Black screen after replacing thermal paste

[tofkop]

Hi,

 

When I boot up my pc my R9 390 shows a black back-lit screen after posting. I hear the windows sound that the login screen is displaying. Removing the drivers in safe mode using ddu let's me use my computer normally. I'm able to install the drivers connecting my screen to the on-board graphics. The moment I connect my screen it crashes.

Sometimes it showed some artifacts before crashing or showing the login screen, see 

https://drive.google.com/file/d/1VEvM_G0uBW8gL7m6IDqt9LD03c_Ed8su/view?usp=drivesdk

. Before I could run occp or other software to test it it crashed.

 

Possible causes:

- I somehow damaged it while replacing the thermal paste. I don't think so but first time I did it after 5years... used noctua nt h1 which is nonconductive.

- I damaged it when the case wasn't properly grounded. The outlet in my new house wasn't properly grounded and shocked me when I touched the case. Disconnected everything soon after and fixed the grounding. 

- I did a windows cumulative update around that time and it broke the driver or something. Using a system restore point didn't fix it. 

 

What i tried:

- Tried using a displayport instead of the hdmi connection. 

- updated windows, bios, and gpu bios. Installing older drivers, system restore points.

- Installing the driver while using the on-board graphics. Testing vram with occp while using the on-board graphics as display. Ran it for 1min and it completed normally, but crashed when I tried it a second time, or when I tried to run it for 1h.

- Disabled windows energy saving setting for pci.

- Tried different pci slot 

- cleaned and reapplied the thermal paste again. Checked for anything out of place.

- I installed another harddrive as well before it gave me this problem, disconnected all extra harddrives keeping only my boot ssd.

- checked the voltages in bios and everything seems ok. Changed one of the power cables to another psu "slot" that was free.

 

Specs:

I5 6500

Sapphire R9 390 trix

Seasonic m12II evo 620w bronze 

Gigabyte GA-H110M-S2H

Kingston 16GB 2133

 

Only thing left to try is baking the card I think. Any other suggestion besides buying a new system?

 

Thanks

[leclod]

18 minutes ago, tofkop said:

Hi,

 

When I boot up my pc my R9 390 shows a black back-lit screen after posting. I hear the windows sound that the login screen is displaying. Removing the drivers in safe mode using ddu let's me use my computer normally. I'm able to install the drivers connecting my screen to the on-board graphics. The moment I connect my screen it crashes.

Sometimes it showed some artifacts before crashing or showing the login screen, see 

https://drive.google.com/file/d/1VEvM_G0uBW8gL7m6IDqt9LD03c_Ed8su/view?usp=drivesdk

. Before I could run occp or other software to test it it crashed.

 

Possible causes:

- I somehow damaged it while replacing the thermal paste. I don't think so but first time I did it after 5years... used noctua nt h1 which is nonconductive.

- I damaged it when the case wasn't properly grounded. The outlet in my new house wasn't properly grounded and shocked me when I touched the case. Disconnected everything soon after and fixed the grounding. 

- I did a windows cumulative update around that time and it broke the driver or something. Using a system restore point didn't fix it. 

 

What i tried:

- Tried using a displayport instead of the hdmi connection. 

- updated windows, bios, and gpu bios. Installing older drivers, system restore points.

- Installing the driver while using the on-board graphics. Testing vram with occp while using the on-board graphics as display. Ran it for 1min and it completed normally, but crashed when I tried it a second time, or when I tried to run it for 1h.

- Disabled windows energy saving setting for pci.

- Tried different pci slot 

- cleaned and reapplied the thermal paste again. Checked for anything out of place.

- I installed another harddrive as well before it gave me this problem, disconnected all extra harddrives keeping only my boot ssd.

- checked the voltages in bios and everything seems ok. Changed one of the power cables to another psu "slot" that was free.

 

Specs:

I5 6500

Sapphire R9 390 trix

Seasonic m12II evo 620w bronze 

Gigabyte GA-H110M-S2H

Kingston 16GB 2133

 

Only thing left to try is baking the card I think. Any other suggestion besides buying a new system?

 

Thanks

So the computer works fine with igpu or not ?

The Ram is ok or not ? if not try one stick at a time

What are you doing messing with thermal paste on gpu ?

 

Try reinstalling W10 ?

You could first try the option which keeps the programs and all

[Latvian Video]

21 minutes ago, tofkop said:

Hi,

 

When I boot up my pc my R9 390 shows a black back-lit screen after posting. I hear the windows sound that the login screen is displaying. Removing the drivers in safe mode using ddu let's me use my computer normally. I'm able to install the drivers connecting my screen to the on-board graphics. The moment I connect my screen it crashes.

Sometimes it showed some artifacts before crashing or showing the login screen, see 

https://drive.google.com/file/d/1VEvM_G0uBW8gL7m6IDqt9LD03c_Ed8su/view?usp=drivesdk

. Before I could run occp or other software to test it it crashed.

 

Possible causes:

- I somehow damaged it while replacing the thermal paste. I don't think so but first time I did it after 5years... used noctua nt h1 which is nonconductive.

- I damaged it when the case wasn't properly grounded. The outlet in my new house wasn't properly grounded and shocked me when I touched the case. Disconnected everything soon after and fixed the grounding. 

- I did a windows cumulative update around that time and it broke the driver or something. Using a system restore point didn't fix it. 

 

What i tried:

- Tried using a displayport instead of the hdmi connection. 

- updated windows, bios, and gpu bios. Installing older drivers, system restore points.

- Installing the driver while using the on-board graphics. Testing vram with occp while using the on-board graphics as display. Ran it for 1min and it completed normally, but crashed when I tried it a second time, or when I tried to run it for 1h.

- Disabled windows energy saving setting for pci.

- Tried different pci slot 

- cleaned and reapplied the thermal paste again. Checked for anything out of place.

- I installed another harddrive as well before it gave me this problem, disconnected all extra harddrives keeping only my boot ssd.

- checked the voltages in bios and everything seems ok. Changed one of the power cables to another psu "slot" that was free.

 

Specs:

I5 6500

Sapphire R9 390 trix

Seasonic m12II evo 620w bronze 

Gigabyte GA-H110M-S2H

Kingston 16GB 2133

 

Only thing left to try is baking the card I think. Any other suggestion besides buying a new system?

 

Thanks

Have you tried not tightening the cooler all the way? Overtightening can cause problems on cpus, so it probably also can do that on GPUs. 

[tofkop]

1 hour ago, leclod said:

So the computer works fine with igpu or not ?

The Ram is ok or not ? if not try one stick at a time

What are you doing messing with thermal paste on gpu ?

 

Try reinstalling W10 ?

You could first try the option which keeps the programs and all

IGpu works fine, haven't touched the ram so I don't think it will have anything to do with it.

Replaced the paste on the GPU since it hadn't been replaced in 5years, and if it could run 10° cooler it would be a bit more silent.

 

reinstalling w10 is an option indeed, will do.

54 minutes ago, Latvian Video said:

Have you tried not tightening the cooler all the way? Overtightening can cause problems on cpus, so it probably also can do that on GPUs. 

Tried it, and after reading that not tightening it all the way could also cause problems I retightened it, but haven't forced anything. Should be fine.

 

Will reboot with a ramstick less since thats the quickest thing to check, after that reinstalling windows.

 

Thanks!

[leclod]

5 minutes ago, tofkop said:

IGpu works fine, haven't touched the ram so I don't think it will have anything to do with it.

Oh, true, you wrote vram

 

6 minutes ago, tofkop said:

Replaced the paste on the GPU since it hadn't been replaced in 5years, and if it could run 10° cooler it would be a bit more silent.

This seems fantasy to me, I don't believe paste gets old and dry. But if you've got a reliable source feel free to share.

 

And if you've got a tv with hdmi you could try it too.

[tofkop]

8 minutes ago, leclod said:

This seems fantasy to me, I don't believe paste gets old and dry. But if you've got a reliable source feel free to share.

Checked some info about my card specially, where most people got good results, and the paste I removed was dry.

8 minutes ago, leclod said:

And if you've got a tv with hdmi you could try it too.

Forgot to mention it but i tried it with another screen using the dvi output with same results.

 

Thanks for the suggestions!

[Mark Kaine]

3 hours ago, tofkop said:

Only thing left to try is baking the card I think

Nah, check if the gpu cooler is mounted correctly first, also the thermal pads.

Too much or not enough mounting pressure can cause these issues.

 

3 hours ago, leclod said:

Try reinstalling W10 ?

Also this.

 

 

[Mark Kaine]

1 hour ago, leclod said:

This seems fantasy to me, I don't believe paste gets old and dry.

not only can thermal paste dry out, it can also completely evaporate 

[tofkop]

22 minutes ago, Mark Kaine said:

Nah, check if the gpu cooler is mounted correctly first, also the thermal pads.

Too much or not enough mounting pressure can cause these issues.

 

Also this.

 

 

Just ordered some new thermal pads. Found a diagram on overclockers.net with the exact dimensions.

 

Ill try all the suggestions this weekend!

[Falkentyne]

2 hours ago, tofkop said:

Just ordered some new thermal pads. Found a diagram on overclockers.net with the exact dimensions.

 

Ill try all the suggestions this weekend!

Black screen after windows loads is almost always improper thickness spacing of thermal pads, leaving the GPU with no contact to the heatsink.  Many of these cards use two or even three different thicknesses on the same card, depending on model and AIB.  Even if you had very poor thermal paste, you wouldn't black screen at idle if there was at least proper contact.  Even poor contact of thermal pads to VRAM or VRM's shouldn't cause shutdowns at idle because they don't get stressed until you put a load on the card.

[leclod]

11 hours ago, tofkop said:

Checked some info about my card specially, where most people got good results, and the paste I removed was dry.

Forgot to mention it but i tried it with another screen using the dvi output with same results.

 

Thanks for the suggestions!

I watched that video, Linus only hints at some thermal paste can degrade over time.

9 hours ago, Mark Kaine said:

not only can thermal paste dry out, it can also completely evaporate 

Found this on Wikipedia :

The consistency of thermal paste makes it susceptible to failure mechanisms distinct from some other thermal interface materials. A common one is pump-out, which is the loss of thermal paste from between the die and the heat sink due to their differing rates of thermal expansion and contraction. Over a large number of power cycles, thermal paste gets pumped out from between the die and the heat sink and eventually causes degradation of thermal performance.^[5]

Another issue with some compounds is the separation of the polymer and filler matrix components occurs under high temperatures. The loss of polymeric material can result in poor wettability, leading to increased thermal resistance

 

[tofkop]

Tried installing ubuntu before reinstalling windows, and had artifacts during the installation.

Just finished replacing the thermal pads following this diagram with the same results.

 

I dont think its worth putting it in the oven, dont want to create a health risk. So yeah... lets see when i can afford a new gpu and the necessary hardware to run it.

 

Thanks for all the suggestions!

[Assistive]

On 7/15/2021 at 2:52 PM, Latvian Video said:

Have you tried not tightening the cooler all the way? Overtightening can cause problems on cpus, so it probably also can do that on GPUs. 

Bro, thank you so much! This advice just saved my STRIX 1080ti. 

[Latvian Video]

19 hours ago, Assistive said:

Bro, thank you so much! This advice just saved my STRIX 1080ti. 

No problem, I didnt imagine my post would help someone 8 months after its posted

[Assistive]

40 minutes ago, Latvian Video said:

No problem, I didnt imagine my post would help someone 8 months after its posted

I literally made an account just to tell you that it helped. I hope it can help others too. Good job my friend.