My 3080 is having implausible and inconsistent voltage readouts and hitting power limit at idle, all while having perfectly normal temperatures. I don't think it's actually reaching the power limit, rather incorrectly reading it's own input voltages. This card had this issue when I got it, but after Precision X1 firmware flashing and new thermal pads because I could see air above the inductors and MOSFETS it was fine for around 2 months. Now the issue is back and not going away, I'd love some help if possible. Obviously warranty is void and I got this used. I have a full electrical lab and test facility available, thermal camera, microscope, and like three electrical engineers happy to help but none of them have worked on GPUs and I'd like to respect their time if someone has an idea what is causing this issue.

While running correctly, this card is fantastic and autoclocks typically somewhere at or above 1965MHz. I do not overclock, nor have I altered any voltage settings from stock since receiving the card.
 
Under any load, however, after about 30 seconds it clocks down to 225MHz and typically stays there, sometimes clocking back up for a minute and running fine but then clocking down again. Whenever it clocks down, it consistently has implausible voltages in GPU-Z. Normally it reads out 12V-12.1V with normal performance, but when it clocks down it will either read between 4V and 10V on the power connectors or it will read 12V on power connectors and 3000V+ on "PWR_SRC Voltage". Whenever it downclocks, Precision X1 reads out at max power, GPU-Z reads out at power threshold and throttle, and temperatures go down from functioning fine at ~65C under load but not max power to 30C when it says full power threshold. This tracks consistently between Precision X1 and GPU-Z. 
 
Things I know about this board:
-It's used, and the previous owner said this was an issue. I get it, buying this was probably dumb.
-I replaced all thermals and did a firmware update immediately when I got it, the card cooled and ran very well after that
-This card ran Starfield very well for 12 hours at a time on release. It also ran a bunch of other games as well as expected for months until last week
-There are no visible cracks on the board. 
-There is no warp on the board
-This board has likely been quite hot before for extended periods. The GPU die was still well cooled, but on receipt there was a visible airgap above the MOSFETS and inductors for the VRMs. All of them. Pictures below if it lets me link them.
-The tape below the wire connector for the LED plate was melted. No components look damaged. Pics also included.
-The firmware updated successfully literally this morning, the issue persists
-The issue persists with different power connectors and in different PCI-e slots, I tested it in two of the four I have available and the issue persists
 
Things I've tested already:
-There is no short through VRMs that I can detect through the coils. Weren't when I got the card, aren't as of ten minutes ago.
-There are no blown fuses that I can find, at least none of the ones big enough to have the amperage rating on them
 
Things I can't really test today:
-I don't have a power supply to apply small voltage and look for heat today
-I don't have a thermal camera to detect heat
-I don't have a multimeter that can test capacitors today, I can test resistance or continuity though
 
Things I can test tomorrow
-I'll have a power supply that can apply discrete voltage
-I'll have a thermal camera to see heat
-I'll have a microscope
-I'll have solder setup and ESD equipment for replacing components
 
Pictures of physical card and GPU-Z readouts"

 
Thanks in advance for anything anyone can suggest testing today or tomorrow in the lab to save time. I'm looking to rule out or isolate issues today to save time tomorrow.
 
 
 
 

Welcome to the EVGA Forums & Congrats on your first Thread & Post
 
--> First thing I would test/ do is try a different PSU in your PC .... PerfCap = Pwr does seem to align with a 12V bus issue
 
Which Nvidia drivers have you tested?
 
Your PC specs & OS?
 
Multilayer PCB on the GPU, so damage could be anywhere.
 
The circuit near the melted tape looks kind of toasty
 
EVGA does allow a secondhand warranty. So it is not clear in your post why you believe there is no warranty on this card.
check the serial number (if your card is not damaged from things not covered by warranty)
 
EVGA Warranty terms
 
 
RE: Limits on New member posts ---> Learn your way around the EVGA Forums, Rules & limits on new accounts Ultimate Self-Starter Thread For New Members

Why do you think the warranty is void?
If it is just because it is bought used second-hand, that's no problem.  I would try getting a warranty replacement.
 
If you verify correct voltage into the card and good power connections (probe each pad on the card), there is clearly a component which is damaged and causing incorrect readings, but good luck finding it.

bertmern
Not sure how to do markdown here so I'll answer in order

Drivers tested-> most up to date and two prior
PC Specs-> i7-6850k, 32gb 3000MHz RAM, ASUS Rampage V edition 10, Corsair RM850i, running Windows 10 on a fresh install on a fresh M2 Samsung 970 EVO Plus
I still haven't had a chance to test the PSU, will do today. Kid came home sick so that was my night.
 
Warranty Void:
Thanks for the links though, I had been looking at the warranty terms page before and assumed since the card came with the tape over the screw nearest the GPU die perforated that it was void. That and I've replaced all the thermals, which wasn't exactly a factory modification even if it ran cooler than I've seen a lot of peoples' readouts. 

1) Hope your little one gets better soon & the whole family does not end up getting said "bug"
 
2) RE: Tape over screws - just a telltale that RMA product needs a closer look mostly as cooler may have been removed.  EVGA does allow you to change TIM, Pads or cooler type - without automatically voiding warranty - as long as you cause no physical damage to the card

Thanks for the update
 
 
Any signs of arcing on the back of backplate?  Just trying to think of what could have happened to that black tape in your OP as it looks burnt