[SOLVED] 2080 burnt connector pins?! What do i do now?

[jack4404]

Hi everyone, I've got a 2080 and 3700x on a hardline loop. Have had it for a while now. I had to replace a faulty power supply (Silver Stone strider gold 850w), because it would just power off my pc almost immediately when turning on. Long story short, i did buy it new, but cannot claim warranty. So i replaced it with a new silver stone et700 80plus gold, and it fixed the problem immediately.
I mention this because other forums about burnt gpu connectors blame the psu. I dont even bother putting a permanent overclock on, its always stock, except for the last 3 days when i started mining on my pc. Years ago ive done the same thing. My memory clock was on +1200 on afterburner (which didnt display artifacting when testing and ran fine for a little while). After a night of mining the pc would stop displaying anything (but still be on completely fine) i quickly realised the gpu is at fault. So i reset cmos and booted and then put the clock at 1000+ and it was fine over night mining. But then when i used it this morning, it showed no display again. I then disconnected the connector and saw the 8 pin was broken? The extra two pin isnt photed (https://ibb.co/CHtxZxh, on the right), luckily i have spare pcie extensions, soon after i saw 3 of the pins were burnt (https://ibb.co/jkrqvTS, https://ibb.co/nw4W77P). At the moment I've used a clean tooth brush and lightly brushed inside the connector with isopropyl alcohol. Although it doesn't look like its done too much. Ive drained my loop now to have a better look at the gpu out of the loop and i thought well, what do i do now?
It isn't really my power supply is it? Have i been that unlucky with two faulty psu's?! If it is my psu, am i even able to claim warranty, and what about my 2080?
Also, the temperatures are still really good, cpu at 63c under load and gpu at 51c under load.
Thank you for your help!

 

[vov4ik_il]

It looks like your extension cables had a poor contact to the card aux power connectors and that caused voltage drop and heating which melted the plastic pin housing. It could have damaged the card too as it would imply asymmetric current through other lines and should have triggered protection if(!) present.
If the card has a warranty, it might be honored or it might be not. Hard to tell. Totally depends on the tech mood when writing the report. Whatever the case is, do not tamper with the card before you send it in as it dramatically reduces the chance of getting your warranty honored.

As for the PSU, do not use high current components with extension and splitting cables, it is a bad idea in general. If you have other connectors, use those and you should be good to go.

P.S. When you deal with overclocking the GPU, you are taking it out of power/thermals envelope it was designed for and it becomes totally experimental. In other words, you are a pioneer and totally on your own. There are many components on the GPU PCB (besides the ones that touch the heatsink) that have absolutely no thermal protection or monitoring and no active cooling.

 

[InvalidError]

As for the PSU, do not use high current components with extension

By 'extension' he likely meant modular cable: the first photo shows the connectors in his hands at one end of the cable and the PSU-side connector a fairly long way down from there in what appears to be one single cable.

The PSU cable connector probably melted due to a bad crimp.

 

[vov4ik_il]

The PSU cable connector probably melted due to a bad crimp.

Well, that might be the case, having bad crimps only on 4 +12V leads off that one connector would be a bit of special luck.
Since all +12v got cooked in one connector, it might be the card eventually had trouble with the current steering. But it is a result, not the cause.
Anyway, it is all just speculation, I would send it in and let a tech figure it out.

 

[InvalidError]

Well, that might be the case, having bad crimps only on 4 +12V leads off that one connector would be a bit of special luck.

The crimps may be bad on all eight pins: it is only the 6+2 connector that has melted off and the other 6-pin splits off from it which means these crimps have doubled-up wires. My guess is that the wires got partially cut from over-crimping - either the crimper was calibrated for thinner wire gauge or not compensated for the fact that wires were doubled up. GND pins got spared simply because current can return through the PCIe slot's 70+ ground pins and through chassis ground via the rear bracket.

 

[vov4ik_il]

@InvalidError, I don't say it is impossible, just expressing reasonable doubt. I also tend to blame the cabling for being out of spec. As to the PCIe current leak, it does happen normally, and that is exactly what current steering is supposed to limit or prevent.

In any case, I would not use this GPU until the connector is either refurbished (every pin in the connector is cleaned with steel pick to bare metal, will void the warranty) or replaced/RMA'd.
The PSU cable has to be replaced.
The hardline loop might be great for cooling the critical hot areas, but there are other areas that get hot and not touch the heatsink, the absence of airflow over the card (which would normally be there with factory cooling) may permanently damage the card.

 

[InvalidError]

@InvalidError, I don't say it is impossible, just expressing reasonable doubt.

Improper crimping is the most common cause of crimped cable/connector failure: not tight enough and the wires pull out, too tight and the wires break off. I crimp some of my own cables every now and almost always scrap at least one connector because I need to tune my crimper to the specific wire gauge and terminal I am crimping. The most likely way the factory could screw up multiple terminals in a single connector is if there was a tooling change and someone forgot to set it up correctly. This is the sort of thing that would be likely to cause a recall due to fire hazard and civil liability for damages caused.

BTW, AFAIK, the stock cables on Silverstone Strider PSUs are either black ribbons or black bulk-sleeved. The blue individually sleeved cables may very well be aftermarket.

 

[jack4404]

Improper crimping is the most common cause of crimped cable/connector failure: not tight enough and the wires pull out, too tight and the wires break off. I crimp some of my own cables every now and almost always scrap at least one connector because I need to tune my crimper to the specific wire gauge and terminal I am crimping. The most likely way the factory could screw up multiple terminals in a single connector is if there was a tooling change and someone forgot to set it up correctly. This is the sort of thing that would be likely to cause a recall due to fire hazard and civil liability for damages caused.

BTW, AFAIK, the stock cables on Silverstone Strider PSUs are either black ribbons or black bulk-sleeved. The blue individually sleeved cables may very well be aftermarket.

@vov4ik_il

Thank you both very much for your reply's. The cables are indeed Phanteks extensions and I've had them for 3-4 years. I'll be ditching them completely and just using the stock psu cables. I doubt I'll have the warranty for the gpu (bought it used with an invoice, in box with everything, right after launch), but I'll look into it anyway. I thought my card was toast, so this is a bit of a relief, but I'll keep my expectations low and my fingers crossed lol. I'll try clean the pins a bit better, and then put it back together.