you have been in the RF and microwaves business for any length of
time, you are probably familiar with a company named Varian. In
the days before you did your parts shopping online, Varian catalogs
populated the desks and bookshelves of many RF engineers who worked
in the radar field, including mine. Did you know that it is named
after the brothers Russell and Sigurd Varian, who started the business
in 1948 to market their high power klystron tubes? Following a number
of reorganizations, it was purchased by Agilent technologies in
2010. This story from Radio Electronics magazine does a
real nice job explaining the workings of a klystron without getting
too deep into the gory theoretical detail.
Thanks to Terry W. for providing this article.
See all available
vintage Radio-Electronics articles.
Klystron - tube
for outer space
By Tom Jaski
When we get to talking to those intelligent beings "out there" on
other planets or even in other solar systems, very likely klystrons
will be the transmitter tubes that will make our communications
possible. Large-power klystrons have been used as amplifiers in
the equipment that bounced radar signals off the moon, Venus and
satellites in orbit. Klystrons have been used to "interrogate" satellites,
and to trigger into action the electronic and mechanical equipment
only for outer space, but for its usefulness wherever microwaves
must be generated, the importance of this tube grows with the
industry's use of higher and higher frequencies
Less romantic, but even more practical, are other
applications for klystrons. Large-power klystrons are used widely
in Europe for UHF television transmitters. Here UHF television has
not become common enough to need many power klystrons. Klystrons
are also the heart of the new "scatter" communications systems in
which the line-of-sight rule about microwave transmission is violated
simply by using very high-power transmitters, large antennas and
Another major use of klystrons
is in experiments with food sterilization. These use high-speed
electrons issuing from linear electron accelerators, and these in
turn are powered by large klystron tubes.
In linear accelerators,
the klystrons provide a mighty push to the electrons passing through
successive drift tubes, eventually speeding them up to almost the
speed of light.
What then are these klystrons, what do they
look like and how do they operate?
Klystrons were invented
just before the start of World War II by the Varian brothers, then
graduate students at Stanford University. I remember their little
shack behind the Ryan laboratory in the hills behind the university,
and the excited talk of a resident in the area who had seen the
barbed-wire fence around this little shack develop a mysterious
red-hot glowing section of wire, True or not, the klystron has played
an enormously important role in the development of radar and microwave
communications, and is now on the verge of taking over industrial
jobs from other tubes.
To start the explanation of klystrons, let us first
look into another item, resonant cavities. Understanding cavities
is essential to understanding klystrons. All RF oscillating circuits
contain resonant elements (Fig. 1-a). As frequency increases, we
must decrease the inductance and capacitance of the resonant circuits.
We decrease the inductance by decreasing the turns until we have
nothing left but a straight wire or even a flat strip of metal.
The capacitance is reduced by lowering the number of plates in our
capacitor and finally by further separating the plates (Fig, 1-b).
Eventually we get to paralleling inductances (Fig. 1-c) since paralleling
two inductors halves their inductance, and the entire process winds
up as in Fig. 1-d or 1-e. The final product is a box or cavity,
the top and bottom representing the capacitor plates and the sides
the paralleled inductors.
Cavities follow certain hard and fast rules, which can be determined
easily from common-sense observation. For example, regarding the
top and bottom plates of the cavity as plates of a capacitor, we
see that they are virtually short-circuited at the edges. This means
that at the edges of the plates we cannot have a charge, and therefore
no field. From this follows our first rule about cavities: the electric
field parallel to a wall must be zero at that wall. Now to maintain
any charge which has a field in the center of the plate and none
at the edges, the voltage distribution must look something like
a sine-wave half-cycle from wall to wall. In fact, this is the simplest
way we can maintain a field in a cavity, the simplest "mode" in
which we can operate it. It follows that the width of the cavity
should be just about a half-wavelength of the microwave energy,
or any multiple of that. And the same goes for the length, if the
cavity is rectangular.
Fig. 1 - Evolution of a klystron cavity: a - lumped tuned circuit;
b - same, highest possible frequency ; - turns paralleled to
decrease inductance; d, e - rectangular and cylindrical resonant
cavities; f - klystron cavity. The last three are all derived
Fig. 2 - Typical klystron, cutaway view.
Fig. 3 - A 10-kw multi-cavity klystron.
Fig. 4 - Cross-section, reflex klystron.
Fig. 5 - Three old-time klystrons, the 417A, 707B 2K25. The
2K25 is still used to generate 3-centimeter waves.
The magnetic field always associated
with an electric field, and always at right angles to it, will then
be parallel to the top and bottom of the cavity. Thus it would cut
the end plates. But since it is a changing magnetic field, it will
induce a current in any conductor within the field, and the end
plates have currents induced in them which set up counter-magnetic
fields equal to and thus cancelling the first fields.
we have the second rule about cavities: the magnetic field must
be zero at any wall which it cuts at right angles. Thus the magnetic
field is confined to the box as well. But with the magnetic field
we do not have the same dimensional problem, for we can swap density
for space. Therefore, the top-to-bottom dimension of the cavity
is not as critical, but does determine the capacity of the cavity
to maintain a certain field amplitude. For just as a capacitor dielectric
would break down if it were too thin for the voltage on the plates,
so a cavity can break down, dielectrically speaking, when the voltage
gets too high between top and bottom plates. Because we design the
cavity carefully as far as dimensions are concerned, we can then
set up standing waves in it, and the cavity can easily be excited
with small charges on the top and bottom plates.
If we make
the cavity an integral part of a vacuum tube, and make part of the
top and bottom into a grid area (punch holes in it or slot it),
this does not drastically change the properties of the cavity. It
can still be excited easily by charge differences between top and
bottom plate. The klystron incorporates one or more of these cavities
with grids in top and bottom. Fig. 2 is a cutaway representation
of a typical two-cavity klystron.
The Bunching Action
At the bottom of the tube we have an electron gun that produces
a narrow beam of electrons. This beam leaves the gun under the influence
of the accelerating grid, which you can see just below the first
cavity. Then the electrons travel on through the two cavities, and
the space between them - called the drift space - to the collector,
which can collect the electrons because of a positive charge on
it. As the electrons travel through the first cavity grids, they
constitute a current through these grids, from one grid to the next
- after all, a current is nothing more than a flow of electrons.
But, since this is a steady flow of electrons, the best that we
could expect would be a steady potential difference on the grids.
If we manage to excite the cavity between the grids in some
way creating an alternating potential between these grids, we will
affect the electrons between them. An electron traveling toward
a grid that is positive will be attracted and speed up and one traveling
toward a negative grid will slow down. If the bottom grid of the
lower cavity is momentarily negative, and the top grid positive,
the electrons approaching the bottom grid from the cathode will
be retarded, while those between the grids approaching the top grid
of the first cavity will be accelerated.
In the next half-cycle
of applied RF, the lower grid will be positive and the top one negative.
Thus electrons which then approach the lower grid will be accelerated,
and the electrons which are then between the two grids will be retarded.
In this way, the grids and cavity with applied RF will form bunches
of electrons, some of which move faster than when they left the
cathode and some of which move a bit slower.
When the RF
applied to the cavity goes through zero, the electrons then passing
through the grids will not. be affected, and will just travel on
at the same velocity. The lower cavity and grid assembly, forming
the bunches, is appropriately called the buncher. (The Varians named
this a rhumbatron.) In the space between the cavities, the drift
space, the electrons that are moving at the original "from-the-cathode"
velocity will join some of those which were slowed down. They in
turn will be joined by some of those that speeded up. Thus the bunches
of electrons in the drift space become denser, and the space between
bunches has fewer and fewer electrons.
Were we to let the
bunches drift too long, the repulsion between electrons would again
scatter them. But we don't give them time to do that. The denser
bunches, now with more electrons, pass through the second set of
grids. Through these grids then pass alternately dense bunches of
electrons and spaces .with none or just a few. This is, in effect,
a pulsed dc. Pulsed dc can look very much like ac if we shift the
base line (different zero level).
The bunches then constitute
a periodically changing current capable of inducing an RF voltage
in the second cavity. Note that the acceleration and deceleration
of electrons between the buncher grids lasted nearly a half-cycle.
The bunches which reach the "catcher" grid are also about a half-cycle
long. They will induce in the catcher cavity an RF of the same frequency
as was applied to the buncher.
Getting Power from
To induce a field in the second cavity,
the electrons must give up energy. It is easy to see how this happens
after the field has built up. Electrons approaching a negative grid
are retarded and impart energy to the grid. Electrons leaving a
positive grid are also retarded, giving off energy. Thus if we time
the bunches (by regulating the initial velocity of the electrons)
to be between the catcher grids only when the first catcher grid
is positive and the second catcher grid is negative, while we make
sure that we have virtually no electrons between the grids when
this situation is reversed, then we draw the maximum energy from
our bunches of electrons. This is the way a klystron is operated.
The collector and accelerator voltages must be precisely adjusted
to get this kind of timing.
If we feed back a portion of
the catcher energy to the buncher, the tube will oscillate. If our
timing is correct, the phase of the RF will of course be exactly
right for the feedback situation, for the bunching occurs when the
second buncher grid is negative, and we get the most energy when
the second catcher grid is also negative. Amplification is obtained,
because the bunches going through the catcher contain many more
electrons, thanks to the time spent in the drift space, than the
bunches coming out of the buncher.
The energy is coupled
into the buncher and out of the catcher cavities with a small loop,
which will contain some of the magnetic lines of force of the fields
and will thus have a current induced in them.
We can of
course use the energy in one of the catcher cavities to excite additional
cavities and grids, and this we do many times to increase the energy
produced by large klystrons. Fig. 3 shows such a large multicavity
klystron made by Eimac, capable of producing 10,000 watts output
in the 720-985 mc range.
The Reflex Principle
there are also klystrons with but one cavity, The principle is illustrated
in Fig. 4. These we call reflex klystrons because the collector
at the end of the tube is given a negative voltage, thus repelling
the electrons. This electrode is usually called a repeller, What
happens here is that the electrons, after being bunched in the grids,
travel on into the drift space above the cavity for a time, then
are repelled back toward the grids. If we repel them with exactly
the right velocity to make them arrive at the grids when the voltages
on these grids are of the correct phase to obtain energy from the
electron bunches, the original field is augmented, and we have oscillation.
So the reflex klystron is used primarily as an oscillator.
Reflex klystrons come in many shapes. Fig. 5 shows three of
World War II vintage, the 417A made by Westinghouse for the S-band
(10 cm), the 707B with an external cavity, also for the same frequency
range, and the 2K25 used most often as the' local oscillator in
3-cm (10,000-mc) radar receivers.
All three are tunable
to a certain extent (Fig. 6), The 417A is tuned by changing the
cavity dimensions with
a tuning lever and screws, the 707B by
modifying the electric fields in the cavity with slugs projecting
into it, and the 25K5 by changing the cavity dimensions with the
tuning "bow". The tuning bow is flexed by the screw. This alters
the position of the more or less flexible top portion of the metal
enclosure, and the top cavity grid with it.
A more modern
version of the reflex klystron, using ceramic insulation, is shown
in the head photo. Such ceramic klystrons are now produced and regularly
oscillate at 25 kmc, while some laboratory models have been used
to generate frequencies as high as 100 kmc. The latter are not in
production, but are strictly experimental tubes.
Fig. 6 - Three klystron tuning methods.
Klystrons can be
modulated in various ways. One is to vary somewhat the reflector
voltage or, in the power klystron, the collector voltage. This has
the effect of changing the velocity of the electrons, and thus the
frequency of oscillation in the klystron is affected. This kind
of modulation is limited within very narrow ranges. Klystrons specially
built with a modulating anode near the electron gun can be amplitude-modulated
by the simple mechanism of making the electron beam vary in density.
Since the amplification of the tube depends on increasing the density
of the electron bunches in the drift space, the effect of the bunching
will be more pronounced when a lot of electrons are available than
when only a few are traveling through the cavity grids. These anode-modulated
klystrons are so constructed that the total voltage between the
cathode and the tube structure (including the cavities) remains
the same. Thus the velocity of the electrons is constant, but the
voltage between the modulating anode and the cathode can vary and
the quantity of electrons with it.
Very often, particularly
in television transmitters, it is actually unnecessary to modulate
the klystron. Here it acts as a power amplifier, and the modulation
can be introduced at an earlier stage. Thus the klystron amplifies
the already modulated signal.
The klystron can be pulse-modulated
by the anode in the types which have this separately insulated anode,
and by turning the collector voltage on and off in the types that
Except when we want to modulate the klystron, the
voltages supplied to the elements must be very stable. Usually they
are supplied from well regulated power supplies. The reasons are
fairly obvious. If the dc voltages on the cavities and collector
or reflector varies, the velocity of the electrons also varies.
And, since the speed with which the electrons travel through the
buncher determines the frequency of the generated rf, this too would
In the reflex klystron the situation is even more
critical. The path the electrons travel must be exactly the right
length to allow the electrons on their return voyage to reinforce
the original bunching action. If the path should be altered, by
a varying voltage, the electrons would arrive at the wrong time
and might partly cancel the bunching. The oscillation would then
soon die out.
As a matter of fact, this device is used to
allow the reflex klystron to operate in different "modes". The path
of the electrons, for oscillation, must always be a multiple of
a quarter-wave-length. But whether the tube has a path of 3 3/4.
or 4 1/4 wavelengths for the electrons, the action is the same.
However, with the longer path, caused by a lower (less negative)
reflector voltage, the density of the beam is somewhat affected,
and the klystron produces less power. By selecting one or the other
modes the klystron can be made to put out at different levels of
power. The 25K5 for example can operate in about five modes, all
producing the same frequency, but with different power levels.
As UHF television becomes more popular, the klystron will
be used increasingly for high-power amplification in the transmitters.
Further increases in UHF scatter communication and in microwave
applications as we progress in the space age is also to be expected.
The klystron, which has proven its mettle in bouncing signals off
our neighboring planets, will most certainly be the power amplifier
for space telephony, once man takes the big jump and starts traveling
between planets in the solar system and to distant stars. It is
a special vacuum tube to be reckoned with for the next few centuries
of man's technological development.