642 BELL SYSTEM TECHNICAL JOURNAL 



tinguished from the steady rustling noise of the shot and thermal 

 effects. 



Ballantine ^^ has recently made calculations and measurements on 

 the noise due to positive ions from collision ionization in which he 

 has shown that the mean square noise voltage is roughly proportional 

 to the gas pressure within the tube and to the 3/2 power of the plate 

 current. Comparing his results with equation (2), it appears that 

 under ordinary working conditions the noise due to collision ionization 

 in a vacuum tube may be of the same order of magnitude as noise 

 from thermal agitation in its plate circuit. The noise level of tubes 

 having a poor vacuum, however, may be much higher. 



Measurement of Tube Noise 



The performance, as regards freedom from noise, of a vacuum tube 

 used in an amplifier may be indicated by a comparison between the 

 noise and a signal applied to the grid. Usually we say that the noise 

 is equivalent to a signal which gives the same power dissipation in the 

 output measuring instrument as the noise, the frequency of the signal 

 being suitably chosen with respect to the frequency characteristics of 

 the amplifier. Since tube noise is distributed over all frequencies and 

 the noise power increases with the effective band width, it will be 

 advantageous to express this input signal in equivalent mean square 

 volts per unit frequency band width, effective over a given frequency 

 range. 



From these considerations it can be seen that the most convenient 

 standard signal for measuring the equivalent input noise over any 

 given frequency range is one in which the mean square signal voltage 

 is distributed equally over all frequencies. With such a signal the 

 equivalent input noise over any frequency range can be measured 

 directly, while if an oscillator is used a number of measurements are 

 required and the result must be computed by graphical integration. 

 A signal which meets these frequency requirements perfectly is the 

 noise of thermal agitation. Accordingly, in the measurements to be 

 described here the standard input signal will be the thermal agitation 

 voltage of a resistance R, connected between the control grid and 

 cathode of the tube under test.^ 



The thermal noise voltage of the grid circuit, referred to the output 

 measuring device, is given by equation (1), where R{f) is the real 

 resistance component of an input impedance consisting of the pure 

 resistance R in parallel with its shunt capacity and that of its leads 



10 Stuart Ballantine, Physics, 4, 294 (1933). 



