CONTACT PHENOMENA IN TELEPHONE SWITCHING CIRCUITS 61 



which softens a tiny point on the contact which is pulled out and 

 fused into metallic contact under the action of the high fields. After 

 rupture by increasing separation or increasing current density, the 

 process may repeat or, as is more likely, the separation is too great 

 for another metallic bridge to form. The high field discharge then 

 sets the stage for the next type of conduction or breakdown. This 

 may be either a series of sparkovers interspersed with metallic arcs of 

 extremely short duration or a gas glow discharge, initially intermittent 

 and then more or less stable. Factors predisposing toward one or the 

 other type of discharge are known thus far only in a most general 

 fashion and much remains to be done before the relation between 

 contact erosion and the transient currents and voltages can be pre- 

 dicted accurately. There is ample evidence that molten metal may 

 be expelled from the immediate contact area at high velocity and 

 may be deposited at distances of at least 0.1 inch. It also appears 

 that both the ionized nitrogen cloud of the "A" transient and the 

 disruptive sparks of the "B" transient may corrode the contacts and 

 their supports at locations and distances which never enter directly 

 into the rupture of the current path. 



We have seen that the line wire contributes to the current surges 

 through contacts due to its properties as an oscillatory circuit, charged 

 repeatedly by the energy stored in the magnetic field of the relay. 

 The surges and the resultant erosion may be reduced in several ways. 

 If a radio frequency choke coil is connected between the contact and 

 the line wire, the discharges of the latter are much reduced, and the 

 "A" type gas glow transient favored. A group of many current 

 surges of 15 amperes peak may in most cases be reduced to one or two 

 of 0.15 ampere or less, and a radical reduction in erosion secured. 

 Unfortunately choke coils are expensive and inconvenient. The usual 

 line wire may be terminated in approximately its surge impedance by 

 shunting both ends to ground with a resistance of about 100 ohms in 

 series with a condenser of the order of 0.01 mf. This heavily damps 

 the line oscillations and greatly reduces the number and severity of 

 the current surges. It is also expensive. Instead of the copper line 

 wire, a material such as iron or permalloy plated copper having a high 

 surge impedance and large high frequency a-c. losses may be used. 

 This seems more practical, but brings up new problems in design, 

 handling, and soldering. 



The most effective means of reducing erosion is of course the well 

 known "spark-killer" (consisting of a condenser and resistance in 

 series, shunted across the contact or load), which can be designed to 

 hold the voltage below the sparkover point at least until the contacts 

 have separated a safe distance. 



