Broken GTX 980M

Damn, if you didn't remove the seriel number you could have RMA'd this card directly to a clevo facility, and I could have helped you. Which is weird because no one would remove that seriel ESPECIALLY RJTech, I assume. Also that liquid stuff looks like the oil we use to clean the surfaces or maybe some of the flux from soldering. It really couldn't be anything else IMO, unless a larger component was cracked and leaking, but on both sides of the card and near no such components ???.

Seems fishy to me, honestly.

Do not know how your coils took physical damage either, because it's sort of just not possible. If that card was not touched by you, then either whoever [put it in when upgrading did it, or it came stock like that which would mean its a defect. and because I doubt it came stock like that or you put it in there and somehow damaged it long ago, and the fact the serial is ripped off, i am suggesting someone is replacing your "Alive" card with a dead one. If someone were to touch my seriel numbers, thats the first thought I'd have. Why would it get removed, and even show signs of being ripped off ? the heatsink doesn't even touch there, the laptop doesn;t make contact with it there, there's no reason, honestly. All it can do it help you RMA it or get warranty and identify the card. Since ALL of those apply to this card currently, there is even less chance you'd touch it.

Unless I'm missing something here or just Call me paranoid. Thats my thoughts. I think the RAM was resoldered on, or some sort of oil from a broken components, or foul play because the seriel is missing. DO not know what caused your cards death, it's not plausible that - that oil or damage was there on a new card installed by your seller if it was new, so no idea's.

Get grizzly Minus 8 pads, they're great ! Fujipoly are way to expensive and make no difference in GPU applications, honestly.

Thanks for the mention. Cheers!

 

Actually the oil is very likely from thermal pads. I repaired a great deal of notebooks and have seen this issue a couple of times on many different models. I've even had this issue myself on my GT 70 back in the days, I mounted the card myself and applied everything, so I know it was impossible from another source.

idk about the serial number tbh.

 

I thought about the pads, they can absorb humidity sure, but not oil, really. Normally pads get dry and in the cases they do not, they end up bonding with the components touching them. Least from my experience.

 

The thing is my GT 70 shorted one of the mosfets of my old GTX 570M because of the oily substance. It was exacly like OP's case, it wasn't just any kind of water, I assume this is actually melted glue mixed with some kind of fluid. I did the mounting and used cheap pads. I've also recently had a Asus G750 which also got shorted by this oily substance, only the pads were literally melted, so idk what happened there, it was disastrous, I should have taken a picture, it literally looked like someone took the pads, threw them into a mixer and threw them onto the card, then heating them up. All of the place was the oily substance as well.

 

RJTech has nothing to do with the serial sticker being removed. I removed it to check components below after receiving the card back, but I saved the torn sticker chunk. I had no idea what I was doing, so I tried poking around. Serial number is missing at least 2-3 digits, but the barcode is "intact" for the most part. The card was indeed the one I sent them as I took the precaution to mark the card where it won't interfering with operation. Had they resoldered the VRAM chips and used an ultrasonic cleaner to rid the flux residue, the VRAM chips would look "cleaner," and the mark would have been washed away.

Given the amount of oil covered on the R22 coils + VRAM chips, it has got to be from the thermal pads. Even Fujipoly themselves admitted here in the Warranty Statement that silicone oil can leach from their products. I spent some time looking at the components up close, and I couldn't find any trace of something big enough that would potentially hold/leach that much liquid.

How R22 and R47 appeared damaged is beyond me as well. Not sure if this has any connection with the coil whine issue I mentioned in the original post. There wasn't anything sharp on the old pads that would have caused such damage. The temperatures looked good, so I never bothered to do a repaste. I just checked the original purchase invoice that the seller sent me, and I noticed that the P750ZM was purchased as a barebone with 980M installed. Hey @win32asmguy, do you remember seeing anything strange on the 980M when you installed the 4790K?

I considered the Clevo RMA option, but given RJTech's estimate of 4-6 weeks + ~$400 fee + 90(?) days warranty and my "what else would fail next" mentality at the time, I decided to proceed with the NP9870-S purchase and shelf the P750ZM. One can only tolerate so much eye strain and frustration on a 11" Chromebook. I originally planned to use the slave 980M from the NP9870-S to resurrect the P750ZM and sell the Sager to recuperate losses. After looking at the bottom panel of the NP9870 long and hard, I conceded that it had a more superior cooling solution than the P750ZM and kept it for good. I thought about selling the P750ZM for parts, but I gave replacing the GPU w/ something else cheaper a try which fortunately happened to work. I thought I'll revisit the 980M when I have more time + better understanding of what happened to the card. I've been looking at causes since August without much progress, probably due to focusing too much on the shorting + power supply rather than the contaminated oil on the pad. This is why this thread was made.

The thermal pads on the copper backplate was also oily as seen here. It was important to note that the thermal pads on the backplate is completely different from the one the main heatsink. Note that only the VRAM chips on the back side had a noticeable amount of oil on them. Front side VRAM chips had nothing on them at time of discovery.

 

I'm planning on placing an order for ~1-2 120 mm x 20 mm strip of 0.5, 1.0, 1.5, and 2.0 mm. Do you think these will be enough to replace all the stock pads in a P870DM + P750ZM?

 

Yes I do think so.

have some rep too for being on the forum, and welcome to NBR !

 

Thank you! Order placed.

 

Honestly I don't see anything really wrong with the card. Thermal pads can leave a bunch of liquid goo behind. I don't think anything you're looking at has anything to do with the failure.

It looks like at some point someone may have scraped up the two inductors (the R22 is the inductor for the 3rd core phase, the R47 is the inductor for either the pci-e or 1.8V voltages). That will have zero effect on their performance though. They are not metal coils internally, but a solid block of bonded iron powder.

You probably had a power FET blow. Usually that's the only failure that can cause a lot of heat to be generated. Power FETs can blow and not look like it. The voltage on anything else like memory is too low to make much heat. Also the laptop would still power up.

Check the resistance between the two giant pins on one side of the mxm slot. This is the card's supply voltage. The resistance should be over 2k ohm. If a fet is dead you'll read 0. A blown cap can also cause a similar failure, but it is less likely.

 

Checked. 7 Ohms, effectively shorted. How do I check which power FET was blown? All of the identical looking ones between/near the R22 inductors had the same resistance across them (~7.7 Ohms).
Nothing unusual was noticed on the package as well.

EDIT: Caps mistaken for MOSFET. Ignore the ~7.7 Ohms measurements. See later posts.