650 BELL SYSTEM TECHNICAL JOURNAL 



and plate voltages.^® This reduction in noise is due to a decrease in 

 current to the floating grid. Using a heater current of 1.3 amperes, 

 a plate current of 0.1 milliampere, a screen potential of 16.5 volts and 

 a plate potential of 30 volts the equivalent input noise was 1.4 X 10~^ 

 volt for the entire frequency range from 10 cycles to 15,000 cycles. 

 Under these operating conditions the floating grid potential was 1.0 

 volt negative with respect to the cathode, the input resistance 1.4 

 X 10^° ohms, the dynamic grid-to-cathode capacitance 6 X 10~^^ 

 farad, and each component of grid current about 4.5 X 10~^^ ampere. 



Discussion of Results 



From the noise data in the preceding tables one can estimate quite 

 accurately the equivalent input noise voltage of each of the four types 

 of tubes at any frequency between 5 and 15,000 cycles, and for any 

 band width within these limits. For example, using the noise data 

 given in Table I the equivalent input noise voltage of the No. 102G 

 tube working over a band having sharp cut-offs at 5 cycles and 205 

 cycles is computed to be 



(P)i/2 = (7//r)i/2 = 2.1 X 10-7 volt. (15) 



For a band width of 200 cycles with mid-frequency at 10,000 cycles 

 this noise is reduced to 1.0 X 10~'' volt. It can be seen that for each 

 type of tube the noise voltage over equal band widths is between 1.5 

 and 4.5 times greater at frequencies below 200 cycles than at the higher 

 frequencies.''^ 



Even at high frequencies the noise voltage is above that expected 

 from thermal noise in the plate circuit which, as stated above, is the 

 absolute minimum to which fluctuation noise in a thermionic vacuum 

 tube may be reduced after all other causes are eliminated. In the 

 case of the No. 102G tubes for instance, using the operating conditions 

 of Table I, and assuming 1100° K. as the temperature of the barium 

 oxide filament, it is found by means of equation (4) that the equivalent 

 input noise voltage produced by thermal agitation in the plate circuit 

 is 2.7 X 10-8 volt for a band width of 200 cycles. The total input 

 noise voltage obtained experimentally at the higher frequencies is 

 greater than this by a factor of 3.8. In like manner it is found that 

 the total input noise voltages found experimentally for the Nos. 264B, 

 262A and 259B tubes are greater than the equivalent input thermal 



^® I am indebted to Dr. J. R. Dunning of Columbia University for pointing out this 

 fact. 



1^ Other investigators have also found an increase in tube noise energy at the 

 lower frequencies. G. F. Metcalf and T. M. Dickinson, Physics, 3, 11 (1932). 



