VIII. A. TUBE RESEARCH AND DEVELOPMENT MAGNETRON DEVELOPMENT Dr. S. T. Martin A. G. Barrett 1. Testing and Design of High-Power 10. 7-Cm Magnetrons MF-8B magnetron was baked out at 600'C for 91 hours and attained a pressure of -6 3.5 x 10 mm Hg at this temperature. After checking the direct current emission, the tube was connected for pulsed operation with the Model 16 rotary gap modulator and later with the 20-Mw modulator. Performance in both cases was poor, with a maxi- mum output of 0. 85 Mw for the following conditions: Anode Voltage 24. 8 kv Magnetic Field 1450 gauss Current 128 amp Pulsed Duration 1. 0 4sec PRF 350 pps Operating time for the tube while still undergoing tests on the pumps totaled 9 hours. A leak at the window metal-ceramic seal occurred while pressurizing the output test tank to reduce sparking and ended further tests. Dissection of the window disclosed that the metal-to-ceramic bond was structurally weak, vacuum-tight. although it had originally tested Only 15 psi pressure was required to cause failure. Cutting the tube open revealed damage that was apparently due to overheating. In addition, the kovar window cup was pushed into the copper output transition piece by the window jig with enough force to make Z 0 an undesired diffusion joint. UJ This radical change in the internal output UJ C' circuit geometry accounted for much of the poor performance of the MF8B magnetron. As a result of MF-8B experience, tests were conducted to determine Lile proper tiesI PRESSURE psi GAUGE HYDROSTATIC UI WIIUUW iau m WiLLI AI-200 ceramic when subjected to Fig. VIII-I Deflection vs. static load, AI-200 ceramic window. -33- static loads parallel to the window axis. Deflections of the ceramic (VIII. TUBE RESEARCH AND DEVELOPMENT) disk were measured as functions of applied hydrostatic loads for two window s using the arrangement shown in Fig. VIII-I. At 110 psi gauge the difference in deflections for one window was 2. 7 mils, and 3.5 mils for the other window. window are given in Fig. VIII-I. The curves for the latter Two windows were loaded to the breaking point. Fail- ure occurred at 120 psi in one case and at 155 psi for the window of Fig. VIII- I . Figure VIII - 2 is a photograph of the window of Fig. VIII - 1 taken from the vacuum side after rupture. Ink has been flowed onto the window to bring out the cracks. In this case the metal- to-ceramic bond had adequate strength and failure occurred in the ceramic disk. On the basis of these tests all windows are now load-tested and then again vacuum-tested before acceptance for assembly to a magnetron. Each window is subjected to a load of 110 psi for one hour. Fig. VIII-2 During magnetron oper- ation pressure in the test tank will be AI-200 ceramic output window after load test. limited to 90 psi so as not to exceed the total pressure of the load test. The gold-wire diffusion method of attaching the window to the magnetron has produced vacuum -tight seals each time it has been used. The seal is now made before mounting the cathode and the end plates by maintaining the tube at 750°C for 90 minutes in the hydrogen furnace. The jig is removed after firing and the tube is ready for final assembly. MF- 9B magnetron was the first tube to be assembled with the latest techniques and construction sequence. This tube appeared to bake out rapidly on the pumps, but leaks developed in the glass portion of the vacuum system and in the end plate structure of the tube and prevented completion of the processing cycle. This tube has now been set aside for rebuilding. Two attempts have been made to rebuild MF- 8B magnetron; the second is still in process. The first attempt, designated MF-8BR, failed when the ceramic disk in the window cracked during assembly to the tube. The second attempt, designated MF-8BS, is in the window-assembly stage. Parts are on hand for assembling MF-10B magnetron, the last tube of the oxide 'cathode series. Revision of the drawings for these magnetrons has been completed. -34- (VIII. B. TUBE RESEARCH AND DEVELOPMENT) MICROWAVE TUBES L. D. Smullin G. Guilbaud C. Prof. L. J. Chu H. Haus L. Roberts Prof. J. Thomas J. Houston H. E. Rowe Krusemeyer L. Stark E. A. W. Boekelheide 1. a. M. H. J. E. Muehe, Jr. Noise and Space Charge Waves Small signal theory for one-dimensional flow Technical Report No. 207 has been prepared for publication. L. D. b. Smullin Experimental study of noise on election beams The apparatus described in the last report has been put into operation and preliminary results have been obtained on the noise- standing wave ratio in the electron beam. A typical example of the general type of standing-wave pattern obtained by moving the first cavity along the electron beam and measuring its output noise power is shown in Fig. VIII-3. Similar data have been taken for a range of beam voltages from 500 to 0 DISTANCE IN CM Fig. VIII-3 Noise current standing wave on electron beam. current < 20 pa, P = 10-6 mm, B = 325 gauss. -35- V = 1500 v, I o = 3.25 ma, intercepted (VIII. TUBE RESEARCH AND DEVELOPMENT) 2000 volts, a range of beam currents from 1 to 5 ma, and a range of magnetic fields from 200 to 800 gauss. The plasma wavelengths obtained in this way have agreed very closely with the calculated values. One notable feature of this curve is the difference in the magnitude of the various minima; from the theory it would be expected that all the minima would be the same size, and, of course, that all the maxima would be the same size. This difference in the size of the various minima is believed to be caused by partition noise generated by very small amounts of current striking the front of the measuring cavity; the intercepted current varies as the cavity moves along the beam, and, as would be expected from the analysis carried out for partition noise at low frequencies, the partition noise increases with the intercepted current. It appears that the intercepted current must be reduced to a small fraction of 1 percent of the beam current before partition noise can be ignored. In our preliminary measurements the intercepted current was 1 to 1.5 percent and the SWR was approximately 7 db. The data in Fig. VIII-3 were obtained with the intercepted current less than 0.5 percent. Present work is directed toward reducing interception to less than 0.1 percent of the beam current. The vacuum in the tube becomes important when it is required that less than 0. 1 percent of the electrons suffer any collisions, and pressure less than 10 -6 mm Hg must be maintained. H. E. Rowe c. Traveling-wave tube design A demountable, wave tube designs. continuously pumped system is being built for testing travelingA shielded electromagnet encloses a drift tube 2.25 inches in diameter and 16 inches long, within which helices, velocity jump sections, and other electrodes can be located and moved as desired. A shielded housing is provided for the electron gun. 2. C. E. Muehe, L. Roberts Operation of Pulsed Magnetrons into a High-Q Load The measurements described in earlier reports, comparing the operation of the magnetron with ballast loads of different impedance with the operation of the magnetron using the switching device have been completed. A technical report is being prepared. H. J. 3. a. Krusemeyer 1-MEV Pulsed Electron Source Tube Mechanical construction difficulties have delayed the completion of the Pierce-gun type, high-voltage diode but it is now in the final assembly stage. culty was in spinning the large cup-shaped cathode electrode. -36- The principal diffi- This was finally achieved (VIII. TUBE RESEARCH AND DEVELOPMENT) with the use of sheet nickel, 0.037 inch thick before spinning. In addition, it was found that the kovar anode disk, pictured in the Quarterly Progress Report, January 15, 1951, deflected enough to break the glass seal (8 5/8 inches in diameter) when the tube was evacuated. This condition was overcome by spinning the kovar into a spherical shape. An oversize bake-out oven and vacuum system have been completed for processing the tube. Tests on a beryllium electron window, 0.003 inch thick, have shown this to be the ideal material if such a thin sheet can be sealed with a vacuum-tight seal. Its energy absorption is low, its thermal conductivity is high, and it has sufficient tensile strength to withstand the tube bake-out at 4500C with no noticeable creep. However, all samples obtainable to date have had too rough a surface for a vacuum-tight gasket seal with either a copper or a gold gasket. A strong braze joint was obtained between the sheet beryllium and nickel-plated stainless steel, with the use of BT silver solder and a titanium hydride reducing agent. However, oxide pockets in the beryllium were reduced in the process, thus forming small vacuum leaks. Because of these difficulties with beryllium the use of a stainless steel window 0. 001 inch thick is planned for the first tube. of beryllium may be carried on later. Further research on the possibilities It is calculated that the stainless steel window will reach a temperature of 6500C at the desired duty ratio of 1:2000. A sample win- dow was held at this temperature for several hours with no sign of creep or vacuum failure. Metallic zirconium is being used as a getter in this tube. One piece will be attached to the cathode heat shield, where it will operate at 300 C to 400 0 C, the range of maxi0 mum hydrogen absorption. Another piece will be attached to the cathode cylinder itself where it will absorb other gases at approximately 800°C. A. b. W. Boekelheide Modulator The modulator has been completed and tested with a 1-p.sec pulse and repetition rates varying from 80 pps to 400 pps. primary of the pulse transformer, Approximately 25 kv can be delivered to the corresponding to 1. 25 Mv at the secondary. At higher voltages corona develops along the glass reservoir bulb of the thyrotron in the vicinity of the anode lead flare which, of course, defines the upper limit. Other com- ponents in the modulator are capable of higher voltages but to use them to full capacity would mean a series-parallel hookup of four thyrotrons. It is not planned to include these extra tubes at this stage. The 75-kv, 0. l-pf capacitor had to be replaced by a pulse-forming network. The capacitor was found to break down at voltages in excess of 15 kv. L. -37- Roberts (VIII. TUBE RESEARCH AND DEVELOPMENT) 4. Spiral Beam Reflex Oscillator Figure VIII-4 shows a cross section of a new type oscillator. It consists of a TE resonant cavity with an electron beam projected along the axis (perpendicular to the E field). An axial magnetic field is also provided, as shown. It was observed that with the beam collected on electrode R, the Q of the cavity was greatly reduced with the magnetic field adjusted to cyclotron resonance at cavity frequency. (See L. P. C. 35, I. Shulman: Smith, Frequency Modulation and Control by Electron Beams, Proc. IRE, 644, July 1947.) If electrode R was held at cathode potential, or lower, the system oscillated at the cavity resonant frequency, with the same magnetic field as above. With a beam of about 10 ma at 700 volts, the power output was about 30 mw at 3000 Mc/sec. Further experiments will be made to determine the effect of variations in the shape of the reflector field, and beam impedance. 5. J. D. Smullin Cut-off Frequencies of Single and Multi-Filar Helices The cut-off frequencies predicted by S. No. E. Thomas, L. Sensiper (forthcoming Technical Report 194) have been experimentally verified. The results will be presented in Technical Report No. 208. 6. L. Stark A New Gaussmeter A new type of gaussmeter is being developed. magnet mounted on jewel bearings. It consists of a small permanent The natural period of the magnet is in the static magnetic field being measured. determined A small coil of copper wire, driven by an audio oscillator, is mounted so that it sets up an a-c magnetic field perpendicular to the field to be measured. of the magnet is reached. One varies the audio frequency until the resonant frequency The impedance of the coil changes rapidly near the resonant frequency of the magnet, and this change is detected as an unbalance in a bridge circuit. See Fig. VIII-5. The resonant frequency of a magnet pivoted in a magnetic field is given by f 1 BM where B is the static magnetic field being measured, M is the magnetic moment of the magnet, I is the moment of inertia of the magnet, and o is the permeability of vacuum. 2 Thus, B is proportional to f2. To get a high ratio of M/I, one must use a small magnet of a strongly magnetic material. Also, it is desirable to choose a magnetic material of high coercive force to prevent demagnetization. A rather crude test model has been constructed and tested. The magnet consists of a nearly cubical piece of Alnico V about 4 mm per side, mounted on pivots and vee jewels. The driving coil consists of about 1400 turns of No. 40 wire. -38- The device ER IN Fig. VIII-4 Cross section of spiral beam reflex oscillator. LMAGNET PIVOTS ABOUT AXIS PERPENDICULAR TO THE PAPER DRIVING COIL Fig. VIII-5 Bridge circuit for measuring resonant frequency for probe magnet. I. 2. 3. 4 5. CYLINDER OF COBALT- PLATINUM MAGNETIC ALLOY SHAFT OF POLISHED CARBALLOY SAPPHIRE JEWELS BRASS DRIVING COILS Fig. VIII-6 Cross section of oscillating magnet gaussmeter. (VIII. TUBE RESEARCH AND DEVELOPMENT) has been tested over the range of 40 to 1200 gauss. At 1200 gauss the resonant fre- quency is 152 cps and the half-power bandwidth is 5 cycles. The audio driving voltage applied to the bridge is about 0.4 VRMS and the bridge unbalance at resonance is about 0.01 VRMS. Both of these voltages seem rather independent of frequency over the measured range of 20 to 152 cps. A more refined model having as its magnet a cobalt-platinum cylinder 0.07 inch in diameter and height is now under construction. diameter. It will fit into a probe 0.28 inch in A cross section of the new design is shown in Fig. VIII-6. J.M. Houston -40-