'The greatest scientific invention of the war' is the level at which a prominent scientist has assessed the cavity magnetron. The atomic bomb may have shortened the war, but after having heard Sir Robert Watson Watt measure radar in terms of the invisible reinforcements by which it effectively multiplied every aeroplane, gun, ship, tank, searchlight, etc., one doubts more than ever whether the war could have been won at all without radar. It is true that radar would have established itself in history by its part in the Battle of Britain alone, in which the magnetron was not present; but at least a year before then it had been realised that radar of quite a different kind would before long be urgently needed. It was in the Battle of the Atlantic, when a few magnetron-equipped radar sets checked the disastrous increase in ship losses.
What was wanted was something to provide a narrow beam of radio waves from an aircraft without reducing its speed. Because sharply directional aerials are necessarily large in comparison with the wavelength, this meant waves not longer than a few centimetres. And because only a very small fraction of the transmitter power reaches and is reflected by a distant target, and only a small fraction of that small fraction can be caught by the receiver, the transmitter power must be large - many times larger than had ever been generated. up to the outbreak of war.
This was the dilemma: the general method of reducing the wavelength is to make the oscillator smaller, whereas the general method of increasing the power is to make the generator larger, in order to prevent it from being burnt up by that part of the input power not converted into useful output. To make matters worse, the tendency was for that undesirable part to become a larger percentage of the input as the wavelength was reduced.
The Klystron, developed in America, had got over the transit time difficulty very neatly, and had shown what could be done by bringing the tuned circuits - highly efficient cavity resonators - right into the valve, to be excited directly by the electron stream; but that stream proved incapable of introducing power of the required many kilowatt order, even in pulses.
The magnetron as then known had produced only a few not-too-reliable Watts; but in I939, Prof J T Randall and Dr H A H Boot at Birmingham University had the idea that if the magnetron principle were combined with directly-excited cavity resonators a great increase in centimetre power would result. An experimental valve designed for 10 centimetre wavelength was on paper in November, 1939, and on the test bench three months later; and when at the first trial it yielded 500 Watts of CW at 9.8 centimetres the originators knew that they had been right. By June, I940, a workable sealed-off type of cavity magnetron, made by the General Electric Co. and Marconi-Osram Valve Co., was giving a peak output of 10 kilowatts. A sample was taken to the USA in August, 1940, as part of the first British contribution to the Anglo-American interchange of technical information, and a very sensational contribution it was. From then on progress has continued, and in 1945 a magnetron was produced by British Thomson-Houston with an output in the 10 cm (3 GHz) band of 2,500 kilowatts, or about 50 times as much as a BBC high-power medium-wave broad-casting station.
Now admittedly this power is on a peak or pulse rating, that is to say, it lasts for periods of the order of 1 microsecond, recurring about 1,000 times per second (the pulse recurrence frequency), so that the mean output is only about 2,500 Watts ; but even so, bearing in mind the size of the 'bottle' (seen in Fig. 1), it still seems rather fantastic.
Fig. 1. High-power magnetron, BTH BM735, with a peak output of 2,500 kilowatts, photographed alongside a 6 inch rule.
Compared with many modern valves of conventional grid-controlled type the magnetron is an amazingly simple device. It consists of a straight cathode surrounded by a cylindrical anode divided into two or more segments. A steady magnetic field must be provided, acting parallel to the cathode throughout the space. Fig. 2 illustrates a pre-war split-anode magnetron with external oscillatory circuit.
Fig. 2. The early type of split-anode magnetron with external tuned circuit, from which the cavity magnetron evolved. See also here.
Under the influence of the anode voltage alone, electrons would take straight radial paths from the cathode. But directly they start to do so, they come under the influence of the magnetic field, which forces them into curved paths, so that an electron starting off towards A1 may actually alight on A2, or even back on the cathode, depending on the anode voltage and magnetic field strength. If the tuned circuit is oscillating, the potentials of the two anodes alternate in opposite phase above and below the HT voltage, and given a suitable magnetic field, react on the electrons, making them take routes such that their energy maintains oscillation.
Fig. 3. Cross-section of the first experimental 10 cm cavity magnetron. C, the directly-heated cathode, was a 0.75 mm. tungsten wire.
Randall and Boot converted this valve to the integral-cavity principle in the beautifully simple way shown in Fig. 3, where cylindrical cavities are cut out of a solid copper anode. There are various ways in which such a set of resonators can oscillate; the most useful is that in which current flows round the cavities as shown, charging adjacent anode segments to opposite polarity. As there is thus a phase difference of 11 radians between any segment and its neighbour (one cycle = 2π) it is called the π mode. The frequency at this mode is fixed mainly by the diameter of the cavity. Oscillations in other modes (for example, from one-half of the anode block to the other) are at somewhat different frequencies, and it is obviously most undesirable that they should occur instead of or in addition to the desired mode. A great improvement in operating stability, and an increase in efficiency from 10-15% to 35-50% was effected by connecting together anode segments of like polarity, as shown in Fig. 4.
Fig. 4. Example of anode construction, showing anti-mode strapping. See also here.
These connections are duplicated at the other end of the anode. A second advantage of these straps is that within narrow limits the working frequency can be adjusted by bending them. Other ways of tuning magnetrons have been devised, but during the war they were generally manufactured for fixed frequency bands, spot frequencies within a band being obtained by selection.
Electrical engineers can hardly fail to be struck by the resemblance of Fig. 4 to the stator of an AC machine. The resemblance is rather more than superficial, if one interchanges current and voltage, electric and magnetic fields. For example: in the stator, currents circulate through windings in the slots, while magnetic fields alternate in the metal between; in the magnetron the reverse happens.
Fig. 5.Typical electron paths in a magnetron.
The oscillating potential distribution around the anode is equivalent to two patterns, such as that indicated roughly in Fig. 3, rotating in opposite directions (compare standing waves). Of the electrons emitted from the cathode, some arrive on the scene at such times and places that they receive energy from the oscillating electric field and are driven back on to the cathode, where their excess energy causes heat. Such a path is No. 2 in Fig. 5, and can be compared with path No. 1, representing the non-oscillating condition. Other electrons give up energy to the anode, and if they rotate around the cathode at about the same rate as one of the field patterns they may continue to energize the oscillations in several loops, such as path No. 3.
For efficiency, it is obviously desirable that the electrons should as far as possible fall into the second class. Nevertheless the others, although not adding directly to the oscillatory output, do provide most of the enormous electronic emission necessary for such high peak powers. For example, the CV76 gives a peak output of 500 kW, and as the efficiency is 50% the input must be 1,000 kW. With an anode voltage of 28,000, the anode current is thus 35 amperes. Most of this is secondary emission due to bombardment by returning electrons, the primary emission being only a few milliamps. After the magnetron has warmed up, the heater current can be switched off. The limiting factor in a magnetron is not so much the heat that can be put into the cathode as the heat that can be carried away from it.
Fig. 6. Cross-section of a cavity magnetron.
Fig. 6 shows the construction of a typical cavity magnetron, which follows the original experimental model of Randall and Boot remarkably closely. The output is extracted by a small loop feeding a coaxial cable or waveguide. The older 'hands', at least, find it hard to believe that this curl of a safety pin carries hundreds of kilowatts of RF, even momentarily. The loop and its external projection, as well as other constructional features, can be seen in Fig. 7.
Fig. 7. Internal view of the CV64, much used by the RAF.
For the 3-centimetre band it is of course necessary to reduce the dimensions; and in order to retain as high an efficiency and output as possible it is usual to have about twice as many cavities as in the 10 cm types. The output coupling leads as directly as possible into a waveguide.
The high conductivity copper that best satisfies the electrical requirements is not at all suitable for accurate and rapid machining, the dimensional tolerances being no more than ±0.0005 inch3}. This problem was solved by using copper alloyed with 0.5-1% of tellurium. The American solution was to build the anode up from laminations, increasing still more the resemblance to a stator.
Fig. 8. CV56 and field magnet.
The magnetic field is normally provided by a permanent magnet of alnico or alcomax. Fig. 8 shows the CV56 with its magnet; an early version of a type used by the Royal Navy.
Fig. 1 is the BM735, the high power magnetron already mentioned, with a peak output of 2,500 kW at an anode voltage of 40-45 kV.
The anode is obviously a much 'earthier' electrode than the cathode, so the normal practice is to earth the anode and supply high voltage negative pulses to the cathode. As the mode and frequency of oscillation depend on the HT voltage, it is important that the pulse reaches its maximum very rapidly and hold it for the duration, dying away again rapidly to zero.
The frequency depends also to some extent on the load, and in fact the waveguide coupling has been used as a means of frequency adjustment within about 1%. For the same reason, variations in load and standing wave ratio must be minimized.
No fewer than 200,000 magnetrons were manufactured in Britain alone for war purposes. In addition to peacetime military requirements, there is no doubt that increasing quantities of magnetrons will be needed for civil uses. The scope of radar in civil aviation is not very clear at the moment, but there can be little question that it will come into general use on ships, for which the optimum wavelength is between 3 and 6 cm, right in the magnetronbs territory.