one thing that keeps really puzzling me about these 12VHPWR failures is that we consistently see the +12V lines burned and not the GND lines, but there's just as much current returning back to the PSU.
if the hotspots are contact resistance variance, why aren't we seeing that on GND too? (or at least as often)
if it's a mechanical pressure issue, why isn't it correlating with connector orientation? the +12V pins specifically are burning, both portrait and landscape.
it's *really* weird.
@gsuberland maybe because there's additional ground current paths through the card edge connector, chassis, etc? Even if the *intended* path is mostly the big chonky minifit, there's enough redundancy that some of those lines failing high resistance aren't going to be as catastrophic.
@azonenberg ... I had not considered that, and it's a kinda scary thought.
@gsuberland As someone who's seen a lot of power flowing in places it wasn't supposed to, that was my very first thought lol.
@azonenberg I bet this is it. also how the heck are they getting away with not burning the hell out of the PCIe slot all the time with that being a ground path? O_o
@gsuberland @azonenberg every signal pair is separated by a ground pin. Half the pci-e slot is ground pins, for signaling. Which reminds me, the 5090 has signaling problems.
