The chip was well beyond depression, the PCB did not like the role of rented mule it wound up in and committed suicide.I don't think the fets were sagging, and most of the TIM print shots show 'good enough' contact and better with tape.
I went through and checked all the cards for contact on the VRM's before running load tests, that with a generous dollop of TIM and the height variation did not play much if any of a role in the GPU seppuku... at least I am not pointing it to that. There had been physical stress on the cap for those fets from the majority of blocks (notice the hot glue for added durability), add that into the rigors of thermal testing and that area was just prime for failure. GPU's are much more 'delicate', and separating the card from the block is a violent process. I can't recommended lots of remounts for a GPU block.
![]()
![]()
But yeah, losses happen, I'm just glad all of the important tests were complete when it happened.
Bookmarks