the Form of Free Triode Vibrations, 205 



Such a simple analysis of: the action of the circuit, of 

 ^course, assumes that the oscillation characteristic is the same 

 for increasing and decreasing values of v a . The oscillograms 

 show that in a practical case these may not be identical and 

 that the difference increases with increase of frequency. 

 The general shape of the oscillograms is, however, in agree- 

 ment with that of the oscillation characteristic shown in fig. 3. 



The fact that the difference between the oscillation 

 characteristic for rising and falling values of v a is most 

 marked when the anode current approaches saturation value, 

 suggests at once two possible explanations which may be 

 put briefly as follows : — 



M 



(a) The relation v g = — ^- v a may not be strictly true if 



grid currents are large, since in that case the grid potential 

 v g will not necessarily be the same as the E.M.F. v go induced 

 in the coil L 2 . 



(b) The thermionic emission may not be strictly constant 

 since the temperature of the filament will be influenced both 

 by the Joulean heating of the filament by the anode current 

 and by the cooling due to electronic evaporation. Both these 

 effects are periodic in character. 



Let us consider the effect of (a) in greater detail. We 

 have seen that for large positive grid potentials the anode 

 potential is very low and the current flowing in the grid 

 circuit may be greater than that flowing in the anode circuit. 

 The vacuous space in the tube between grid and filament for 

 these conditions can no longer be regarded as of infinite 

 resistance, as is the case when v g is negative. A considera- 

 tion of the actual magnitudes involved shows that the 

 "effective resistance" p of this space for large positive 

 values of v g may be of the order of one or two thousand 

 ohms. Thus if we distinguish between the grid potential v g 

 and the potential v go impressed on the coil L 2 , we have 



1 + *P L 2 + 5? ' 



P P 



The ratio — 2 was clearly negligible for the circuit used tor 



p l 



the oscillograms, but - — - would be comparable with unity 



at the higher frequency when v g ^> v a0 '-i-Va. There is thus an 



angle of lag tan -1 -^~ at any instant which would account 



