My design uses SN65220DBV for ESD protection on a input bus that is meant to carry 3.3V logic signals. SN65220 device GND pins (2 and 5) are connected to product chassis GND and device A and B pins (6 and 4) are each connected to a signal. The intent is that each signal will be independently clipped between 0V and 3-5V. On a unit which was returned from the field I found that the pin4-to-GND path on one device had failed short and always had a resistance of approx. 15 ohms. The fault seems to be bipolar, by which I mean I measured the same small resistance which way I attached the leads of a DMM in resistance mode and the same small Vf no matter which way I attached the leads of a DMM in diode mode.
The faulted SN65220 was shorting out an input channel that had passed test before shipping the unit, so the damage must have occurred in the field. I am trying to determine what sort of event may have damaged it. As a transient supression device, it seems unlikely that is was itself damaged by an ESD event. Many transient devices can fail if DC voltage is applied to them which is beyond their clipping voltages, so I tried emulating this. On the surviving channel of the part (pin6-to-GND) I intentionally applied DC voltages below GND and above 6V. As expected this caused damage to the part. However, the negative DC voltage seemed to create a fault that was unidirectional, and the positive overvoltage seems to have blown the channel completely open. Neither conditions really replicated the bidirectional 15ohm short seen on the channel that was damaged in the field. Do you have any ideas on possible root causes of a persistent, bidirectional short?