How to Construct a 56 Megacycle Magnetron Transmitter
September 1932 Radio News

September 1932 Radio News [Table of Contents] Wax nostalgic about and learn from the history of early electronics. See articles from Radio & Television News, published 1919-1959. All copyrights hereby acknowledged.

Magnetrons and klystrons are fairly ubiquitous in society these days for use in heating, radar, industrial processes, cooking, and even lighting. They were probably the first useful means of producing high power microwave signals. The concept was first brought to fruition in the early 1920s as a laboratory curiosity and rapidly developed into a practical type of device with many applications and spin-off products like the klystron, the traveling wave tube, and the cross-field amplifier. This article from a 1932 edition of Radio News magazine reports on the state of the art a decade after the magnetron's inception.

How to Construct a 56 Megacycle Magnetron Transmitter

Here it is, something new in radio! A new oscillator principle for the ultra-short waves employing the gridless double plate tube in a manner comparable in many respects to the Gill-Morill method. This article, a feature of the Radio News series on opening up the ultra-short wave field, will be invaluable to the technician and earnest experimental amateur

By James Millen

The American amateur was the first to make practical use of the 200-meter band and the first to develop the utility of still shorter and vastly more important communication channels. A new field of research, the ultra-short waves, now challenges his ingenuity, and we are sincere in our belief that here again the "ham" will make a genuine contribution to ultra-high-frequency technique. The magnetron, in particular, offers fertile possibilities, and its application to commercial enterprise may be, to our way of thinking, materially accelerated by its exploitation and development in the amateur ranks. This article finds dual justification in the effort to stimulate such experimentation, and in presenting a practical 56-megacycle magnetron transmitter.

As we pointed out in the original article of this series, there exist a variety of methods whereby ultra-short-wave energy can be set in motion. However, the necessity for efficiency (reasonably high-power output for practical input powers) and stability places a definite limitation on the systems serviceable for useful communication. The magnetron, today, offers the most economical method for generating quasi-optical power. As it is an electronic device, its functioning is perhaps best understood by indicating its similarity to more conventional systems. It is not particularly difficult to design the usual sort of tube oscillators for wavelengths between 5 and 10 meters, and by the utilization of their harmonics to extend this range to a still lower minimum. However, as might be imagined, the stability of such systems leaves much to be desired, and the power output is generally inadequate. Also, as may be readily understood, maximum frequency limitations are necessarily imposed by considerations of the capacity and inductance by which resonance is determined. An additional complication, the fact that as the frequency is still further raised the period approaches the times required for the electrons to complete their inter-electrode cycle, imposes further limitations - at the same time offering a solution to the problem. It was found that, under proper conditions, oscillations could be sustained the frequency of which was dependent on the geometry of the tube or on the potentials applied to the elements rather than upon the LC characteristics of the circuit. Such systems have been described categorically as Barkhausen-Kurz circuits in deference to their two most-prominent investigators. It is logical and true that such arrangements are capable of delivering higher powers, at very short wavelengths, than those with which we have become familiar on the conventional short waves. It was also discovered that the power output could be increased by resonating the circuit to the natural electronic frequency, and such transmitters have come to be known as Gill-Morill circuits and are comparable in many respects with the magnetron system.

The magnetron is not a new tube. It has been with us for well over a decade, and was originally designed as a tube in which electron control was effected by magnetic influence rather than electrostatically. In other words, the grid of the tube was a magnetic coil, and, peculiarly enough, the magnetron was first used as a radio-frequency amplifier and oscillator in the neighborhood of 8000 meters!

The magnetron used in ultra-short-wave work varies from its prototype in several essentials. The magnetic coil no longer functions as a grid (in the control sense), but its effect is similar to that of a "bias." By increasing the magnetization current, the space current can be cut off - corresponding to an increase in negative bias in a conventional triode circuit. At optimum magnetic bias, the space current is reduced while space charges build up within the tube. Connected with a suitable circuit, a negative-resistance characteristic permits these charges to dissipate and re-accumulate, producing a cycle the time constant of which is partly determined by the spacing of the elements, the intensity of the magnetic field and the potentials applied. Subsequent experiments have shown that, as in the Gill-Morill circuits, the efficiency of the magnetron is greatly increased when the circuit resonance approaches the natural electronic period.

The magnetron used in the experiments and transmitter to be described is the GE-239, and is shown in Figure 1. The anode is split into two semi-cylindrical sections mounted coaxially with the heavy tungsten filament. It is an air-cooled tube, used principally as an oscillator. The lowest operable wavelength is 0.75 meters, and at 1 meter, with a plate potential of 1500 volts, has a plate impedance of 5000 ohms. The maximum operating anode potentials are 1500 volts d.c. and 2000 volts r.m.s. a.c. The maximum d.c. anode current is .075 ampere, and the maximum plate dissipation 60 watts. The inter-electrode capacities are anode-to-anode (filament grounded) .5 mmfd.; anode-to-filament (other anode grounded) .7 mmfd. The tungsten filament draws 5 amperes at 5 volts.

The mechanical characteristics of the tube are indicated in the photographs. The anodes are connected from the top, and the base plugs into a standard 50-watt socket. The overall length is 10 inches; diameter 2 1/4 inches.

The output of this tube, in both power and stability (when properly operated) is definitely superior to other arrangements, its output at .75 meter being vastly better than that available with the B-K G-M circuits.

Figure 2 shows the experimental layout from which the various data accompanying this article were secured. The circuit, as may be observed from Figure 3, is simplicity itself.

It consists of the tube between the anodes of which is connected the center-tapped coil. (As will be described, this coil was changed in the course of the experiments, and an exterior capacity shunted across it. The simple inductor is shown in Figure 9 and the coil plus the tuning condenser, are shown in Figure 10.) The field coil has been removed from its place around the magnetron for the sake of clarity.

The oscillating frequency is largely determined by the inductance and the capacity of the circuit. For the highest frequency (400 megacycles), the inductance of the anode leads and the capacity between them is sufficient. In this case the external anode leads are short-circuited with a copper bar or ribbon) about one inch from the glass, and the high voltage is connected to a tap at the center of the bar.

The Field Coil

The field coil is wound with 70 pounds of number 14 enameled wire, on a form having an inside diameter of 5 inches. The length of winding is 5 inches, 74 turns to the layer, with a total number of 2665 turns. Each layer is separated with paper insulation .015-inch thick. The coil is excited with a potential of 100 volts through a suitable resistor, providing a maximum current of 5.8 amperes and a maximum continuous field strength of 830 gauss with a dissipation of 580 watts. The outside diameter of the coil is 11 inches. The variable series resistor should be capable of carrying the maximum excitation current and should have a resistance of at least 17 ohms.

The effect of the magnetic field on the operating characteristics of the magnetron is, as we have already suggested, somewhat similar to that of a bias of the grid of a triode oscillator. At low values of field, below 20 amperes, corresponding to a small bias, the plates heat excessively; the plate current is high and the efficiency is low. As the field is strengthened, the circuit becomes more stable, operating with greatly improved efficiency and output. The current in this region varies from 3 to 5 amperes. Plate-current "cut-off" can be obtained with an excitation current of about 7 amperes.

In general, the requirements of the oscillatory circuit, in respect to efficiency, stability, etc., are similar to those of the dynatron oscillator, although, of course, the frequency range is much greater.

As the filament is tungsten, it is possible to operate it over a wide range of voltages without encountering the trouble experienced with thoriated filaments. It was found, however, that for best stability the filament should be operated near its rated voltage - though a 10 or 15 percent. variation has a relatively small effect.

The first tests, while altogether satisfactory in respect to output, left a great deal to be desired from the standpoint of stability. Curve A, of Figure 4, illustrates the manner in which frequency varied as the plate voltage was changed. In this original experiment the tuned circuit consisted of 4 turns of 1/4-inch copper tubing, 2 inches in diameter, mounted directly on the anode leads, the tuning capacity being the inter-electrode capacity of the tube (about .5 mmfd.) and the distributed capacity of the coil and leads.

The frequency was approximately 56 mc. The circuit and operating conditions under which this test was made probably represented the most unfavorable, with respect to frequency stability, that would be encountered in practice.

The total change in frequency, i.e., about 200 kc., seems rather appalling to one familiar with the operation of low-frequency apparatus. As a matter of fact it is not much worse than that encountered in a self-excited push-pull oscillator operating on the same ultra-high frequency. Such as oscillator, however, cannot be considered satisfactory. An investigation (with the idea of improving stability) was therefore begun, and while it is not fully completed, it represents definite steps in the right direction.

Stabilizing the Circuits

Each portion of the magnetron circuit has a relationship to frequency stability. Variations in the magnetic field were very bothersome, so it was deemed advisable first to investigate this. It was discovered that when a certain magnetic field strength was employed, variations in the field current had a minimum effect on frequency change. The optimum field current, from this point of view, was about 4 amperes. This is important since the field-supply filtering is something of a problem, due to the high current required. In these tests a motor-generator furnished the field power and when the correct adjustment was attained, a monitoring test disclosed the fact that such disturbances as commutator ripple and minor line-voltage variations were no longer of major importance. The magnetron was not operating quite at its maximum efficiency at this field setting, but the drop in efficiency was small enough to be considered important in the light of the improved stability obtained.

The frequency variation, with change in plate voltage, was still the same, however, so the "low C" circuit was abandoned in the hope that a certain amount of tuning capacity would be beneficial. The magnetron operates most efficiently into an oscillatory circuit of high impedance, making a "high C" circuit theoretically undesirable. Nevertheless a tuned circuit was constructed employing a coil of 3 turns of 1/4-inch copper tubing 1 1/2 inches in diameter, tuned with a condenser of about 20 mmfd. capacity. Since the inter-electrode capacity of the tube is only about .5 mmfd., this additional capacity effected a considerable improvement in stability, without serious loss in efficiency.

The operation of the circuit, as a whole, was quite different from the original tests. Changes in field current, plate and filament voltages caused only a relatively small variation in frequency. The curves in Figure 4 show the contrast between the "low" and "high C" circuits.

A definite value of field current which minimized the effects of commutator ripple, etc., was found as before, but was considerably lower in value, being approximately 3 amperes instead of 4 amperes (Figure 5). It is interesting to note that the optimum operating value of field current, with regard to frequency instability caused by field variation, is approximately the same at all plate voltages, as is shown in Figure 6. As before, better output was obtained with the field current reduced slightly from that value giving most stable operation. The choice of field current will therefore depend somewhat on operating conditions. If a well-filtered supply is available, efficiency and output are the deciding factors. Where the field supply is subject to variation, stability is the predominant consideration .

The value of R is not critical as far as efficiency or frequency vs. plate-voltage variations are concerned. However, it was found that by the correct adjustment of R, frequency variations due to field strength changes could be still further reduced materially. The actual value varies between 100 and 400 ohms and depends on the plate voltage - higher resistance being required for higher potentials. Some energy is necessarily dissipated in this resistor, but this loss is relatively unimportant in view of the main goal of a stable circuit. Also, some compensation of this loss is achieved through an improved anode circuit efficiency.

A d.c. Filament Source Preferred

After a fair degree of stability had been attained, as indicated in curve B, Figure 4, another source of trouble became evident. This was vibration of the filament structure of the tube, due to the filament current, which was alternating current, reacting in a "motor" effect with the strong d.c. field. When the filament was light by d.c., this trouble was eliminated.

Modulating the Circuit

Referring to the curves in Figure 7, it is immediately seen that the relation between the plate current and plate voltage is essentially linear, and the circuit, therefore, lends itself admirably to modulation. From the slope of the curve, the effective plate impedance is found to be about 9000 ohms, which, while somewhat higher than that commonly encountered in the usual transmitting tubes of similar power, should present no serious difficulty from the standpoint of obtaining a satisfactory impedance relationship in the modulator output circuit.

Photograph, Figure 8, shows the complete set-up of the magnetron, with its attendant apparatus, including modulation and speech equipment. The modulator consists of two type -50 tubes in push-pull, connected through a suitable transformer to the oscillator plate circuit. Since the magnetron was operating with an input of about 35 watts at 650 volts, a fair percentage of modulation was obtained with the two -50's operating at the same plate potential. In any case the modulation should not exceed 70% if reasonably good fidelity is desired. The speech equipment and amplifier circuits are quite conventional.

The radiating system was a simple vertical, half-wave, copper-rod antenna approximately 8 feet long, and fed by means of a balanced transmission line.

Transmitting Tests

Field tests were made up to distances of about 5 miles, the signal being checked in comparison with that of a conventional modulated oscillator. Two types of receivers were used, one being a super-regenerator, and the other the latest type of ultra-high-frequency superheterodyne described in Radio News for August. It was, of course, impossible to check the effects of frequency modulation of the super-regenerator since the receiver itself is subject to a degree of frequency modulation from the suppressor frequency that is comparable with that of the worst transmitter. Due to this fact, only slight differences between the transmitters were apparent on the first receiver (the signal from the magnetron was appreciably sharper, but the tone quality was not noticeably better.)

On the other hand, when using the super-heterodyne, the magnetron was found to give a clear-cut signal of good quality, in contrast to a decidedly broad and wobbly signal of low intelligibility from the modulated oscillator.

The tests definitely checked the various circuit requirements that laboratory experiments had indicated to be desirable. Of the two transmitters, each employing similar modulating equipment, the magnetron gave a much stronger signal due, of course, to the higher degree of efficiency. Increased power at these frequencies does not result in a corresponding extension of range, since the behavior of the signal tends to comply with optical laws. However, the stronger signal was found to be much more effective in location subject to local interference.

The possibility of modulating the r.f. output of the circuit by supplying the modulating power to the field coil will doubtless occur to many experimenters. However, aside from the difficulty of controlling the high excitation current with any depth of audio-frequency variation, field circuit modulation might be undesirable due to the excessive change in radio-frequency with field current change. Also, the r.f. output does not vary uniformly with the field, except over rather small limits and when using a somewhat restricted range of plate voltages.

Higher Frequency Possibilities

The tests described above were confined to the neighborhood of 56 megacycles. A vast amount of experimentation still remains before the possibilities of the magnetron are fully realized on frequencies above 300 megacycles (below 1 meter). Work in this region, which is not being used at present for any practical purpose, affords a highly interesting and unusually fertile field for the experimenter, and the magnetron, at the present time, is the tube best suited for this phase of exploration.

* The National Co.

Posted November 26, 2021
(updated from original post on 10/21/2014)