EVGA 2080 Ti XC burst into flame !!

Let's see EVGA try to fix this by sending out some thermal pads like they did with the 1080 SC series.
 
Well, with the Christmas season rapidly approaching, at least nVidia was gracious enough to throw in a free yule log for that $1200+ price tag. Be sure to make a wish.


On a serious note:
Damn that sucks! Glad no one got hurt. I hope this works out for you and you are taken VERY WELL care of, especially if there is collateral damage to any other components in your rig.
 
sirmonkey1985 said:
i'd actually contact them privately first get that all sorted out and then post it on their forums.. just throwing it on there forums immediately could potentially cause them to just turn around and deny the whole thing and refuse to warranty it.
Click to expand...
What a ridiculous way for a company to behave, fuck them! Lying shysters! I've fucking had it with the shit of this world, what a turd we live on.
 
triwolf said:
What a ridiculous way for a company to behave, fuck them! Lying shysters! I've fucking had it with the shit of this world, what a turd we live on.
Click to expand...

EVGA is very good with RMA and actually takes care of their customers. Not sure why people are speculating they'd tell the op to stuff it or deny warranty. It's a legitimate warranty claim.
 
Slade said:
EVGA is very good with RMA and actually takes care of their customers. Not sure why people are speculating they'd tell the op to stuff it or deny warranty. It's a legitimate warranty claim.
Click to expand...
Seems lots of people love to invert everything, have a card fail, post on their forums, get denied an RMA. Seems like some evil devil logic to me, if EVGA did try to do that, my statement holds.
 
Slade said:
EVGA is very good with RMA and actually takes care of their customers. Not sure why people are speculating they'd tell the op to stuff it or deny warranty. It's a legitimate warranty claim.
Click to expand...

Every time I've seen a card catch fire, I've also read about the manufacturer fighting the warranty claim. If EVGA gets their card back and says it was the OP's PSU that caused it and they aren't replacing it, then what? Because the cards are so new and there are so many problems with the 2080Ti, I'd bet that EVGA takes it back just for the good PR regardless of the cause of the fire.

But if your 2 year old 1080 catches fire, I'd expect a fight to get a manufacturer to get you a new one under warranty.
 
I still have 80 days on a 1080 GTX warranty and I'd expect them to honor it to the end.
 
kirbyrj said:
Every time I've seen a card catch fire, I've also read about the manufacturer fighting the warranty claim. If EVGA gets their card back and says it was the OP's PSU that caused it and they aren't replacing it, then what? Because the cards are so new and there are so many problems with the 2080Ti, I'd bet that EVGA takes it back just for the good PR regardless of the cause of the fire.

But if your 2 year old 1080 catches fire, I'd expect a fight to get a manufacturer to get you a new one under warranty.
Click to expand...


According to their forum they've already sent him a new card after getting in touch with him.
 
Just so you know guys, Nvidia is outsourcing the manufacture of their chips to Cyrix :giggle:
 
So as it stands, we have a batch of 2080 ti GPU which should have failed testing get shipped out as FE cards and AIB cards. Wonder if there is enough RMA cards to pin down a specific week of GPU manufacture where these GPU came from.

Also wondering if there is a way to run a benchmark or stress test which utilizes or targets the weak part of the GPU in order to induce failure or show it will fail quicker.
 
Slade said:
So as it stands, we have a batch of 2080 ti GPU which should have failed testing get shipped out as FE cards and AIB cards. Wonder if there is enough RMA cards to pin down a specific week of GPU manufacture where these GPU came from.

Also wondering if there is a way to run a benchmark or stress test which utilizes or targets the weak part of the GPU in order to induce failure or show it will fail quicker.
Click to expand...

If there is, I doubt Nvidia will disclose what the weak link is.
 
I guess the super unlucky ones who pulll the bad gpu chips will have runaway thermals and memory corruption after some time.
 
Quite certainly cracked MLCC failure, from a mechanical overstress. There are two screw fixing holes nearby and pcb panelization "mouse bites" ( https://www.electronicdesign.com/boards/pcb-designers-need-know-these-panelization-guidelines ). So screws tightened improperly bending the PCB, or PCB cut from the panel using not suitable tools/forces/methods or just some other kind of mechanical stress to the PCB or directly to the MLCC.

EEVblog #1037 - Solving Ceramic Capacitor Cracking - https://www.youtube.com/watch?v=QgKY5QWehME


Photos of original board from https://xdevs.com/guide/evga_2080tixc/ :


1.png 2.png
 
I have got the replacement card from EVGA the day after they contacted. So far seems to be working, but still an nVidia PCB though.

Doesn't seem to be EVGA's fault here, since Asus one with reference PCB also blown up at the exact same spot. Just happened to be EVGA who got the batch.

Really happy with their rapid response on this issue. Hopefully this issue will be resolved soon and finding the root cause of such incident.

thobie said:
Quite certainly cracked MLCC failure, from a mechanical overstress. There are two screw fixing holes nearby and pcb panelization "mouse bites" ( https://www.electronicdesign.com/boards/pcb-designers-need-know-these-panelization-guidelines ). So screws tightened improperly bending the PCB, or PCB cut from the panel using not suitable tools/forces/methods or just some other kind of mechanical stress to the PCB or directly to the MLCC.

EEVblog #1037 - Solving Ceramic Capacitor Cracking - https://www.youtube.com/watch?v=QgKY5QWehME


Photos of original board from https://xdevs.com/guide/evga_2080tixc/ :


View attachment 120839 View attachment 120840
Click to expand...

After it burst into flame, I tried to take a look on what caused the issue, there is definitely no stress on PCB. Very likely to be the faulty resistor that nVidia stated.
 
shansoft said:
I have got the replacement card from EVGA the day after they contacted. So far seems to be working, but still an nVidia PCB though.

Doesn't seem to be EVGA's fault here, since Asus one with reference PCB also blown up at the exact same spot. Just happened to be EVGA who got the batch.

Really happy with their rapid response on this issue. Hopefully this issue will be resolved soon and finding the root cause of such incident.



After it burst into flame, I tried to take a look on what caused the issue, there is definitely no stress on PCB. Very likely to be the faulty resistor that nVidia stated.
Click to expand...

You can’t see mechanical stress happened during manufacturing or past mishandling of the card/pcb or even cracked MLCC with your eyes. One bend is enough. There are soft termination type MLCCs avilable also which can handle more bending/stress.

Those resistors are for current sensing and in series of the supply line - and fail open. Decoupling caps are in parallel of the supply line and fail short -> Crash! Boom! Bang!

PCB could short also but it would need some major fuck up somewhere in the chain, either design, manufacturing or leaving flux/other corrosion to happen and allow things to evolve.

http://www.ieca-inc.com/images/Ceramic_capacitor_Failure_Mechanisms.pdf

https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20110015296.pdf
 
My week and a half old RTX 2070 (Asus Turbo) went down almost the same way, as the OP's card did. Mine didn't burst in flames, but these was smoke coming out of it. Full version:
I bought my card on day one, waited almost two weeks to get it (because the supplies were low). When I got it I played some games and did a bit of 3D modeling/rendering. Never overclocked it, everything was stock. While under load for a few hours (gaming) the card temperatures were maxing at about 77°C and went down to about 34°C while in idle (measured by TechPowerUp GPU-Z). There were absolutely no issues during that period (11 days). Then one morning I turn on my computer, try to do some internet browsing, and suddenly these are some weird noises (I am not a native English speaker, so it's hard to describe it... :S) coming out of my PC. The PC shuts down immediately and I see a bit of smoke coming out of it. I plugged everything out of the electrical socket and opened the case. The smoke was only coming out of the GPU (out of the fan). I plugged the GPU out and turned on the PC. Everything else works just fine. My card wasn't burned, or melted like OP's did. At least not on the outside. I didn't open it, because I didn't want to void the warranty. I filled the RMA form and sent it back to the retailer. This happened to me 9 days ago.
 
That's the Vpp power supply, and I don't see nearly enough caps to work with those inductors.

Two 0603 capacitors will not hold up a switcher, it just ain't happening.

Compare it to the same section on the older expensive card, which uses the same PS section/chips/inductors.

The most interesting thing to me is there is no values for the Ipp on the datasheet; it's obviously not as much as the other two power supplies, with their multiple chips, inductors, and capacitors, but it sets the bias level for turning shit on, and if it's not right, it gets really hot.

Burning how.

BTW, if the current limit resistor opens, the supply turns off, and has since 1985 or so. It Shorting is a problem, but that's not the way an 0.005 ohm resistor fails, lol.
 
Is this Nvidia's idea of a firesale now? Selling video cards that are literally on fire? For the price they should throw in some fire insurance to cover the cost of your house burning down.
 
Blade-Runner said:
Is this Nvidia's idea of a firesale now? Selling video cards that are literally on fire? For the price they should throw in some fire insurance to cover the cost of your house burning down.
Click to expand...

Most likely. Nothing like this has ever happened before with a GPU or any other piece of electronics for that matter.
 
When my asus dual 2080 ti died last Friday I could smell burning from the card pretty strongly. No burn marks that I could see. But it also has a thick backplate I didn't want to take off. I've been using my old video card and no problems since.
 
TahoeDust said:
I don't think I would buy a reference RTX card right now.
Click to expand...

the problem is finding which AIB cards aren't using the reference boards.. even most of EVGA's cards all use the reference boards just with different heat sinks, i think there's only 1 or 2 models that don't but unless it specifically says it's using a custom one there's no way to really know.
 
sirmonkey1985 said:
the problem is finding which AIB cards aren't using the reference boards.. even most of EVGA's cards all use the reference boards just with different heat sinks, i think there's only 1 or 2 models that don't but unless it specifically says it's using a custom one there's no way to really know.
Click to expand...


Often if you look for a custom loop water block for RTX 2080 Ti, find the loop that does not work with Refence card, then take note of models that do, there should be your non-reference RTX cards.
 
Last edited:
nEo717 said:
Often if you look for a custom loop water block for RTX 2080 Ti, find the loop that does not work with Refence card, then take note of models that do, their should be your non-reference RTX cards.
Click to expand...

smart, never even thought about that.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
As an Amazon Associate, HardForum may earn from qualifying purchases.
You must log in or register to reply here.
Back
Top