We are seeing mysterious button presses on one of our iot lightswitches at i3.

Herein lies the debugging so far.

So one set of identical buttons would be triggered by the welder, or by a set of fans turning off.

I swapped the lightswitches, same thing happened, so we thought it was where the switches were plugged in.

Put an osciliscope on the 5v line, and sure enough, there was some fluxuation when the fans turned off.

So we put an APC line conditioner on.

No change.

Okay fine, 5V UPS because we have those lying around. It doesn't fully put the output on the battery though, so no change.

Put a few ferrite beads on the 5v line.

No change.

It was moved to a pile of lead acid, off a nice power supply.

No change.

Finally I just put it on a battery, not powered at all.

It still triggered.

So it's RF magic!

There is a suggestion to put stiffer pullups and caps across the buttons, we shall see how that works out.