200 BELL SYSTEM TECHNICAL JOURNAL 



made by adjusting the attenuation so as to bring the deflection of the 

 microammeter as near mid-scale as possible. Fractions of a db were 

 estimated by means of the deviation from the standard mid-scale 

 reading. Thus what was measured in each case is the insertion loss 

 necessary to produce a standard electrical output. 



Noise as a Function of Applied d.-c. Voltage 



The contact noise in several different types of granular resistance 

 elements was measured as a function of the applied d.-c. voltage, all 

 other variables such as resistance, frequency range, temperature, etc., 

 being held constant. The first of these measurements to be described 

 is that obtained by using a standard handset telephone transmitter. 

 The circuit used for coupling to the high-gain amplifier is shown in the 

 insert of Fig. 2, the essential parts being an input transformer having a 

 high-turns ratio, a d.-c. voltage supply, and a standard a.-c. signal 

 generator. The resistance of the carbon transmitter was about 

 50 ohms. 



The results of the measurement are shown in Fig. 2 where mean 

 square contact noise voltage is plotted as ordinate and the d.-c. voltage 

 directly across the transmitter is plotted as abscissa, the scale being 

 logarithmic in each case. Measurements were made as the transmitter 

 voltage was varied from 0.00145 to 4.5 volts. This is the greatest 

 possible voltage range in which contact noise can be observed in this 

 instrument since the contact noise is masked at the higher voltages by 

 carbon burning and at the lower voltages by the thermal noise of the 

 transmitter resistance. Thus the total noise at 0.00145 volt is only 

 slightly above thermal noise and the measured value consists of thermal 

 plus contact noise. The two effects have been calculated separately 

 and the latter plotted as a cross. Using this method of plotting it is 

 seen that there is a straight line relationship between contact noise and 

 voltage over the entire lower range. These experimental data can be 

 accurately represented by the equation 



V} = Const. F«, (1) 



where V^ is the mean square contact noise voltage, V is the d.-c. 

 voltage across the transmitter, and a is a numerical constant having in 

 this case the value 1.85. 



By this procedure the contact noise in a number of types of carbon 

 transmitters, filled with carbons of various origins, was measured as a 

 function of voltage. In each case the relationship given by Eq. (1) 

 was followed very closely over a wide range of voltages. The value of 



