Quote from: RobertJasiek on February 13, 2025, 16:31:54The root cause of the problem is the flawed 12 pin standard in all its revisions. In detail, high current is not balanced through different pins but only two pins carry most of the current. With then up to 23A per cable, the cable diameter is too small by far and the plugs are also way too insufficient to prevent or handle the more than 150°C in them, especially since the cables and plugs are not cooled directly. Using up to four 8 pin connectors would solve the issues. Nvidia's server cards have much more robust sockets, plugs and cables so their higher power does not create problems; such would look ugly in desktops so Nvidia and OEMs went for the thin and cause melting, if not fire.
STOP BLAMING ENDUSERS WHEN THE MANUFACTURERS FAIL!
Watch der8auer videos for evidence why it is not user errors but a terrible 12 pin standard terribly applied to RTX 5000!
What we really have here is 12x Molex "MiniFit" series pins & connections.
Each individual pin (as rated) can handle UP TO 9A using 16-28AWG wire. (9Ax6=54A Max)
However, if bad/ineffective crimps are used in the cabling those pins would not be carrying any of the current.