MAGNETRON AS GENERATOR OF CENTIMETER WAVES 327 



It might be surmised that the trouble in the unbroken strap case was due 

 to a component of // = 7 wliich did not couj)lc to the output circuit. The 

 phenomenon was studied in an operating magnetron with small electro- 

 static probes built into several resonators. The n — 7 components were 

 identified, their relative intensities being approximately those expected. 

 It is of some interest to note that the "moding" encountered here differed 

 from that seen in magnetrons of longer wavelength. The magnetron 

 seemed invariably to start in the w mode, falling into oscillation in the 

 n = 7 mode more rapidly as loading increased. As might be expected from 

 this behavior, the mode boundary on a performance chart is little afifected 

 by the rate of rise of the pulse as it had been in other magnetrons. Later 

 evidence made it appear that the change in the operating voltage of the 

 n = 7 mode brought about by the redesign of the resonator system was 

 a decided improvement. 



The necessity of using strap breaks led to an unforeseen difficulty. Un- 

 der the influence of RF electric forces, the overhanging ends of the straps 

 at the strap breaks moved together and shorted, causing failure after only 

 a few hours of operation. The trouble was eliminated by removing the 

 overhanging portions of the straps with no noticeably harmful effects. 



In a 3 cm. magnetron, strapped as heavily as the 4J50, circuit losses 

 become very important in determining the over-all efficiency. A noticeable 

 improvement in unloaded Q was effected when an atmosphere of prepurified 

 No was substituted for the C02-alcohol mixture which had previously been 

 used to prevent oxidation during the final brazing. The C02-alcohol 

 method had been abandoned for another reason, namely, that chemical 

 analysis showed carbon deposited on the steel pole-pieces underneath the 

 copper plating. 



If one determines the circuit efficiency for matched load by impedance 

 measurements on a non-oscillating magnetron [see equation (25) of PART I], 

 one may then calculate its value for any load. From measurements of 

 over-all efficiency as a function of load conductance, the dependence of the 

 electronic efficiency on this conductance may be determined. The curve 

 of Fig. 19 was obtained in this way for the 4J50. The fact that the elec- 

 tronic efficiency, as seen in Fig. 19 is practically independent of load, that 

 is to say, of electronic conductance, together with some observations of 

 the load sensitivity of the "moding" in the 4J52, led to consideration of 

 a new design for the resonator system. It was found invariably that the 

 boundary for mode change in the 4J52 magnetron moved to higher currents 

 as the load impedance was changed to values further removed from the 

 frequency sink and power maximum in any direction on the Rieke diagram. 

 This suggests that, as the circuit conductance [Gs of equation (36)J is 



