ELECTRICAL NOISE IN SEMICONDUCTORS 



953 



these are typical of the variations encountered. There is a general ten- 

 dency for noise current to increase in proportion to bias current, l)ut in 

 limited regions the individual units may have slopes considerably dif- 

 ferent from unity. It would perhaps be more logical to plot cui'rent 

 densities rather than total cm-rents, but because of the general form of 

 the relations this makes little difference in the overall picture, und 

 there is some difficulty in estimating the appropriate area for the point 

 contact units. There is an almost unlimited number of different ways 

 of representing noise data. For example, noise current, current density, 

 voltage, or available power may be expressed as a function of various 

 bias parameters. Of a good many combinations tried, none gave an 

 outstandingly simple picture of noise behavior, and the representation 

 used in Fig. 2 is probably as good as any for an overall picture of diode 

 noise. 



The noise behavior of transistors depends on two bias parameters. 

 Selection of the emitter cm-rent and collector voltage for the parameters 

 usually leads to a rather simple representation. It often turns out that 

 the noise behavior as an amplifier over the commonly used range of bias 

 ^•alues depends largely on the collector voltage and is relatively inde- 

 jjcndent of the emitter bias. Data of this sort were shown for point 

 contact transistors in a previous reference, and have been given for 

 an n-p-n transistor by Wallace and Pietenpol. A somewhat more com- 

 plete family of curves is shown in Fig. 3 for a recent n-p-n transistor. 



A few attempts have been made to determine the effect of tempera - 



a 20 



0.2 



20 



40 



60 



0.6 0.8 1.0 2 4 6 8 10 



COLLECTOR BIAS IN VOLTS 



Fig. 3 — The noise figure of an 7i-p-n transistor depends in a fairly simple way 

 on emitter current and collector voltage. 



