Probably Bound to Happen: Increased Power Draw for DLSS 5 Reportedly Claims Its First Victim After a User Finds an RTX 5090 with a Melted Power Ada...

Soldering would fix the resistance issue, right, but what about the voltage imbalance between pins, it wouldn't fix that would it? I would think some kind of hardware would be needed on the PCB to regulate the voltage per-pin, IIRC NVIDIA removed this hardware function.
 
igor_kavinski said:
So you solder again every time it melts? Fun job!
Click to expand...
Or just solder it right for the first time?

The excessive heating is caused by the bad connection / poor surface contact of the pins combined with load imbalance. If it's soldered then you remove a hole from the swiss cheese model.

Also plastic melts away at a lower temperature than solder melts at.
 
Grimlakin said:
Right but because testers ignored them then.. doesnt mean you can ignore them now.
Click to expand...
Right, but if it could be ignored before but not now, that means this connector is worse with no safety margin.
Grimlakin said:
Thats like running a 5090 and high end system with multiple drives on a 800 watt chinesium power supply that is 80 plus bronze but nowhere to be found on the registry.. it might work... or you might need the fire department.
Click to expand...
Undersizing your power supply is a choice. With the cable you have no choice. If I could replace the connector with four PCIE power cords I would, but I can't. You shouldn't need to buy a $150 gizmo to prevent your $10000 GPU from catching fire.
 
Brent_Justice said:
Soldering would fix the resistance issue, right, but what about the voltage imbalance between pins, it wouldn't fix that would it? I would think some kind of hardware would be needed on the PCB to regulate the voltage per-pin, IIRC NVIDIA removed this hardware function.
Click to expand...

The imbalance seems to be caused by poor connection between one or more pins and another pin picks up the slack. Soldering the connection would ensure all pins have a solid connection which should eliminate the imbalance issue which would then eliminate the heat.

On my 5090 I have one pin that's drawing an excessive amount of power and another that's not. I suspect the one that's not drawing a lot of power is a bad connection somewhere which causes the other one to compensate.

igor_kavinski said:
So you solder again every time it melts? Fun job!
Click to expand...

It shouldn't melt since it wouldn't get hot anymore. The connector on the GPU itself is soldered to the PCB and it doesn't melt because the the heat doesn't build up there due to a good solid connection. That would then extend to the newly soldered cables.
 
MadMummy76 said:
Right, but if it could be ignored before but not now, that means this connector is worse with no safety margin.
Click to expand...
Discussion about the merit of the connector not withstanding

The point was that the influencer/author/whatever you want to call them had access to all the tools, and was under the pretense of testing the card - but ignored a key tool in their toolbox. Had they been watching everything - I dunno if you could have prevented it, but you could have got some pretty interesting data if nothing else.

I have to admit, while I don't fault HU for the card failing, but they did miss a good opportunity. The conspiracy theory is they intentionally ignored it hoping for flames, because smoke and fire = more clicks. I don't necessarily believe that, but ... they did post the video for clicks.
 
Brian_B said:
Discussion about the merit of the connector not withstanding

The point was that the influencer/author/whatever you want to call them had access to all the tools, and was under the pretense of testing the card - but ignored a key tool in their toolbox. Had they been watching everything - I dunno if you could have prevented it, but you could have got some pretty interesting data if nothing else.

I have to admit, while I don't fault HU for the card failing, but they did miss a good opportunity. The conspiracy theory is they intentionally ignored it hoping for flames, because smoke and fire = more clicks. I don't necessarily believe that, but ... they did post the video for clicks.
Click to expand...
I think there is already enough data to draw a conclusion what more can they add? I believe they just didn't expect it to happen to them, well like everyone else.

But also it is important as a tester to make the test environment as close to real-world as possible, if you employ all the tools the average person doesn't have access to then the testing isn't very real world. TBF I don't know what were they testing when it happened and would it have mattered.
 
Peter_Brosdahl said:
So, and sorry if I missed this in the thread, what if there's a poor connection at the PSU side?
Click to expand...

It burns there as well. In HUB's instance, the melting was at the power supply side. They had to replace the PSU in their case, not the video card, but had they left it running in that scenerio, it would have eventually burned the GPU side as well, but they caught it in time.

Because ultimately, the issue is the whole format and form factor itself, not just the pins or connectors, but also the wire, the load imbalance, it's several issues that combine into one outcome.
 
You must log in or register to reply here.
Become a Patron!
Back
Top