Professional Documents
Culture Documents
Perkins - Development of Pulse Doppler Radar (Westinghouse) IEEE 1984
Perkins - Development of Pulse Doppler Radar (Westinghouse) IEEE 1984
Perkins - Development of Pulse Doppler Radar (Westinghouse) IEEE 1984
Seeker Devetopment S
I/ ~~~
P oeeker, DPN 53 Ae at1
~~~
~
a
< \ //4 \\ <~~~~~~~~~~~~~~~~Flyoff y ALa-153
f BOMAMultiple Fixed S
. f~~~~~~~~~~~~~~~or TakNg|avy F 4J /
*High PRF
Deve
/oprl ent
e Flyl ;
| *~~~~~~~~
PRF Switched
During Ant Bearn cORV-
ec.hnORT"o)g APQ-164
5
I ~~~~~~~~*
Digital Target ~ +*Low
Track While ~~ ~ ~ ~Sidelobe ofwaeControlled
~ ~~* AWACS
)
| ~~~~~~~~~Report & Track , /Suveillance Ant IContracts & FSD and ProductionI
| ~~~~~~~~*
Multiple Target I/*High Power Flyoftt
AN/APY 1
} ~~~~~~~~~Tracking J/Transmitter
§ /| *~~~~~~~~~~~~~~~~~~~~~~~~
Pulse Mode
Interlaced
for Over-Horizon Long Range
* Software Controlled
1950 1960 1970 1980
Fig. 1. Pulse doppler radar.
Author's present addresses: L.C. Perkins, 18843 First Place SW, Manuscript received March 8, 1984.
Seattle, WA 98166; H.B. Smith & D.H. Mooney, Westinghouse, Box
1693, Baltimore, MD 21203. This paper was prepared by the authors and formed the basis for their
talk at the Awards Luncheon at NAECON, May 23, 1984, Dayton, OH.
0018-9251/84/0500-0292 $1.00 (© 1984 IEEE
292 IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. AES-20, NO. 3 MAY 1984
Transmitter Carrier Main-Beam Clutter
.
Forward-Hemisphere -
Sidelobe Clutter
Incoming Target
A. CW Doppler Spectrum
Clutter- Target
Free
Space I
I1
---
g i
Fc - Fr Fc Fc + Fr f
B. Pulse Doppler Echo Spectrum
Fig. 2. Comparison of doppler spectra for airbome cw and pulse doppler radars. a) cw doppler spectrum; b) pulse doppler echo spectrum
THE BOEING BOMARC RADAR ..... Boeing's breakthrough came in late summer of 1951
in an all night brainstorming session between Glaser and
The U.S. Air Force IM-99 Bomarc project was begun Perkins. Mulling over the problems with the aid of
in 1949 in response to a proposal submitted jointly by blackboard sketches, the solution was revealed when they
Boeing and the University of Michigan. varied the parameters on a spectral diagram of
Each Bomarc missile had its own target seeker radar hypothetical doppler target and clutter echoes.
because its 400-mile range took it beyond accurate The explanation starts with Figure 2a which shows:
ground radar control. It was known from the initial CW doppler transmitter carrier (amplitude not to scale);
Boeing/Michigan studies that the airborne radar would the echo from an incoming target as seen from a missile
have to have some form of ground-clutter discrimination platform in level flight; the antenna main-beam ground
to be effective against incoming bombers at all altitudes. clutter with the beam at a depressed angle; and the
The first model of Bomarc (IM-99A) sidestepped the antenna side-lobe clutter from the ground both ahead of
clutter problem by employing a high altitude attack and behind the missile.
geometry which allowed the use of a conventional pulsed For pulsed operation we have the central portion of
radar because targets at a range less than the Bomarc's the spectrum of a pulsed transmitter and adjacent lines
altitude would suffer no clutter interference. spaced at the repetition frequency, Fr; the envelope of the
Meanwhile, Cecil K. Stedman, head of Boeing's array is a sin X/X characteristic. The received signal
Physical Research Unit, addressed the full clutter spectrum in Figure 2b is a similar array except that each
problem. He hired Raymond A. Glaser, who had been "line" is a replica of the doppler-shifted target and
working on doppler radar at MIT's Radiation Laboratory, clutter signals of Fig. 2a. It is clear that if the prf is too
to head a group of mathematicians and radar engineers- low, the target at one line will be overlapped by the
including Leroy C. Perkins-to design an effective anti- clutter from the next higher line.
clutter airborne target seeker. Glaser, from his MIT If the carrier frequency is 101' Hz (X-band) and
experience, believed that the transmitter-spillover noise Bomarc's horizontal speed is 2600 ft/sec the sidelobe
problem of a CW doppler radar would prevent it from clutter will spread 50 KHz. If the target is closing at a
ever achieving the necessary 20-mile detection range on a ground speed of 1000 ft/sec its echo will have a
100-square foot target. Therefore, his group set about maximum additive doppler shift of about 70 KHz.
looking at pulse doppler schemes such as those suggested Therefore, to create the clutter-free doppler space for the
by W. W. Hansen and others at the MIT Radiation target as in Fig. 2b, the prf must be a minimum of 50
Laboratories [5]. KHz plus 70 KHz or 120 KHz!
Since the Air Force was not convinced that the Glaser and Perkins spent the rest of the night
potentially simpler CW doppler system could not be made assessing the implications of this radical departure of prf
to work, they also funded the Ryan Aircraft Company, of from the usual one or two KHz. Items considered were
San Diego CA, to develop a CW seeker. the following: Range determination becomes ambiguous
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. AES-20, NO. 3 MAY 1984 293
beyond about 0.8 mile. This was not a problem, since the to start a Westinghouse generic R&D pulse doppler radar
Bomarc needs only line-of-sight rate to steer a collision program with in-house funding. Also, in mid-1953,
course. If the prf is fixed at one particular value, a Westinghouse received a contract to study the Hawk
closing target echo will be eclipsed periodically by the missile system for the Army. As the study progressed it
transmitter pulse with probable loss of target resulting. was concluded that pulse doppler ws an attractive choice
Eclipsing can be avoided by increasing the prf gradually for the seeker radar. Since an experimental demonstration
and when it gets too high (say 250 KHz) cut it in half seemed highly desirable for both programs, a laboratory
and start over. effort was started in September 1953 with David Mooney
No adequate source of high-power, coherent as the principal investigator. A partial radar was
transmitter pulses was known. The solution was to have assembled and was demonstrated to the Army evaluation
the Varian klystron people put a control grid in one of team in early 1954. The demonstration hardware
their high-power oscillator tubes; then injection-lock the consisted of a fully coherent grid-pulsed, 1-watt-peak
phase of each succeeding rf pulse to the stable local power, 1/12 duty-cycle klystron amplifier fed by a cavity
oscillator (STALO). stabilized microwave klystron oscillator; a 30-MHz i-f
TR tube recovery time was too long to have full receiver; a single narrow band doppler filter, or velocity
receiver sensitivity immediately following the transmitter gate; receiver range gating and range tracking provisions;
pulse. Recovery time can be shortened by pulsing the TR and a conical-scan angle track arrangement. The
keep-alive electrode. demonstration consisted of manually acquiring the signal
By late fall of 1952 a demonstration breadboard of the from an unaugmented moving automobile at about one
high-repetition-rate radar had been completed and mile range, and tracking it in range, velocity and angle.
installed in a rooftop "penthouse" overlooking Boeing While the performance was impressive, the Army
Field where real moving targets were to be seen. A C-47 concluded that a less risky approach was to go with a
transport airplane was tracked from takeoff out to a range CW doppler semi-active seeker.
of about 6 miles. The in-house program was continued, with
Following the penthouse testing, the breadboard set concentration on airborne intercept and AEW
was moved into a Curtis C-46 Commando to measure applications. Airborne intercept (AI) radar was a major
ground clutter from the air. Later the set was again Westinghouse product line and a pulsed doppler (PD)
moved into a trailer and hauled to Fort Lawton version would provide a good base for development.
overlooking Puget Sound where successful chaff and Mooney and Smith were to be part of all Westinghouse
ECM tests were made using a special ECM B-29 from pulse doppler developments with over 25 pulse doppler
Wright Patterson Air Force Base. patents jointly and individually.
In the summer of 1953 the Air Force was convinced The next model [Fig. 3], was configured to fit in the
that pulse doppler was a viable mechanization for the cabin of the DC-3 company aircraft and was instrumented
Bomarc B target seeker, and terminated the Ryan CW with a magnetic tape recorder and scope camera for
backup development work. recording clutter.
Boeing gave a bidders' briefing and requested It had improvements in a number of different ways: It
proposals for the development of a Bomarc B prototype had a klystron transmitter of 100 watts average, and a
radar; RCA's Los Angeles Radar Division was the duplexer to permit use of a single search-track antenna.
successful bidder. It turned out to be a tough Fig. 4 shows a typical block diagram of such a radar.
assignment-four and a half years later RCA was still The development of the higher power transmitter was
having trouble meeting the stringent requirements. a formidable task, and nearly stopped the program.
In the meantime Westinghouse had developed a Klystron amplifiers of that era had to be cathode gated
prototype of such promise that, by the fall of 1958, the with a pulse on the order of 20 kilovolts (at one ampere),
Air Force, with Boeing's concurrence, selected with a pulse repetition rate of over 100 KHz. In addition,
Westinghouse to produce the target seeker, DPN-53, for the jitter requirement on the pulsing was on the order of a
the Bomarc B missile. 301 seekers were delivered. few nano-seconds. Very low ripple voltages had to be
maintained to prevent putting sidebands on clutter.
Eventually, the transmitter was developed, and the
MEANWHILE AT WESTINGHOUSE . system was flown over a four month period during 1955,
and the nature of airborne clutter was verified and
The effort [6, 7] at Westinghouse differed analyzed [8-12].
significantly; the Boeing effort was directed to the single Of all the problems encountered, the one requiring the
application of the Bomarc target seeker, while the most work was to get a receiver subsystem that was free
Westinghouse effort was devoted to a wide variety of of spurious signals, or "birdies", caused by harmonics of
applications. In addition, there were a number of the prf, or high order mixer products. The necessity to
fundamentally different mechanization approaches. perform range gating and "main bang" blanking early in
In April 1953 Harry Smith persuaded top Defense the receiver before single sidebanding, makes birdie
management in Baltimore (Nick Petrou and Sy Herwald) freedom very difficult. This appeared to be impractical
Platform Antenna
Motion Position
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. AES-20, NO. 3 MAY 1984 295
with a continuously variable prf, such as required in the
Boeing approach. Our solution was to use discrete fixed
prfs, all of which were harmonically related to the i-f
center frequency. Even with this, several receiver local
oscillators must vary in frequency in order to position
main beam clutter into a notch filter, or to position a
target in a narrow band-pass filter. Great care in the
selection of frequencies, and in choosing filter
characteristics were an essential part of achieving birdie-
free operation.
While the elimination of birdies, and the achievement
of a high-prf transmitter were the major difficulties with
the early pulse doppler radars, there were a number of
lesser problems:
One was getting a suitable duplexer which would
protect the receiver, but be fully recovered in a
microsecond or so, since weak targets from long range
can appear just after a transmit pulse. In the first high-prf
transmitter, the duplexer consisted of a ferrite circulator
and a ferrite switch to gate off the receiver during
transmission. The switch provided enough isolation, but
had dredful drift problems, even at room temperature. A
testing program for a suitable TR gas tube turned up one
brand of one type which worked well enough at the high
prf and low peak power to be useful. In the meanwhile,
the Westinghouse tube laboratory worked on a gas tube
especially aimed at the problem. Suitable tubes were
eventually developed, and all subsequent Westinghouse
radars have used such tubes.
Another difficult problem was getting a flight worthy
STALO [131, that would not spread main beam clutter
due to vibration-induced sidebands and pick-up. The early
stalos were cavity stabilized klystron oscillators, with
special shock and vibration isolation, and were adequate
for the fixed tuned klystron amplifiers. Much later, when
suitable components became available, stalos became
frequency synthesizers, that were channel tunable to be
compatible with wide band TWT's.
A rather mundane problem was developing a suitable
r-f target simulator for system sensitivity minimum
detectable signal checking. Our initial simulator was a Fig. 5. First U.S. Air Force X-Band Al radar (1957)
hom fed crystal detector, which was modulated by an
audio amplifier for doppler simulation. While it worked,
it was unsatisfactory as a calibrated MDS test. A motor
driven belt was trid next, with little reflectors attached to
the belt, but the reflectors flew off at high speed (the first
"flyoff"). Finally, we found that a serrodyne provided
the solution by giving a known amplitude at a set- a mechanically rotating switch which used a stream of
frequency, and compatible with multiple prfs. mercury as the contactor. It worked, but as the mercury
System timing stability was a problem with the got dirty, the switch got very noisy. Later versions solved
vacuum tube technology of the time and was solved by the problem with semiconductor switches.
dc on critical heaters and care in wiring, grounding and The detection threshold in those days was a fixed
shielding. All pulses were generated from a common high level, and the noise level was set manually. The lack of
frequency crystal oscillator, and used start and stop constant false alarm rate (CFAR) technology forced one
triggers. to set the threshold higher than desired, so the early
The doppler filter bank required some kind of systems generally had sensitivities [14] less than
commutating switch, to sequentially test the output with a theoretical. Later radars used a cell-averaging CFAR
threshold. The best device we could find at that time was which solved the problem.
U.S. AIR FORCE INVOLVEMENT ..... 8) Mainbeam clutter acquisition and clutter track was provided to
work with clutter reject filters and controlled frequency oscillators to
As a result of these developments, the Air Force -notch out" main beam clutter.
decided to fund Westinghouse to build advanced models 9) It was provided range-vs-azimuth,a nd doppler-vs-azimuth
displays as well as a "pause-to-measure-range" to get range for the
of Al radars, which would demonstrate full functional search display.
capability. A breadboard X-band model [Fig. 5] similar
to the previous flyable model, and a C-band model [Fig. There were several state-of-the-art limitations. At the
6] with extended range performance and pre-prototype time wide-band TWT power amplifiers, suitable
packaging were contracted in July 1956 (AF transistors or integrated circuits were not available. Due
33(616)3700). The Westinghouse Al radar had a number to hardware size, only one or two receiver channels
of significant differences from the Boeing Bomarc seemed feasible for all but ground application. Rapid
approach: recovery TR gas tubes were yet to be developed. Despite
these limitations, working radars were realized.
I ) It used a dual mode transmitter, with a high duty cycle (0.46) for
search, and a lower duty cycle (0.1) for track, to optimize detection The X-band radar was accepted by the Air Force in
range, while measuring true range in track. October 1957, after demonstration of aircraft acquisition
2) The transmitter was a klystron amplifier rather than an injection and tracking from the rooftop lab, and flight tested in the
locked klystron oscillator. DC-3 in 1958. Meanwhile, the high power C-band APG-
3) It measured true range by comparing ambiguous range from 55 was built, and was flight tested in a B-66 in 1959.
several fixed prf's. [15]
4) The multiple prfs distributed '-eclipsing' (due to the necessary
receiver blanking) during ''search". EARLY NAVY INVOLVEMENT .....
5) These multiple prf's were fixed, as a necessary step in avoiding
spurious signals due to harmonics and PRF lines from clutter. In 1957 the Navy began to cosponsor the PD effort.
6) All prfs and i-f local oscillators were derived from a single
source, to eliminate spurious signals. This resulted in a long range track-while-scan pulse
7) It used a bank of narrow band doppler filters to speed up the doppler radar, the APQ-81 [Fig. 7] which was started in
search mode. May 1957 (NOas 51-7500). The APQ-81 was intended to
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. AES-20, NO. 3 MAY 1984 297
Fig. 7. U.S. Navy track-while-scan pulse doppler radar, AN/APQ-81 (1960)
be used on the AWS-404 or ''missileer" airborne simplified version of the Al radar, and had the following
multishot fleet defense system; that system was characteristics:
discontinued but the radar development continued.
Rooftop simultaneous tracking of multiple airborne targets I) It used a fully coherent klystron amplifier transmitter, with a grid
was achieved and later demonstrated in flight test, for pulsing.
including range measurement within the search dwell 2) It used multiple, fixed prf's to distribute eclipsing during search.
3) During track it switched prf when eclipsing occurred.
time. A key to the TWS operation was the programming 4) It used a STALO that could tolerate severe missile vibration.
of the prf discretely and continuously over several values
during a dwell time of the antenna beam. The -81 used As a result of evaluation by Boeing, the
digital processing of data to establish track files; it used a Westinghouse version was selected for production.
medium duty cycle (.08) and multiple range gated Over the life of the program, over 300 DPN-53's
channels which increase the information rate and provided were produced and deployed, constituting the first
time gating for additional clutter rejection. These operational pulse doppler radar; BOMARC missiles were
techniques along with an extremely low side lobe antenna successfully fired against drones at all altitudes.
provided the technology needed for AWACS, and their
use became more attractive with the advent of digital LATER NAVY INVOLVEMENT: AWG 10, 11,
realization of the fast Fourier transform. Such digital 12 .....
techniques were reduced to practice on the Future
Weapon Control System (FWCS) a Navy development During the period 1959-65 Westinghouse proposed a
contract that followed the APQ-81. system for the Navy F4 that consisted of hardware and
techniques that had evolved from the U.S. AF and Navy
WESTINGHOUSE BOMARC SEEKER, DPN-53 ..... developments and the DPN-53 seeker. The result was the
AWG-10 fire control system using the APG-59 pulse
The credibility of pulse doppler radar as a solution to doppler radar [Fig. 9] which was also procured by the
the clutter problem was enhanced also by the production United Kingdom (AWG-1 1 & 12). In this same time
and fielding of the AN/DPN-53 pulse doppler target period, Ferranti Limited (of the United Kingdom)
seeker. This equipment was developed and produced in assigned a score of people to work at Westinghouse-as
the late 1950s by Westinghouse for the USAF IM99B part of a royalty agreement-a linkage which proved
BOMARC under contract to Boeing (a precursor of the particularly beneficial on the British AWG-1 1 and 12.
arrangement that would emerge 16 years later for the Over 1000 AWG 10, 11 and 12 systems were delivered
AWACS and 23 years later on the B-i). starting in 1963. This was one of the first all-solid-state
In March 1956 Westinghouse decided to spin off a designs with built-in test and fault isolation. Another
development of a pulse doppler target seeker that would unique feature was the ability to switch from PD to pulse
compete for the production buy of the Bomarc, since operation (the latter using CHIRP techniques). The
Boeing had contracted with RCA for the development display accommodated both modes, i.e., velocity vs.
model. The Westinghouse model [Fig. 81 was based on a azimuth and range vs. azimuth.
ANTENNA
Fig. 10. U.S. Air Force AWACS long range surveillance pulse doppler radar (Transmitter not shown), (1975).
produced by Hughes. (The AWG- 10 preceded the AWG- working in the sidelobe clutter area. Actually these
9 but was assigned nomenclature later). benefits derive from the time gating reduction of clutter
By the latter part of the 1960s, analog to digital such as used inthe AWACS radar. The time gated
converters had achieved sufficient dynamic range and approach usually results in a lower duty cycle, hence the
speed to permit quantizing the received radar signal into term "medium duty cycle" would be more descriptive
in-phase and quadrature channels; this digital format but "medium prf" has become the term in common use.
could be rapidly processed to perform filter bank and The velocity ambiguities of discrete clutter spikes brought
clutter elimination functions by a digital Fast Fourier on by the "'medium" prf were made more tolerble by use
transform. These techniques were used on a of a clutter "guard channel". The F-15 fly-off tested
Westinghouse candidate for the F-15 which lost a flyoff these concepts which were incorporated as a mode of the
to a Hughes model in 1970. This outcome was reversed APG-63 produced by Hughes for the F- 15. A single
on the next two flyoffs (for the AWACS and F-16). In signal/guard channel arrangement was devised for the
the later 1970s Hughes also designed and later produced Westinghouse WX radar described next. This was used in
the radar for the F- 18 (no flyoff). the F-16 fly-off and is used in the APG-66 production
During this period the U.S. AF sponsored work [16, radars [18] produced by Westinghouse for the F-16.
17] on "medium prf"; the use of a prf lower than that In the early 1970s control of false alarms was carried
required to yield a clutter free area was made possible by to extreme in a Westinghouse tail waming radar that won
the development of lower sidelobe antennas and the another U.S. AF flyoff (with AIL) in 1976. The ALQ-
feasibility of using digital processing. A motivation for 153 has virtually no false alarms (1 or less per day) and
the lower prf was to obtain better tail chase capability by over 200 have been delivered for the B-52 fleet.
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS VOL. AES-20, NO. 3 MAY 1984 303