578 



BELL SYSTEM TECHNICAL JOURNAL 



anode also behaves in this way and its erosion rate differs but little from the 

 "end sparking" type of tube; in fact, the practical gaps to be described in 

 Il-(f) are essentially of this type. 



With these facts in mind it would appear that gaps could be designed to 

 meet a variety of pulsing conditions if the total number of ampere hours for 

 a pre-assigned life were known, for the electrode areas could be so adjusted 

 that the changes in gap spacing would be as small as required. Analysis of 

 the gradients associated with the end sparking type of gap shows that there 



0.25 X 10 



5 

 O 



D 0.15 



CATHODE 

 LOSS 



CATHODE LOSS 

 FROM SPARKING ARE/ 



TU 



2 3 4 5 



PULSE DURATION IN MICROSECONDS 



Fig. 11 — Cathode loss, in grams per coulomb, as a function of i)ulse duration showing 

 effect of gap design. 



can be a considerable build-up on the anode before there is much change in 

 the maximum gradient which determines the spark-over voltage. 



Experience with gaps designed for a variety of pulsing conditions showed 

 that substantial anode build-ups could be tolerated without interfering with 

 operating conditions, but not as much as theory would predict for an unex- 

 pected factor had a controUing influence on gap life. This factor was the 

 failure of the spark to keep moving under certain conditions with the result 

 that spikes were grown on the anode which introduced a rapid deterioration 



