48 



BELL SYSTEM TECHNICAL JOURNAL 



such a circuit will oscillate steadily or "sing" with no coupling but that 

 provided by its internal capacities. This phenomenon is frequently 

 encountered in vacuum tube amplifiers and at times proves quite 

 troublesome. 



Tubes can readily be constructed in which r\ and r 3 are so large 

 as to exert no influence on the behavior of the tube and may be ignored 

 in the above equations. However, even in such tubes there is an 

 effective input conductance, either positive or negative, depending 

 upon the character 2b of Z 2 . 



Case 2, High Frequencies. For very high frequencies terms of the 

 first order in p are negligible compared to terms of the second order, 

 and Eq. 5, becomes, 



Pida+dCs+acz) 



Y e =' 



G + C 3 



indicating that as the frequency is raised the effective input impedance 

 approaches that due to the condensers alone. Under these circum- 

 stances the grid absorbs very little power, but the amplification is 

 lowered because the input is to an extent short-circuited by the 

 electrode capacities. Fig. 20 shows the variation in voltage am- 



O0 !00 300 100 500 600 700 COO 800 000 1 



Wave Length-Meters 



Fig. 20 



plification against wave lengths in meters for high frequencies. The 

 two curves are for different E p , the higher E P giving a larger amplifica- 

 tion because r P of the tube is lower. It is seen that the amplification 

 at 1,000 meters is about three times as large as the amplification 

 at 100 meters and the amplification at both values of E p tends to 

 approach zero as the frequency becomes infinite. 



Nichols suggests 2 '' that the reduction in amplification for a given 

 frequency can be avoided by shunting the grid-plate capacity C$ with 



38 For cases in which Z 2 is neither pure resistance nor reactance, see Van der Bijl, 

 1. c, p. 210. 



29 H. W. Nichols, Phys. Rev., Vol. 13, p. 411, 1919. 



