Evga 3080Ti XC3 Red Light Problem

[igotek]

I replaced the thermal pads and paste with a better ones. (6 pcs in a rig + 6 pcs in a rig)
I power on the rigs. One rig reboot itself after 24 minutes mining. I saw GPU2 has an error in HiveOS screen. Then GPU2 is no longer exist in my motherboard.
I tested it severa ltimes with different risers and motherboards. No help. Then I realised a red light under the PCIE power sockets. There are 3 red light.

One is for PCIE power 1 socket. I m sure, the first light is on if i dont plug the first PCIE power socket. Secod is for PCIE power 2 socket. I m sure, the second light is on if i dont plug the first PCIE power socket.

Third is for ? I dont know what is for. The third red light is on all the time. I think there is an error. I m not sure i can RMA or not because I change thermal pads and paste. What should the problem is ? Any idea ?

[arielbit]

Do you have a modular psu? Check each wire connection, not just the whole socket.

[philipma1957]

I replaced the thermal pads and paste with a better ones. (6 pcs in a rig + 6 pcs in a rig)
I power on the rigs. One rig reboot itself after 24 minutes mining. I saw GPU2 has an error in HiveOS screen. Then GPU2 is no longer exist in my motherboard.
I tested it severa ltimes with different risers and motherboards. No help. Then I realised a red light under the PCIE power sockets. There are 3 red light.

One is for PCIE power 1 socket. I m sure, the first light is on if i dont plug the first PCIE power socket. Secod is for PCIE power 2 socket. I m sure, the second light is on if i dont plug the first PCIE power socket.

Third is for ? I dont know what is for. The third red light is on all the time. I think there is an error. I m not sure i can RMA or not because I change thermal pads and paste. What should the problem is ? Any idea ?

Try a one gpu rig with a completely different psu.
see if it works.

try it on a windows rig just the 1 card.

and once again do not use the original psu.

[batsonxl]

Did you check the temps of card. Something getting hot there maybe one gpu not touching properly and overheats.

[igotek]

I watched the temp of the graphic cards in HiveOS. I know there was no any temp problem.

PSU isnt modular one. I tried the same PCIE power sockets in another gpu - no problem. The PCIE sockets is well.

I also tried it with different psu. I also tried it in Windows. The same problem.
I read many articles at evga forum. Someone says one of the fuse on the graphic card blew.
is there any way to find out which one ?

I also mailed the problem to evga support to learn RMA is valid or not because I replaced the thermal pads and the thermal paste.
The graphic cards are newly bought - just one week ago. - amazon.

Problem: Nothing shows in PCIE slot and red lightis on under the PCIE power socket of the graphic card.

[FP91G]

After replacing the thermal paste and thermal pads in an unauthorized service, the RMA procedure in 99% does not work.
Using a multimeter, you need to check the fuses on the video card. Just because a fuse is bad doesn't mean it's the problem. You can solder a thin wire and test the video card, but usually the test of the video card is done using a laboratory power supply unit with a reduced current so as not to burn other elements in the event of a short circuit.

[igotek]

After replacing the thermal paste and thermal pads in an unauthorized service, the RMA procedure in 99% does not work.
Using a multimeter, you need to check the fuses on the video card. Just because a fuse is bad doesn't mean it's the problem. You can solder a thin wire and test the video card, but usually the test of the video card is done using a laboratory power supply unit with a reduced current so as not to burn other elements in the event of a short circuit.

I use napkin to clean the thermal pads pieces. (Some thermal pad pieces is really hard to remove.)
I use toothbrush to remove all napkin pieces.
Then I use alchol to clean. I always change all pads on graphic cards. (Not only memory side.)
Normally, I m so carefull. But my litlle sweety 11-month-years old baby was crying. I was in a rush that graphic card.
I think there was a little piece napkin. I m not sure.

24 minutes ETHW mining was good. Then rig reboot itself and I saw that gpu has an error.
Maybe the napkin burnt and short circuit.
I clean the graphic card properley again. There is no burnt parts.
Maybe fuse protection. I think I can change the fuse or short circuit, it will work.

I know it s my fault. I want to fix it. I was used to repairing cell phones at the past. I have tools for it.
But i never repair graphic card. By the way, This graphic card has two pcie power socket.
First one is for PCIE socklet 1, Second one is for PCIE socket 2.
Third one is for report a problem. But what problem ? Shor circuit ?

How could I learn which ones are fuse ?


Rig1:


Rig2:


Graphic Card:

[arielbit]

Maybe the 1 sec of LED1 of other graphics card is POST (power on self test).

You used a tooth brush, are you sure the hard part of the brush didn't hit a component like a resistor and removed it from its solder?

[igotek]

Maybe the 1 sec of LED1 of other graphics card is POST (power on self test).

You used a tooth brush, are you sure the hard part of the brush didn't hit a component like a resistor and removed it from its solder?

Sure, there is no any damage.

There are R005 fuse, one bottom back to pcie, 5 front next to pcie power sockets. I get bip signal when i measure with multimeter.

[JayDDee]

Have you tried the GPU in a X16 slot? That will confirm it's the GPU at fault or not. I haven't found any spec for those lights
but it seems like they represent the 3 sources of power for the card: the slot and the 2 aux connectors. That suggests there's a problem
with the slot power source.

Edit: there are only 2x8 pin aux connectors for that 3080ti, that's at the limit. Make sure you're using risers with 6 pin
power and not cheating with any splitters.

[philipma1957]

Have you tried the GPU in a X16 slot? That will confirm it's the GPU at fault or not. I haven't found any spec for those lights
but it seems like they represent the 3 sources of power for the card: the slot and the 2 aux connectors. That suggests there's a problem
with the slot power source.

Edit: there are only 2x8 pin aux connectors for that 3080ti, that's at the limit. Make sure you're using risers with 6 pin
power and not cheating with any splitters.

I told him to do a 1 gpu windows test with a different psu.

not sure he did that.

But I think he did try different mobos with different risers.


If he has an identical card he could measure and comare readings on the good card to the bad card.


https://www.youtube.com/watch?v=2o2Pu9cP7N0


some info could apply

[JayDDee]

Have you tried the GPU in a X16 slot? That will confirm it's the GPU at fault or not. I haven't found any spec for those lights
but it seems like they represent the 3 sources of power for the card: the slot and the 2 aux connectors. That suggests there's a problem
with the slot power source.

Edit: there are only 2x8 pin aux connectors for that 3080ti, that's at the limit. Make sure you're using risers with 6 pin
power and not cheating with any splitters.

I told him to do a 1 gpu windows test with a different psu.

not sure he did that.

He apparently has multiple identical cards so swapping the bad card with another card (only the cards, remove them from the risers) would quickly identify whether the card is at fault.

Plugging a suspect card into an x16 slot reliably rules out any riser issues (power or data) for any card problems. Running a different card in the
same slot with the same risers would rule out a slot issue as well as a riser issue.

Since the cards are (assumed) identical it should also rule out a power issue.

Eliminate everything it isn't and you're left with what it is.

[rdluffy]

I'm interesting and curious to see what's the problem of your card too.
I follow a channel on YouTube of the techs who do the RMAs for the brand Galax, and they are great tech and modders, they have access to document sheet, I tried to find if there is any source of sheets of EVGA, but I think it's exclusive and not shared to public.
Since you are familiar with cel phones repairing, it could help, but I have no success.

A piece of paper napkin is unlikely to cause a short circuit.
As a miner I did a lot of troubleshot, and I suggest you to wait the asnwer of EVGA, their support are good, maybe a tech can assist you.
Another thing you can do is to disassembly the GPU and mount again, simple but sometimes worked with me.

Trying the GPU alone in motherboard with 16x slot too like users already said.

[FP91G]

How could I learn which ones are fuse ?

If you don't know where the fuse is, then don't proceed with the repair. There may or may not be fuses, you need to look at the diagram. If there is no video card cdiagram, then you can check the resistance of each power circuit after the throttle for a short circuit.
This may be your mistake, or it may not be yours, but just a coincidence. You need the help of a good master and not advice on the forum.

[igotek]

Have you tried the GPU in a X16 slot? That will confirm it's the GPU at fault or not. I haven't found any spec for those lights
but it seems like they represent the 3 sources of power for the card: the slot and the 2 aux connectors. That suggests there's a problem
with the slot power source.

Edit: there are only 2x8 pin aux connectors for that 3080ti, that's at the limit. Make sure you're using risers with 6 pin
power and not cheating with any splitters.

I told him to do a 1 gpu windows test with a different psu.

not sure he did that.

But I think he did try different mobos with different risers.


If he has an identical card he could measure and comare readings on the good card to the bad card.


https://www.youtube.com/watch?v=2o2Pu9cP7N0


some info could apply

Sure, I tested it  x16 slot with different psu and mobo on windows 10.
I will send it for warranty. I will pay if they want for repair.

[rdluffy]

Sure, I tested it  x16 slot with different psu and mobo on windows 10.
I will send it for warranty. I will pay if they want for repair.

EVGA accepted your invoice to possible RMA?
I always heard good things about EVGA warranty, I know they allows thermal paste / pads replacement without lose warranty, and this is a good reason why modders usually prefer EVGA.
Here in my country Galax also accepts, all another brands say you lose warranty if you void the adhesive seal or open the card (Gigabyte, MSI, Zotac etc)
Good luck, maybe it is a faulty gpu and was only coincidence

Keep us updated

[igotek]

Evga advised me to contact local amazon for warranty.
Evga accepts to check it for RMA if local amazon doesnt help me about it.
But there is international shipping fee + custom fees.
I will write what will happens.

[philipma1957]

Evga advised me to contact local amazon for warranty.
Evga accepts to check it for RMA if local amazon doesnt help me about it.
But there is international shipping fee + custom fees.
I will write what will happens.

Good luck.

[FP91G]

Evga advised me to contact local amazon for warranty.
Evga accepts to check it for RMA if local amazon doesnt help me about it.
But there is international shipping fee + custom fees.
I will write what will happens.

How long will your video card spend on the road?
If EVGA is loyal to the replacement of thermal pads and thermal paste outside of authorized services, then it can do diagnostics in a good service and check all power lines for a short circuit before shipping. If the repair will be expensive, then send the video card to the EVGA warranty service.

[jagraven]

I have the same problem on an EVGA FTW 3080Ti - red power light on the card.  I too have replaced the thermal pads and paste, but it was a good few months before this problem arose. I can't remember if I tried it in a full X16 slot so I'm going to give it a try, but interesting the same fault is appearing on two different cards.