One of the major unknowns in the entire problem related to burned RTX 40s and the cables is the power supply unit (PSU). Although PSUs have always been a subject of concern, they usually didn’t catch fire themselves. Instead, the integrated connector would melt down, similar to the graphics cards. But how does a PSU affect the connector and its temperature? Hwcooling’s investigation sheds light on this fiery triangle between the power supply, the 12VHPWR cable, temperature, and GPU.
There are a few things to consider here. Firstly, the GPU is suppressed by a Chroma 63106A load module to establish stability in consumption and amperes. This unit doesn’t have active cooling or indirect airflow, causing temperatures to be slightly higher than a normal GPU inside a tower.
Regarding the power supply and 12VHPWR cables, the model can indeed affect temperature. According to PCI-SIG, 12VHPWR cables must support up to 55 amperes, and from that point on, it depends on the PSU’s capacity. Tests were conducted using a FLIR camera at 600 seconds of load with a controlled ambient temperature of 19°C (66.2°F). Twelve PSU models from different manufacturers were tested, with power capacities ranging from 850W to 1000W.
The temperature differences found raise many questions about certain models and manufacturers. Two main metrics were analyzed: temperature in the connector housing and temperature in the cable insulation. From these measurements, maximum and average temperatures were calculated for each.
The results show that MSI and ASUS models reach over 80°C (176°F) at maximum temperature, which is unacceptable. This happens because some of their cables meet the specified AWG 18 standard instead of the expected AWG 16. Another issue with the ASUS and MSI models is that their pins have larger gaps, which reduces contact and increases transient resistance, ultimately causing the connector to overheat.
Choosing a power supply is crucial for a GPU and the cable’s temperature, as temperature disparities can be significant. In some cases, the worst-performing models have temperatures up to 52% higher than the best ones. This is an alarmingly huge difference.
In conclusion, the power supply and cable play an essential role in preventing RTX 40s from burning out. Better-made models and higher-quality cables are advantageous when considering the purchase of an RTX 40.