Nvidia's RTX 50 series had a rocky launch, and the latest concerns about the RTX 5090's power connector issues further add to the turmoil.
The company recently supplied RTX 5080 labeled incorrectly with a 5090 shroud, creating quality control issues. According to recent reports, even with the revised 12V-2x6 standard, the RTX 5090's 16-pin connectors still melt under certain conditions.
The issue dates back to the RTX 40 series, when Nvidia introduced the 16-pin 12VHPWR PCIe power connector. Early models experienced substantial failures as a result of faulty cable insertion or overbinding. In response, manufacturers introduced L-shaped connectors and brightly colored plugs to prevent these errors. The updated 12V-2x6 standard was developed to address these concerns, ensuring power was only supplied when the connector was fully inserted.
Despite these advancements, some RTX 5090 Founder Edition users have reported burning and melting power connectors. Tech expert Der8auer's analysis suggests that the problem could be caused by unbalanced current distribution across the pins. During his tests, he discovered that some pins managed a low current load, while others reached dangerously high levels (up to 22 amperes), resulting in severe heat accumulation. His thermal imaging revealed socket temperatures rising above 140°C in minutes.
However, conflicting reports complicate the situation. Andreas Schilling, senior editor of Hardwareluxx, didn't see any such current in balances, while Falcon Northwest, a high-performance PC builder, tested many RTX 5090 FE devices without encountering the overheating issue. Some also believe that repeated plugging and unplugging of cables may contribute to the problem, potentially weakening the connector's performance over time.
Interestingly, Nvidia tested engineering samples with up to four 16-pin connectors, indicating worries about power delivery. However, the final design used a single 16-pin port, which now appears to be a possible weak link. Some AIB designs have per-pin power to mitigate these risks, although Nvidia's FE is said to lack this functionality.
For the time being, users must proceed with caution, ensuring cables are properly installed and keeping an eye out for excessive heat accumulation. Nvidia has yet to release a formal response; without considering the high stakes of a $2,000 flagship GPU, an answer is immediately required.