REFLEX OSCILLATORS 



507 



occurred only on one end of the repeller characteristic and was absent on 

 the other. The key to this situation lies in the fact that M t and A6 do not 

 vary in the same way when the repeller voltage is changed and the fre- 

 quency shifts as shown in (8.12). As a result, the resulting limiting function 

 does not shrink uniformly with repeller voltage, since the contribution 

 Ge changes more rapidly than G^ . Hence we should need a continuous 

 series of pictures of the limiting function in order to understand the situa- 

 tion completely. 



AMPLITUDE OF OSCILLATION, V >- 



Fig. 29. — Theoretically derived curves of electronic conductance vs amplitude of oscil- 

 lation. Curve G" shows the variation of the resultant electronic conductance when 

 the repeller space contribution and the cathode space contribution are in phase addition. 



Suppose we consider Fig. 29 and again assume in the interests of simplicity 

 that Mt and A0 vary at the same rate. In this case we observe that in the 

 region aa' the conductance varies very rapidly with amplitude. This would 

 imply that in this region the output would tend to be independent of the 

 repeller voltage. If we refer again to Fig. 27 we observe that the output is 

 indeed nearly independent of the repeller voltage over a range. 



We see that these facts all fit into a picture in which, because of the more 

 rapid phase variation of 6 1 than 6 with repeller voltage, the limiting function 

 at one end of the repeller voltage characteristic has the form of Fig. 28. 

 accounting for the hysteresis, and at the other end has the form of Fig. 29, 



