Except all attempts at reproducing the problem so far have failed to nail it to any of the probable causes currently known or even combinations of two. Since no single cause or even combination of two causes could duplicate the result so far, either an adapter with a combination of 3+ flaws is required, there is some other factor yet to be identified or possibly as Curmudgeon put it, some if not most of the reports may be hoaxes.
Or like Steve from GN put it, there could have been running design changes during manufacturing and nobody was notified of them.
Copper melts at 5 times that temp (1085 C) , so yes low. PVC is between 100 and 500 depending on modifiers and if you look at the failures I would say that "smoldering mess" is pretty correct here with cables burnt half of an inch back. Paralleling the load on that many conductors is just asking for a single point failure overloading the whole system. From what I have gathered bending the connection seems to be a big contributor. Do I think that 33 amps on a 16 ga marginal connector will melt solder (and plastic insulation too), ohhh yea. Crimps, while tinned are made of copper and shouldn't be any more failure prone due to heat than the conductor itself.
Not really: each pin on the PCIe connector can take about 1A and there are about 70 total ground pins on there. You could cut off all cable grounds and possibly still might be fine. The IO bracket is connected to ground too, so another 10-20A could go through that too. Nobody is talking about the ground pins from a plug-melting perspective because there are so many alternate paths available for DC ground return. The only thing losing cable grounds might affect is the PSU's transient response due to increased wiring stray inductance - fast signals (transients) follow the path of least inductance.
The wiring insulation is only rated for 105C, which means whatever you do, you cannot allow the temperature to get any hotter than that and that makes the 235C melting temperature of lead-free solder absolutely moot.
Oh come on. 105c is the maximum field rated temp of the insulator in operation, not when it melts. That means that you can actually run it at 105c. Not that anyone in their right mind would. Beyond that point what does the smoke point of the PVC have anything to do with this case as it "is" clearly melted/burnt completely off of the wire one half inch + back on some of the conductors with oxidized copper showing? Which by the way copper oxidation happens at a temperature range of 200c -900c in normal atmospheric gasses. Prelim thermal imagery of the outside of the insulation and connector is in excess of 150C. That says nothing about conductor temp inside. Furthermore I am absolutely sure that nothing with this stuff is being "allowed" or happening within design specs. These things are burning themselves off of the card. All that I said is that a solder joint melts and fails at a very low temperature in contrast with a copper crimp connector. Which I still stand by. My personal preference on a high amp larger gauge connection is a crimp connector with dielectric paste regardless. It handles flexing better and you don't have cold joint/fluxing issues when you make it up. On the other hand I would be absolutely lying to say that I know what is causing this failure. All I can see is that the several of the pins on the connector overheated, melted the insulation, zapped a couple of pins and ultimately failed. It also looks like soldered conductors were coming off of the connector in some cases which either indicates mechanically cracking or melting and release of the joint.
It may not melt but prolonged exposure will still accelerate its degradation, release volatile compounds and become brittle given enough time. That is why places where wiring is expected to get exposed to temperatures above 90C on a regular basis usually replace the insulation with fiberglass sleeves or use silicone instead of PVC.
Most PC buyers don't want to spend $50 per connector and in-cabinet wiring doesn't need gorilla-proof connectors since the 500lbs gorillas hardly ever come inside the housing. It would be a grossly undue burden for 99.9% of people.
Doesn't matter what a top-end GPU costs. the HPWR connector adopted in the ATX 3.0 spec is intended to eventually replace all 6/8-pins cables, which would eventually include $150 GPUs that are just over the PCIe slot's power budget too.
Facebook
Twitter
Instagram