294 BELL SYSTEM TECHNICAL JOURNAL 



The results were found to fit well into the picture of starting time behavior 

 discussed in PART I. Thus, increase of frequency and decrease of load 

 both result in a greater rate of RF voltage buildup and permit a greater rate 

 of rise of DC voltage through the range where oscillation starts. The mag- 

 netron may thus be driven harder and more current passed. In agreement 

 with these results, changes in the rate of voltage rise applied by the pulser, 

 accomplished by changing its characteristics, also permitted oscillation to 

 continue to higher current and voltage values. The dependence on B and 

 rc/ra presumably are to be accounted for by the dependence of GeiVRp) on 

 these quantities. It appears that GeiVRf) increases with both B and Tc/ra. 



Empirical relations of the critical current, h , and the electronic efficiency, 

 77e , at which x mode failure occurs, to the load conductance, Gs , were ob- 

 tained.-^ Whereas /c decreases with increasing Gs , Ve decreases. 



Since the work was done at different frequencies and interaction space 

 geometries, it was found convenient and instructive to express voltages, 

 currents, and conductances as reduced variables of the type introduced by 

 Slater, that is, as ratios to the voltage, Vo , and current, h , at the intersection 



of the Hartree line and the cutoff parabola, and the conductance, Go = t^ , 



' O 



respectively, /o and Vo may be determined from the expressions for the 

 cut-off parabola and the Hartree line [expressions (8) and (16) of PART I, 

 respectively]. Plotted in terms of the reduced variables 7c//o and Gs/Go , 

 it was found that the data relating h/Io to Gs/Go and Tje to Gs/Go pre- 

 dicted approximately the same relationship independent of frequency and 

 rc/ra. . Thus, quite independent of the fundamental significance of these 

 particular reduced variables to the functional dependence of the quantities 

 involved, it appeared that they would be useful in the design of a new 

 magnetron whose design parameters would lie in or near the range for which 

 data were available. 



The purpose of the work was to produce a tunable magnetron which 

 would operate satisfactorily to currents in excess of those needed to meet 

 the power output specifications. It was hoped to do this by redesign of the 

 magnetron itself. This course rejected as unsatisfactory the alternative of 

 limiting the rate of voltage rise on each equipment with which the mag- 

 netron would be used. The general goal of the proposed changes in mag- 

 netron design were to increase the rate of buildup of RF oscillation while 

 at the same time keeping the electronic efficiency at a reasonably high value. 

 These goals are not independent but oppose one another as has been seen. 

 Thus, if fc/^a is increased, h increases, but Tje decreases. If Gs is increased 

 to increase r]e , h drops. The hope of success in the venture lay in the prob- 



Y 1 



25 Gg was determined from the relation G, = 77^ =] — 7^- [see equation (36) of PART 



I], using a calculated value for the total resonator inductance and the measured value of 

 the loaded Q. 



