I may have broken my 1080 Ti

Joined
Mar 29, 2021
Messages
25
Hi all!

I'm new here but I've been a PC enthusiast for over 2 decades. I recently did a DIY water-cooling for my CPU and it worked great so I tried to add my 1080 Ti to the loop and things didn't go as planed.
I have an Asus Turbo 1080 Ti. Before the water cooling I tried an overclock in MSI Afterburner. I was running +250 on the core and +400 and the memory. I noticed a little bit of artefact at those settings so I left them there and proceed to water-cool the card. I did exactly like that, with the exact same card (except my OC less than his)



After all this was done I got a successful boot and ran TimeSpy Extreme again, with the same OC and got no artefact and a similar score. While I was analyzing the results, I got a bunch of random artefacts on my 3 screens followed by a shut down. Not good! Then I tried for many hours just to get my PC to boot. The debug LEDs on the mother board were stuck at VGA. I put the stock cooler back on to no success. After going insane and doing the same thing over and over again, I got a boot!
Thing is, since then, I get random BSOD or just system restart whenever the GPU is working. I can play DOOM for an hour no problem before it crashes, or sometime just 5 minutes. Even just watching a video in VLC can make the system crash.
I have no idea how I could have kill a card doing what I did. I thought it was maybe the alcohol 50% I used to clean the card (I learned about that after), so I teared the card down again, cleaned it with an electronic cleaner, and put the card in the oven at 175F for an hour to help evaporate any water that could have been stuck somewhere. Thermal paste was replaced by MX-4.
One thing I noticed is the core is running way hotter than before. At idle I was getting in the low 80F at 20% fan speed, now I need to have fan at 40% to get the same temperature. At 20% it's now in the upper 80s. Also under load, I was previously able to stay in the 160F at about 50% fan speed, now I'm hitting over 80% to keep it at those temperatures. So it's definitely running hotter for some reasons. I have confidence in my application of the thermal paste. I even did it twice to make sure.
For now it's running at 50% power limit, -400 on the core and -500 on the memory, but it did not improve much if at all. Still crash way to often.

I'm kind of lost now and not sure what to do next to repair the card. Bad time to buy a new one right now, and I'm sure it can be fixed.

Any ideas? Thanks!
 

Attachments

  • 20210319_175047.jpg
    20210319_175047.jpg
    646.5 KB · Views: 0
looks like there is grease or residue on the card.

I prefer to clean cards in the sink with hot water and dish soap, it degreases pretty well and doesn't leave residue like alcohol. Will need to let it dry in front of a fan for something like 24 hours to make sure it is dry.

Did you give it a thorough once over to ensure you didn't knock any SMD's off?
 
It looked greasy when I first tried it on water. After one week of crashing I did another tear down and cleaned with the contact cleaner attached to this post. It look much cleaner now, but still unstable.

I did not see anything get knocked off the card, but I took pictures to document the whole process so I can open it again and compare to see if anything is missing. This will have to wait for the week-end since I need my computer for work. Maybe I will try to clean it in the sink too.

I'm thinking maybe one VRM phase died and is pushing too much current constantly? If I knew the check points on the card I could verify if the reported voltage and the actual voltages matches.
 

Attachments

  • 20210330_181232.jpg
    20210330_181232.jpg
    375.4 KB · Views: 0
Having not looked at 1080 coolers; that doesn’t look like enough. Was that block designed for the card? What block is it? I expect more coverage on ram and the like, at least for 2000 and 3000 series.
 
Ok. Realized that’s a medium page. Not a weird pic.
I suspect you’re cooking the VRM or something else that needed to be cooled. I would put the stock cooler back on and see if it is stable. And try to find a full coverage block. Someone who knows these more might know what else needs to be cooled, but I suspect you’ve got not enough coverage for that.
 
I installed an Arctic Accelero Xtreme on a GTX 1080 Ti a while back. I do remember installing heatsinks on these guys here.

1.jpg
 
Ok. Realized that’s a medium page. Not a weird pic.
I suspect you’re cooking the VRM or something else that needed to be cooled. I would put the stock cooler back on and see if it is stable. And try to find a full coverage block. Someone who knows these more might know what else needs to be cooled, but I suspect you’ve got not enough coverage for that.

Thing is the stock cooler is back on since 1 hour after I started the project. Something got damaged in the one 3DMark run I did. I have issue with the stock cooler on. It crashes and run hotter than it did before with the same cooler.
Full coverage block don't exists for this PCB. Also, I wanted to not buy something but build something myself (see how that turned out).

I installed an Arctic Accelero Xtreme on a GTX 1080 Ti a while back. I do remember installing heatsinks on these guys here.
Those look like capacitors to me and should not need cooling as far as I know. Even with the stock cooler the card is very unstable now so something got damaged, I need to find a way to test components and replace the bad one(s).

Thanks for the replies, I appreciate it.
 
Looking at similar parts on the aftermarket coolers for 3000 series- those are all actively cooled now. RazorWind may know more.
 
  • Like
Reactions: travm
like this
Thing is the stock cooler is back on since 1 hour after I started the project. Something got damaged in the one 3DMark run I did. I have issue with the stock cooler on. It crashes and run hotter than it did before with the same cooler.
Full coverage block don't exists for this PCB. Also, I wanted to not buy something but build something myself (see how that turned out).

That's not necessarily surprising as the original thermalpads, thermal paste, etc. have been altered. Maybe take it apart and repaste the GPU?
 
That's not necessarily surprising as the original thermalpads, thermal paste, etc. have been altered. Maybe take it apart and repaste the GPU?

I repasted the GPU when I put the water-block, then again when I put the stock cooler back on, and another time again just in case but it didn't change anything. The thermal pads still looks good like in this picture, but cleaner. This picture was taken when I first teared down the card before any other modification
 

Attachments

  • 20210319_171041.jpg
    20210319_171041.jpg
    664.4 KB · Views: 0
looks like there is grease or residue on the card.

I prefer to clean cards in the sink with hot water and dish soap, it degreases pretty well and doesn't leave residue like alcohol. Will need to let it dry in front of a fan for something like 24 hours to make sure it is dry.

Did you give it a thorough once over to ensure you didn't knock any SMD's off?
Not sure if trolling or...
 
Not sure if trolling or...
He's probably not. Assuming you dry it off thoroughly, you actually can clean computer parts that way. You can even run them through a dishwasher. You really need to rinse with distilled water and then blow it all out with compressed air, though.

Am I understanding correctly that the OP's problem is that he tried out water cooling this card with a CLC, but it started crashing, so he put the stock air cooler back on, and now it reports hotter GPU core temperatures and crashes frequently, even if it's underclocked?
 
Am I understanding correctly that the OP's problem is that he tried out water cooling this card with a CLC, but it started crashing, so he put the stock air cooler back on, and now it reports hotter GPU core temperatures and crashes frequently, even if it's underclocked?

100% spot on
 
He's probably not. Assuming you dry it off thoroughly, you actually can clean computer parts that way. You can even run them through a dishwasher. You really need to rinse with distilled water and then blow it all out with compressed air, though.

Am I understanding correctly that the OP's problem is that he tried out water cooling this card with a CLC, but it started crashing, so he put the stock air cooler back on, and now it reports hotter GPU core temperatures and crashes frequently, even if it's underclocked?
He's not. I've washed lots of motherboards in the dishwasher. Give them proper time to dry and they are fine.
 
You should never use tap water on electronics. Distilled, maybe, but not normal sink water.

I personally only use 91% isopropyl when cleaning components.
 
Screenshot of idle and under load attached. Temps under load are low cause it only let it run under load for a few seconds.
My worries are if a VRM is dying, I the reported voltage and actual voltage can be different? I didn't find how to check voltage with a DMM
 

Attachments

  • Idle.png
    Idle.png
    182.8 KB · Views: 0
  • Load.png
    Load.png
    227.5 KB · Views: 0
Screenshot of idle and under load attached. Temps under load are low cause it only let it run under load for a few seconds.
My worries are if a VRM is dying, I the reported voltage and actual voltage can be different? I didn't find how to check voltage with a DMM
There should be some huge solder pads on the back of the core area where a capacitor could be installed that are a good place to check the actual voltage. Put the black probe on a good ground, such as the slot cover plate, and the other probe on the big pads on the back of the core. One pad in the pair will be ground, and the other should be vcore. It may differ slightly from what PX1 reports, but it should be close. Obviously, you need to be very careful that you touch ONLY the pad you intend to probe, hence why I suggested using the large ones.
 
There should be some huge solder pads on the back of the core area where a capacitor could be installed that are a good place to check the actual voltage. Put the black probe on a good ground, such as the slot cover plate, and the other probe on the big pads on the back of the core. One pad in the pair will be ground, and the other should be vcore. It may differ slightly from what PX1 reports, but it should be close. Obviously, you need to be very careful that you touch ONLY the pad you intend to probe, hence why I suggested using the large ones.

I got this right?
 

Attachments

  • Probe.png
    Probe.png
    5.7 MB · Views: 0
Oh stop. Cleaning with hot water and dish soap is very effective. Its pretty much the only way to clean vintage hardware as well.
Yeah, generally I only try to use isopropyl but I've had boards come in that were used by heavy smokers and the dishwasher is simply the only thing that will clean those.
 
Just make sure it's been "off" and unplugged long enough the caps have had time to discharge fully, but that's a gut reaction on my part. Other than that - clean it, dry it, isopropyl it.

Remember, a lot of this stuff is BUILT underwater in the first place (that's how modern lithography works - and yes, I know that's INSIDE the chip, but still).
 
Just make sure it's been "off" and unplugged long enough the caps have had time to discharge fully, but that's a gut reaction on my part. Other than that - clean it, dry it, isopropyl it.

It's unplugged every night, for a minimum of 8h. I have this problem since March 19. I'll also try to clean & dry it this week-end but I have little hope. I may buy another one just to be able to work from home correctly, while I try repair this one. The price of those cards is just insane right now :(
 
Funny - My cousin has an Asus 1080 Ti that is exacting somewhat the same way. His was watercooled on a full coverage block for a couple of years and then he started having worsening crashes until he couldn't even successfully log into Windows anymore, even on a fresh install. Temperatures all report as OK, and he has a plenty large Seasonic 850W power supply. Before this point we'd try stress testing it, and it would seem fine, and we even underclocked it just in case but issues kept happening more and more when he was active playing anything. We removed the block, cleaned it up real well and put it back on air - no difference in his PC. His PC is a Core i9-9900K system on an Asus Maximus mainboard.

We put it in my daughter's computer (which has never had an nVidia card in it) on an Asus Crosshair VI Hero, and it is working * mostly * OK. I can't complete Timespy without it crashing, but her desktop, web, video and Fortnite use all seem to work just fine (which is still better than he was getting on it).

Next step is we're going to try re-flashing the BIOS on the card... because what else CAN we do, really? It's worth a shot.
 
That heatsink you have over the MOSFETs isn't making contact with any of the pads for the inductors or other surrounding components, is it?
 
Funny - My cousin has an Asus 1080 Ti that is exacting somewhat the same way. His was watercooled on a full coverage block for a couple of years and then he started having worsening crashes until he couldn't even successfully log into Windows anymore, even on a fresh install. Temperatures all report as OK, and he has a plenty large Seasonic 850W power supply. Before this point we'd try stress testing it, and it would seem fine, and we even underclocked it just in case but issues kept happening more and more when he was active playing anything. We removed the block, cleaned it up real well and put it back on air - no difference in his PC. His PC is a Core i9-9900K system on an Asus Maximus mainboard.

We put it in my daughter's computer (which has never had an nVidia card in it) on an Asus Crosshair VI Hero, and it is working * mostly * OK. I can't complete Timespy without it crashing, but her desktop, web, video and Fortnite use all seem to work just fine (which is still better than he was getting on it).

Next step is we're going to try re-flashing the BIOS on the card... because what else CAN we do, really? It's worth a shot.
That sounds like a pretty clear-cut BGA failure. You could conceivably try reballing the GPU and memory if you're equipped to do it, but that's not exactly a trivial task.

The OP's problem is possibly different in that the card at least seems to think that it's running hotter than usual. Actually... Now I just thought of something.

BlueScreen128 - What is the condition of your power supply? What voltage do you get when you measure the 3.3V and 12V inputs on the card?
 
That sounds like a pretty clear-cut BGA failure. You could conceivably try reballing the GPU and memory if you're equipped to do it, but that's not exactly a trivial task.

The OP's problem is possibly different in that the card at least seems to think that it's running hotter than usual. Actually... Now I just thought of something.

BlueScreen128 - What is the condition of your power supply? What voltage do you get when you measure the 3.3V and 12V inputs on the card?

Good point - I haven't baked it yet... Into the oven it will go this weekend
 
That heatsink you have over the MOSFETs isn't making contact with any of the pads for the inductors or other surrounding components, is it?
No it's not. It wasn't clear to me what component needed the heatsink. On the article I used as a guide, he put them on the inductors. But when I looked at my card, the only component that had thermal pad were the MOSFET so that's why I put the heatsink on them. I will say that they got HOT to the touch, I could no keep my finder on them. Haven't touched the other components, the computer shutdown before I could do all the tests I wanted.

What is the condition of your power supply? What voltage do you get when you measure the 3.3V and 12V inputs on the card?
It's the same good old PSU I use everyday since 2013. It's a Corsair TX850M. I will measure those voltages also tomorrow night.
 
No it's not. It wasn't clear to me what component needed the heatsink. On the article I used as a guide, he put them on the inductors. But when I looked at my card, the only component that had thermal pad were the MOSFET so that's why I put the heatsink on them. I will say that they got HOT to the touch, I could no keep my finder on them. Haven't touched the other components, the computer shutdown before I could do all the tests I wanted.


It's the same good old PSU I use everyday since 2013. It's a Corsair TX850M. I will measure those voltages also tomorrow night.
No shit? As it happens, I have a TX650M right here that, with 100% certainty, will kill a GTX690 after about 30 seconds under load. Wouldn't surprise me if your 8 year old power supply is a contributing factor here.

Have you tried this card in a different system, or with a different power supply?
 
No shit? As it happens, I have a TX650M right here that, with 100% certainty, will kill a GTX690 after about 30 seconds under load. Wouldn't surprise me if your 8 year old power supply is a contributing factor here.

Have you tried this card in a different system, or with a different power supply?

Not yet as I don't have an extra PC here but it's on my list for tomorrow. I will test it and another PC.
It's not impossible, but I would be very surprised if the PSU broke at the exact moment I try to modify the video card. I would be glad to replace the PSU instead of the GPU!

Funny story about the PSU, I originally wanted to buy the 650W version because it was plenty for my 4670K / GTX 770 at the time, but there was a special on the 850W making it was the same price as the 650W, and the 750W was the most expensive of the three. So obviously I got the 850W.

For my own knowledge, how can your PSU kill a card? As I understand it, power is drawn by the card and not pushed by the PSU. Could it be a short in the card itself?
 
Not yet as I don't have an extra PC here but it's on my list for tomorrow. I will test it and another PC.
It's not impossible, but I would be very surprised if the PSU broke at the exact moment I try to modify the video card. I would be glad to replace the PSU instead of the GPU!

Funny story about the PSU, I originally wanted to buy the 650W version because it was plenty for my 4670K / GTX 770 at the time, but there was a special on the 850W making it was the same price as the 650W, and the 750W was the most expensive of the three. So obviously I got the 850W.

For my own knowledge, how can your PSU kill a card? As I understand it, power is drawn by the card and not pushed by the PSU. Could it be a short in the card itself?
The most common I've seen is that, on cards where the power transistor gates are driven directly from the power supply rails, the gate voltage on those transistors will droop approximately in parallel with the voltage produced by the power supply.

When the gate voltage to a transistor drops, the transistor doesn't turn all the way on, burning off a whole bunch of energy as heat as it passes current. If it burns enough energy off as heat, it eventually destroys the transistor. When this happens on the high side of a transistor, it frequently causes the input voltage to be shorted into the switch node, or output of the VRM.

In less technical terms, low voltage supplied to the card causes the VRM to run so hot that it burns up and applies the input voltage directly to the GPU itself. GPUs are surprisingly tough, and you can usually repair the card by replacing the failed transistor, but it's not guaranteed, and you have to diagnose it first, which can be tricky.
 
Oh stop. Cleaning with hot water and dish soap is very effective. Its pretty much the only way to clean vintage hardware as well.
Huh, I wonder why LN2 overclockers are always so worried about condensation getting under the surface mount components? If water gets under something it's not going to be easy to see that. What you fail to grasp is that the components which are exposed to water during manufacturing are not then plugged into a system at full voltage, are they? They go through many steps including heating to remove any contaminants. Quit spreading oversimplified falsehoods for the topic at hand: modern hardware.
 
Huh, I wonder why LN2 overclockers are always so worried about condensation getting under the surface mount components? If water gets under something it's not going to be easy to see that. What you fail to grasp is that the components which are exposed to water during manufacturing are not then plugged into a system at full voltage, are they? They go through many steps including heating to remove any contaminants. Quit spreading oversimplified falsehoods for the topic at hand: modern hardware.
The key difference is that the competitive overclockers are dealing with condensation while the system is running. This causes two problems:
1. It can create short circuits, which cause the hardware to malfunction
2. Because there is a voltage being passed through the condensed water, it causes the solder joints to corrode very rapidly. Like, rapidly enough that you can sometimes see it happen in real time.

Assuming you dry the board assembly off thoroughly before you plug it in and power it on, it's perfectly fine to clean PCBs with tap water, if something better isn't available. You just need to be really sure you dried it off thoroughly - particularly under the BGA components.
 
What is the condition of your power supply? What voltage do you get when you measure the 3.3V and 12V inputs on the card?

Just checked that. I got 12.00V exactly on all 12V pins. VCore is about 2mV different than I see in MSI Afterburner, under load and at idle so that's good. I'll try the card on another system in a couple hours.
What about the 3.3V? AFAIK the card only get 12V from the connectors so I'm not sure what you meant there.
 
Huh, I wonder why LN2 overclockers are always so worried about condensation getting under the surface mount components? If water gets under something it's not going to be easy to see that. What you fail to grasp is that the components which are exposed to water during manufacturing are not then plugged into a system at full voltage, are they? They go through many steps including heating to remove any contaminants. Quit spreading oversimplified falsehoods for the topic at hand: modern hardware.

Just because you want it to be bad doesn't make it so. There is a growing community of vintage hardware hobbyists, who have to clean some really nasty components. Dish soap and hot water is the method. I have a (too large) collection of old hardware as well, and basically all of it has gone in the sink for a cleaning. Really helps when I need to re-cap a motherboard as well. Or video card. They all still work after a good scrub.

Like so:

IMG_20210125_201632748.jpg
 
Just checked that. I got 12.00V exactly on all 12V pins. VCore is about 2mV different than I see in MSI Afterburner, under load and at idle so that's good. I'll try the card on another system in a couple hours.
What about the 3.3V? AFAIK the card only get 12V from the connectors so I'm not sure what you meant there.
Are these measurements taken while it's idling, or actually doing something? Try taking another measurement while it's doing something that would normally cause it to crash. Edit: It's really the 12V that we're interested in.

The six and eight pin connectors only provide 12V, but the system also provides 3.3V to the card through the slot. This is frequently used to power things like the BIOS chips and VRM control logic. On this card, it may also be used to power the transistor gates, and as I mentioned before, if the gate voltage is low, the transistors will run hot. If you look at the PCI-E pinout, you'll see that pin 8 on both sides of the connector is the +3.3V input. You'll need to follow the traces to somewhere you can probe with the card plugged in.

This may be a dumb question, but... have you tried this card in a different system, or a different card in the system in question?
 
Just make sure it's been "off" and unplugged long enough the caps have had time to discharge fully, but that's a gut reaction on my part. Other than that - clean it, dry it, isopropyl it.

Remember, a lot of this stuff is BUILT underwater in the first place (that's how modern lithography works - and yes, I know that's INSIDE the chip, but still).
Honestly at this stage of the game all of these things are designed with pull downs for capacitors, and anti static protections to the point you could likely wash it in the sink beside a vandegraaff generator (disclaimer, not recommended).


Just because you want it to be bad doesn't make it so. There is a growing community of vintage hardware hobbyists, who have to clean some really nasty components. Dish soap and hot water is the method. I have a (too large) collection of old hardware as well, and basically all of it has gone in the sink for a cleaning. Really helps when I need to re-cap a motherboard as well. Or video card. They all still work after a good scrub.

Like so:

View attachment 344305
Whoa, I use a toothbrush :D, thats intense.
 
Back
Top