EVGA GTX 1070/1080 GPUs Catching Fire Due To VRMs Overheating

[thesmokingman]

People sometimes record video to post when asking for help. Guessing that's why he was recording. Seems he dropped the camera and frantically plugged out wires after the spark.

Not buying it. There was no power to the gpu, no pcie cables plugged in, ie. no voltage thru the plugs, no power, fans don't even spin till it gets to X temp which is not gonna happen as the card is inactive. Again, perfect timing and my BS meter is going off.

The VRMs not likely to fail by themselves, but likely to heat up surrounding components like mosfets which can fail in poorly ventilated cases. Good in depth article on the issue, which basically states that its an EVGA design flaw of their ACX cooler.

http://www.gamersnexus.net/news-pc/2661-evga-mosfet-failure-possible-from-thermal-runaway-scenario

Dude, did you bother reading anything? It's not the cooler that is faulty.

 

[Dave2150]

Did some reading. Only affects cards with non-reference PCB. Meaning only FTW models. And ones shipped before Aug. 31. Non issue otherwise. SC cards like mine are not affected.

Wrong. Affects ALL EVGA cards with the ACX cooler, so all 1070's and 1080's apart from the founders edition. ONLY the founders edition has the thermal pads fitted, the rest have nothing.

 

[Azix]

Not buying it. There was no power to the gpu, no pcie cables plugged in, ie. no voltage thru the plugs, no power, fans don't even spin till it gets to X temp which is not gonna happen as the card is inactive. Again, perfect timing and my BS meter is going off.

yeah didn't notice the pci-e cables weren't connected. But there was power to the GPU, unless you think he used some trick to move the fans and light the leds.

BS meter needs tuning. Go listen to trump and come back.

Trolling is not allowed and neither are Political comments
Markfw900
Anandtech Moderator

 

[amenx]

Dude, did you bother reading anything? It's not the cooler that is faulty.

I get that the lack of thermal pads creates the problem, but in conjunction with the ACX cooler seems to exacerbate it. I could be wrong, but then the article needs to be reworded for better clarity.

The ACX coolers have fallen prone to a number of design flaws over the past few years, but this one is one of the easiest to prevent. EVGA's lack of a thermal interface between the PCB baseplate and the ACX heatsink means that the cooler could actually be doing the opposite of cooling. Heat build-up on the silicon looks for an interface to transfer to the copper or alloy sink. That heatsink doesn't do anything unless there's a way for the heated device to transfer its heat, and could end up trapping the heat in an area of effectively zero airflow (see some of our tear-downs to better understand this). There's no airflow in that part of the card -- the air is designed to hit the heatsink and dissipate heat through the greater surface area provided by fins, not somehow ethereally phase through the heatsink and cool those hot zones.

Are all other GPU coolers the same in that regard?

 

[ConsoleLover]

Of COURSE this is a MAJOR issue! Someone paying that much money and having his card burn out is just unacceptable! That is weeks without a card, that is going through all the trouble to return the card, sometimes it can be a pain, even supposedly "good" companies and again it leaves you without a computer for a long time, until you get a replacement back.

 

[thesmokingman]

I get that the lack of thermal pads creates the problem, but in conjunction with the ACX cooler seems to exacerbate it. I could be wrong, but then the article needs to be reworded for better clarity.

Are all other GPU coolers the same in that regard?

The ACX coolers have fallen prone to a number of design flaws over the past few years, but this one is one of the easiest to prevent. EVGA's lack of a thermal interface between the PCB baseplate and the ACX heatsink means that the cooler could actually be doing the opposite of cooling. Heat build-up on the silicon looks for an interface to transfer to the copper or alloy sink. That heatsink doesn't do anything unless there's a way for the heated device to transfer its heat, and could end up trapping the heat in an area of effectively zero airflow (see some of our tear-downs to better understand this). There's no airflow in that part of the card -- the air is designed to hit the heatsink and dissipate heat through the greater surface area provided by fins, not somehow ethereally phase through the heatsink and cool those hot zones.

I'm not sure what this quote is going to but it seems to mucking the wording up along the way. But yea, all other gpus are the same in that respect. Forget to put the thermal pad between the vrm/mosfets to heatplate(baseplate/coolingplate/memory-vrm plate/etc plate) and the heat has no where to go. That is bad news for any cooler.

 

[Timmah!]

Seems this needs to be posted for more clarification:

in regard to black screens and 100 percent fan speed:



Recently, there are many customers wanting to know more details about the EVGA GeForce 1080 FTW “Black Screen” issue and we believe we need to make a clear statement to clear all kinds of speculation.

The EVGA GeForce GTX 1080 FTW “Black Screen” issue was fixed as of the first week of September and this has been confirmed by all users who have received the replacement units with the issue resolved. Also, all new production has been confirmed to not have this issue after the first week of September.

This previous black screen issue was caused by On Semiconductor (VRM IC Manufacturer) VRM IC’s being out of spec, triggering the OCP (Over Current Protection) and will not cause any damage to your system. The percentage of the IC’s that were out of spec has been confirmed by On Semiconductor and is approximately 3% to 4%. Since then, EVGA and On Semiconductor have worked out the solution and RMA replacement action a couple months ago as we stated above. At this moment, all of the EVGA GeForce GTX 1080 FTW’s have been corrected and all products in the field are working properly.




in regard to burned out cards due to lack of VRM cooling:


The test used in the referenced review from Toms Hardware (Germany) is running under Furmark, an extreme usage case, as most overclockers know. We believe this is a good approach to have some idea about the graphics card limit, and the thermal performance under the worst case scenario. EVGA has performed a similar qualification test during the design process, at a higher ambient temperature (30C in chamber) with a thermal coupler probe directly contacting the key components and after the Toms Hardware (Germany) review, we have retested this again. The results in both tests show the temperature of PWM and memory is within the spec tolerance under the same stress test, and is working as originally designed with no issues.

With this being said, EVGA understands that lower temperatures are preferred by reviewers and customers.

During our recent testing, we have applied additional thermal pads between the backplate and the PCB and between the baseplate and the heatsink fins, with the results shown below. We will offer these optional thermal pads free of charge to EVGA owners who want to have a lower temperature. These thermal pads will be ready soon; and customers can request them starting Monday, October 24th, 2016. Also, we will work with Toms Hardware to do a retest.


If you work for EVGA, you need to put that in your signature
Markfw900
Anandtech Moderator

 

[nerp]

Wrong. Affects ALL EVGA cards with the ACX cooler, so all 1070's and 1080's apart from the founders edition. ONLY the founders edition has the thermal pads fitted, the rest have nothing.

Sorry, but I believe you're mistaken. Only FTW cards produced before Aug. 31 are affected. The reference PCB doesn't even need the thermal pads because it's not affected. EVGA is still ending them out to non FTW owners to be nice, but it's pointless for non-FTW cards.

 

[Markfw]

Of COURSE this is a MAJOR issue! Someone paying that much money and having his card burn out is just unacceptable! That is weeks without a card, that is going through all the trouble to return the card, sometimes it can be a pain, even supposedly "good" companies and again it leaves you without a computer for a long time, until you get a replacement back.

You need to chill. I have had TWO of the FTW cards running at 100% load 24/7 for MONTHS, and have had no problems. Of course they found a way to make it happen, but it must be a small issue, or I would have had a problem. The only difference is that I over-ride the default fan speed, and make sure they run at 73c or less, NOT 83c or less. And have you heard of an advanced RMA ? where they cross-ship ?

Please don't blow this out of proportion.

 

[moonbogg]

All I know is, I want to see another GTAV sticky bomb mod where the bomb looks like an EVGA 1080 FTW. Toss the high end EVGA GPU at a nearby vehicle and watch the explosion take place. Its a weapon.


Trolling is not allowed
Markfw900
Anandtech Moderator

 

[Annisman*]

Welp my gtx 1080 with hybrid cooler virtually no air flow to it besides the case fans because I accidentally snipped the blower power wires when installing it a few months ago, and it hasn't lit up in flames or shown any sign of problems.

 

[escrow4]

I wouldn't purchase a GPU that isn't Asus or Gigabyte as a distant second (quality is occasionally so-so). MSI has crappy memory chips on the 1070, EVGA will keep your Note 7 company, Gainward and Zotac and the rest are second tier.

 

[Shmee]

My MSI has Samsung chips which OC well, so I am not sure what you mean by the crappy memory comment. Also, Zotac has gotten a lot better, and has some really good cooling designs recently.

 

[TheRyuu]

MSI has crappy memory chips on the 1070

Pretty sure every company making 1070's has moved to Micron because of supply issues. It's only become an issue because of incorrect timings/voltage being set in the BIOS.

As for the VRM problem, I believe I have an affected model (1080) although I have not had any problem up to this point although I haven't really done anything remotely stressful aside from some quick testing with Doom. I do have a case fan on the side (albeit set to the slowest speed) which hits the side of the 1080 so perhaps this gets some airflow surrounding the unit while the ACX fans are off

I just ordered some thermal pads from amazon (free 2 day shipping) since they're cheap. I guess I'm going to have to reseat the GPU heatsink as well with fresh thermal paste if I'm going to be removing it to apply the pads. Guess I can put some better thermal paste[1] on it while I'm at it since I already have some.

Although I think I know where I'm putting the pads did anybody make any pictures of where they should be? Just want to double check. I didn't see anything on the EVGA site like instructions although I assume they'd be shipping them with the thermal pads.

[1] http://www.newegg.com/Product/Product.aspx?Item=9SIA1K639W8999

 

[96Firebird]

I would like to see this lower the price of the 1070 FTW so I can get a good deal on it, but that probably won't happen...

Just like everyone said the price of the 970 would tank after the memory config was revealed, but that didn't happen either!

 

[LTC8K6]

I wouldn't purchase a GPU that isn't Asus or Gigabyte as a distant second (quality is occasionally so-so). MSI has crappy memory chips on the 1070, EVGA will keep your Note 7 company, Gainward and Zotac and the rest are second tier.

The Micron memory chips are actually excellent, as usual.

Where did you get the idea that the Micron memory chips on the cards were faulty?

 

[Timmah!]

If you work for EVGA, you need to put that in your signature
Markfw900
Anandtech Moderator

I dont work for EVGA, i am just personally invested in this, as i own 2 FTWs. And it kinda looked like this topic went in circles, with people discussing things without having complete picture on what is going on, as its arguably bit more complicated, thats why i took liberty to repost those official EVGA stances on both "issues" just to bring some clarity into discussion.

 

[Markfw]

I dont work for EVGA, i am just personally invested in this, as i own 2 FTWs. And it kinda looked like this topic went in circles, with people discussing things without having complete picture on what is going on, as its arguably bit more complicated, thats why i took liberty to repost those official EVGA stances on both "issues" just to bring some clarity into discussion.

Then you need to clearly state Something like that, and source and quote the material

 

[guskline]

I have owned a number of EVGA cards and have always been pleased with the quality and performance. However, like every pc component, sometimes someone has a faulty one.

 

[Timmah!]

Then you need to clearly state Something like that, and source and quote the material

OK, will do next time.

 

[swilli89]

You need to chill. I have had TWO of the FTW cards running at 100% load 24/7 for MONTHS, and have had no problems. Of course they found a way to make it happen, but it must be a small issue, or I would have had a problem. The only difference is that I over-ride the default fan speed, and make sure they run at 73c or less, NOT 83c or less. And have you heard of an advanced RMA ? where they cross-ship ?

Please don't blow this out of proportion.

Actually you should chill. Its not very wise to minimize a potentially fire-hazard situation to fellow posters. This could be a serious issue for some owners out there with the defective GTX 1070's.

 

[Maverick177]

Another 1080 caught on fire: https://www.youtube.com/watch?v=O2-VaOPOvgI

 

[thesmokingman]

Another 1080 caught on fire: https://www.youtube.com/watch?v=O2-VaOPOvgI

That's the same video.

 

[Dave2150]

Sorry, but I believe you're mistaken. Only FTW cards produced before Aug. 31 are affected. The reference PCB doesn't even need the thermal pads because it's not affected. EVGA is still ending them out to non FTW owners to be nice, but it's pointless for non-FTW cards.

Wrong.

"EVGA’s Entire Range Of Nvidia GTX 1080 & GTX 1070 Featuring The ACX Cooler Is Affected
There were several investigations by a number of reputable technology outlets to try and identify the culprit behind it all. It was discovered that the issue originates from the voltage regulation modules which overheat leading to component failure and MOSFET blow out. The issue affects EVGA’s entire range of Nvidia GeForce GTX 1080 and GTX 1070 cards equipped with the ACX cooler, all of which lack direct VRM cooling. This includes GTX 1080 & 1070 ACX, SC, FTW & FTW DT models. Essentially all dual fan models except the Classified."

Source: http://wccftech.com/nvidia-gtx-1080-1070-evga-cards-dying/

EVGA Source: http://forums.evga.com/badHOT-PCB-Layout-on-GTX-1070-FTW-m2565921-p5.aspx#2567638

 

[thesmokingman]

Wrong.

"EVGA’s Entire Range Of Nvidia GTX 1080 & GTX 1070 Featuring The ACX Cooler Is Affected
There were several investigations by a number of reputable technology outlets to try and identify the culprit behind it all. It was discovered that the issue originates from the voltage regulation modules which overheat leading to component failure and MOSFET blow out. The issue affects EVGA’s entire range of Nvidia GeForce GTX 1080 and GTX 1070 cards equipped with the ACX cooler, all of which lack direct VRM cooling. This includes GTX 1080 & 1070 ACX, SC, FTW & FTW DT models. Essentially all dual fan models except the Classified."

Source: http://wccftech.com/nvidia-gtx-1080-1070-evga-cards-dying/

EVGA Source: http://forums.evga.com/badHOT-PCB-Layout-on-GTX-1070-FTW-m2565921-p5.aspx#2567638

More craziness than brains in that quote man. The way the evga acx cooler is designed is essentially the same as every open fan cooler before it. The cooler only direct cools the core and the cooling plate covers the other vrm/memory ic and the coolers fan pushes air down over the plate.

https://www.techpowerup.com/reviews/MSI/R9_290X_Lightning/4.html

290x which pulls down like 3 times the power draw, zomg no direct vrm cooling on the actual cooler?