REGENERATION THEORY AND EXPERIMENT 



697 



30° 





 360° 



240 270 300 



Fig. 11 — The transfer factor diagram for the amplifier of Fig. 9 with the feed-back 

 attenuator set at zero decibel. The four curves shown correspond to different 

 amounts of the 60-kilocycle load. 



was closed. Two sets of curves are shown, one for each of the low- 

 frequency crossings. These are plotted against the loading amplitude. 

 The agreement between the experimental and predicted values is close 

 for the higher gain crossing at small loading amplitudes, but a di- 

 vergence is apparent at high loads. For the lower crossing there is a 

 divergence of 1.5 decibels at low loads, which changes sign and be- 

 comes greater at the higher loads. These divergences may be ascribed 

 to a variety of causes among which probably the most important are 

 the effects of harmonics upon the amplifier loading, overloading of the 

 measuring apparatus by harmonics of the loading electromotive force, 

 and phase shifts introduced by the selective circuits. The last two 

 causes may be eliminated by improved technique, but the first cause in 

 general introduces a fundamental difficulty, particularly important 

 when large nonlinearities are involved. 



Negative Impedances 



One of the early forms of stability criterion mentioned in the first 

 section was that relating to the measured impedance of the circuit. 



