
ductors and an average winding-pitch of 10. On examina-
tion it will be seen that the winding, though simplex, is trebly
re-entrant (symbol (g^ ) , making 6 circuits though there
are only 4 brushes, in correspondence with the 4 poles. The
equivalent ring, as shown, will be a 6-pole ring with 6
brushes.

Fig, 36 is a 6-pole, four-circuit winding, having 32 con-
ductors with an average winding pitch of 6. This produces
a duplex-winding ; there being two independent windings each
singly re-entrant. Hence there are four circuits (symbol O O).
The equivalent ring will, of course, have a 4-pole field and 4
brushes.

Fig. 37 is a 4-pole, eight-circuit winding, having 32 con-
ductors, with an average winding pitch of 10. It results in a
duplex, doubly re-entrant winding (symbol (q) (q) ).
The equivalent ring has S poles and 8 brushes.

Fig. 38 is a 4-pole, six-circuit winding, with 30 conductors
and a winding-pitch of 9. As there is the common factor 3

izecoy Google

L,j,i,„,;,„ Google

Four-Circuit. Duplex,Singly re-entrant

E.&Elf.SponL'.^Lonaon.

.oogic

Eight-Circuit. Dupux, Doubly re-entrant.

the
livalent

D, Google

„, Google

Symbol OOO

Six-Circuit. Triplex,Sinoly re-entbamt

Phe
liraleat

Rm^Windnig

E.&Elt.SpoiiL'' London.

„ -. „Gooi;lc

^^^^^^ ^^P^^ Symbol (S) Caj)

Twelve-Circuit. Duplex,Trebly re-entrant

ivalent

D, Google

Armature Winding Schemes. loi

between' J Z andj/, there will be three independent windings,
coloured respectively red, green and black, and each is singly
re-entrant, so that there are 6 circuits (symbol O 0). This
triplex jvindingahould be compared with Fig. 35, which also
results in & six-circuit winding. The equivalent ring has, of
course,;als6 three independent windings.

Fig. 39 is a"6-pole, twelve-circuit winding, with 36 con-
ductors, and. an average winding pitch of-8, resulting in a
duplex trebly re-entrant winding (symbol (flj^ (flo) )■ ^^
requires but ^ broad brushes, though, as is obvious from the
equivalent ring diagram; it has 12 circuits in parallel one with
another.

It will be noted that Figs. 34 and 36 depict two cases in
each of which there are 6 poles and 32 conductors. They
differ, however, in the winding-pitch, with the result that one

Fig. 40.

is. a two-circuit and the other a four-circuit winding. The
latter would yield double the current at half the voltage.

" Further 'Examples of Drutn- Windings.— rKn example of a
lap-winding is afforded by the 6-pole Scott and Mountain
generator, page 160, in which Z = 496, K = 248, g= 2,
c = 6,_j'i = 41, /a = -39,J' = 2, the conductors lying in 124
slots, 4 conductors per slot As there are 6 poles there are 20§
slots per pole. The coils are grouped to span over 20 slots,
conductor No. I (upper) in No. I slot being united to con-
ductor No. 2 (lower) in No. 21 slot; and No. 2 in No. 2i slot
is connected back to No. 3 (upper) in No. i slot, which in turn
is joined to No. 4 (lov/er) in No. 21 slot, as shown in Fig, 40.
This is, then returned to No. i (upper) in No. 2 slot, and so
forth. If the conductors were numbered consecutively, begin-
ning with No. I, in No. i slot, those in No, 21 slot would

itizecy Google

102

Dynamo Design.

become Nos. 41, 42, 43 and 44. Hence if No. i !s joined at
^e front end to No. 42, and No. 42 at its back end laps back
to No. 3, the pitches are respectively y^ = 41,^1 = - 39, and
■y = 2. This is therefore a simplex singly re-entrant winding.
As an example of a wave-winding we may take the lO-pole
generator of Kolben (page 216 and Plate VI.)- This has 874
conductors lying in 437 slots, i.e. two conductors per slot.
Now 874 = 87 X 10 + 4. Hence if ^ = &•;, it follows that
c = 4, and the winding will be doubly re-entrant, or is a series-
parallel wmding. The winding table may be cOnsthicted as
follows, beginning with conductor No. i : —

1st round . . .

175

363

349

4,6

5*3

610

697

784

871

2iu]iauitd . . .

871

171

358

345

432

519

604

693

780

867

ardroond . . .

867

167

aS4

341

438

S>5

600

689

776

863

44thionnd . . .

703

790

177

a64

35'

438

525

613

699

45th lonnd . . .

699

786 ■ 873

'73

360

347

434

Sai

608

695

87lh round . . .

S3I

5i8|7os

792

179

266

353

440

527

8Eth round . . .

527

614 1 701

788

The winding becomes re-entrant after 87 rounds plus 4
steps. It had become all but re-entrant by returning to No. 3
(instead of No. l) after 44 rounds plus 2 steps.

Mixed Wave and Lap- Winding. — None of the foregoing
formultC take any account of certain symmetrical windings in
practical use which are mixed. Fig. 42 is a simple example
of such a winding, essentially a wave-winding, of which each
element consists of 4]loops in series. Windings of this general
character lend themselves to small or medium sized armatures
with former-wound coils, as those of tramway motors, or for
special high-voltage construction. An example is to be found
in the high-voltage 4-pole dynamo of Messrs. Brown, Boveri
& Co., page 205. This is a winding of great interest. There
are 59 slots, receiving 59 former-wound groups of coils. Each
group is made up of three separate "sections," so that the
number of " sections " is 177, and there are 177 segments to

oy Google

Armature Winding Schemes.

103

the commutator. These sections are connected up as a wave-
winding. But each section itself consists of 4 loops or turns.
There are therefore 24 wires througli each slot, making 1316
conductors in all. They may be regarded as 177 "sections,"
each consisting of 8 conductors united together ; or, for
purposes of calculation we may regard the whole thing as a
wave-winding of 354 conductors, and then substitute 8 con-
ductors for the 2 in each loop. Now to make a singly re-
entrant wave-winding of 3S4 conductors we must have an

average pitch J> such that /J* ± 2 = 354 ; whence j; = 8(

As a matter of fact the average pitch chosen is %% and the

two actual pitches aren't = fij.y^ = gi, the winding table being

as follows : —

Firal round j — q2 — 179 270 3

-94-^81—272 — 5
31 87 91 87

izecoy Google

I04 Dynamo Design.

and so forth.*- Butthesteps of pitchgi are all of them laps of
4 turns, while the steps of pitch ^"j are mere connexions down
to the commutator and then on to the next set of 4 turns in
the succession, as in the following scheme : —

(Upper

2 91

^272

3 ■ 92

/ \/

Commutator Bar I 9O

Now as there are 354 "groups" in 59 slots this is 6
" groups " per slot, three " upper " of odd number and three
" lower " of even number. As there are 6 in a slot we may take
No. I slot as containing groups i to 6, No. 2 slot groups 7 to
12, and so forth, so that No. 16 slot will contain 91 to 96.
Then group 92 will be the first lower group in No. 16 slots,

Fig. 42.

and the slot-pitch for the former-wound coils will be from
No. I slot to No. 16 slot, or the slot-pitch for the coils spans
over 15 teeth.

Number of Brusk-sets. — The number of places on the
commutator at which it is necessary or advisable to place a
set of collecting brushes can be ascertained from the winding
diagrams. All that is necessary is to draw arrows marking
the directions of the induced electromotive-forces. This has
been done, for example, in the radial diagrams Figs. 34 to 39.
Wherever two arrow-heads meet at any segment of the
commutator there a positive brush is to be placed : and at
every point from which two arrows start in opposed direc-
tions along the winding, there is the place for a negative
brush.

izecoy Google

Armature Winding Schemes. 105

For all lap-windings, and for ordinary parallel ring-wind-
ings, there will be as many brush-sets as poles, and they will,
be situated symmetrically around the commutator in regular
alternation, + and — ,'at angular distances. apart equal to the
pole-pitch. It must be remembered that the number of brush-
sets does not necessarily show the number of circuits through
the armature. Take the case of a 4-pote machine with four
sets of brushes at 90° apart from one another. If the winding,
is a simplex, singly re-entrant lap-winding, there will be 4
circuits. But if the winding is a duplex,.ora doubly re-entrant
lap-winding there will be 8 paths. If a triplex singly re-,
entrant lap-winding there will be 12 circuits.

For wave-windings, whether series or series-parallel, and
for series ring-windings, if the arrow-heads are similarly drawn
it will be found that there are required but two brush-sets,
whatever the number of poles ; and the angle between the -f-
set and the — set will be the same as the angle between any
N-pole and any S-pole. Thus for a lo-pole machine with .
wave-wound armature, the brush-sets may be 36° apart, or
they may be 3 x 36° = 108°, or 5 x 36° = 180° apart. But
it must ^ain be remembered that though there may be only
two brush-sets, the number of circuits through the winding is
not necessarily 2. For if the winding is duplex or if it is
doubly re-entrant, the number of circuits will be 4. If both
duplex and doubly re-entrant 8.

There are, however, some further considerations that
deserve attention.

Reduction in number of Brusk-sets. — Cases occur when it
may be desirable, with a parallel-winding (for which the
number of brush-sets would naturally be equal to the number
of poles), to reduce the number of brush-sets. In the case of
4-poIe tramway motors mere convenience of access dictates
the reduction of the number of brush-sets to 2. Now, if a
wave-winding is adopted the number will naturally be 2, not
4, If a parallel winding is adopted the number 4 may be
reduced to 2 by the application ofMordey's device of cross-
connecting the segments of the commutator. Let us, how-
ever, consider what is the result, without resorting to either

oy Google

io6

Dynamo Design.

of these expedients, of simply using, with a parallel-wound
4-pole armature, 2 brushes instead of 4. Suppose the ma-
chines to be generating 120 amperes; then if 4 brushes are
used there will be 4 circuits, each carrying 30 amperes, and at
each "brush " the current will be 60 amperes (Fig. 43). If now
2 of the brushes are removed, and the dynamo still generates
120 amperes, the current through each of the two remaining
brushes will be 120 amperes; while internally there will be
only 2 circuits. But these will not take equal shares of the
current since, though the sum of the electromotive-force in
each circuit is the same, the resistance of one is three times

FIG. 43.

that of the other. So the currents will be about 90 amperes
in one circuit, and about 30 amperes in the other as in Fig. 44-
Assuming that no spark-difhculties occur in collecting I20
amperes at either brush the arrangement will work perfectly.
But the heat losses will be greater than before. For, if the
resistance of one-quarter of the winding be taken as 0*05 ohm,
the heat loss will be : —

fVith 4 brusha 4 X 30 X 30 X 0*05 = 180 watts.

With -2, brushes f J 5^ ^ J! ?? JS °'°M = 270 watts.
I 3 X 30 X 30 X 005 f '

It is not an uncommon thing in the case of 6-pole

oy Google

Armature Winding Schemes.

107

slow-speed exciter machines to see only 4 brush sets instead
of 6.

Increase in number of Brusk-sets. — In cases where wave-
windings are used, requiring, as we have seen, only two brush-
sets, it is often advisable to use more sets than two. This is
particularly the case where the current to be collected is
several hundred amperes. In fact, though in one sense only
two sets are required, and these situated at an angular distance
apart equal to the angular distance from one N-pole to any
S-pole, there is no harm done if as many sets are employed
as there are poles. Consider a singly re-entrant simplex wave-
winding for an 8-pole machine such as Fig. 31, Whenever

any brush bridges across between two adjacent bars of the
commutator it short-circuits one " round " of the wave-winding,
and this "round" is connected at three intermediate points to
other bars of the commutator. So, if the short-circuiting brush
is a -H brush, no harm will be done by three other -J- brushes
touching at the other points. If these other brushes are
broad enough to bridge across two commutator bars, then
they may have the effect that commutation may go on at
them also, three " rounds " instead of one undergoing com-
mutation together. Or, what amounts to the same thing, the
duration of act of commutation for any one " round " will be
prolonged, much as it would be if for the one brush there
were simply substituted one of greater breadth. Certain it is

izecy Google

lo8 Dynamo Design.

that the commutation is in general imprcwed by using more
brush-sets than two. Many makers of multipolar machines
with wave-windingS( habitually use the full number all' round
the commutator. As examples, see the Kolben lopole
machine, p. 216, and the Oerlikon 12-pole machine, p. 188.

Choice of Number of Circuits. — From the considerations
already discussed it will be seen that it is possible to have
windings that give any desired (even) number of circuits in
machines having any number of poles. It was not knoivn
until recent years that this could be so ; that, for example, one
might have a 6-pole machine with 4 circuits, or an 8-pole
machine with 6 circuits. A few considerations on the choice
of alternatives may be desirable. In large multipolar genera-
tors it is as a rule inadvisable to have more than lOO or 1 50
amperes in any one circuit [Special machines for electro-
chemical work form exceptions.] Suppose then it were
desired to design a 6-pole machine to give an output of 400
amperes. If designed with 2 circuits as a singly re-entrant
wave-winding, there would be 200 amperes per circuit. If
with a duplex singly re-entrant wave-winding, or a simplex
doubly re-entrant wave-winding, there would be 4 circuits each
carrying 100 amperes. If with a triplex singly re-entrant
wave-winding, or with a parallel lap-winding, 6 circuits with
66 ■ 6 amperes each. In each case except the last there might
be only 2 brush-sets ; but in each case 6 brush-sets would be
preferable. From this last point of view there is nothing to
choose. But the 2-circuit winding is too thick, and the 6*
circuit. winding involves an unnecessarily great number of con-
ductors and connexions. The 4-circuit winding is distinctly
preferable. Again, suppose a 12-pole slow-speed machine
were desired for a high voltage and to give out 300 amperes.
A parallel winding with 12 circuits each carrying only 25
amperes would be absurd : a 2-circuit winding would be dis-
tinctly preferable.

Thus it will be seen that wave-windings, with their many
possibilities of different groupings in series, series-parallel, etc.,
offer distinct advantages over lap-windings, and they possess
the further incidental advantages of equalizing any inequality

izecoy Google

Armature Winding Schemes. fo9

in the magnetic fields of the various poles, and, in general, of
requiring fewer conductors and end-connexions than lap-
windings do, Arnold has given the following very striking
e:xaniple of the adaptations of wave-winding:^.

Taking the formula (p. 96) for series parallel \grouping

yi^ _ 2 K ± 1 ^ ^ ^jjj applying it to the case of a 6-pole machine
: P

with 290 conductors and 145 commutator bars, we may have,
with one and the same size of core-disk, and the same size of
conductors, the following cases : —

With the same core-disk, a doubly re-entrant lap-winding
would give : —

A disadvantage of series-groupings is, that in general they
require an odd number of slots, making construction of the
disks in segments a not too easy matter, unless the number
of slots is divisible by 9, 15, or 21. Some makers find com-
mutation less satisfactory in these machines than in those
with parallel grouping.

Equalizing Connexions. — It was noted above, that if
for any reason the poles are of unequal strength, parallel-
windings, whether lap-wound or ring-wound, work unequally,
the current no longer dividing itself equally between the
various circuits that are in parallel. Asa result the heating
is no longer a minimum. To mitigate this evil it is now
customary to provide parallel-wound armatures with equalizing
connexions, which are cross-connexions between those parts of
the winding which are, or ought to be, at equal potentials.

As a matter of history such cross-connexions were intro-
duced many years ago for other reasons.

Such cross-connexions will obviously have the tendency to

izecy Google

no Dynamo Design.

equalize the amounts of current collected at the various sets of
brushes. In multipolar machines, any two or more points in
the winding that are during the rotation at nearly equal
potentials maybe connected together. If there were perfect
symnietry in the field system no currents would flow along
such connectors ; but, owing to imperfect symmetry the
induction in the various sections of the winding may be

Fig. 45.— Armature with Equalizing Rings.

unequal and the currents not equally distributed. Thus in a
lo-pole machine with parallel winding, suppose two of the poles
to be badly excited owing to some defect of the exciting
bobbins, then the sections of the armature winding as they pass
those poles will not generate the full electromotive force, and
at this instant there will be an abnormal amount of current
drawn from the other sections, tending to set up sparking. If
now there are chosen 5 equidistant points on the winding and

izecoy Google

Armature Winding Schemes.

these are joined together by a connexion of low resistance, by
being united to a copper ring, this adjunct will, at those instants
when these five points are near the commutation-points, tend
to equalize the distribution of current. But to be effective
several such equalizing rings are needed, each independent of
the other, and eacli connected down to the winding at points
spaced out at distances apart equal to twice the pole-pitch.

As an example, suppose a lopole machine having Z = 480,
with a parallel lap-winding, and that we decide to have S
equalizing rings. As there are g6 conductors within the
double-pole pitch any conductor (No. 1 for example) will be
joined to the 96th, beyond it, and so on around the first ring.
As there are to be 8 rings, if the first ring isjoined to conductor
No, I, the next ring must be joined to the conductor that is
the eighth part of the distance along the winding from No. i
towards No. 97, that is to a conductor 12 places further on,
namely No. 13, and so forth. Then the connexions to the rings
may be tabulated like a winding-table as follows : —

First ring ....

193

289

385

Second ling . .

109

ao5

301

397

Thiidring . .

121

si;

3«3

409

Foorih ring . ,

133

229

3*S

421

Fifth ring . .

i4i

337

433

Sixth ring . .

157

253

349

445

Seventh ring

169

26s

36'

457

Eighth ring. .

iSi

277

373

469

It will be obvious that it is expedient for perfect symmetry
that in designing an armature to be furnished with equalizing
rings, Z should be chosen such that the number of rings and
the number of poles are both of them factors of Z. In the
case of double-current machines (see as example, Fig. 73) with
connexions to yield three-phase currents, such three-phase
connectors serve as equalizers even though no three-phase
current is being drawn from the armature. Fig. 45 shows
such equalizing rings in an armature of the Thomson-

oy Google

tlrz Dynamo Design. ^

Houston Co. They are sometimes arranged at the back of
■the armature (see the machine of the English Electric Company,
■Fig. 82), or sometimes inside the commutator, or at the
back of it {see the Scott and Mountain machine, Fig. 69). Id
some cases they are placed over the armature periphery like
binding wires. The theory of equalizing connexions has been
treated very fully by Arnold.? In order to discuss . the
application of equalizing connexions to wave-windings he has
suggested an ingenious " reduced scheme " or diagram in
which he takes the various numbered sections of a wave-
winding and rearranges them like a two-pole ring winding
having equivalent properties. This is best understood by
comparing Figs. 34 to 39, each of which represents a wave-
winding together with the "reduced diagram" of the same
winding. When such diagrams are made for 4-circuit or
6-circuit windings, it at once becomes obvious which coils are
or ought to be equipotential, and the points to be joined by
equalizing connexions can be seen. Arnold has patented
equalizing connexions in wave-windings. One difficulty,
namely that wave-windings require odd numbers of slots,
giving rise to unbalanced groupings that are unsymmetrical,
Arnold purposes to obviate by interpolating a single lap in
the wave-winding in any section which has one element
too few.

To remedy inequality of poles in series windings,
Mr. B. G. Lamme, of Pittsburg, has devised * the method of
laying in the same slots a separate closed winding of low
resistance connected down at symmetrica] points, like the
ordinary equalizing connexions, to two or more insulated
rings. This virtually makes a parallel connected two (or
more) phase closed winding, in which, unless the inductive
actions are unequal, there will be no currents ; but in which if
the inductive actions of the individual poles are unequal,
balancing currents will be induced.

' Eleklrotcch. Zcitsihr. xxiii. i\^-23a md 233-235, 1902 ; and in his work
Gltkkstrom-Masciinai (1902).
' U.S.P. No. 646,092.

izecoy Google

CHAPTER VI.

ESTIMATION OF LOSSES, HEATING, AND PRESSURE -DROP.

In this chapter we propose to consider these questions from
the designer's point of view, as they are leading features of any
design and require to be accurately predetermined from the
drawings. On account of the diminishing importance of
bipolar machines and of those with smooth core armatures,
we shall consider, both in this chapter and the next (which
deals with the design of continuous-current machines), the case
more particularly of multipolar machines and machines with
slotted armatures.

The losses occurring in any dynamo or motor come under
six heads, as follows : —

A. Copper Losses. — These consist of the sum of the C*R
losses in armature and series coils (if any) and increase with
the load, but are independent of the speed.

B. Iron Losses. — These are made up of the eddy-current
and hysteresis losses produced in the armature core-plates
owing to the changes of flux-density to which they are sub-
jected in each revolution. They vary slightly with the load,
and are always variable with the speed. There are also certain
losses in the case of machines with toothed armatures due to
the production of eddy-currents in the pole-pieces.

C. Excitation Losses, that is, the watts expended in heat,
in driving the magnetizing current around the magnetizing
coils ; which losses must be debited against the dynamo, as
they lessen the efficiency.

D. Commutator Losses. — These consist o^

(1) C'R loss on account of contact resistance.

(2) Brush friction loss.

oy Google

114 Dynamo Design.

(3) Losses through sparking, and through eddy-
currents in the commutator bars.
Of these, Nos. (i) and {2) are as a rule the only ones
necessary to consider. There are also local circuits in the
brushes producing a small loss of energy.

E. Friction and Windage Losses. — The former is the loss
due to friction of bearings, which depends only upon the
load. The tatter is the loss occasioned by the armature
churning the air. It is independent of the load but varying
with speed.

F. Secondary Copper Losses. — We will consider these
separately.

(A.) Copper Losses.

Let w^ represent the total copper loss of the machine.
„ r^ „ (hot) resistance of the armature.

„ r^ ,. „ „ „ series coils.

„ C^ „ full-load armature current.

„ / „ total length of armature conductor

injeel, including end connectors.
„ s „ section of the armature conductor in

sq. inches.
Then wc have for the total resistance of the conductors on
the armature, considered as all in series irrespective of their
groupings

0*000008 X \ I +0-034

<'--)K]

at a temperature olfJZ. This formula becomes at tempera-
tures of about 60° C. (compare p. 42),

r = 9'5 ^ ^.
10* X s'

The actual resistance proper of the armature r depends
on the form of winding'employed, and the number of circuits
in parallel from brush to brush, the rules for which are given
on p. ^j. For all] bipolar machines and simple series-con-
nected . multipolar armatures, with two circuits only, the rule
becomes ra = r -i- 4.

izecoy Google

Losses, Heating, and Pressure-Drop. 115

For simple multipolar parallel-connected armatures run-
ning in fields of/ poles, wc have

r, = r^p\
because there are as many circuits as poles.
Then the copper loss of the machine is

«'. = (C^ X O + (C^„ X rj.

(B.) Iron Losses. — For the calculation of the hysteresis
loss, we can either make usi of the formula given on p. 9,
or, better still, refer to a curve obtained by test upon the
actual iron, Siich a curve is shown in Fig. 2, the ordinates
giving directly the watts lost per pound of iron at the
different flux-densities given by the abscissas, and at 30
periods per second. To find then the hysteresis loss in a
slotted armature, for instance, we proceed as follows. First,
calculate the number of complete m^[nctic reversals, thus

where / is the frequency, / the number of field-poles.
Next calculate the actual flux-densities B^ and B5 in teeth
and armature core respectively. A reference to the upper
curve of Fig. 2 will give the corresponding number of watts
lost per pound of iron at these flux-densities. Multiplying
the two numbers so obtained by the total weight of the iron
in teeth and core, and adding the two results we obtain the-
hysteresis loss in the armature at/ = 30 periods per second-
If the frequency of reversal is either higher or lower, than
this, the hysteresis loss will be proportionately greater or less.
Instead of computing these losses from the weight of the iron
we may compute them from the volume (cubic inches) by the
curves given in Fig, 3 on p, 13, or estimate it from Table III.,
p. 12.

The eddy-current losses are proportional to the square of
the flux-densityj to the square of the frequency, and to the

itizecy Google

1 16 Dynamo Design.

square of the thickness of the armature plate. They may be
calculated from the formula (see p. I2),

Watts lost per cub. inch = (40'64 X /^ X /" X B') lO"",

where t represents the thickness of plate in inches. As, how-
ever, this formula takes no account of the short-circuiting of
individual plates caused by the machining of the armature,
the results given by the formula will always be found to be too
small. The error is partly compensated for by the fact that
the formula is based upon the specific resistance of iron at
o° C, and as the armature will always be fairly hot, the
increase of resistance will diminish the eddy-current loss.
But a reference to the lower curve of Fig. 2, p. 10, will usually
give good results, and it is easier to apply as it gives directly
the eddy-current loss per pound at different flux-densities for
the standard thickness of English armature plate, viz. 25 rails,
and for 30 periods per second. It contains a correction-factor
to cover the loss due to the after-tooling of the core, but should
the frequency (or thickness of plate) be greater or' less than
the one for which the curve was plotted, the flnal result must
be raised or lowered in proportion to the square of the
frequency or plate-thickness. Table IV. on p. 14 and the
■curves of Fig. 3 on p. 15 are also useful.

We have assumed above that the volume of active iron is
the same as the actual volume of iron in the armature. This,
is of course not strictly true, as some of the teeth and perhaps
a small portion of the core may be missed by the flux going
from pole to pole. The error is, however, negligible, and is on
the right side. It tends to correct for the eddy-current loss
in the pole-faces, which is impossible to calculate. The iron
ioss of the armature is hence

■where If and ai^ are the, eddy-current and hysteresis losses
respectively, evaluated separately as above.

(C.) Excitation Losses. — If r^ is the resistance (hot) of the
shunt winding, calculated by means of one of the resistance
formulse already given, and V the electromotive-force at its

::,y Google

Losses, Heating, and Pressure-Drop.

(V_^v).X.'as.he.

terminals at full load, we have

actually used in excitation. To these must be added the loss
in the shunt regulating resistance, if any, giving a total loss of
Wx watts at full load. The watts required for excitation
purposes by shunt machines vary in practice from one to ten
per cent of the output, according to the size of machine. As
a guide to the designer in this direction, the table below may-
be useful.

Ouipu. of machine
in kilowatts.

EiciB.|[oi>l»>inpcranc.
of full had output.

>°

3-5

100

2-75

200

2'S

300

a as

5«>

2-0

aooo

1-75
IS

(D.) Commutator Losses. — The contact resistance between
commutator and brushes depends mainly upon (i) the ma-
terial of the brush ; (2) the bearing pressure ; {3) the peripheral
speed of the commutator ; and (4) the current-density in the
brush. Other causes, such as the condition of commutator
and brushes, weight and spring of the brush-holders, etc., will
also influence the contact resistance to a greater or less
extent

Carbon brushes are worked in practice at current-densities
of about 40 amperes per square inch for machines of medium
and large output. For small machines this figure may be
considerably exceeded, but the maximum permissible figure
is 80 amperes per square inch. Copper brushes are generally
worked at 150 to 200 amperes per square inch, and sometimes

::,y Google

Ii8 Dynamo Design.

even higher for small maclimcs. The bearing pressures found
in actual machines are \'2% to I'S lb. per square inch for
Copper brushes, and i • 5 to 2 lb. per square inch for carbon
brushes, this figure being exceeded for tramway motors on
account of vibration. Peripheral commutator speeds vary
from 1500 to 2500 feet per minute according to size of
tnachine. The latter figure is occasionally exceeded with large
tramway generators, Hobart uses sometimes 3060 feet per
minute.

Recent tests made by Professor Arnold ' upon commutator
losses show that both for carbon and copper brushes, the
contact resistance decreases rapidly with increasing current
density relatively to the peripheral speeds, it being more
marked in the case of carbon. According to his experiments
the contact resistance of carbon brushes for current-densities
of 50 to 30 amperes per square inch and peripheral speeds of
1200 to 2400 feet per minute may be taken as being 0'023 to
0-039 ohm per square inch of contact. For copper brushes
the corresponding values maybe taken as being 0"00077 to
0"C023 ohm per square inch. So that we can safely assume
as outside values of the contact resistance per square inch
For carbon brushes . . 0*04 ohm.

„ copper „ . . 0'003 ohm.

These values enabling us to easily calculate the commutator
loss brought about by contact resistance for any machine.

Example. — In a large electro-metallurgical dynamo by Brown
with an output of 4000 amperes, there are about 160 square inches
of brush contact surface, or 80 square inches for entry ahd 80 for
exit of the current, collecting about 50 amperes per square inch.
Assuming the contact resistance at 0-02 ohm per square inch we find
the whole C^R loss will be

2 X 4000 X 4000 X -^ — ■ = 8000 watts.

The contact resistance is not constant, but varies approxi-
mately inversely as the current density ; thus, with 1*5 lb.
■ %«t EI&trotechms€ke Ztitschrift, No. i, 1899.

izecoy Google

Losse:, Heatings and Pressure-Drop 119

per square inch pressure the resistance is about 0*04 when
the current density is 20 amperes to the square inch, and
goes down to about 0'02 when the density is 40 amperes to
the square inch. Hence it follows, that the drop of potential
due to this contact resistance is nearly constant at all loads,
and may be taken at from o-8 to i"0 volt at each side,
positive and negative of the commutator, or from i'6 to
2 volts on the whole machine.

The loss arising through the friction of the brushes against
the rotating commutator depends upon the bearing pressure
of the brushes, the peripheral speed of the commutator, and
the coefficient of friction between the two. If brushes and
commutator are in good condition this latter may be taken as

For carbon brushes , . . 0*3

„ copper „ . . . 0'2

In order then to calculate the watts lost through brush
friction, we simply multiply the total pressure on the com-
mutator (in pounds) by the peripheral speed in feet per minute
and by the friction coefficient, which gives the losses in foot-
pounds per minute, and then reduce to watts by dividing by
33,OCX3and multiplying by 746. Hr ^inrp^^^ —

foot-pounds per minute may be brought to watts by multi-
plying by this figure, or by dividing by 44-2 which is its
reciprocal.

Example. — Taking as example the same machine, if we assume
the brush pressure as i^ss lbs. per square inch, and the friction
coefficient as o'j, since the peripheral speed of the commutator is
3350 feet per minute, we have as the friction loss

160 X i'55 X 0*3 X 3350 X 746 -T- 33.000 = 5650 watts.

In estimating commutator-losses it must be borne carefully
in mind that with brushes or commutator in bad condition,
the losses (both mechanical and electrical) will probably
come out considerably greater than the above calculations

indicate.

oy Google

X2Q Dynatno Design.

Commutator Heating.

Let Wi represent the total commutator loss, electrical
and mechanical, in watts,
„ Sa represent the heat radiating surface of commu-
tator in square inches,
„ V represent the peripheral speed of the commu-
tator in feel per minute.
„ 6, represent the final temperature rise, in degrees.
Centigrade.

Then, according to tests made by Professor E. Arnold,

Q 46-5 X -Wj

Sa (I -t- 'OCOS v)'

According to Messrs, Parshall and Hobart, the rise of tem-
perature of the commutator *ill seldom exceed 20" C. per watt
per square inch of peripheral radiating surface at a peripheral
speed of 2500 feet per minute ; for ventilated commutators
this figure may be considerably improved upon.

(E.) Friction and Windage iowi-i'.— These are naturally
very difficult, if not impossible, to calculate with any accuracy,
and arc usually estimated by the designer from previous ex-
perience of the same type of machine as a percentage of the full-
load output. Direct-coupled machines will have smaller fric-
tion losses than belt or rope-driven machines, and low speed
dynamos smaller mechanical tosses than high-speed machines-
of the same output. For belt- or rope-driven machines running
at the usual speeds found in practice, the mechanical losses
may be taken as being from 3 to I per cent, for outputs of la
to 300 kilowatts. For an approximate method of calculating
the mechanical losses of dynamos, the reader should consult
the writings of Mr. Fischer-Hinnen.'

Efficiency. — Having estimated the separate losses, it be-
comes a very simple matter to calculate the efficiency of th^
machine for the load at which the calculations were made.
3ty the efficienc}' we mean simply the relation between the
power actually delivered electrically at the terminals of the
' Continues Current Dynamos, London, 1899.

izecoy Google

Losses, Heating, ami Pressure- Drop. 1 2 i

dynamo to the mains, and the power applied mechanically at
the shaft to turn the armature, both qualities being for con-
venience expressed in watts, so that we may write, the
efficiency as

Output watts _
Input watts

There is little advantage in adhering to the old terms
"electrical efficiency," "gross efficiency," etc., as the above
definition includes everything within the true use of the term.
If W is the output in watts of the dynamo or motor, and if
■ the sum of all its losses as estimated above, we have

-= ij.

' W + «-

as the true or commercial efficiency expressed as a percentage.
The efficiency will differ at different loads, since the watts lost
constitute a different proportion at different outputs. The

' core-losses are nearly constant at all loads, and so is the loss

: of energy due to excitation by the shunt coils. The question
what the efficiency will be at i-load, or ^-load, or at I J load
depends largely on the proportion of the various losses. At

' no-load there are hysteresis and eddy-currents in the iron

, core-body, excitation losses, and friction. As the load in-
creases there is added the loss by healing in the copper of the
armature, and in the series coils ; and these losses increase with

' the square of the current. Consequently the efficiency, which
at no-load is zero, goes up to a certain maximum, which, if the
design is good, should be at the normal full-load ; but it

' should be high also at half-load and even at one quarter-

. load.

The form of the efficiency curve is shown by a typical

■ example in Fig. ^6, which also shows the values of the
separate losses at different loads. This set of curves relates
to a particular 6-pole 2iX) kilowatt machine, supplied by the
General Electric Company to the Central London, Railway
and described by Messrs, Parshall and Hobart.' Fig. 47
' Eltelric Ceiieraleri, p. 190^

D, Google

122 Dynamo Design.

shows how the excitation in this same machine automatically
increases with the load by the compounding action of the
series coil,

(F.) Secondary Ccpper Losses.

In addition to the ordinary ohmic loss of power due to
the resistance of the armature conductors, there -are certain
obscure causes of loss that lower the measured efficiency of

t I 1 1 1 1

...

MP_€_200fc, -l3S^p^.-S00.,.

■

NCY

.^A

LUUU

r.''

—

—-'

',^

,s^^

COHt

2flOO

■

n^<i

s _

COMM

ITATO

^glllg^

pasri«

.UDW" ■

ES5'

320 3B0 - 400

machines. One of these is the production of eddy-currents
in pole-pieces, heating them and wasting some of the power
applied to drive the armature. Another is the production of

oy Google

Losses, Heating, -and Pressure-Drop. 123

eddy-currents in the copper conductors themselves. Akin to
this isthe actual increase of resistance which occurs if for any
rsason the current in the conductor does not distribute itself

7,000

COOO
S.DOO

-6-a0OK.v-l3Srpni..-6O&v.

COKIPOUNDINQ ' CURVE

FOR BOO VOLTS.

equably in the cross-section. This may (and does) occur in
the following manner.

For bar-armatures, rectangular bars set edgewise in slots
are almost universal. Round bars are very rarely found in
continuous-current machines. Smooth-core armatures do not
lend themselves to bar-winding, because solid copper bars set
<jn the outside of a smooth-core are liable to a serious waste of
energy that does not occur in small-wire windings. When the
conductors present any considerable breadth, there is a
tendency to set up eddy-currents in them as they enter or
leave the magnetic field, owing to the fact that one edge of the
bar may be passing through a field the density of which is

oy Google

124 Dynamo Dedgn.

very different from that of the field through which the other,
edge is passing. For example, if the surface speed is 3000
feet per minute, and the bars are ^ inch wide, it may be that
the front edge may be in a field of density, say, of 40,ocX5 h'nes
per square inch, while the other edge is one of 30,cxxi only.
If the active length of the conductor is, say, 12 inches, then it*
front edge will be cutting magnetic lines at the rate of
288,000,000 lines per second, and therefore the induction in
that edge will be 2 ■ 88 volts ; while in the hind edge the corre-
sponding induction will be only 2 '16 volts. The difference,
or 0*72 volts, will tend to set up an eddy current flowing up
one edge and down the other edge of the bar. Suppose the
bar to be \ inch thick ; if for the purpose of argument it is
regarded as equivalent to two parallel bars \ thick and \ wide
united at the ends, the resistance round this elongated loop
will be that of a rod of copper 24 inches long and of 0*25 X
0*125 square inches of cross-section. At 6c" C. this resistance
will be only o ■ 000607 ohm ;
and an electromotive-force
of 0-72 volts would set up
a current of 1 186 amperes !
But as the electromotive-
force is 0-72 between the
extreme edges only and has
lesser values towards the
middle of the width, the
eddy-current up and dowD
the strip will be less than
this. Even if one takes a twentieth part of the value so found
as being more probable it is serious enough from the heating and
waste of power that it' entails : for at no-load there would be
this waste in each conductor as it approached and left each
pole. At full-load there might be no actual eddy-current.
For if the full-load current through the conductor were 150
amperes, then superposing upon this an eddy-current of 59
amperes in the two halves of the conductor (as in Fig. 49),
the result would be that in one half of it the current would be
75 + 59 = 134 amperes, and in the other half 75 — 59 = 16

izecoy Google

Losses, Heating, and Pressure-Drop. 135

amperes. The 150 amperes no longer distribute tliemselves
equally through the breadth of the strip ; and the total heating
of the strip would be precisely the same as if the eddy-current

Fig. 49.— EDDY-CuaRENT in Widi Bak.

■were really there. At the peripheral speeds actually used, it
IS found impossibleby any shaping of the pole corners to avoid
excessive heating of solid copper bars on a smooth core if
their width exceeds 0"2 inch.

In sunk windings these losses practically do not occur,
unless the slots are very narrow and deep so that there is a
magnetic leakage across the slot from tooth to tooth. In the
case of a conductor being made up of several wires in parallel,
the separate wires must not not lie in different slots, for reasons
similar to those already discussed.

To eliminate such eddy-current losses Crompton ' pro-
posed several methods of twisting or imbricating around one
another two or more strips, so as more effectually to neutralise
the eddy-currents. He introduced the use of bars made of
stranded wire compressed into a rectangular form, each wire
being oxidized or lightly insulated.

Calculation of the Pressure-Drop. — It was shown in
Chapter I. (page 37) how the saturation curve cf a dynamo
machine may be constructed, that is, the curve connecting the
ampere-turns upon the magnetic circuit and the useful flux
produced by them in the air-gap. Let O C in Fig. 50 repre-
sent such a saturation curve, the ordinates representing the flux
Tjeing cut by the conductors and the abscissae the ampere-turns
producing it. Now the fundamental equation for the induced
«iectromotive-force (page 79) tells us that

E=A«ZN„-i-io",

■or that E = Na X a constant depending upon the construction

and operation of the machine, and which may be denoted by/

' See JOur. Inst. Ela. Engineers, xix. 140, 1850.

oy Google

126 Dynamo Design.

Consequently the ordinates of the curve represent the induced
electromotive- force to a different scale, that is, E = y Nj, and
we may therefore regard it as being the curve of induced
electromotive-force of the machine for different excitations at
constant speed — in other words, it is the " no-load charac-
teristic" of the machine. Assuming then that Fig. 50
represents this curve for a particular machine, we can see from
it what the pressure-drop at constant speed and excitation
will be, and incidentally determine the amount of compounding
that must be employed in order that constant pressure may

Fic. 50,— Saturation Curve.

be maintained at the terminals of the load. Let the ordinate

O V represent this constant pressure. Evidently at no load
an excitation of X 1 ampere-turns are required. Now at the
full load of Ca amperes, there will be three causes tending to
lower the pressure, namely the effects of {a) ohmic resistance
of armature and series coils ; (1^) demagnetizing action of the
armature ; and (c) distortion of the armature flux, the effect of
which only becomes of importance in slotted armatures. The
first two are easily allowed for. The lost volts at full load
are evidently

« = (C, xO + (C,xrJ...

iLCD, Google

Losses, Heating, and Pressure-Drop. 127

the last term not being required in the case of shunt machines.
Adding this quantity to the ordinate O V we obtain O E
as the pressure that must actually be generated in the
armature, and this requires an excitation of X3 ampere-
turns. As it is frequently convenient to check the dimensions
of an armature conductor in a preliminary design by means of
this quantity, the table appended below, giving average values
oie, may be of use in,this direction..

n»chb?mkuJ«tit

ShuDl nudilnei.

Compvund machinec.

■ 5

100

ao3

500

With regard to the demagnetizing ampere-turns of the
armature, we know generally that these arc the ampere-turns
lying in the angle of brush lead. Assuming that the brushes
will be set just under the pole-tips at full load, the demag-
netizing turns are given by number of conductors lying between
tite adjacent pole corners multiplied by the current in them.

These ampere-turns we multiply by the dispersion co-effi-
cient V, because they have to be neutralized on the field system,
then add the result to X„ and set them oft' as X^ ; and by
projecting this value up to the curve and across, find the
point E,.

Now with a smooth core armature, the distortion of the
ilux in the air-gap does not produce a pressure-drop. In
Fig. 51 let AB represent the width of the pole-face to scale,
^d E F the flux-density in the air-gap Bj. Then the area

oy Google

138

Dynamo Design.

A B C D is proportional to the useful flux N^ , and at no-load
we may regard this flux as being distributed uniformly alongf
the air-gap as indicated by the rectangle. But at full-load the
flux is heaped up at the forward pole-horn and withdrawn
from the hindward horn, as indicated by the figure A H G B.
and as the permeability of the air-gap is constant, the area of
this figure is equal to the area of the rectangle A B C D.
For instance, if X^ are the ampere-turns required for the
air-gap (flux-density = B3), and X^ arc the ampere-turns lying
under the poles and producing the distortion, we have

Line F E proportional to X^ ;
„ AH „ X,- X^;

„ B G „ X^ -I- X^ ;

and consequently no diminution of the total flux takes place
in a smooth core armature.

We see then that Fig. 50 gives us the compounding for a
machine with such an armature. We see that if the load were

*,^^^

...^^-f^

L.
C

U
11

Fig. 51. Fig. 52.

switched off" (the speed remaining the same) the volts at the
terminals of the machine would rise from the value O V to
that of OE]. Consequently the compounding ampere-turns
are given by (X, — Xv). and the shunt ampere-turns by Xf

oy Google

Losses, Heating, and Pressure-Drop. 129

" If the machine were merely shunt-wound, it would require to
have inserted in the shunt-circuit a regulating rheostat which
could be adjusted to give X, ampere-turns at no-load, and X^
ampere-turns at full-load.

But for toothed armature machines there must be made an
allowance for the distorting effect, due to the fact that
the permeability of the teeth is not constant. As before, let
(in Fig. 52) the rectangle A B C D be proportional to the
useful flux, E F representing the flux-density Bj in the air-
gap. If the teeth were of constant permeability, the flux
from the pole-face could be regarded as being of the same
value as at no-load, but distributed differently, as shown by
the figure ABGH. But the increased flux-density at the
forward pole-horn causes the permeability of the teeth at this
point to have a much lower value than they have with the flux-
density B,, while, on the other hand, the permeability of the
teeth under the hindward horn has increased on account of the
diminished flux-density in them. As a result, the line H F G
takes a bent form as shown by the curve K L ; and the sftape
of this curve is the same as that of the saturation curve over
this range. As can readily be seen from the figure, the area
A K L B is considerably less than tha area A H G B, that is,
there is a diminution of the useful flux N^, and consequently a
corresponding pressure-drop, and the diminution will be as a
rule greater, the greater the flux-densities in the teeth.

One way of estimating the number of compensating
ampere-turns required to overcome the effect produced by
the distortion of the useful field is as follows. In Fig. 53 let
O L be the saturation curve of the machine, the ampere-turns
required for no-load, and for the full-load induced electro-
motive force on no-load (and, therefore, without the extra
allowance for distortion), being set off" upon its scale of
abscissa OX as Xi and Xj respectively, these being esti-
mated as shown above. Now mark off"

0A = X3- X^; OB = Xi-l-X^

upon O X. The point A then represents the hindward pole-
hom, and point B the forward pole-horn. Had .distortion

oy Google

130

Dyn%vto Design,

been absent, the ampere-turns required to produce E, volts
would have produced a flux across the gap proportional to
the area of the piece A B C D. But as distortion is present,
the flux is proportional to the smaller area A B L K. All we
■have to do now is to shift the point F higher up the curve to
a point such as V, so that the area A' B' L' K' becomes equal
•to the area A B C D. This gives a new point X^ along O X,
representing the full-load ampere-turns required. Conse-

^r^

Fig. 53.

quently, we see that, if the machine is compound-wound, the
series ampere-turns must be X( - X,, and the shunt-turns Xj
in order that the terminal volts may be O V at full-load. If
a shunt machine, the resistance of the shunt rheostat must be
capable of reducing X^ ampere-turns to Xi ampere-turns. If
the machine had no shunt resistance, then the drop from fult-
" load to no-load would be (O Ej - O V) at constant speed.
It is unnecessary to say that the above methods of prede-

::,y Google

Losses, Heating, and Pressure-Drop. 131

termining the pressure-drop and amount of compounding will
not give an exact result. Such a result would be quite
impossible to arrive at by any process of calculation only,
and as a matter of fact great accuracy is not required* If the
machine is shunt-wound, the regulating rheostat will in practice
have sufficient margin each way to cover the inaccuracy, while
compound windings are in practice adjusted in the test-room by
the method of experiment. It becomes, therefore, necessary
only to predetermine the pressure-drop from the point of view
of the winding space required- on the magnet bobbins, and

FIG. 54.

from this point of view the above method will be found to
give extremely good results.

■ Resistance of Shunt Regulator. — Knowing the values of
Xi and Xj as found above (Xj and Xj for machines having
smooth core-armatures) it is a simple matter to determine the
necessary resistance for the shunt regulator. Let, in Fig. 54.
O C be the no-load curve of the machine, the no-load and full-
load ampere-turns being given by Xj and X<, respectively.
Then we have to find the value of shunt regulating resistance
in order that the machine may deliver current at the constant

K 2

Digil.ze.:,, Google

1 3 2' Dynamo Design,

electromotive-force V ; it being assumed that the regulator
is to be short-circuited at full-load. First, obtain the points
P and Q by projection, and Join O P and O Q.

Let the number of shunt turns be denoted by S,.

Then shunt current at no-load = X, -r- S ;

and „ „ „ full „ = X, ^ S^'; hence

..= -^^'-^ = tan«.
., + ..= §^^=tan«.

Along O X mark off a piece O R equal to the number of
shunt turns S, and to the same scale as Xi and X*, etc. Erect
a perpendicular R T. Then

r, __ RS
^^-'-,~Rt■

That is (R T — R S) gives directly the value of r^ in ohms ;
it being read off from the scale of electromotive-force, as
indicated by the figure.

On the opposite page, Table X. gives a list of suitable
materials for rheostats, and the respective coefficients for
calculating the lengths required for giving prescribed amounts
of resistance.

The Adjusting Shunt. — To adjust the operation of com-
pounding coils, they are often shunted with an iron resistance.
As the load rises this shunt heats, and its resistance rises
relatively to the compounding coils of copper, so increasing
the compounding at the maximum load.

Inherent Regulation. — One way of considering the regu-
lating properties of a machine, is to observe by experiment
(or calculate from the saturation-curve) how many volts the
potential will rise if, being excited with the full ampere-turns
necessary to give no-load voltage at the terminals under full-
load, the excitation is maintained, but the load taken off the
armature. The resulting rise of voltage may be called the
inherent regulation, as distinguished from iha pressure-drop.

oy Google

Losses, Heating, and Pressure-Drop. 133

TABLE X.— Resist A jiCB Materials for Rheostats.

NiircofMiteriil.

Sp«.«c«.U,a„„

R=-iIS™=[«r
dtgr«C.

Ppecific
Gr-vily.

Coefficlcxtto
Copp«^cJiHlue-

Microhnii
culx.

^riich"

Conslanlaji . .

19-7

zcroornegative

8-S

30'8

German silver .

8-2
I. 8

0-0004^

8'S
8-S

<3
i8s

... . . . .j

3-94
4-73

0-00450

7-8
7-81

6*
7-4

Krappln . . .

33-5

000077

8-8

516

Mmginese coi-per

100-6

41-8

000004

8-7

Manganin. . .

46-7

18-4

o' 00033

8-94

Neusilber . . .

15-6

000020

8-5

Kickeltn . . .

33a

I3'i

000030

9-0

.7-.

000033

9-0

«7

Nickel stetl . .

29
75

114

1 ooooso j

8-4
8-S

18
4.5

FlaUnoid ■ - ■

3*5
5«

11-8
201

1 O-0002. '

8-5
8-7

246

9' 7

000039

8-6

•S*

RlieoEtan . . .

473
100

iS-6
39-4

J o-oooi3 1

3'6
8-6

30
6a

N.B. — The composiiion of these alloys varies much as to its proportion
according to the source of manufacture. For calculating the resistances of wires
or strips of these mateiials, the simplest procedure is to calculate them as if of
copper, and then multiply the resistance so found by the coefEcienl given in the
last column of the table. As Iheir specific Gravities do not differ greatly from ibat
of copper (8-8), they will all (except iron| weigh approximately the same as a wire
of copper of same fauge and lengtb.

' Contuning to per cent. tin.

oy Google

Dynamo Design.

CHAPTER VII.

THE DESIGN OF CONTINUOUS-CURRENT DYNAMOS.

The calculations and formulae required by the dynamo
designer have been already mostly given and explained in
the preceding chapters. A careful study of these, and the
detailed calculations given of the three representative ma-
chines in the present chapter, will make the methods adopted
in designing continuous-current dynamos sufficiently clear.
Beyond giving a number of working data, and an order of
working that may be adopted in designing, we shall rely upon
the worked-out examples of the succeeding chapter to give
the reader an insight into the principles of dynamo design.
It would be useless to do otherwise, as so much depends
upon the skill and experience of the designer, the type of
the machine, and the conditions of the specification as to
speed, output, voltage, regulation, and heating limits to
which he is obliged to conform, that no hard-and-fast rules
applicable to every case can possibly be given. The follow-
ii^ remarks and working constants are to be taken as apply-
ing only to modern machines of fair and large sizes — that is,
to slotted drum armatures with multipolar field-magnets.
which are assumed throughout, except where anything is
explicitly said relating to other types.

There are two principal ways of designing a dynamo to
fulfil specified conditions as to freedom from sparking, heat-
ing and efficiency ; the output, speed and voltage being
assumed to be the same in each case. With a given number
of poles on the field-magnet frame, and of conductors upon
the armature, the effects of armature reaction may be kept
down either (i) by working the teeth at normal fiux-densities
(say 100,000 lines per square inch, and under) and with a

izecoy Google

Rules for Design. 135

wide air-gap, or (2) by forcing the magnetism of the teeth and
working with a smaller air-gap ; the high reluctance of the
teeth with such lai^e magnetic densities acting like an exten-
sion of the air-gap. The former method corresponds to
Continental practice, and the latter to American practice in
continuous -current dynamo building. It would appear that
the second method is considerably the better from the point
of view of avoidance of sparking (both being the same with
regard to efficiency and heating), and therefore we adopt it
here, as giving a better commercial machine.

There is another aspect in which the plans followed in
designing may differ. One may begin by following general
experience as to speeds, sizes and electrical proportions, and
having proceeded to sketch out the main features of the
design, may then proceed to calculate the power wasted as
heat in the various parts, and so estimate the efficiency, and
then, after so finding the various items of heat-waste, return
and amend the first calculations according as to whether we
have found any part to heat too much or too little. Or,
instead, one may begin, as the result of experience, to as-
sume and allot in advance the various permissible losses of
power in the various parts— so many watts in the iron core,
so many in the armature copper, so many in the field-magnet
coils. One will then have a definite idea as to how much
cooling surface will be necessary, and what will be. the allow-
able current-densities in the copper and flux-densities in the
iron. This procedure settles many points in advance. Similar
considerations have long governed the design of transformers,
and their advantage has gradually been acknowledged by
dynamo designers.

In another respect also dynamo design has developed.
Formerly the dynamo designer built his machines without
knowing the precise voltage which they would give at any
particular speed, and left the speed to be determined by trial
after the machine should have been completed; then adding a
pulley of such size as would suit the conditions ol driving.
But now-a-days, when nearly all dynamos are direct-driven
from engine or turbine, the speed is prescribed beforehand by

oy Google

136 Dynamo Design.

conditions fixed by the steam-engine builder or the turbine-
constructor. Therefore now all dynamo design proceeds on
the supposition of a prescribed speed. Further, in designing
a series of dynamos of different outputs from small to large,
it must be remembered that engine-conditions govern the
selection of speeds, and that it will not do to assume that
a lOOO-kilowatt dynamo can run at the same speed as a
lO-kilowatt dynamo. Neither will it do to assume that the
speed may be varied inversely ' as the number of kilowatts.
A rule more near to practice is that in a series of steam-
engines of given type, the speeds vary about inversely to the
square root of the capacity. If a lO horse-power engine runs
at 800 revolutions per minute, then an engine of lOOO horse-
power will not run at jj[^ of the speed, but at about -j^j of
the speed, namely 80 revolutions per minute.

Working Constants and Trial Values.

{a) Flux-densities. — As average values for the magnetic
s parts of the machine at full load, we

densities in the var

ous parts of

may take :—

Flux -density in—

LJDCl per

AtmBture body

„ teetli

Air-gap . .

Hagoet corea .

4S,

60,000
130,000

oco to 5S,

,, 7oke .

(70.C

100 10 100

\ 35.00

For sparkless commutation, the density in the armature-
teeth must not differ very much from the above value.
Mr. H. S, Meyer recommends an apparent density of 140,000
to 155,000, The density in the armature core-body should be
less, and is determined by the permissible iron-loss.

{b) Length of Air-gap. — This should not be less than half

If (his might be assumed, the design of a seiics of dynamos would be much
siroplified, as Mr. S. H. Short has shown, since then all armatures might be made
of same axial length, and all field-magnet poles at same size, the number of them
being simply increased, 4, 6, 8, 10, \l or more, in simple propoilion to the
requited capacity.

oy Google

Rules for Design. 137

the width of a single slot, even with highly saturated teelh.
If the slot is partly closed, take three-quarters of the maxi-
mum width as a trial value for the length of a single air-gap.
[c) Number of Poles. — With such values for flux-density,
and air-gap length as taken above, the armature ampere-con-
ductors per pole, at full-load, should not much exceed 14,00a
To get a rough idea of the numberof poles required, we simply
multiply the total number of the conductors, Z, around the
armature by the full-load current in each, divide the product
so obtained by 1 4,000, and take the nearest even integer as the
number of poles. But this assumes the number of armature-
conductors to be known. Another criterion ^ is the prescribed
output of current, since, to avoid sparking troubles, it is wise
not to attempt to collect more ^ than 200 amperes at any one
row of brushes. As the number of rows of brushes in either
the positive or the negative set is equal to the number of
pairs of poles, the total number of rows of brushes will be
the same as the number of poles. Hence it follows that a
trial value as to the proper number of poles can be found by
dividing the prescribed full-load current by 100. Thus, if the
machine is to give 950 amperes, 10 poles will be adequate.
{Special machines, such as electrolytic machines and very
slow-speed exciters to be mounted on the shafts of large
alternators, are exceptions.) In general this rule gives too

' Wiener (Practical Calailathn, Snd edition, 1903, p. 2870) advises to malce
the choice of poles depend only on the speed, rhe object being to limit the number
of merssls of magnetization per second in the annatuic-core, and so keep down
the kcmatuie iron-losses. He states the number of cycles per second at 10 lo 15
in slow-speed machines, increasing to as many as i!5 or even 35 in high-speed
machines. His rule is equivalent to saying that for slow-speed machines one can
find the appropriate number of poles by dividing the number of revolutions per
minute into izoo or iSoo, and taking the nrarest even integer higher than the
•quotient. But such a rule leaves out of sight the size of the machine : and it
would be absurd to give the same number of poles to u 40-kiIowalt and to an
800-kilowalt machine, limply because each of them ran at, say, i xo revolutions
per minute. Fischer- Hinnen {Cotilhwoui Current Dynamot, l£99) recommends
4 poles as appropriate for machines between 6 or 20 kilowatts and 100 or 150 kilo'
watts ; and 6 poles as appioprinle up to 200 or 300 kilowatts.

' Nevertheless satisfactory machines exist with fewer poles than by this rule ;
for example, the Berlin generators of the A. E. Gesellschaft, giving 3600 amperes
nnd having only iS poles, and the 6-pole generators of the Oeilikon Company,
at the Volta Company in Rome, giving 1500 amperes.

iLCD, Google

138

Dynamo Design.

few poles for the smaller sizes, and too many for the lai^er
sizes of dynamo.

(d) Current Densities. — As an approximate guide to the
sizes of the conductors required for the different parts, we
may take : —

c™,D...,»^ 4XT.S

Squ.nMil.

CircuUi HHi
pwAmpe,..

1 •5'»
In annature cmductors . .J to

1 3O0O

667
500

S47

[2400
Ij 4000

4>7

53'

318

In field-roagnrt coils . . .t| lo

|; Soo

1670
1250

1126
1591

The section is, of course, finally determined by the per-
missible heating and voltage-drop. For field-magnet windings
see the rules in Chapter III. pp. 49 to 56.

{e) Number ef Armature Conductors. — Under the standard
conditions of flux-densities and gap-lengths adopted above,
the number of ampere-conductors per inch of periphery (at
full-load), should come out at about 600. Or, writing Z.as
the total number of armature- conductors, C^ as the total
armature- current at full-load, and p as the number of poles,
the current in any one conductor will (for parallel-wound
armatures) be = C, -j-/. Hence the total number of ampere-
conductors all round the armature (sometimes called the
" circumflux") will be Z x Ca-f-^; and this, by the above
rule, ought not to exceed 6co times the number of inches all
round, or 6oo7ri/, where d is the diameter in inches. This
gives, as a formula for calculating the trial-value of Z,
1885 X p_y^d

For armatures with series or series-parallel windings, where

izecoy Google

Rules for Design. 139

there are^c circuits (see pp. 85 and 97) through the armature
(instead olp circuits) the rule becomes

But these rules give values that are often wide of the mark.
Another, and for some purposes better rule is this : — First
obtain a trial value for the magnetic flux N, and then calculate
Z from the electromotive-force and speed. Using « for the
revolutions per second, and E as the volts at no-lpad, the
formula for parallel-wound armatures is

<N ■

[7]

In every case the trial-number when obtained will need to be
adjusted so as to give a proper multiple for winding.

Example. — Required the proper number of armature conductors
for a dynamo M P — 8 — 200 kw. — 375 r.p.m. — 125 v. — 1600 A.
Taking the diameter of armature as 48 inches, and the trial value for
the flux N = 4 X 10' ; by formula [a], the value of Z comes out 45 r ;
by [y] it comes out 500. The actual value of Z in the machine (as
built at Schenectady) was 480.

(_/") Number of Commutator Segments. — The value of the
average volts per bar of the commutator furnishes to a certain
extent a sparking criterion of a machine, as they measure the
inductive action of each individual section of the winding. If
the suitable number of volts per segment is known, one at
once obtains an estimate of the number of segments in the
commutator by dividing the prescribed voltage of the machine
by the suitable number of volts per segment, and then (for
parallel-wound machines) multiplying by the number of poles.
Or, in the case of armatures with series or series-parallel
windings, multiplying by the number of circuits.

' Arnold {Die Ankerwichelungcii, 3rd ed., 1899, p. 178) gives a formula as. .
follows: — That the number of commutator s^menls must tat be Itss than^-ayj
(o o'04 times the product of Z into the square-root of the cnitent in any one
circuit of the oimature. For example, in a 4-pole machine witti tour drcuJis, each
carrying tooainpeies, andbaving 336 armature conductors, Arnold's rule iroidd ftX

D, Google

140 Dynamo Design.

For machines having flux-densities as assumed above, the
following values are found.

For Macbinu working al

^pL-lS™-"

*""p.'?lr""*

500 to J 50 volts ....

300 to 230 VOIIS ....
lOOlOIIOYolU . . . .

51 to IS

410 10
3

35 to 100
30 to 50
35 >o 35

(f) Size of Armature. Steinmetz coefficient. — The ques-
tion next arises, given the specification of a machine to be
built on the above lines, how is the designer to begin with the
calculations ? Obviously he must get some idea how large
the armature must be. Previous experience of the same type
and size of machine will of course be sufficient to go upon,
but failing this, use must be made of an empirical rule con-
necting the output of the machine with the over-all dimensions
of the armature core. The simplest of such empirical rules
is that originated by Mr. Steinmetz, that the product of the
diameter of the iron core-body into its length is equal to the
kilowatts of output of the machine multiplied by a certain
coefficient ; or in symbols

Where <r, the Steinmetz coefficient, will have a value, if d
and / are both in inches, not differing much' from 3. In old

(be Dumber of commutator segments as not test than 0*037 x 336 X Jxtxi =: 123.
The machine actually had i63, two conductors making one loop as the element
of the winding. Had the elements been d two loops each the number of segments
would have been 84, which is too small. The machine would have sparked in
all probability. Nevertheless it is certain that good machines have been con-
structed for which the constant was lower than o'os?, even ns low as o'al5.

' If d and / are given in millimetres, the values of a will range at about 2000,
£oingdownin large modern machines to 1250, or going up in small and si ow.speed
machines to 3500 or 400a,

There is a rational basis for this Steinmetz coefficient, for if it is assumed ihat
there k a best average peripheral speed for the conductors moving in a field of

oy Google

Rules for Design. • 141

machHies and machines of relatively small output, or of slow
speed, the value of v may go up to 5 or 6 ; for large and
well-designed machines it may fall below 2. Of the various
machines mentioned in thic book the lowest value of <r is I '44.
being that of the Hobart design of 1600 kto., p. 229, While
the highest is 13 "4, being that of the very slow speed exciter
of Kolben & Co., p. 217. , -

best arerage den^ty, and lha7there is a. fixed limit, of temperature rise and a fixed
ratio fot the armaluie losses to tbe normal output, then the output ought lo bear
a constant ratio to the working surface of Ihs- armature — Iberefore proportional
Vid-Kl.

Let V be the peripheral speed in inches per second, B the average flux-density
(In lines per square inch) in the air-gap, if the ratio of pole span to pole pitch,
W the full-load oufpul in kilowatts, q the number of ampere-conductors per inch
periphery, then the total polar area surrounding the armature xi tv. d y. 1 1(. if
square inches. Now all the work done by the machine lakes place under the
poles ; and ire may write

Work done per second = peiipheral force x peripheral velocity,
which may be written

Output = force per unit surface x total active surface x peripheral velocity,

= Bx;XifX*Xi^X/Xi' (eigs per second).
W = Bx?X*XTX</x /Xf-t-lo'MWIowatts).

Now asiigning to B, q, if, and v the respective average values 50,000, 600, o' 7,
and 500, values which are found in good non-sparking machines, this becomes

W=033><rfX/,
whence

3-03 = rfx/-hW,

thus arriving rationally at the Steinmeti formnla. Moreover, as the value of the
constant was fixed by non-sparking conditions, it is clear that the limit of the
output of the machine depends on healing alone ; d x I being a measure of tbe
cooling surface. Thui under the assumption (which Is found to be true in prac-
tice) that the limit to the output of machines of this type is put by heating, not by
sparking, it follows that the value of ibc constant depends almost entirely en (A^
ptriphfral sfeed. That ia part of the reason for tbe differences observed in the
examples given above, and furnishes a strong argument in favour nl high periphtml
tpeeds, and therefore of armatures lending toward the fly-wheel type,

Kapp has given a rule, which in British units i,d and / in inches, and peri-
phetal speed v in feM per minute), comes to this:—

where W is the fuU-load output in kilowatts, and e a coefficient which varies be
tween tbe valoes of 35,000 for small ring-wound machines and slow-speed machines
down to 7000 for large well -ventilated drum nucbinet.

oy Google

I4S

Dynamo Design.

E«mpl«.

Kilow«ii.

CoJiBcKDt.

r, Engl. El. Mfg. Co., p. aio

IIOO

1-96

Hobftti's design, p. 119 .

1600

1-44

400

3-7

Siemens & Halske, p. 231 .

1000

3- 16

Brown, Boveri & Co., p.iiS

480

2-2S

■

p. K>3

194

3'47

Kolben&Co., p. »t6

aSo

4*35

Walker Co., p. 173 .

440

3-6

Gen. Elect. Co., p. 309 .

550

3-S7

Bipolar

JohDson & Phillips (1887) .

8-4

Having thus obtained a trial-value for the product of
diameter and length of the armature core, it remains to
separate it.into these two factors. One method of doing this
is to take the highest permissible surface speed ; divide it by
the number of revolutions per minute and by ir, thus getting
the largest permissible diameter as one of the desired factors.
Another guide, assuming the number of poles to be fixed, is
to assume (as is a fair rule for cast-steel pole-cores of circular
section) that the length of the armature-core wilt be equal to
half the pole-pitch at the armature surface ; in which case
the ratio rf// = 2 j>/ir ; whence d* = 2 kw X / X <r -j- ir.

Another rule connecting output and size is

kw = 0-064 X d}l X revs, per min. ;

where the coefficient 0-064 'S a sort of mean, and will be
greater for machines of greater specific output, therefore
larger in the case of large modern machines than for small
or old types. If the length / is assumed to be known we

may deduce

= 7-9V',

revs, per min. X / '

{h) Assignment of Losses of Energy. — Losses in energy
due to ohmic heating of the copper, to hysteresial and eddy-

izecoy Google

Rules for Design. 14,3

current heating in the iron, and to friction, are inevitable.
They are discussed in the preceding chapter. Such losses must
be kept down, because they lower the efficiency; and because
an undue rise of temperature in any part is not permissible.
Experience has shown that if a machine is to be so designed
as not to overheat in any part, and to make the total loss of
the minimum value compatible with economy of material, its
various losses must be rightly apportioned out. The follow-
ing may be taken as the apportionment of the losses in
machines of different sizes : —

OutpuHnKJo.

Efflcicaey
pacaol.

«™u«,l™..

'^..

Loii.

3-3

^£r

FHction acid
WiBdag..

5 to 40

3. a

2-5

0-5

■0

10 „ 60

3-D

2'I

0-4

40,, 100

3"2

a-8

1-6

0-4

75 » 300

z-8

2-3

I -55

300 „ 500

■ 2-4

1-8

i-S

0-3

400 „ 1000

'■5

'■35

0-35

Such a table would, however, be misleading unless it is
borne in mind that the values may vary considerably even in
machines of the same size for different speeds and under
different conditions of working. For example, a 93 per cent,
efficiency is in general too high for a 7g-kilowatt machine,
unless it is of very large size for its output. Such tables may
however be drawn up with some accuracy for a standard
series of machines of some one type, such as a series of slow-
speed tramway generators, or a series of high-speed lighting
machines. The following is a table of average values in a
series of standard generators built by Messrs. Kolben and Co.
(see next page).

Parshall gives the following apportionment for a 550-
kilowatt tramway generator : — Armature copper 2-25 ; arma-
ture iron 2*25 ; magnets 0-75 ; commutator, etc., o"75 ; total

izecoy Google

144 Dynamo Design.

6 per cent, Rothert gives for similar machines : — Armature
copper 2'7 ; armature iron 1*8; magnets I'g"; commutator
o'3 ; total 6'3 per cent

te.il

Si««J.

ruii-L«d

1™.

EiciUtbn

Lou.

te:

Lt^.

SSo

926

Soo

1000

1,200

1,000

100

450

1500

1900

2,200

1,900

aoo

350

93 "4

2800

5600

4,200

3.600

400

300

94-0

5000

6000

7.9a'

6,800

Soo

100

94-4

59«5

6400

9.500

8,000

750

100

950

Sooo

7100

13.500

11,000

N.B. — The first four of these niachines are high-spred L'ghting micbines,
itnitable for coupling direct to a h[gli-i:pced engine, and the loss due to the outer
hearing is included in friction losses. The last two are slow-speed traction
geneiatois; and for these the efficieac]' figure does not include any loss at the
bearings.

((■) Centrifugal Forces. — If a mass is whirling with a radius
of R inches, at V revolutions per minute around an axis, the
centrifugal force is 0"0009i38 "BiY* poundals per pound of
peripheral matter, or O'O00O284 R V pounds per pound of
peripheral matter. This rule can be used to estimate the
centrifugal forces on armature conductors.

£'ar(7«/&— Suppose an armature conductor weighing 0*39 lb. to
be revolving at 150 revoltitions per minute ; the radius of the arma-
ture being 31 inches. The centrifugal force on it will be

0-39 X 0-0000284 X 31 X 150 X 150 = 7'73 Ih.
{/) Calculation of Binding Wires. — In the case of smooth
cores the conductors must be secured in their places by a.
number of external bands, known as binding wires. In the
case of toothed core-disks the conductors may be held in by
■wedges of hornbeam or of hard white fibre driven in under
the tops of the teeth ; or in the case of straight teeth binding
wires may be used instead of wedges. Binding wires must be

izecoy Google

Rules for Design. 145

strong enough to resist the centrifugal forces, and yet at the
same time must occupy very little radial depth that they may
not interfere with the clearance between the armature and the
pole-faces. The almost invariable practice is to employ a
tinned wire, of hard-drawn brass, phosphor bronze, or steel,
which, after winding, can be sweated together with solder into
a continuous band. Mr. Wall, of Sheffield, manufactures a
special " plated steel " wire for binding, in sizes of 18, 22, 28,
36, 48 and 56 mils diameter respectively. Phosphor bronze
will withstand a tensile stress of from 65,000 to 120,000 lb.
per square inch. Steel varies from 125,000 to 230,000 lb. per
square inch.

To estimate the proper size and number of binding wires
required we may remember that on a pound of material at a
radius of R inches, revolving at a given speed, the centrifugal
force will be that given by the formula on p. 144. Or if </ be
the diameter in inches, and « the revolutions per second,
the centrifugal force per pound of matter will be = 0*012 dn*
pounds' weight.

Suppose we know the mass h;, (in pounds) of one conductor,
multiplying this by the total number of conductors Z, we get the
total mass of armature conductors, and dividing by tt we find
the mass that will be effectively projected in any one direction.
Putting this into the formula, and dividing by 2, we find the
total tensile force to be borne by the binding wires at one side;
and dividing again by the maximum tensile stress which the
material can stand, we obtain the net theoretical total cross
section of the whole of the binding wires. Taking a factor of
safety of 10, and a value of 100,000 lb. per square inch as the
tensile stress for steel or for phosphor bronze, we get total
necessary section of binding wires in square inches

_ ift X Z X 0'i02 X </' X »' X 10 .
■ ~ 2 X n- X 100,000 '

_ I -623 X Wi X Z X <^' X «'
1,000,000

From this total necessary section, and the appropriate wire-

Digitizecoy Google

146 Dynamo Design.

gauge, the number of wires is then calculated, and they are
then arranged in suitable belts.

Example.— Wt = 0-39 lb, ; Z = 1536; d = 62"; w = 2-5 revs.
per sec.; total necessary section works out to o'386 square inch.
Referring to wire-gauge tables we find a No. 17 S.W.G. of diameter
56 mils, has a cross-section of 0-00246 square inch. Dividing o- 386
by o' 00246 we find that 156 wires are needed. These may be
arranged as follows, 5 belts of 16 wires each over the core body, and
4 belts of 19 wires each over the extended ends of the winding, i.e,
I belts of 19 wires each over each end.

Under each belt of binding wires a band of insulation is
laid. This usually consists of two layers, first a thin strip of
thin vulcanizad fibre or of hard red varnished paper, slightly
wider than the belt of wires, and then a strip of mica (in short
pieces) of about equal width. Some makers lay a small strap
of thin brass under each belt of binding wires, having tags
which can be turned over, and soldered down, to secure the
two ends of the binding wire from flying out.

Order of Procedure in Design.

The specified conditions to be fulfilled are that the
dynamo, running at a prescribed speed (fixed by the choice
of engine), shall give out its current at a prescribed voltage,
and that when running at normal full-load of a prescribed
number of amperes it shall have a prescribed efficiency.'

The method of getting out a preliminary design to a
given specification is, therefore, as follows : —

The example which is here given is continued for the
remainder of this section.

Example. — To design a tramway generator to give, at 150 revolu-
tions per minute, 600 amperes at 500 volts, the efficiency being 93
per cent,

' Othei detaiU mny be prescribed, for examplr, the limit !□ which tempemtnTe
ii allowed to rise, the amount to which the machbe shall be compounded or over-
compounded, the efiiciencj at } or { load, the permissible amount of temporary
overload, end the like.

,1.0, Google

Rules for Design. 147

(i) Find the full-load kilowatts by multiplying together
the volts and the full-load amperes, and dividing by 1000.
Example. — As above, 300 kilowatts.

(2) Assume a suitable value for the Steinmetz co-efficient
(remembering that its value is reduced the higher the per-
missible peripheral speed), and proceed by multiplying the
required number of kilowatts by this coefficient to find the
product of (/ X /.

Examfk. — Taking o- = 3'i3, we find 300 X 3'i3 = 939. That
is to say, we know that d and / must be such that d 1^ / (in inches)
= 939.

(3) Fix a trial-value for the number of poles (see p. 137
.above).

Example. — As the full load is (100 amperes, a 6-pole design would
do ; but to be quite sure of avoiding trouble as to sparking let us
take 8 poles, so that with a parallel- wound armature there will be 8
circuits and only 75 amperes in each circuit, or 150 amperes to col-
lect at any one row of brushes.

('4) Fix the value of d and I separately, putting down
trial-values by experience, and test them by observing what
surface-speeds they correspond to, and whether any of them
agrees with the rules laid down above (p. 142). In particular
see whether the proportions chosen are suitable for the
number of poles provisionally selected.

Example. — We may at once put down a number of trial values of
dax\6. /as follows:

d^l

Peripheral
Speed

93«

3063

938

363"

930

»434

938-

2277

5»

93«

2041

Calculating the peripheral speeds that correspond to the different
diameters at 150 revolutions per minute, we see that the diameter
which gives us the surface speed nearest to the moderate value of
2400 feet per minute is 62 inches. The periphery of this armature
will be 62 Xt= i94"75 inches. Dividing by 8 we get the pole-

L 2

:.y Google

14*^ Dynamo Design.

pitch at the armature face as 24 34, and as the pole-fece will cover
about 80 per cent, of this, the pole-arc will be about ig inches long ;
making the actual pole-face about 19 hy 15 inches, which is a suit-
able shape.' We may take it then that the values of d and / will be
61" and 15".

(S) Fix the value of Z. The preferable mode of doing'
. this is, now that the size of pole-face is approximately known,
to assume a provisional value for the flux from the pole, and
calculate Z from the voltage by the fundamental formula

E = «ZN -f- 10*,

or from the variety of it given on p. 139 above. The pro-
visional value of N is found by assuming a suitable flux-
density, and multiplying this into the area of pole-face pro-
visionally found.

Example. — The pole-face being 19 X 15 = 285 square inches, if
the flux-density at the pole-face be taken at 45,000 lines per square
inch (a low estimate allowing for increase at full load), then N the
flux from one pole will be about 13,000,000 lines. Now at 150 revo-
lutions per minute the value of « will be 150-- 60 = a ■ 5 revolutions
per second. Hence if E is 500 volts we provisionally find Z as

500 X 10* -1- (2*5 X 13,000,000) = 1558.

But, for parallel windings (lap- win dings) it is preferable that Z shouM
be an even multiple of the number of poles, in this case 8. Now

■ 1538 is not a multiple of 8, but 1536 is an even multiple, therefore
we will take Z as 1536. But before finally deciding on this value we
must test it by the other requirements. If we apply the fonnula [a]
on p. 138, we find Z = 1885 X 8 x 62 -^ 75 = 1579 as the highest
number permissible if there are not to be more than 600 ampere-
conductors per inch periphery. So Z = 1536 is satisfactory.

At this point we ought to pause and test the consequences
of our procedure so far. Let us test the number of poles by
the rules given on p. 137.

■ If it ii requiied th&t the poles should be ol cast steel and circular in sectionr
Ihete might be some ndvaotige in having a more elonfiated shape for the pole-
face, foreiample, aijinchcsby 14 inc?:ei, which would be the size had the 67.incl»

■ «3te-d;sl: been selected.

izecoy Google

Rules for Design. (4g[

Example. — ^The number Z is 1536, and each conductor carries 75-
amperes ; multiplying these together, and dividing by 14000, we get
S'3, justifying our selection of 8 poles.

Further, as the- speed is « = zj revolutions per second and the
Jiumber of pairs of poles 4, the frequency of magnetization in the
■armature is ro cycles per second, which ensures that iron-losses inthe
-armature can be kept low.

(6) Fix the number of commutator segments. Lap-
wound armatures may have one segment for each loop of two
conductors. Modem practice is against having more loops
than one • per segment, so as to keep down the average
voltage per segment In all cases it is well to keep the
number of segments high.

Example. — Z = iS3^i or i9* conductors per pole. That makes
^6 loops per pole, with an average voltage of 500-^96 = S"2 volts
per loop. As it is undesirable that the volts per segment should be
unnecessarily increased, we will decide to have 768 segments in total,
or 96 per pole. Arnold's rule, p. 139, prescribes as the minimum
number o-o37XZXv'Ci where Cj is the current in one conductor.
In this case we have 0-037 X 1536 x VTs — 49^- ^°'' 7^*^ '^
above this number, whereas if we had tried having two loops per
segment and only 384 segments we should have gone below the per-
missible hmit.

It being then decided that the commutator shall have 768 seg-
ments, since each segment cannot be much less than 0-2 inches
broad, the commutator will have to be nearly 150 inches in periphery
or say 45 inches in diameter.

(7) Next settle upon the. style of armature-winding.
Modern practice tends toward preserving the utmost sim-
plicity, that is to say, it favours the lap-wound drum executed
as a barrel-winding so as to have ample cooling surface, the
conductors being in two layers, and with two, four or six con-
ductors in each slot. It is true that some designers still prefer
to use series-parallel windings, as they have the advantage of
enabling fewer armature conductors to be used for the same

' For « cose to the conlrary, see Brown, Boveti and Co.'s S-poIe machine,
p. z^. In motors undei 50 H.P. it is, usual to have more than one tucn pet

oy Google

150 Dynamo Design.

voltage, and these conductors are thicker ; and (as shown on
p. 109) they give rise, in case of irregularity in the strength
of the poles, to less internal heating from unequal currents
in the different circuits. Some examples of designs with
series-parallel windings will be found in the following
machines ; —

Kolben and Co.'s 10-pole, 250 kilowatt machine, p. 216.

Oerlikon Co.'s 12-pole, 350 kilowatt machine, p. 188.

Brown, Boveri and Co.'s High Voltage 4-pole, p. 205.

At one time designers favoured the custom of adapting
one and the same armature core so that it could be wound
with the same number of conductors for 125, 250, and 500
volts provided the magnet had 4 or 8 poles: for if the 250
volts were simply a parallel (lap) winding with as many circuits
as poles, the 125-volt armature would be also a duplex
parallel (lap) winding (p. 83), with twice as many circuits as
poles, and the soo-volt armature a series-parallel wave-
winding with only half as many circuits as poles.

The object of fixing the type of construction is that an
estimate may be made of the available cooling-surface. For
barrel -winding the length to which the oblique end parts of
the winding extend out will be about equal to half the pole-
pitch on each side of the core-body.

Example. — The pole-pitch is 24*34 inches. Adding half this, or
say 12J inches, to each face of the armature core-body, which is
already 15 inches long, makes the over-all length of the armature
(without the commutator) 40 inches.

(8) Next decide upon the apportionment of the various
losses. This might have been done at an earlier stage. The
figures given on p. 143 will assist in apportioning the various
losses. The exposed surface of the armature should have
not less than 18 to 20 square inches for each kilowatt of out-
put (see p. 69), otherwise the temperature cannot be kept
within permissible limits with these percentages of loss. If
on reckoning out the losses it is found that the armature
surface is insufficient, the armature must be re-designed of
larger size.

izecoy Google

Rules for Design. 151

Example. — Assuming the copper-loss in the armature to be 3 ■ 5
and the iron-loss to be 2-0 per cent., or in total 4^ per cent, of the
output, the total watts wasted in the armature will be 4J per cent, of
300,000 watts, or 13,500 watts. The periphery of armature being
194' 75 inches and the over-all length 40 inches, the total area of the
cylindrical surface is 7790 square inches. Each square inch must
therefore radiate away i'73S watts. As the surface speed is 2434
feet per minute, a reference to the curves of Fig. 22, p. 70, will
show a probable rise of temperature of 45° C. On this estimate it
would be preferable to reduce the copper loss by using a slighdy
thicker conductor.

According to the rule given above, the surface of the armature
ought to be from 18 or zo times 300, »>. 5400 or 6000 square inches.
As it has 7790 there should be ample surface.

(9) Next we may settle upon the number and dimensions
of the slots. The former depends upon the type of the
armature winding used and the number of commutator seg-
ments. It is almost universal now-a-days for all large ma-
chines to wind with copper strip, two layers (or sometimes
four layers) deep. But whether the slot is made wide enough
to carry two, four, or six conductors depends on the condi-
tions. Putting four or six conductors in one slot simplitics
the construction and saves labour. It also saves some space,
and should be adopted if there is fear of not having suffi-
cient tooth-section for magnetic purposes. In soo-volt
generators the depth of the slot varies from l to 2 inches
or so. Assuming a proper trial value for the current density
(see p. 138) in the conductors, their proper section can be at
once provisionally assigned. And, if the grouping has been
chosen, the necessary area of each slot can be reckoned out
by aid of a space-factor (p. 45), It must then be considered
whether this leaves an adequate section for the tooth. Since
the average flux-density in the air-gap is, say, 50,000 lines per
square inch, and the appropriate flux-density in the teeth is
130,000 lines per square inch, one would expect the teeth to
take up only five-thirteenths of the periphery. But It must
be remembered that the iron of the teeth is not continuous,
there being insulation, and often air-ducts, between the lamina-

oy Google

152 Dynamo Design.

tions. In the case of slots with parallel straight sides it is
usual to find the width of the tops of the teeth about as great
as, or slightly narrower than the width of the slots : and as
the teeth slope slightly down to their roots their mean width
will be less than that of the slot, and is indeed generally
about three-quarters of that width. The number and arrange-
ment of the conductors should now be readjusted to suit
winding conditions.

Example. — As each conductor has to carry 75 amperes, the
appropriate section at 2000 amperes per square inch will be 37,500
square mils (— 0-037500 square inch). If we decide to place these
6 in a slot the total copper section pet slot will be 0-225 square inch.
Taking the space-factor as 0-4, we have the area of slot as 0-225
-f-o'4 = 0-562 square inch at least; and as there will be 256 slots
and 256 teeth in a total perimeter of 194-75 inches, each slot cannot
well be tnore than 0-38 inch wide, each must be at least i'5 inches
deep.

(10) The internal diameter of the core may now be fixed,
by ascertaining the requisite radial depth of the core to give
an adequate cross-section of iron below the teeth. As a trial-
value one may take either the face-diameter of the armature
divided by the number of poles, or a length equal to half the
pole-arc. But the final adjustment of this radial depth de-
pends purely upon the permissible iron-loss, as this governs
the flux-density that can be used.

Example. — d = 62 inches, and as there are eight poles the trial
value of the radial depth is 7jinches; or.thepolearcbeingiginches,
the half of this is 9J inches. We may provisionally take a mean value
such as Z\ inches, which, if the slots are estimated at r J inches deep,
brings the internal diameter to 42 inches. Now the flux through
the armature-core is ^ N or 6,500,000 lines. The nett length of iron
from front to back being about 1 1 ■ 7 inches, the nett sectional area
will be about 100 square inches, giving a flux-density of about
65,000, which is satisfactory (see p. 136), and will ensure that the
iron losses are not too great.

(11) Next, settle the dimensions of the air-gap by the
principles laid down on p. 39.

izecoy Google

Rules for Design. 153

Example. — As the slots are 0-3805 inches wide at the top, the
length across the gap ought not to be less than 0-5 inch. This is
sufficient as a clearance on an armature 63 inches in diameter, but
unless there is some reason to the contrary, it would be with ad-
vantage larger : so we will adopt o ■ 65 inches.

(12) Fix the approximate dimensions of the magnet-pole
cores. These must have sufficient cross-section to carry the
full-load flux, including that which forms by leakage the stray
field ; and they must be long enough to receive the exciting
bobbin. The flux will be v x N^ ; where v is the coefficient
of dispersion, N^ being taken at its full-load value. A good
trial-value for the length of the pole-core, if cylindrical, is to
make it equal to the diameter of the section, though this may
generally be reduced after the magnetic circuit calculations
have been made to ascertain what provision must be made
for excitation. Another rough way of obtaining a trial value
for the length of the pole is to take 20 times the length of
the air-gap, if the machine is to be shunt-wound, or 40 times
the length if it is to be over-compounded. The section
necessary is fixed by the permissible flux-density (sdc p. 136).

Example.— the no-load armature flux per pole being 13,000,000
the full-load flux will need to be (for a shunt machine or compound-
wound, but not over-compounded machine) say 13,500,000. Taking
the coefficient of dispersion as i- =^ i "21 at full load, it follows that
N«must be 16,400,000. Then taking 105,000 lines per square inch
as a suitable value for the flux-density, there will be required 157
square inches. Hence the pole-core, if circular, must be 14 inches in
diameter; or, if square, about isj inches each way. Taking the
other rule, if the air-gap is o'65 inch, multiplying this by ao gives
13 inches as a suitable trial-value for the length. We may therefore
take 13 inches provisionally as its length.

(13) The necessary cross-section and size of the yoke may
then be fixed : the section being as before fixed by the appro-
priate flux-density.

Example.— T^t yoke has to carry 4 N„ lines, in this case
8,200,000, at full-load. Suppose it to be of cast-iron with an ap-
propriate density not exceeding 40,000 lines per square inch. Then
about 205 or 210 square inches will be needed. Being of cast-iron, a

oy Google

154 Dynamo Design.

broad semi-oval section, flat in the inner face, will be appropriate, and
if a breadth of 30 inches with a thickness of g inches at the middle
be adopted, the over-all diameter of the magnet frame will be about
loS inches.

(14) All is^now provisionally ready for the commencement
of the real calculation of the machine. A drawing should be
sketched out to scale, using the trial-values adopted so far.
This drawing will enable the designer to judge of the ulti-
mate dimensions and appearance of the machine. From it
a complete set of calculations must now be made (on the
principles laid down in the preceding chapters) for (l) ex-
citation, (2) heating, (3) sparking, and (4) efficiency, as is
done in the case of the two machines discussed below. On
examining the results of such calculations it is then easy to
see in what manner it would be desirable to alter the design
in order to fulfil more completely the specified conditions.
Finally, as the outcome of such considerations, other designs
ought then to be made, and worked through, differing in
various ways from the first one, but fulfilling the terms of the
specification. For example, if the sparking-criteria are only
barely fulfilled, it might be worth while to recalculate after
slightly increasing the diameter of the armature : or if they
are amply fulfilled, the diameter might be slightly reduced.
Or if the percentage of the heat-losses in any of the parts —
say the teeth — comes out either higher or lower than the
amount known by experience to be advisable, then the design
might be modified so as to give either a higher or a lower
flux-density, as the case may be, in that part When a few
such variants on the first design have been made it becomes
a simple matter to pick out that design which appears to be
the best all round, cost of materials and cost of manufacture
being the most important final consideration.

Machines intended to be used as over-compounded gene-
rators must be designed a little more liberally than those
designed for same speed and voltage as shunt machines, so
as to allow for the increase of magnetic fiux and additional
excitation losses at full load ; or, what comes to the same
thing, if of equal dimensions, they must for the same speed

izecoy Google

Rules for Design. i55;

be rated as of, say, from S to 7 per cent, lower output ; or, if
rated at the same output, their speed must be increased from
5 to 7 per cent.

Other Procedure in Design. — It is, of course, passible to
follow a different order of procedure in designing. Rothert,
in an excellent paper in the Elektrotechniscke Zeiischrift for
1901, gives the following : — Using for the estimating of the
armature dimensions a constant which differs in different
types of machine, but which, for a given diameter, determines
the length of the core, he then chooses the number of poles
in dependence chiefly on the prescribed speed. His rule for
this is that the frequency (i.e. the product of number of pairs
of poles into revs, per second) shall lie between 17 and 20 for
500-volt machines, or between i8'5 and 22 for those over-
compounded to SSo volts. From this he selects the armature
dimensions so that, taking the pole-span about 72 per cent, of
the pole-pitch, the pole-face shall be approximately a square,
about the middle of which is centred a cylindrical steel pole.
He does not calculate up the efficiency until after the main
dimensions have been settled, as he finds it always to conje
out right, in the case of large machines, if they are only
approximately correctly .designed in other respects. Much
more important he regards the cooling question, which can,
however, be controlled by providing due ventilation. The
two main factors, however, which are of vital influence in
selecting dimensions are the proper magnetic saturation of
the teeth, and the economy of material attained by using high
current-densities in the copper. As to the former, he uses
151,000 lines per square inch (apparent) at full load. As
armature current-density, he takes 1700 to igco amperes per
square inch. Allowing a temperature-rise of 35 deg. C, and
surface speeds of 17CO to 2300 feet per minute, he finds this
to correspond to a waste of about 161 to 2' 13 watts per
square inch of peripheral armature surface. For the stationary
coils of field-magnets he allows a current -density of gco to
lo6o amperes per square inch ; and with a permissible tem-
perature rise of 35° C, a corresponding waste of 0'77 watts
per square inch of cylindrical surface.

izecoy Google

156' Dynamo Design.

Criteria of a good Design, — A well-designed machine must
not spark at any load up to an overload of 50 per cent above
normal. It must not spark (see p. 159) at any load up to an-
overload of 25 per cent., even though the brushes be fixed. It
must not overheat (see Chapter VI.). And it must be neither
too heavy nor too costly in manufacture.

A good criterion is tfie ratio of the flux-density in the gap
to the ampere-conductors per inch of periphery. The former has
usually values approximating to 50,000, while the latter is
usually from 5C0 to 600. This ratio is, therefore, of the order
of magnitude of 80 to 100. It may be briefly called the
stiffness-ratio. If higher, the machine is unnecessarily heavy ;
if lower, it may be prone to spark at high loads.

Another criterion of goodness of commutation is the value
of the stiffness-ratio as compared with the volts per segment of
the commutator. The latter varies (see p. 140) in machines of
diiTerent voltages. In loo-volt machines the voltage per bar
may be taken as about 3. In these, then, the stiffness-ratio
(of 80 or 100) divided by 3 gives the commutation-ratio as
from 27 to 33. In 500-voIt machines, taking voltage per bar
as about 6, gives 13 to 16 as the commutation-iatio. Any
lower values than these should be looked upon with suspicion.

Yet another criterion is to compare the number of ampere-
turns of excitation needed at full load to drive the flux through
the gap and teeth, with the whole number of ampere-con-
ductors (at full-load) that lie under one pole-face. This is a
comparison in effect between the magneto-motive force that
can resist distortion, with the ampere-turns tending to distort
the field, A couple of examples from machines known to
commute well at all loads will suffice : —
, In Parshall's 550 kilowatt generator, p. 209, the number
of ampere-turns spent on gap and teeth is 6600 per pole,
while the distorting ampere-conductors under one pole amount
to 14400, making the ratio of the former to the latter 0*45.

In the Scott and Mountain 150 kilowatt generator, p. 161,
the ampere-turns for gap and teeth amount to 8350, while
the distorting ampere -conductors are 13000, making the ratio
0-64, . . -

izecoy Google

Rules for Design, 157

As to weights, machines with a low peripheral speed
always weigh more than those of equal output with a high
speed ; and those with cast-iron yokes more than those with
steel yokes. The armature weight (apart from the shaft)
ought to bs approximately proportional to the kilowatts if
equal peripheral speeds are attained.

Hobart has considered ' the problem of designing series of
generators of standard patterns to cheapen manufacture. His
designs favour a high surface speed, lai^e commutator, and
high current-density in the armature, from 2340 to 2520
amperes per square inch.

Specific Utilization of Material. — Mavor has introduced*
the conception of the " active belt," meaning by this term the
entire mass of the armature periphery down to the roots of
the teeth, consisting of iron, copper, and insulation. It is in
this active belt that the whole inductive generation takes
place, and on this active belt that the mechanical forces arc
exerted. Mr, Mavor found the number of ergs per second per
cubic centimetre at unit velocity 111 unit field to be about 5.
But the work done per line in moving a current across a field
is simply proportional to the current : so that Mr. Mavor's
figure is a measure of the current -density in the gross section
of the active belt.

We may extend still further this conception of a belt of
.active material, and may consider not only the mean number
of amperes that traverse each square inch of it parallel to
the shaft, but also the mean number of magnetic lines that
traverse each square inch of it radially, and the speed with
which it moves forward tangentially. Let us consider the
number of watts generated per cubic inch of the active belt.
If d be the diameter, / the length of the core-body, and s the
depth of the slot (or length of the tooth), the total volume of
the active belt will be w dls. Hence : —

Watts per cubic inch = —-;-;-.

Now, writing for E the value n Z N/ -i- c 10' ; and re-

' ypurnal Imlitution EUttrical Engineer!, xn\ 170, 1931.
' Ibid., xxxi. J18, 1901.

oy Google

i.5S Dynamo Design.

membermg that ii^v-^ird, where v is the peripheral velocity
in inches per second, we have : —

This we may decompose into three factors, thus : —

Watts per cubic inch = ■ — — ,- • — %-, • — »•
'^ ivcds •adl lo'

These three factors we may severally write : —

7.0. -T- ■JTcds = a = gross current- density per square

inch in, active belt
N^ -r irdl = ^ = gross magnetic density per square

inch in active belt.
I'-T-io* =Y= a quantity proportional to the

peripheral velocity.
If, then, we take out these three factors a, 0, and y for
any particular machine, we have at once a means of com-
parison between its design and that of other machines in
respect of the specific utilization of materials. Some makers
manage to crowd many amperes through the copper : in their
machines a will be large. Other makers contrive to have a
Tiigh average flux-density in the belt : in their machines fi
will be large. Others drive their machines with a high surface
speed, and so increase the specific output of a given quantity
of active material. Owing to the conditions that are neces-
sitated by sparkless commutation a cannot be very high unless
^ is high also, though yS may be high without a being so. And
7 may be high or low, quite independently of a or /3.

The Author has therefore made a detailed examination of
more than fifty modern generators, including the machines
mentioned in this book, to ascertain the values of these three
factors of specific utilization. The values of a for machines
of the type principally dealt with lie mostly between 3C0and
460, a few being outside these limits. The values of ^ lie
mostly between 30,000 and 45,000, the extreme values being
22,000 and 58,000. The values of 7 lie mostly between
0-OCOC04 and 0-000009 > but in a few cases exceed the latter
figure. The watts per cubic inch of active belt run from

izecoy Google

Rules for Design. 159

about 45 to 1 20 ; but in one case go down to 1 5, and another
case, Hobart's 1600 kw. generator, reach 162. Smooth-core
machines are not included in these calculations, nor arc-
lighting machines, nor magnetic machines, nor any small
designs. See the. Table on p. 234.

Sparking Criteria. — A rule much used by designers is
that the flux-density under the backward pole-horn at full load
shall not be reduced below about 13,000 lines per square inch.
Or if X^ denote the ampere-turns per pair of poles required
for the double air-gap, X^ the number of ampere-conductors
under one pole, and Bj the average flux-density in ttie gap at
full load, then

B3 X (Xj. — X^) -r- Xg shall not fall below 13,000.

Mr, Kapp has suggested two other criteria, involving the,
use of arbitrary coefiicients, which are here stated in British
units.

Let Bj be the flux-density (lines per square inch) in the gap,
„ Xj be the number of ampere conductors per inch of armature

periphery,
„ K be the total number of commutator segments,
„ k^ be the number of commutator segments short-circuited

together by any one brush,
„ Y( be an empirical constant,
„ Y3 be a second empirical constant,

„ g be the length across one air-gap (iron to iron), in inches,
„ d be the diameter of the armature, in inches.

Then t\\t first criterion is that

Yi = B3 X -&, -j- Xo ;
where for good results, in

slotted drum armatures, Yi should not come less than 38
.1 ring „ „ „ „ 60

.The second criterion is that

Y, = K^H-^(t -f--^,);
where for good results
with metal brushes, Yj should not fall below the value i - 2
„ carbon „ Yj „ , o*6

izecoy Google

Dynamo Design.

CHAPTER VIII.

EXAMPLES OF DYNAMO DESIGN.

We now proceed to analyse the designs of two machines of
different types, as examples of the foregoing principles and
methods.

Example I. — Shunt-wound Multipolar Machine
WITH Slotted Drum Armature.

Built by Messrs. Ernest Scott and Mountain,

M.P.— 6— 150— 450 — 250 volts— 600 amps.

{Shown in Figs. 55 & 70, and Plate II. For description,

see page 198.)

izecoy Google

Examples of Dynamo Design. 161

The leading dimensions and particulars as obtained from
the drawings are given in the schedule form below.

GENERAt Spec [FlC AT ION.

Full load (kilowatls) . . ..... 150

„ ((ennind volts) 350

„ (amperes) 600 ■

Eevolaltons per minute 450

Peripheral speed (feel per minute) ..... sSjg

Number of poles ........ 6

Nature of load ........ Lighting

Dimensions.
Armature: —

Cote disks, external diameter (inches) .... 33

„ internal „ „ . . . . ;3

Number of slots . . . 1 24.

Depth of slot (inch) 1-625

Width „ ,, 0-4

Fitch of slot at armature Cice (inch) . . . . - 0*840

„ ,, average (inch) ...... o'jqb

Depth of iron in core, under teeth (inches) . . , j • 87:

Gross length of core (inches) . . . . . . ■ i

Iron 9

Diameter of finished armature (inches) .... 33

Number of conductors ....... 496

Arrangement ■ . . ■ . . . . 4 in slot

Slfle of winding ........ parallel

Dimensions of each conductor, bare (inches) . . . o'7 x o'll

11 ,1 <t insulated (inches) . O'73xo*i4

Section of each conductor (square inch) . . . , 0*077

Mean length, one armature turn (inches) . . . . 66

Fiild-Magnits : —

Diameter of bote (inches) ...... 33'6lS

Polar angle (degrees) ....... 43

Turns per pair of poles 360a

Mean length of one tnrn (inches) ..... 43

Diameter of wire, bare (inch) 0*09

Section of wire (square incb) ...... 0*0050

Shunt current (amperes) 4>ia

CommulatoT . —

Diameter (square inches) jl

Number of segments . . . . . . . 248

Active length (inches) 7*5

An inspection of the drawings shows that the magnet cores
of steel are of circular section, bolted on to the yoke, the pole
pieces being in one piece with the magnet cores.

izecoy Google

'162

Dynamo Design.

• The field-frame is cast in two pieces and bolted together.
All the field-bobbins are connected in series. The armature
slots are straight, and of the dimensions given above. There
are two ventilating apertures, each f inch wide, and the core
disks are insulated with varnish, deducting altogether i8 per
cent, from the gross section.

The armature winding has six parallel circuits, and six sets
of brushes at 60° apart.

We will first construct the saturation curve of the machine.

We have : —

E = « X Z X N„-s- 10",
or

E = t^o X 496 xK-~ io« ;
60

E = 0-03000372 Na.

The dispersion coefHcient of this machine is v = i * 17.
Hence we have : —

N„

300

8,100,000

9,480,000

380

7,530,000

8,800,000

260

7,000,000

8,MO,O0O

230

6,180,000

7,230,000

300

5,400,000

6,320.000

From the drawings we obtain : —

Length of m

n magnetic patli in magnet yoke (inches)

„ two magnet cores (inches)

28
>re .. . IS

„ ,, two teeth ,

,, „ Iwo air-gaps (inch)

[ And for the magnetic areas : —

For the yoke (square inches) . . . . . . ■ = 44

, „ nu^el cores (square inches) = 78'5

„ armature body „ ..... =53-0

The polar angle being 43° we have for the number of
teeth under one pole

124 x_43 :

36a

14-7.

itizecy Google

Examples of Dynamo Design.

■63

Owing to the high flux-densities in the teeth the flux will
spread considerably, so we will assume 16 teeth as the number
under 1 pole.

The pitch of the slots at armature face is 0*84 inch, and
hence the width of the tooth = 0-840 — 0-400 = 0'44.

Consequently the area of the teeth under each pole may
be taken as : —

16 X o*44 X 9 = 64 square inches.

The air-gap area will be taken as the mean of the tooth
area and pole-face area, or

140 -f 4 _ JQ2 square inches.

We have now all the data for constructing the saturation
curve and no-load characteristic. It will be sufficient here to
work out two points on the curve, as the method is the same
for them all.

Taking the point E = 260 arid E = 230 as examples, we
obtain the two tabulations given below.

The values of ampere-turns per inch, S, have been taken
from the magnetization curves given in Plate I., though it is
quite probable that the actual brands of cast-steel and of iron
stampings used by the makers had slightly different values.

The value for the air-gap might be obtained by the use of
the gap-coefficient 0'3I33, as explained on p. 28,

E =

230.

N„ = 6

180,000.

N« = 7,23o,oo(

M^Si^.

M.teriia.

ts;-

MsKntlic
Stcdon.

Flui
Dfiuliy.

ViaMofiffOB

Curve.

'=r

Yoke ,

cast Steel

Id.

3 X44

81,300

597

Two magnet 1

cast steel

78-5

9.. 100

3* ■

895

Two Bir-EUps

air

0-625

102

60,700

((perXi"«:l')J

11,887

T--" lurpZg.)

3'z5o

97,000

A™...„ ;> tan 1
core i^slampings )

•5

2x53

58,400

2-7S

Total anipeie-turns per pair of poles

13,495

Dynamo Design.

P.r.of

M,M™a.

Ltislh.

sISiSSf Dai"i!r.

"Cuive."™

Tutiu.

Yoke

ca« steel

2x44

93,200

884

Twomagnetl
cores J

cast steel

78s

104,300

1,680

Two air-gaps

ait

0-625

102

68,600

[fper4',^och)}.'3.43« |

Two teeth

sum'^"ngs} 3-'So

109,000

107

348

Armature

U7..} '^

.XS3

66,000

Total ampeie-turns per pair of poles 16,402

It) the same way we calculate other points of the curve
obtaining : —

When

E = 200, necessary ampere-tums = 11,251
E = 280 „ „ = 20,298

E = 3CO „ „ =25,915

By plotting the curve connecting these five points we obtain
the working part of the saturation curve, as shown in Fig. 56.
We have then for E = 250 volts ;

Necessary ampere-turns at no-load = Xi = 15.300.

We will now proceed to find the necessary ampere-turns at
full-load. These will be greater than those required at no-load
by an amount depending upon : —

1. The value of the full-load lost volts.

2. Amount of armature demagnetization.

3. „ „ distortion.

Now the resistance of the armature including brush-leads
and carbon brushes is o*oo8i ohm brush-to-brush at the
working temperature.

The resistance of the series coils is 0-00083 ohm.

Hence the total resistance of the main current circuit in
the machine is (o-oo8i -I- O'OO083) = 0*00893 ohm.

oy Google

Examples of Dynamo Design. 165

The full-load drop is therefore : —

e= 600 X 0-00893 = 5-3 volts.

Now the terminal voltage of the machine at full-load is 250.
Hence the armature must generate at fuU-load 250 + S'3

= 255-3 volts.

-It

^M^f^

~g — s s

oy Google

1 66' Dynamo Design.

Finding this point on the scale of ordinates of the curve
and projecting it across we find the point X, on the scale of
abscissa:, which corresponds to the ampere-turns required per
pair of poles at full-load if armature reaction were entirely
absent This makes Xj = 15,900.

Now, the number of slots lying between the pole-tips is

360

and in each slot there are four conductors carrying 100 amperes
•at full-load. Hence the demagnetizing turns of the armature
at full-load, and upon the assumption that the brushes are
■moved right under the pole-tips, are

5'8 X 4 X 100 = 2320.

Multiplying this number by v, the necessary compensating
ampere-turns per pair of poles are therefore

2320 X I'I7 = 2715.

Adding, then, these ampere-turns to Xj we find Xg = l8;6r5
as the ampere-turns necessary at full-load, assuming that there
were no drop of pressure due to the diminished permeability
of the teeth at the forward pole-horn, due to distortion of the
flux. But this is not the case. We must therefore allow for
this as explained on p. 129.

For this we have : —

Ampere-turns > „ «L x „+ x 4 X ico
under one pair poles) 360

= 6500.

We set off, therefore, 6500 ampere-turns on each side of the
point Xj upon the scale of abscissa;, and obtain thus the points
A and B, which represent the hindward and forward pole-
horns respectively. If distortion of the main flux were absent,
the latter would be proportional to the area A B C D. But as
this is not so, it is proportional to the smaller area A B L K.
In order to make this area equal to that of the rectangle, we

izecoy Google

Examples of Dynamo Design.

167.

must shift the point F higher up the curve to the position F,
so that

. Area A' B' L' K' = area A B C D.
In this manner we obtain the point X4 as the necessary
ampere-turns at full-load.
Their value is

X( = 19,700.

Comparing Our calculated results with the actual values of
the running machine, we have : —

0.,,.,.

CilcuLiied ValuiL

A„..,V..u..

At no-load , ,
At full-load .

«5.3<»

19.700

14,800

17,800 .

The discrepancy between the values Calculated from the
drawings and the values found by the makers, is probably
due to the quality of iron actually used being better than that
assumed in the calculations.

The full-load excitation is made up as follows :—

Shunt'turns per pair of poles, 3602 carrying 4- 10 amperes.

Series-turns per pair of poles 5 carrying 600 amperes.

Total ampere-turns per pair of poles ; —

3602 X 4' I = 14,800 shunt ampere-tums.
S X 600 = 3,00c series ampere-turns,
.■. Total ampere-turns =. 17,800

I. Copper Loss : —
Armature.

Losses.

= 600 X 600 X 0-004
: 1440 watts.

Series Coils.

w„= 600 X 600 X 0-00083 ;

= 29S watts ;

.-. 7V, = 1738 „

oy Google

i68' Dynamo Design.

2. Iron Loss. — The number of cubic inches of iron in the
teeth is : —

I ■625 X 0*396 X 9 X 124 = 720 cubic inches.
The frequency of reversal of magnetism is : —

3 X 450 -i- 60 = 22*5 cycles per second.

At full-load the flux-density is 130,003 lines per square
iiich. Reference to the curve (page 13) shows that at 130,000
lines per square inch, and taking rj (the hysteretic constant>
as 0*003, which is a probable value for armature stampings,
the hysteresis loss will be about 0*038 watts per cubic inch
of iron at 1 cycle per second. Hence the hysteresis loss in
the teeth is : —

720 X 0*038 X 22*6 = 620 watts.

Similarly, on reference to the curve of eddy-current losses
on page 15, we find that, at this flux-density; the eddy-current
loss for I cubic inch of iron at i cycle per second for plates
of 20 mils thickness is 0"0C024 watts. Now the stampings of
this armature are only 18 mils thick, therefore the eddy-current
loss in the teeth is : —

720 X 0*00024 X (22 * 5)' X J° .°'-va = 70 watts ;

making a total iron-loss in the teeth of 690 watts.

The number of cubic inches of iron in the armature core-
body is : —

{(29-75)' XO-78X9} - {(18)' X 0-78 X 9}
= 39^5 cubic incher.

At full-load the flux-densily is about 69,000 lines per
square inch. From the curves, taking as before the hysteretic
constant as 17 = 0-003, we find that 0'oi3 watts are lost per
cubic inch of iron at i cycle per second. Hence, the hysteresis
loss in the core-body is ; —

3905 X 0*013 X 22*6 = 1140 watts.

For the eddy-current loss, we find that at this flux-density,
and at i cycle per second, and for stampings 20 mils thick.

::,y Google

Examples of Dynamo Design. 169-

o-oocx>8 watts are lost, so that the eddy-current loss in the
core-body will be : —

3905 X o-coooS X (22-5)' X ^°'°?^|* = 123 waits,

(0*020)^

making the total iron-loss in the core-body 1268 watts.

Adding together the losses in the teeth and core-body, we
have as the total iron loss of the machine : —
""i = 1958 watts.

3. Excitation Loss. — The total resistance of the shunt
winding is 6r ohms, therefore the current through the shunt-
coils at full-load is —

^._ = 4'i amperes,

«'j:= 4-1 X 250
= 1025 watts.

4, Commutator Losses.— Upon the commutator are pressed
twenty-four brushes (4 per pole), and the area of each brush
is about 1*75 square inches, making a total area of brush
contact of 42 square inches. The 6co amperes go in through
21 square inches and come out through the other 21 inches.

Assuming the contact resistance to be 0*03 ohm per

square inch, we have for the CjR loss at the commutator : —

2 X 600 X ^oo X o ■ 03 -r 2 1

= 1032 watts.

The peripheral speed of the commutator is 2275 feet per

minute.

Assuming brush pressure to be i ■ 5 pounds per square inch
and the friction coefficient to be 0'3, we have for the friction
of the commutator

I '5 XO'3 X 42 X 2275 X 746

33.COD

= 1000 watts.

Hence the full-load loss by brush resistance and friction is

t Wi = 2032 watts. ^

oy Google

170 Dynzmo Desipt,

5. Friction and Ventilation Losses. — Owing to this being a
rope-driven machine the friction losses will come out rather
high, say 3 per cent of the output, that is
w. = 4500 watts.

The total full-load loss is obtained by taking the sum of
the separate losses, that is

ai = w^ + 'a/. + w_^+ m^ + w^ ;
- ■ - ■*-^' I7f8'-P i9s'8 + 1025 + 2032 + 4500;"
or, in total, 11,253 watts; say 11-25 kilowatts.
Therefore the full-load efficiency is : —

■n = ; = 93 per cent.

(J50+ 11-25) ^^^

Probable Heating, — («) Armature. From the drawings the
heat-radiating surface of the armature is found to be about
2500 square inches.

The peripheral speed is

33X3-I4-6X4S° = ,835 f„, per minute.

The watts lost in the armature at full-load are ; —
Iron loss ..... 1958
Copper loss ..... 1440

making the watts wasted per square inch of radiating surface
---"- =1-36 watts per square inch.

A reference to the lower curve of Fig. 22 shows that the
temperature-rise per watt per square inch, for a peripheral
speed of 2835 feet per minute, is for large well-ventilated
armatures 25 deg. C.

Hence we have

(9a = 25 X 1-36= 34 deg. C.

ifi) Field-magnet system. Here the radiating surface is
about 342 square inches per bobbin, or a total of 2052 square
inches. The watts lost in the shunt coil have been already

oy Google

Examples of Pynamo Design. 171

estimated at 1025, and the watts lost in heat in the scries
winding are

(600 X 600 X 0-00083) = 298 watts.

Hence the probable rise of temperature of the field coil is

(1025 + 298) X 75

2052

e^ = 49 deg. C.

g ^ 46-5 X 2032 - ^

' 21 X 3-14 X 9'S (i T-o'ooos X 2275)
e, = 35-8 deg. C.
Sparking.— '^e have already found the value of the cross-
magnetizing ampere-turns, namely,

Xi = 6500
The ampere-turns required for the gap and teeth at full-load
are about 16,700, the flux-density In the former being 65,000.
Hence the flux-density under the entrant pole-horn is
approximately

16.700 — 6?oo ,. . .'

65,000 X — '—-2 ^— = 39,700 lines per square inch,

16,700

which is amply sufficient for commutation fsee p. 1 59),

Applying the sparking criteria described on p. 159 and

taking the formulie there given, we find the present data are

Bj = 65,000

X. = 570

K = 248
k^ =2
£■ =0-31

'^ =33
whence Yi = 65,000 x 2 -r 570 = 228.

Y, = 248 X 0-31 -r 33(1 + 2) = 078
Thus Yi and Yi are both above the minimum values pre-
scribed on p. 1 59, and we may assume that the machine will
not spark. ' ' . ■' -

oy Google

X72 Dynamo Design.

Fig. 57 gives the test-curves of the performance of this
machine running on the testing-bed of the factory.

I NEOAi^Nes PER p4le

Fig. 57.— Factory Tests op Scott and Mohntain 6-Pole Generator.

izecoy Google

Examples of Dynamo Design. 1 73

^jr.w«^^.//.— Over-compounded Multipolar Traction

Generator with Slotted Drum Armature.

Built by The Walker Manufacturing Company.

M P — 10 — 440 — 85 — 550 volts — 800 amperes.

The leading dimensions and data given below and in
Figs. 58, 59 and 60, have been kindly placed at the disposal
of the Author by Mr. S. H. Short, formerly the company's
chief engineer.

general cpe
Full-load kilowatts .

ctjicaiion.

440

terminal volts

S50

amperes

800

No-load terminal volts

500

Revolutions per minute

Peripheral speed, feet pet

minute .

2000

Number of poles

Nature of load .

Traction

Dimensions (in

itich units).

llldLUL C^—

Core-disks, external diameter

„ internal ,

Number of slots.

464

Depth of slot

1-75

Width „

0-3

Pitch ,. (average)

0-6

Depth of iron in core .

9-25

Gross length of core .

18-5

Iron „ „

13-8

Total number face conductors

1856

Conductors per slot .

Style of winding

Parallel

Dimensions of conductor, bare o*o6

XO-7I

» It n

insulatedo-o8

XO-7S

Section of conductor .

0*0426

Mean length, one turn

103

D, Google

Dynamo Design.

Field-magnets.

Diameter of bore

Polar angle ....

Turns per pair of poles, shunt

2200

„ „ „ series

.19

Mean length, one shunt turn

„ „ „ series „

Diameter shunt conductor, bare .

0-162"

„ „ „ insulatej

0-185

Section (square inches)

0'0206

Dimensions series conductor.

bare . . . . ^oio

XO-28

Dimensions series conductor.

insulated . . . 5 of o

X 0-32

Section of series conductor .

X 0-21

Shunt current at no-load

IO-7

„ full-load .

11-8

Commutator.

Diameter . . . . . ^ ■ 70

Useful length 8-5

Number of segments .... 928

Bars per brush . , . , . . 3*3

Brushes per pole .,..". . 3

Size of brushes . ■ , 3 X r.^ X 2J

Area at Commutator face , . . i -95

In this machine, which is a representative type of tram-
way generator, the wrought-iron magnet-cores are cast in
with the magnet-yoke, and are of square section. The pole-
horns are secured to the poles after the field bobbins hav.i
been slipped on. The series winding is not wound over the
shunt winding, but in a separate compartment of the bobbin,
as may be seen from the section of the latter given in Fig. 59.
The armature slots are quite straight with slightly rounded
, and of width equal to the average width of tooth.

izecoy Google

E.a«pUsofDynan.oD.i^-

175

,1,1.0, Google

176

Dynamo Design.

There are four ventilating ducts in the armature, each 0"S625
inch wide, while the paper insulation between the core-disks
deducts 1 5 per cent, from the gross section of the core. The
ratio of nett length to gross length of armature iron is thus
0"74S. There are ten parallel circuits in the armature, and
ten sets of carbon brushes at 36 degrees apart around the
commutator.

The design of this machine may now be analysed in pre-

Jttf

sf j;«

!•

...*<-.- ^

, ♦r....,

raV

**W—

cisely the same manner as in the last case. The first thing
to do is to construct the saturation curve and no-load cha-
racteristic. We have

E = ^ X 1856 X N„ X io-»;

O'coco263 X N^-

The dispersion coefficient of this machine has been deter-
mined experimentally by the makers, and found to be 1*13
and nearly independent of the load. Hence

N.,

460

17,500,000

19,750-000

500

19,000,000

21,500,000

S20

19.750,000

22,300,000

560

21,300,000

24,900,000

S8o

22,000,000

24,900,000

600

22,800,000

25,800,000

iLCD, Google

932

= 233

= 256

9-25

■3

■8

= 128

Examples 0/ Dynamo Design. 177

From Fig. 58 and the data already given above, we
obtain

Length of mean magnetic path in —

Magnet yoke

Two magnet cores

Armature core

Two teeth

Two air-gaps

And for the magnetic an
For the yoke

„ magnet cores
„ armature core

The actual number of teeth under one pole is

464 X ^^^ = 34- 8.

As the air-gap is on the whole lai^e compared with the
diameter of the armature, and as the teeth are worked at
high flux-densities, the flux will spread considerably, and for
this reason we will take the number of teeth transmitting the
flux as being 37. The average pitch of the slots is 0'6, and
their width o ■ 3 inch. Hence the area of the teeth is

37 X (0'6 — o*3) X 13-8 = 154 square inches.

Also, owing to the high densities in the teeth, and the
rounded comers of the latter, the area of the air-gap will be
very nearly the same as that of the pole-face. This latter is

This figure, would adequately represent the air-gap area if
there were no ventilating ducts in the armature core. Re-
ducing the polar area obtained above in the proportion of the
length without ducts to the length of pole-face, we obtain

Area of air-gap = ^'' - — ? = 348 square inches,

10-5

DigitizecoyGOOgle

178

Dynamo Design.

The two tabulations below give the working out for two
points of the saturation curve, namely, when E = 520 and
E = 560. The tooth flux-densities given are the true values,
obtained from the apparent values by means of curve A A of
Fig. 7, The values of ampere-turns per inch (S) have been
obtained from the magnetization curves of Plate I.

E = 520 ; N„ =

19,750,0

oo;N„ = ..,3oo

000.

PutofHmchuut.

Miluial.

UiEDctic

i
MugiKiic \ Flmi-

233 1 47.600

Yoke .

Cast iron

5°

4600

Two masnetl

Wrought iron

256 1 87,000

28' 5

.000

Two ur-g&ps

Air

348 56.700

Co' 3-33]

iS.ioo

Two teeth .

Irot> stampings

3-S

154 i23.a»

700

2450

Armature core

118 , 77,000

264

Totals

mpere-iums per pair of poles

26,414

E = 560 ; Na =

21,300,

X)0; N„

= 24,000.000.

PoctoTMachlDt

Mauriil.

Mugueric

Magnttfc
Ana.

Flm- 1 ^»!™ 0* '

':sr

Voke . .

Cast iron

J33

51,400 115

5750

Two magnetl

Wrought iron

256

93,600 45

•575

Two air-gaps

Air

348

61,100 ,[0-3133]

19,200

Two teeth .

Iron stampings

3'5

i '"

131,000 1400

4900

Armature core

83,000 u

365

Total

ampere- tu

ns per pair of poles

31,790

In the same way we calculate other points of the curve,
obtaining

When E = 460, necessary ampere-turns = 20,710
= 500 „ „

= 600

= 24,040
= 34.960
= 39.100

oy Google

Examples of Dynamo Design. 1 79

By plotting the curve connecting these six points we
obtain the working part of the no-load characteristic, as
shown in Fig. 61. " We have then

Necessary ampere-turns at no-load = X, = 24,250.
We will now proceed to find the necessary ampere-turns
at full-load. These will be greater than those required at no-
load by an amount depending upon —

(i) Amount of over-compounding asked for.

(2) The value of the full-load lost volts.

(3) Amount of armature demagnetization.

(4) „ „ distortion.

Now, there are 1856 conductors in series round the whole

armature, that is, 928 turns. From the data already given

(p. 42), we thus have for 40° C.

Q-2 X Q28 X 103 ,

r = -2 2 — -: =i = 1 -72 ohms

12 X 0-0426 X 10"

and the resistance of the armature at this temperature is thus

r = ^; vi =■ 0*0I72 ohms.

' 4 X (sr

There are 19 turns of series conductors per pair of poles,
the mean length of one turn being (78 -f- 12) =6'5 feet.
Hence the total resistance of the series winding at 40° C. is
_ 9'2 X &•$ X 19 X 5
" I og X 1,000,000
= 0-00542 ohm.
Hence the total resistance of the main current circuit in
the machine is (00172 -1- 0-0054) = 0-0226 ohm. The full-
load drop is therefore

e = 800 X 0*0226 = iS' I volts.

Now, the terminal pressure of the machine at full load is
to be 550, corresponding to an over-compounding of 10 per
cent. Hence the armature must generate at full load (530 +
i8*i) = 568 volts. Finding this point on the scale of ordi-
nates of the curve, and projecting it across, we find the point

N 2

izecoy Google

Dynamo Design.

NOUCEO ELeCTROMOTIVE-FORCe - N„ - 0-00O0263

Fig. 6i. — Saturation Curve of Waikhr io-Pole Genekator.

.oogic

Examples of Dynamo Design. i8i

Xj on the scale of abscissas, which corresponds to the ampere-
turns required per pair of poles at full load if armature
reaction were entirely absent.

Now, the number of slots lying between the pole-tips is

and in each slot there are four conductors carrying a little over
80 amperes at full load. Hence the demagnetizing turns of
the armature at full load, and upon the assumption that the
brushes are moved right under the pole-horns, are
1 1 ■ 5 X 4 X So = 3680.

The compensating ampere-turns per pair of poles are
therefore

3680 X 1*13 = 4160.

Adding then these ampere-turns to Xj we find Xj as the
ampere turns necessary at full load, assuming that there is no
drop of pressure due to the diminished permeability of the
tetth at the forward pole-horn, due to distortion of the flux.
But this will not be the case, owing to the high flux-densities
in the armature teeth. We must therefore allow for this as
explained on p. 129. For t^is we have : —

Ampere-tarns under j. = 2J ,< 464 ,< 4 x So
one pair poles j 360
= 11,100.

We set off", therefore 11,100 ampere-turns from each side of
the point X3 upon the scale of abscissse, and obtain thus the
points A and B, which represent the backward and forward
pole-horns respectively. If distortion of the main flux were
absent, the latter would be proportional to the area A B C D.
But as this is not the case, it is proportional to the smaller area
AB LK. In order to make this area equal to that of the
rectangle, we must shift the point F higher up the curve to
the position F', so that

area A' B' L' K' = area A B C D.

oy Google

l82 Dynamo Design.

In this manner we obtain the point X^ as the necessary
ampere-turns at full load. Their value is
X, = 38,500.
Comparing our results with the actual values of the running
machine we have —

Ooipul.

1 Calcul.led ViJuH.

Ac,a.lV.l.«.

At no-load

1 24.250

23,600

At full load

38.500

4I,I0Q

Showing that the calculated value is about 2\ per cent, too
high at no load, and &\ per cent, too low at full load, which is
good enough. Had the magnetic properties of the iron used
for this machine been definitely known, a somewhat better
result might have been obtained.

Calculation of Full-Load Efficiency. —

(1) Copper loss. This is

w^ = Soo X 800 = 0-0226
w^ = 14,500.

(2) Iroft loss. The weight of iron in the teeth is

175 X 0-3 X 13-8 X 464 X 028
= 944 pounds.

The frequency of reversal is

X 5 = 7 ' I periods per second.

They are worked at a flux-density of about 132,000 lines
per square inch at full load. From the curves of Fig. 2, p. 10,
we see that at a flux-density of 80,000 lines per square inch
and a frequency of 30 <" the hysteresis watts per pound are
about 2' I, and the eddy loss o-S watt per pound. Therefore
the hysteresis loss in the teeth is

944_x_2-i_x 7^ X (1320005^*
30 X {80000)" *
= 1040 watts.

oy Google

Examples of Dynamo Design. 183

And the eddy-current loss in the teeth is

944 X o-8x (7-i)' X (132,000)'

(30)^ X (80000)^

= 115 watts,

making a total iron loss in the teeth at full load of 1155 watts.
The weight of the armature core is

(?5jJ:^X3I4l6x I28)xo-28

= 8750 pounds.

At full load it is worked at a flux-density of about 85,000
lines per square inch. From the curves of iron-loss we see
that at 30 periods per second and at this flux-density the
hysteresis loss is about 2-3 watts per pound, and the eddy-
current loss 0*9 watt per pound. Hence the hysteresis loss
in the core is

8750 X 2 '3 X ^~-- = 4760 watts.

And the eddy-loss is

8570 X 0*9 X y--,a = 440 watts.

So that the total iron-loss in the core is 5220 watts. Adding
this to the loss in the teeth we have as the total iron-loss of
the generator at full load

w. = 6375, say 6380 watts.

(3) Exciiadon loss. There are 2200 shunt-turns per pair
of poles. Taking the previously calculated value of full-load
ampere-turns, we have as the shunt current at full load

38,500 - (19 X 800) .„ ,
-J__i^ ^ ^ — i = lO'o amperes.
2200
Hence

w^ = io*6 X 550 = 5840 watts.

(4) Commutator loss. There are altogether 30 carbon
brushes upon the commutator, the section of each at the
commutator face being i ■ 95 square inches. The total area
of contact is thus 58-5 square inches. Assuming the contact

izeciy Google

184 Dynamo Design.

resistance to be o 03 ohms per square inch of contact area
(p. 118) we have as the C* R loss at the commutator

2 X 800 X 800 X - — 3- = 1320 watts.

29-75

. The peripheral speed of the commutator is

70 X 3_.4.6 xiS „ ,560 fe« per minute.

Assuming the brush pressure to be i ■ 5 pounds per square
inch, and that the friction coefficient is o"3, we have as the
friction loss of the commutator

I '5 X 58 '5 X 1560 X o'3 X 746

33.000

= 930 watts.

Hence the total loss by brush resistance and by friction is
ifj = 2250 watts.

(5) Friction and ventilation losses. — Taking these as r per
cent, of the fuU-load output — an ample estimate — we have

af,= 4400 watts.

The total full-load loss is the sum of the five losses above,
or

'^^ = 33.390 watts.
Therefore the full-loud efficiency is

440

-. = 0-c

(440-1-33 -9)
i; = g2'8 per cent.
Heating. —

(«) Armature. From the drawings the heat-radiating
surface of the armature is found to be about ii,000 square
inches. The peripheral speed is

90 X 3- I4I6J< 85 _ ,^ f^^^ „i„„„,
12 *^

Digiiii.,, Google

Examples of Dynamo Design. 185

The watts lost in the armature at full load are
Iron-loss . , 6380
Copper-loss . . 800 X 800 X 0-0172 = 11,000

making the watts wasted per square inch of radiating surface

-— — = I "58 watts per square inch,

A reference to the lower curve of Fig. 22 shows that the
temperature-rise per watt per square inch for a peripheral
speed of 2000 feet per minute is 30° C. Hence we have

^« = 30 X 1 • 58 = 4; • 5 deg. Centig.

{p) Field-magnet System. Here the radiating surface is
about 2000 square inches per bobbin, or a total of 20,000
square inches. The watts lost in the shunt coils have been
already estimated at 5840, and the watts lost in heating the
series winding are

(800 X 800 X o'oos4) = 3460 watts.

Hence the probable rise of temperature of the field-
coils is

(5840+3 460) X ;5
20,000

^« = 3S deg- Centig.
{c) Commutator. For the probable heating of this part
of the machine we have (p. 120}

n 46' 5 X 2250

' 70 X 3'i4 X 8-5 (I + o'ooo5 X 1560)
B^ = 17-5 d^. Centig.

Sparking. — We have already found the value of the cross-
magnetizing ampere-turns, namely,

The ampere-turns required for the gap and teeth at full-
load are about 25,000, the flux-density in the former being

oy Google

1 86 Dynamo Design.

about 61,000. Hence the flux-density under the entrant

pole-horn is approximately

, 2C,ooo — 11,000

oi.ooo X -^- — — ^ —
25,000

= 34,000 lines per square inch,

which is a sufficient value for sparkless reversal.

Applying the tv/o criteria of sparking (p. 159), we have
the following data : —

B3 = 19,000 lines per square inch ;
d = go inches diameter of core ;
K = 928 total segments of commutator ;
^2 = 3 ' 3 number of segments of commutator short-
circuited at once ;
g = O'S inches across gap ;

Xq = 1856 X 80 -i- 90 7r = 525 ampere-conductors per
inch at full-load ;
whence we get :

Yi = 19,000 X 3 ■ 3 -^ 525 = 1 19 ;

Ya = 928 X O'S -7-90 (i + 3'3> = '■2-
Now for a sparkless result (see p. 159) in this class of
machine the conditions are that Yi should not be /ess than
38, nor Ya less than 1 ■ 2. From both points of view, there-
fore, the criterion is satisfied by the design. As a matter of
fact, the machine runs quite sparklessly with adjusted brushes ;
and even with_;fxtf!^ brushes runs nearly sparklessly at all loads
up to 25 per cent overload.

We may now proceed to describe a number of modern
designs by various makers who have kindly furnished data
and drawings to the Author.

Oerlikon Go's Dynamos. — For many years past the
Oerlikon Machine Works near Zurich have produced excel-
lent machines. Till 1892 the chief designer was Mr, C. E.
L. Brown. After that date Mr. Kolben and, later, Dr. Behn
Eschenbui^ have been amongst those mainly responsible for
the types produced. With the Oerlikon works originated the
multipolar type of generator of which Fig. 62 is an example,

■oogle

Examples of Dynamo Design. 187

a type which since 1890 has been extensively followed in the
United States as well as In Great Britain,

Plate III. shows an Oerlikon MP 4 — 265 — 370 ma-
chine. The general aspect is given in Fig. 62. Of this
pattern a 4SO-volt generator is at work in the Central London
Railway, and a 5SO-volt one at Zurich. Both machines are

Fig. 63. — Typical 4-P0LE Generator of the Oerlikon Works.

identical in all respects except as to the number of slots and
conductors, corresponding to the different voltages.

The yoke is of cast steel, the diameter over yoke being
75 inches, and the length of yoke parallel to shaft is \T'i
inches. The four field-coils each have, in the 550-voIt
machine 3200 turns, and in the 450-volt machine 2600 turns,
the diameter of the shunt-wire in the former case being
0-079 inch, and in the latter o'oS? inch. The higher voltage

1 88 Dynamo Design.

machine, that is to say, the Zurich generator, is compounded,
there being 3^ turns of strip copper conductor on each pole,
this conductor measuring o'i38 inch by 6'7S inches insu-
lated. The pole-pitch is 31-5 inches, the pole-pieces being
rectangular, length parallel to shaft 19' 3 inches, and pole-arc
23-6 inches. The length of air-gap in both machines is
0"49 inch. The diameter of armature is 39^4 inches, length
over conductors 26-8 inches, and length between core-heads
19-7 inches ; there being one ventilating duct in the armature
o'gS inch wide. ■

The machine for Zurich has 243 slots and 480 conductors
each 0'ii4 inch by 0"925 inch, the slots being o'2s6 inch
wide and I "02 inch deep; and in the machine for London
there are 208 slots and 416 conductors each o-i38 inch by
I '04 inch, the slots being 0-335 inch wide and I* 18 inch in
depth. The winding in both machines is a four-circuit doubly
re-entrant winding (symbol /q) ) with four sets of brushes
and four parallel paths through the armature. The commu-
tator is 22 ■ 8 inches in diameter, there being 240 segments in
the Zurich machine and 208 in the London machine, the length
of segment being about 9 inches.

Plate IV. shows an Oerlikon traction generator M P 12 —
500—100 — 550 volts— 900 amperes. A general view of the
machine is afforded by Fig. 63.

This machine, of which two were constructed for the Basel
tramways, was required to fulfil somewhat unusual conditions
which were specified as follows: —

When taking the undermentioned amounts of power the
electric output of a generator shall be as follows r —

Horse-power (metrii:) . . .130 300 500 750'

Kilowatts outi>ut .... 778 zo6 347 510

Volis at terminals . . . 550 550 550 550

Revolutions per minute , . . 100 100 100 96

[This allows for a 4 per cent, drop in the engine-speed at
- top load.]

Ths generators must be able to develop an output of 347
kilowatts for a continuous run of 18 hours without the tem-
perature rise in any part exceeding 35° C. They must be

:,3,t,zec.yGOOgk'

Examples of Dynamo Design. 189

able to endure an exceptional overload up to 520 kilowatts
for two hours, and a temporary one up 10 675 kilowatts.

izecoy Google

I go Dynamo Design.

Mechanically they must be able to stand a casual doubling
of the speed, and electrically must stand a test of 2000 volts
between winding and frame. At a constant speed and with
a fijied position of the handle of the exciter rheostat, the
voltage shall rise from 550 to 588 volts, or shall fall to 512
volts when the load of the machine being at first 250 horse-
power, shall be respectively reduced to zero or raised to
500 horse-power. If the speed of the steam engine is raised
or lowered 3 per cent the resulting change of voltage shall
not be greater than 12 per cent Each generator is a pure
shunt machine-

This generator, which represents the normal type " G 120 "
of the Oerlikon Company, has the following principal dimen-
sions : — The diameter of the armature is g8j^ inches, the length
between core-heads 14J inches, and there is a single ventilat-
ing duct about 0*8 inch wide. There are 1326 conductors
each 0"472 by]|o*I38 inch, two such being placed in each of
663 slots ; each slot being 0-236 inch wide and I'iSinchin
depth. Th^ gap-space is 10 millimetres or 0'3g37 inch.
The core-segments are mounted on a cast-iron spider. The
winding is a series- parallel wave with six circuits from brush
to brush, the winding-step being ^-i =yi= III, This reduces
the number of conductors to half that which would have been
necessitated had a parallel winding (i2-pole, i2-circuit) been
adopted. The commutator is composed of 663 segments of
hard-drawn copper built up upon a cast-iron ring, and the
commutator risers connecting the segments with the winding
are of iron. The commutator is about 71 inches in diameter
and has an active length of 6 inches. There are 12 ranges
of carbon brushes with 6 brushes in each range, mounted on
a bronze support. The yoke of cast steel, bored on its inner
face, is cast in two parts. The 12 pole-cores are cylindrical,
of cast steel, with a diameter of I3*82S inches. Their basal
faces are turned off" to fit the bored face of the yoke, and
each is secured with two screws. The shunt-coils on the
bobbins consist each of 950 turns of a wire o'i4i inch in
thickness. The steel yoke is stiffened by a single rib of
girder section. It will be noted that the pole-cores are in
this machine relatively short. The whole magnet-frame

.oogle

Examples of Dynamo Design. 191

stands on two feet at the sides upon two cast-iron foot-steps
which are secured into the concrete foundation. The front

bearing is screwed down to a separate foot-step. The design
of these bearings is separately shown in Fig, 64. They

192 Dynamo Design,

are provided with oil-rings for automatic lubrication. The
weight of the magnet-frame and pole-cores is about 9I tons,
with about ij tons of copper in the twelve magnet-bobbins.

i is

The armature weighs about 1 1 tons ; there being about
4-7 tons of iron stampings, 1320 lb. of copper conductors
and 880 lb. of commutator segments.

Tests made on the completed machines show the following
results. Shunt winding resistance 38 ohms, armature resist-

oy Google

Examples of Dynamo Design.

■3

D, Google

194 Dynamo Design.

ance (brush to brush) 0'O2 ohm. Efficiency at all loads from
350 to 500 kilowatts about 94 per cent The temperature-
rise after 12 hours at full>load wa^ about 25° C. At all loads,
and even with sudden changes of 300 to jooo amperes, and
with fixed position of brushes, the machines were reported
to show no sparks at the commutator. Fig. 65 gives a
graph of these tests and shows the no-load characteristic of
the machine.

Another recent Oerlikon machine is shown in Fig. 66,
which is a lighting generator supplied to Bordeaux. This is
M P 10 — 165^110 — 280 volts — 590 amperes. It has 440
slots with six conductors in each slot, the coils being former-
wound with three conductors in the upper and three in the
lower half of the coil. The commutator has 440 segments.
The coils are joined up as a lap-winding, the end of one to
the beginning of the next, and each junction is united by an
inverted butterfly evolute riser to two segments of the com-
mutator situated 88 segments apart (corresponding to the
double pole-pitch), thus tending to equalize the currents to
be collected at the brushes. The dimensions of the slot are
I '497 by 0-295 inch. The magnets are shunt-wound, with
all ten coils in series ; each coil having 972 turns of a wire
O' 131 inch in diameter covered to a diameter of o* 150 inch.
The no-load flux is 5 ' 8 megalines. The efficiency is 90 per
cent, from half-load to full-load. Steinmetz coefficient 2 ■ 24,
Ampere-conductors per inch, 660.

A lai^e electrolytic generator, furnished by the Oerlikon
Company to the Aluminium works at Rheinfelden, is shown
in Fig. 67.

In this machine, M P 32 — 560 — 55 — 80 volts — 7000 am-
peres, the large number of poles is necessitated by the very
large current output and slow speed. The pole-cores are cast
solid with the yoke of cast steel, no pole-shoes being used. The
Steinmetz coefficient in this machine works out to 5*1, the
figure being high as the result of the low speed. The arma-
ture is 177 inches, or 14 feet 9 inches in diameter, the length
between core-heads being 16-2 inches, the ratio of diameter
of armature to length being very large. The armature is

izecoy Google

Examples of Dynamo Design.

O 2

izecy Google

196 Dynamo Design.

parallel-wound with 32 parallel circuits, so that the current in
one. conductor is 225 amperes ; the end connexions at both
ends of the armature are made of evolutes terminating in
copper segments, held exactly as the segments in an ordinary
commutator by end clamping plates.

This construction enables a set of equalizing conductors
to be added at the commutator end of the armature, as shown
in the figure. There are 544 slots in the armature, there being
two conductors per slot, each conductor having a cross-
sectional area of 0*124 square inch, the slots being 0*885
inch in depth and O'Si inch wide. The commutator is
118 inches in diameter, and has 544 segments, each seg-
ment being 13 inches in length; it is bracketed out from
the armature spider. There are 32 sets of brushes with
12 brushes per set. The field-bobbins are connected in two
parallels of 16 bobbins in series, each bobbin having 175
turns of copper wire 0-374 '"ch in diameter, wound in 7
layers of from 28 to 22 turns per layer. The flux per pole at
no-load is 8-03 megalines.

Fig. 68 shows an Oerlikon generator MP 6 — 285—450 —
90 to 190 volts— 1 500 amperes, supplied to the Volta Electro-
cheimcal Company at Rome.

This machine has cast-steel poles bolted on, the pole-core
and pole- shoe forming a single casting. The poles are
slotted radially with a very large single slot about 4"7 inches
long and i"97 inches wide, this slot being, of course, pro-
vided to prevent distortion of the pole-face flux, this pre-
caution being especially necessary in this machine, owing to
the fact that very heavy currents are carried per unit length
of periphery, the ampere-conductors per inch periphery
working out to 835 at the full-load rated output of 1500
amperes ; which is large for a 6-pole machine. The yoke is
62^ inches in diameter over all ; the armature is 35^ inches
in diameter, and has 187 slots, each slot being 1-44 inches
deep and 0-258 inch wide. There are in each slot two
conductors of 0-093 square inch section, the current through
each conductor being 250 amperes, there being thus joo
amperes to be collected at each set of brushes. The arma-

izecoy Google

Examples of Dynamo Design.

oy Google

198 Dynamo Design.

ture has a lap-drum six-circuit winding, the end" loops of the
former-made conductors being bent down and held in place
by an end-clamping shield. The commutator is 21 inches in
diameter and of very massive construction, the segments
being securely held at both ends and clamped In the manner
plainly shown in the drawing. The field-coils have each 600
turns of wire of 0-189 '"ch diameter bare and 0-204 i"<^h
insulated,

Messrs. Scott and Mountain make a standard line of
generators from a l2-pole 78-inch bj' 13-inch generator of
280 kilowatts and 90 revolutions to a 4-pole 42-kilowatt
machine at 680 revolutions, the larger sizes being, of course,
for direct coupling, and the smaller ones rope-driven. These
machines throughout are characterised by solid mechanical
construction, the large relative size of bearings, in all sizes,
being especially noticeable. The mechanical construction of
the armature is simple, the armature laminations being held
upon a spider by stout bolts. In the larger sizes, the com-
mutator is bracketed out from the arms of the spider, and in
the smaller sizes the commutator is built up on an extension
of the armature hub, the whole being held against a shoulder
on the shaft by a threaded ring kept home by a grub-screw.
Slotted drum armatures and barrel-windings are used through-
out ; the pole-cores are of cast steel, while both cast-iron and
cast-steel yokes are used, the former in the larger sizes. The
armature conductors, instead of being bent round and in one
continuous piece at the back of the armature, are clamped
. together with a copper clip and the whole then soldered, this
construction being considered to give special advantages in
repairing. In the machine having four conductors per slot
(see Plate II.) the conductors are first taped, then a pair of
them are wrapped with manila paper and placed in the slot,
which again is lined with varnished millboard. In this par-
ticular example the total thickness of insulation between
conductor and core is '075 inch. In connecting to the
commutator, the commutator-risers are let into the com-
mutator-bars, and then both soldered and riveted, thus

izecoy Google

Examples of Dynamo Design. 199

making an excellent joint both mechanically and electrically.
The binding-wires are insulated from the armature-core with

one turn of varnished millboard and mica slips. The com-
mutator construction possesses no unusual features. Equal-
izing rings are used in the larger sizes, built up af^inst the

izecoy Google

20O Dynamo Design.

back of the commutator, and held by a cast-steel clamping-
ring. In a particular case of the i2-pole 28-kilowatt gene-
rator (see Fig. 69) equalizing rings, six in number, are used,
the.se rings being built up with the commutator, behind the
commutator-risers, and insulated as shown. The six copper
rings are i inch in depth and \ inch thick, the insulation
between the rings being O'Og inch in thickness and the in-
sulation at the ends J inch thick.

This firm aims at high flux-densities throughout, running
the flux up to B = 140,000 or more at roots of teeth and
over 100,000 in the magnet -cores ; and using also a fairly high
gap-density. The field-bobbin construction in these machines
is a detail worthy of note. The bobbins are made with sheet-
iron cores and thick teak flanges, which have a good appear-
ance, but which would seem to take up a good deal of valuable
space.

Fig. 70 shows a detail drawing of bobbin construction
for the Scott and Mountain 6-pole generator depicted in
Plate II. For insulation over the sheet-iron cores two layers
of varnished canvas and one complete layer of press-spahn
o-o6 inch thick are used. According to the makers, the use
of these sheet-iron cores, thus enabling the winding readily
to communicate its heat to the frame of the machine, permits
the use of very high current- densities in the field-bobbins.
They are thus enabled to use current-densities of over j:ooo

oy Google

Examples of Dynamo Design. 201

amperes per square inch in the field winding, and by thus
shortening the necessary winding space, the over-all dimen-
sions of the machine may be reduced, and consequently the
cost.

FRONT. .Jbin, 1 (itfpa-)h
BACK . , Z (Imver- ) .,
FRONT. . 3 <,tffmT-} „
BACK. . 4-(lnwer) ,.
and, -so fbrth/.
Fig. 71.-

N^l oA-e to 2 (lifyyer) m..
^'H , .,3 itifper) „

.,2/ .

Fig. 71 shows the winding scheme of the 6-pole machine
described, the design of which is analysed at the beginning
of the present chapter, p. i6o.

The compounding conductor is of rectangular strip, wound
edgewise, as shown in the detail drawing. These coils are

, C-'OOgle

202 Dynamo Design.

wound bare, then opened out slightly and taped. Connexion
of one series coil to another is made by flexible copper
couplings bolted on. Carbon brushes are used throughout
without exception in machines by this firm, the current
density in the carbon being about 30 amperes per square
inch in the larger machines and about 15 amperes in the
smallest size. An extract from a very usual form of guaran-
tee for these machines, supplied by the makers, is as follows: —
There shall be no sparking due to variation of load within
the limits of no-load and 25 per cent, overload, the machine
to run continuously with practically no sparking or burning
of the brushes, and without blackening the commutator.
The machine to stand a momentary overload of 50 per cent.
without sparking with fixed brushes, and the armature to
stand an alternating potential of 20CO volts without damage.

Messrs. Brown, Boveri and Co., of Baden (Switzerland) have
constructed many types of machines for continuous currents.
A leading feature of most is the barrel-winding in two super-
posed cylindrical layers, patented in November 1892 by
Mr. C. E. L. Brown.

Fig. 72 shows the normal type of belt-driven machine
with cast-steel yoke, and steel pole-cores. All the larger
sizes have laminated pole-shoes screwed on. In the case of
compound machines the series and shunt-coils are separately
former-wound, the series-coils being nearest the armature.
The spider is of cast iron ; the core-disks insulated from one
another. The binding wires are of bronze. The pole-cores
are each secured by one central screw and a steady-pin.

In the Author's work on Dyttanto-electric Machinery are
given several other examples of the machines of Brown,
Boveri and Co., including a large 8-pole electrolytic generator.

Fig- 73 depicts an interesting machine which departs
from the normal type in one respect. It is a double-current
machine; being furnished not only with the ordinary com-
mutator to yield continuous currents, but also with three slip-
rings, that it may at the same time furnish a three-phase
alternating current The rocker-ring is bracketed out from

oy Google

Examples of Dynamo Design. 203

the yoke, while the brushes for the slip-rings are supported
from the pedestal of the bearing. The chief data of this
machine, which was constructed for the lighting station at
Alloa (Scotland), are as follows :— M P (coat, and 3-phase)—
8 — 194 — 350— 490 volts (or 300 A volts) — 396 amperes. The
armature core-body is 42' I liy I2'4 inches, with 128 slots
1*9 inch deep and 0-49 wide. In each slot are 12 con-
ductors, four-deep, each having a section O'yi^ inch by

Fig. 72.— Brown, Boveri and Co.'s Normal Tvpe (1901).

0*13 inch. The over-all length of the armature windings is
about 25 inches. The gap is 0-355 '"<^h. The magnet-cores
arc 7"9 inches in radial length, and io'2 inches in diameter.
The outside diameter of the yoke is 71 '9 inches. The com-
mutator is 32"4 inches in diameter, and the segments 6*9
inches gross length, there being 384 segments. The magnets
are shunt-wound with 1S20 turns on each bobbin, the wire
being o'o83 inch in diameter. The space-factor in the

M By Google

Dynamo Design.

izecoy Google

Examples of Dynamo Design. 205

slots is 0-315 ; that of the magnet-coils 0-568. The current
density in the armature is 1920 ; in the shunt-coil 962 ; at the
brush contacts 40 amperes per square inch. The flux-density
in the gap is 42,cxx>, and in the teeth 116,000 at no-load.
The no-load excitation is 6600 ampere-turns per pole, of
which the gap and teeth require about 5360. The cross-
magnetizing ampere-turns are about 6600, and the demag-
netizing 3000. Ampere -conductors per inch of periphery 520.

Figs. 74 and 75 depict a specially interesting machine of
Brown, Boveri and Co., for a high voltage. This is M P 4 — 20
— 700 — 1000 volts — 20 amperes.

In this small machine, working at high pressure, great
care is bestowed upon the question of insulation throughout
the design. The chief data of this machine are as follows : —

Outside diameter of yoke 35 inches, length parallel to
shaft II '4 inches, of cast steel. The magnet-cores are
circular in section, having a diameter of T^ inches, and the
cores, and at the same time the pole-pieces, are attached to
the yoke of the machine by a single steel bolt ; the fact that
the seatings both at the yoke and pole-pieces are turned
and thus possess a rounded surface, making this possible.
The armature is 15 inches in diameter, the length between
core-heads being g'Ss inches. There are 59 slots and 1416
conductors ; there being thus 24 conductors per slot, arranged
in the slots in two taped sets of 1 2 conductors each. Round
wire of a section of 0*0037 square inch bare and 0-0070
square inch insulated is used, and the total thickness of in-
sulation between conductors and core amounts to 0*07 inch.
The winding has a two-circuit series-parallel grouping de-
scribed on p. 102 ; and throughout great attention is given
to the insulation of the end turns and connexions. But the
design of the commutator is the most noteworthy feature of
this machine. Owing to the fact that only twenty amperes
have to be collected, the question of insulation was the para-
mount one to be considered. There are 177 segments, or
three per slot ; the end clamping plates of this commutator
are unusually substantial.

Mica 0*035 '"ch thick is used between the segments and ,

M By Google

2o6 Dynamo Design.'

the end insulating rings project far beyond the end of the
segments, and are not turned off" flush, as is usually the case
with machines of lower voltage. On the whole the construc-
tion is very simple, that of the commutator especially so ;
the design being very open throughout and such that there
is little chance of dust or dirt collecting, which might lead

Fig. 74 — High-Voltage Dynamo of Brown, Boveri and Co.

to a breakdown in the insulation. Fig. 75 gives a sectional
view of the armature.

Of late Messrs. Brown, Boveri and Co, have designed special
machines to be coupled direct to Parsons' steam turbines.
The very high speeds have necessitated sundry modifications
in design. The armatures are relatively smaller in diameter
and of greater length, and the field-magnets are of the pattern
devised by Deri ' with his special mode of cross-componnding.

' See EltktroUchnisihe Zeitschrijt, vol, !uiiii, p, 817, September 1902.

izecy Google

Examples of Dynamo Design. 207

These field-magnets are built up of concentric stampings with
closed slots at the inner peripliery, and wound in a manner

oy Google

2o8 Dynamo Design.

resembling the stator of a two-phase induction motor. Owing
to the high peripheral speed carbon brushes cannot be used
at the commutator.

The Genera! Electric Company of Schenectady, with which
is associated the British Thorn son- Houston Company, of
Rugby, has produced- many hundreds of machines for trac-
tion or lighting. So far back as 1893 it exhibited at the
Chicago Exhibition a multipolar generator of 1500 kilowatts,
having 12 poles and running at 75 revolutions per minute,
This machine is described by Messrs. Parshall and Hobart in
their work on Electric Generators (London 1900), in which
they give very full constructional data of this and of three
other machines, viz. : —

M P 6—200 — 135 — 500 volts — 400 amperes ;

M P 10 — 300 — 100 — 125 volts— 2400 amperes ;

M P 6—250 — 320 — goo volts— 455 amperes.
Of the three, the second, which is a lighting machine, is not
a satisfactory design. Judged by modem standards ; while the
first is excellent. Its armature is lap-wound with 1760 con-
ductors in two layers, in 220 slots, barrel-wound. The current-
density is 1760 amperes per square inch in the armature,
6670 in the commutator risers, 800 in the shunt-coil, 770 in
the series-coil, 44- 5 at the brush contact. The flux-densities
at full-load were as follows : — 76,000 lines per square inch in
armature core-body, 121,000 (apparent) in teeth, 45,000 in
the gap, 96,000 in steel pole- core, 70,000 in steel yoke. The
excitation percentages were allocated as follows : —

Ai no-lDad. Ai full-laad.

Annalure core 4-4 5'0

Teeth 73 10-4

Gap ........ 5S'9 63'o

Pol=-co.e 17-3 ai-S

Yoke . 12-! 14-0

Compensation fur demagnetization ..... 244

,, distortion . . . .. 5*0

tooo 143-3

The excitation at no-load required 7630 ampere-turns per
pole ; at full-load 10,990. The heat-loss in the armature was
1*70 watts per square inch of radiating surface, and the

izecoy Google

Examples of Dynamo Design. 209

temperature rise 30° C. by thermometer or 37 by resistance
measurement. The losses in percentage of the nett full-load
output were : — armature iron i '38, armature copper 4*4, com-
mutator and brushes 0'735, excitation i "21, including series-
coils and rheostat. The curves given in Figs. 46 and 47 relate
to this machine.

Other machines of the General Electric Company are de-
scribed in the Author's Dynamo-electric Machinery, including
a 6-pole 400 kilowatt belt-driven machine, and a 6-poIe
150 kilowatt machine designed by Mr, Parshall.

Mr. Parshall has also published ' very complete data of a
slow-speed S5o-kilowatt generator of the General Electric
Company's design, which, though a rather heavy machine for
its output, gave a very satisfactory performance from the
point of view of cool and sparkless running, Fig. 76 shows the
armature in .section. Let us treat this machine as though we
had to design it, adopting the order of procedure of p. 146.
We are to produce a multipolar generator working at a
terminal pressure of 500 volts (at no-load), and of 550 volts
at the full-load of lOOO amperes, the engine speed being 90
revolutions per minute. Obviously it is to be over-com-
pounded. The prescribed efficiency is 94 per cent, at full-
load. As this is a slow-speed machine, the Steinmetz co-
efficient cannot be low ; let us take it at 3 ■ 5, Then, by (2)
on p, 140, 550 X 3-5 = 1925 = d X I, which we must pre-
sently fix. As the full-load is looo amperes, if we would not
attempt to collect more than 200 amperes at any one row of
brushes, we must have at least 10 poles and 10 rows of
brushes {5 positive and 5 negative). With a lo-pole machine
with steel poles, one would expect the armature diameter to
be five or six times the length of the core- body. Then, since
J y. I are to equal 1925, two approximate factors would be
100 and igi. The dimensions actually used are </ = 96 and
/ = 20'S ; so that the Steinmetz coefficient is actually 3* 56.
The peripheral speed is 2263 feet per minute, and the per-
phery 302 inches. This gives 30*2 inches for the pole-pitch

i. Oct. 1900 i and Eleelrician, kIvJ. 670, No». zj.

oy Google

Dynamo

oy Google

Examples of Dynamo Design. 2 1 1

at the armature face. Taking the pole-arc at about 75 per
cent, of the pole-pitch, or 22 inches, the area of pole-face
(which is not quite rectangular, being bevelled at the outer
comers to a polygonal form) will be about 22 X 20 = 440
square inches. If we take 42,000 lines per square inch (at
no-load) as a suitable pole-face density, that would make the
flux from one pole to be N = 18,480,000, or i8*48 megalines.
Now using the formula of (5), p. 148, since n (the revolutions
per second) = i " 5 and E, the no-load voltage, is 500, we get
for the trial value of Z the number of armature conductors : —

500 X IO*'-r(l*5 X 18,480,000) = 1803.

The actual number in the machine is 1800 grouped in 300
slots, 6 conductors in each slot. Testing this by the rule that
it is inadvisable to have more than
600 ampere-conductors per inch of
peripnery, we find 1800 conductors
each carrying lOO amperes (since
there are 10 paths for 1000 amperes)
occupying 302 inches periphery,
making 595 amperes per inch peri-
phery, which is satisfactory. Further,
as M = I J, and there are 5 pairs of
poles, the frequency of magnetization
will be only ■ 7^ cycles per second.
There will be, of course, 900 seg-
ments in the commutator, and as
these ought to be about 0'3 inch
wide, the total periphery of the ^
commutator ought to be about
270 ; it is in fact 272, the diameter being 86 inches, and
the length of the segment about 9 inches. As the arma-
ture periphery is 302 inches, and there are 300 slots, the
tooth-pitch will be I -006 inch. The slot should be about
half this; it is in fact 0*525 inch wide. As 6 conductors
each carrying 100 amperes pass through the slot, and as the
current-density in the copper will be about 1 500 amperes per
square inch, each conductor will need to be about o ■ 065 !>quare

P 2

oyGoogle

212 Dynamo Design.

inch in section, and the total section of copper in any one
slot will be about 0-39 square inch. If the space-factor be
assumed at about 0*4, this will show that the gross section of
the slot must be nearly i square inch. As it is 0-525 inch
wide it must therefore be nearly 2 inches deep. In fact, the
slots are made exactly 2 inches deep, allowing for a wedge at
the top. The copper conductor is o-o64i square inch in
section. The slots being 0-525 inch wide, the gap must not
be much less. It was actually 0*375 '""^h. As this is to be
an over-compounded machine there must be allowed a long
pole-core, say, as a trial value, not less than 40 times the
length of the gap, since the teeth also are long ; or, say,
15 inches. The actual length was 18 inches. The principal
data being thus accounted for, it will now suffice to add the
following data as given by Mr. Parshall.

Nett iron length of armature core-body I4'9.

Internal diameter of armature-core 71,

Yoke (cast steel), internal diameter 138*25.

Yoke, external diameter I49"6.

Yoke, diameter over ribs 1 57 ' 5.

Yoke, length parallel to shaft 24.

Number of equalizing rings 10.

Number of equalizing points per ring 5-

Fitch of winding is over 2g teeth.

Armature-spider has 5 arms, with 1 5 dovetail notches to
receive cores.

Style of winding, lap-wound, barrel-drum.

Average volts per segment of commutator, 6* I.

Breadth of carbon -brushes 1 inch, or 3 segments.

Amperes per square inch brush-contact 40,

Shunt-w ire makes 1 154 turns per bobbin, and consists of
780 turns No. 9 B. and S., and 374 turns No. 10 B. and S.

Voltage-drop at full-load is as follows: — 12*6 volts due
to copper armature resistance (at 60° C.) ; 2*4 volts due to
brush contacts; o-6 due to resistance of the compound
winding ; or in total 19 volts. Hence, to give 550 voJts at
terminals, the internal electromotive-force generated at full-
load must be 569 volts ; which, assuming speed constant.

::,y Google

Examples of Dynamo Design. 213

means that the armature-flux at full-load must rise to
20,360,000 lines per pole. Assuming a dispersion ratio of
I • 125, this makes the values of the flux per pole in the pole-
core 20*8 megalines at no-load (500 volts) and 23*6 mega-
lines at full-load (550 volts).

The excitation is given by Mr. Parshall as follows : —

1 N».Load.

Full.L<«d.

-"fJS:i^"^'

FiSi-LotS''

Core-body .

. ' 59,000

67,000

190

320

Teeth . .

. ' 108.000

119,000

340

900

Gap . . .

. , 4*, 500

48,500

5000

5700

Pole- core

. i 78,000

88.000

880

1530

Yoke . . .

69,000

79,000
Totals . .

640
1 7050

1000

9450

There are 180 conductors on the armature per pole, each
at full-load carrying 100 amperes, making i8,000 armature
ampere-conductors per pole, of which about 20 per cent., or
3600, are demagnetizing, and about 80 per cent., or 14,400, are
cross-magnetizing. Total ampere-turns allowed for on each
magnet-pole at fuU-Ioad at 550 volts 12,350.

The heat-waste in iron in the armature was estimated at
o*88 watts per pound ; hence, as core weighs 12,600 lb., core-
loss was 11,000 watts. Armature resistance, brush to brush,
at 60° C, 0-0125 ohm. Hence CR. loss for 1000 amperes was
12,500 watts. Total armature loss 23,500 watts. Peripheral
radiating surface 12,000 square inches ; therefore i -g/ watts
per square inch. Observed rise of temperature after 8 hours'
run at full-load, by thermometer 26° C, by resistance 38° C.
Excitation losses: total C^R loss per bobbin, at 60° C,
422 watts. External cylindrical radiating surface of i bobbin
1350 square inches; therefore 0-312 watts per square inch.
Observed rise of temperature after 8 hours' run at full-load,
by thermometer on surface of shunt coils 26° C, by resist-
ance 45° C. C'R loss at brush contacts 2400 watts ; in
commutator segments 400 watts. Friction loss at com-

izecy Google

214 Dynamo Design.

mutator 870 watts. Total watts lost in commutator 3670 ;
radiating surface 2400 square inches; therefore r'53 watts
per square inch. Observed rise of temperature after 8 hours'
full-load run, 22° C. This makes the total losses as follows : —
armature 23,500, field-magnets 4220, commutator 3670 ; total
31,390 watts. Hence the efficiency (excluding friction at
bearings) is 550 -h- 581*39 = 0'945 = 94i per cent. As-
suming a permissible temperature-rise (by thermometer) of
30° C, Parshall gives the following handy rules as to the
requisite amounts of radiating surface : —

Parshall's 550 Kilowatt Generator.

Constructional data of a larger General Electric Co.'s
generator, M P 14 — lOOO — loo — 575 volts — 1740 amperes,
are given by Hobart in an article in the Elektrotechnische
Zeitschrift, xxii. p. 650, August 1901, where they are com-
pared with those of kindred machines of equal output by
Rothert, and by Siemens and Halske (Vienna), see p. 232.

The firm of Kolben and Co., of Prag, has made itself
known for the excellent types which Mr. Kolben has produced
during recent years.

Fig. 79 depicts a small 4-pole machine of this firm, of
3 kilowatts output. Though it has four poles, two of them
only are wound, the other two being consequent poles at the

, .., ,Goo;;lc

Examples of Dynamo Design. 215

sides of the magnet-frame. Running at I loo revolutions per
minute, it generates 25 amperes at a pressure of 123 volts at
the terminals. The magnet-frame and cores are of cast steel ;
the bearing supports of cast iron. The external diameter of
the core-disks is g-ii inches; the internal 4 inches. The
length between core-heads is about 4i inches. There are 6^
sots, each 0*788 inch deep and 0-167 inch wide. In each

Fig. yg.^KoLBKN'a 4-P0LE 3 Kilowatt Bynamo,

slot are 6 conductors, making 414 conductors in all, their
diameter being 0-087 inch bare, covered to o* 1 10 inch. The
gap-space is o*ii8 inch. The commutator, of 69 segments
is 4 inches in diameter, 2 inches long, and the mica insulation
is 0-024 inch thick. There are two sets of carbon brushes
with two brushes on each set, of a size allowing I square inch
for 30 amperes. On each pole-core are wound a shunt coil
of 2300 turns of a wire 0-040 inch diameter covered to
0-055 diameter, as shunt, and 28 turns of a series winding
0'173 inches in diameter covered to 0-193 inch. The
efficiency at full-load is 85 per cent.

D,j,l,:ec.y Google

2i6 Dynamo Design^

Plates Vt VI. and VII. show a Kolben traction generator,
M P lo — 250 — 125—550 volts— 454 amperes.

The ten pole-cores are cast in one piece with the yoke, the .
whole being of cast steel, with cast-steel pole-pieces screwed
on. Plate VII. shows in detail the construction of the field-
magnet bobbins and pole-pieces. The pole-pieces are skewed
in order to ensure that the armature conductors in revolving
shall come gradually into the field. This machine is com-
pounded, each pole having 5 J turns of copper strip 5 '32
inches wide and 0^059 inch in thickness wound outside the
shunt-turns. The outside diameter of the yoke is 105 inches,
the maximum radial thickness being 5-9 inches; the pole-
cores are 13 97 inches in diameter. The field is bored to
67 • 25 inches, and the diameter of armature is 66 ■ I inches,
the gap therefore being 0"575 inch long. The armature is
wave-wound, the winding being a series -parallel, having four
circuits in parallel, with ten sets of brushes. There are 437
slots, each 0-256 inch wide and 0*984 inch deep, and two
conductors of o"o652 square inch section, in each slot ;
there being thus in all 874 conductors. The space-factor,
that is to say, the ratio of copper section to slot section, is
o'Sr6.

The end connexions of the armature conductors are made
by joining them together" at separate insulated copper seg-
ments, held round the armature exactly like commutator
segments, this construction being exceptionally good me-
chanically ; the commutator risers are then simply sweated
into cuts in these segments. The commutator has a diameter
of 39* 35 inches ; there are 437 segments, or one segment per
slot. Mica 0-036 inch thick is used for insulation between the
segments. This generator is for direct coupling to engine, a
flange being provided for the purpose on the end of the shaft.
The armature spider is secured firmly to the shaft by means
of steel rings pushed on to a shoulder on the 'spider while hot,
the subsequent contraction effectually gripping the spider to
the shaft.

The winding-scheme of this machine is specially con-
sidered and described on p. 102 above.

izecoy Google

Examples of Dynamo Design.

oy Google

2i8 Dynamo Design.

Fig. So illustrates a special type of generator, namely, a
very slow-speed exciter, destined to be mounted on the end
of the shaft of a large alternator, revolving at only 75 revo-
lutions per minute. This entails peculiar variations in the
construction.

The diameter of the core-disks is 33-8 inches ; the length

Fig. 81.— Tramway G
Manvfa

between core-heads 15*1 inches; and as the output is only
38 kilowatts the value of the Steinmetz coefficient reaches
the abnormal value of 13 '4. The commutator has a dia-
meter only slightly less than that of the armature, the risers
being necessarily very short. The machine is shunt-wound.

Several other machines of Messrs. Kolben and Co. are
described in the Author's larger work.

izecoy Google

Examples of Dynamo .

oy Google

220 Dynamo Design.

The English Electric Manufacturing Company (Dick,
Kerr and Co,),'of Preston, produce a standard type of genera-
tor designed by Mr. S. H. Short, depicted in Fig. 8i. A
sectional view of a 12-pole machine is given in Fig. 82. This
is an iroo kilowatt generator, running at lOO revolutions
per minute. It has a heavy cast-iron yoke ; the laminated
pole-cores being cast in solidly, and a cast-iron pole-shoe

I attached by screw-bolts. Fig. 83 shows the laminated pole-
cores. The pole-shoes, as will be seen from Fig. 84, which
gives a view of a magnet-frame, are in two halves, being
secured in V-notches punched in the laminated pole-cores,
the two halves being clamped together by bolts, a space
being intentionally left between the two halves to assist in
preventing distortion of the pole-face flux. In order to

oy Google

Examples of Dynamo Design. 22 1

secure a well-graded fringe at the pole tips they are notched,
as is well shown in Fig. 84.

The construction of the shunt-bobbins is shown in Fig. 85.
The bobbins have cast-metal end-plates cast in open design,

Fig. 84.— Magnet-Frame of io-pole Machine.

to render the ventilation as good as possible. The shunt and
compounding turns are wound side by side and not in super-
posed layers as is usual. The compounding turns are wound
edgewise, of copper strip.

The armature is a simple lap-winding, each element being

oy Google

Dynamo Design,

Fig. 85. — Field-Magnet Bobblns.

a loop of strip copper in one piece. Fig. 8/ shows the com-
plete armature -core. Equalizing rings, or connexions short cir-
cuiting points of the winding at approximately equal potential

Kic. 86.— Armature-Core Stampings.

izecoy Google

Examples of Dynamo Design. 223

are used to facilitate commutation ; and their arrangement is
plainly shown in the figure, between the commutator risers
and end-plates of the armature.

Fig. 86 shows the core-plates of this machine, and the

Fig. g

process of manufacture, from the blank (numbered i in the
figure) to the complete section (numbered 4). One of the core -
plate separators to keep the laminations apart for the formation,
of ventilating ducts is also shown (numbered 3).

oy Google

2 24 Dynamo Design.

Fig, 88 shows the armature-spider ready machined. It
has six arms, with dove-tail grooves for holding the core-
plates. The commutator seating is plainly shown, as also
the seatings and bolt-holes for securing the commutator to
the spider.

88. — Armature-Spider.

Fig. 89 shows the brush gear ; the brush-rocker is carried
' on a massive cast-iron ring, which is bracketed out from the
yoke, as is shown in Fig. 8r, a worm-wheel being used to
shift the rocker for adjustment of the brushes.

izecoy Google

Examples of Dynamo Design. 225

The firm known as La Compagnie de I'lndustrie Elec-
trique, of Geneva, has done much continuous current work for
lighting and power, as well as for electrolytic purposes. Figs.
90 and 91 depict one of their electrolytic generators designed
by M. Thury, viz.

- ~ — ■"^** — 4000*.

This machine, which is at work at Ch^vres, is of the vertical
type, and is driven by a 1200 horse-power turbine at a speed
varying from 90 to 120 revolutions per minute. The output

Fig, 89.— Bkush-Gbar.

is 4000 amperes at 208 volts. This generator has the yoke
and pole-pieces of the pattern peculiar to the designs of
M, Thury. Each shunt-bobbin is wound with 448 turns of
wire of 33 square millimetres section, and the 12 coils are
connected as usual in series. The armature has 468 slots,
each slot containing one conductor of 120 square millimetres
section ; the end connections being made by butterfly connec-
tors of 140 square millimetres section.

The commutator is 1000 millimetres in diameter, has 234
segments, with two working lengths of 390 millimetres, a steel
ring being shrunk on the outside surface of the commutator,

Digitizecoy Google

2 26 Dynamo Design.

at its middle point, to guard against any excessive centrifugal
strain on the heavy commutator segments ; this ring is, of
course, adequately insulated from the surface of the com-

F[G. 90. — Thury's Electrolytic Generator ; Section.

izecoy Google

Examples 0/ Dytiamo Design. 227

mutator. The 4CMD0 ampere current is collected by twelve
sets of brushes, each set consisting of 24 carbon brushes, the
24 brushes being again subdivided into two sets of twelve
brushes. For other Thury machines see the Author's larger
work.

Fig. 91. — Thury's Electrolytic Generatok : Plan.

A generator by the International Electrical Engineering
Company, of London, is shown in Plate VIII., M P 8 — 450 —
250 — 500 volts — 900 amperes, having cast-steel poles cast in
one piece with the yoke. It has laminated pole-shoes bolted
on after the shunt-bobbins are placed in position. The out-

<3 2

izecoy Google

2 28 Dynamo Design.

side diameter of yoke is 91 inches, radial length of pole-core
15 inches, and the bore of field is 49*7 inches, and the arma-
ture beinp 49 inches in diameter, the length of air-gap is
about o*35 inch. The total flux from one pole at no-load
is 20*15 megalines, and as the area of pole-face is about
280 square inches, the pole-face density is about 73,000 lines
at no-load, which is rather high. In this machine in fact
both the magnetic fluxes and current densities are pushed as
high as possible, but a fairly high armature surface speed,
combined with careful design as regards ventilating capa-
bilities, enables this to be done, without excessive heating.
There are 1350 turns on each field-bobbin, and the poles
being circular the mean length of one turn is 58-6 inches ;
the shunt-winding space is 1 1 ■ J inches in length and the
wire is wound to a depth of 2 " 75 inches.

There are 200 slots and 800 conductors otiS inch by
o'443 inch, and consequently there are four conductors per
slot, the slots being 0-394 inch wide and itS inch deep.
The winding has eight parallel circuits and eight sets of
brushes. The armature has three ventilating ducts and the
stampings are held by two substantial end-castings bolted on
to a spider of simple design. The commutator is bracketed
out from the main armature-spider, being secured by screws ;
the diameter of the commutator is 33 inches and the segments
are 124 inches over all.

Mr. H. M. Hobart, who has written on dynamo construc-
tion in conjunction with Mr. Parshall, has contributed to the
subject of dynamo design a paper ' in .which he gives par-
ticulars of a large number of machines of his own designs.
Amongst these is noticeable a large generator, M P 22 — 1600
— 85 — 550 volts — 2900 amperes, which has the high peri-
pheral speed of 4000 feet per minute and the remarkably low
Steinmetz coefficient of i'44, showing great economy of
material. The chief data of this machine are as follows.

' faurn. Inst. Ekcl. Snf., vol. xx:
by Hobart in Electrical Raiino, vol. 1. p

izecoy Google

Examples of Dynamo Design. 229

Armaturi : —

Core disks, external diameter (inches) .... 177

„ inlernal ,, ,, 148

Number of slots ........ 440

Depth of slot (inch) i'34

Width „ „ 0-55

Pitchofslot at armature face (inch) 1-26

Depth of iron in core, under teeth (inches) . . 'S'S^

Gross length of core (mches) ,,..,. 13

Iron 8-6

Diameter of tinished armature (inches) .... 177

Number of conductors ....... 2640

Arrangement . . . . . . . . , 6 in a slot

Style of winding parallel (lap)

Dimensions of each conductor, bare {inches) . . . 0-53 x o'liS

Section of each conductor (sqaare inch) .... C'o6z5

Minimum width of tooth 0-735

Nnmber of ventilating ducts 7

Fiild-Magniis : —

Diameter of bore {inches) ...... 177'79

Pole arc ratio (per cent.) ...... 72

Diameter of magnet-core ...... 15

Length of ditto ig'iS

External diameter of yoke {inches) ..... 254

Gap 0-394

i'lux in magnet-cores (megKlines) ..... 17

Flui. density in pole-cores (steel) gjiOOO

„ in gap at pole-face ..... 64,000

„ in yoke (steel) 35, 000

,, in teeth (apparent) ..... 148,000

„ in core- body ...... 63,000

Commulalor : —

Diameter (inches) ........ 13S

Number of segments ....... 1320

Active length (inches) 9-6

Other data are as follows : —

Current density in armature conductor .... 2128

Space-factor of slot O'SI

Average volts per segment of commutator . . 9*2

Current density in brush face . . . . . 3^'^

Armature ampere-turns per pole 7900

Amperes in one conductor ...... 132

Ampere-conductors per inch peripheral .... 627

Ampere-tums per pole at no-load 13,000

izecoy Google

230 Dynamo Dedgn.

The losses are as follows at full-load : —

Armature iron loss (watts) 31 ,cx)o

., copper lews 24,400

Commutator resistance loss .... -Si 800

,, friction loss -...,. 5i30O

., stray losses 300

Excitation, shunt 13,000

•I „ rheostat 2,000

<• •• diverting thunt (see p. 131) . . . 1,000

Total . , 86,800

Total constant losses 52,600

„ variable losses 34,*oo

Commercial efficiency, Tull-load (per cent.) 94-9

„ half-load 92-8

1 , „ quarter-load 8S ' t

The number of watts wasted per square inch of peripheral
surface with a temperature-rise of 60° C. was 3-9 in the arma-
ture, 2-4 in the commutator, and o'jS in the magnet-coils.

The core stampings weighed/'/tons, the armature copper
I'l tons, commutator segments i."7 tons, the magnet copper
2*7 tons, the pole-cores 10 tons, total magnet-yoke (with
feet) 33 tons.

Another machine, MP 16 — 1000 — 90 — 500 volts — 2000
amperes, described in the same paper by Mr. Hobart, and
constructed by the Union Elektrizitats Gesellschaft, for Shef-
field, is fully described in the Elektrotechniscke Zeitschrift for
Jan. 16, 1902, vol. xxiii. p. 45.

Fig- 92 gives a view of a machine MP 8— 3Cnd — 150 —
— - volts — 600 amperes, by the Electric Construction Com-
pany of Wolverhampton, The yoke is of soft cast iron,
and the pole-cores of sheet-iron stampings cast in ; with soft
iron forgings for pole-shoes screwed on. The armature-core
is built up of charcoal iron stampings of 1 1 inches radial
depth, assembled to a gross length of 15 inches including four
air-ducts. It is clamped between cast end-plates, which have
brackets to carry the barrel-windings. The armature and

izecoy Google

Examples of Dynamo Design.

izecoy Google

232 Dynamo Design.

commutator are mounted on a cast-iron spider keyed at the
commutator end to the shaft, but expanded at the other end
into a wide driving-flange that is bolted to the boss of the
fly-wheel. On the inner part of the spider are cast oblique
webs to serve as fans. The armature-winding consists of
1536 conductors of strip of a sectional area stated to be
0-0636 square inch. The commutator, which is 42 inches
in diameter, with a working face 11^ inches long, has 768
segments. Each section of the winding consists of a single
jointless loop bent edge-on at the end away from the com-
mutator, and former-shaped. The external diameter of the
armature is 62 inches. The slots have straight sides, and
are lined with separate insulation. The insulation resistance
is stated at 100 megohms, and the temperature rise after
8 hours' full-load run is stated not to exceed 28° C. The
conductors are secured in the slots by wedges of hard wood,
and by binding wires over the projecting ends. The rocker-
ring runs on rollers bracketed out from the yoke, with an
adjusting worm-wheel gear. The shunt-current at normal
full-load is 8"3 amperes; the shunt-winding being former-
wound with coned ends. The machine is over-compounded
by 10 per cent, the full-load voltage rising to 550 volts. The
series-winding is also former-wound, of rectangular strip,
wound edge-on, of two flat spirals united together at the inner
periphery, thus bringing both free ends to the surface. At
full-load the series- winding takes up about i ■ 2 kilowatts,
being therefore about 0033 ohm in resistance.

A large traction generator constructed by Siemens and
Halske, of Vienna,^ was shown by this firm at the Paris Exhi-
bition of 1900. Its type is MP 14 — 1000 — 95 — 550 volts, thus
giving 1820 amperes at full-load. The armature has a diameter
of 98 ■ s inches ; the length between core-heads is 2 1 • 3 inches.
There are five ventilating ducts, each about 0-4 inch wide.
There are 1 144 conductors, each 0-157 'n<^h X 0*71 inch, and
four of these conductors are placed in each slot, there being
therefore 286 slots, the size of the slots being 0-51 inch wide
' ZeUschriJtfur EUktrotechnik, vol. nviii. p. 551, 190a

oy Google

Examples of Dynamo Design. 233

by I "97 inches in depth. The winding is series-parallel with
10 circuits. The commutator is 82 inches in diameter, with
572 segments, mica 0'03l inch thick being used for insulation
between the segments. The minimum length of air-gap is
0-35 inch, and the maximum length 0-47 inch, the average
■ flux-density in the gap being 58,000 lines. The yoke is of
cast steel, the overall dimensions of the machine being 156
inches or 13 feet.

The fourteen field-bobbins have each 770 turns of wire, the
ampere-turns at full-load coming out to about 18,500, the
excitation required for this being about 13 kilowatts, or i'3
per cent, of the full-load output of the machine. There are
14 sets of brushes.

The temperature-rise after 24 hours' full-load run is 30" C.

The total weight of the machine is just over 100,000 lb.,
or 44J tons.

Summary. — The machines which have now been described
are summarized in the following Table in order to afford a view
of the several values of the more important coefficients adopted
in their design. The coefficients a and /S in the later columns
are explained on p. 158 above, and signify respectively the
gross values of the average current-density and of the average
magnetic density in the active belt at normal full-load. The
values in the last column of kilowatts of output per cubic inch
of the active belt are, like the preceding, deduced from the
normal full-load output for which the machines are rated.
All makers do not, it is well known, adopt the same basis
for the rating of their machines, for some allow a higher
temperature-rise than others. It is quite impossible to reduce
the values to a common basis in this respect. Nevertheless
the values given are not altogether unfair as a means of com-
parison. If in any machine the peripheral speed is high,
there is necessarily a higher specific utilization of material.
And if in any machine in which this is high the a and ^
■densities are also high, the output per cubic inch of active
belt, and therefore the output in proportion to weight and
cost, will also be high. The 8-pole machine of the Inter-

oy Google

Dym

' Design.

Maker or Doiiut.

Sptdficiuon

Brown Bo*eri&Co.

194 35°

490

2-4749'5 575:38601302 27400:62-5

Electric

300 150

Soo

600

3-1 1 75,59212430 246 39500 43

English
Elcc. Manfg. Co.

100 ,00

550

2000

1-96 166 6102830 390 51400 103

General Elec. Co.

SSo 90

550

000

3-95 100 595,2760 297 3270044-4

Hobart . . .

1600 85

55o^9t>o

1-44 132 627I4000 467 47000 163

International
ElecEng. Co. .

450 250

500

900

1-96 M2 585 ■3200 496 47500 140

Kolben & Co. .

;o 75°

260

370

4-6 >35 403 4600 399 18200 67-4

Xolben & Co. .

350 125

550

455

4-35! 114 4772200 485 35600^ 74

Oerlikon Co. . .

265 370
265 370

450
550

510
360

2-9 25s 430,3860 365 29000 92
2-9 180 34813860 340 33800,106

Oerlikon Co, . .

1 MP

165 no

280

590

4'48" 59 660 Il6o'440 2500O146

Oerlikon Co. . .

i MP

■I

330 100

550

600

4-2 ICO 429'264o,363 374001 63

Scolt & Mountain

ISO 450

250

600

2-42,100 473 3940' 296 37800' 80

Siemens k Halske

1000 9S

550

1820

2-07 182 6722460 310 50500 70

Walker Co. . .

440 85

550

800

3-6 80 585J2C00 336 40500 44

xce™.

N*L

DKSiONS.

! 1 i

Brown Bo veriaCo.
(High \ oltagel

20 ;oo

000

7-35 10.3002740

273 25900

Kolben & Co.
(Esciter)

MPX

'S«

254

13-4 1 127 5i6i 660

437 34700

Oerlikon Co.

(Electrolytic)
Oerlikon Co.

(Electrolytic)
Thury

(Electrolytic)

MPC
MPC

MPC

560
832

55
450

80
190
208

7000

.500

+000

5-1 '228j447:254o
2-i8 250 835:4200
3-i5|333,435 3S8o

1 1 1

505 aB6oo
580, 24070
368' 29100

117

D, Google

Examples of Dynamo Design. 235

national Electrical Engineering Company, and the 22-pole
machine designed by Mr. Hobart, are cases in point. The
high specific values attained in these machines unquestionably
indicate the way to future economy in design. Everything
points to the adoption of high-speed steam-turbines for all
steam-driven dynamos of large power. With such speeds as
these machines entail, very high surface-speeds will be reached ;
and design must be modified to meet these conditions. Greater
axial lengths and relatively smaller diameters of armature
will be a necessity ; while with the high commutator speeds
carbon brushes cannot be used. Both these influences will
render greater the difficulties of sparkless commutation, and
make more needful than ever the most careful attention to
the question of saturation of teeth and of pole-pieces, and the
combating of armature distortion. But they will also bring
about a higher specific utilization of material.

With the introduction of large gas engines and the com-
mercial production of cheap gaseous fuel, it would seem likely
that for all generators exceeding 1000 kilowatts, gas engines
will be employed rather than steam engines, in case water-
power is not available. This development will again influence
dynamo design : and as is very evident, the dynamos of
largest output are precisely those in which the best ventilation
can be attained, and in which the highest specific utilization of
electric and magnetic materials is possible.

izecoy Google

D, Google

APPENDIX I.

DIX II.

Space
Factor.
(No Bed-
ding.)

Feet
per Ohm

Lbs. per
Ohm (bare)

Obms per
Lb. (bare)

Nnmler

'lig.)

ai 15° C.

at.5°C.

«,s-c.

S.W.G.

550

■7401

"S7I5

7611

-00013139

0000

)66

■7369

13592

5693

-00017566

000

•7339

1.895

4361

■00022933

262

■73M

10310

3276

■00030523

J03

■7260

8839

240B

-00041526

■7211

7481

1725

■00057980

■7155

6236

1199

-00083424

■7098

5286

861-2

■0011612

■7032

4413

778-6

-0016660

■6944

36.4

402-5

■0024841

■6875

3039

284-6

■0035137

■6794

*5i3

■94-7

■005 '355

-6697

2038

128-1

0078090

■5528

1601

78-97

.015225

■6408

1318

53-33

■018750

„

■6316

1063

34-82

■028717

■6139

831-4

21-300

■046949

■5939

628-6

12-178

-082115

■5566

509-2

7-9966

■12515

■557'

402-3

4-9884

■20047

■5483

3o8'o

2-9240

■34200

■5'9>

226-2

1-5773

■63399

-4830

157-1

-76037

1-3152

■4811

127*3

■49951

2-0020

■4559

100-6

■31173

3-2079

■4262

77-00

■ 18273

5 ■4727

■39'3

56 '59

-098651

10-137

■3712

47-43

■06948.

14-392

nGOQljIc

D, Google

APPENDIX U.

'■433
1-482

i-633

[■147
i-764

(-3673
1-6325
i-2603

■3070
■6273
■0816
■6513

■3070
■0387
■82354
■63617
-51912

■32473
■25877

Space
l-'actor.
NoBed-

IFcetper
1 Ohm

Ohm(Ce)

di..g.)

at 15° C.

74551

20782

133"

74339

16481

8371 '4

73509

13070

5265 ■!

73005

10370

33i4^8

72541

8219-8

30825

72022

6518-7

1309-7

71160

S'69-3

823-59

70249

4100-9

518-02

69061

3251-0

325-01

68093

2578-0

104-86

67176

2044-5

128-83

66140

1621-5

81 033

64784

w86o

50-974

62740

1019-6

32-041

60958

808-66

20-155

59270

641-35

12-678

57592

508-57

7 '9534

55853

403-28

5-0127

55256

319-76

3-I5I5

52774

253-65

19830

51296

201 ■ 10

I -2464

49051

159-51

-78416

47814

116 50

■49327

45480

100-31

■31019

44035

1 79-549

■15493

41164

; 63-064

-12258

39070

50-030

-077146

35249

, 39-679

-041526

Ohms per
Lb. (bare)
at If C.

-000075125
-00011945
■0001S993
-00030168

00048020
00076353

0019304
0030768

0077623

OI2J4I

019618

031210

078877
2573

9949
31731

■5043
-8023
I '2753

2-0273

3 ■ 2239
5-1271

Number
B. and S.

izecy Google

D, Google

APPENDIX III.

SCHEDULE FOR CONTINUOUS
CURRENT DYNAMO DESIGN.

Type of Machine

Design calculated by—
Machine constructed by
In opercttion at

Poles; -Kw. ; . Revs. p. min. ;

.Volts; Amperes.

Type. _____

Weight complete '
Voltage, No Load-
Over compounding.

COMMERCIAL TEST.

Load.

VolU.

CUCTHIL

Arm.' iShuDt.

LosMS-Wa.ls.

'Tota

Effi-
1 citney.

^*- Arm. 1 Arm. , E«c[- j Fiic
Copper| Iron. Iiuion. lion

No Load ... 1

J Load .

1 i ■ ^

;

J Load .

i ! ■ '

iLoad .

j ■

Full Load.

iJLoad .

Test made after hours' continuous run ; Armature

Current amps.; Shunt Current -amps,;

E.M.F._ ._. .volts. Resistance in shunt_ ohms.

Measured Resistance of Shunt Winding : cold (_ ^C.) ohms. ;

hot ( °C) .ohms.

Remarks ; . . .

Test made by.

3,3,l,ze.:,, Google

DIMENSIONS.

{uSlth

widlhofcacbduci
eflec:tivt lei^hofcoi

intemal diameter con
numlMr ot ilou

>li« or Mctlan of mndoctor (ten)

mean leneih of conductor per turn
pltsh orwlDdlnl tfr»nE ind hack)

diameter of iMtra &t Bald

MAaNET OOBE.

length radially
leng'h parallel lo shaft
width or diameter

BOBBIH.

leBgih o"er all

>1» ot ihont »lF«

YOKE.

COHHUTATOB.

GOmiUT&TOil BRUSHES.

ELECTRICAL.

COMMUTATOR.

annge volti beiveen ban
reversal density + pole face density

FIELD COILB.

umber of bobbins in
can length of one tu

mpere^' no load (shu
■nperes, per square i

Rdrop'ciortv^u)

nte.dlation
<:h.faUload

COMPOUND

WIHWFO.

rang
itnibc

"f turns it, K

„,.,^»

REACTIONS.

ARMATURE.

J. ,. per inch periphcrs

full load
„ mills, gap ai»l teeita + beneath
pole
density in gap under backward pole.hora

FIELD.

::,y Google

iiill|=

o|.stj|«j|

5|h« tl

S-SUili

^'«

«^.

?!S;

„,.

Jli^

^^i"!

■' =

"I'l-^

^jh

iii|

iSJiisls

•i r

alts.-;.

-. I

■g

■1

5ik

^is

■ E K : S^ i *g i a

his •

111 IK? _

III ^fi,M i

ill

M;J°|H

:I1T

, Cioogic

TEMPERATURE TEST.

!|!|f!|f:|Jj

AimituTc. ' Shunt. I Seria.

JVil'i'jJ's

After faouis nm

Remarks .
Dale of lest..

Test made by -..

WEIGHTS AND COSTS.

Machine Puts. Mueriil.

Yoke

Poles 1

Armature Core ...
Armature Copper...
Commutator ... |

Shunt Coil 1

Series Coil 1

Armature Spider ... |

Cost or Fin

SHED PbODU

™T....

^Z^

i».i.

PcrKw.

— -

Armature Shaft ... [
Brush Gear, &c. ... •

Bedplate & Bearings ,

Totals

K bad in quutity frai

^^^n. '

Google

INDEX.

Active Belt, Definition of 157

Air^ap, Average flux-density in .. .. .. .. ,, 136

„ Calculation of .. .. .. .. 3;, 136, i;z, i6z, 177

Alioth, Messrs , Winding Pole-cores ., .. .. .. ., 58

Allgemeine Etektficitais GesetUchaft 137

Alloys, for Rheostats .. .. .. ., ,. .. .. 133

" Anibroin" .. .. .. ..' ,. ,. .. ,, 71

Amortisseur .. .. .. .. .. ,. .. .. 17

Ampere-tums .. .. ., .. .. .. .. 3, 5,63

„ Calculation of 27,129,163,178

Annealing of Iron .. .. .. .. .. .. .. .. S

Apportionment of Losses 122, 143, 144, 150, 214, 230-

" Armalac " Varnish ^^

Atmature Conductors, Estimation of .. .. .. .. .. 148-

„ „ Number of .. ,, .. 138, 148, 211

„ Size of, how to find .. ,. .. .. 140, 147, 209 -

Armatures, Core-bodies of .. .. .. .. 29, 38, 77, 153 '

„ „ Dimensions of .. .. .. 140, 1^2

„ Equalizing Rings in .. .. .. .. 109, 200, 212, 223

„ Healingof 68,170,184

„ Insulation of ;. 77

„ Length of 140,150

„ Losses in ii;, 151, 168, 182, 213

„ Magnetic Density in .. 36, 136, 163, 208,- 213, 229

„ Number of Circuits in 97

„ Surface, Estimation of .. 68, 150, 151, 214

„ Teetli, Flux-density in

30,39,136, 151, 15s, 178. 200,208,213,229
„ Temperature-rise in .. .. 68,151,170,185,213

Winding, Theory of 78,82

Arnold, Professor E. on Adaptations of Wave- windings .. .. 109

„ „ on Commutator Losses .. .. .. 118

„ „ on Equalizing Connexions .. ,. ,, 112

,, „ en Formula for Commutator Heating .. no-

„ I, on Trcdetermination of Dispersion „ .. 26

izecoy Google

Dynamo Design.

Aruold, Profeisor E.,oa Reduced Diagrams.. .. .. .. 112

„ „ on Rule for number of Commutator Segments 149

„ „ on Series-Parallel Winding „ .. 85, 96

Asbestos 7') 75

Asphaltum 71

^j'^/ca S' P^rrc, on Magnetic Shunts 24

Baily, F. G., on Rotational Hysteresis 16

Bar Armatures .. .. .. .. .. .. .. 123

Barrett, W. F., researches on Aluminium-iroa .. .. 6, 9, 11

Bedding of Wires 48

Binding Wires, Calculation of .. .. .. .. 144

„ „ of Bronie .. .. .. .. .. 202

Bitumen 71

Bronze used for Binding-wires .. .. .. .. .. 143, 202

Brown, Boveri If* Co. .. .. .. .. .. .. 149, 150

„ „ Barrel- winding ,. ,. ., ,. ., 202

,, „ Double Current Machine .. .. .. 204

„ ,, High Voltage Dynamo .. .. .. 10;, 205

„ „ Multipolar Generators .. .. 142, 203, 205

„ „ Normal type of Generator ., ,. 203

Brown, C. E. L., Barrel-winding in two layers .. .. .. 202

„ Method of Piling Coib 58

Brown, E., on Heatingof Magnet coils .. ,. .. .. 67

Brush-sets, number of .. .. ., .. .. .. .. 10;

Blushes, Permissible Current-density in 117, 202, 20;, 208, 2 1 ;, 229

„ Pressureof 118,119,169,184

„ Resistance of Contact of .. .. .. .. .. n8

Carbon Brushes, Current-density in 117, 118,202,205,208,212,:

„ „ Resistance of

C^Aar/, /f. X, on Stray Field

Centrifugal Forces

Characteristic, Ejitemal .. .. .. .. .. 1

„ No-load .. .. .. .. .. 126, 165, 1

Cloth, " Empire " ,.

„ Mica ...

Coefficient of Dispersion ..

„ „ „ lncrer.se of at FuU-load

oy Google

Index.

Co:ninutation, Criterion of Goodness of ,. .. i;6

„ Ratio 156

Commutator, Diameter of, how to find 149, 211, 314

„ Fixing Number of Segments .. .. .. 139, 149

„ Heating of ., .. .. .. 120, 171, 185, 214, 330

„ Losses of EnerEY in .. 114, 117, 119, 169, 183, 213, 230

„ Peripheral Speed of itS, 120

„ Risers 138,208

Compagnie de Plnduslrie Eltctrique {see Thury) .

Compensating Ampere-turns 129, i6'i, 181, 208, 213

Compounding, curve of .. .. .. .. ,. .. 123, 172 .

Compound- winding. Calculation of .. .. 130, 166, 181, 213

Conductors, finding number of 138,148,211

Constantan 133

Constants, in Design 136,155

Copal 71

Cooling-surface, Estimation of, in Armature .. .. 68. 150, i;i, 214.

„ „ „ in Magnet-coils .. 65, 66, 68, 214

Copper Brushes, Current-Density in 117,118

„ Electric Resist ajice of ,. ,, ,. .. .. .. 41

„ losses in, Estimation of ,. .. 64, 113, 114, 167, 182

„ Secondary Losses in .. .. ., .. .. 114, 122

.„ Weight of 40,56

Copper-losses in Armatures .. .. 64,143,144,151,167,182,209

„ „ in Rheostats 117,209

.,. „ in Series Coils 167

,, ,, in Shuni-coils 64, 113, 117, 127, 143, 169, 183, 209, 213
Core-body of Armature, Siieof ., .. .. .. .. 140, 152

„ „ Insulation of .. .. .. .. .. 77

Core of Magnet Pole, to find .. ,. -. .. 153

„ „ Length of 153,212

Criteria of Good Design ,. 156

Cromplott, R. £., on Eddy-current Losses in Armature Conductors 125
Current-densities in Armature .. 138, 152, 155, 157, 205, 208, 21 1, 239

„ „ in Brush Contacts 117,118,205,208

„ „ in Commutator Risers .. .. .. 138, 208

„ ,. in Copper .. .. 65, 138

„ „ in Magnet-coils ,. .. 52, 138, 155, 200, 205, 208

D AM PEKS, Magnetic 17

Demagnetizing Ampere-turns .. .. .. ,. ., 127

Densities, Current .. .. .. .. .. .. 52,65, 138

„ Flux, Average .. .. ,. .. 36, 136, 208, 213, 229

R 2

izecoy Google

244 Dynamo Design.

ZJ^f-^j Method of Cross-ccmpcHinding 206

Design, Methods of 134, I46> 'S!

iJrf/miW, C, on Density of Flux in Cores 39

Dielectric Strength of Insulating Materials .. 7$

/>iV«i on Rotational Hysteresis 16

Dispersion, Magnetic .. .. .. .. ■• .• •- 18

„ „ Coefficient of 17,18,23

„ „ Increase at Full-load 27

Distortion of Field 1:7, 129, 156, 166, 181, aoS

Doubly Re-entrant, meaning of .. .. .. .. .. 83

Duplex Winding, meaning of .. .. ... .. 150

Ebonite 71

Eddy-cutrents 9

„ „ Calculation of Loss due to 168,183

„ „ in Armature Conductors .. 133, 124

„ „ in Pole-pieces .. .. .. .. .. ., 122

„ „ Law of .. ,. .. ., i«

Edge-wound Strip 48,57,66,201,332:

Efficiency, Apportionment of Losses in 143,144

„ Curves of ,. .. .. .. .. 122, 172, 192

„ Estimation of .. .. .. 121, i;4, 184

Electric Construction Co., Form of Field Magnet Coil .. .. 67

„ „ „ Multipolar Generator ., .. .. 230

" Empire Cloth " .. .. .. .. .. .. .. 7t, 77

" Enamelac " Varnish 72

Enclosed Motors, Heating of ., . .. .. .. 68

Ends of Coils, Methods of Fixing 59

Energy Losses, Assignment of 142,230

Engine Speeds, Variation of with Size .. .. ,, .. .. 136

English Electric Manufacturing Co.,lTs.z'mti Generator .. 142, 220
Equalizing R ngs .. .. ,, ., ., jog, no, 200, 212, 22Z

Equivalent Ring Winding - 99

Etson, IV. B., on Heating of Magnetic Coils .. .. ,, 65

„ „ on Stray Field.. .. .. .. .. ,, 33

£'k'/«^, 7. -^., Hysteresis Tester 9

„ on Hysteresis 16

„ „ Papers on Magnetism 8

Excitation 3,27

,, Calculatinn of ,, .. .. .. 162, 176

„ Losses due to .. ,. ,, ., 113, 116

izecoy Google

Ferranii, Messrs., on 'Edge-VJoimd Strip „ 49

„ „ on Heating in Strip Winding 66

Field-magnet Bobbins, Calculation of Heating ., .. 170, 185

„ „ „ Windings .. 51,52,53,167

„ „ Construction of ., .. ,. .. 57

Heating of 65,67

„ „ Ventilation of 61

Fischer-Hinnen, J., on Dynamo Losses .. ,. ,. .. 110

„ „ on Proper Number of Poles ,. ,. ., 137

„ „ on Rules for Fringing .. .. .. .. 35

Flux, Magnetic ,. .. .. .. .. ».. .. ,. 3

„ „ Useful 17

Flux Density 3, 6, 178, 213

„ Apparent and Actual .. .. .. .. .. 32

„ Average .. .. ,. .. .. .. 136, 178

Forbes, George, Rules for Penneance .. .. .. ., ,. 25

Forces, Centrifugal .„ ,. ., .. ,. ,, .. 141.

Formers used for Winding Coib 58

Former-wound Armature Coils .. ,, .. .. 102, 198,232

Frequency of Magnetic Reversals .. .. 1 1, 115, 149, 155, 211

Friction, Coefficient of, at Commutator .. .. .. 119, 169

„ at Commutator .. .. .. .. 143, 169, i34, 2:3

„ Lossof Energy due to,. .. 114,119,120,144,170,184,230
Fringing, Allowance for 28,35,163,177

GrtB? &• Co., Method of Fixing Coil Ends S9

„ „ Space-Factor in Slots .. ,. .. .. ,. 46

Gap, Air, Determination of .. .. .. .. 39, 152, 153, 211

Cap-Coefficients 2', 37, 163

General Electric Co. (Schenectady), Fluxes in Dynamo of .. 23

,, „ „ Method of Insulating Coils .. 60

„ „ „ Multipolar Generators,

139, 142, 208, 209, 214
„ „ „ Efficiency Curves .. ., 121

German Silver I33

Glass 71,72

„ and Sulphur ., .. .. .. ., .. .- 7/

Coldsborough, IV. £., on Distribution of Flux in Cores .. .. 38

„ „ on Stray Field 26

Gutta-Percha 71

oy Google

14^ Dynamo Design.

HEAT-Wa'te in Magnetisation 9, ii, 14> I'S

„ Calculation <ti {te: also Irot) Losses) .. i68, 182

Heating, Estimation of (j» n/jii Temperature-Rise) 113, iS4t i7o. '84-

„ ofAnnatures 69,170,184,213

„ of Magnet Coils 61,64, 117, 170, 185, 213

/f^/^-i'AiKc, //, and ^. ^a/, on Stream Lines 39

Hering, Ciw/, on Stray Fieid 23

Hobart, H. M., (tire alio Parsia/l), on ComniuUtor Speed,- .. I18

„ „ on Insulating Slots 75

,, „ on Multipolar Generators 141. U^

„ „ on Standardization of Design .. .. .. i:7

Hobart, H. M. &• Parshall, Efficiency Curves I2t

„ „ „ Temperature-Rise of Commutator .. I30

Hopkittson, Dr. John, on Coeilicient of Dispersion .. ., ■■ 2o

„ „ on Retardation cf Magnetism .. .. i&

Hysteresis or Magnetic Fatigue 8

„ Calculation of Loss due 10 .. 11,168,182

„ Constants of .. .. .. .. .. .. .. 11

„ Law of 9, i»

INDIA-ROBBER

•jt

„ and Asbestos (Vulcabeston) ..

71.7s

hard (Ebonite)

- 7S

„ Varnish (Scott's)

.. 72

Inherent Regulation

.. 132

Insulating Materials Classified

„ „ Dielectri; Strength of ..

- 75

Insulation of Binding Wires

.. 19»

ofBobbins

57.

,60,200

„ of Commutators

■■ 20s. 2

>s.

216, 235

„ of Core-Bodies

.. 46,

77.

198, 2C5

„ of Equalizing Rings

■■ I9»

„ of Field-magnet Coils

57,

• S9>

,60,202.

„ of Former- wound Coils

.. 58

of Slots

".. 46,

75,

. 77> 1^8

„ Test of

75.

190, 202.

International EUclrical Engineering Co., E-pole Generator of ,. 227
Iron, Magnetic Properties of Various Brands (^see also Curves of Plate !,)■

„ Resistance of Iron Wire for Rheostats 133

Iron Losses, i.e. Waste of Power in Iron .. .. '6, 143

„ Estimation of .. 113, 115, 168, ifa

izecoy Google

Japan Varnish

yohnsoit &* Phillipi, Bipolar Generator

Kapp, Gisisrt, on Predetennination of Dispersion .. .. .. 26

„ „ Rules for Calculating Weights of Coils ,. ,. 56

„ „ Method of Fixing Ends of Coils .. ., .. 59

n „ Rule for Dimensions of Armatures .. .. .. 141

„ „ Sparking Criteria IS9

Kiese^hr, and Sulphur 77

A'f/i**» &• Cf,, 10-pole Exciter 217

„ „ 4-pole Generator .. .. .. .. .. 21S

„ „ on Losses in Generators (Table oO I44

' „ „ ic-pole Traction Generator .. ., ,. ,. 216

„ „ Winding Scheme of Series- parallel Armature ,. 102

Kruppin .. ,. ., ., .. .. „ ,. ..133.

Laminc, B. G., Balancing Windings ,, .. .. .. .. 112

Lap-windings, Example of .. .. .. .. .. .. 101

„ Rules for ,. .. .. .. .. .. 90,92,93

Linen, Insulating Properties of .. .. .. .. .. .. 76

Losses, Apportionment of .. .. 122, 143, 144, 150, 214, 23U

„ Assignment of ., ,. ,, .. .. .. .. 142

„ Estimation of 64,113,120,167,182

in Copper.. .. 64, 113, 117, 127, 143, 151, 167, 182, 209,213
„ in Iron 16, 113, 115, 143, 168, 182

Magnet Cores, Average I

„ Yoke, „

Magnetic Dampers
Manganese Copper
Manganin ,,
Manila Paper

Marble

Material, Snecific Utilizatio

izecoy Google

348

Dynamo Design.

Mavor, H. A., on Active Belt 157

„ on Bedding of Wires .. .. .. 49

., on Stray Field 23

" Megohmite" ,. .. .. .. .. .. .. 71, 76

Meytr, H. S., on Iron 6

„ on Tooth Flux-Density 136

Mica 71,73.75

„ Canvas 7S

„ Long Cloth 73. 7S

,. Paper 75.77

„ Proper Thickness of 75,205,215,216,233

„ Shellacked (micanite) 71, 72, 73, 74, 75, 77

Micanite (or "made mica") 72, 73, 74, 75, 77

Mordey, W. M., Cross-connexions in Armatures .. .. .. 105

„ on Rotational Hysteresis 16

Keu, Levine, and Havill, on Heating'of Magnet Coils ' ..
Ncusilber ..

Nickel Steel

Nickelin

Nielhammcr, F., on Temperature Rise

No load Characteristic 126, 165, 1

Oiled Canvas ' ' 72, 73

.. Paper 7r, 75

Oils, as Insulators.. ., ,. .. ,. ,. .. .. 71

Oerlikon Machine Works, on Heating of Armatures .. .. 69

on Healing of Magnet Coils .. .. 66

on Insulating Matetials .. .. 72,74,77

4-pole, 265 kw. generator.. .. .. 187

i2'pole, 500 k"- generator for Basel iBB, 19a
lo-pole, 16; kw. generator for Bordeaux 194
3a-pole, 560 kw, generator for Rheinfelden 194
6-pole 285 kw. generator for Rome 137, 196

Order of Procedure in Design 134,146,155

Output, Relation of Size to
Over-all Length of Armature, Esti
Over-compounded Machines, Design of

140, 142, 158, 233

izecoy Google

Paper, as Insulator 71, 77

Papier-MSch^ 71, ^^

Paraffin Wax 71

Paraffined Compositions, Insulating Properties of 76

.. Paper 74,75

„ Slate 75

Parchment, Vegetable 71

/'«rJAtf//,/f.i^, Curve for High Flux-Densities 6

„ Efficiency Curves .. .. .. ., .. 121

„ on Apportionment of Losses .. .. .- 143

„ on Temperature- Rise of Commutators .. 120

„ lo-pole, 550 kw. generator ., .. .. .. 209

„ „ „ „ Sparking Criteria of 156

Parshall, H. F., *• Hobart, " Electric Generators " .. 8, 77, 208

„ „ Multipolar Machine, Described by 2c8

Penneability, Magnetic .. .. .. .. ., ,. .. 3

„ Variation of, at pole-face 129,181

Peimeance (Magnetic Conductance) 24

Phosphor-bronze .. ,. .. ., ., ,, 133, 145, 202

/■(Vuw, /?., on Magnetic Dispersion 23

ritchof Winding, Definition of 86

Platinoid ., „ ,. ,. ,. .. ,. ., •• '33

Pole-core, Fixing Din
Fole-pieces, Eddy-e

Poles, Estimation of Number of.. 137, I47i 148, 155

„ Testing Numbtr of .. .. .. .. ., ., 148

- Porcelain .. .. .. .. ,. ., ., ,. 71, 72

■" Press-spahn" .. ., ,. .. ,. ,, 71. 72, 75, 2co

Pressure-drop, Calculation of .. .. ,. .. 12;, 164, 179

Procedure in Design .. .. ,, .. ,. 134, 146, 155

Pi,ffer, W. L., on Stray Field 23

Radial Diagrams of Armatures
Rectangular Wires, Advantage of
Re-entrant Winding, Meaning of
Regulation, Inherent
Regulator, Shunt, Calculation for
Retardation of Magnetism

Rheostan

Rheostat, Shunt, Calculation for

oy Google

250

Dynamo Design.

^oAr, on Annealing Iron..

Rotational Hysteresis

Rothirt, Alexandre, on Apportionment of Losses

„ „ on Magnetic Dispersion

„ „ on Procedure in Design ..

„ „ on Space-futor of Coil ..

Saakey's Iron

Saturation Curve

Schedule for Costs

„ Dynamo Design

„ Magnetic Calculations

Scotl &• Mountain, Messrs., Equalizing Rings, r

-pole Generator

„ „ „ Flux- Densities used by

„ „ „ Guarantee for Standard Machine ..

:, „ „ Standard Generators, description of

., „ „ Shunt Bobbins, construction of

„ „ „ e-pole, ijokw. Generator, analysis of :

„ „ „ Test Curves of

„ „ „ Winding Diagram of

ScoiPs Rubber Varnish

Secondary Copper Losses

Segments of Commutator, estimation of number of .. .. 139,

Series-Parallel Winding 85, 961 138,

„ „ „ Advantages of., .. -. .. 109,
Siemens &• Halske, 14-pole Generator of .. .. .- 142, :
Silk

Size in Relation to Output

Shellac

„ and Mica (Micanite)
„ Varnish
Shellacked Cardboard

Short, Sidney H.

„ „ Standard Generators designed by ..

„ „ Walker Co's. Machine designed by

Slate

Slots, Estimation of Conductors in

„ „ Depth of

„ „ „ Dimensions of

,. „ „ Number of,.

Width of

140, 142, 158,

72, 73> 74, 75,

','. 76,

izecoy Google

Index. 25 1

Slots, Insulation of 46, 75, 76, 198

„ Space-factor of 46,151,212,216

Space-factor ., .. .. .. .. 43, 47 63, 151, 212, 21&

Sparking Criteria .. .. .. .. .. .. .. 154, >S9

„ „ Kapp's Rule IS9

„ Limits affect Size for given output .. .. .. .- 141

Specific Utilization Coefficients, Table of 23+

„ „ „ Discussion of .. .. .. 233

„ „ of Material 15?, 233

„ Gravity of Alloys .. .. .. .. ■■ '33

Speed of Dynamo, Variation of with site .. 136, 140, 141, 144, 23 j

" Stabilite" 7I1 7^

Steam Turbine Dynamos, design of .. .. .- .. .. 23J

Steatite ■ 71

Steel, Mild Cast, for Dynamos 6

,, „ „ Flux-Density .. ,. .. ,. .. ■• 136

Sleinmetn, C. P., Coefficient, choice of .. .. .. 140, 147

„ „ Discussion of .. .. .. ,, 140

„ on Strength of Dielectrics .. .. .. .. 77

Sterling's Varnish 72

Stiffness-Ralio, as a Critirion of Good Design 156

Stone-ware 71.72

Stranded Copper Conductors .. .. .. ., .. 45, 12;

Stray Field 18

Sulphur 71

„ and Powdered Glass 77

„ and Kieselguhr 77

Surface Speed in Relation to Heating ., .. .. .. 6g, 141

Table of Specific Utihzalion Coefficients

Tape, Insulating .. ,. .. .. .. .. 5

Teeth, Density of Flux in 30, 39, 136, 151, 155, 178, 200, :

„ Estimation of Flux-Density in 30,

„ Width of, in reUtion to width of slots

Temperature, Electrical Measurement of

Temperature Rise, Estimation of .. .. 113, 151, I

„ „ Permissible .. 65,69,113,151,170,1

Thomson- Houston Co., Armature of

„ „ Multipolar Generators

Thitry, Electrolytic Generators

Timmermann, A. H. &* C. £., on Heating of Armatures _..

.7=

.75

213.
SI.

230
163

44,
70
88,

213

184
233

iLCD, Google

Dynamo De:ign.

Trial Values m Design .,

for Diameter of Armature

ftr Length of Armature ..

for Number of Commutator Segments ..

for Number of Conductors

for Number of Poles

for Radial Dcfth of Iron Core-bpdy .,

for Size of Magnet Cores
Turbine, Steam, Design of Generators for ,,

142, 147
142, 147
139. 149
138, 148
137. 147

Utilization, Specific, of material

Varnish, "Armalac" 72

„ " Enamelac " .. .. .. .. .. .. 72,76

„ Insulating Properties of ., .. .. .. .. 76

„ Japan .. .. ,. .. ., .. .. .. 72

„ '■ Scott's" Rubber 72

„ Shellac .. ,, .. .. ., ., .. .. 76

„ "Sterling's 72,76

Ventilating Ducts, Allowance for 29,162,176

„ in Core-body 162, 176, 223, 228, 230, 232

„ in Field-Magnita 6i

Ventilation of Armatures .. 29, 69, 226, 229, 232

Vitrite 71

" Vulcabeston " ., ,. .. .. ,, ,. 71, 75

Wave-Windings, Balancing Circuits for

„ Examples of ,. .. 83, 102, it

„ Peculiarities of 97i "

„ Rules for

Walker Co., ic-pole, 440 kw. Generator

I, 11 11 » Design of ..
„ „ „ „ General Specificalior
Wedges used in Armature Slots ., .. .. i,

Weight of Generators in Relation to Peripheral Speeds
Weights of Copper, Calculations for

I, 92, 94
.. 142

.. 176

izecoy Google

Index. 253

Westinghouie Co., Use of Copper Dampers .. .. „ „ 17

WA«/tfr, 5. J., on Bedding of Wires 49

Wiener, Alfred, on Dispersion .. .. .. 36

„ „ on Dynamo-Machines .. .. .. 33

„ „ on Estimation of Number of Poles 137

Willesden Paper .. .. .. ,, .. .. .. 71

Windage Loss .. ., ., .. .. .. 114, 143

Winding, Choice of 108, 149

„ Formula for Armatures .. .. .. .. S2

Wire, Binding 144, 202

H-'iwrf, /^. //., Curves for Magnet-Winding 56

Wood 71, 72

„ Pulp, Preparations of 71

Yoke, Calculating Dimensions of

33. '53, 162, 177

izecoy Google

D, Google

iLCD, Google

D, Google

iLCD, Google

D, Google

iLCD, Google

D, Google

iLCD, Google

D, Google

. OF COMPOUNDING O

D- Magnet

of
ben Ixajnway
4o-2SO-I25

.,, Google

D, Google

iLCD, Google

izecDy Google

INDEX TO ADVERTISERS.

Anderston Founqry Co., Ltd., Glasgow . ■ I5

Armfield & Co.. J. }., Ringwood 6

Back & Manson, New Broad Street, E.C. (Roeblings) . 3

Brotherhood, Peter, Belvedere Road, S.E. .
Callender's Cable & Construction Co., Ltd., Victoria

Embankment .......

Connolly Bros., Ltd., Manchester

Davis & Son, John, Derby

Dick, Kerr & Co., Ltd., no Cannon Street, E.C. .
General Electric Co., Ltd., Queen Victoria Street, E.C.

Harling, W. H,, Finsbury, E.C

Harris, J. F. & G., Finsbury, E.C.
Hindley & Sons, E. S., n Queen Victoria Street, E.C. . ig
KOLBEN & Co, Prague, Austria . . . ' . -23

London Electric Wire Co., Ltd., Golden Lane, E.C.
Nalder Bros. & Thompson, Ltd., Queen Street, E,C.
Phosphor Bronze Co., Ltd., Sumner Street, S.E. .
Price's Lubricants, Battersea, S.W.

ROBEY & Co., Ltd., Lincoln

Scott & Mountain, Ltd., Newcastle-on-Tync
Tangye Tool & Electric Co., Ltd., Birmingham .
Wright, Clark & Wallis, 157 Southwark Bridge Road,

London, S.E. ........

itizecoy Google

ADVERTISEMENTS.

The phosphor bronze CO.

87 SUNHER STREET, SOUTHWARK, LONDON, S.E.

BoUing and Wire UUla— BAGOI BIBEET, BIBBnNaHAM.

SOLE MAKSBS OP Tan CSLBBBATBD

"OOQ WHEEL" AND "VULCAN" BRANDS

PHOSPHOR BRONZE INGOTS
AND CASTINGS.

A special quolily in the form of Phosphor Branie Spring and

other Wire. Rods. Plates, Strips, Doctor Blades. Ao.

Also supplied in Billels. Wire. Bars, and Ingots,

SILICIUM BRONZE

ELECTRICAL WIRE,

Sbe Qtmlitia (lee Wreulan and price LitU).

For Overhew) Telegraph and Tolephooe Linos, £c., as lued by the chief

OoTutDmenbi, Kailwa; and TelepboDe Compaaies throaghoat the Wotld.

NOeSTRUOTIBILITr.

ROLLED AND DRAWN PHOSPHOR BRONZE,
GDN METAL, BRASS, TIN, AND GERMAN SILVER.

Bronze, Gun Metal and Brass Castings in
the rough or machined if required.

izecoy Google

ADVERTISEMENTS.

The phosphor bronze CO.

LIMITED,

Sole Makers of the following ALLOYS :
PHOSPHOR BRONZE.

« DURO METAL " ^^^^^^ »,^,

For Roll Bearings. Wagon Brasses, &c.

PHOSPHOR TIN AND PHOSPHOR COPPER.

"Cog Wheel" Brand. The beat qualities made.

PLASTIC METAL.

BABBITT'S METAL.

"Vulcan" Brand. Seven Grades.

"WHITE ANT" METAL, No. I.

and Superior to

"WHITE ANT" BRONZE.

Superior to Fenton's Metal for Car Bearings.

PHOSPHOR WHITE LINING METAL.

Equal to White Brass No. 2.

PleoK apply for Calaloguei eoniaining fail yaHicidari lo the
Campany'e Head O^ee,

87 SUMNER STREET, SOUTHWARK,

I.CONDON, S.E.

DigiliLcGOOgle

ADVERTISEMENTS.

ROEBLING

HIQH CONDUCTIVITY ELECTRICAL

Copper Wires

FOR EVERY ELECTRICAL, TELEGRAPHIC AND
TELEPHONIC USE.

HARD-DRAWN

COPPER TROLLEY WIRE

A SPEiciAiL.rr'y.

Biire and Insulated Wires and Cables of every description.

Iron, 5teel and Copper Wire Rope.

COLUMBIA RAIL BONDS

"BEST IN THE WORLD."

117 & 119 LIBERTY ST., NEW YORK, U.S.A.

H. L, SHIPPY, Treasurer.

ENQLISH RKPRKSKNTATIVKB:

Messrs. BACK & HAHSON, 36 New Broad Street, London, E.G.

Telephone, No, 2592 London Wall, Telegrams— Back Manson, London.

,j ...... nGoogIc

ADVERTISEMENTS.

J. R & Q. HARRIS

Mahogany, Walnut, Oak, Teak,

Whitewood, Pine, Pitch Pine &

General Timber Merchants.

litSoulding ]IIci,ni:tf£i,otupei?8.

MOULDINGS MADE TO ANY DESIGN AND IN ANY WOOD.

CASINGS

BLOCKS ■ ■■ ■ CLEATS

Ana alt Wood Fittings for Electrical Purposes.

Offices and Yards —
S8< WILSON STREET, LONDON, E.C.; 42 OBANGE STREET,

ORAVEL LANE, SOUTHWARK, S.E., So.
Mills— PALMER'S ROAD, OBEEN ST., BETHNAL OREEN, &o.

Please Write to 58V WILSON STREET, B.C., for Illustrated Lists.

RODING JOINERY WORKS, ILFORD, E.

IRON & WOODEN BUILDINGS MADE TO ORDER.

JOINERY OF BVBRV DESCRIPTION.

Please Write foK> EatlmateB.

„Googlc

ADVERTISEMENTS.

GENERAL ELECTRIC Co., Ltd.

ENGINEERING WORKS,

®itt0n, near BIRMINGHAM.

iie:a.x> office :

71 QUEEN YICTORIA ST., LONDON, E.G.

O.E.C. 300 Kilowatt Three-Phase GeneratoT, 7,200 Voitt, 390 Revs, per minute.

Electrical Machinery of all kinda for Lighling,
Traction and Transmission cf Power.

. - Gooijlc

ADVERTISEMENTS.

BRITISH EMPIRE
TURBINE.

Made in Six Types, Horrzonfal and Vertical,
and ali sizes from 3 in. diameter.

CHEAPEST TO BUY. NO WASTE OF POWER.

EASIEST TO FIX. NO BELT TO COME OFF.

SIMPLEST TO OPERATE. NO SEAR TO BREAK DOWN.

EFFICIENT AT WORK. ABSOLUTELY RELIABLL

DIRECT DRIVING ELECTRIC GENERATING TYPE.

Turbine and Dynamo, on same shaft and bedplate.

for generating Electricity fnom Water-power

without any Intermediate transmission.

Sole fl>anufacturers :

J. J. Armfield & Co.

20 MARK LANE, LONDON.

Works: RINGWOOD, HANTS.

L,j,i,„,;,„ Google

ADVERTISEMENTS.

PETER BROTHERHOOD

MECHANICAL ENGINEER,

Belvedere Road, Westminster Bridge,

I^ONOON, S.E.

IMPROVED, ENCLOSED, SIMPLE, SINGLE CRANK OR
COMPOUND TWO OR THREE CRANK ENGINES,

Fitted with the vary latest PerfiBcted Byatem of Automatic Forced
Lubrication.

HIGH EFFICIENCY, NOISELESS RUNNING, PERFECT GOVERNING,

HPROVED PATENT SIMPLE OR COMPOnD THIEE CYLIHDER ENGINES

STEAK AHD ELECTRICALLY DRIVEH AID COMPRESSORS.

Contraotora to the ADMIBALTY, VAB OFFICE, COLONIAL and

INDIAN GOVERNHENTS, and to ths AUSTBIAN, OEBHAN,

JAPANESE and other FOBEION GOYEBNHENTS.

ADVERTISEMENTS.

Ltibrkants for Electrical Installations.

The numerous novel engineering and mechanical devices which occur in
modern electrical installations, and the hit;h temperatures, pressures and speeds
which are common in these, have called for various new systems to provide for
the effective luhricatior of the different bearings. The most prominent among
these systems of lubrication are " splash,". "forced" and "rings or chains.''
The feature common to these three systems as distinguished from the older
methods of oiling is that the lubricant remains in use, with small periodic
additions to replace waste or loss, for long periods, and is only renewed at
intervals of, it may be, many months. Such systems call for oils which are
not seriously affected or changed in character by long use upon the bearings or
by exposure to the action of water, steam or air.

These conditions can only be attained in oils which refuse to combine
chemically or mechanically with water, which, separating readily from water,
may be easily filtered to eUminate dirt or other impurities picked up in use and
which neither become oxidised nor develop gummy secretions in use.

The following oils are recommended to electrical engineers as meeting in the
fullest manner the conditions in which they are interested.

For Gas and Oil Engines. — Price's Gas Engine Oil, which is universally
recognised as the standard lubricant for internal combustion engines of all
kinds, and is used and recommended by the leading makers of these for the
lubrication of their respective engines.

For Steam Cylinders and Valves. — Price's Sherwood Valve Oil, a charcoal
filtered oil of dark amber colour, perfectly free from all bituminous residues.
Fluid at normal temperatures, body at 212" Fahr. = three to four times that of
best tallow, flashing point 525° to 550° Fahr,

For Steam Engines. — Price's '' Grosvenor '" Engine Oil or Compound
Engine Oil.

For '^splash'" Lubrication. — Price's Sherwood Valve Oil, as above.

For "Forced' Lubrication. — Price's "Pioneer" Machinery and Engine
Oil, a pure hydro-carbon of moderate viscosity, separating readily and com-
pletely from water.

For "Rings or Chain" Lubrication.— '?i\c€% Compound Engine Oil or
Price's Pioneer Machinery Oil.

For Dynamos and Electric Motors. — Price's Dynamine, a rich pure hydro-
carbon of amber colour and medium body.

Further information upon this subject and full particulars of these oils wiU
be found in the undernoted pamphlets — "Some Aspects of Lubrication";
" Descriptive Catalogue of Oils " — of which copies may be obtained from

PRICE'S PATENT CANDLE COMPANY LTD.,
Oil Manufacturers and Refiners.

Belmont Works, Battersea, London, S.W.
Bromboro Pool Works, near Birkenhead.

::,y Google

ADVERTISEMENTS.

JOHN DAVIS & SON

(DERBY), Z.T1>.

All Saints' Works, Derby,

26 VICTORIA STREET, WESTMINSTER.

CELLULOID SLIDE RULES

For ftll Tecbnicftl Cklcnlationi.

CtllnloM Slide Rules lUb Spin Logologaritlunlc SMe

Lt-Col. H. C. DUMLOP, TLF^ ACS. JACKBOX, Baq^ X.A.

Circular Calculators.

MATHEMATICAL DRAWING INSTRUMENTS.

ADJUSTABLE DRAWING BOARD TRESTLES.

THE UNIVERSAL DRAFTING MACHINE.

STEAM EHGIHE INDICATORS
REVOLUTION COUNTERS.

ELECTRICAL ACCESSORIES :

Damp-Proof Lamp FittinsfSi

Damp-Proof Switches, Fuses, etc.

Automatic Recordlnfi: Micrometer,

Wire Gauge.

J
I

■ 1r

WRITE FOR NEW CATALOGUE.

Section A— InstiTunent Department. Section B— Electric Department.

ADVERTISEMENTS.

Nalder Bros. & Thompson, Ltd.

Voltmeters, Ammeters,

Recorders,

Circuit Breakers &

Switchboards.

Recording InetmmentB'

NEW
CATALOGUES

r^ist oi

£. BOBERTS, 6 Eo
SIMON, BEBBY&(
WH. HoOEOCH It

Qkagow.
YANDAH, HUtSI

Priory, Biiminf
BOBT. BOWBAN,

LDCIEM ESPIB, 11

OSWALD HAES, 6t} Uargaret street, Sydiie;.

BALHEB, LAUBIE & CO., Oaloutta.

Telegram.: "Occlude, London." Telephone Moa. : 124 A 6124 Bank.

ADVERTISEMENTS.

CONNOLLY BROS., Ltd.

Insulated Wire & Cable Makers,
BLACKLEY, MANCHESTER.

TeUjrunt: "COMMOLLTB, BLACKLET." Telephone : Ho. 2361.

SPECIALITIES :

WIRES & CABLES

Up to one square inch section of Copper. Insulated with
Vuloanisfld Indiarubber.

Flexible Dynamo Cables.

siariTCHBOARD cabil.e:s.

Ouaranteed to an; required Voltage.

Insultting Tapes and Cloths (or Dynaio and Motor Windings.

SOLE MANUFACTURERS OF "BLACKLEY' TAPE.

a«lf-Adh«il¥& Tlu Bait Ta^ Ka Oat.ld. Go.«rlii< In the lUrkat

WKITE FOR OUR LATEST CATALOaVBS * REFERENCES.

AOENCIES AHD STOCKS IN LONDOH,
OLiSOOW, NEWCASTLE ON-TYHE AND HELBODBHE,

ADVERTISEMENTS.

■1 ^

i :3

iil

in I

[in

Sfl"

rli

s -s

-.18 ass

III I

" a

z s

O i

Q s

Z '.

O I

S °

1. s"

s E

I 3

I s

z-
J

,, Google

13 ADVERTISEMENTS.

ESTT^BlalSH&O 1879.

The LONDON ELECTRIC WIRE

T«l«i«m.: f AMP A NY T tH Tdephon.!

•BlMtrio, London.' Vivllir Ail X i lilUi ""■ **^ '^^^

Anchor Works, Playhouse Yard,

Oolden |l<ane, I^ondon, E.C.

And CHURCH ROAD WORKS, LEYTON, E.

ManutacturBra of

Sllk-Covored WirOS, Copper, "Platinoid," "Eureka," German-
Silver, etc., for Electrical Instruments.

Cotton-Covered and Braided Wires, Strips and

Cables for Dynamos.

Cables of Llg^ht, Medium, and Hlffh Insulation

for Electric Lighting and Power, Telegraphs and Telephones.

Flexible Cables for Dynamo Connections.

Line Wires for Electric Ligrhtingr, Electric Bells and

Telephones, etc.
Flexible Cords of all descriptions for Incandescent Lamps, etc.
Dynamo Brushes, Copper Gauze :— Ordinary type, " Sparkless "

and Self-Lubricating.

"Platinoid," "Eureka" and other Hlffh Resistance

Wires.
Fusible Wires for Cut-outs.

Jointing Materials of all descriptions.

Varnishes and Sundry Appliances.

Contraotors to H.H. GoTernment and the leading Eleotrio Light
and Telephone Companies.

ENGINEERS' SPECIFICATIONS CAREFULLY WORKED TO.

PRICE LISTS AND SPECIAL QUOTATIONS ON APPLICMTION.

ADVERTISEMENTS.

Telegrams :

" DICKER, LONDON."

ALL OODE8.

DICK, KERR & CO.

Limited.

Standard Q>ntinuous Current
Generator,

Head Office: Works:

110 CANNON STREET, PRESTON, LANCS.

LONDON, E.C. KILHABNOGK, M.B,

, Cioogle

ADVERTISEMENTS.

P=3

Accessil

omy.
Darabi

& MIDDLES

On Admiralty an

RY GO.

1 E-H

Efficiency.
Econ

Works: GLASGOW

;=>

c«

•|

5 Bfl

^"2

C/5

csr

u"J

•1

*-z

, CiOoqIc

ADVERTISEMENTS.

TANQYE

TOOL
and ELECTRIC CO., Ltd.

BIRBXINOHABa.

DYNAMOS
MOTORS

ELECTRICALLY-
. DRIVEN
TOOLS

CATiLOOUEB ON APPLIOATIOH

L.., ,„Gcx'iglc

ADVERTISEMENTS.

R0BEY&C0.1TD., LINCOLN

Upwards ot
22,000 Engines

at Work In
All Parts of the

World.

Illuitrated

C&tB)o£uei poit fi'ee

on Application.

Coupled Compound HoriiantAl Fiied Entinu.

bsjlkMrs of

STEAM > ENGINES (simple & CompouDd)

ALSO

GAS & OIL ENGINES

Over 260 Gold

and other
Medals Awarded

BUNCHES AND

AGENCIES

IN Aa PANTS CF

THE WOULD.

Compound Vertical Engines for Electric Lighting.

LOUDON Olcts md Sliow Rooms— 19 IJUEEll YICTOMA STREET, E.C.

ADVERTISEMENTS.

TANQYE TOOL

and ELECTRIC CO., Ltd.

ARK PRCPARKD TO OFPIR

Electrically 'Driven Machine Tools

LATHES, MILLERS, DRILLERS, &C

AND IN CONJUNCTION WITH

TANGYES Limited

FOE

ELECTRICALLY-
DRIVEN
• MACHINERY

OF ALL KINDS.

DYNAMOS WITH STEAM, GAS OR OIL ENGINES

■ MOTORS for Drirag Pimps, Overhead Travellers, etc.

CATALOaUES & FULL PARTICULARS ON APPLICATION.

.oogle

ADVERTISEMENTS.

E.S. HiNDLEY&SONS

Works ! r'

LONDON

XX QUEEN
!

STEA»
and

For ELEC

D,g,l,i.aD, Google

ADVERTISEMENTS. 20

t ERNEST SCOTT & MOUNTAIN, L">

1
I

Engines, Dynamos, Motors, Pumps, &c.

iLCD, Google

ADVERTISEMENTS.

^^ HAMILTON HOUSE, \J

VICTORIA EMBANKMENT, E.G.

Telephone : No. 19tl Holfaorn. Telegraraa : Callendep, London.

Laying CallendBr Mains for

DUDLEY TRAiyiS

CABLE & CONSTRUCTION CO.

Dij.iei, Google

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Design of dynamos"

See other formats