Quote Originally Posted by RichardH View Post
Does the picture here include a GFCI on either the grinder's or the welder's outlet? I would have expected them to trip. (I know GFCI detects an imbalance between hot-hot or hot-neutral currents; do they also trip if current is detected on ground?)

Considering how the current smoked the 120v receptacle, it'd be wise to inspect the ground wiring from the 120v receptacle back to the panel's bus, and do the same for the 240v circuit, particularly the ground wires inside the welder.

I'm trying to grasp why the electrical path would prefer the building ground over the heavy ground clamp wire. It'd guess it's because the short was actually on the torch lead, so a path existed from the ground clamp, to the building ground via the grinder, through the case, and to the torch lead. This would have been a much lower resistance path than through a spark gap to the torch electrode, which would explain why it wouldn't strike an arc, yet current was flowing through the grinder as soon as you hit the pedal.
A GFCI on the grinder wouldn't have helped, as they do not break the ground line even if it tripped. A GFCI on the welder circuit might not have tripped either. Hard to say as the balance between hot lines should have still been ok. I'm not really sure as I am not sure of the ground current detection system used in a modern one. Last time I took one apart, they were still just called a GFI and didn't do anything but watch for imbalance between conductors via a small transformer. In theory the welding output is completely isolated from the input, so the GFCI might not even detect it. The ground wire to the grinder was no doubt a terminal bolted tightly to the chassis, and the grinder was bolted to the table so that ground path was pretty good. The point of highest resistance was probably the oxide layer on the plug and outlet, hence where it went poof. Combine that with the ground clamp of the welder being on a dirty part of the table or part, and a layer of oxides here and there, and the resistance could be much higher. You can see why ground loops can do the damnedest things. The torch lead bonded to the case does seem to be the best fit with what happened, though.