[SOLVED] Burnt Thermal pads on rx 580?

[klaudmjj]

I recently bought a second-hand RX 580. I took it apart today as I was just getting sick of it constantly ramping its fans up to 100% and I saw that one of the thermal pads, not sure what component it was but it looked burnt out.
I changed the thermal paste on the GPU die and that's about it but I didn't have any thermal pads to replace so I just left it. Should I be worried about it?
Here Is a picture its the 2nd from the bottom on the right:

https://imgur.com/a/V5jFEuA

View: https://imgur.com/a/V5jFEuA


For anyone wondering about results, ran Witcher 3 on high settings for about 30 mins, fans were running at 2900rpm (I believe 3100 is max), and the temp was hovering about 69-70c. After thermal paste change, fans ran at 1500rpm, and temps were hovering around 64-65c. The thermal paste used was noctua nt-h1.

 

[XxDarkMario20xX]

The temps is nothing much to worry about not unless its gose well over 85C and you could always use MSI afterburner to speed up fans and such if your still worried you can try buying some pads and replace them but buying 2nd hand stuff can be abit risky abit so would really have the seller to test it out in front of you first!

 

[Phaaze88]

RX 580 just has GDDR5 memory, so nothing to worry about. This type is easily cooled on fans alone.
Can't say the same with GDDR6 and 6X.

I don't know of any apps that can read RX 580 vram thermals yet?

 

[klaudmjj]

A weird thing has happened. I decided to test another game, risk of rain 2 and I uncapped my fps so my GPU usage would be 100%. Typically when my GPU gets to about 70c its fans kick up to about 100%. Now after the thermal paste change, it stays at around 1000-1500rpm and my temps rise to high 70's. I didn't touch any fan profiles. I just left everything default.

 

[Phaaze88]

"tested another game..." Ok, but you should compare both changed and unchanged settings. I don't expect Game A to behave like Game B, even if they're running the same graphics engine.
I don't know what to make of it otherwise.

 

[klaudmjj]

"tested another game..." Ok, but you should compare both changed and unchanged settings. I don't expect Game A to behave like Game B, even if they're running the same graphics engine.
I don't know what to make of it otherwise.

Before when I ran Risk of rain 2 uncapped, my gpu never went past 70 as the fans would be on 100%. Now the fans stay the same no matter the temperature.

 

[klaudmjj]

I don't know if this is related to my thermal paste change or not but;
After me testing the thermals on different games I saw that I had a driver update available. I downloaded it and proceeded to install the new drivers. Mid way through my screen went black as normal but it stayed black for a good 30 seconds, and I could see my cursor.
Then my pc just powered off as if someone had just unplugged it.
After my pc restarted, it took longer than usual, about 5-10 seconds longer to load into windows.
My 2nd monitor wasn't getting signal to it. I went into device manager and I saw that there was a problem with my gpu aswell as under "other devices" "Pci Data acquisition and Signal processing controller", "PCI memory controller" and "SM Bus Controller all had a problem with it. I have updated my windows to the latest drivers.

 

[Phaaze88]

Do you remember what driver version you were running previously?
Does your cpu have an integrated gpu? Remove the gpu, plug the monitor into the motherboard, and Display Driver Uninstall that new driver.
Download and install the older, working driver.

Before when I ran Risk of rain 2 uncapped, my gpu never went past 70 as the fans would be on 100%. Now the fans stay the same no matter the temperature.

The default fan profile works based on the average temp of the gpu's components. You 'fixed' the gpu cooler after the repaste.

 

[Flayed]

RX 580 just has GDDR5 memory, so nothing to worry about. This type is easily cooled on fans alone.
Can't say the same with GDDR6 and 6X.

I don't know of any apps that can read RX 580 vram thermals yet?

That pad is on the VRM not the VRAM

 

[klaudmjj]

Do you remember what driver version you were running previously?
Does your cpu have an integrated gpu? Remove the gpu, plug the monitor into the motherboard, and Display Driver Uninstall that new driver.
Download and install the older, working driver.



The default fan profile works based on the average temp of the gpu's components. You 'fixed' the gpu cooler after the repaste.

I managed to fix the issue,

I had to install my motherboard drivers again and that seemed to fix the problem and I was able to install the Radeon software again, this time it prompted me to either downgrade or repair, I clicked repair and it installed 21.2.2 without issues.

However now, whenever I tried to open Radeon software it would say, no AMD driver has been detected. Again I had to go to device manager and roll back my drivers to 21.2.1. Now everything works fine.

Now I am facing another issue, now when I load up into a game, my fans stay at 1000rpm until it reaches about 75c and then they start to rise to about 2000rpm where my temps stay on about 75c. If I max out the fans the temps drop to 70c which leaves me basically where I started. I want to replace the thermal pads on the GPU as I believe when I took apart my GPU, I broke that 'stock perfect' bond the thermal pads had and maybe that's what is causing my issue, the question is what thickness thermal pads should I use? I can only find 1mm or 0.5mm on amazon as I want it to come tomorrow next day, I've read online that my specific card uses 0.75mm pads, so I think using 0.5mm pads would be better as I want the GPU die to contact the heatsink perfectly.

 

[Phaaze88]

That pad is on the VRM not the VRAM

/blush
Well that's embarrassing...

Now I am facing another issue, now when I load up into a game, my fans stay at 1000rpm until it reaches about 75c and then they start to rise to about 2000rpm where my temps stay on about 75c. If I max out the fans the temps drop to 70c which leaves me basically where I started.

What's the rest of your hardware, and how are the fans set up?

I want to replace the thermal pads on the GPU as I believe when I took apart my GPU, I broke that 'stock perfect' bond the thermal pads had and maybe that's what is causing my issue...

I'm thinking the problem may be more related to the 'browned' MOSFET than the actual thermal pads. The thermal pads can be reused if they're not torn.
If the MOSFET is damaged, that's another story. Are you able to clean off the brown on it? Does it look any better on the opposite site of the PCB?

 

[klaudmjj]

/blush
Well that's embarrassing...


What's the rest of your hardware, and how are the fans set up?


I'm thinking the problem may be more related to the 'browned' MOSFET than the actual thermal pads. The thermal pads can be reused if they're not torn.
If the MOSFET is damaged, that's another story. Are you able to clean off the brown on it? Does it look any better on the opposite site of the PCB?

i5 7600k not oc
ASRock z170 extreme 4
DDR4 running at 2800mhz
500GB nvme SSD
1TB HDD
RX 580 Pulse 4GB
EVGA 650w 80+ gold PSU (Supernova G2 I believe)
The case is the Corsair Icue 4000x with stock fan config, 3 intakes. Custom fan profile fans reach 100% when 60c is reached on cpu.
I didn't touch the fan profile on gpu.
I cleaned the actual chip that was browned and it didn't seem damaged, ill take it apart again and take more detailed pictures.

 

[Flayed]

I don't think that 580 will have a long and happy life. Maybe you should sell it before it dies lol

 

[klaudmjj]

https://imgur.com/a/gaA875M

View: https://imgur.com/a/gaA875M


Here are some more pictures, the yellow highlighted area is where the card backplate was significantly hotter compared to other parts of the backplate. I have made a mistake when taking it off and I forgot about the two little screws which hold the thermal pads that connects the back of the MOSFETs to the backplate. After cleaning off the brown residue I was able to see the writing and I can't see any damage. I did dab a little bit of thermal paste on that mofset and also put a bit more thermal paste on the gpu die.

EDIT: Okay this might explain why I'm seeing higher temps, my power consumption on the gpu now is 150W. Before, it used to be between 120-130W. I don't know why the power consumption has increased. However when I change the power limit to -15% , the power consumption is now around 130W, and the temps stay at 70c max with the rpm of the fans at 50%.

 

[Phaaze88]

Pic 1: That is the VRM area that's circled.

Pic 2: It's not very clear... next.

Pic 3: Looks normal. Pads are a bit dirty at the edges, but GDDR5 ram isn't hard to cool.

Pic 4: Ok.

Pic 5: The area around the cap to the right of the 'browned' MOSFET is noticeably darker. Source of the leak?

Pic 6: In the VRM area, I see a mixture of torn thermal padding, hair, dirt, and oil(from thermal pads). Some of the browning is noticeably darker - burns or simply higher concentrations of oil?
It's normal for pads to leak oil and darken areas of the PCB - the oil is harmless.


Definitely need to clean up that area on the back of the PCB and on the backplate - torn thermal padding is useless and needs to be replaced... but that's assuming that cap also didn't leak.

 

[klaudmjj]

Pic 1: That is the VRM area that's circled.

Pic 2: It's not very clear... next.

Pic 3: Looks normal. Pads are a bit dirty at the edges, but GDDR5 ram isn't hard to cool.

Pic 4: Ok.

Pic 5: The area around the cap to the right of the 'browned' MOSFET is noticeably darker. Source of the leak?

Pic 6: In the VRM area, I see a mixture of torn thermal padding, hair, dirt, and oil(from thermal pads). Some of the browning is noticeably darker - burns or simply higher concentrations of oil?
It's normal for pads to leak oil and darken areas of the PCB - the oil is harmless.


Definitely need to clean up that area on the back of the PCB and on the backplate - torn thermal padding is useless and needs to be replaced... but that's assuming that cap also didn't leak.

If the cap could be the source of the leak, what does that mean. I agree that area does look a bit darker compared to the other ones,

 

[Phaaze88]

That would be a possible cause of the card crashing or becoming unresponsive under load.

 

[klaudmjj]

That would be a possible cause of the card crashing or becoming unresponsive under load.

Wow okay, that makes a lot more sense now, at the start when I got the card probably a week into having it, I did get a lot of just random black screens for a couple of seconds, occasionally the screen would just stay black until I forced a restart. It has stopped doing that, I don’t know why it has stopped however I am still confused on why the power consumption has increased. Any ideas? And is the leak something I should really worry about?

 

[Phaaze88]

Power consumption increased either due to the drivers, or because of enhanced cooling improving gpu performance... but I've not known the old RX 500 series cards to do that.

As for the leak: it'll be on it's way out the door sooner or later. It'll fail and fail until it inevitably croaks.

 

[klaudmjj]

Power consumption increased either due to the drivers, or because of enhanced cooling improving gpu performance... but I've not known the old RX 500 series cards to do that.

As for the leak: it'll be on it's way out the door sooner or later. It'll fail and fail until it inevitably croaks.

Thank you for your help. Learnt a lot actually. I guess my plan is now to sell the card while it works and get an upgrade perhaps.

 

[klaudmjj]

Right so the new thermal pads came, I got some information on my specific card, they use 0.75mm Pads on the GDDR5 and 1mm pads for the MOSFETs. Couldn't find any 0.75mm pads, but since GDDR5 is not too hard to cool I wasn't worried. Replaced the MOSFET ones and also the thermal pad between the backplate and the card itself and viola! It Maxed out at 72c with the fans spinning at 50%. The average was 70c.
I inspected the caps for leaks and I couldn't find any.

 

[Phaaze88]

Cool! So it was just the mangled padding.