Wax nostalgic about and learn from the history of early electronics.
See articles from Radio &
Television News, published 1919 - 1959. All copyrights hereby acknowledged.
The microwave klystron was invented in 1937 by brothers
Russell and Sigurd Varian. If you have been in the microwave
design business for a couple decades, you undoubtedly recognize
the company name of
Varian Associates, especially if you worked in the aerospace
or defense electronics business. There is a
video on YouTube of a segment on
Varian done sometime around 1990 by Walter Cronkite. There is
also a historical piece on
Varian Associates on the Communications & Power Industries
website. This circa 1952 article covers the fundamentals of
klystron operation and reports on the increasing use of klystrons
in high frequency application - including by amateur radio operators
exploring the top end of the bands.
RF Cafe visitor Joe Molon (KA1PPV)
sent some photos of a klystron he worked with at television
WEDW in Bridgeport, Connecticut. See those pics and his
Microwave Klystron Oscillators
By Joseph Racker* and Lawrence Perenic†
Part 1. Practical operating data on klystrons as used in
present-day communications equipment.
Microwave frequencies are being used to an ever-increasing
degree in commercial television and communications links, amateur
transmission, and a long list of governmental and industrial
electronic devices such as radar, instrument landing, guided
missiles, and air traffic control. Consequently, the field of
microwave techniques offers an excellent opportunity to technicians
and engineers, as well as providing interesting equipment for
Fig. 5 is a block diagram of a typical transmitter and receiver
operating at microwave frequencies. As seen in this diagram,
all stages except the microwave oscillator in the transmitter
and input circuit and local oscillator in the receiver, operate
at conventional frequencies. This illustrates a very important
fact, namely, a thorough understanding of microwave oscillators
and their associated circuits provides the reader with a very
substantial basis for servicing and building many microwave
systems. In other words, generally speaking, a microwave transmitter
is no more than a conventional type transmitter which uses a
microwave oscillator. A microwave receiver is a standard superheterodyne
circuit using a microwave local oscillator and input circuit.
Fig. 5 - Block diagram of a conventional
microwave transmitter and receiver.
There are a number of methods of generating energy at microwave
frequencies including lighthouse oscillators, magnetrons, traveling
wave tubes, and klystrons. The klystron oscillator, which is
the subject matter of this article, is by far the most commonly
used oscillator, particularly in commercial equipment.
Sylvania's reflex klystron is designed to operate at wavelengths
be tween 6-7 cm.
For reasons that need not be discussed in this article, it
is extremely difficult and inefficient to generate microwave
energy using conventional techniques and tubes. Of the microwave
oscillators listed, only the lighthouse tube uses standard oscillator
circuits. This type of oscillator, however, has an upper frequency
of about 4000 mc. All of the other oscillators, sometimes called
transit time oscillators, employ new and different techniques
Since entirely new techniques are involved, the authors have
divided this article into two parts. The first part describes
the basic theory of klystron operation so that a reader who
has never had any previous microwave experience can understand
how it works. The second part covers the practical aspects of
the klystron oscillator, namely, its construction, characteristic
curves, methods of tuning, modulation, and servicing. Needless
to say, the reader cannot absorb the material presented in the
second part unless he understands the theory outlined in the
Fig. 1 - Electron flow in a diode tube.
Conservation of Energy Law
A very familiar universal law - conservation of energy -
which is rarely used in electronic theory becomes very important
in klystron and other transit time oscillator operation. They
are called transit time oscillators because the energy required
to sustain oscillations is obtained from the electron stream
while it is in transit between cathode, or electron gun, to
plate, or collector. In order to understand how this energy
is transferred, the conservation of energy law must be applied
to the electron flow.
Consider the flow of an electron that leaves the cathode
of the diode, shown in Fig. 1, and travels toward the plate.
When the plate is positive with respect to the cathode, the
electron is accelerated toward the plate. Whenever any mass,
including an electron, is accelerated it picks up energy. From
the law of conservation of energy it is known that this energy
must come from some other element in the system.
Actually, in this case, the energy comes from the battery. This
is most easily seen at the instant that the electron actually
reaches the plate. At this time the electron would normally
neutralize a positive charge and the plate potential would decrease.
However, the battery has been expending energy and drawing an
electron away from the plate, as the transit electron approaches
it, so that when the electron actually reaches the plate there
is no net change in charge and the plate potential remains constant.
The most important fact, however, is that when an electron is
accelerated, it gains energy, and this energy comes from some
element in the system.
Similarly, if the plate potential is negative, the electron
is retarded, or slowed down, by the plate. When an electron
is retarded, it is giving up energy. Thus when it approaches
the plate it repels an electron on the plate toward the battery.
This; in effect, means that energy is being returned to the
Now let us apply this law of conservation of energy to an
electron flowing in an electric field such as can exist in the
cavity described in a later paragraph. If the electron is accelerated
by the field it is gaining energy which must come from the field
and make it weaker. Hence an electron that is accelerated by
the field causes the field strength to decrease. Conversely,
an electron that is retarded by the field, loses energy to,
and increases the field strength. In this article, the direction
of the electric field will be depicted by an arrow and the magnitude
of the field by length of arrow. An electron flowing in the
direction of the field is accelerated by it (some texts use
the reverse convention), while an electron flowing against the
field is retarded by it.
Flow in Cavity Resonator
At microwave frequencies a special type of tuned circuit,
known as a cavity resonator, or simply a cavity, is required.
The cavity is a hollow metallic box which can be rectangular,
cylindrical, spherical, or a number of other shapes. Several
common types of cavities are shown in Fig. 2. The cavity is
usually about one-half wavelength long (wavelength as measured
in the cavity), although it can also be any multiple number
of half wavelengths long.
Fig. 2 - Several common types of cavities.
In any tuned circuit, such as the conventional LC tank at
conventional frequencies, there is a continuous exchange of
energy from magnetic to electric field, and vice versa. For
example, in the LC circuit the energy stored in the magnetic
field around the coil is transferred into the electric field,
building up in the condenser, as the current in the circuit
declines. During the next half cycle, as the condenser discharges,
the condenser electric field energy is transferred into the
magnetic field which is being generated around the coil.
In a cavity this continuous exchange of energy takes place
in the air within the cavity. In a one-half wavelength rectangular
cavity, for example, the electric field exists sinusoidally
along the length of the cavity as shown in Fig. 3. The field
is always maximum at the center and reduces to zero at the ends.
The magnetic field in the cavity also has a sinusoidal variation,
being maximum at the ends and becoming zero at the .center.
In Fig. 3, three instantaneous values of the electric field
distribution are shown. In Fig. 3A, the electric field is at
its maximum intensity. In Fig. 3B, which occurs a short time
later, the entire field is declining (and beginning to build
up the magnetic field). In Fig. 3C, the field has undergone
a half-cycle of operation (since 3A) and is at its maximum
Fig. 3 - Electric fields in rectangular guide.
Just as in any tank circuit, the intensity of the electric
field during each successive cycle would decrease slightly,
due to small energy loss - equivalent to resistive loss - unless
some method of replenishing this energy is available. This can
be done by use of an electron stream. Assume that, as shown
in. Fig. 4, a small slot is inserted at the center of a rectangular
cavity and an electron stream is directed through this slot.
When the electric field of the cavity is so directed (positive
half-cycle) that it accelerates the electrons, energy is transferred
from the field to the electrons and the field intensity diminishes
and oscillations are damped out. However, if the field is so
directed (negative half-cycle) that it retards the electron
flow, then energy is transferred from electrons to field, and
oscillations are sustained. This is the basic principle of klystron
Fig. 4 - Electron stream flowing through
the slot in a rectangular wave guide.
In the foregoing example, if the electron stream were passed
through the cavity with uniform intensity, it would increase
the field during one half-cycle and decrease it during the next
half-cycle and there would be no net exchange of energy. However,
if we could arrange the electron stream so that the density
of electron flow during the field negative half-cycle is much
greater than the density during the positive half-cycle, then
there would be a net exchange of energy from electron stream
to the field. The process by which the electron beam is "bunched"
in this manner is known as bunching or "velocity modulation."
The velocity modulation action is best understood by considering
the operation of the simplest type of velocity modulation oscillator
called the positive grid, Barkhausen-Kurz, or retarding field
oscillator. In this type of oscillator, shown in Fig. 6, the
grid is operated at a positive potential with respect to cathode
and plate, and the plate is negative with respect to the cathode.
Fig. 6 - The electron motion in a positive-grid
(Barkhausen-Kurz) type oscillator having constant electrode
The operation of this circuit will be considered first under
d.c. conditions and then with an a.c. voltage applied to the
grid. In addition, although the complete action involves many
electrons, it will be simpler to first investigate the behavior
of a single electron. Then, later; the reasoning thus obtained
will be extended to entire groups of electrons.
Consider the flow of an electron leaving the cathode of the
tube shown in Fig. 6 with the tank circuit shorted out (no a.c.
voltage on grid). As an electron leaves the cathode it is accelerated
toward the grid by its high positive potential. By the time
the electron reaches the grid its velocity is high and it may
either hit the grid, delivering its energy in the form of heat,
or - more likely - it will pass through the space between grid
wires into the region between grid and plate. In the grid-plate
region the electric field is in the opposite direction because
the plate is negative with respect to the grid. This field tends
to slow down the electron and for this reason the oscillator
is sometimes called the "retarding field" oscillator.
If the plate voltage is sufficiently negative, the electron
will come to rest at some point in space between the grid and
plate. The attraction of the grid then causes the electron to
reverse direction and move back toward the grid. The electron
then swings back and forth past the grid (path M, N, 0) until
it eventually hits one of the grid wires. The phenomenon is
very closely parallel to that of the oscillation of a damped
pendulum (damped because electron loses some energy during each
If no other elements were introduced in the circuit many
individual electron oscillations would occur in the space between
plate and cathode, causing equivalent energy oscillations in
the grid circuit. The exact phase and amplitude of these oscillations
between any two electrons would depend upon the time at which
the electrons were emitted and the space charge at that time.
It is obvious that under these conditions no useful oscillator
energy can be supplied to the grid circuit, since the electron
oscillations are random and cancel each other.
Now assume that an a.c. voltage is superimposed upon the
grid. (tank circuit no longer short circuited). The frequency
of this signal is so high that by the time the electron travels
from cathode to the grid, the voltage has changed one-half of
a cycle, for example from maximum positive to maximum negative,
or from zero to zero, and so on. This is shown in Fig. 8 where
(A) plots the a.c. voltage and (B) the oscillatory (no energy
loss) electron path. Let us define the velocity, vo,
as the velocity of the electron at any point in the cathode-plate
space with grid potential at the d.c. value shown in Fig. 6.
Consider the relative velocities of the electrons leaving
the cathode during grid a.c. potential of A, B, and C shown
in Fig. 8A. An electron leaving during time A, travels at a
velocity less than vo in the cathode-grid plane because
during this time the grid is always at a negative a.c. potential.
It loses more velocity between grid and plate since during this
time the grid is positive. Thus, its over-all velocity is less
Fig. 8 - Graph of (A) grid a.c. voltage and
(B) electron position in space leaving cathode when the time
base is equal to zero.
An electron leaving the cathode with grid potential at B
will travel at about vo, since during both its cathode-grid
and grid-plate paths the grid is positive half the time and
negative half the time. Finally, an electron emitted at time
C will travel at a velocity greater than vo, since
the grid is positive (a.c.) when it is in the grid-cathode plane
and negative in the grid-plate plane. It is readily seen that
the electrons emitted at time C tend to catch up with electrons
emitted at time A so that electrons will tend to oscillate in
bunches instead of completely at random. Now useful oscillatory
energy can be supplied to the grid circuit.
Basically, the klystron, as shown in Fig. 7, consists of
an electron gun, "buncher" and "catcher" grids, and a collector.
The electron gun is similar to those found in cathode-ray tubes
and functions to provide a steady stream of electrons. The buncher
and catcher grids are part of the cavity resonators, slotted
in the center using the same principles shown in Fig. 4. The
electron stream is velocity modulated by the buncher grids and
converted to microwave energy in the catcher grids. Electrons
passing through the catcher grids are removed from the circuit
by the collector. The entire assembly is enclosed in a vacuum
Fig. 7 - Basic representation of klystron.
The electron gun directs a constant intensity stream of electrons
through the buncher grids. The electric field, due to the cavity
action between these grids, is varying sinusoidally. Accordingly,
some electrons are accelerated, some maintain the same velocity,
and others are retarded as they pass through these grids.
The electrons emerge from the buncher having various velocities,
but the electron stream still has an essentially uniform density.
The electrons then flow through a field-free "drift space."
It is assumed, for simplicity, that in this region there are
no d.c. or r.f. fields, and that any space-charge effects are
In this drift space, the electrons that were speeded up by
the buncher begin to catch up with the slower moving electrons
ahead of them. In a similar manner, the electrons which were
slowed down by the buncher lag behind more and more until they
are overtaken by electrons that left the buncher at a later
time. This bunching process, similar to that occurring in the
positive grid oscillator, eventually results in the breaking
up" of the electron beam into groups, or bunches, of electrons.
These bunches of electrons are separated by regions in which
there are comparatively few electrons.
The electrode arrangement thus far described is useless in
the sense that no output signal has been obtained. In principle,
an ordinary plate might be installed at the end of the drift
space and be used to collect the signal from the electron beam.
The voltage of the plate would rise and fall as if was struck
by the bunches of electrons. Unfortunately this method of signal
collection is not practical because the frequencies at which
velocity modulated tubes operate are so high that stray capacity
of the external load circuit would short circuit this energy.
Therefore another cavity is used to absorb this energy. This
cavity, known as the catcher, is placed at the point in the
drift space where maximum bunching occurs. The field of this
cavity is so phased that it always is negative with respect
to the bunched electrons. Hence energy from the bunched electrons
is transferred to the catcher cavity and oscillations are sustained.
The field is positive across these catcher grids during the
time that there is a region of relatively few electrons and,
therefore, little loss of energy occurs during this part of
Proper. phasing between catcher and buncher grids is effected
by feeding back some of the catcher cavity energy into the buncher
cavity. The remainder of the oscillator energy is coupled through
a coaxial cable to the desired load.
After passing the catcher grids, the electrons are moving
at a greatly reduced velocity and are finally removed from the
tube by a positive collector plate. The collector plate potential
must be positive enough to attract all the electrons, but not
so positive that electrons will strike at a high velocity and
cause secondary emission. It is obvious that any random electron
flow detracts from the over-all efficiency and stability of
the system. Hence, the importance of effective removal of electrons
after they have passed catcher grid.
In the second article on this subject, some of the practical
aspects of klystron operation will be considered. (Concluded
RF Cafe visitor Joe Molon (KA1PPV) sent the following after
seeing this article (posted with permission).
WEDW CH 49 Transmitter Klystron
Enjoy your site and check in almost everyday. Good stuff
to know. I enjoyed the vintage piece on Varian.
I've been in TV broadcast for over 35 years, so I guess
that makes me vintage, and I certainly remember Varian.
We used them in high power
(30kw) UHF TV transmitters. We even ran two in parallel
to make it into a 60 kw transmitter back in 1981.
That's when pulsers first came out and for a short time
we had the most efficient UHF transmitter in the country.
We were getting around 66% with pulsers as opposed to 33%
I've attached two pics of the tube. Un-mounted and mounted
in the old GE transmitter. For reliability you can't beat
the new solid state transmitters though.
They were used at WEDW CH 49 Bridgeport CT in a GE 30 kW
analog TV transmitter. These were photographed probably
in 1982. The transmitter used one for visual and one for
aural service. Later we connected two visual tubes together
in parallel to form a 60 kW transmitter. We had a similar
transmitter at WEDN Norwich which used the same tube type.
RF Cafe began life in 1996 as "RF Tools" in an AOL screen name web space totaling
2 MB. Its primary purpose was to provide me with ready access to commonly needed formulas
and reference material while performing my work as an RF system and circuit design engineer.
The Internet was still largely an unknown entity at the time and not much was available
in the form of WYSIWYG
All trademarks, copyrights, patents, and other rights of ownership to images and text
used on the RF Cafe website are hereby acknowledged.