|Excerpt from Reflex Oscillators by J.R.
Pierce and W.G. Shepherd Bell System
Technical Journal, July 1947, p. 460.Reprinted with permission, Copyright
Fig. 58.-External view of the W.E. 707-A reflex oscillator tube. This tube
is intended for use with an external cavity and was the first of a series
low voltage oscillators.
Fig. 59.-X-ray view of the W.E. 707-A shows the method of applying an
external cavity tuned with a piston.
Fig. 61.-A narrow tuning range cavity for the W.E. 707-A of the type used
radar systems. The inductance of the cavity can be adjusted by moving
into it. This view also shows the adjustable coupling loop.
Fig. 60.-Sketch showing a piston tuned circuit for the W.E. 707-A which
permit operation from 1150 to 3750 mc.
Fig. 63.-A 3 centimeter reflex oscillator with an internal resonator. This
tube is a further development of the earliest internal resonator reflex
oscillator designed at the Bell Telephone Laboratories.
SMEC UPDATE: We have examples of the 707A, 707B, and 723A/B, as well
2K25 Klystrons at the museum. These items were saved for future
to view by K. D. Smith, an engineer at Bell Laboratories.
For more details on Klystrons, be sure to check out our set of BSTJ's.
A Reflex Oscillator With An External Resonator-The 707
The Western Electric 707A tube, which was the first reflex oscillator
extensively used in radar applications, is characteristic of reflex
oscillators using inductance tuning. It was intended specifically for
service in radar systems operating at frequencies in a range around 3000
megacycles. Fig.58 shows a photograph of the tube and Fig. 59 an x-ray
view showing the internal construction. A removable external cavity is
employed with the 707A as indicated by the sketch superimposed on the
x-ray of Fig.
59. Such cavities are tuned by variation of the size of the resonant
chamber. Such tuning can be considered to result from variation of the inductance
of the circuit.
The form of this oscillator is essentially that of the idealized
oscillator shown in Fig. 59. The electron gun is designed to produce a
cylindrical beam. The gun consists of a disc cathode, a beam forming
electrode and an accelerating electrode G1, which is a mesh grid formed on
radius. The gun design is based on the principle of maintaining boundary
conditions such that a rectilinear electron beam will flow through the
resonator gap. The resonator grids G2 and G3 are mounted on copper
One of these has a re-entrant shape to minimize stray capacitance in the
resonant circuit. These discs are sealed to glass tubing which provides a
vacuum envelope. The discs extend beyond the glass to permit attachment to
the external resonant chamber. The shape of the repeller is chosen to
provide as nearly as possible a uniform field in the region into which the
A wide variety of cavity resonators has been designed for use with this
oscillator. An oscillator of this construction is fundamentally capable of
oscillating over a much wider frequency range than tubes tunable by means
The advantage arises from the fact that the interaction gap where the
electron stream is modulated by the radio frequency field is fixed. As
discussed in more detail in Section X, this results in a slower variation
the modulation coefficient with frequency and also a slower variation of
cavity losses and gap impedance than in an oscillator in which tuning is
accomplished by changing the gap spacing. A cavity designed for wide range
frequency coverage using the 707A tube is shown in Fig. 60. Using such a
cavity it is possible to cover a frequency range from 1150 to 3750
megacycles. The inductance of the circuit is varied by moving the shorting
piston in the coaxial line. For narrow frequency ranges, cavities of the
type shown in Fig. 61 are more suitable. In such cavities tuning is
by means of plugs which screw into the cavity to change its effective
inductance. Power may be extracted from the cavity by means of an
coupling loop as shown in Fig. 61.
The 707A was the first reflex oscillator designed to operate at a low
voltage i.e. 300 volts. This low operating voltage proved to be a
considerable advantage in radar receivers because power supplies in this
voltage range provided for the i.f. amplifiers could be used for the
oscillator as well. Operation at this voltage was achieved by using an
interaction gap with fine grids, which limits the penetration of high
frequency fields. This results in a shorter effective transit angle across
the gap for a given gap spacing and a given gap voltage than for a gap
coarse or no grids. Hence, for a given gap spacing a good modulation
coefficient can be obtained at a lower voltage. Moreover, since drift
results in more efficient bunching at low voltages, a larger electronic
admittance is obtained than with an open gap. This gain in admittance more
than outweighs the greater capacitance of a gap with fine grids, so that a
larger electronic tuning range is obtained than with an open gap. The
successful low voltage operation of the 707A established a precedent which
was followed in all the succeeding reflex oscillators designed for radar
purposes at the Bell Telephone Laboratories. The 707A is required to
a minimum power output of 25 milliwatts and a half power electronic tuning
of 20 megacycles near 3700 megacycles. The power output and the electronic
tuning are in excess of this value over the range from 2500 megacycles to
3700 megacycles in a repeller mode having 3 3/4 cycles of drift.
A Reflex Oscillator With An Integral Cavity-The 723
The need for higher definition in radar systems constantly urges operation
at shorter wavelengths. Thus, while radar development proceeded at 3000
megacycles, a program of development in the neighborhood of 10,000
megacycles was undertaken. Although waveguide circuit techniques were
employed to some extent at 3000 megacycles, the cumbersome size of the
made its use impractical in the receiver and hence coaxial techniques were
employed. The 1" by 1/2" guide used at 10,000 megacycles is
receiver design and also desirable because the loss in coaxial conductors
becomes excessive at this frequency. Hence, one of the first requirements
an oscillator for frequencies in this range was the adaptability of the
output circuit to waveguide coupling.
In considering possible designs for a 10,000 megacycle oscillator the
scaling of the 707A was studied. This appeared impractical for a number of
reasons. The most important limitation was the constructional difficulty
maintaining the spacing in the gap with sufficient accuracy with the glass
scaling technique available. Also, variations in the capacitance caused by
variations in the thickness of the seals caused serious difficulties in
predetermining an external resonator. Contributing difficulties arose from
the power losses in the glass within the resonant circuit and the problem
making the copper to glass seals close to the internal elements.
Consideration of these factors led to a new approach to the problem, in
which the whole of the resonant circuit was enclosed within the vacuum
envelope. This required a different mechanism for tuning the resonator,
since variation of the inductance of a cavity requires relatively large
displacements which are difficult to achieve through vacuum seals. The
alternative is to vary the capacitance of the gap. Since the gap is small
relatively large change in capacitance can be achieved with a small
displacement. This sort of tuning permits the use of metal tube
constructional techniques, and these were applied.
As a matter of historical interest an attempt at this technique made at
Bell Telephone Laboratories is shown in Fig. 62. This device was
together by a sealing wax and string technique and was not tunable in the
first version. It oscillated successfully on the pumps, however, and a
second version was constructed which was tuned by means of an adjustable
coaxial line shunting the cavity resonator. Adjustment of this auxiliary
line gave a tuning range of 7.5%. Such a tuning method is fraught with the
complications outlined in Section IX.
An early reflex oscillator tube of the integral cavity type designed at
Bell Telephone Laboratories was the Western Electric 723A/B.
This design was superseded later by the W.E. 2K25 which has a greater
frequency range and a number of design refinements. From a constructional
point of view the two types are closely similar, however, and to avoid
duplication the later tube will be described to typify a construction
served as a basis for a whole series of oscillators in the range from 2500
to 10,000 Mc/s.
Before proceeding to a description of the 2K25 tube it seems desirable to
recapitulate in more detail the design objectives from a mechanical point
view. These were:
1. To provide a design which would lend itself to large scale production
one sufficiently rugged as to be capable of withstanding the rough use
inherent in military service.
2.. To provide output means which permit coupling to a wave guide in such
manner that installation or replacement could be accomplished in the
simplest possible manner.
3. To provide a tuning mechanism for the resonant circuit which, while
simple, would give sufficiently fine tuning to permit setting and holding
frequency within one or two parts in 10,000. In addition, in order to
field installation it was desired to have the tuning mechanism cheap
to be factory installed and discarded with each tube.
4. The oscillator was required to be compact and light in weight to
facilitate its use in airborne and pack systems.
Figure 63 (at the end of this section) shows a cross-section view of the
final design of the 2K25 reflex oscillator. The resonant cavity is formed
part by the volume included between the frames which support the cavity
grids and also by the volume between the flexible vacuum diaphragm and one
of the frames. This diaphragm also supports a vacuum housing containing
repeller. The electron optical system consists of a disc cathode, a beam
electrode and an accelerating grid. These are so designed as to produce a
slightly convergent outgoing electron stream. The purpose of this initial
convergence is to offset the divergence of the stream caused by space
after the stream passes the accelerating grid and to minimize the fraction
of the electron stream captured on the grid frame on the round trip. The
repeller is designed to provide as nearly as possible a uniform retarding
field through the stream cross-section.
Power is extracted from the resonant circuit by the coupling loop and is
carried by the coaxial line to the external circuit. The center conductor
the coaxial line external to the vacuum is supported by a polystyrene
insulator and extends beyond the outer conductor to form a probe. Coupling
to a wave guide is accomplished by projecting this probe through a hole in
that wall of the wave guide which is perpendicular to the E lines so that
the full length of the probe extends into the guide. The outer conductor
connected to the wave guide either metallically or by means of an r.f.
bypass or choke circuit. A more detailed section on such coupling methods
will be given later.
The tube employed a standard octal base modified to pass the coaxial line.
Thus if a standard octal socket is similarly modified and mounted on the
wave guide it is possible to couple the oscillator to the wave guide and
power supply circuits simply by plugging it into the socket, just as with
any conventional vacuum tube.
The tuning means for this type of oscillator tube presented a serious
problem. This will be appreciated when it is realized that the mechanism
must permit setting frequencies correctly to within one megacycle in a
device in which the frequency changes at the rate of approximately 200
megacycles per thousandth of an inch displacement of the grids. In other
words, the tuner was required to make possible the adjustment of the grid
spacing to an accuracy of five millionths of an inch. The design of the
mechanism adopted was originated by Mr. R. L. Vance of these Laboratories.
The operation of the tuner can be seen from an examination of the
cross-section and external views of Figs. 63 and 67. On one side of the
a strut extending from the base is attached to the repeller housing. This
strut acts as a rigid vertical support but provides a hinge for lateral
motion. On the opposite side the support is provided by a pair of steel
strips. These are clamped together where they are attached to the vacuum
housing support and also where they are attached to a short fixed strut
the base. A nut is attached rigidly to the center of each strip. One nut
a right and the other a left handed thread. A screw threaded right handed
one half and left handed on the other half of its length turns in these
and drives them apart. The mechanism is thus a toggle which, through the
linkage provided by the repeller housing, serves to move the grids
to one another and thus to provide tuning action.
The 723A/B was originally designed for a relatively narrow band in the
vicinity of 9375 megacycles. It operates at a resonator voltage of 300
and the beam current of a typical tube would be approximately 24
milliamperes. The design was based on the use of repeller voltage mode
with the manufacturing tolerances lay between 130 and 185 volts at 9375
megacycles. It is difficult to establish with certainty the number of
of drift for this mode. Experimental data can be fitted by values of
6 3/4 or 7 3/4 cycles and various uncertainties make the value calculated
from dimensions and observed voltages equally unreliable. This value is,
however, of interest principally to the designer and of no particular
in application. The performance was specified for the output line of the
oscillator coupled to a 5/8" x 1 1/4"' wave guide so that the
projected full length into the guide through the wider wall and on the
of the guide. With a matched load coupled in one direction and a shorting
piston adjusted for an optimum in the other the oscillator was required to
deliver a minimum of 20 milliwatts power output at a frequency of 9375
megacycles. Under the same conditions the electronic tuning was required
be at least 28 megacycles between half power points.
For reasons of continuity a more detailed description of the properties of
the 3 centimeter oscillator will be given in a later section. The 723A/B
oscillator served as the beating oscillator for all radar systems
in the 3 centimeter range until late in the war when the 2K25 supplanted
At the time that the 723A/B was developed the best techniques and
available were employed. In retrospect these were somewhat primitive and
course this resulted in a number of limitations of performance. Since the
tubes designed as beating oscillators commonly served as signal
in the development of ultra-high frequency techniques and equipment the
wartime designer of such oscillators usually found himself in the position
of lifting himself by his own bootstraps. In spite of these limitations
later modifications of the 723A/B which led to the 2K25 did not
fundamentally change the design but were rather in the direction of
extending its performance to meet the expanding requirements of the radar
art. The incorporation of the resonant cavity within the vacuum envelope
resulted in a major revision of the scope of the designer's problems. He
assumed a part of the burden of the circuit engineer in that it became
necessary for him to design an appropriate cavity and predetermine the
correct coupling of the oscillator to the load. The latter transferred to
the laboratory a problem which in the case of separate cavity oscillators
had been left as a field adjustment.+++++++++++
( picture John Pierce Holding An Early Traveling Wave Tube.)