GTX 1080Ti Failsafe

[Joseph.b211]

Hello, I Recently purchased a MSI GTX 1080ti 11g card. The person I bought it from claims the card worked up until he replaced the replaced the thermal compound. I was wondering if anyone would be willing to help me out and give me any advice. When I have the card installed in a computer with no pcie 8 pin power connectors the system will boot and all fans will spin even the ones on the card but no display signal. When I have one 8 pin connector plugged in the one closest to the I/O the same thing happens the system will boot and all fans will spin even the ones on the card but no display signal. BUT when I try this with the other 8 pin connector the system wont turn on and I hear a click in the Power supply like its shorting out. The same thing happens when I have both plugged in. I believe the issue would have to do with whatever that 8 pin powers I have some EE experience so trying anything isn't an issue.

[DaJMasta]

If it broke after replacing thermal compound, the problem is most likely physical damage from the removal or replacement of the heatsinks, so I'd just look over the board carefully and look for missing parts (sometimes difficult to tell from intentionally omitted parts) or parts with visible damage.

Replacing any of those would be a starting point, but I want to be perfectly clear: this card is incredibly complex and many of the parts are difficult to identify, source, or rework without experience.  This kind of repair is not one well suited to beginners, if it's repairable at all (a knick out of the GPU could happen with heatsink removal and replacement and would not be fixable), so if you're going to attempt it, expect for it to take a lot of time and understand that the likelihood that it will work fine in the end is low.

[Joseph.b211]

Yeah I understand that, this is just alittle projected I wanted to take on to try to keep busy. I looked over the whole card and I didnt see anything usual no bulging caps or any burns on the PCB. I cleaned everything with rubbing alcohol and the qtip had no type of burn residue. I did notice that towards the bottom of the VRMs an inductor (The components labeled SFC) had a burnt smell to it. Do you think a bad            VRM inductor or a few Inductors gone bad will cause this type of issue?  I just think its weird that i get different results depending on what PCIE connector i use on the gtx 1080 that leads me to think that there is some sort of power delivery issue with that half of the PCIE power circuit

[Runco990]

The card should start with ONE connector, although it could crash once you go into a game and load it.  For desktop it will run fine on ONE connector.  My guess is someone "knowingly" sold you a bad card.  One circuit has a short.  You could try ohming out the vrms and find the shorted one and remove it.  It should start with one or 2 missing.  They run in parallel.

Mining card???

[Joseph.b211]

I bought the card knowing that is wasn't working i grabbed it for a good deal and I figure it was worth a shot to fix. Im not sure if it was a mining card the person I bought it off of was very nice and didnt say they mined with it but you know anything is possible. Do you know what values I should be looking for?

[wraper]

If seller was telling the truth, the first thing you need to check is if corners of silicon die are not chipped. If they are, the best thing you can do is selling it further to another idiot. Also if you see some graphics card or motherboard sold with statement that some small part was knocked off, don't delude yourself that it will be an easy fix. It might as well be GPU IC failure and small part knocked off intentionally to make it more attractive.

BUT when I try this with the other 8 pin connector the system wont turn on and I hear a click in the Power supply like its shorting out

Likely a high side switch shorted in VRM. Probably it already overvolted and killed GPU. Measure if 12V from that connector is not shorted to GND. If you are very lucky might be just shorted ceramic capacitor.

[wraper]

The card should start with ONE connector, although it could crash once you go into a game and load it.  For desktop it will run fine on ONE connector.

No, it will not. They are not connected in parallel and each powers only a part of VRM.  But likely you will see a message about lacking PCI-E power connection on display.

[Joseph.b211]

Okay Ill check to see if any there's any continuity between the positive and negative. If there is do you think there is any hope in the card? The die it self is in good condition no chips or anything honestly its strange nothing burnt either I took some up-close hd video of the card with my phone I can post those if it helps.

[wraper]

Okay Ill check to see if any there's any continuity between the positive and negative. If there is do you think there is any hope in the card? The die it self is in good condition no chips or anything honestly its strange nothing burnt either I took some up-close hd video of the card with my phone I can post those if it helps.

As long as GPU is not dead or PCB burned, it's fixable. GPU can be replaced too but not economically viable.

[Runco990]

No, it will not. They are not connected in parallel and each powers only a part of VRM.  But likely you will see a message about lacking PCI-E power connection on display.

Odd... my evga 1080ti did, as I had forgotten to plug in the other cable initially.  I did get a post. 
In any case... I have fixed a card or 2, but usually a cap was bloated.  VRM's are another story except here is a short which will put the power supply into protection.

I have a flir camera, so idiot me would likely feed 12V into the card on the bench using a current controlled power supply and look for what gets warm to begin with.  Found a lot of things like that.

[Joseph.b211]

Okay so I got the meter out and did some testing and this is what I found using the pinout i drew please see attached picture. From what I can tell the the second, third, and fourth pin on the second pcie plug is shorted somewhere due to the continuity on pins 3-4-5-6-8. What do you think my next step should be? all my inductors have continuity to ground and none to 12v

edit everytime i try to upload the picture it says failed security check so here is a link
https://ibb.co/C8k66yk

[wraper]

Yep, there is a short on second connector. Which means a high side MOSFET shorted in VRM (might be a low side MOSFET as well in the same phase + GPU might be dead). Or small chance of shorted ceramic or polymer capacitor.

[Joseph.b211]

Okay I can probably test the mosfits. If I understand right the mosfets are #fdpc 5018sg I'll try to get pinout to test them. What do you think the best way to test them are?

[wraper]

It's a dual Mosfet. Without desoldeing inductors, you can only measure gates basically. Check resistance between gates and ground or inductor connected to the mosfet. And see if some of them is shorted or has low resistance compared to others. https://www.onsemi.com/pub/Collateral/FDPC5018SG-D.pdf

[Joseph.b211]

Okay, so halfway though some measuring I discovered something. I was checking resistance and continuity between the mosfets and I noticed there was what looked to be a solder ball between some pins but for some reason the pins aren't on that data sheet, but I checked the same for the continuity between those pins and all the other mosfets none of them were connected. Im going to try to remove the solder joint hopefully this will do something. Let me know what you think I'm going to attach a picture of what I am talking about

[wraper]

Okay, so halfway though some measuring I discovered something. I was checking resistance and continuity between the mosfets and I noticed there was what looked to be a solder ball between some pins but for some reason the pins aren't on that data sheet, but I checked the same for the continuity between those pins and all the other mosfets none of them were connected. Im going to try to remove the solder joint hopefully this will do something. Let me know what you think I'm going to attach a picture of what I am talking about

Likely it's one which failed. Solder melted and was squeezed out because MOSFET reached very high temperature.

[Joseph.b211]

If i remove that MOSFET what is the likely hood of the card being able to post you think (as long as the takes away the shorted PCIE Connector)? I don't have an issue with replacing it but id just like to know the likely hood of it working before I get more time and money invested?

[MathWizard]

I have a broken GPU, the main chip seems to be cooking itself, so it probably is toast already. IDK, it's 1 of many broken thing's I own that I look forward to fooling around with, once I know what to expect from some more theory.