I have an odd problem with ADum2401 isolators.
I am looking into a problem where a spurious output pulse is being pruduced, when the device is subjected to noise produced by an arcing event.
WHen the noise is produced the outputs give a high pulse of about 1us in duration. This altering the contents of a shift register, and causing misoperation. The arc is produced by disconnection of an AC signal at approx 400kHz and 340V.
The PCB is quite well decoupled, but the 2401 is in a separate potted box on a 2-layer PCB. The decoupling is more or less in conformance to the data sheet - there is 1 off 100nf.
Nothing I do in filtering the supplies, or inputs/outputs makes any difference.