The circuit below is used on both ends of a 10-foot cable. The PCA9615 I.C.s catastrophically and permanently fail, sometimes at both ends of cable simultaneously. Failed I.C.s drag down our power supply and get hot - they die. Replacing the I.C.s fixes the problem.
Failures may occur after hours or days of normal operation, or immediately after a disconnect/reconnect, and possible after (unauthorized) hot-swapping. The cable carries the PCA9615 dual differential bus, ground, and +24V. +5V and +3V are generated from +24V independently at each end of the cable. We are not using the hot-swap logic, but have EN tied to +3.3V. The datasheet includes no Limiting Values identified for hot-swap, and its hot-swap description mentions nothing about protecting the I.C., so I presume the hot-swap logic is intended not to protect the device from damage, but to prevent the device from corrupting data when connected to a live bus.
The failure can be recreated by intentionally hot-swapping under ESD-safe conditions, so failures are probably not likely cause by ESD.
No microcontroller is available to operate the hot-swap EN signal at one end of the cable.
Is hot-swapping without using the hot-swap logic known to cause device failures?
What is the failure mechanism?
The PCA9615 is a nice, inexpensive, compact solution to extending I2C through a cable. We'd like to keep using it, and we want to keep using it if possible.