The Nvidia GeForce RTX 4090’s melted power cable issue continues to cause confusion. As more information is dug up and presented, the big picture isn’t always clear. The RTX 4090 may be one of the best graphics cards out there, but it’s currently in trouble with supply, pricing, and whether the adapter is “safe” for long-term use. It would take Nvidia to issue an official statement about the observed issues, as there seem to be so many variables that even the biggest tech sites and video channels have been unable to come to any useful conclusions.
For example, on Saturday we reported findings from Igor’s lab regarding the root cause of the 16-pin power connector fusing problem. Igor discovered that one of his cables was constructed in such a way that a critical connector would “easily” fail, and suggested crimping rather than soldering the connector joints on the cable. I suggested to his manufacturer.
Igor’s reasoning was basically that soldering is not ideal for connectors that may be subjected to pressure, stress, or strain. can become brittle, bubble, or introduce corrosive compounds into the wiring. Instead, manufacturers must use wires that are inherently flexible and crimped joints that are said to flex easily. However, Igor’s destructive testing of the cable did not show any problems with the socket melting.
And on Sunday, Gamers Nexus (GN) Steve Burke Video published (opens in new tab) About this same topic. GN did more extensive testing, using 5 cables selected from different sources, 3 powerful RTX 4090 graphics cards, running for several hours at stock and overclocked (>600W) settings, melting Tried very hard to reproduce the socket. GN has also invested in new thermal imagers to get to the bottom of the issue. Equipped with 5 cables, 3 GPUs, an infrared camera and probes, the test began.
During the course of our investigation, GN removed the adhesive strain relief tape to expose the metal cable/connector area within the GPU power plug. Carefully done “without hurting them”. Burke recalled that Igor’s investigation found that he was using a 150V cable, but with all five of his adapters that GN had, the used I found that all of the component cables were labeled with a rating of 300V (14AWG, 105C).
The partner GN spoke with said that the 300V cable on the spec, which means Igor’s cable must have been underspec. The specs say that for the power supply he needs 300V 14AWG, but the 4 “sensing wires” could use a lower gauge. There were other differences as well. For example, Igor said that removing the sleeve could “quickly tear” the metal of the underlying thin plate, but no such problems were observed with the GN. Check out the chart below for other key differences between Igor and his GN’s cables.
With more soldering points, a more evenly distributed power feed, and 300V rated wires (labeled), GN tried various “worst case” connector destruction methods, but I was not able to reproduce the overheating issue. The graph below shows that in a continuous stress test of up to 8 hours, the worst damage only increases the temperature around the connector by 8°C.
image 1 of 2
In summary, the GN experienced no failures despite running some of the highest power overclocking settings with very severely damaged cables. There was no melting of cables or sockets, smoke or fire. Note that Igor’s test also showed no melting. He only theorized how it happened.
Survey: 150V wires are less common
Following the revelations about different soldered cables with different component wires (150 or 300V), both GN and Hardware Luxxe reached out to readers on Twitter to evaluate the combinations of cables shipped. According to GN, out of 130 responses, only 7 received adapters using 150V rated cables. This is only 5.4% of all adapters investigated so far. Most of these users seem to have purchased Zotac branded graphics cards.
of Hardware Luxx results (opens in new tab) It was pretty bad overall. 79% of his respondents received a 16-pin adapter with 300V wiring. however, 12VHPWR adapter thread (opens in new tab) So far, there are only 29 responses to the survey. Still, it’s interesting that his percentage of 16AWG adapters on the surface seems to be much higher in Germany than in the US based on the site’s demographics.
If you have an adapter with AWG16 150V wire, don’t panic. So far, there is no correlation between wire gauge and incidence of connector melting. Still, given that there are at least 15 “documented” cases of adapters at the moment, anyone with an RTX 4090 should expect an official statement from Nvidia and its partners about what’s going on. It’s a good idea to check the adapter regularly to see if it’s getting particularly hot…or don’t use the adapter (or card) for the time being.
5 cables from Tom’s Hardware
We’ve tested 5 different RTX 4090 cards, one of which is the Asus RTX 4090 ROG Strix, which currently has a CPU reviewer, so we didn’t check the adapter cable. All came with what appeared to be identical adapter designs, pulled back the tape to see they all use 300V 14AWG wire.
These may be from the same supplier as the samples Gamers Nexus checked, but I haven’t done an in-depth analysis as I would like to continue using the adapters and cards. is Nvidia RTX 4090 Founders Edition, MSI RTX 4090 Suprim Liquid X, Gigabyte RTX 4090 Gaming OC (opens in new tab)When Colorful RTX 4090 Vulcan OC.
