I bought a GTX 1070 a while back, but never put it to use because it had a broken fan. One of the two fans did not spin, and it was the fan that has the tachometer signal cable coming out of it.
Obviously, this meant that the GPU would not register fan-speed, so the fan-controller would ramp up the speed of the other fan to 100% even though the GPU was almost room-temperature (40C/100F) and no load.
I have since replaced the fans, and I know for a fact they are working correctly because Iāve tested them separate from the GPU, even the tachometer signal is working fine.
The GPU still does not register fan-speed, so I fear that the fan-controller is broken.
Itās for sure not a software problem. Iāve tested and re-tested a bunch of drivers, even on a fresh install of Windows. Iāve updated the firmware on the card to the latest version. That all does not seem to work. Iāve used all kinds of monitoring software, and they all report ā0 rpmā when I hear the fans (both, since theyāre repaired/replaced!) spinning at 100%.
Does anyone have any idea how to fix this? Can it be fixed?
I know the hack of using an adapter cable and plugging the fans into a motherboard CPU connector. Thatās not what Iām looking for.
I actually intend to sell the card since I have no direct use for it myself, but so far nobody seems interested (locally at least). So if I could fix this, I would have a much better chance I believe.
The warranty on this card has expired, Iām afraid, so itās either repair it myself, or sell it very cheaply if at all.
Could be a broken trace or cold solder joint on/near the fan header or controller.
In theory, you can run a bodge wire from the pin on the header to the corresponding pin on the controller. The difficulty will be finding said trace. A multimeter is not necessarily going to help.
I have re-flowed the solder on the fan-connector on the back of the PCB but that didnāt fix the issue, sadly. So I tried to find out where the signal was going.
The PCB has about 10 layers, and I see no connection on the two outer-most layers except for the ground, so I honestly have no idea where the internal traces might be going. I canāt find a schematic for the PCB online either, so I think this is impossible to repair for me.
Also, have you tried looking up datasheets for the ICās nearby to those connectors, to find out which control the fan speed (if it is any; hopefully the GPU die isnāt the thing controlling fan speedsā¦)?
The parts on the front near the fan header (top left header is for LEDs), are similar to the ones near the PCIe-power connectors. So I guess they are power conditioning.
Most likely was a resistor that was ripped off the board, or a cap shorted to ground, causing the fan control to not work properly. But considering how much cheaper youāll sell the card, it may not be worth your time to try to find the problem and fix it.
If you did want to fix the card, the quickest thing to do would be to visually inspect that back corner and see if anything looks like it was ripped off, and/or use rubbing alcohol in that area, and seeing where the water evaporates first, showing where a possible short to ground is.
Edit: Hereās what the pads could look like if a capacitor/resistor was ripped off the PCB
Further, the card has a 2mm steel backplate protecting the board, which looked like it was never removed before (considering the screws were very tight and in factory-condition). The orange circle is one of the screw holes of the backplate.
The āspotsā you see on the bottom and right are dust particles that were trapped between the PCB and backplate, no water damage or anything.
ā
I actually just managed to sell the GPU to someone (ā¬200 / $230 for those wondering), so I wonāt pursue this further. Next time Iām in a similar position, I might try the alcohol trick. Thanks
I would have tried that, but I bought a new 1070 Ti while I was waiting for the replacement fans to arrive. Since I have a Ryzen CPU, I didnāt have integrated graphics. So when I repaired the 1070, I didnāt have any use for it anymore.