682 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1956 



maximum current of one millianipere. None of these units show any ap- 

 preciable change. 



7.3 All of those tests in the past sub-section had to do with continu- 

 ous dc or ac power being supplied to the units under test. However, in 

 actual operation the units may be subject to voltage pulses due to 

 power line pulses, accidental shorts, etc. In order for the rectifier to be 

 useful, it should be able to take an overload for a period of time suffi- 

 ciently long to allow a protective device to operate. Pulse tests have been 

 performed on the medium size rectifier. These devices are able to with- 

 stand over 300 amperes for times of the order of 50 microseconds. How- 

 ever, the fastest circuit breakers operate in about 20 milliseconds and 

 for this period, these units can stand onl}^ approximately 50 amperes 

 before failing. Since these units have such a low forward resistance at 

 the operating currents (Fig. 7), any small increase in voltage across the 

 diode will change the current through the device to a very large cjuan- 

 tity. Therefore series protective resistances may be necessary where 

 the possibility of short-circuiting the device is high. Such operation 

 would reduce the efficiency of the unit and is to be avoided if possible. 

 Another type of protection may be afforded through the use of a high 

 impedance, high current inductor. This type of protection is quite bulky 

 and heavy and suitable only for stationary apparatus. Another common 

 possibility of burnout of the devices occvu's when using a capacitance 

 input in conjunction with the rectifier. When the circuit is turned on, 

 large currents will flow to charge up the capacitors and consequently 

 burn out the rectifiers. One possible protection from such operation is 

 the use of a series resistance in conjunction with a time delay relay. The| 

 series resistance will limit the initial capacitor charging current and the 

 time delay relay will short out the resistance after the capacitors have 

 reached near their maximum charge. 



7.4 Dissection of burned out units have indicated that the failure 

 takes place through small spots on the device. This can be explained by 

 the fact that some small areas of the device have slightly better forward 

 characteristics. These areas will tend to conduct most of the forward 

 current. Therefore most of the power will be dissipated there and these 

 areas will become even more conducting leading to a channeling of the 

 forward current through these spots with the consequent burnout. The 

 best way to avoid such mishaps would be to make a more uniform de- 

 vice. Experiments are in process along this line. Another less satisfactory - 

 method would be the control of contact resistance such that the current 

 would be limited in any particular area by the contact resistance. Simi- 

 lar ideas must be considered when paralleling these diffused junctioiii 



