Third Edition

merrill I. Skolnili

McGRAW-HiLriNTERNATION-
Electrical Engineering Series
 chapter

Radar Transmitters

10.1 INTRODUCTION
The radar systems engineer would like a transmitter to provide sufficient energy to detect
a target, be easily modulated to faithfully produce the desired waveforms, generate a sta¬
ble signal so that doppler signal processing can be performed without transmitter noise
masking the doppler-shifted received signal, provide the needed signal bandwidth and tun¬
able bandwidth, be of high efficiency, be of high reliability, be easy to maintain, have long
life, be able to operate with a minimum of personnel, be of a size and weight suitable for
the intended application, and be of affordable price. All of the above can be obtained, but
seldom all together in one transmitter. Compromise is necessary.
Some radar transmitters have to generate large peak power as well as large average
power; but it is the average power (which relates to energy) that is the measure of radar
performance rather than peak power. It was seen in Chap. 2 that the range of a radar is
proportional to the fourth root of the radar transmitter’s average power. To increase the
range of a radar by one order of magnitude (a factor of ten) the transmitter power has to
be increased by ten thousand. Although there have been radars with average powers greater
than a million watts, power cannot be increased without limit since high-power transmit¬
ters are heavy, take up space, and can consume much prime power (the power taken from
the local power company) or fuel for motor-driven electrical generators.
An indicator of the performance of a radar is the product of the antenna area times
the transmitter’s average power. Tom Weil1 described quite well the problem of choosing
between high transmitter power and a large antenna as follows:

690
 10.1 Introduction 691

It obviously would not make sense for a radar to have a huge, costly antenna and a
tiny, inexpensive transmitter, or vice versa, because doubling the tiny part would allow
cutting the huge part in half, which would clearly reduce total system cost. Thus, mini¬
mizing total system cost requires a reasonable balance between the costs of these two
subsystems. The result, for any nontrivial radar task, is that significant transmitter power
is always demanded by the system designers.

Radar transmitters have been based on either a power amplifier, such as a klystron,
or a power oscillator, such as a magnetron. In the early days of microwave radar in the
1940s and 1950s, the magnetron power oscillator was used almost exclusively since it was
the only high-power microwave tube available at the time. It did an outstanding job in
making microwave radar a reality in World War II, but it had many serious limitations.
Magnetrons are noisy devices that limit the MTI improvement factor that can be obtained.
Although they can produce high peak power (megawatts), they are not capable of large
average power, and their signal output cannot be readily modulated to produce pulse-com¬
pression waveforms. All of these disadvantages are overcome with amplifiers such as the
klystron, traveling wave tube, and the transistor. Modem high-performance radars almost
always employ some sort of power amplifier as the transmitter. The magnetron appears
to be limited to those applications where its relatively small size and lower cost are more
important than its limitations.
Most of the discussion in this chapter is about the RF power source. A transmitter,
however, is more than just the active RF power source. It includes the exciter and driver
amplifiers that provide the signal to be amplified if the power source is an amplifier. If
the transmitter generates a pulse waveform, a pulse modulator of some type is needed (ex¬
cept for RF power sources that are self-pulsed by the input waveform, as are transistors).
There must be a d-c power supply for generating the necessary voltages and currents to
operate the RF power device; means to remove the heat dissipated, including a heat ex¬
changer when liquid cooling is used; protection devices for dissipating high-voltage arc
discharges; safety interlocks; monitoring devices; isolators; high-voltage cable; insulat¬
ing-oil tanks (to immerse high-voltage cathode bushings to prevent corona and high-volt¬
age breakdown); and lead shielding of X-rays when high voltages are used. Not all high-
power radar transmitters need all of the above, but an RF power source is useless without
the ancillary devices required to make it function.
The efficiencies of RF power sources typically might range from about 10 percent to
about 60 percent. This is the RF conversion efficiency, defined as the ratio of RF power out¬
put available from the device to the d-c power input to the electron stream. It is the efficiency
of interest to the tube or RF power source designer, but it is not the best measure for the
radar systems engineer. A better measure is the transmitter system efficiency, which is the
ratio of the RF power available from the transmitter to the total power needed to operate
the transmitter. The total power includes the power to generate the electrons at the cath¬
ode, the power to generate any electromagnetic fields required for containing the electron
beam, the power to cool the device, and any other power needed for the proper operation
of the transmitter. If, for example, the RF conversion efficiency were 40 to 50 percent,
the transmitter system efficiency might be 20 to 25 percent, or less. Thus one usually
doesn't want to start with a power source whose RF conversion efficiency is only 10 to
15 percent, unless the power is so low that efficiency is not an important consideration.
692 chapter to * Radar Transmitters

For maximum efficiency, most high-power RF power sources operate saturated, mean¬
ing they are either completely on or completely off, with no intermediate power levels. This
is all right for a radar that generates a rectangular-like pulse waveform. There are times, how¬
ever, when it might be desired to have an amplitude-tapered, or shaped, pulse (for example,
to reduce the time sidelobes in pulse compression waveforms or to minimize the effects of
RF interference to other users of the electromagnetic spectrum). Highly shaped transmission
waveforms are seldom found in high-power microwave radar systems because of their lower
efficiency. Transistor amplifiers can be operated in what is called class-A operation so that
there is a linear relationship between the output and input signal amplitudes. The efficiency
of a class-A amplifier, however, is much less than that which would be achieved with the
same device operated class-C. (Class-C amplifiers are nonlinear and are self-pulsing in that
they generate pulses when the RF drive is turned on and off.) Thus, a microwave radar trans¬
mitter is almost always operated in saturation and not as a linear device.
High reliability and long life are important for a transmitter. The life of most RF
power sources can be many thousands or many tens of thousands of hours, as will be dis¬
cussed when describing the individual devices later in this section. If a transmitter’s mean
time between failures (MTBF) is not as long as expected, factors other than the RF power
source are often at fault. Likely candidates are fans and blowers, the wrong type of coolant,
RF connectors and coolant fittings damaged, coolant lines clogged, and leads that are bro¬
ken, mishandled, or abused. Conservative mechanical and electrical design and procure¬
ment practices that guarantee reliability from suppliers are needed for a trouble-free trans¬
mitter (or anything else). The user of RF power sources also needs to help in avoiding
less than the achievable reliability. For example, A. S. Gilmour2 states “If the truth were
known, it could well be that over 50 percent of all failures are the fault of the users, rather
than the tube manufacturers.”

Summary of Radar RF Power Sources The different RF power sources available for high-
power radar application include the klystron, traveling wave tube, solid-state transistor
amplifier, Twystron, magnetron, crossed-field amplifier, amplitron, grid-controlled vac¬
uum tube, extended interaction amplifier, gyrotron, and others. None provide all the de¬
sirable features that might be wanted. Some are no longer as popular as they once were.
The choice depends in large part on the application and its constraints. Each of these de¬
vices will be briefly summarized below and-described in more detail later in the chapter.
All except the magnetron are power amplifiers. The gyrotron can be either an amplifier
or an oscillator.

Klystron This is an excellent radar tube when it can be employed. It has high gain and
good efficiency, and is capable of higher average and peak power than most other RF
power sources. It can have wide bandwidth (in the vicinity of 8 to 10 percent relative
bandwidth*) when its power is large, long life (tens of thousands of hours), low interpulse
noise, and good stability for doppler processing. When large peak powers are generated
it requires very high voltage and X-ray shielding.

‘Relative, or fractional, bandwidth in percent = (Af/f0) x 100, where Af = absolute bandwidth and f, - center
frequency.
 l o. i Introduction 693

Traveling Wave Tube (TWT) TWTs have slightly less power, slightly less gain, and
slightly less efficiency than a klystron; but they are capable of wide bandwidth, especially
at modest power levels. At high power levels the TWT bandwidth is lower than what can
be achieved at lower powers, but it is still relatively large.

Hybrid Klystrons These are similar to klystrons, but with one or more of the resonant
cavities replaced by multiple cavities similar to what are used in a TWT. There are at least
three versions: tire Twystron, the extended interaction klystron, and the clustered cavity
klystron. Bandwidths can be 15 to 20 percent, or more. Extended interaction klystrons
have also been used for modest power millimeter-wave transmitters.

Solid-State Transistor Amplifiers These are capable of wider bandwidth than most other
RF power sources. They operate with low voltages, ease of maintenance, and have promise
of long life. They are inherently of low power so that a large number of individual de¬
vices must be combined to generate sufficient power for most radar applications. For ad¬
equate efficiency, they should be operated at high duty cycles, which require that they
generate long pulses and employ pulse compression.

Magnetron The magnetron is generally smaller in size and utilizes lower voltage than
the klystron; but its average power is limited and it has poor noise and stability charac¬
teristics, which restrict its ability to cancel clutter when used in an MTI radar.

Crossed-Field Amplifiers These are capable of high power, good efficiency, and wide
bandwidth, but are of relatively low gain (about 10 dB). Lower voltage, just as with the
magnetron, means that X-rays are not usually a problem. Crossed-fteld amplifiers are gen¬
erally noisier and less stable than most other RF power sources.

Grid-Control Tubes These are modern versions of the classical triode and tetrode vac¬
uum tubes originally introduced early in the twentieth century. They are a good power
source for UHF radars, but they have been largely replaced by solid-state devices.

Microwave Power Module This is a combination of a modest power TWT driven by a

solid-state device that competes favorably with high-power solid-state modules in some
applications.

Gyrotrons These can produce very high power in the millimeter wave region, but they
require large magnetic fields. They have had only modest application in operational radar
systems.
RF power sources may be grouped into four general classes: (1) linear-beam tubes,
(2) solid state, (3) crossed-fteld tubes, and (4) others not included in the first three. The
klystron, traveling wave tube, magnetron, and crossed-fteld amplifier are slow-wave de¬
vices in that the phase velocity of the electromagnetic wave in the RF structure is slowed
so as to be approximately equal to the velocity of the electron stream in order for (d-c)
energy from the electron beam to transfer to electromagnetic energy in the (RF) signal.3
The gyrotron, on the other hand, is a fast-wave device in that the phase velocity of the
electromagnetic wave exceeds the speed of light in the interaction region.
694 CHAPTER 10 ® Radar Transmitters

10.2 LINEAR-BEAM POWER TUBES

In the linear-beam tube the electrons emitted from the cathode are formed into a long
cylindrical beam that receives the full potential energy of the electric field before the beam
enters the RF interaction region. The klystron, traveling wave tube, Twystron, and ex¬
tended interaction amplifier are examples of linear-beam tubes. The last two are basically
hybrid devices that combine the technology of the klystron with the RF structure of the
TWT. An axial magnetic field is used in linear-beam tubes to confine the electron beam
and keep electrons from hitting the RF structure. Transit-time effects, which can limit con¬
ventional vacuum tube performance at high frequencies, are used to good advantage in
linear-beam tubes to density modulate the uniform d-c electron beam to create bunches
of electrons from which RF energy can be extracted.
Linear-beam tubes can produce much higher power than other power sources. Kly¬
strons are capable of more than a megawatt of average power. High power is a result, in
part, of their larger size and high voltages. Thomson-CSF in France produced a UHF kly¬
stron that delivered more than one MW CW power. It was 5 m (16.4 ft) long and weighed
1400 kg (3000 pounds).4 On the other hand, A-band klystron and TWT transmitters pro¬
ducing many kilowatts of average power can be made light enough and small enough to
fly in the nose of military fighter/attack aircraft.

Klystron A sketch of the principal parts of a klystron is shown in Fig. 10.1. At the left
is the cathode which emits a stream of electrons that is formed into a narrow cylindrical
beam by the electron gun. The electron gun consists of the cathode that is the source of
electrons, a modulating anode or other beam-control electrode to provide a means for turn¬
ing the beam on and off to generate pulses, and the anode. The electron emission density
at the surface of the cathode is less than that required for the electron beam, so a large-
area cathode surface is used and the emitted electrons are caused to converge to a narrow
beam of high electron density. The multiple RF cavities, which correspond to the LC res¬
onant circuits of conventional lower-frequency amplifiers, are at anode potential. Elec¬
trons are not intentionally collected by the anode, as in some other tubes; instead they are

gun
 10.2 Linear-Beam Power Tubes 695

removed by the collector electrode (shown on the right) after the beam has given up its
RF energy to the output RF cavity.
The RF input signal is applied across the interaction gap of the first cavity. Those
electrons which arrive at the gap when the input signal voltage is a maximum (peak of
the sinewave) experience a voltage greater than those electrons which arrive at the gap
when the input is at a minimum (trough of the sinewave). Thus the electrons that see the
peak of the sinewave are speeded up and those that see the trough are slowed down. The
process whereby some electrons are speeded up and others slowed down is called veloc¬
ity modulation of the electron beam. In the drift space, electrons that are speeded up dur¬
ing the peak of one cycle catch up with those slowed down during the previous cycle. The
result is that the electrons of the velocity-modulated beam become “bunched,” or density
modulated, after traveling through the drift space. A klystron usually has one or more ap¬
propriately placed intermediate cavities to enhance the bunching of the electron beam,
which increases the gain. If the interaction gap of the output cavity is placed at the point
of maximum bunching, power can be extracted from the density-modulated beam. The
gain of a klystron might be 15 to 20 dB per stage when synchronously tuned (all cavities
tuned to the same frequency), so that a four-cavity (three stage) klystron might provide
over 50 dB gain.
After the bunched electron beam delivers its RF power to the output cavity, the en-
ergy ol the election beam that remains is dissipated when the spent electrons are removed
by the collector. The energy dissipated by the collector is energy lost and reduces the ef¬
ficiency of the tube. If the collector is insulated from the body of the tube and a negative
voltage is applied to the collector, the electrons in the spent beam will have lower kinetic
energy so that less heat is produced when they impact upon the collector.5 This results in
an increase in the efficiency of the tube. There is a spread, however, in the velocities of
the electrons in the beam; so if the potential is too negative, some of the slower velocity
electrons will be returned to the walls of the RF section of the tube and be collected as
body current, with a decrease in efficiency. This problem is overcome by employing a
collector with several segments insulated from one another and with different negative po¬
tentials so that electrons with different velocities can be separated and collected at their
optimum potential. Figure 10.1 shows a single-stage collector, but both the klystron and
TWT usually employ multiple-stage depressed collectors for greater efficiency. The mul¬
tiple stages (three or four might be typical in a radar tube) are at intermediate voltages,
which allow catching the spent electrons at a voltage near optimum.
According to Weil, a klystron with a peak power of 1 MW requires a voltage of
about 90 kV. Gains might vary from 30 to 70 dB, bandwidths from 1 to 8 percent? and
efficiencies from 40 to 60 percent (with depressed collectors).
A long solenoid (not shown in Fig. 10.1) with iron shielding around its outside di¬
ameter surrounds the high-power klystron to provide an axial magnetic field that confines
the electrons to a relatively long, thin beam and prevents the beam from dispersing. Cool¬
ing might have to be provided for the electromagnets. In a high-power klystron, from 2
to 5 percent of the beam power might normally be intercepted by the interaction struc¬
ture, or body of the tube. If the beam were not properly confined by the external mag¬
netic field, the stray electrons impinging on the structure of the tube could cause it to
overheat and possibly be destroyed. Protective circuitry is normally employed to remove
696 chapter 10 ® Radar Transmitters

the electron beam voltage in the event the magnetic field fails to keep the beam properly
focused.
The electron beams of klystrons and TWTs also can be confined with permanent mag¬
nets. They do not require power or cooling, and the various protective circuits needed with
solenoids are eliminated. Permanent magnets have been used with high power tubes, but
they are quite heavy.6,7 A significant reduction in weight, however, can be obtained with
a periodic permanent-magnet (PPM) focusing system that consists of a series of magnetic
lenses. These lenses employ washer-shaped disk magnets separated by iron pole pieces.
The PPM replaces the uniform field of a solenoid with a periodic, essentially sinusoidal,
field having the same rms value as the uniform field. Samarium cobalt is an example of
a magnet material that has been widely used for tubes requiring permanent magnets. In
the past, PPM focusing was usually not suited for large average-power tubes, but it has
been successfully applied to high peak-power klystrons, as described next.
High-power klystrons have been used ever since the 1950s for linear accelerators to
generate energetic beams of charged particles for research on the physics of high-energy
particles. Many advances in klystron capability have been obtained from the development
of klystrons for this purpose. These advances have, of course, been of benefit to radar as
well. Although the invention of the klystron amplifier was reported in 1939, before the
invention of the magnetron, it was not used or further developed significantly during World
War II. It did not find its way into radar application until the development of a 20-MW
peak power klystron, used in one of the first linear accelerators, was reported in 1953 by
Stanford University. Thus the high-power klystron, which is a very important power source
for radar, was a by-product of basic research in science. Work continued at the Stanford
Linear Accelerator Center (SLAC) to develop high peak-power klystrons for electron-
positron colliders. The klystron for the Next Linear Collider (NLC)8 is at a frequency of
11.424 GHz, with a peak power of 60 to 75 MW using periodic permanent magnets made
of neodymium-iron boron. The PPM with 10 pounds of permanent magnets replaces a
1/2-ton, 10-kW focusing solenoid. The NLC uses 6528 klystrons, which means that a
total of 65 MW of solenoid power is avoided. The use of PPM, as well as having a tube
with 60 percent efficiency, reduces “the NLC electric power bill by tens of millions
of dollars per year.” The tube requires an electron beam voltage of 490 kV. The average
life of the 5-band klystrons used for the previous Stanford Linear Collider (SLC) was
50,000 hours, and it is expected that a similar life will be obtained with the A-band NLC
klystron. The manufacturing cost was said to be $30,000 per tube. If the tube delivers
75 MW with a pulse width of 1.5 /jls and a prf of 120 Hz, its average power is 13.5 kW.

Bandwidth of a Klystron The frequency of a klystron is determined by its resonant cav¬

ities. When all the cavities are tuned to the same frequency, the gain of the tube is high
but the bandwidth is narrow, usually a fraction of one percent for a tube of modest power
output. This is called synchronous tuning. To maximize the klystron’s efficiency the next
to last (penultimate) cavity is tuned upward in frequency and is outside the passband. Al¬
though the gain is reduced by about 10 dB, the improved electron bunching results in
greater efficiency and in 15 to 25 percent more output power.9 Broadbanding of a multi-
cavity klystron may be accomplished by stagger tuning the cavities, similar to the method
for broadbanding a conventional multistage IF amplifier. Stagger tuning a klystron is not
 10.2 Linear-Beam Power Tubes 697

precisely analogous to stagger tuning an IF amplifier because of interactions among the

cavities that can cause the tuning of one cavity to affect the tuning of the others. The
VA-87 four-cavity S-band synchronously tuned klystron amplifier with a 20 MHz band¬
width and a 61 dB gain can have a 27 MHz bandwidth and a 57.6 dB gain when tuned
for maximum power.10 When stagger tuned it has a 77-MHz bandwidth (2.8 percent) and
a gain of 44 dB.
Theory shows that the bandwidth of a klystron can be significantly increased by in¬
creasing its power and its beam perveance (which is defined as the beam current I divided
by beam voltage V to the 3/2 power, or p = I/V312). A 10 MW peak power klystron, for
example, can have an 8 percent bandwidth, as compared to a 200 kW tube which might
have a 2 percent bandwidth, and a 1 kW tube having only a 0.5 percent bandwidth.10 High-
power multicavity klystrons can be designed with bandwidths as large as 10 to 12 percent.

Frequency Changing, or Tuning6,9 Conventional narrowband klystrons may have their

frequency changed mechanically over a relatively wide frequency range. The individual
cavities of a klystron can be changed in frequency (or tuned) by having a flexible wall in
the resonant cavity (tuning range of about 2 or 3 percent), by a movable capacity element
in the cavity (10 to 20 percent tuning range) or by a sliding contact movable cavity wall.
It can be tedious to tune a multiple-cavity klystron because of the interactions among
the several cavities. Conventional gang-tuning is complicated since the resonant cavities
do not have the same tuning rates. The channel tuner mechanism6 avoids the problem of
frequency tracking of the resonant cavities by pretuning the cavities, generally at the fac¬
tory. The tuning information is stored mechanically within the tuner mechanism. When a
particular frequency is selected, the tuner mechanism provides the correct tuner position
for each cavity to furnish the desired klystron frequency response. The channel tuner mech¬
anism is in a box attached to the klystron with gears to simultaneously set the tuning
plungers at each cavity to their predetermined positions for a given frequency. The tun¬
ing plungers can be actuated manually or remotely by push buttons and a servomotor. The
frequency can be changed in seconds.

Power Some of the highest power radar transmitters have used klystrons. The ability of
a klystron to produce higher power than other microwave power sources is, in part, due
to its geometry. The regions of beam formation, RF interaction, and beam collection are
separate. Each can be designed to best perform its own particular function independently
of the others. The cathode, for example, is outside the RF field and need not be restricted
to sizes that are small compared to a wavelength. Large cathode area and large interelec¬
trode spacings may be used to keep the emission current densities and voltage gradients
to reasonable values. The only function of the collector is to dissipate heat. It can be a
shape and size best suited for satisfying the average or peak power requirements without
regard for conducting RF currents, since none are present.

Efficiency It is said in Ref. 8 that Robert Symons found that the best values of effi¬
ciency reported for klystrons worldwide followed the relation

RF efficiency (percent) = 90 - 20 X microperveance [10.1]

698 CHAPTER 10 ® Radar Transmitters

where the microperveance is the perveance //V372 times 106, where 1 = beam current and
V = beam voltage. Thus the lower the microperveance (or perveance) the higher will be
the klystron efficiency. The perveance affects other properties of the klystron, including
its bandwidth and power. Higher efficiency often requires, therefore, a reduction in band¬
width and lower power.

Reliability and Life High-power transmitters employing power vacuum tubes have
sometimes had the unwarranted reputation for poor reliability and short life. There is much
evidence to the contrary for the klystron tube. Gilmour11 reports the mean time between
failures (MTBF) of eleven different applications of klystrons in radar systems (not iden¬
tified by type or power). The MTBF for these examples varied from 75,000 hours to 5000
hours, with an average value of about 37,000 hours for all eleven applications. (There are
8760 hours in a year.) The VA-842 high-power klystron tube used by the U.S. Air Force
in the original Ballistic Missile Early Warning System (BMEWS) had a demonstrated life
in excess of 50,000 hours. Symons12 reports that one of the BMEWS tubes he designed
in 1958 was still operating after 240,000 hours when the radar in Greenland was replaced
by the solid-state Pave Paws radar'.

An S-Band Klystron The venerable VA-87 klystron built by Varian (now Communica¬
tions & Power Industries) was widely used in the FAA’s S-band Air Surveillance Radars
commonly found at major airports. It was a six-cavity tube tunable from 2.7 to 2.9 GHz,
the frequency band reserved for air-traffic control radars. It had a peak power of from 0.5
to 2 MW, average power of from 0.5 to 3.5 kW, 50 dB gain, 45 percent efficiency, and a
one-dB bandwidth of 39 MHz. It demonstrated a mean-time-between-failure (MTBF) rate
of 72,000 hours. A similar tube was used in the Nexrad Doppler Weather Radar, but with
bandwidth from 2.7 to 3.0 GHz.

Traveling Wave Tubes (TWT) Like the klystron, the traveling wave tube is also a linear-
beam tube with the cathode, RF circuit, and collector separated from one another. The
klystron and the TWT were invented at different times in different parts of the world, but
they are similar to one another. There is continuous interaction of the electron beam and
the RF field over the entire length of the propagating structure of the traveling wave tube.
In the klystron, on the other hand, the interaction occurs only at the gaps of a relatively
few resonant cavities. The chief characteristic of a TWT is that it has wide bandwidth.
Low power TWTs with a helix slow-wave RF structure are capable of octave bandwidths.
With the high peak powers required of most radar applications, the bandwidths available
with high-power TWTs are, however, much less than an octave.
The major parts of a TWT are indicated in Fig. 10.2. A helix is shown for the slow-
wave RF structure even though the helix is seldom used in TWTs found in radar appli¬
cations. The electron beam is similar to that of the klystron. Both the TWT and the kly¬
stron employ the principle of velocity modulation to cause the electron-beam current to
be periodically bunched (density modulation). The electron beam passes thorough the RF
interaction circuit known as the slow-wave structure, or periodic delay line. The velocity
of propagation of the RF signal is slowed down by the periodic delay line so that it is
nearly equal to the velocity of the electron beam. This is the reason that the helix and
 10.2 Linear-Beam Power Tubes 699

Figure 10.2 Representation Gun

Attenuation Collector
of the principal parts of a Cathode,. \r anode
traveling-wave tube.
Heater-

Electron , | Helix ( , |~~

t (interaction region) r
beam RF RF
input output

other microwave circuits used in TWTs are called slow-wave structures. The synchronism
between the electromagnetic wave propagating along the slow-wave structure and the
d-c electron beam propagating inside the helix results in a cumulative interaction which
transfers d-c energy from the electron beam to increase the energy of the RF wave, caus¬
ing the wave to be amplified. Just as in the klystron, an axial magnetic field keeps the
electron beam from dispersing as it travels down the tube.
After delivering their d-c energy to the RF field on the slow-wave structure, the elec¬
trons are removed by the collector, which is usually a multistage depressed collector, as
was described for the klystron. It is easier to design a depressed collector for a TWT than
for a klystron since the spent electron beam of a TWT might have a 20 percent spread in
velocity, but the klystron might have a velocity spread of almost 100 percent.5 Because
the efficiency of a conventional TWT is usually lower than that of a klystron, the increase
in efficiency in the TWT provided by the depressed collector has a greater relative effect
than with a klystron.
Although a helix is shown in Fig. 10.2 as the slow-wave structure, it is seldom found in
TWTs used for radar. The helix TWT is limited to voltages of about 10 kV and a peak power
output of a few kilowatts,1 which is generally too low for most radar applications. Other types
of RF slow-wave structures have to be employed instead, and these do not have as wide a
bandwidth as the helix. A modification of the helix known as the ring-bar circuit can be used
if the peak power is less than about 100 to 200 kW. One example is the Raytheon QKW-
1671 A, a tube suitable for air-surveillance radars. It has a peak power of 160 kW, duty cy¬
cle of 0.036, 70-p.s pulse width, 45-dB gain, and a 100-MHz bandwidth. The Air Force Co¬
bra Dane phased array radar uses 96 QKW-1723 ring-bar TWTs, each with a peak power of
175 kW and average power of 10.5 kW. The Cobra Dane operates from 1175 to 1375 MHz.
Powers greater than 200 kW are obtained with the coupled-cavity circuit, of which
the so-called “cloverleaf ’ is an example. The bandwidth, however, is less than that of lower
power TWTs. The individual unit cells of the coupled-cavity circuit resemble klystron res¬
onant cavities. Several tens of these klystron-like cavities are used for the slow-wave struc¬
ture of a high-power TWT.13 There is no direct coupling between the cavities of a kly¬
stron, but in the traveling wave tube, coupling is provided by a long slot in the wall of
each cavity. There are two slots in each cavity (input and output) that are 180° apart in
rotational position so they act similar to a folded waveguide.
An example of a TWT using a cloverleaf coupled-cavity slow-wave structure is the
S-band VA-125A. It is liquid cooled and is capable of 3 MW peak power over a 300-MHz
bandwidth, 0.002 duty cycle, 2-/jus pulse width, and a gain of 33 dB. It was originally de¬
signed to be used interchangeably with the VA-87 klystron, except that the VA-125 TWT
has a wider bandwidth and requires a larger power input signal because of its lower gain.
700 chapter 10 ® Radar Transmitters

The bandwidth of a coupled-cavity TWT can be from 10 to 15 percent. When the

power of a TWT is increased, its wide bandwidth decreases. On the other hand, when the
power of a klystron is increased, its narrow bandwidth widens. As the power of these two
tubes increase, their bandwidths become comparable. With high power, the klystron tends
to be the preferred tube since it doesn’t experience the stability problems of the TWT
Although the TWT and the klystron are similar in many respects, a major difference
is that there is feedback along the slow-wave structure of the TWT, but the back coupling
of the RF energy in the klystron is negligible. If there is a mismatch at the input of the
TWT and if sufficient energy is fed back to the input, undesired oscillations can result.
To reduce the amount of feedback energy to an insignificant level, attenuation has to be
inserted in the slow-wave structure. The attenuation may be distributed or lumped, but it
is usually found within the middle third of the tube. The loss introduced to attenuate the
feedback also reduces the power of the forward-traveling wave, and is therefore undesir¬
able. The loss in the forward wave can be avoided by the use of severs, which are short
internal terminations designed to dissipate the reverse-directed power without seriously
affecting the power in the forward direction. The number of severs depends on the gain
of the tube; one sever is used for each 15 to 30 dB of tube gain.1
The efficiency of TWTs is less than that of the klystron because of the necessity for
including attenuation or severs. Efficiency is also reduced by the presence of relatively
high RF power over an appreciable fraction of the entire structure.
In some traveling-wave tubes with coupled-cavity circuits, oscillations appear for an
instant during the turn-on and turn-off portions of the pulse.1 These are called rabbit-ear
oscillations because of their characteristic appearance when the RF envelope of the pulse
waveform is displayed visually on a CRT. They are undesirable in some military appli¬
cations since they might provide a distinctive feature for recognizing a particular radar.
Weil1 describes some of the ways rabbit-ear oscillations can be avoided.

TWT MTBF The mean time between failures (MTBF) is given by Gilmour11 for nine
different types of coupled-cavity TWTs. (The type of TWT, frequency, and power are not
mentioned.) The MTBFs of these nine classes vary from a high of 17,800 hours to 2200
hours, with an average of 7000 hours for all nine classes of tubes. TWTs for space ap¬
plications, which are of lower power than radar tubes, are said by Gilmour to have MTBFs
of the order of one million hours.

Hybrid Variants of the Klystron By judiciously combining the best features of the kly¬
stron and the traveling wave tube, an RF power source can be obtained which has band¬
width, efficiency, and gain flatness better than either the conventional klystron or TWT.
This is achieved by replacing one or more of the klystron resonant cavities with broader
bandwidth cavities that are more like the coupled-cavity circuits used in traveling wave
tubes. There have been three variants of the klystron in which this is done: the Twystron,
the extended interaction klystron, and the clustered-cavity klystron. Such combinations of
klystrons and TWTs are sometimes called hybrid tubes.14

Twystron The bandwidth of a klystron is limited by the output resonant cavity. It can¬
not be made broadband without a decrease in efficiency. Since coupled-cavity slow-wave
 10.2 Linear-Beam Power Tubes 701

circuits have broader bandwidth than klystron resonant cavities, replacing the output cav¬
ity of a klystron with a TWT coupled-cavity circuit can significantly increase the band¬
width as well as achieve a slight increase in efficiency. Although the output is a TWT
slow-wave circuit, the driver portion of the tube (the intermediate cavities and the input
cavity) consists of resonant cavities that are stagger tuned. Such a tube is called a Twys-
tron, a trademark name of Varian (now Communications & Power Industries, or CPI). The
VA-145 Twystron has demonstrated a 14 percent 3-dB bandwidth (12 percent
1-dB bandwidth), 48 percent efficiency, and 41 dB gain at midband.1

Extended Interaction Klystron (EIK) In this device, the single-gap resonant cavity of
a klystron is replaced by a resonated slow-wave TWT-like circuit. The use of slow-wave
coupled resonators can be applied to the prior cavities as well as the output cavity. The
extended interaction amplifier (EIA) klystron can have a high average power; for exam¬
ple, 1 MW CW at X band using a five-cavity resonator.15 It has broader bandwidth than
a klystron, but less than that of a TWT. EIKs also have been used for low-power mil¬
limeter wave tubes. A 150-W average power 95-GHz klystron, advertised by CPI Canada,
is claimed to have a 1.5-kW peak power, 500-MHz bandwidth, 25 percent efficiency,
45-dB gain, and to weigh 4.5 kg. There is also an extended interaction oscillator, or EIO,
which has been used at millimeter wavelengths.

Clustered-Cavity Klystron The technique of grouping cavities was extended in what

has been called a clustered-cavity klystron (CCK). In this tube, Fig. 10.3, the individual
intermediate cavities of a multicavity klystron are replaced by pairs or triplets of artifi¬
cially loaded low-<2 cavities with Qs of one half or one third that of the single cavity they
replace.16 Similar groups of cavities are used in both the EIA and the CCK, but there is

Figure 10.3 Comparison of a RF RF

conventional staggered-tuned klystron
(top) and a clustered-cavity klystron
(bottom). The shaded circular regions
represent bunching of the electrons.
I (From Symons and Vaughan,14 Copyright
I 1994 IEEE.)

RF
input
Intermediate cavity clusters

lq m
—#---11J-f
LULULLJ
Collector ■
' Cathode
Clustered-cavity™ klystron
702 chapter 10 ® Radar Transmitters

no inductive or other coupling between cavities in the CCK as there is in the P.TA Theory
indicates that bandwidths of 30 percent should be obtained in megawatt klystrons using
fifteen intennediate cavities in triplets. In practice, 20 percent bandwidths have been ob¬
served. This form of structure also provides the greatest bandwidth in the shortest length.
Clustered-cavity klystrons might be more complex and costly than a klystron, but they are
less complex and of less cost than a comparable TWT or Twystron.1 Symons,12 the in¬
ventor of the clustered-cavity klystron, states that two of these tubes can be used instead
of the two nan'ower-band klystrons in the AWACS radars. Redundant operation is provided
without a large weight penalty, since either of the clustered-cavity klystrons provide full
operational capability similar to the redundancy commonly employed in FAA radars.

10.3 SOLID-STATE RF POWER SOURCES

Use of Solid-State Amplifiers for Radar Transmitters The solid-state RF power generation
device usually of interest for radar- application has been the transistor amplifier, both
silicon bipolar and gallium arsenide FET. An individual transistor amplifier device is
inherently of low power and low gain, but it operates with low voltages and has high
reliability.
A single microwave transistor might have an average power capability from a few
watts to over a hundred watts, depending on the frequency and the duty cycle. The lower
the frequency the greater can be the power. To increase the power, transistors may be op¬
erated in parallel, and with more than one stage to increase the gain. A single power mod¬
ule might, for example, consist of eight transistors, with four operating in parallel as the
final stage, two in parallel as the next to last stage, and two in series as the driver stages.
To achieve the high powers required for most radar applications, the outputs of many solid-
state devices have to be combined in some manner. Combining of many devices can be
achieved with microwave circuitry or by combining in space (radiating from many indi¬
vidual antenna elements of an array antenna).
Solid-state power devices of a given average power cannot be operated at high peak
powers as can vacuum tubes. According to Borkowski,17 “a microwave transistor capable
of perhaps 50-W average power cannot handle-much more than 100 to 200 W of peak
power without overheating during the pulse.” For this reason solid-state amplifiers for
radar generally operate in the vicinity of 0.1 duty factor, instead of the 0.001 to 0.01 duty
factors common with high-power vacuum tube RF power sources. Thus when solid-state
devices are employed in radar transmitters they have long pulse widths and require pulse
compression to obtain useful range resolution. Long pulses are not always desired by the
radar engineer, but they have been accepted as one of the prices to be paid for the use of
solid state.
There are at least four ways that solid-state devices can be employed in radar: (1) as
a transmitter for a low-power application, (2) as a high-power transmitter where a large
number of individual transistors are combined with microwave circuitry, (3) with many
modules distributed on a mechanically steered planar array (such as a 3D radar), and (4)
with a module at each of the many elements of an electronically scanned phased array
 10.3 Solid-State RF Power Sources 703

(also called an active aperture). In the last two, the power from the many solid-state trans¬
mitter modules is “combined in space.”

Low-Power Transmitter The solid-state device is used as a direct replacement for a vac¬
uum tube when the radar waveform is of low power and of high duty cycle or CW. Ex¬
amples are the FM-CW radar altimeter, doppler (police) speedmeter, and the airborne
doppler navigator. The solid-state transmitter has been highly successful in such applica¬
tions. It has been difficult, however, for solid-state to replace the small magnetron in the
civil marine radars found on many ships and pleasure boats because this radar market is
highly competitive and low price is important for success. The same appears to be true
for the absence of solid-state transmitters for the nonradar application of microwave ovens
for the household market.

High-Power Transmitter The solid-state transmitter has replaced the high-power vac¬
uum tube in some air-surveillance radars. A large number of transistors are combined to
produce a single output that feeds a conventional antenna. (This was at one time called a
“solid-state bottle,” but such transmitters are housed in cabinets which do not resemble
“bottles.”) Two such transmitters will be described.
One of the first solid-state radars to have its tube transmitter replaced by solid state
was the AN/SPS-40, a modest UHF 2D shipboard air-surveillance radar used mainly by
United States ASW (antisubmarine warfare) destroyers to provide conventional air-
surveillance for keeping track of ASW aircraft.18 It was developed by Westinghouse, Bal¬
timore (now known as Northrop Grumman). This was a good example of a direct re¬
placement since the tube transmitter operated with a long pulse (60 /as), pulse compres¬
sion, and a moderate duty cycle (0.018), so that the solid-state transmitter could utilize
the same waveform as did the original radar system. The basic transistor building block
operated from 400 to 450 MHz, with 400- to 500-W peak power, 8-dB gain, and 55 per¬
cent efficiency. A module consisted of two stages with a total of 10 silicon bipolar tran¬
sistors that produced 2500-W peak power out when the input was 120 W. There were 112
of these modules combined in two groups of 56 each to produce 250-kW peak power and
4 kW of average power. Each of the two groups of 56 modules was housed in its own
cabinet. There was a third cabinet with the driver, power supplies, and some other de¬
vices. The transmitter was designed so that no damage occurred when a full short circuit
was applied across the load. Both liquid and forced air cooling were used; and if the liq¬
uid cooling was lost, the transmitter could operate with 80 percent power (200-kW peak)
with only air cooling. The loss of one module reduced the transmitter power output by
0.08 dB. The transmitter had good reliability due in part to its built-in spare modules. The
solid-state transmitter for this radar cost more and was larger than the vacuum tube trans¬
mitter it replaced; but it was considered a success.
The Ramp (Radar Modernization Project) radar system was an L-band (1250 to
1350 MHz) air-traffic control Primary Surveillance Radar (PSR) located at major airports
across Canada.19 It was developed by Raytheon Canada and had a range of 80 nmi and
an altitude coverage of 23,000 ft against a 2 m2 aircraft target with 80 percent probabil¬
ity of detection. It used a solid-state transmitter with a peak power of 28 kW and an av¬
erage power of 1.2 kW, which corresponded to a duty cycle of 0.068.
704 CHAPTER 10 • Radar Transmitters

There were a total of 14 modules used in the RAMP PSR. Each module consisted of
42 identical class-C 100-W peak-power silicon bipolar transistors arranged in a 2-8-32
configuration to produce 2350 W of peak power. As described by Merrill,20 “The tran¬
sistors were arranged in a one-driving-four ‘unit amplifier’ format, with eight unit ampli¬
fiers in parallel so that 10 transistors were drivers while 32 were output devices.” The
50-pound air-cooled module had a measured efficiency greater than 25 percent and a power
gain greater than 16 dB.17
The 14 modules were combined as pairs to form seven transmitting channels. Only
six of the seven channels were needed to meet the system requirement for a minimum
peak power of 21 kW. The extra seventh channel permitted maintenance and repair to be
performed on a failed channel while the remaining six channels were continuously avail¬
able. The extra channel, therefore, allowed the radar to maintain a high availability. [The
theoretical power output when A out of 7 channels are operating is Pout = P- (Nil)2, where
/J7 = the power delivered by all seven channels.] The antenna for this radar was a 33 ft
wide by 22 ft high reflector with 33.5-dB gain. It rotated at 12 rpm.

Modules Arranged on a Mechanically Scanned Planar Array, AN/TPS-59 Individual

transmitter modules can be arranged on a mechanically scanning array antenna by plac¬
ing one module at each element, but it has been more usual in such a radar to employ one
module at each row of the antenna. This is the arrangement in the AN/TPS-59,17 a trans¬
portable L-band 3D air-surveillance radar developed by GE, Syracuse, N.Y. (now Lock¬
heed Martin) for the U.S. Marine Corps for air defense and ground-control of intercept
(GCI). It was designed to detect a 1-m2 fluctuating target within a 200-nmi range with
90 percent probability of detection, but it was also required to cover a volume out to 300
nmi in range and an altitude up to 100,000 ft. The rotating planar array antenna is 30 ft
high by 15 ft wide and consisted of 54 rows, each with 24 dipole elements. The radar can
operate within 1200 to 1400 MHz with 54-kW peak power and 9.7-kW average power,
which results in the relatively high duty cycle of 0.18. A single pencil beam is electron¬
ically scanned over an elevation angle of 20° as the antenna rotates 360° in azimuth.
At each row is a transceiver, which is a miniature radar containing transmitter, re¬
ceiver preamplifier, duplexer, phase shifter for steering in elevation, logic control, cool¬
ing, and power supply. Each transmitter module has ten 100-W amplifier units consisting
of two 55-W silicon bipolar transistors (with 7-dB gain) driven by a smaller 25-W de¬
vice. There are a total of 540 modules on the antenna.
Fixed-site variants of this radar are known as the GE-592 and the AN/FPS-117. The
latter uses a 24 ft by 24 ft antenna with 44 rows. It is a minimally attended radar whose
antenna operates within a radome for use in northern regions. The tactical mobile version
of this radar is the TPS-117, which was shown in Fig. 1.8.

Active Aperture, Electronically Steered Phased Array An example of the composition

of a T/R module as might be used for an active-aperture phased array antenna is illus¬
trated in Fig. 10.4. The T/R switches select between the transmitter and the receiving
paths. The circulator, which might be the largest and heaviest component on the module,
performs the function of the duplexer. The receiver protector (which is a diode limiter)
provides further protection of the low-noise amplifier (LNA). The same phase shifter is
 10.3 Solid-State RF Power Sources 705

Figure 10.4 Example of the TRANSMITTER--

composition of a T/R module that might
be used for an active-aperture phased
array radar.

--RECEIVER

used for both transmitting and receiving. Many other parts of a T/R module are not shown
in this illustration, such as the module controller and the power conditioner.21 The mod¬
ule controller obtains a beamsteering command from a central computer and calculates
the correct settings for the phase shifter. To minimize power consumption, the power am¬
plifiers and the LNA might be gated off when the controller is on. The module controller
might also perform self-testing and reporting of the status of the module. The power con¬
ditioner is important in keeping the efficiency of the module as high as practical.
The Pave Paws electronically steered array radar, also known as the AN/FPS-115, is a
UHF radar developed by Raytheon that was the first all solid-state active aperture electron¬
ically steered phased array radar. Its function was to detect and warn of sea-launched bal¬
listic missiles fired at the United States. This radar was discussed in Sec. 9.9. It operated
from 420 to 450 MHz with a peak power of 600 kW and an average power of
150 kW, which corresponds to a duty cycle of 0.25. Its diameter was 72.5 ft, with room to
grow to 102 ft. An individual module delivered 340 W of peak power with 39 percent effi¬
ciency.17’22 The pulse width was 16 ms. There were two phased array faces per site to cover
approximately 240° in azimuth. Each antenna face had the capability of operating with 5354
elements, but only 1792 active transceiver modules were used for transmission. Extra ele¬
ments and a narrow beam were used on receive. (The remaining elements were for future
growth.) A transmitting module was made up of seven Class-C silicon bipolar transistors in
a 1-2-4 amplifier configuration. It has been said22 that “system performance is maintained
with as many as 200 modules per face inoperative.” An enlarged version of the Pave Paws
radar replaced the parabolic torus antennas used in the original Ballistic Missile early Warn¬
ing System (BMEWS). This version had 850-kW peak power with a 0.30 duty cycle.

THAAD Ground Based Radar This is an example of an active-aperture phased array

radar with 25,344 elements, a very large number. It was discussed in Sec. 9.9 and its pic¬
ture shown in Fig. 9.36. With such a large number of elements, the cost of an individual
T/R module must be kept small. If, for example, the cost of an individual T/R module
were $1000 each, the total cost of just the modules for this radar would be twenty-five
million dollars.
06 chapter 10 ® Radar Transmitters

Solid-State Devices Used in Radar1 As has been mentioned, the transistor amplifier has
been the device usually used for radars with high-power solid-state transmitters. At the
lower microwave frequencies, the silicon bipolar transistor is usually used; and at the
higher microwave frequencies, it has been the gallium arsenide (GaAs) FET transistor. At
the higher frequencies, the solid-state power device can also be incorporated as part of a
microwave monolithic integrated circuit (MMIC).
The silicon bipolar transistor has been used at microwave frequencies below about
3 GHz (S band). According to Olson,23 a typical internally matched silicon bipolar tran¬
sistor operating as a class-C amplifier over the frequency range from 2.7 to 2.9 GHz with
a pulse width of 50 jus, 10 percent duty cycle and a supply voltage of 40 V, can have a
minimum power output greater than 100 W, a minimum gain of 6.5 dB, and a minimum
efficiency of 35 percent.
The power output of the silicon bipolar transistor decreases with increasing frequency.
At the higher microwave frequencies, the gallium arsenide FET, often in the form of a
MESFET (metal semiconductor field-effect transistor) is capable of greater power than
the silicon bipolar transistor. Transistors should be operated at a high duty cycle since the
peak power output for pulsed operation cannot be significantly increased over that of CW
operation. At X band, the power output of such devices might be 10 W. Other devices that
have been considered for solid-state power sources include23 GaAs HEMT (high electron
mobility transistor); GaAs-based pseudomorphic HEMT, or PHEMT; GaAs heterojunc¬
tion bipolar transistor (HBT); and devices employing unconventional materials such as
silicon carbide and semiconductor diamond for high temperature, high power operation.24
At the higher microwave frequencies where compactness in size is desired, the mi¬
crowave monolithic integrated circuit (MMIC) has been of interest for T/R modules. Ac¬
tive and passive circuit elements are formed on a semi-insulating semiconductor substrate,
usually GaAs to create system architectures that are difficult to realize with less integrated
technologies. The benefits of MMIC are due in large part to the batch processing of both
the active and passive components on the same substrate. Borkowski17 lists the advan¬
tages of MMIC for radar as low cost, increased reliability, increased reproducibility, and
small size and weight. The nonrecurring costs of engineering design of MMICs, however,
can be high and design might require a relatively long time. Since MMICs are not well
suited for tweaking of the circuits once manufactured, the designs must be tolerant to vari¬
ations in the processing. Olson23 states that the power available from MMICs is about
10 W in the frequency range from 3 to 10 GHz and then decreases with increasing fre¬
quency at a rate of 6 dB per octave.

Advantages of Solid State The solid-state transistor amplifier has been of interest for
radar transmitter applications because of the following:17

• Individual solid-state devices have long MTBF (mean time between failures).
8 Maintenance is relatively easy with the modular construction of solid state. (A de¬
fective module is pulled out and replaced by another.)
* Very wide bandwidths can be obtained (up to 50 percent or more).
® No cathode heater is required (no warm-up time and no heater power to reduce the
overall transmitter efficiency).
 10.3 Solid-State RF Power Sources 707

8 Solid-state devices operate at much lower voltages (order of tens of volts) than RF
power tubes (order of tens of kilovolts).
• No pulse modulator is required. (When operated as a class-C amplifier, the transis¬
tor is self-pulsing in that it automatically turns on when the RF drive signal is ap¬
plied and it automatically turns off when the drive signal is turned off.)
• Solid-state transistor amplifiers have low noise and good stability (important for de¬
tection of small targets in the presence of large clutter echoes).

Another attribute often claimed for a solid-state radar is that many individual devices
can fail without significant effect on the overall transmitter power (graceful degradation).
The power output in dB varies as 20 log r, where r = ratio of the number of operating
devices to the total number of devices.17 This is correct in principle, but in practice there
can be catastrophic failure modes for a solid-state transmitter and, eventually, the mod¬
ules that fail must be replaced, even if they fail “gracefully.” Except for its long pulses
and high duty cycle the solid-state transistor is well suited for use in an active-aperture
phased array where each element contains its own transmitter/receiver module.
When solid-state transmitters were first proposed for radar, it was said they would be
lighter in weight and lower in cost than vacuum tube transmitters. It is not obvious that
this has occurred. In some cases in the past when a solid-state transmitter replaced a high-
power vacuum-tube transmitter in an existing radar, the solid-state transmitter was heav¬
ier and cost more.

Systems Implications of Solid-State Devices Pulse radars typically have been character¬
ized in the past by low duty-cycle waveforms with typical values of duty cycle ranging
from approximately 0.001 to 0.01. Power vacuum tubes are well suited for low duty cy¬
cles. For a given average power their peak power can be increased by a factor of 1000 or
more with little penalty other than the practical problem of making the insulators able to
stand off the higher voltages. Semiconductor power devices, on the other hand, cannot be
efficiently operated at low duty cycles. For a given average power, the peak power might
be less than ten times the average power. Thus replacing a vacuum-tube transmitter with
a solid-state device usually means the radar must use high duty-cycle waveforms. High
duty cycles mean long pulse widths which have the disadvantage of long minimum ranges.
When a short minimum range is important,-more than one pulse width might have to be
transmitted. The long pulses of solid-state transmitters require pulse compression to
achieve good range resolution. The technology of pulse compression has been widely used
in radar, but it does have some limitations that short-pulse waveforms do not have. Sel¬
dom, however, is the cost of pulse compression and the cost of multiple waveforms con¬
sidered as a solid-state transmitter cost, even though they increase the total cost of the
radar system and are not needed with many vacuum-tube transmitters.
There are at least two reasons why the cost of radars with solid-state transmitters is
often higher than those with comparable vacuum-tube transmitters.12 One is that the ef¬
ficiency of solid-state transmitters is generally less than that of a high-power vacuum-tube
transmitter. The other is that the cost of obtaining power is greater when the total power
is obtained with more than one power-source unit. An advantage of low-power solid-state
devices for computers and low-power transmitters is that they can be made very compact.
:hapter 10 • Radar Transmitters

A lot can be placed on a small chip. When the solid-state device has to handle high power,
however, as it would for a radar transmitter, there can be a problem in dissipating the heat
generated by the power sources. Solid-state devices that have to handle high power have
to be spread out over a greater extent of circuit board area to avoid exceeding the heat
transfer limits. The size and weight of the solid-state amplifier are therefore determined
by the power densities that the amplifier can handle rather than by the size of the indi¬
vidual components. Thus the advantages of photolithographic fabrication of high-density
low-power solid-state integrated circuits are not available when the power is high. Spread¬
ing the solid-state devices over a wide area in order to provide for heat dissipation can
result in lower efficiency due to higher power-combining losses in transmission lines and
in combiners. Dissipation of the higher heat levels requires heavier heat sinks and results
in a heavier transmitter.12
It has been known for a long time that the cost of a single high-power vacuum tube
varies as the square root of the output average power. Thus, the lowest cost tube trans¬
mitter for a given total power output is the transmitter that uses a single high-power tube
rather than multiple tubes. Symons12 indicates that the cost of a transmitter made up of
multiple devices, such as is necessary when using high-power solid-state devices, increases
almost linearly with the number of devices. At high powers, therefore, it should be ex¬
pected that the single vacuum-tube transmitter will be of lower cost than a solid-state
transmitter made up of many modules generating the same total power.
Individual solid-state devices can have a much longer life and lower failure rate than
an individual vacuum-tube power source. The life of a solid-state transmitter, however, is
determined not by the life of a single transistor or a single module but by the life of all
the modules and the many other components that make up a transmitter. Vacuum-tube
transmitters, when well designed and properly operated, have been known to achieve very
long lifetimes, mentioned previously in this chapter. Maintenance of a solid-state trans¬
mitter should be easier than maintenance of a high-power vacuum tube transmitter, but
the life of a properly designed and operated vacuum tube transmitter should be quite long
and not be a serious system problem.
Compared to the high-power vacuum tube, the solid-state transmitter has advantages,
but it also has some serious limitations. It is not obvious that current solid-state trans¬
mitter technology will cause the high power RF vacuum tube to disappear. Solid-state will
be used when its particular advantages are more important than its limitations and higher
cost. As happens often in engineering, the radar systems engineer has a number of choices
when it comes to selecting the type of RF power source to use for a particular radar sys¬
tem application. The solid-state transmitter is just one of many possibilities that have to
be considered for any particular application—unless the customer insists otherwise.

MAGNETRON

The magnetron has been the only high-power RF power source used for radar that is a
power oscillator rather than a power amplifier. It is a crossed-field device in that its elec¬
tric field and its magnetic field are perpendicular to one another. The compact size
and efficient operation of the magnetron at microwave frequencies allowed radars to be
 10.4 Magnetron 709

small enough to fly in military aircraft, be mobile for ground warfare, and even be used
on submarines.

Coaxial Magnetron A major improvement in the power, efficiency, stability, and life of
the original magnetron architecture came about with the coaxial magnetron introduced in
the 1960s. The key difference was the incorporation of a built-in stabilizing cavity sur¬
rounding the conventional magnetron. Figure 10.5 is a sketch of the cross section of the
circular geometry of a coaxial magnetron. At (he center is the “fat” oxide-coated cathode.
Surrounding the cathode are a number of RF resonant cavities defined by the radial vanes.
Between the cathode and the resonant cavities is the interaction space where the electrons
interact with the d-c electric field and the static magnetic field in such a manner that the
electrons give up their d-c energy to the RF field. The crossed electric and magnetic fields
cause the electrons to be “bunched” almost as soon as they are emitted from the cathode.
After bunching, the electrons move along in a traveling-wave field that is almost the same
speed as that of the electrons.
The frequency of a coaxial magnetron can be changed by mechanically moving one
of the end plates, called a tuning piston, of the stabilizing cavity. (The end plate is located
in the plane of the paper of Fig. 10.5 and is not shown.) The tuning piston can be posi¬
tioned mechanically from outside the vacuum by means of a vacuum bellows.
There is also an inverted form of the coaxial magnetron (an inside-out) version with
the anode and resonant cavities in the center and the cathode around the outer perimeter
of the tube. It is supposed to provide better performance at higher frequencies when the
cavity becomes small and the regular type of coaxial magnetron would result in a small
cathode.

Figure 10.5 Cross-sectional

sketch of a coaxial cavity
magnetron.
CHAPTER 10 ® Radar Transmitters

tt Mode It is not easy to describe the theory of operation of a magnetron in a simple

manner, so no attempt is made to do so here. A magnetron can oscillate at a number of
different, closely spaced frequencies due to various possible configurations of the RF field
that can exist between the cathode and the resonant cavities. These different RF field con¬
figurations, along with coupling among the many cavity resonators of the magnetron, re¬
sult in different modes of oscillation. The magnetron can shift, almost unpredictably, from
one mode to another (which means from one frequency to another) as the voltage changes
or as the input impedance that the magnetron sees changes. The shift from one mode to
another (often called moding) is especially bad since it can occur when the radar antenna
scans and views different environments. It is important to avoid moding.
The preferred magnetron mode of operation is the so-called tt mode that occurs when
the RF field configuration is such that the RF phase alternates 180° (tt radians) between
adjacent cavities. The advantage of the tt mode is that its frequency can be more readily
separated from the frequencies of the other possible modes. (An V-cavity magnetron has
N/2 possible modes of oscillation. The tt mode oscillates at only a single frequency, but
the other modes can oscillate at two different frequencies, so that the magnetron can os¬
cillate at a total of N - 1 different frequencies.)
In the coaxial magnetron, the output of every other resonant cavity is coupled to a
stabilizing cavity that surrounds the anode structure, as indicated in Fig. 10.5. The output
power is then coupled from the stabilizing cavity. The cavity operates in the TE0n mode
with the electric lines closed on themselves and concentric with the circular cavity. The
RF current at every point on the circumference of the cavity has the same phase, so that
the alternate slots which couple to the stabilizing cavity are of the same phase as required
for 7r-mode operation.

Coaxial Magnetron Life The power that can be produced by a magnetron depends on
its size. A larger size means more resonators, which then makes it more difficult to sep¬
arate the various modes of oscillation in a conventional magnetron. The coaxial mag¬
netron, however, with stabilization controlled by the TE011 outer cavity permits stable op¬
eration with a larger number of resonant cavities, and thus with greater power. The anode
and cathode structures of a coaxial magnetron can also be bigger than those of a con¬
ventional magnetron, which further allows operation at larger powers. The larger struc¬
tures permit more conservative design, with the result that the coaxial magnetron exhibits
longer life and better reliability than conventional magnetrons, in addition to having a
more stable operation. The operating life of a coaxial magnetron tube can be between
5000 and 10,000 hours, which represents a five- to twenty-fold increase compared to con¬
ventional magnetrons.25 It has been said26 that a VMS-1143 S-band coaxial magnetron
operating at 3 MW of peak power in an AN/FPS-6 height-finder radar exceeded a life of
50,000 hours. This tube was likely one-of-a-kind, but it indicates that a magnetron does
not necessarily have to have a short life, as once was the case.
Just after World War II and during the 1950s, the life of the early magnetrons was as
low as 200 hours mean time between failures, which probably explains why some have
had the impression that power vacuum tubes were inherently unreliable and of short life.
The demonstrated long lifetimes of the coaxial magnetron and even longer lifetimes of
the modem klystron and TWT linear-beam tubes offer ample evidence that the use of
 10.4 Magnetron 711

power vacuum tubes need not result in unreliable radars. (The reader might try to recall
how often, when watching TV or listening to the radio, the transmission shut down and
went off the air because of a failure in the tube transmitter.)

Systems Aspects of the Magnetron The magnetron transmitter was used in a large num¬
ber of different types of radar transmitters. At one time it was the most popular radar trans¬
mitter. Its use, however, has diminished greatly because of the more demanding require¬
ments of modern radars that it cannot readily meet, but which can be satisfied by other
RF power sources. Weik7 describes the problems encountered with the use of the mag¬
netron in his chapter “Transmitters” in the Radar Handbook. The major limitations of the
magnetron are its limited average power and poor ability to see moving targets in heavy
clutter.
Although the magnetron can produce a peak power of several megawatts, its average
power is limited to about one or two kilowatts. This may be sufficient for some medium-
range radars and for civilian air-traffic control radars that use large antennas, but it is not
large enough for many military radar applications. The magnetron is usually smaller in
size than many other types of RF power sources, but this is due in part to its low average
power.
During World War II, air-surveillance radars did not have an MTI (moving target in¬
dication) capability, except for the S-band AN/CPS-1 MEW (Microwave Early Warning)
radar which only became available in small numbers near the end of the war. In order for
radar to detect aircraft in World War II, targets had to be in the clear outside of the clut¬
ter. The bomber aircraft of that time unknowingly cooperated since they were not designed
to fly at low altitude below the radar coverage. The early analog MTI radars could use
only a single or a double delay-line canceler, so they were limited in MTI improvement
factor (or clutter attenuation) to about 20 dB. The magnetron itself limits the improve¬
ment factor to perhaps 30 or 40 dB. Therefore, the magnetron was not the limiting factor
in early MTI radar performance. This changed when digital signal-processing of MTI sig¬
nals became available that allowed much better values of MTI improvement factor. Now
power amplifiers such as the klystron, TWT, or the transistor have to be used—and not
the magnetron—in order to detect small moving targets in heavy clutter, consistent with
the full capabilities of digital doppler-signal-processing.
There are several other factors that are not favorable with the magnetron. Its pulse
widths are limited from just under 0.1 /as to about 100 /as, but this is usually not a prob¬
lem. However, modulating the pulse with frequency or phase to achieve pulse compres¬
sion is quite difficult with a magnetron and has not been done operationally. The signal
is not coherent from pulse to pulse so that in MTI radar the phase of the coho (reference
oscillator) in the receiver has to be reset every time a new pulse is transmitted. The mag¬
netron frequency drifts with time, which requires that the frequency of the local oscilla¬
tor in the receiver be continually tuned to the transmitter’s frequency (whatever it might
be). Magnetrons are noisy and can produce electromagnetic interference at frequencies
outside their design frequency range.

Marine Radar Magnetrons The magnetron has proven to be a tube well suited for civil
marine radars. These are small devices that generate peak powers between 3 and 75 kW
I chapter 10 • Radar Transmitters

with low average powers of a few watts to a few tens of watts. The marine radar customer
demands reliability. When a commercial vessel goes to sea, its captain wants the radar to
still be operating when the ship returns to its home port. An example of a magnetron for
a civil marine radar is the third generation MG5241 manufactured by EEV of Chelms¬
ford, England. It is an 18-cavity A-band magnetron, which produces a peak power of
12.5 kW with an efficiency of 43 percent. The manufacturer claims an expected typical
life of over 10,000 h and guarantees a minimum life of 3000 h. Its weight is 625 g (1.4
lb) and it has a volume of 315 cm3. It operates at a fixed frequency within the band 9.38
to 9.44 GHz.

1.5 CROSSED-FIELD AMPLIFIERS

The crossed-field amplifier (CFA) resembles the magnetron in that it employs a magnetic
and electric field that are perpendicular to one another.1'28 It is also similar in appearance
to the magnetron, except that the RF circuit is interrupted to provide the input and output
connections, Fig. 10.6. CFAs have high efficiency (40 to 60 percent), use lower voltage
than linear-beam tubes, and are lighter in weight and smaller in size. They are of wide
band (10 to 20 percent), with high peak and average power, and have good phase stabil¬
ity; but their gain is relatively low. They are sometimes used with a high-gain, but lower
power, TWT that serves as the driver for the CFA.

CFA Operation There are several different types of crossed-field amplifiers, and they all
employ a slow-wave circuit, cathode, and input and output ports. For radar, CFAs usually
have the form diagrammed in Fig. 10.6, which is reentrant with distributed emission. Dis¬
tributed emission means that, like the magnetron, the cathode is adjacent to the full length
of the RF structure. Electrons are emitted from the cylindrical cathode, which is coaxial
to the RF slow-wave circuit that acts as the anode. The electrons, under the action of the
crossed electric and magnetic fields, form into rotating electron (space-charge) bunches,
or spokes. These bunches drift along the slow-wave circuit in phase with the RF signal

re 10.6 Simple representation of a

sed-field reentrant CFA showing the drift
on and the control electrode.
 10.5 Crossed-Field Amplifiers 713

and transfer their d-c energy to the RF wave to produce amplification. The spent electrons
that remain after their energy is extracted are collected by the slow-wave anode structure.
The electrons that are not collected after their energy is extracted at the output are per¬
mitted to reenter the RF interaction area at the input, which is the reason such a tube is
called reentrant. Some of the reentering electrons contain modulation (bunched electrons)
that will be amplified in the next pass around the RF circuit. To prevent this, a drift space
is included between the output and input ports. In the drift space, space charge forces
cause the electron bunches to disperse, removing any modulation that accompanies the
reentering electrons.

Cold-Cathode Emission In high-power CFAs, the electrons can be generated by cold-

cathode field emission rather than thermionic emission with a heated cathode. Cold-
cathode emission requires the presence of both the d-c voltage between cathode and an¬
ode as well as the RF drive signal applied to the tube. The initiation of electron-emission
build-up results from the relatively small number of free electrons present near the cath¬
ode surface. Emission is sustained by those electrons not collected by the anode and which
are returned to the cathode by the action of the RF field and the crossed electric and mag¬
netic fields. When these electrons strike the cathode they produce secondary electrons that
maintain the electron emission process. There is little pulse-to-pulse time jitter in the start¬
ing process, and the buildup time is quite rapid (<10 ns).28

Insertion Loss A CFA has low insertion loss, perhaps less than 0.5 dB. Sometimes this
can be an advantage in a multistage transmitter. By omitting the application of d-c volt¬
age to the final stage of a multistage transmitter, the lower level RF drive can be fed di¬
rectly through the final stage with little attenuation. This allows a radar with such a trans¬
mitter to have two power levels, which might be of interest for some system applications.
The low insertion loss of a CFA means that the RF drive power will appear at the
output tube with little attenuation. In a low-gain amplifier, such as the CFA, the input
power that appears at the output can be a sizable fraction of the total, perhaps one-tenth
or more. The conversion efficiency of a CFA is defined as

RF power output - RF drive power

Efficiency = [10.2]
d-c power input

When the RF drive power is included in the output power rather than omitted as it is in
Eq. (10.2), it is sometimes called the power-added efficiency. That is, power-added effi¬
ciency is the total RF power out divided by the d-c power in. Tube engineers like to quote
the power-added efficiency instead of Eq. (10.2), since it results in a higher value.

Forward- and Backward-Wave CFAs The interaction of the electron bunches with the
RF signal may be with either a forward traveling wave or a backward traveling wave. The
type of interaction is determined by the slow-wave circuit. A forward-wave interaction
takes place when the phase velocity and the group velocity of the propagating signal along
the slow-wave circuit are in the same direction. (The group velocity is the velocity with
which energy is propagated along the slow-wave circuit, and the phase velocity is the ve¬
locity of the RF signal on the slow-wave circuit as it appears to the electrons.) To achieve
chapter 10 ® Radar Transmitters

amplification, the phase velocity must be near the velocity of the electron stream. A back¬
ward-wave interaction, as in the CFA device known as the amplitron, takes place when
the phase velocity and the group velocity are in opposite directions. The forward-wave
CFA can operate over a broad range of frequency with a constant anode voltage, and with
only a small variation in the output power. On the other hand, the power output of a back¬
ward-wave CFA, with a constant anode voltage, varies with frequency. It is like a volt¬
age-tuned amplifier. The power output can vary 100 percent for a 10 percent change in
frequency.29 It is possible, however, with conventional modulator techniques to operate a
backward-wave CFA over a wide band with little change in output power. The line-type
modulator can be compensated for the power variation with frequency and can hold the
variation of output power within acceptable levels. When the anode voltage is properly
adjusted, the bandwidth of a backward-wave CFA might be 10 percent. Operation of a
forward-wave CFA is more like that of a TWT, and it can obtain bandwidths up to about
20 percent.28

High-Gain CFA1'30 The gain of conventional pulsed crossed-field amplifiers is typi¬

cally 8 to 15 dB. By designing the cold cathode as a slow-wave circuit, and introducing
the RF drive at the cathode emitting surface, it is possible to achieve 15 to 30 dB of gain
in a high-power pulse CFA with power, bandwidth, and efficiency comparable to con¬
ventional designs. The RF output is taken from the anode slow-wave circuit. This is known
as a high-gain CFA or a cathode-driven CFA. Its higher gain means that less drive power
is required. The cathode-driven CFA can also be made to have lower noise than a con¬
ventional CFA by 10 to 20 dB if a slightly different configuration is used. It has been
said31 that signal-to-noise ratios greater than 70 dB/MHz were obtained, which are claimed
to be comparable to linear beam tubes and 20 to 30 dB better than conventional CFAs.
However, both high gain and low noise cannot be obtained simultaneously with the same
configuration.
A 1.2-MW peak power .S'-band cathode-driven CFA operating from 3.1 to 3.5 GHz
(12 percent bandwidth) achieved 23-dB gain at saturation and 30 dB at reduced peak
power, with an efficiency of 53 percent when employing a drive power of 6 kW.

Modulating a CFA The CFA can be pulsed-modulated by turning on and off the high
voltage between the anode (at ground potential) and the cathode (at a large negative po¬
tential). This is called cathode pulsing. A forward-wave CFA with cold-cathode electron
emission, however, can be pulsed without the need for a high-power modulator as required
for cathode pulsing. The d-c operating voltage is applied continuously between cathode
and anode. The tube remains inactive until the application of the RF input pulse starts the
electron emission process, and amplification then takes place. At the end of the RF drive-
pulse the electrons remaining in the tube must be cleared from the interaction area to avoid
causing feedback that generates oscillation or noise. In reentrant CFAs, the electron stream
can be collected after removal of the RF drive-pulse by mounting an electrode in the cath¬
ode, but insulated from it, in the region of the drift space between the RF input and
output ports. This is called a cutoff, or control electrode, and was indicated in Fig. 10.6.
A short positive pulse is applied to the cutoff electrode at the termination of the pulse
to quench the remaining electron current. This method of modulation is called d-c
 10.6 Other RF Power Sources 715

operation. Weil1 has said that d-c operation has been seldom used because it requires a
large capacitor bank to limit droop of the pulse and because an arc in a d-c operated tube
requires a crowbar protective device to quench the arc, which interrupts the operation for
a few seconds instead of for only a single pulse.
It is also possible to turn the CFA on and off with just the RF drive pulse, without
the need for a positive pulse applied to the cutoff electrode at the end of the drive pulse.
The is called RF keying, and is a simple method for modulating a CFA. It has not been
widely used, however, since there are factors other than modulator size that determine the
best method of modulation.1

System Implications CFAs in the past have been employed for high-power air-surveil¬
lance radars, for power applied at the subarrays of a high-power phased array radar, and
as a power booster following a magnetron oscillator. The low gain of CFA, however, re¬
quires that there be multiple stages. When used in an amplifier chain, the CFA is gener¬
ally found in only one or two of the highest power stages. It can be preceded by a medium-
power, high-gain traveling-wave tube, a combination that takes advantage of the best
qualities of both tube types. The TWT provides high gain and the CFA allows high power
to be obtained with high efficiency and lower voltage.
The electrons in the rotating space-charge bunches do not have identical velocities so
that random currents are induced in the slow-wave structure, which generates bioadband
noise. The noise levels in a CFA, therefore, will be higher than those of a linear-beam
tube by about 20 to 30 dB.32
The possibility of pulsing the CFA in a simpler manner than with a cathode pulser
has been an attractive feature; but d-c operation and RF keying have limitations that have
made cathode-pulsing preferred even though it is heavier. Since high voltage remains on
the tube between pulses with d-c operated CFAs, serious levels of noise may be gener¬
ated even though there is only a small amount of beam current flowing through the tube.
With cathode pulsing, on the other hand, high voltage is removed between pulses so that
noise is not normally encountered. The increased interpulse noise of the CFA without a
cathode pulser, as well as its high level of in-band spurious noise can prevent attaining
good MTI performance (large MTI improvement factors) and low time-sidelobes in pulse-
compression systems.
The CFA does not seem to generate as much interest as it once did, and it appears to
have been overtaken by linear-beam tubes when high performance is required.

10.6 OTHER RF POWER SOURCES

In addition to the RF power sources already mentioned, there have been several other RF
devices that been used or proposed for radar application.

Microwave Power Modules (MPM)~~ The microwave power module, outlined in Fig. 10.7,
combines in a single unit a solid-state MMIC (monolithic microwave integrated circuit)
amplifier driving a moderate-power helix traveling-wave tube, along with an integrated
16 chapter 10 ® Radar Transmitters

gure 10.7 Microwave power

odule.
RF
RF
input output

Prime
power

power conditioner in a compact lightweight package. It provides an RF power source with

high efficiency, wide instantaneous bandwidth, low noise, and average power levels from
several tens to several hundreds of watts. It is smaller and lighter than comparable TWT
and solid-state RF power sources and is also capable of operating at high ambient tem¬
peratures. The gain of an MPM (nominally 50 dB) is divided between the solid-state dri¬
ver and the TWT power booster in the ratios from 20/30 to 30/20. The MPM is claimed
to be competitive to a TWT or a solid-state power amplifier for radar applications as well
as lor electronic warfare. It leverages the advantages of both solid-state and vacuum-tube
technologies while minimizing their disadvantages.
The MPM seems best suited for the higher microwave frequencies, from S band
to Ku band, approximately 2 GHz to 40 GHz. It is expected that an MPM operating
over a frequency range from 6 to 18 GHz would produce a peak power of 100 W (up to
a 100 percent duty cycle) and be able to be packaged in a space 5/16 in X 4 in X 6 in.
The MPM is claimed to have an efficiency three times better than that of a comparable
solid-state power amplifier and an improvement in noise 30 dB better than a TWT alone.
It has been suggested that the MPM can be employed in C- and X-band phased arrays!
synthetic aperture radars for unmanned airborne vehicles (UAV), missile seekers, and air¬
borne turbulence-warning radars.
Although the MPM has been considered mainly for wideband (nonradar) applications
operating over one or more octaves, it can also be used for radar with more usual band-
widths. An example is a C-band module described by Smith, Armstrong, and Duthie34
which can operate over the band from 4 to 6 GHz with an efficiency greater than 35 per¬
cent and CW power greater than 125 W. If its operation is restricted to frequencies from
4.6 to 4.9 GHz, the output CW power is 200 W and its efficiency is 50 percent. It is of
low weight and small size, producing more than 70 W per pound and 4 W per cubic inch.
A limitation for radar application is that it requires a high duty factor (preferably unity or
perhaps 50 percent).

Grid-Control Tubes This is the microwave version of the classical triode or tetrode vac¬
uum tube. These tubes employ a cathode to generate electrons, an anode to collect them,
and one (if a triode) or two (if a tetrode) control grids in between. A voltage applied to
 10.6 Other RF Power Sources 717

the control grid acts as a gate, or valve, to control the number of electrons traveling from
the cathode to the anode. By varying the voltage on the control grid, the number of elec¬
trons that reach the plate also varies. The process by which the electron density of the
electron stream is modulated by the signal on the control grid to produce amplification is
called density modulation. The grid-controlled tube is capable of high power, wide band¬
width, good efficiency, and inherent long life; but it is of low or moderate gain. It can be
used only at the lower radar frequencies.
The performance of density modulated grid-control vacuum tubes is limited by the
time it takes for the electrons to transit from the cathode to the anode. The transit time
should be small compared to the period of the RF signal to be amplified. For this reason,
grid-control tubes have been limited to frequencies below 1 GHz. To minimize transit¬
time effects, the complete RF input and output circuits and the electrical interaction sys¬
tem are within the vacuum envelope.
Grid-control tubes have been used with success in HF over-the-horizon radar, and in
VHF and UHF radars, including the U.S. Navy’s E2C airborne early warning radar,35 and
the U.S. Air Force’s AN/FPS-85 satellite surveillance radar.36 Outside of radar, the grid-
control vacuum tube has been widely used in commercial VHF and UHF TV transmit¬
ters. It is not likely, however, that the grid-control tube will be used much for radar in the
future. Solid-state transmitters, even though they may cost more than vacuum tubes, seem
to be preferred by those who buy radars at the frequencies where grid-control vacuum
tubes have been used previously.

Inductive Output Tube (IOT), or Klystrode3 This device dates back to work in 1939 by An¬
drew Haeff,37 who tried to extend density modulated vacuum tubes to microwave fre¬
quencies. Haeff called his device an Inductive Output Tube (IOT). He described an IOT
that produced 100-W CW power at 450 MHz with 35 percent efficiency and 10-dB gain.
This was quite good for its time. Nothing further happened since interest then was mainly
on velocity modulated linear-beam tubes.
The IOT was independently reinvented about 30 years later by D. H. Preist and
M. B. Shrader under the name Klystrode. The name was chosen to signify that the device
resembled the klystron in the region between the anode and the collector and it resem¬
bled a tetrode in the region between cathode and anode. According to Priest and Shrader,38
Haeff realized that the conventional triodes and tetrodes were limited by their use of in¬
tercepting grids, so in his IOT he replaced the wire grids with an aperture that did not in¬
tercept the electrons. A coaxial magnetic field confined the electron stream, as in a kly¬
stron or TWT. The action of the grid was to density modulate the electrons, as in a
conventional triode, to form bunches of electrons. RF energy was removed from the
bunched electron beam by passing it through a resonant cavity that extracted the kinetic
energy of the high-velocity electrons. The spent electrons were not collected by the res¬
onators, but by a separate collector. The IOT was like the klystron, except that the den¬
sity modulation of the electron beam was performed by a grid rather than by an input res¬
onant cavity and drift space that induced velocity modulation on the electrons as in the
klystron.
The Klystrode was developed mainly for UHF TV. It can produce many tens of kilo¬
watts average power at high efficiencies (50 to 60 percent) with power gains of 18 to
718 chapter 10 ® Radar Transmitters

25 dB. Although it has not been used in radar, it has the potential to provide better per¬
formance for UHF radars than the tetrodes used previously.

Constant-Efficiency Amplifier (CEA) One reason for mentioning the IOT and the Klystrode
in the above is that they can be modified to produce an amplifier whose efficiency is ap¬
proximately independent of the power output. Such a device would be of interest for radar
when it is desired to shape the radar pulses in order to reduce the time sidelobes of pulse-
compression waveforms or to reduce the out-of-band interference generated by a rectan¬
gular waveform. Conventional radar RF power sources, such as discussed in this chapter,
cannot operate with shaped pulses without a loss in efficiency. A so-called Constant Effi¬
ciency Amplifier (CEA), however, can be obtained by combining the Inductive Output Tube
(IOT) with a multistage depressed collector similar to that used in klystrons and TWTs.12
The CEA was developed for the television industry. It is claimed that with a CEA the prime
power requirements of a TV transmitter can be reduced to one-half compared to a con¬
ventional tube transmitter39 and to one third the prime power of a silicon-carbide solid-
state transmitter.40 The CEA, however, does not operate at frequencies higher than UHF.

Gyrotrons41-41 The power available from the microwave-radar power sources discussed
thus far in this chapter decreases as the frequency is increased. This is due to the reso¬
nant structures of the devices becoming smaller with increasing frequency (decreasing
wavelength) and the difficulty in removing heat dissipated in small structures. Conse¬
quently the power output of a particular type of generator varies approximately inversely
as the square of the frequency. The gyrotron, on the other hand, does not have this type
of frequency dependence since it does not employ a resonant slow-wave structure. In¬
stead, it is based on a fast-wave structure such as a smooth circular tube (one where the
phase velocity of the electromagnetic wave is greater than the speed of light). The diam¬
eter of the gyrotron circuit can be several wavelengths and the electron beam need not be
placed close to the RF structure. Since the size limitations of conventional microwave
power sources with resonant circuits are not present in gyrotrons, their power handling
capability can be considerably greater. The gyrotron is of interest as a potential source of
high peak and average power at millimeter wavelengths. It has also been considered for
operation at microwave frequencies, but it has not been able to compete with more con¬
ventional microwave power sources.
The gyrotron, also known as a cyclotron resonance maser, employs a strong exter¬
nally applied axial magnetic field to cause electrons within the circular fast-wave struc¬
ture to rotate at the electron cyclotron frequency, which is coc = eB0/my, where e = elec¬
tron charge, m = electron rest mass, B0 = axial magnetic field, y is the relativistic factor
which is [1 + (e/mc2)V0], c = velocity of light, and Vf) = beam voltage. The beam volt¬
age and the corresponding electron velocity are high enough to cause relativistic effects.
The electrons follow helical paths around the lines of the magnetic field in the presence
of an electromagnetic wave with a transverse component of electric field. The electrons
become phase-bunched in their cyclotron orbits as a result of the relativistic mass change
of the electrons. Electrons that lose energy to the electromagnetic wave become lighter
and accumulate phase lead and catch up with the electrons that gain energy and become
heavier and accumulate phase lag.
 10.6 Other RF Power Sources 719

The frequency of the gyrotron is determined by the magnetic field and not by the
characteristic size of the interaction region as it is in microwave power tubes. The mag¬
netic field must be quite large in order to generate cyclotron oscillations at the higher fre¬
quencies. For this reason, the magnets used for millimeter wavelength gyrotrons are usu¬
ally superconducting, which can be a burden for some applications, especially if the device
has to be operated in a cryostat at liquid helium temperatures. However, the development
of magnets based on high-temperature superconducting materials and efficient closed-
cycle refrigerators, or cryocoolers, offers the possibility of using supercooled gyrotrons
in mobile platforms such as ships and aircraft.
Since the diameter of the gyrotron RF circuit is normally large compared to the wave¬
length of the electromagnetic wave it generates, it is possible to have a higher order mode
or multiple modes of the electromagnetic field. Operation in more than one mode can re¬
sult in operation at more than one frequency. The design of a gyrotron requires that care
be given to insure stable, single-frequency operation in only one mode.
The gyrotron oscillator at 94 GHz can produce CW power greater than 100 kW and
peak pulse power of 1 MW, with efficiencies of about 30 percent. A quasi-optical gyro¬
tron was tunable from 80 to 130 GHz (a half octave) by varying the magnetic field.44 The
power was relatively constant over this frequency range, averaging about 60 kW.
The gyrotron can be operated as an amplifier as well as an oscillator. Generally, more
power can be obtained from a gryrotron oscillator than the gyrotron amplifier, but the am¬
plifier might have some advantages in radar when doppler processing is important—just as
at microwave frequencies. The electron beam of a gyroklystron passes through two or more
resonant cavities with standing-wave fields separated by drift spaces. A gyroTWT is one
which operates with traveling wave fields, similar to a microwave TWT. The gyroklystron
has a smaller relative bandwidth than the gyroTWT, but it is capable of higher gain, effi¬
ciency, and output power. A gyrotwystron operates similarly to a microwave twystron in that
it uses standing-wave fields to bunch the electrons and a traveling-wave field to extract the
energy. The relative bandwidths of millimeter-wave gyrotron amplifiers are generally less
than the relative bandwidths that can be obtained with microwave power amplifiers.
The specifications for a particular experimental gyroklystron designed for radar op¬
eration at a center frequency of 94 GHz required that it have an average power of 10 kW,
peak power of 80 kW, efficiency of 20 percent, and bandwidth of 600 MHz.45

Multiple-Beam Klystrons46 In a conventional klystron with a single electron beam, the

power can be increased by increasing the already high beam voltage. Instead of increas¬
ing the beam voltage to obtain greater power, it is possible to employ many electron beams
that pass through individual channels located in a single multichannel drift tube. The to¬
tal power is the sum of the power extracted from each of the lower-current electron beams.
The number of beams has been from 6 to 61. Such a power generator is known as a mul¬
tiple-beam klystron (MBK).
The significant reduction in beam voltage results in reduced size and weight com¬
pared to a conventional klystron of comparable power. Its magnet and power supply are
smaller and lighter. The geometry of the multiple beams of the MBK allows an increase
in bandwidth because of an increase in perveance. The lower voltage also can eliminate
the need for lead shielding to screen against X-ray radiation.
I chapter 10 ® Radar Transmitters

The high-power multiple-beam klystron was first seriously examined in the U.S. in
the 1950s, but interest was not sustained because the needs for high power were satisfied
by other more conventional types of klystrons. Tube engineers in the Soviet Union began
seriously investigating the MBK in the 1970s, and were successful in producing RF power
sources that have been widely used in Russian radar systems. An example of an MBK
produced and marketed by the Russian Company ISTOK is their IKS-9007, a six-cavity,
36-beam klystron. It operates at 3.3 GHz with a 200-MHz bandwidth (6 percent), peak
power from 500 to 800 kW, duty cycle of 0.02, gain of 40 dB, and an efficiency of 40 to
50 percent. The beam voltage is 28 to 32 kW. The klystron tube weight is 25 kg and the
solenoid magnet weight is 95 kg, which is said to be 2 to 3 times less than single-beam
klystrons of similar performance.
ISTOK has also applied the multiple-beam technology to the traveling wave tube and
the inductive output tube.

).7 OTHER ASPECTS OF RADAR TRANSMITTERS

Pulse Modulators47,48 The function of the modulator is to turn the transmitter on and off
to generate the desired waveform. When the waveform is a pulse, the modulator is some¬
times called a pulser. Each RF power source has its own particular characteristics that de¬
termine the type of modulator to be used. The magnetron modulator, for instance, has to
handle the full pulse power. The transistor amplifier, on the other hand, requires no mod¬
ulator at all since the transistor is turned on and off by the presence or absence of the in¬
put pulse. The full power of the klystron and the traveling-wave tube can be switched by
a modulator handling only a small fraction of the total electron-beam power if the tubes
are designed with a modulating anode or a shadow grid. The crossed-field amplifier (CFA)
is often cathode-pulsed, requiring a full-power modulator. Some CFAs, however, can be
d-c operated, which means they can be turned on by the start of the RF input pulse and
turned off by a short, low-energy pulse applied to a cut-off electrode. Similar to the tran¬
sistor amplifier, some lower power CFAs require no modulator since they can be turned
on and off by the start and stop of the input RF pulse.
The basic elements of one type of radar modulator is shown in Fig. 10.8. Energy from
an external power source is accumulated in the energy-storage element at a slow rate dur¬
ing the interpulse time. The charging impedance limits the rate at which energy is deliv¬
ered to the storage element. When the pulse is ready to be formed, the switch is closed
and the stored energy is quickly discharged through the load to form the d-c pulse that is
applied to the RF power device. During the discharge part of the cycle, the charging im¬
pedance prevents energy from the storage element from being returned to the energy
source.

Line-Type Modulator In this device a delay line, or pulse forming network (PFN), is
used as the storage element. The switch can be a hydrogen thyratron, mercury ignitron, a
silicon controlled rectifier (SCR), saturable reactor, or other device that can initiate the
discharge of the PFN to form a rectangular pulse. The shape and duration of the pulse are
 10.7 Other Aspects of Radar Transmitters 721

Figure 10.8 Basic elements of one type of

radar pulse modulator.

determined by the passive elements of the PFN. The switch has no control over the pulse
shape other than to initiate it. The pulse ends when the PFN has discharged sufficiently
to allow the switch to close and recover its voltage hold-off capability. The trailing edge
of the pulse is usually not sharp since it depends on the discharge characteristics of the
PFN. The line-type modulator is simple, compact in size, and can tolerate abnormal load
conditions. It has been widely used in the past for magnetron pulsing.

Active-Switch Modulator The switching element in this type of modulator has to be

able to be turned off as well as on. Originally the switch was a vacuum tube and the mod¬
ulator was called a hard-tube modulator to distinguish it from the gas-tube switch often
used in a line-type modulator. In addition to the vacuum tube, the switch can be a semi¬
conductor device such as a silicon controlled rectifier.
There are three types of active-switch modulators: (1) cathode pulsers that control the
full beam power of the RF tube, (2) mod-anode pulsers that are required to switch at the
full beam voltage of the RF tube but with little current, and (3) grid pulsers that operate
at a far smaller voltage than that of the beam.
Unlike the line-type modulator, the switch in the active-switch modulator controls
both the beginning and the end of the pulse. The energy storage element is a capacitor.
To prevent droop in the pulse shape due to the exponential nature of a capacitor discharge,
only a small fraction of the stored energy is extracted from the capacitor for the pulse to
be delivered to the tube. In high-power transmitters with long pulses the capacitance must
be very large. A large capacitance might be obtained with a collection of capacitors known
as a capacitor bank.
The active-switch modulator permits more flexibility and precision than the line-type
modulator. It can provide excellent pulse shape, varying pulse durations and prfs, includ¬
ing mixed pulse lengths and bursts of pulses with close spacings. It is, however, of greater
complexity and weight than a line-type modulator.

Crowbar Protective Device Power tubes can develop internal arc discharges with little
warning. When an arc occurs in an unprotected device, the capacitor bank discharges large
currents through the arc and the tube can be damaged. The tube can be protected with a
device called an electronic crowbar, which places a short circuit across the capacitor bank
722 chapter 10 ® Radar Transmitters

to transfer its stored energy. When a sudden surge of current due to a fault in a protected
power tube is sensed, the crowbar switch is activated within a few microseconds. The cur¬
rent surge also causes the circuit breaker to open and deenergize the primary source of
power. The name “crowbar” is derived from the analogous action of placing a heavy con¬
ductor, like a crowbar, directly across the capacitor bank. A crowbar is required for a high-
power active-switch modulator because of the large amount of energy that is stored in its
capacitor bank. Crowbars are not usually needed with line-type modulators which store
less energy in their pulse-forming network. Line-type modulators are designed to dis¬
charge safely all the stored energy each time a pulse is generated.

Transmitter Noise and Spectrum49 An RF power source can produce spurious, unwanted
outputs as harmonics of the fundamental frequency, adjacent-band (out-of-band) noise,
and in-band noise. Harmonics and adjacent-band noise can be reduced 30 to 60 dB by us¬
ing high-power filters. Shaping of the pulse to make it more rounded and less rectangu¬
lar reduces out-of-band signal energy. In-band spurious signals and noise cannot be read¬
ily filtered since these unwanted signals are within the frequency range of the desired
signal spectrum. The in-band noise is greater in some RF power sources than in others.
For example, Weil states49 that the noise level in a 1-MHz bandwidth of a conventional
CFA is typically 50 to 60 dB down, but is down 90 dB or better in linear-beam tubes.
Section 3.7 discussed the effects of equipment instabilities on the amount of clutter
cancellation, or improvement factor, that can be achieved in MTI radars. The typical noise
level in conventional CFAs can set a limit on the achievable MIT improvement factor to
perhaps the vicinity of 45 dB or so. Linear beam tubes, on the other hand, are capable of
very high MTI improvement factors except for limitations introduced by their modulators
and power supplies. The ripple on the modulator voltage and the variation of the high-
voltage power supply (HVPS) must be sufficiently small to obtain the large improvement
factors needed in high-performance radars.
In a staggered prf MTI system (Sec. 3.3) the variation of the interpulse period causes
a variation in the HVPS voltage, which can be a significant source of transmitter insta¬
bility. The consequent reduction in improvement factor that would result needs to be cor¬
rected, as indicated by Weil.49
The transmitter and its modulator can also distort pulse compression waveforms and
introduce spurious time-sidelobes. Active-switch, modulators are more likely to allow low
pulse-compression time-sidelobe levels as compared with the time sidelobes produced by
line-type modulators.

REFERENCES
1. Weil, T. A. “Transmitters.” In Radar Handbook, 2nd ed. M. Skolnik (Ed.). New York:
McGraw-Hill, 1990, Chap 4.
2. Gilmour, A. S., Jr. Microwave Tubes. Norwood, MA: Artech House, 1986, Chap. 16.
3. Granatstein, V. L., R. K. Parker, and C. M. Armstrong. “Vacuum Electronics at the
Dawn of the Twenty-First Century.” Proc. IEEE 87 (May 1999), pp. 702-718.
 References 723

4. News article from Microwaves & RF, (November 1984), p. 31.

5. Smith, M. J., and G. Phillips. Power Klystrons Today. New York: John Wiley, 1995,
Sec. 7.2.3.
6. Staprans, A. “Linear Beam Tubes.” In Radar Technology, E. Brookner (Ed.). Boston:
Artech House, 1977, Chap. 22.
7. Gilmour, A. S., Jr. Ref. 2, Chap. 4.
8. Phillips, R. M., and D. W. Sprehn. “High-Power Kystrons for the Next Linear Col¬
lider.” Proc. IEEE 87 (May 1999), pp. 738-751.
9. Staprans, A., E. W. McCune, and J. A. Ruetz. “High-Power Linear-Beam Tubes.” Proc.
IEEE 61 (March 1973), pp. 299-330.
10. Dodds, W. J., T. Moreno, and W. J. McBride, Jr. “Methods for Increasing the Band¬
width of High Power Microwave Amplifiers.” IRE WESCON Conv. Rec. 1, pt. 3 (1957),
pp. 101-110.
11. Gilmour, A. S., Jr. Principles of Traveling Wave Tubes. Boston: Artech House, 1994,
Sec. 18.4.
12. Symons. R. S. “Tubes: Still Vital After All These Years.” IEEE Spectrum 35 (April
1998), pp. 52-63.
13. Gilmour, A. S., Jr. Ref. 11, Chap. 13.
14. Gilmour, A. S., Jr. Ref. 2, Chap. 11.
15. Luebke, W. and G. Caryotakis. “Development of a One Megawatt CW Klystron.” Mi¬
crowave J. 9, no. 8 (August 1966), pp. 43-47.
16. Symons, R. S., and J. R. M. Vaughan. “The Linear Theory of the Clustered-Cavity™
Klystron.” IEEE Trans. PS-22 (October 1994), pp. 713-718.
17. Borkowski, M. T. “Solid-State Transmitters.” Radar Handbook, 2nd ed., M. Skolnik
(Ed.). New York: McGraw-Hill, 1990, Chap. 5.
18. Lee, K., C. Corson, and G. Mols. “A 250 kW Solid State AN/SPS-40 Radar Trans¬
mitter.” Microwave J. 26 (July 1983), pp. 93-105.
19. Dyck, J. D., and H. R. Ward. “RAMP’s New Primary Surveillance Radar.” Microwave
J. 27 (December 1984), pp. 105-113.
20. Merrill, P. R. “A 20 kW Solid-State L-Band Transmitter for the RAMP PSR Radar.”
Microwave J. 31 (March 1988), pp. 165-173.
21. Chilton, R. H. “MMIC T/R Modules and Applications.” Microwave J. 30 (September
1987), pp. 131-146.
22. Hoft, D. J. “Solid State Transmit/Receive Module for the Pave Paws Phased Array
Radar.” Microwave J. 21 (October 1978), pp. 33-35.
23. Olson, F. A. “Microwave Solid-State Power Amplifier Performance: Present and Fu¬
ture.” Microwave J. 38 (February 1995), pp. 24-46.
24. Trew, R. J., J-B Yan, and P. M. Mock. “The Potential of Diamond and SiC Electronic
Devices for Microwave and Millimeter-Wave Power Applications.” Proc. IEEE 79
(May 1991), pp. 598-620.
724 CHAPTER io • Radar Transmitters

25. Butler, N. “The Microwave Tube Reliability Problem.” Microwave J. 16 (March 1973),
pp. 41-42.

26. Advertisement of the Electron Device Group, Varian Beverly Division, Beverly, MA.
(Varian is now known as CPI, Inc.)
27. Weil, T. A. Ref. 1, Sec. 4.2.

28. Gilmour, A. S„ Jr. Ref. 2, Sec. 13.3.

29. Weil, T. A. “Comparison of CFA’s for Pulsed-Radar Transmitters.” Microwave J. 16

(June 1973), pp. 51-54, 72.

30. Kaisel, S. F. “Microwave Tube Technology Review.” Microwave J. 20 (July 1977), pp

23-42.

31. Anonymous. “Cathode-Driven Crossed-Field Amplifer.” Microwave J. 31 (February

1988), pp. 208-209.

32. Sivan, L. Microwave Tube Transmitters. London: Chapman & Hall, 1994, Sec. 7.4.
33. Abrams, R. H., Jr. “The Microwave Power Module: A ‘Supercomponent’ for Radar
Transmitters.” Record of the 1994 IEEE National Radar Conf, Atlanta, GA, pp. 1-6,
IEEE No. 94CH3359-7.

34. Smith, C. R., C. M. Armstrong, and J. Duthie. “The Microwave Power Module: A
Versatile RF Building Block for High-Power Transmitters.” Proc. IEEE 87 (May 1999),
pp. 717-737.

35. Yingst, T. E„ et al. “High-Power Gridded Tubes—1972.” Proc. IEEE 61 (March

1973), pp. 357-381.

36. Reed, J. E. “The AN/FPS-85 Radar System.” Proc. IEEE 57 (March 1969), pp. 324-
335.

37. Haeff, A. V. “An Ultra-High-Frequency Power Amplifier of Novel Design.” Electron¬

ics 10 (February 1939), pp. 30-32.

38. Preist, D. H., and M. B. Shrader. “The Klystrode—An Unusual Transmitting Tube
with Potential for UHF-TV.” Proc. IEEE 70 (November 1982), pp. 1318-1325.
39. Symons, R. S. “The Constant Efficiency Amplifier.” NAB Broadcast Engr. Conf. Proc.
(1997), pp. 523-530.

40. Symons, R. S., et al. “The Constant Efficiency Amplifier—A Progress Report.” NAB
Broadcast Engr. Conf. Proc., 1998.

41. Granatstein, V. L., and I. Alexoff. High-Power Microwave Sources. Boston: Artech
House, 1987.

42. Gilmour, A. S., Jr. Ref. 2, Chap. 14.

43. Felch, K. L., et al. “Characteristics and Applications of Fast-Wave Gyrodevices.” Proc.
IEEE 87 (May 1999), pp. 752-781.

44. Manheimer, W. M. “On the Possibility of High Power Gyrotrons for Super Range Res¬
olution Radar and Atmospheric Sensing.” hit. J. Electronics 72, nos. 5 and 6 (1992),
pp. 1165-1189.
 Problems 725

45. Blank, M., B. G. Danly, and B. Levush. “Circuit Design of a Wideband W-Band Gy-
roklystron Amplifier for Radar Applications.” IEEE Trans. PS-26 (June 1998), pp.
426-432.

46. Gelvich, E. A., et al. “The New Generation of High-Power Multiple-Beam Klystrons.”
IEEE Trans. MTT-41 (January 1993), pp. 15-19. See also the ISTOK Web Site at
www.istok.com.
47. Weil, T. A. Ref. 1, Sec. 4.8.
48. Ewell, G. W. Radar Transmitters. New York: McGraw-Hill, 1981.
49. Weil, T. A. Ref. 1, Secs. 4.6 and 4.7.

PROBLEMS

10.1 One way to define the efficiency of a transmitter is RF power out, Pmn, divided by the
prime power in, Pin. (a) Plot the power dissipated, Pdis = Pm - Pout, as a function of the
transmitter efficiency, e, for a fixed power out. [Make the ordinate the ratio (power dissi-
pated)/(power out).] (b) If the output power has to be 30 kW, what power will be dissi¬
pated with a transmitter having a 15 percent efficiency? (c) If the transmitter efficiency
can be increased to 50 percent, what is the amount of power to be dissipated? (d) What
disadvantages occur with low efficiency?

10.2 (a) In a solid-state transmitter (a solid-state “bottle”) with 300 modules, what would be
the reduction in output power if 20 modules were to fail? (b) What fractional reduction
in radar range would this cause?

10.3 (a) If 10 percent of the modules in an active-aperture phased array radar fail, what would
be the reduction in transmitter power? (b) What would be the reduction in the maximum
radiation power density? (c) What would be the reduction in radar range?

10.4 For an air-traffic control radar application, compare the advantages and disadvantages of
a solid-state transmitter, klystron transmitter, and magnetron transmitter.

10.5 (a) If one wanted a radar transmitter with a 10 percent bandwidth, what options are avail¬
able to the radar system designer and which RF power source appears the most desirable?
(You may have to make some assumptions about the application.) (b) If one wanted a
radar transmitter with 40 percent bandwidth, what options are available and which of these
options might you choose? (Include the reason for your choice.) (c) Are there any unde¬
sirable consequences with your choice for (b) that you might have to live with?

10.6 What are the advantages and disadvantages of the gyroklystron (amplifier) for radar ap¬
plication at 94 GHz (millimeter waves)?

10.7 How can a tube designer achieve a large bandwidth with a klystron type of power tube?

10.8 When might the systems engineer choose to use a traveling wave tube over a klystron for
a high-power radar application?
26 chapter 10 • Radar Transmitters

10.9 For a high-power UHF radar transmitter application, compare the advantages and disad¬
vantages of solid state, the grid-control vacuum tube, the constant-efficiency amplifier,
and the klystron.
10.10 When might a magnetron be desirable for use in radar applications?
10.11 What factors might be involved when a radar systems engineer tries to choose among the
crossed-field amplifier, TWT, and Twystron for some radar application?
10.12 If the R&D genie were to grant a radar systems engineer many wishes, what improve¬
ments might the radar systems engineer want to have for a radar transmitter?