Help to fix/diagnose Gigabyte GTX 980TI - repair[Solved]

[Chaos666]

Hi ...

I know its a long shot .. but...
I got Gigabyte GTX 980TI Gaming G1. It just collects dust in my closet. it just stopped working one day. I disassembled the card. I checked for burn marks etc... Nothing. When I put the card in my pc it won't boot. Fans are spinning on the GPU but pc won't boot. I suspect there is a short somewhere or something.. I'm a pc enthusiast and I have some basic electrical knowledge but I don't know to much about GPU pcb. Is there someone who would help me fix or at least diagnose the card? I'm just curious what is wrong with it.
I own a digital multimeter and can make hires photos of the card if needed.

this is the model I'm talking about

 

[tyler_jrb]

Basics is to either stick the card in and using a multi-meter see what voltages you get on the main vrm's on the pcb and go from there.

Or with the card removed measure resistance values on the core + memory etc.

Core will be in the single ohms range. Memory will be much higher in the hundreds usually.

You can also measure that you have all correct voltages throughout the board with the backplate removed. If i have the correct pcb.

Go from ground to any of the large 8 black smd capacitors (ones in a line), should be around 1v.

Go from ground to both legs of any of the 6 memory vrm capacitors about 1 inch below the main power connectors sticking through the pcb. Usually around 1.3-1.35v.

Much easier on the core side of the pcb though as you can easily measure from the coils.

After that its just a lot of diagnosing everything else. Main powers 3.3v,5v,12v. And if you have a short diagnosing and repairing. If no short checking drivers etc.

MUCH easier to watch Elitech on youtube as he does a lot of gpu repairs which i think will help you out a lot.

 

[LeMson]

To the oven and reflow the silicon!

 

[LuckyBenski]

To the oven and reflow the silicon!

Amazing.

 

[Chaos666]

I'm gonna revive my own thread..
After a few years in my closet I decided to give it another go..
I was able finaly to boot pc with this card in the second slot. Bios was corrupted so I flashed it.
Card seems to work now but clock fluctuate a lot.
Memory goes from 3500 to 810... And vddc is jumping from 0.8 to 1.1

 

[str]

I'm gonna revive my own thread..
After a few years in my closet I decided to give it another go..
I was able finaly to boot pc with this card in the second slot. Bios was corrupted so I flashed it.
Card seems to work now but clock fluctuate a lot.
Memory goes from 3500 to 810... And vddc is jumping from 0.8 to 1.1

The main issue will be two ferrite beads on the card that are very likely to be degraded and they will eventually blow which could result in significant damage to the PCB and components. The solution is to very carefully desolder the ferrite beads and then fix up the card so it works again. There is a YT video about replacing the ferrite beads on the 980TI G1 where it's suggested to replace one of them with a 0 ohms resistor or a solder bridge to get the card working again.

After I resolved the ferrite beads issue, I still had fluctuating clocks and voltage that I was able to reliably reproduce using a benchmark and it was randomly occurring in games. After diagnosing everything in the PC, it turned out to be my Z97 ITX motherboard had a fault that was causing the GPU to stall.

There is still the possibility that a design fault is causing the card to degrade in other ways which could mean it will eventually have problems again but mine has been watercooled in a Ryzen build for a while now and it's working perfectly.

 

[Bluntwrapped]

I'm gonna revive my own thread..
After a few years in my closet I decided to give it another go..
I was able finaly to boot pc with this card in the second slot. Bios was corrupted so I flashed it.
Card seems to work now but clock fluctuate a lot.
Memory goes from 3500 to 810... And vddc is jumping from 0.8 to 1.1

Sounds like pretty normal behaviour. Depending on which power profile the card is using (sitting idle windows desktop > 3D) the clocks will fluctuate. If while running the 3D profile (P0 if I remember correctly) I’d guess it’s a driver problem and potentially thermal throttling. If both core and mem fluctuate together it’s probably more likely to be driver related. Which bios are you using on the card and how did you flash it?

 

[Chaos666]

The main issue will be two ferrite beads on the card that are very likely to be degraded and they will eventually blow which could result in significant damage to the PCB and components. The solution is to very carefully desolder the ferrite beads and then fix up the card so it works again. There is a YT video about replacing the ferrite beads on the 980TI G1 where it's suggested to replace one of them with a 0 ohms resistor or a solder bridge to get the card working again.

After I resolved the ferrite beads issue, I still had fluctuating clocks and voltage that I was able to reliably reproduce using a benchmark and it was randomly occurring in games. After diagnosing everything in the PC, it turned out to be my Z97 ITX motherboard had a fault that was causing the GPU to stall.

There is still the possibility that a design fault is causing the card to degrade in other ways which could mean it will eventually have problems again but mine has been watercooled in a Ryzen build for a while now and it's working perfectly.

that was exactly the case... i found this video

and fixed the card myself
after soldering a wire to see if its the ferrite beads card boosts normaly to 1450Mhz and memory stays at 3500
https://www.3dmark.com/3dm/67511364

then i desolder both ferrite beads and added short wires to connect the pads (didn't took any pics though).
I am just waiting for my 0 ohm resitors to come in so i can fix it properly. It's nice to repair a gpu especially now when its so hard to get them.

thx for the answer anyway...