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

Full text of "BSTJ : Early Fire-Control Radars for Naval Vessels (Tinus, W.C.; Higgins, W.H.C.)"

The Bell System Technical Journal 

Vol. XXV January, 1946 No. 1 

Early Fire-Control Radars for Naval Vessels 

By W. C. TINUS and W. H. C. H1GGINS 


FOR a number of years before the war a very intensive development 
effort was under way in the Army and Navy laboratories, and in sev- 
eral commercial laboratories, on the application of radio methods to the 
location of objects at a distance. The equipment which resulted was 
eventually called "Radar" equipment by the Navy and this term is now 
almost universally used. The urgent needs of the war have resulted in the 
very rapid development and extensive application of this new science 
during the last few years. 

Radar equipments of many different types have been designed to perform 
specific functions on land and sea, and in the air. These equipments 
have had an important part in the winning of the war and the recent re- 
laxation in secrecy regulations now permits publishing some of the story. 
In this present article a description of the Mark 3 and 4 Fire-Control 
Radars for Naval Vessels will be given, together with a little of the history 
that preceded their development. 

Historical Background 

When the Bell Telephone Laboratories began active radar development 
work early in 1938 an effort was made to set technical objectives for this 
work that would avoid duplication of the intensive work then under way in 
the Army and Navy laboratories, and that would advance the art toward 
the solution of some of the recognized basic problems. The general ob- 
jectives were to increase the accuracy of radar measurement of location and 
to increase as much as possible the operating carrier frequency. The 
reasons for these objectives are discussed in the following paragraphs. 

The state of the art at the time under discussion has been partially de- 
scribed in a recent paper by Maj. Gen. R. B. Colton. 1 The work he de- 
scribed and directed was carried out at the Signal Corps Laboratories at 
Fort Monmouth, New Jersey and was directed principally toward solving 

1 "Radar in the U. S. Army" by Maj. Gen. Roger B. Colton, published in the Proceedings 
of the I. R. E., November, 1945. 



the ground forces' problems of aircraft warning and searchlight control. 
At the same time intensive work was being pursued at the Naval Research 
Laboratory at Anacostia, D. C. under the direction of Dr. A. H. Taylor, 
Dr. R. M. Page and Mr. L. C. Young. Their work was directed primarily 
toward developing radar equipment that would be useful aboard ship, and 
it was from them and from the engineers of the Navy Department that the 
principal inspiration and guidance for the work described in this paper were 

The first military application in which radar equipment proved its use- 
fulness was in the detection of approaching aircraft. For this kind of 
application the radar is not required to locate the approaching planes with 
very great accuracy and the experimental radars of 1938 and 1939 per- 
formed this function in quite a useful way. The fact that the first appli- 
cation of radar was a strictly defensive one may account in part for the 
great interest and support given radar work in England and in this country, 
while apparently much less radar work was done before the war by the 
scientists of Germany and Japan. Thus, when radar later became a power- 
ful and versatile aid to offense, the enemy nations found themselves years 
behind in development. 

Very early in their work the men of the Naval Research Laboratory 
recognized the potential ability of radar to help solve the fire-control prob- 
lem. Since this problem determined the design of the radar systems to be 
described later in this paper a brief general discussion of fire control is given 
here. The term fire control refers broadly to the means by which a gun or 
other weapon is aimed and fused so that, when fired, the projectile will hit 
or burst near the intended target. A fire-control system includes two 
major parts: first, a locating device for determining the present position of 
the target; and second, a computing device which analyzes the present 
position data, computes the target's course and speed, and the position the 
target will occupy at the future time when the projectile arrives at that 
point, and finally furnishes the correct aiming and fusing information to 
the guns. A modern fire-control system does these things in a continuous 
manner so the guns remain correctly aimed and can be fired at any time 
during the engagement. 

Before the war the present position of the target was ordinarily de- 
termined by optical instruments. Operators tracked the target by con- 
trolling their telescopes in such a way that the target remained on the 
crosshairs in their eyepieces. Thus the azimuth and elevation angles 
were found. Another operator measured the range to the target with an 
optical range finder, or indirectly estimated range from the angular extent 
of the target and its estimated size. 

The accuracy of this optical system in determining azimuth and elevation 


angles is very good provided the target can be seen clearly. This proviso is 
a serious limitation under many typical operating conditions. It is fre- 
quently difficult to see a target at a range of several miles on account of 
haze even on a relatively clear day, and at night or in fog or smoke screen 
the usefulness of a telescope is almost nil. The optical range finder is 
subject to the same limitations as the telescope and in addition leaves 
much to be desired in the matter of accuracy and continuity of data even 
under the best visibility conditions. This is due to the fact that optical 
range finders are triangulation devices which inherently have accuracy 
limitations. The need for a long and very stable base line between the 
prisms of an optical range finder is difficult to meet aboard ship, and the 
principle of operation makes inevitable a rapidly decreasing accuracy with 
increasing range. Thus, as the effective range of guns increased, the need 
for more accurate means for measuring range became more acute. 

In its earliest forms radar offered at once a potential means for measuring 
range with much better accuracy than that of the optical range finder. 
This was due to the different principle on which radar works. A pulse of 
radio frequency energy is sent out to the target and the echo signal is re- 
ceived back at the source. The velocity of the waves en route is the 
same as that of light, and is one of the basic physical constants. To measure 
range accurately with radar required only the development of techniques 
for producing short transmitted pulses and for measuring accurately the 
short intervals of time between the transmitted pulse and the returning 
echo pulse. Both of these were the kind of problems which yield readily 
to electronic solutions. The early work in Bell Telephone Laboratories 
thus included the production of shorter transmitted pulses than were 
being commonly used, and the development of improved range measuring 


The second important general objective for the early work at Bell 
Telephone Laboratories was to devise equipment which would operate at 
frequencies much higher than had been previously used. The need for 
higher-frequency operation arose from the fact that for a given size of 
antenna the beam width decreases with increasing frequency while the 
gain increases. Narrow beams are required to obtain accurate angular 
data while increased gain is desirable since it obviously provides increased 
range for a given transmitter power and receiver noise figure. These factors 
are illustrated by the curves of Figs. 1A and IB which show the relationship 
between beam width, antenna gain and antenna size expressed in wave- 
lengths. The curve labeled "uniform illumination" yields maximum gain 
and minimum beam width for a given antenna size but produces unwanted 
side lobes of undesirable amplitude. For this reason the illumination is 
usually graded over the antenna aperture to reduce minor lobes. The gain 





















> — 

10 100 


Fig. 1A— Antenna beam width vs. aperture 


5. __ 



10 100 

Fig. 113 — Antenna gain vs. aperture 


and beam width obtained in this manner are shown by shaded area labelled 
usual design values. The need for higher frequency or shorter wavelength 


was apparent to all of the early experimenters since physical limitations 
restricted the size of antenna which could be installed conveniently aboard 
ship. However, development effort along these lines had previously been 
hampered by lack of suitable vacuum tubes. 

In spite of the vacuum tube difficulties the Laboratories work was started 
in the range from 500 to 700 mcs, a region several times that then in use at 
the Army and Navy laboratories. The best tubes available were those of 
the doorknob type which have been described in the literature by A. L. 
Samuel 2 and are illustrated in Fig. 2. The smallest of these was used in 
the receiver input circuits and two of the middle sized ones were used in 
the tramsmitter oscillator. These triodes operated at quite high frequencies 
by virtue of the very small spacing between their electrodes, a feature which 
made them fragile and demanded the development of plate modulation. 
Earlier radars had generally used grid keyed oscillators, i.e., the plate 
voltage was applied to the oscillator continuously together with a high grid 
bias voltage. The bias was removed momentarily by the keyer to emit a 
pulse. In order to obtain a useful pulse output from the doorknob oscillator 
tubes it was found essential to remove all stress from them except during 
the pulse. This was accomplished by using a direct coupled pulse am- 
plifier or modulator, effectively in series with the oscillator and the power 
supply. Here again in 1938 no really suitable tubes were available for 
the modulator service since it also demanded a highly intermittent duty. 
However, since the modulator duty did not require the tubes to operate at 
very high frequency it was possible to use rugged high-voltage triodes which 
had been designed for continuous service, and to obtain the required pulse 
current capacity by paralleling a number of tubes. The earliest radar 
modulators used in the Laboratories employed a group of Eimac 100-TH 
tubes. Later, in the CXAS and Mark 1 Radars, six tubes similar to the 
W. E. 356A were used in parallel. 

After a great deal of laboratory work an experimental equipment was 
assembled and demonstrated to the Army and Navy in July 1939. This 
early radar was notable in that it operated at what was then considered a 
very high frequency and also in that it employed a single antenna only 
about 6 ft. square. The transmitter and receiver were connected to the 
common antenna by a duplexing technique to be described later, which had 
been applied at lower frequencies by engineers at the Naval Research 
Laboratory. The results of these first field tests were encouraging and 
both the Army and the Navy ordered one prototype model equipment to 
be known as the CXAS. This radar was to operate at 500 or 700 mcs and 
was to incorporate a number of new features which were designed to make it 

2 Proceedings of I. R. £., Vol. 25, page 1243, 1937— "Negative Grid Triode Oscillator 
and Amplifier for Ultra High Frequencies." Digest in Oct. 1937 B.S.T. J. 




convenient to operate and to provide a range accuracy that would be useful 
in fire control. Since this early radar is of cosiderable historical im- 
portance it will be described in some detail. 

Fig. 3 — CXAS — Antenna 

The CXAS Radar 

This equipment was divided into three major assemblies and the circuits 
were arranged so the three could be installed at some distance from each 
other. The antenna (see Fig. 3) consisted of a cylindrical parabolic re- 
flector about 6 ft. square with an array of eight half-wavelength dipoles 
along the focal line. With shipboard use in mind the reflector was per- 
forated to minimize wind resistance and the dipole and coaxial line feed 



system was made weatherproof, which was accomplished by making the 
line system pressure-tight and filling it with dry gas. The gas-line system 
was extended to include the radiating elements by covering the latter with 
pyrex test tubes sealed to the support with a packing gland as shown in 
Fig. 4. A device was included in each dipole assembly for supplying the 
two half-wavelength radiating elements with balanced voltages from the 
unbalanced line, while a coaxile line harness including impedance matching 

Fig. 4— CXAS— Dipoles 

transformers was used to connect the several dipole assemblies and provide 
a matched load to the single transmitter-receiver line. A schematic dia- 
gram of this arrangement is shown in Fig. 5. The contemplated use of 
this radar was for surface targets or low-flying planes and rotation was 
provided only in azimuth. A gas-tight rotary joint was developed to carry 
the f " coaxial line through the azimuth axis (Fig. 6). 
The operator's cathode ray oscillograph indicator and all of the essential 













Fi". 5 — CXAS — Antenna schematic 

Fig. 6— CXAS— Rotary joint & transmission line fittings 



Fig. 7 — CXAS — Indicator desk — covers removed 

operating controls were combined into an assembly called the Indicator 
Desk, a photograph of which is shown in Fig. 7. This was intended for 
indoor mounting below the antenna in such a position that the azimuth 


hand wheel on the desk could be connected to the antenna turntable by a 
shaft. The indicator employed a 7" cathode ray tube and displayed the 
radar signals by what is now known as a Class A sweep with a full scale of 
100,000 yards. A pioneering feature of this indicator was the provision of 
a series of electronic range marks to increase the accuracy with which target 
range could be read. Earlier indicators had used a ruled mask for the 
range scale and had suffered in accuracy due to parallax, sweep non-lin- 
earity, drift of sweep position, etc. The CXAS provided sharp pulses to 
mark the 10,000-yard intervals along the sweep line, and smaller pulses to 
mark the intervening 2,000-yard intervals. This system was free from 
the errors of the ruled mask and permitted range readings accurate to 
±200 yards throughout the 100,000-yard scale. Provision was also made 
to expand any desired 20,000-yard segment of the scale to fill the entire 
tube screen so that signals could be examined more closely. The ranges 
corresponding to the 10,000-yard intervals were designated by illuminated 


Fig. 8 — CXAS — Range mark system 

numerals located directly below the electronic scale. The presentation 
obtained with this arrangement is indicated in Fig. 8 which shows the 
electronic calibration marks, transmitted pulse, and an echo at 43,000 
yards on both the full and expanded scales. 

The third part of the CXAS equipment was an assembly known as the 
Transmitter-Receiver or Main Unit. It was designed to be unattended in 
normal operation and contained the Pulse Generator, Radio Receiver, 
Power Control Panel, Radio Transmitter, and H.V. Rectifier, which were 
all built as removable drawer type units. A side compartment in the 
Main Unit also housed the duplexing circuits, gas equipment for the trans- 
mission line, and some built-in test equipment, including a wavemeter and 
monitoring rectifier. The Main Unit and its sub-units are shown in Figs. 
9 to 14, respectively. A single \" coaxial transmission line provided con- 
nection from the Main Unit to the antenna. 

In order to use a single antenna for both transmission and reception, 
means had to be provided to effectively disconnect the receiver during the 



















Fig. 9— CXAS— Main unit 

transmitted pulse and to effectively disconnect the transmitter when the 
echo is received. If this were not done, a large part of the transmitted 


energy would be dissipated in the receiver. Also, the minute received 
energy would be partially lost in the transmitter output circuit thus re- 
ducing the maximum range. Because of the extremely short time in- 
tervals between transmitted and received pulses, ordinary switching methods 
cannot be used. A duplexing technique mentioned earlier was therefore 
developed to provide this function. In the CXAS Radar this switching 
was obtained by connecting the transmitter and receiver to the antenna 
transmission line through adjustable lengths of coaxial line which were 
preset for a given operating frequency to be effectively an odd multiple of 

Fig. 10 — CXAS — Modulation or pulse generator 

one-quarter wavelength long. During the transmitted pulse, a small 
amount of the transmitted power overloaded the first tube in the receiver 
and provided a low impedance at that point. Due to the line length 
between receiver and junction point this low impedance is reflected as a 
high impedance at the junction point with the result that very little power 
is lost in the receiver line. At the end of the transmitted pulse the output 
impedance of the transmitter consists only of the small inductance of the 
output coupling loop and this impedance is reflected by proper choice of 
line length as a very high impedance at the junction joint with the receiver 
line. Thus, most of the received echo is directed to the receiver input 





Fig. 12 — CXAS — Power control panel 

Fig. 13— CXAS— Transmitter 



circuit. The adjustable duplexing transmission lines may be seen in the 
side compartment of the Main Unit on Fig. 9. 

The equipment just described could be operated over a small frequency 
band of about 40 megacycles in the neighborhood of either 700 or 500 meg- 
acycles. The transmitter, receiver, and duplexing circuits were tunable 
over the entire range, but it was necessary to set up the antenna for one 
band or the other. This was accomplished by installing the proper one of 
the two sets of dipoles furnished, and installing or omitting a set of wedges 

Fig. 14— CXAS— High voltage rectifier 

which tilted the reflector wings to change the parabola focal length. This 
antenna set a precedent in design in that the dipoles and coaxial harness 
were designed for fairly broadband operation and were entirely free from 
field tuning adjustments which had been very troublesome in earlier 

The CXAS Radar was demonstrated to the Navy in December 1940. 
After a few tests it was decided by the Navy to standardize on the 700- 
megacycle band. One of the principal reasons for this was that the tests 
had proved the superiority of shorter waves for surface target work; the 


CXAS having regularly out-performed much higher powered equipment 
operating at 100 or 200 megacycles for this service. The reason for this 
can best be uuderstood by reference to Fig. 15 which illustrates what happens 
when a radio beam is directed horizontally over water. The beam breaks 
up into an interference pattern of several rays due to reflection from the 
surface; the position of the lowest ray depending only upon the height of 
the antenna measured in wavelengths above the water. Since the mount- 






Fig. 15 — Effect of surface reflection on elevation beam 

Table I 

Operating Frequency Tunable 680-720 mcs. 

Antenna Dipole array of 8 half-wave radiators, reflector 6' x 6', 

beam width 12 degrees, gain 22 db. 

Transmitter Pulse Power Approximately 2 kw. 

Pulse Repetition Rate 1640 PPS 

Pulse Duration Variable in 5 steps from 1 to 5 microseconds. 

Receiver-Superheterodyne 1 mc bandwidth, 30 mc IF frequency. 

Receiver Noise Figure Approximately 24 db. 

Range Calibration Electronic marks at 10,000 and 2.000 yard intervals. 

ing height available aboard ship is fixed, the use of shorter wavelengths 
made it possible to keep the lowest ray more nearly horizontal where it 
could intercept a target's superstructure at greater distance. 

The principal characteristics of the CXAS Radar as set up for operation 
at 700 megacycles are given in Table I. 

This equipment gave useful results on surface targets at ranges of 10 
miles or more (depending on the size of the target) and the range accuracy 


of about ±200 yards was then considered very usable in surface target 
fire control. The target azimuth could also be determined to a precision 
of one or two degrees by rapidly swinging the antenna back and forth and 
observing the point which gave a maximum echo signal. This angular 
information was hardly good enough for fire control use. The equipment 
was also of some use against low flying aircraft as a means of getting better 
range data for fire control. Minor equipment difficulties were not entirely 
solved; in particular the doorknob triodes in the transmitter had a very 
short life under the high voltage pulse operating conditions. They had, of 
course, been designed originally for CW communication use and strenuous 
development effort to make them more suitable for the intermittent high 
power radar use had not been very successful. 

The Mark 1 Radar 

In spite of the obvious unsolved development problems the Navy im- 
mediately ordered 10 equipments, similar to the CXAS, for use in the 
Fleet. These were first called the FA Radio Ranging Equipment but the 
designation was later changed to Radar Mark 1. Several changes were 
made to better adapt the equipment for installation aboard ship, the princi- 
pal one being a servo driven antenna pedestal of the amplidyne type which 
was furnished by the General Electric Company. The servo system elim- 
inated the antenna drive shaft problem while retaining control from a 
handwheel on the control desk. The desk was also modified to provide 
dials reading both relative and true azimuth bearing, the latter being ob- 
tained by interconnection with the ships gyro compass system. 

The first Mark 1 Radar was shipped by the Western Electric Company 
in June 1941 and installation on the USS Wichita was completed at the 
Brooklyn Navy Yard early in July 1941. This was the first fire control 
radar in our Fleet and the first of many thousands of radars of all types 
which the Western Electric Company was destined to build for the Navy 
in the following four years. 

The Mark 2 Radar 

While the ten Mark 1 radars were being built, development work was 
proceeding at top speed on major improvements designed to increase per- 
formance, eliminate operating troubles, and to make this new device fit 
better into the existing fire control situation aboard ship. The older opti- 
cal devices were neatly integrated into a system, many features of which 
were automatic. For example, the gyro stabilized telescopes and optical 
range finder were assembled into a compact rotating armored box called a 
director, located high on the ship. Target data from the director was sent 


automatically by synchro data transmitters to the computer below decks, 
which solved the fire control problem and likewise transmitted automatically 
the correct information to the guns. For the new radar target locating 
device to fit into the existing system it was necessary to make its angle 
finding function operate more in the manner of the telescopes. Not only 
was it desired to determine target angles more accurately but it was nec- 
essary to track target position continuously and smoothly. Finally, to 
take care of the anticipated need for rapidly changing back and forth during 
an engagement from optical to radar data it became apparent that the 
same operators should handle both jobs. Thus it was decided that the 
system should provide the existing operators with oscilloscopes to sup- 
plement their telescopes, and to arrange them so either could be used as 
desired. Further to coordinate the data it became obvious that the radar 
antenna should be connected with the optics in such a way that the two 
were always pointed in the same direction. This would make it possible 
to leave the existing data transmission system alone and would avoid any 
break in data when changing from optics to radar or vice versa. For 
example, if a visible target disappeared behind a fog bank the telescope 
operator would simply move his head to look at his oscilloscope and data 
would continue to flow smoothly to the computer and to the guns. 

Thus the engineers of the Navy decided the new radar device could be 
fitted into the existing fire control system. Any other decision would 
likely have required modification of many parts of the system, and would 
have delayed the extensive use of fire control radar by a matter of years. 
The Bell Telephone Laboratories were accordingly asked to modify and 
improve the radar design to make possible the coordination of optics and 
radar as just discussed. The new radar was to be called Mark 2 and was 
to be similar to the Mark 1 but modified to provide continuous tracking in 
azimuth with an accuracy of ±15 minutes of arc, and continuous tracking 
in range with an accuracy of ±50 yards. Further, the operator's oscil- 
loscopes and controls were to be put into small units that could be mounted 
alongside of the telescopes in the director, and the antenna was to mount 
on the director. These requirements demanded some important forward 
steps in radar development which will be described in some detail. Before 
Radar Mark 2 got into production a much higher powered transmitter was 
developed and with this change the equipment was re-named Radar 
Mark 3. 

The Mark 3 Radar 

The general arrangement of apparatus for this radar differed from the 
Mark 1 principally in the indicators, which were designed to mount in the 



already crowded gun director. These indicators are shown in Figs. 16, 17 
and 19. Fig. 16 shows the range operator's oscilloscope, called the Control 
and Indicator, which was located near the optical range finder. Adjacent 
to this unit was mounted the range unit, shown in Fig. 17 by means of 
which the operator could select the target to be followed and continuously 

Fig. 16— Control & indicator— Radars Mark 2, 3 & 4 

maintain accurate range readings. A typical installation of these two 
units is shown in Fig. 18. The third unit, shown in Fig. 19, is called a 
Train or Elevation Indicator and, in Radar Mark 3 (which was for sur- 
face fire only) this indicator was mounted adjacent to the Train (azimuth) 
Operator's telescope. 
In addition to the Train Indicator, the azimuth operator was provided 



Fig. 17— Range unit— Radars Mark 2, 3 & 4 

Fig. 18— Mark 3 Radar — Range operator's position on Cruiser Honolulu (Navv Photo 




with a Train Meter of the zero center type which indicated the direction of 
deviation from true target position. One of these meters of early design 
can be seen in Fig. 38. Two meters of later design are shown in Fig. 39 
mounted immediately below optical telescopes. The pulse generator, 
receiver, transmitter, rectifiers, etc., were located below decks in the Trans- 

Fig. 19— Train or elevation indicator— Radars Mark 2, 3 & 4 

mitter-Receiver, or Main Unit which was very similar in appearance to 
the Main Unit of Radar Mark 1 shown in Fig. 9. 

Two types of antennas were provided for this radar: a 6 ft. by 6 ft. 
parabolic array similar to the Mark 1 antenna, and a 3 ft. by 12 ft. parabolic 
array. Either one or the other of these antennas was mounted on top of 
the gun director and rotated with it in azimuth. Both were provided with 
azimuth lobe switching to be described later. Because of the relatively 
narrow elevation beam of the 6 ft. by 6 ft. array, this antenna required gyro 



stabilization in elevation to take care of pitch and roll of the ship. Such 
stabilization was not required with the broad elevation beam obtained with 
the 3 ft. by 12 ft. antenna; and in addition, this wider antenna provided 
more accurate tracking due to the narrower antenna beam in azimuth. 
Installations of these antennas aboard ship are shown in Figs. 20 and 21 
and 22, 23 and 24. 

Fig. 20— Radar Mark 3 antenna (6' x 6') on Cruiser Honolulu (Navy Photo 144-6-42) 

Antenna Lobe Switching 

The problem of measuring angles accurately with a relatively broad 
radio beam has been faced many times in the radio direction finding art. 
The most successful attack has made use of the fact that while the nose of a 
radio antenna beam is blunt, the sides of the beam are relatively steep; i.e., 
while the rate of change of signal amplitude with angle is very low near the 
nose of the beam it becomes substantial down on the side of the beam. 
A very well known application of this principle is the airway radio range 
wherein two very broad overlapping beams define a narrow path by uti- 
lizing the points where the two overlap with equal intensity. A somewhat 
similar scheme in which the antenna beam is switched rapidly between 
two positions has been applied in radar, and in an early form was first used 



in this country by the Signal Corps in the work described by General Colton, 
to which reference has been made. 

The use of two^antenna beam (or lobe) positions to obtain more accurate 
radar angle d$0jjj$ referred to as lobe switching and the operating principle 
is illustrated in Fig. 25. The antenna beam is shown in two positions: 
position 1 being directed to the right, and position 2 to the left of the me- 
chanical axis of the antenna. The antenna beam is caused to switch rapidly 
between these two positions, and simultaneous with this switching a small 
horizontal displacement of the indicator Class A sweep is introduced. In 

Fig. 21— Radar Mark 3 antenna on Destroyer Porter (Navy Photo 2711-42) 

this manner the signals received in the two beam positions may be viewed 
separately. The speed of switching is made sufficiently high to minimize 
flicker and the effect of fading signals. It will be noted from this diagram 
that the signal strength received from target A is the same for both beam 
positions thereby producing equal "pip" heights on the indicator screen. 
However, for target B the signal amplitude is greater in position 1 than in 
position 2 and the "pip" amplitudes on the indicator differ correspondingly. 
If the operator wishes to track target B it is only necessary for him to 
rotate the antenna until the two "pips" are of equal amplitude. Smooth 









flow of azimuth data will be obtained if the operator continuously maintains 
equal amplitude of the two "pips". 

In the Signal Corps equipment to which reference has been made, sep- 
arate antennas were used for transmission and reception with lobe switching 


Fig. 24— Radar Mark 3 antenna (3' x 12') on Cruiser San Francisco after Pacific battle 

(Navy Photo 34133) 

applied only to the receiving antenna. Space limitations aboard ship 
made it mandatory to accomplish all functions using a single antenna. 
This required the development of a lobe switching device capable of with- 
standing the high peak power during the transmitted pulse; a problem 
which had not been faced in the Signal Corps equipment. It was further 
desired to provide a weatherproof lobe switching device, free from radio 










Fig. 25— Principle of lobe switching 





Fig. 26 — Mark 2 & 3 — Antenna schematic 

frequency adjustments, in order to simplify operation and maintenance. 
The manner in which these objectives were met is described below. 

To obtain lobe switching of the antenna beam, use was made of the fact 


that the beam position depends upon the relative phase of the excitation 
applied to the radiating elements of the array. If all elements are excited 
in phase, as in Radar Mark 1, the beam will be normal to the line of the 
array, while gradually increasing phase difference across the array will 
result in displacement of the beam. For small angles of beam shift, en- 
tirely satisfactory results may be obtained by shifting the phase of excitation 
applied to one-half of the array with respect to the other, and this expedient 
results in a much simpler phase shifting mechanism than would be required 
to obtain uniform phase change. This system was used in Radar Mark 
3 and its application is illustrated schematically in Fig. 26. It will be 
seen that this array is identical to that used for Radar Mark 1 except for 
the central section of transmission line in which a lobe switching unit has 
been added. In this unit the phase of excitation to one-half of the array 
is retarded with respect to the other half by connecting a capacitive re- 
actance alternately across one feed line or the other to obtain the two beam 
positions. Switching is accomplished by the use of a motor driven rotary 
capacitor shown in Section A-A. The rotor is a semicircular aluminum 
casting which is maintained at substantially ground potential by very close 
spacing to the grounded metal housing. The two stators are small metal 
plates which interleave with the rotor during approximately one-half 
revolution and are connected through half- wavelength coaxial lines to the 
antenna transmission lines. The purpose of the half-wavelength stub lines 
is to avoid physical limitations which would otherwise be encountered in 
connecting the rotary capacitor to the lines. Allowance is made in these 
stubs for end-loading caused by stray capacitance of the stator plates and 
supporting insulators. It will be seen that during nearly one-half revo- 
lution of the rotor one of the stators is engaged to shift the antenna beam in 
one direction while during the other half revolution the other stator is 
engaged to produce the other lobe position. The switching occurs during 
the small interval in which both stators are engaged by the rotor. Signals 
received during this interval are blanked out in the indicator. The rotor of 
the lobe switcher is driven at about 30 RPS by an induction motor mounted 
within a weatherproof housing. The motor shaft also carries cam operated 
contacts to produce image spacing on the indicators, control signals for the 
Train Meter, and blanking during the lobe switch interval. The entire unit 
is gas tight and is filled with dry gas through the transmission line. 

The value of the lobe switching capacitor and its position along the feed 
line must satisfy two conditions: first, the phase shift must be such that the 
antenna beam will be displaced by the desired amount; and second, the 
impedance at the feed point must be such that equal division of power will 
be obtained in the two halves of the array. In the first Radar Mark 3 
antenna (6 ft. by 6 ft. parabolic array) a beam displacement of about 3.0 



degrees was chosen as a suitable compromise between target angle sen- 
sitivity (steepness of beam) and reduction of signal amplitude "on target". 
This displacement required a phase shift of approximately 53 degrees 
between the two halves of the array. From transmission line theory it can 
be shown that this phase difference will be obtained with a capacitive 
reactance equal to the characteristic impedance of the feed line when con- 
nected at a point 0.176 wavelength from the feed point. It can also be 
shown that this condition satisfies the requirement for equal power division 
to the two halves of the array. A capacitor of the required value (about 
3 micromicrofarads) can readily be built to withstand the peak transmitted 
power by proper condenser plate separation. A frequency variation of 
about 40 megacycles can be tolerated without materially affecting the 
antenna performance. 

Table TT. — A ntenva Characteristics 


Apperture in Wavelengths 



Beam Width in Degrees (hetween half power 
points in one way pattern) 



Antenna Cain in db 

Beam Shift in Degrees 



Radai Mark 4 





A lobe switching unit similar to that described above was also applied to 
the 3 ft. by 12 ft. antenna. Pertinent information regarding beam widths 
and lobing angles for both antennas (together with information on the 
antenna for Radar Mark 4 to be described later) is given in Table II. 

The effective beam widths as used in these radars were somewhat narrower 
than the values given above due to the square law characteristic of the 
second detector in the receiver, and the deflection sensitivity was such that 
the specified tracking accuracy of ±15 minutes of arc could readily be 
achieved. The "on target" position or axis of the antenna (lobe crossover) 
was carefully aligned with the optical telescopes at the time of installation 
so that either optics or radar angles could be used. The symmetrical design 
of the antenna made this alignment substantially independent of small 
changes in operating frequency. 

To minimize target confusion the signals presented on the Train or 
Elevation Indicator (azimuth operator's oscilloscope) consisted only of 


those received from the target being tracked by the range operator, all 
others being blanked out in the indicator circuits. 

Accurate Range Measurement 

The second major problem which required solution to adapt radar to the 
fire control problem was the provision of means for accurate and continuous 
range tracking. It was obvious that what was required was some sort of 
electronic range mark on the indicator sweeps, the position of which could 
be varied by a rotary device whose motion could be used to transmit range 
information to a remote point over a synchro system. The range mark 
could then be aligned with the target "pip" on the oscilloscope. For 
accurate data transmission it was necessary to obtain a linear relationship 
between angular rotation of the range handwheel and corresponding range 
to the marker on the radar indicator screen. 

One method which was first employed by the Signal Corps made use of 
the fact that the transmitted pulses were generated at a periodic rate from a 
sine wave oscillator of fixed frequency; the pulse being produced at a fixed 
point in each cycle. By transmitting this same sine wave through a linear 
phase shifter a new pulse could be generated whose position in time, rel- 
ative to the transmitted pulse, could be varied by rotation of the phase 
shifter. In the Signal Corps equipment a special goniometer was used to 
produce the phase shift and the accuracy obtained was considered adequate 
for the intended purpose. However, non-linearity of the phase shifting 
device, though small, was much greater than could be tolerated in the 
Navy fire control system. A study indicated that large scale manufacture 
of special phase shifters, hand adjusted to meet the stringent accuracy re- 
quirements was out of the question. It was therefore decided that a two 
speed system be used, in which the phase shifter errors would be divided by 
the gear ratio to the high-speed unit in much the same way that accurate 
synchro information is transmitted by a "coarse" and a "fine" synchro. 
The manner in which this was worked out by Bell Telephone Laboratories 
and applied to Radars Mark 3 and 4 is described below. 

The method of range measurement can perhaps best be understood by 
first examining the method of presentation used on the cathode ray tube 
indicator for the range operator. This presentation is shown in Fig. 27 
in which it will be noted that a Class A sweep is used to display the trans- 
mitted pulse and received echoes. This horizontal sweep, however, differs 
from the simple sweep of earlier radars in several respects. First, the 
central portion of the sweep is expanded to permit more accurate viewing 
of signals appearing within this region; second, a downward deflection 
called the range "notch" is produced in the approximate center of the ex- 
panded section; and third, the circuits are so arranged that the notch 


remains centered as the range unit phase shifters are rotated thus causing 
all of the signals (rather than the notch) to move across the screen. Range 
measurement is made by rotating the range unit handcrank to place the 
desired signal in the center of the range notch on the indicator. This 
type of presentation has several advantages. It permits the full 100,000-yard 
range to be viewed at all times so that new targets may be immediately 
detected, and permits accurate viewing of the desired target in the expanded 

Fig. 27 — Mark 3 & 4 — Range presentation 

center of the sweep where best focus is obtained. For smooth range track- 
ing it is only necessary for the operator to rotate the range unit hand crank 
to keep the desired signal centered in the range notch. 

A block diagram of the range measuring system, together with the circuits 
used to obtain the cathode ray indicator presentation described above, is 
shown in Fig. 28. A base or reference oscillator generates a sine wave of 
1.639 kc, one cycle of which corresponds to a radar range of 100,000 yards. 
This wave, after amplification, is applied to a non-linear coil pulse gene- 
rator 3 which generates short pulses (one positive and one negative pulse 

3 "Magnetic Generation of a group of Harmonics," E. Petersen, J. M. Manley, L. R. 
Wrathall— August 1937, B. S. T. J., October 1937. 




per cycle) ; the positive pulses being used for keying the transmitter. These 
pulses are rich in odd harmonics of the base oscillator frequency. By 
rectifying these pulses to reverse the negative pulses, even harmonics of 
the base frequency are obtained and the 18th harmonic (29.5 kc) is se- 
lected by means of a filter. This harmonic frequency and the original 
base frequency are applied to two phase shifters whose shafts are geared 
together in the ratio of 18 to 1. Since one revolution of the one speed 
phase shifter corresponds to 100,000 yards, one revolution of the 18-speed 
unit corresponds to only 5550 yards with the result that range errors caused 
by non-linearity of this phase shifter are reduced by a factor of 18. The 
phase shifters employed are similar to those designed by Bell Telephone 
Laboratories for use in a phase measuring bridge 4 and are linear to within 
±1.5 degrees or about 0.4 per cent. The possible range error introduced 
by imperfections in the 18-speed phase shifter was therefore only 23 yards, 
well within the design requirements. It remains to be shown how this 
accurate range information was applied to the indicator. 

The output of the 18-speed phase shifter in the range unit is connected 
to the Control and Indicator where the phase shifted sine wave is used to 
generate short, rectangular pulses of about 600 yards duration. One pulse 
is produced for each cycle of the 29.5 kc wave so that 18 of them occur during 
the 100,000-yard sweep interval. It is desired that only one of these pulses 
appear as a range notch on the indicator screen and this pulse is selected 
from the others by a pedestal pulse generated from the output of the one 
speed phase shifter. It will be noted that as the phase shifters are rotated 
by means of the range unit hand crank, the desired pulse from the 18-speed 
phase shifter will remain substantially centered on the one-speed pedestal 
pulse. After further shaping, the selected pulse is mixed with the received 
signals in the second video amplifier and is then applied to the vertical 
plates of the cathode ray indicator to form the "range notch". The range 
notch is also transmitted to the Train Indicator and Train Meter where it 
is used to prevent any signal from affecting those instruments except the 
one being tracked by the range operator. 

Since it is desired to have the range notch appear in the center of the 
100,000-yard sweep on the indicator, the sweep trigger pulse must occur 
50,000 yards in advance of the notch. This trigger is obtained by se- 
lection of another pulse from the accurate phase shifter, this time using a 
one-speed pedestal produced by an input of reversed phase. The pulse 
thus selected is used as a trigger for starting a saw-tooth sweep wave with a 
duration corresponding to 100,000 yards radar range. Expansion of the 
center portion of this sweep is obtained by adding to this wave a second 

* L. A. Meacham, U. S. Patent 2004613. 



saw-tooth wave having maximum rate of change in the center of the sweep; 
the latter being derived from the range notch selection pedestal. The 
combined sweep is then applied to the horizontal plates of the cathode ray 
tube. The return trace is blanked by applying to the control grid of the 
cathode ray tube a voltage obtained by differentiating the sweep waveform. 


As mentioned earlier the transmitter oscillator tube problem was one 
of the major obstacles in the march of radar development to higher fre- 
quencies. Intense development effort on many possible types of tubes was 
underway in several laboratories in this country and abroad during 1939 

Fig. 29 — VV. E. 700-type magnetron — one side removed 

and 1940. The first significant improvement came in England where work 
with multicavity magnetrons showed that this device was probably the 
answer to radar's need for a highly intermittent duty oscillator suitable for 
high power in the microwave region. A sample of this device was brought 
to this country by the Government and was tested in Bell Telephone Lab- 
oratories in October 1940. It produced pulses of several kilowatts at a 
frequency in the neighborhood of 3000 mc. A tremendous development 
of this device got under way immediately 5 and the multicavity magnetron 

6 "The Magnetron as a Generator of Centimeter Waves," J. B. Fisk, H. G. Hagstrum, 
and P. L. Hartman, B. S. T. J., January, 1946. 



became the key piece in the enormous development of radar equipment for 
still higher frequencies during the war. However, at the beginning of 1941 
there were still many unsolved problems in 3000 mc radar other than that 
of the transmitter tube. On the other hand, the systems problems had 
been quite satisfactorily solved in the 700-mc region. The decision was 
therefore immediately made to extrapolate the British design down to 

Fig. 30 — Mark 3 & 4 — transmitter 

700 mc in order to obtain a higher powered and more satisfactory oscillator 
for the existing systems. This was done at top speed and a picture of the 
resulting tube is shown in Fig. 29. This was the first type of multi-cavity 
magnetron to go into production in this country. Concurrent with the 
design of the new magnetron the vacuum tube department of the Lab- 
oratories developed an improved tetrode modulator tube which was many 
times as efficient for radar pulse service as the triodes formerly used. This 
tube was designated W.E. 701-A. 
A new transmitter using the magnetron and two of the new modulator 


tubes was rushed through development and produced in time to go with 
the first accurate fire control radars. This transmitter provided a peak 
power output of about 40 kw with a pulse duration of 2 microseconds. It 
resulted in a material increase in reliable range, with satisfactory tube life. 
The new transmitter, shown in Fig. 30, was made mechanically inter- 
changeable with the old and was applied retroactively also to the Mark 1 


The use of the high-power transmitter required additional protection 
for the receiver during the transmitted pulse in order to prevent damage 

Fig. 31— W. E. 702— TR tube 

and to permit the receiver to recover rapidly for reception of nearby echoes. 
The duplexing equipment was therefore modified to inclu de a gas switching 
tube in the receiving transmission line. This was a refinement of the 
method used earlier by the Naval Research Laboratory. 

The switching tube (W.E. 702A) was developed specifically for this 
purpose and is shown in Fig. 31. It was the first of the "TR" tubes of this 
general form and consists of a hydrogen-water vapor filled glass chamber 
with three copper electrodes. 6 This tube was mounted in the center of a 
half-wavelength coaxial line short circuited at each end, the outer con- 
ductor being connected to the outer electrodes and the center conductor 

6 "The Gas Discharge Transmit-Receive Switch," A. L. Samuel, J. W. Clark, and W. W. 
Mumford, this issue of B. S. T. J. 


to the middle electrode. Input and output connections were tapped on 
this half-wave line near the short circuited ends. During reception this 
assembly introduces negligible loss in the receiving line. However, during 
the transmitted pulse a small amount of the transmitted power ionizes the 
gas in the switching tube and effectively short-circuits the receiver line. 
This device, which in later forms came to be called a "T-R Box", is located 
near the receiver input and the length of line between it and the junction 
with the transmitter line can be adjusted to an odd multiple of quarter 
wavelengths to present the desired high impedance at that point during 


The receiver delivered with early Mark 3 equipments was identical to 
that used in Radar Mark 1. It was of the superheterodyne type em- 
ploying one stage of RF amplification (doorknob tube), 316A oscillator tube, 
and doorknob first detector. The intermediate frequency amplifier had a 
bandwidth of about 1 megacycle at a midband frequency of about 30 mega- 
cycles. The second detector and video stages were located in the indicating 
equipment. A photograph of this receiver is shown in Fig. 11. 

Since in microwave work the controlling noise is that produced in the 
receiver, it is desirable to reduce this noise to the theoretical limit of thermal 
agitation in the input circuit. However, in 1939 tube limitations and 
circuit design techniques at these frequencies resulted in performance far 
short of this goal. The amount by which the receiver noise exceeds the 
theoretical minimum has been termed the receiver "noise figure" and in 
this early receiver the noise figure was about 24 db. It was recognized that 
considerable improvement in maximum range could be obtained by reducing 
this receiver noise. 

Shortly after first deliveries of Radar Mark 3 a new tube (GL-446 or 
"lighthouse" tube) was made available by the General Electric Company 
which showed promise of providing a substantial improvement in the 
receiver noise figure. An amplifier using this tube was accordingly de- 
signed by Bell Laboratories in which coaxial cavities were used for tuning 
elements. Two stages of amplification were used to replace the single 
"doorknob" tube stage previously employed. The new amplifier resulted 
in a reduction of the receiver noise figure to about 9 db and provided a 
marked improvement in maximum range capability of the radar. These 
amplifiers were manufactured and shipped to the Fleet for field installations 
on early equipments and were included in productions on equipments 
shipped subsequently to availability of the amplifiers. A photograph of 
the receiver with the two amplifiers installed is shown in Fig. 32. 

Another field modification provided automatic gain control of the signal 













selected by the range operator. This was supplied in the form of an ex- 
ternal unit which controlled the gain of the receiver IF amplifier to reduce 
signal fluctuations produced by fading. 

The first production Mark 3 Radars were delivered to the Navy in October 
1941, and the first two installations were completed on the main battery 
directors of the U.S.S. Philadelphia at the Brooklyn Navy Yard that month. 

Radar Mark IV 

During the development work on Radar Mark 3 the Navy pointed out the 
need for a fire control radar for use with the 5-inch Naval guns against 
enemy aircraft. The Bell Telephone Laboratories was therefore requested 
to further modify the radar design to meet this need. The anti-aircraft 
equipment was first designated FD, later becoming known as Radar Mark 

For antiaircraft fire control a new coordinate had to be added to the 
target- locating system; namely, elevation angle. Again it was desired that 
the additional information be obtained from the single antenna with a 
precision equal to that already obtained in azimuth. This problem was 
approached in a manner similar to that used for the Mark 3 antenna and is 
described below. 

Two Plane Lobe Switching 

In considering two plane lobe switching methods it appeared that the 
desired result could be obtained by mounting two 3 ft. x 6 ft. parabolic 
arrays one above the other. This arrangement was tried and resulted in 
the array shown in Fig. 33. It provided two plane lobe switching with an 
antenna only slightly larger than the 6 ft. x 6 ft. antenna used before and 
had comparable gain and beam width (see Table II). 

A schematic diagram of the array is shown in Fig. 34. Here it will be 
seen that there are two horizontal dipole arrays, each mounted along the 
focal line of a cylindrical parabola. The dipoles are in four groups and the 
interconnecting harness is criss-crossed and joined to the feed line at the 
center. Symmetrically placed around the feed point are four stub lines 
connected to the lobe switcher stators. Here again a semi-circular rotor is 
used for the lobing shifting capacitor. It will be observed that during each 
quarter turn of the rotor two stator plates are engaged, and the sequence is 
such that the beam shifts left, up, right, and down during one rotation. 
A separate Indicator was provided for the Pointer (elevation operator). To 
avoid signal confusion on the two Indicators it is necessary to show only 
left-right signals on the Trainer's oscilloscope and up-down signals on the 
Pointer's oscilloscope. This is accomplished by means of cam operated 
contacts in the lobe switcher which blank the indicators during the required 



intervals. Other contacts on this assembly provide left-right and up- 
down image spacing. 


Fig. 3i — Mark 4 — antenna on gun director 

Except for the new antenna and the additional Train or Elevation In- 
dicator for the Pointer (elevation operator), this radar was identical to 
Radar Mark 3. The first demonstration of a development model of Radar 
Mark 4 was made at Atlantic Highlands, New Jersey, in September 1941 
and this model was installed aboard the destroyer U.S.S. Roe the latter 





Fig. 34 — Mark 4 — Antenna schematic 

Fig. 35— Mark 4 antenna on Battleship Tennessee (Navy Photo 1908-43) 






part of that month. Initial production deliveries of these radars were made 
in December 1941. 

Typical installations of the Mark 4 Antenna on the secondary battery 
directors of a battleship are shown in Figs. 35 and 36. The main frame 
installation for Mark 3 and Mark 4 on a battleship is shown in Fig. 37 

Fig. 37 — Radars Mark 3 & 4 — main units on Battleship New Jersey (Navy Photo 181809) 

while typical installations of the train and elevation operator's units in the 
director are shown in Figs. 38 and 39. 

Application and Use op Mark 3 and 4 Radars 

The Mark 3 radars, designed for use against surface targets only, were 
generally installed on the main battery directors of battleships and cruisers. 
The Mark 4 radars for use against either surface targets or aircraft were 
generally installed on the secondary battery directors of battleships and 
cruisers, and on the one and only dual purpose director on destroyers. 
Thus a battleship usually had two Mark 3 and four Mark 4 equipments and 
a destroyer one Mark 4. Practically every ship in the fleet, of destroyer 



size or larger, was equipped with one or more of these equipments early in 
the war. A total of 139 Mark 3, and 670 Mark 4 radars were built, in- 
cluding those used ashore at schools. Although some of these equipments 
were replaced by more modern designs before the end of the war and some 
were lost in battle, there were still approximately 85 Mark 3 and 300 Mark 4 

J""ig. 38 — Mark 4 Radar — trainer & pointer operators' positions on Aircraft Carrier Saratoga 

(Navy Photo 177347) 

radars in service in the fleet on V-J day. The first four Mark 4's, Serial Nos. 
1, 2, 3 and 5 installed on the battleship Washington were used until the 
middle of 1945, although newer designs had been going on all new vessels 
for more than a year. 

These early equipments were the "guinea pigs" of fire control radar. 
They were the instruments with which our fleet learned to fight effectively 



at night and thereby gain a large advantage over the enemy whose radar 
was feeble and inaccurate. They played a part in every one of the early 
battles and most of the later ones in the Pacific. They controlled the cruiser 
Boise's guns in October 1942, when she blazed away at night at a vastly 
superior fleet in the Solomons and made the enemy pay 10 to 1 for the 

Fig. 39 — Radar Mark 4 — trainer & pointer operators' positions on Destroyer Barton 

(Navy Photo 181775) 

damage they succeeded in doing. They were with the cruiser San Francisco 
on a night in November 1942, when a small U. S. force sank 27 enemy ships, 
almost completely destroying a large Japanese convoy bound for Guadal- 
canal when our hold there was at best precarious. The Mark 3 steered 
the big guns of the battleship South Dakota in the Solomons on the dark 
night of November 4, 1942, when she sank a major Japanese war vessel 
eight miles away with two salvos. Even in engagements in broad daylight 
when optics could be used for target angles these radars still played a vital 


part in furnishing accurate range which made 5" gunfire against aircraft, 
for example, deadly at long range. Thus on October 16, 1942, when the 
South Dakota was attacked by planes she shot down an even 38 out of 38 

The rapid and widespread application of this rather complex electronic 
equipment was not accomplished without pain and confusion. It is beyond 
the scope of this paper to discuss the enormous problem of training in 
operation and maintenance that had to be solved, or of the tactical revo- 
lution in Naval warfare that fire-control radar produced. It is sufficient 
here to say that these and other problems were solved by heroic efforts of 
hundreds of officers and civilians in the Navy Department ashore and the 
thousands of officers and men of the fleet. Their problems were made more 
difficult by weaknesses in the equipment which were revealed by battle 
experience as the new science of radar got its baptism of fire. In every 
possible case the Laboratories attempted to remove the causes of recurring 
troubles by redesign and the furnishing of improvement kits of parts for 
installation in the fleet. The many lessons of experience learned from the 
Mark 3's and 4's were immediately applied in the design of the many more 
modern radars for the same and other types of service. 

The authors of this paper wish to express their gratitude to the many 
Navy men with whom they have worked in connection with these equip- 
ments, and whose whole-hearted cooperation during difficult times made 
possible the successful development of these fire-control radars. They 
also wish to thank their colleagues in Bell Telephone Laboratories who 
worked as a team to make this important equipment possible, and the men 
of the Western Electric Company for their help on the many engineering 
problems which arose during production and use in the field. It is the hope 
of all who were concerned with this development that accurate radars, like 
other radars, will find peaceful use in a peaceful world, but it is also the 
determination of these engineers that as long as we need a Navy, we will 
try to provide it with radars as much superior to those of any possible 
enemy as they were in the recent war.