MP 1,1-5,1 Mac Pro 5.1 randomly switch off

[goodbye]

Hi to everyone, I'm new here, but I've been reading a lot in this forum, so thanks for all the helps that I got here.
I hope i'm posting this in the right place.
I've a problem with my Mac Pro 5.1:

Mac Pro 5.1 mid2012 (1.39f11)
2x3,46Ghz 6-core intel xeon
128 Gb ram
NVIDIA GeForce GTX 980 Ti 6143 MB Flashed (387.10.10.10.40.105) powered by boostA&B Aux (6pins+6pins to 8pins) and by a SATAto6pins connector.
NVIDIA GeForce GTX 120
USB 3.0 PCIe
boot from a Sata SSD HIGHSIERRA-17G65
RAID0 with 16Tb.

Suddenly my Mac Pro logs out, many time every day. The screen get black and the login screen appears. It happends randomly. One week ago it started switching off randomly, many times. I opened it, i cleaned the fans, i check with TG Pro the sensor temperature (X58 IOH 70ºC-76ºC), I fresh install OS (no timemachine), but the problem goes on.
The apple hardware test and the Apple Diagnostic say:

[07/12/20 03:06:51] Voltage (VeAS) -- PCIe Boost A, 12V (test #1) - Sensor Reads Within Operating Range
- Check to ensure that sensor reads within operating range.
[03:06:51] ERROR -- 3 [Sensor reading below low point] -- (0.00) TEST FAILED
[07/12/20 03:06:51] * TESTING FAILED [Elapsed time = 00:00:05] *

and

[07/12/20 03:18:12] Video Controller: Mezzanine (test #94) - PCIe 16 Link Width Verification Test
- Verifies that the pcie link width is 16
[03:18:12] ERROR -- 90 [Linkwidth test failed.] -- TEST FAILED


What can i do? Is there a way to solve this problems?


I really thank you if you can help me to solve this!
G.

 

[goodbye]

I know that the first error is the left 6pins aux in the backplane board.
Someone can help me with the second Test Failed?

Thanks!

 

[Macschrauber]

Check the chips near the pcie power connector.

People ripped off the chips with the pcie power cable plugging off uncarefully.

 

[h9826790]

Sounds like GPU driver issue more than power issue. Remove the GT120 and see if problem still exist.

It's very meaningless to run AHT with 3rd party hardware. It often generate erroneous failure message that lead you to the wrong direction.

 

[ctrlzone]

i had both this issues as well
for the shutdowns you could try to lower NB temps with excessive fan speeds.
the logout's indeed sound like GPU issues, using only 1 card solved that.

 

[goodbye]

Good mornig,
thanks to everyone for your replay.
@Macschrauber the chips are ok, actually it's the first time that i unplug the 6pins cable
@h9826790 i forgot to say that i run the AHT only with the GT120, i haven't the original gpu. do you think that it would help to find out the problem, running AHT with the 980ti?
@ctrlzone thanks! i will try to take off the gt120. it's weird because everything was working fine for one year.

So, you don't suggest to buy a new backpanel board or a new power supply?

thanks!