Any hope for a dead RTX4000 GPU?

[jayjr1105]

This was a garbage freebie from work. A small component (maybe a VRM?) just went up in smoke (see pics). I contacted PNY and pushed pretty hard but they denied my RMA claim saying it could had been a surge from the PSU that made that component fry. I thought about getting an attorney involved for some type of small claims deal but I'm not sure it's worth it. It's well known in the community that PNY pulls this crap for RMA's. The client machine is operating just fine with the same power supply with the new card but they simply won't budge.

Anyways, going forward with it. Is there any hope for repair? Or just sell for parts? Might fetch a couple hundred as it's a $1000 card new.

https://www.badcaps.net/forum/attach...1&d=1640296661

[DrasticRepairs]

Re: Any hope for a dead RTX4000 GPU?

That's a current ballancing MOSFET. It's not even really needed, but should be replaced anyway.
The reason it burnt up, though, is because one of the Vcore or VMEM MOSFETs shorted. If it's VMEM, say goodbye. If it's vcore, it's most likely repairable.
So you'll have 2 MOSFETs, and possibly a gate driver to replace.

And uhh... that's not an RTX 4000. Those haven't come out yet. More likely an RTX 2060 or 1060.

[jayjr1105]

Re: Any hope for a dead RTX4000 GPU?

That's a current ballancing MOSFET. It's not even really needed, but should be replaced anyway.
The reason it burnt up, though, is because one of the Vcore or VMEM MOSFETs shorted. If it's VMEM, say goodbye. If it's vcore, it's most likely repairable.
So you'll have 2 MOSFETs, and possibly a gate driver to replace.

And uhh... that's not an RTX 4000. Those haven't come out yet. More likely an RTX 2060 or 1060.

Quadro RTX 4000

[DrasticRepairs]

Re: Any hope for a dead RTX4000 GPU?

Quadro RTX 4000

Oh. That makes sense. Looks the same as a 2060, mostly. Still the same diagnosis. If you have the tools I can help you diagnose and repair it. You'll need a multimeter, a hot air station and a soldering iron, preferably, with a bevel tip.

[jayjr1105]

Re: Any hope for a dead RTX4000 GPU?

Oh. That makes sense. Looks the same as a 2060, mostly. Still the same diagnosis. If you have the tools I can help you diagnose and repair it. You'll need a multimeter, a hot air station and a soldering iron, preferably, with a bevel tip.

Yep, I have all those things.

[SMDFlea]

Re: Any hope for a dead RTX4000 GPU?

Yep, I have all those things.

Do you have a microscope, can you get any better pictures of the burnt part, can you see any part numbers on it at all.

[Topcat]

Re: Any hope for a dead RTX4000 GPU?

Please attach images to the thread and not use 3rd party image hosting. I fixed the initial post.

[jayjr1105]

Re: Any hope for a dead RTX4000 GPU?

Do you have a microscope, can you get any better pictures of the burnt part, can you see any part numbers on it at all.

I do, just a USB one but I'll try later.

Please attach images to the thread and not use 3rd party image hosting. I fixed the initial post.

Sure thing

[jayjr1105]

Re: Any hope for a dead RTX4000 GPU?

Here's some images. I cleaned it up with some 99% ipa

[DrasticRepairs]

Re: Any hope for a dead RTX4000 GPU?

Here's some images. I cleaned it up with some 99% ipa

Okay, cool.
First, you want to pull that burnt chip off. You can test the card without it. Next, you want to measure resistance on the 12V inputs. So the 6/8 pin connectors and the 3 far left pins on the pcie fingers.
That will tell us which rail the shorted MOSFET is on and verify that one really is shorted.

A picture of the whole front of the PCB would be helpful.

Once you have confirmed that there really is a short, you can look for the VCORE and VMEM, power stages, DRMOS, or MOSFETs, depending on what they used. The part number of those will be necessary to find the high gate and measure the resistance on that for each phase.

[jayjr1105]

Re: Any hope for a dead RTX4000 GPU?

Well it got a little more interesting. It's burned on the opposite side in the same location...

[jayjr1105]

Re: Any hope for a dead RTX4000 GPU?

So a small update. I did take some time to heat up and remove the upper and bottom components that were cooked. They sort of crumbled off but they are off and the area is cleaned up. The layers seem to be badly burned underneath both components and unfortunately there is still a short to ground somewhere. Whether it's the cooked layers or a surface mount component somewhere I have no idea. I may be at the end of the road with my skillset.

[Bozog]

Re: Any hope for a dead RTX4000 GPU?

I think there is still hope that the GPU survived this, but clearly the PCB is badly damaged and almost certainly the source of the short. It might be a challenging repair