MAGNETRON AS GENERATOR OF CENTIMETER WAVES 341 



methods of treating the parts, such as preglowing in vacuum, were used. 

 These largely eliminated the difficulty. 



An interesting feature of the problem was the relatively infrequent oc- 

 currence of hydrogen in magnetrons built at the Western Electric Co. 

 The reason for this was finally traced to the precaution, prescribed by the 

 safety engineers, that the sintering of the copper plate on the pole pieces 

 be done in an atmosphere of 95 per cent nitrogen and only 5 per cent hy- 

 drogen. All sintering and brazing operations in our Laboratories had been 

 done in a 100 per cent hydrogen atmosphere. Pole pieces sintered in the 

 predominantly nitrogen atmosphere were found to be nearly as free of hy- 

 drogen as the vacuum preglowed parts. However, the use of the zir- 

 conium getter was continued in models built at the Laboratories as added 

 insurance against recurrence of the difficulty. 



The other operational fault of the 3J21 magnetron was a tendency to 

 oscillate in a mode other than the x mode under certain circumstances. 

 Examination of a large number of magnetrons, opened after operation, 

 showed that the only ol servable causes were resonator distortion, the presence 

 of brazing flux or other foreign material, and off-center cathode location. 

 As in other magnetrons, the tendency to "mode" was greater when heavily 

 loaded or when operated with a sharply rising voltage pulse. \\'estern 

 Electric engineers concluded, as the result of an extended study of magne- 

 trons in which the cathode position was changed during operation by 

 flexing the input lead, that a cathode position 2 to 3 mils off center toward 

 the output resonator was desirable. A tool was developed with which a 

 permanent change in cathode location could be made on operating magne- 

 trons. By its use many "mody" magnetrons were reclaimed. 



An interesting phenomenon, discovered during the development of the 

 3J21 magnetron, is that known as the "cold start." Whereas most magne- 

 trons may be started with difficulty with a cold cathode, the 3J21 was found 

 to start readily under these conditions over several discrete bands of mag- 

 netic field intensity. Although the phenomenon is not well understood, 

 some correlation with the work of Posthumus^^ has been made. 



In making tests on the 3J21 magnetron at a duty cycle of 0.001 and a 

 pulse duration of 0.1 /iS, the cathode was found to operate at a consider- 

 ably higher temperature than at 0.5 ns pulse duration and the same duty 

 cycle and average input power. This behavior has since been observed in 

 the 725A and 4J52 magnetrons. It is thought to result from unfavorable 

 conditions during the periods of rise and fall of the voltage pulse, which 

 become a greater percentage of the total pulse duration, as the pulse dura- 

 tion is reduced. The effect presumably would not appear if the pulse 

 shape were truly rectangular. With present modulators, however, it 



