wnjj wrote:
That’s some impressive data there. I believe the answer to your 7.9mA question is likely number 2. The Onsemi device lists a typical delay of 2ms of fault current before tripping. That equates to a half period of 250Hz.
I agree with this assessment. As I understand things, GFCIs are often designed to not respond to high frequencies precisely to avoid false trips due to noise from switching power supplies or motors starting up.
It's also fairly likely that power on the chassis seen from the converter is at least partly coming out of its output, rather than leaking from its input side. That wouldn't cause the GFCI to trip at all, of course, as it's a "normal" load so far as the GFCI can tell rather than leakage since the input and output sides are galvanically isolated from each other (aside from sharing a common chassis ground).