FLUCTUATION NOISE IN VACUUM TUBES 639 



milliamperes, and ordinates represent mean square noise voltage 

 across the output measuring device expressed in arbitrary units. The 

 change in space current was obtained by varying the filament heating 

 current while the plate voltage remained constant. Tubes having 

 thoriated tungsten, tungsten, and barium oxide cathodes were used. 



At low space currents where no space charge is present the thoriated 

 tungsten and tungsten filaments each give a pure shot effect, the mean 

 square voltage increasing linearly with the space current. As the 

 space current is increased further and space charge sets in, the shot 

 voltage in each tube goes through a maximum and decreases with 

 oncoming temperature saturation as suggested by equation (5). With 

 the approach of complete temperature saturation the noise, however, 

 does not decrease to zero in accordance with this equation. If it were 

 possible to reach complete temperature saturation the residual noise 

 would not be due to the shot effect, but rather to thermal noise in the 

 plate circuit of the tube, positive ions and secondary emission within 

 the tube, and other contributing causes. Usually this condition is 

 approached in the better commercial tubes so that the contribution of 

 true shot noise is a small part of the total noise. 



If the methods used in obtaining equation (4) are applied to equation 

 (5\ it is found that the shot noise in the plate circuit of the tube 

 produces the same effect in the output measuring device as a signal 

 applied to the input circuit whose magnitude at the grid expressed in 

 mean square volts is 



F = 2ej{dildjy{r,l^,fF. (6) 



This equation shows that the level of shot noise at the input is lowered 

 by an increase in the cathode temperature, which increases the degree 

 of temperature saturation, and by an increase in the ratio yu/r^, which 

 by definition is the transconductance of the tube, but is independent 

 of the external load resistance. It should be remembered, however, 

 that shot noise in the plate circuit should not fix the noise level in 

 low noise vacuum tubes and that never, as is sometimes done in the 

 literature, can the noise of an amplifier be calculated as pure shot 

 noise in the plate circuit, for in the absence of space charge the tube 

 would not be an amplifier. 



Although space charge can counteract the effect of random electron 

 emission from the cathode so that shot noise is reduced, other factors 

 can alter the flow of current in such a way that the noise is increased. 

 This is the case when changes in emission occur over small areas of the 

 cathode, giving rise to an additional fluctuation which has been 

 termed flicker effect.^ This type of noise is particularly noticeable 



^ J. B. Johnson, Phys. Rev., 26, 71 (1925); VV. Schottky, Phys. Rev., 28, 74.(1926). 



