"Fixed" a shady GTX 1080 today. It can't be that easy, right?

[koralski]

Hi to all, hope your day is going great.

I bought a GTX 1080 EVGA today, for 5 euro. The card had a reported problem of "crashing with a black screen and artifacts" under load.

I got the card in my hands and instantly tried undervolting and underclocking it, to no avail. It still crashed when I entered Furmark. I flashed the latest VBIOS for the card, that still didn't do the job. And then as a final try, I decided to look at the board.

This is what greeted me.

A stinky dusty card with what looks like cheap thermal paste and flux around the die. I cleaned up the card, the power phases, VRMs ; etc. Changed the thermal paste and tightened the back screws as needed. (They were loose before.)

And, OH WOW! IT WORKED!

It can't be that easy, right? Tell me what was the culprit? Was the memory overheating due to not making sufficient contact with the thermal pads? Was it just luck on my part?

What was it, I need suggestions.

Thanks for reading

 

[StViolenceDay]

How fast it was crashing?

If it crashed immediately on enterint furmark - the issues is not thermal-related. Card with a cooling in the form of the coin on the GPU would be able to run furmark for 3-5 seconds before overheat. Lack of GDDR5X cooling has even slower effect, many-many minutes. The rule of thumb that overheating effects are never "immediate"/"the same second".
The sudden "healing" of immediate crash can be caused either by "temporal fix of some unstable contact due to tightening", or some dirt had low resistance and touched solder ball used foe high-speed memory communication.

If it was crashing in 5+ seconds - in addition of already listed causes - it may be "not even cooler contact". 1080 has a single thermal sensor on the die, so it may didn't know that other chip corner was extremely overheated (but this is more common for 1080tis with a huger die)

In general - use it for a month to see if the fix is temporal or stable

 

[koralski]

How fast it was crashing?

If it crashed immediately on enterint furmark - the issues is not thermal-related. Card with a cooling in the form of the coin on the GPU would be able to run furmark for 3-5 seconds before overheat. Lack of GDDR5X cooling has even slower effect, many-many minutes. The rule of thumb that overheating effects are never "immediate"/"the same second".
The sudden "healing" of immediate crash can be caused either by "temporal fix of some unstable contact due to tightening", or some dirt had low resistance and touched solder ball used foe high-speed memory communication.

If it was crashing in 5+ seconds - in addition of already listed causes - it may be "not even cooler contact". 1080 has a single thermal sensor on the die, so it may didn't know that other chip corner was extremely overheated (but this is more common for 1080tis with a huger die)

In general - use it for a month to see if the fix is temporal or stable

5,6,7,8,9 seconds. It was not instant, but for sure it wasn't minutes. Thanks for the clarification.

The thermal progression was less of a progression and more of a spike. The card idled at 40-42 degrees Celcius, and then instantly it spiked to 67,68 and crashed.

I ran Furmark for 20 minutes and after cleaning and changing the thermal paste the progression is smooth. With a top peak of 70 degrees Celcius.

It has indeed, not been fixed.

Should I try and reflow it? Any suggestions?

Here's what the PCB looks. I tried the card today and it crashed. Not instantly but still crashed.

 

[chinmi]

what's this part? is it a wire that connect those 2 parts?

 

[StViolenceDay]

Reflow may help if your problem is related to damaged (unstable contact) solder ball under core/VRAM.

When thay is the case - the card typically reacts to a board bending during test run. Not an extreme bending, but like "bend one corner moving it 1.5cm, fixing others in place", just with your fingers. Try a lot of different bending directions and angles. If the problem is with solder balls - typically some bending can be found that leads to immediate crash.

If no bending leads to a crash - reflow is rare useful.

 

[koralski]

what's this part? is it a wire that connect those 2 parts?

View attachment 400602

No it was a little hair

Reflow may help if your problem is related to damaged (unstable contact) solder ball under core/VRAM.

When thay is the case - the card typically reacts to a board bending during test run. Not an extreme bending, but like "bend one corner moving it 1.5cm, fixing others in place", just with your fingers. Try a lot of different bending directions and angles. If the problem is with solder balls - typically some bending can be found that leads to immediate crash.

If no bending leads to a crash - reflow is rare useful.

I tried that, to no avail. It just instantly crashes. I think that one is a writeoff. Still a good deal for the shroud. I'll probably use it as a donor/sell it.

 

[rusty caterpillar]

I tried that, to no avail. It just instantly crashes. I think that one is a writeoff. Still a good deal for the shroud. I'll probably use it as a donor/sell it.

Send it to North West Repair(YT channel) but is in US, he will re ball(reflow never does much) it and fix it, not sure worth the cost, ask him first maybe.

 

[dgianstefani]

Send it to North West Repair(YT channel) but is in US, he will re ball(reflow never does much) it and fix it, not sure worth the cost, ask him first maybe.

Absolutely not worth the cost, even just of shipping.

If was 2080 Ti or better, maybe.

 

[Macro Device]

Not a very educated guess but since your GPU crashes no matter the angle it seems you have a problem in VRM. Test it with a multimeter.

 

[koralski]

Not a very educated guess but since your GPU crashes no matter the angle it seems you have a problem in VRM. Test it with a multimeter.

How do I measure it? How do I conclude something's wrong with it?

 

[Macro Device]

How do I measure it? How do I conclude something's wrong with it?

Funny part is I know what to do but I don't know how to do it. Watch some Youtube tutorial or ask someone who knows their tools. Good luck!

 

[Marcus L]

If all else fails bake that bad boy, done this on a couple of occasions and they continued to work for quite some time afterwards, would only recommend as a last resort and on an older chap card like this where you don't stand to lose much

 

[Shrek]

Reflow may help if your problem is related to damaged (unstable contact) solder ball under core/VRAM.

Really need to use flux for a good reflow.

 

[Beertintedgoggles]

If all else fails bake that bad boy, done this on a couple of occasions and they continued to work for quite some time afterwards, would only recommend as a last resort and on an older chap card like this where you don't stand to lose much

I was about to say the same. If you have an actual hot air gun you can do a bit better with that but I've baked a few 8800/9800GT Nvidia cards back in the day that were known for solder issues with a surprisingly high success rate (I actually can't remember when it didn't work).

 

[Marcus L]

There are some basic guides floating around though I couldn't be bothered to take the time to find one, just asked Grok and got this reply:
seems about right to me, I may have left mine in for 10 mins thinking back but it was probably 10 years or so go so details are hazy. But hey at $5 nothing risked nothing gained, would be awesome if you can bring it back to life and a proper bargain, decent for a 2nd 1080p gaming rig with current AAA exceptions it should still hold up really well.

Baking a non-working GPU is a last-ditch effort to potentially reflow solder joints that may have cracked or become loose, often due to overheating or age. This process is not guaranteed to work, can further damage the GPU, and may void any remaining warranty. Since you've exhausted all other options and are okay with the potential loss, here’s a concise guide based on common practices for this unconventional repair method:
Temperature and Duration:

• Temperature: 385°F to 400°F (195°C to 205°C). This is close to the melting point of lead-free solder (around 217°C) used in modern electronics, allowing reflow without damaging components.
• Duration: 8 to 10 minutes. This ensures the solder reaches reflow temperature without overheating other components.

Steps:

1. Preparation:
   • Preheat a conventional oven to the specified temperature. Use an oven thermometer to verify accuracy, as oven dials can be imprecise.
   • Disassemble the GPU, removing the cooler, heatsinks, and any plastic components to prevent melting or damage.
   • Clean the GPU board with isopropyl alcohol (90% or higher) to remove dust and residue.
   • Place the GPU on a baking sheet, supported by balls of aluminum foil to elevate it slightly and ensure even heat distribution. Avoid direct contact with the baking sheet.
2. Baking:
   • Place the GPU in the preheated oven, ensuring it’s level and stable.
   • Bake for 8 to 10 minutes. Monitor closely to avoid exceeding this time, as prolonged heat can damage components.
   • Optionally, place a small piece of solder or a temperature strip on the board to confirm it reaches reflow temperature (it should melt slightly).
3. Cooling and Reassembly:
   • Carefully remove the GPU using oven mitts and let it cool naturally at room temperature for at least 30 minutes. Do not rush cooling with fans or refrigeration, as rapid temperature changes can cause thermal shock and further damage.
   • Reassemble the GPU, applying fresh thermal paste to the GPU chip before reattaching the cooler.
4. Testing:
   • Reinstall the GPU in your system and test it. If it works, monitor for stability, as this fix may be temporary.

Precautions:

• Perform this in a well-ventilated area, as heating electronics can release fumes.
• Avoid using a microwave or toaster oven, as they provide uneven heating and can cause arcing or further damage.
• Be aware of the risks: overheating can degrade components, and there’s no guarantee of success.

This method has been discussed in tech communities (e.g., posts found on X and web forums like Tom’s Hardware) as a folk remedy for GPUs with solder-related issues, but it’s not a professional repair technique. If it fails, the GPU may be permanently damaged, but since you’re prepared for that outcome, it’s a low-risk experiment for a non-expensive card.

 

[Shrek]

I suggest NOT using an oven that is used for food (I use a hot air gun).

 

[Marcus L]

I suggest NOT using an oven that is used for food.

I like to live on the wild side

 

[Shrek]

I tend to wick flux below the chip before heating with a hot air gun, so avoiding using an oven.

 

[rusty caterpillar]

If all else fails bake that bad boy, done this on a couple of occasions and they continued to work for quite some time afterwards, would only recommend as a last resort and on an older chap card like this where you don't stand to lose much

Reflow don't last and with the oven you baking stuff that can't take that heat(for example small SMD), IMO is just waste of time.

 

[Shrek]

Works great if one uses flux and a hot air gun lets one concentrate the heat.

 

[Marcus L]

Reflow don't last and with the oven you baking stuff that can't take that heat(for example small SMD), IMO is just waste of time.

Countless examples of people doing it successfully would disagree, I would never recommend this for an expensive GPU though if you buy cheap and it's the last option it can and does work in certain circumstances, hence the time and temperature has to be just right, in the OP case he spent $5 so if everything else fails he loses nothing, I've done it on a couple of occasions and used them for months afterwards obviously you YMMV but that's the territory you're in when you decide to bake a GPU lol

 

[rusty caterpillar]

Countless examples of people doing it successfully would disagree, I would never recommend this for an expensive GPU though if you buy cheap and it's the last option it can and does work in certain circumstances, hence the time and temperature has to be just right, in the OP case he spent $5 so if everything else fails he loses nothing, I've done it on a couple of occasions and used them for months afterwards obviously you YMMV but that's the territory you're in when you decide to bake a GPU lol

I would put it this way, he gambled 5 bucks, he can gamble some time and energy now, toxic fumes to

 

[Marcus L]

I would put it this way, he gambled 5 bucks, he can gamble some time and energy now, toxic fumes to

Too* stop talking rubbish though, not like he's baking it at 400c and leaving it in the oven for 2 hours, stop moving your goalposts, obviously you have "I know better complex"

 

[CheteriagalenTriagalnik]

You paid 5$ for the card and it doesn't work anyway. If you wanna try your luck go ahead. Just make sure you're in a well-ventilated area and take precautions around fire if you decide to go through with it.

Another thing that could be worth trying is putting it through an ultrasonic cleaner - the type usually used to clean jewelry, then dipping it in alcohol and letting it dry thoroughly. The problem here is you'll struggle to find one that can actually take a full GPU.

As an owner of a 1080 that still works perfectly, it would be a pretty good deal if you actually got it to work. And if you break it even more - well, you only lose 5 bucks.

 

[Shrek]

However, before reflowing, I'd test the VRAM

One chip looks suspect