Due to the latest recall of CableMod adapters, cases of melted 12VHPWR plugs seem to be on the rise again. The plugs have now been in use for a year -> I think they are starting to corrode. At these high currents, even a minimal deterioration of the plug connection (fraction of an ohm) is enough to fry the plug. What you normally do to prevent corrosion: remove the plug and reinsert it to "renew" the connection. But you shouldn't do this too often, igorsLAB once did a connector material analysis and the coating on the pins is so thin that after "a few times out & in" the coating is already off and then the bare copper appears, which corrodes even more. A vicious circle.
"Optimal solution approach" for the problem:
No real solution, but the graphics card driver should be able to detect when the connector starts to melt: By a drop in voltage at the graphics card connector. Tools such as HWiNFO can measure the input voltage of the graphics card at the 12VHPWR connector, so the graphics card driver can certainly do the same. On my TUF 4090, two different voltages are listed, in extreme cases there could be up to six (12 pins), but I assume that several pins are connected in parallel again after input. So let's assume that 450 watts are consumed, i.e. 225 watts per input voltage for my 4090. A tenth of this is probably enough to melt the plug, i.e. 22.5 watts. However, this would mean that the graphics card would only receive 10.8V because the bad contact is a series connection. In addition, this supply voltage would even be outside the ATX specification. If this happens, the graphics driver should immediately stop the power-hungry graphics calculation and display a green screen (NVIDIA!) with a message such as: "Your 12VHPWR connector is probably melting! Please turn off the PC immediately and check the connection: can the plug still be disconnected? BINGO, if not: OUPS".
"Realistic (faster) solution" for the problem:
HWiNFO monitors the input voltages of the graphics card and displays a flashing warning message if one of the voltages drops or the input voltages deviate noticeably from each other! Hopefully there will still be enough time before gaming is stopped to save the plug ...
Greetings to all and Merry Christmas
"Optimal solution approach" for the problem:
No real solution, but the graphics card driver should be able to detect when the connector starts to melt: By a drop in voltage at the graphics card connector. Tools such as HWiNFO can measure the input voltage of the graphics card at the 12VHPWR connector, so the graphics card driver can certainly do the same. On my TUF 4090, two different voltages are listed, in extreme cases there could be up to six (12 pins), but I assume that several pins are connected in parallel again after input. So let's assume that 450 watts are consumed, i.e. 225 watts per input voltage for my 4090. A tenth of this is probably enough to melt the plug, i.e. 22.5 watts. However, this would mean that the graphics card would only receive 10.8V because the bad contact is a series connection. In addition, this supply voltage would even be outside the ATX specification. If this happens, the graphics driver should immediately stop the power-hungry graphics calculation and display a green screen (NVIDIA!) with a message such as: "Your 12VHPWR connector is probably melting! Please turn off the PC immediately and check the connection: can the plug still be disconnected? BINGO, if not: OUPS".
"Realistic (faster) solution" for the problem:
HWiNFO monitors the input voltages of the graphics card and displays a flashing warning message if one of the voltages drops or the input voltages deviate noticeably from each other! Hopefully there will still be enough time before gaming is stopped to save the plug ...
Greetings to all and Merry Christmas