1248 THE BELL SYSTEM TECHNICAL JOURNAL, NOVEMBER 1952 



and substantially so, there are normally no serious consequences due to 

 the very rapid transitions and consequent low thermal energy generated. 



Transistors may not be able to tolerate excess dissipation on this 

 basis if the circuits are slow, that is with transition times in excess of 

 perhaps a few tenths of a microsecond. Such conditions may arise, for 

 example, if loads are inductive. In many such cases, shunting capacitor 

 networks will often permit a rapid transition with consequent transfer 

 of current to the inductive load. 



The frequency response of point contact transistors can be sufficiently 

 good to insure switching type operation with rise times of the order of 

 0.1 to 0.01 us. Fall times may be somewhat longer due to the hole storage 

 effect. In regenerative circuits, operating speeds are faster than might be 

 imagined from the small-signal frequency cutoff. Reliable operation with 

 rise times of 0.1 ixs is obtained with only nominal attention to frequency 

 cutoff. Speeds of the order of 0.02 jus require a 10 mc. lower limit. Present 

 junction transistors are substantially slower. 



Accurate life estimates are difficult to make due to the rapid rate of 

 development, the relative age of the transistor and the number of 

 parameters involved. A given device is quite likely to be obsolete and 

 forced to give way to an improved version before sufficient models can 

 be obtained for life tests. A small quantity of transistors having proper- 

 ties similar to those of Fig. 20 and 21 have been operated for over 6,000 

 hours with an indicated life of 30,000 hours. Other similar transistors 

 with longer life histories have indicated lives of better than 70,000 

 hours. The pattern appears to be similar to that of electron tubes — an 

 early failure and change rate followed by a very slow exponential rate. 

 It is believed that life is extended by low power operation and is de- 

 creased by high temperature operation. 



The relatively high noise level of transistors does not appear to be a 

 significant problem at present when considered in terms of automata. 

 Systems employing switching type circuits in pulse communication will 

 of course be concerned. It is suggested that the non-concern for noise 

 in non-transmission type systems is largely a reflection of the ease with 

 which high magnitudes of state changes are obtained. With design 

 trends toward low power and low operating levels, noise will undoubtedly 

 set a lower limit of level operation in such systems also. 



The extreme resistance of the transistor to shock and vibration with 

 a consequent absence of microphonism may in some applications result 

 in effective lower noise. Shocks in excess of 20,000G have resulted in no 

 damage. No evidences of current modulation in excess of noise have 

 been detected with vibrational forces of the order of lOOG at frequencies 



