590 BELL SYSTEM TECHNICAL JOURNAL 



the same switch voltage as that of Cun-e 2 with a resultant increased residual 

 ionization and a decrease of maximum operating voltage at the higher pulse 

 repetition rates. 



If, instead of doubling the current, the pulse duration be doubled, a 

 similar increase in residual ionization is produced. Curve 4 was obtained by 

 doubling the pulse duration (1.5 ^s) and using a 30 ohm load. Thus, while 

 the current is the same as Curve 2, the current pulse has twice the duration. 

 It will be observed that for these pulses, doubling the time of pulse is the 

 equivalent of doubling the current. 



One might expect that the minimum operating voltage would also decrease 

 as the pulse repetition rate is increased. ?Iowever, experimentally it is 

 found that, for these tubes, the minimum operating voltage is nearly con- 

 stant and, therefore, independent of residual ionization. This result is 

 produced largely because the maximum breakdown voltage of the gaps at 

 the extremely high rate of voltage-time change encountered in triggering at 

 the minimum is little affected by this amount of residual ionization. 



Since the minimum is nearly constant the operating range of voltage of 

 these tubes is a decreasing function of the pulse repetition rate, current, and 

 pulse duration. This is in general true of all fixed spark gaps; however, the 

 amount of decrease of operating range depends on the spark gap spacing, 

 gas atmosphere and geometry of the electrodes. 



(e) Dissipation and Switching Efficiency 



In Il-(d) we considered the voltage-time relationships leading to the 

 simultaneous breakdown of series gaps. In this subsection we will consider 

 the voltage and current relationships with time during this breakdown, and 

 their bearing on spark dissipation and switching efficiency. 



In Fig. 19 (a) are shown a voltage-time and current-time trace obtained 

 oscillographically with a pair of IB 22 gaps. The voltage is measured across 

 both gaps and corresponds to the dotted traces shown for switch voltage in 

 Fig. 17 (a) . The current pulse is shown in proper time relationship with the 

 voltage trace. Similar traces are obtained for any pulse duration and peak 

 current. These, then, may be considered as typical of all pulses produced 

 by spark switching with these gaps. 



In Fig. 19 (b) is plotted the impedance of both gaps with time, from which 

 we see that the impedance of this switch falls rapidly in a small fraction of a 

 microsecond to an average value of only a few ohms while the main current 

 pulse is passing. The tail of the trace showing a negative impedance is due 

 not to the gaps but to inductance inherent in their leads. 



The solid trace, Fig. 19 (c), shows the product of voltage and current in 

 kilowatts plotted against time. The integrated area of this plot corresponds 

 to the dissipation per pulse of both gaps. This area is independent of the 



