212 BELL SYSTEM TECHNICAL JOURNAL 



of temperature depending upon the conditions of the experiment, i^ut 

 that the total change is not more than ?> or 4 db in the temperature 

 range from 90 to 300 degrees Kelvin. These observations are con- 

 sistent with the fact that Otto ^ has reported a decrease of noise with 

 increase of temperature while Meyer and Thiede " have reported the 

 reverse effect. 



The nature of the surrounding medium seems to affect the intensity 

 of the noise but little. The noise of the contact under oil seems to be 

 slightly less, one or two decibels, than when the contact is in a vacuum 

 of 10~^ mm. of mercury. The noise in air seems to be intermediate 

 between these two extremes. This leads us to believe that the noise 

 mechanism is not associated with the medium surrounding the contact. 



QUANTIT.\TIVE VALUES OF CONTACT NOISE 



Equations (1), (2) and (4) may be combined to give the expression 



T} ^ K V"R^ log (F2/F1). (5) 



This is the general empirical equation found for noise in granular 

 resistance elements as a function of voltage, resistance, and frequency. 

 The average values for a and /3 are respectively 1.85 and 1.25. The 

 constant K is dependent on the material, shape, temperature, etc. of 

 the resistance element. The following representative values of this 

 constant were obtained for some of the resistance elements which we 

 have measured : 



Single carbon contact 1 .2 X 10"'" 



Western Electric No. 395-B telephone transmitter 1.3 X IQ-i' 



100,000 ohm carbon grid leak 1.1 X IO-21 



For different single carbon contacts this constant did not vary more 

 than 20 per cent as long as a given type of carbon was used, and a 

 change in the type of microphonic carbon produced a variation by not 

 more than a factor of two. 



The contact noise in a Western Kiectric No. 395-B telephone trans- 

 mitter under actual working conditions (R = 45 ohms, T^ = 2.5 volts. 

 Fx = 200 c.p.s. and F2 = 3000 c.p.s.) is given by Eq. (5) as 9.8 X 10"* 

 volts. The output signal of this transmitter for standard voice opera- 

 tion is about 0.1 volt. The spread between signal and contact noise 

 is so great that this noise is not a disturbing factor in the standard 

 carbon transmitter as used in telephone service. This is not true, 

 however, in the case of high quality carbon transmitters used for studio 

 work and public address systems. The sound fields under the condi- 

 tions wherein such instruments are apt to be used are much less intense 



