Last month we reported that some users started reporting that they had the power connector NVIDIA GeForce RTX 4090 it ended up burnt and/or melted due to excessive heat. At first, the origin seemed to be the incorrect use of an adapter cable that allows you to get the 16-pin connector used by the aforementioned chart and other models of the green giant, but circumstances have changed over the weeks.

As we said, at first the problem seemed to come from the incorrect use of the adapter cable, which, frankly, needs to be handled with much more care than a typical eight-pin or six-plus-two connector than the usual one. It is found in the computers we have in our homes. The beginnings then moved to the specifications of the adapter cable, because 150 watts seemed unreliable and 300 watts unreliable, but then RTX 4090 units with a burnt power connector started to appear after using a 16-pin “native” power supply, without an adapter.

In order to determine the true origin of the burnt power connectors, the Gamers Nexus YouTube channel conducted and published an in-depth investigation to conclude that the main reason is incorrect positioning by the useralthough he also pointed out the remote possibility that there is “foreign object debris” in the connector and that the connector itself is unreliable, at least from a lay user’s perspective.

The glitch that Gamer Nexus was able to recreate in their labs involved plugging the 12VHPWR power connector into the RTX 4090 without fully inserting it, then bending the cable down to make things worse on purpose. After several minutes of testing, the high-power resistor in the connection generated heat in excess of 250 degrees Celsius, resulting in smoke and visible bubbles as the connector melted in the socket.

Although he shifted some of the blame to the manipulation of the connectors, a Nexus player described the problem according to his version as “a combination of user error and design oversight”, with the origin being the fact that the connector was not inserted all the way. The absence of a “click” mechanism or a more robust and easy-to-verify retention also contributes to this, which increases the chance that the 12VHPWR connector was not correctly positioned. This is in contrast to the 6-pin and 8-pin connectors, which have a distinctive latch that ensures proper placement, plus the cables, at least ostensibly, are more resistant to twisting.

Another thing that doesn’t seem to help is the size of the RTX 4090 and RTX 4080 cards, which is exaggerated in many cases, so in many cases they fit very tightly, which of course affects the connector that is in contact with the side cover that allows access to the top of the motherboard. When the cover is closed, there is a good chance that the cover will touch the graphics card connector and loosen or dislodge it, causing the problem described above.

Previous inquiries have suggested that depending on the manufacturer, the connector may have a higher chance of burning/melting, but research by Gamer Nexus has concluded that this factor is not particularly relevant. The YouTube channel’s findings leave the door open that the same could happen with the RTX 4080, so we should have information on that in the coming weeks.

NVIDIA is also investigating to uncover the origin of the failures, seeing more and more that it points more and more to the connector and less to some adapters that many already question.