INTERMITTENT BEHAVIOR L\ OSCILLATORS 15 



employed, the diagram of Fig. 19 differs markedly from that of Fig. 10. 

 Specifically the phase shift corresponding to a given value of attenuation is 

 greatly increased. As in previous cases the curve of over-all loop transmis- 

 sion may or may not loop the point (1,0) depending upon the relative 

 frequency scales. Thus if the points A and A' of Figs. 18 and 19 correspond 

 to the same frequency the Nyquist diagram passes near the point (2, 0) 

 indicating instabihty. If the points B and B' correspond to the same 

 frequency the loop passes very near to the point (1,0) and instability is 

 likely. 



By making the tuned circuits very selective or by reducing the thermal 

 time constant of the lamp circuit the points C and C may be made to cor- 

 respond to the same frequency. In this case the loop passes to the left of 

 the point (1, 0) and the system is absolutely stable. The same result may 

 be secured more easily by making one of the tuned circuits much more 

 selective than the other. This is ordinarily accomplished by increasing the 

 Q and impedance level of the grid circuit while keeping the Q and impedance 

 level of the plate circuit much lower so as to provide a suitable power output 

 to operate the lamp bridge. 



IX. The \'aristor Stabilized Oscillator 



A circuit which differs from that of Fig. 16 only in that the lamps are 

 replaced by varistors is shown in Fig. 20. At low levels of oscillation the 

 impedance of the varistors is relatively high, the loss of the limiter is low 

 and the amplitude of oscillation rises. At some higher level the varistor 

 impedance is reduced, the bridge approaches balance to the fundamental 

 frequency, and a stable condition is reached. Because the initial un- 

 balance of the bridge is opposite to that of Fig. 16 a reversal of phase is 

 necessary to establish oscillation. 



The stable condition reached differs from that of the lamp stabiHzed 

 oscillator in that the varistor goes through its entire range during each high- 

 frequenc}^ cycle. The lamp resistance changes by only a small amount 

 during any one cycle, its resistance depending on an integration of many 

 previous cycles. Two important facts arise from this difference. Har- 

 monics are produced in the bridge and, in so far as the varistors face react- 

 ances of these harmonic frequencies, intermodulation may produce currents 

 of fundamental frequency but shifted in phase with respect to the original. 

 Thus the bridge may produce a phase shift which is a function of level of the 

 oscillation frequency. A degradation of frequency stability results from 

 such a condition. More important to the present problem is the fact that 

 all modulation frequencies are transmitted alike. A small modulation is 

 reversed in phase and magnified by an amount depending upon the bridge 

 balance but not upon the modulation frequency. 



