Well, I think I see what happened.
Removing the other long screws I found that all of them were in contact with the ground plane due to the solder mask being worn off. The short screws which had their nuts on this side also had the same issue. Almost every hole the solder mask wore at the edges allowing the screws to be part of the ground plane. A V3 PDB with a larger gap between the hole and the copper pour could eliminate this entirely:
With the top plate removed things got ugly quick:
A quick check shoed that the electrical tape did not fail though and the power leads were not in contact with the frame...so that's good:
The PDB however was ugly:
So much is burned it's hard to tell just what happened:
Can't tell from the photos - but the insulation on the power wires is all intact...discolored from the nearby heat...but intact:
The arm is damaged though which gives a good indication that it was part of the circuit/problem.
Looking at one of the other arms I believe the story comes togeter...look closely:
I'll move the arm aside so it's more obvious:
I think there were two things that contributed to this failure - and they can both be avoided with some extra precautions on the V2 board and eliminated entirely on a V3 board.
Both issues are due to the solder mask failing.
The first is the ground pour coming too close to the holes allowing the nuts/heads to wear the solder mask when tightened. The exposed copper was not visible until I removed the screws so it wasn't from my wrench it was from the screws/nuts themselves. Maybe I over-tightened them a touch, but....
The second issue is that it appears the arms can move just enough to wear through the solder mask on the positive side as well. I did sand the edges of the arms even here (the slight bevel is visible in some of the photos) but I didn't get these little tiny bits past the screw holes very well. I'm going to go back and touch them up to help prevent this from happening again. Even so the arms can apparently move just enough (I can't even budge them by hand...but in a crash...) that after a few hard crashes they may compromise the solder mask and become part of the power plane.
When the arms became part of the power plane and the screw was part of the ground...a short happened - which explains why I heard my beeper go off (even though upon later inspection the pack wasn't low enough for the beeper to fully activate like it did) and saw a small puff of smoke. That was likely the screw melting through I saw.
E-mail sent to twitchity to order a replacement arm and a new screw. I don't want to speak for him, but I suspect he'll do a v2.5 or v3 of the PDB and widen the gap around the holes on the ground plane and/or remove some of the pour from the positive side around the arms. It's a fairly easy fix and as far as I know he doesn't have a bunch of PDB's already made so a modification should be easy to do before ordering more.
Anyone currently flying a V2 board I'd suggest taking a few precautions. Making sure you've sanded the edges of the arms where they extend in the frame is the first thing. Second I'd add a wrap of electrical tape around the arms where they extend - at least on the bottom where they contact the PDB. Third I'd add some kind of insulating washer under the nuts/screw heads on the ground side. I'm not sure if nylon washers would be a good idea though - I think they'd weaken the frame by preventing the screws from being fully tightened. I'm thinking maybe some mylar sheet or even bits of posterboard. It needs to be something strong enough to handle the torque of the screws being tightened without tearing...but still non-conductive. I'll have to do a bit of brainstorming to come up with something.
If you were careful about sanding ALL of the arm edges, and you fly sedately...this probably isn't a big concern. I had quite a few hard crashes today (the one in the video above was the last and one of the worst) and was flying well outside my comfort range for most of the time I was out there.
I'll also add that I've been thinking about doing a bulk buy on FC props with my co-worker who's getting the G10 frame. But after today I'm really starting to second guess that. They're a lot cheaper than the HQ's...but I've been going through them really quick. Even mildly rough landings I've been breaking them or at least getting "white roots" on them. I've been averaging 2-3 FC props a day flying a pack or two at lunch in the backyard. The FC's I had on today lasted about 45 seconds...one good loss of orientation out at a distance and 3 of them snapped at the root when the ground and my quad got acquainted. I swapped to the HQ's and despite several harder crashes I only managed to take out one prop....even that last crash in the video above didn't even result in any white spots on the HQ's. I'm still not 100% sure the HQ's are fully worth their extra cost...but they do hold up a LOT better than the FC's it seems. They cost about 3x more...I'm not sure they hold up 3x better...but they do hold up quite a bit better and I'm probably going to try and stick with them going forward. At least as long as my budget allows. I'll probably still pick up some FC's for testing and bashing...but the HQ's really are far better props.