Why did my video card fry?

1

1.1.2.3.5...

Gawd
Joined
Jul 13, 2006
Messages
530
The setup...

mobo: evga 750i ftw
cpu: x3380 @3.66ghz
mem: 2gb x 4 gskill ddr2 1100 @ ddr2 1066
video cards: 2 evga 285 gtx sc at stock sc speeds in a parallel water loop.
Two 7200 rpm WD hdds. One for each os.
powersupply: BFG 1kw psu
http://hardocp.com/article/2007/02/27/bfg_tech_bfgr1000wpsu_power_supply/1
installed OSs: Windows 7 64 professional latest drivers fully updated
Ubuntu: latest distro fully updated with latest beta nvidia drivers

I was running a NAMD all atom simulation in Ubuntu. This loads my cpu completely and seems to heat the video cards up pretty good, but nothing over 50C. NAMD does not use very much memory so there was not much of a memory load.

I heard a *pop* and I smelled the death of electronics as my pc instantly shutoff.

I pulled the video cards to check it out, and it looks like some power regulator fried on the top (slot 1) 285.

The rest of the system and the other video card are working fine (thankfully.)

Attached are some pictures.
dscn1270.jpg

dscn1271e.jpg

dscn1272e.jpg

dscn1273g.jpg

dscn1274w.jpg

dscn1275.jpg


http://img222.imageshack.us/img222/9418/dscn1270.jpg
http://img831.imageshack.us/img831/3985/dscn1271e.jpg
http://img809.imageshack.us/img809/4165/dscn1272e.jpg
http://img823.imageshack.us/img823/112/dscn1273g.jpg
http://img18.imageshack.us/img18/7103/dscn1274w.jpg
http://img87.imageshack.us/img87/9318/dscn1275.jpg
 
Last edited:
Shit happens, more than likely from an overclock. although the cards were kept cool under water >50 celcius the current running through the card could have been too much for that poor card and it kicked the bucket.

If you were running at stock speeds etc.. I have no idea other than the first two words of this post. ;)
 
I kept the video cards @ stock superclock speeds....

I think it might be my psu's fault....the point being, if it is my psu's fault it would be nice to know so I can replace it before getting the 285 replaced and reinstalled. I am running on an old 8800gt right now, but eventually the two 285s gotta go back in....

If its just back luck, I am cool with that. But I would like to be sure, or at least do my best to eliminate any risk factors I may have missed.

Is there any chance the beta driver I was running in Ubuntu could have had a hand in this?
 
gpu cool, and power circuitry cool are two different things!
 
I'd replace your psu. Those bfgs were made 4 years ago and were nothing to talk about back then.
 
Given that the rest of the system still works with the same PSU and nothing else has failed, I'm somewhat reluctant to believe it's the PSU at fault, rather just a fairly typical mid-term graphics card failure, but I do agree with the sentiment of llose's post, those BFG PSUs are pretty naff, and if you don't have problems now, you may do not long from now.
 
They're not explicitly bad PSUs, but they're middle-tier quality PSUs at best, and a middle-tier PSU at that age is at least something to keep an eye on.

The VRM cooling issue is certainly one worth mentioning with waterblocks and heavy load, although any card without fault would just shut down if the VRMs were overheating, running them close to their max for a couple of years can take its toll. Most water setups for cards that large though encompass VRM cooling in an all-in-one block so in theory that shouldn't have been issue. No way to find out now of course.
 
After having lost at least 3 or 4 video cards and 2 motherboards over the past few years, I can say, if I were in your shoes, I would replace the power supply. No question about it.

In all cases, the voltage regulators are what popped on the video cards and the motherboards I lost. After you move a power supply that "seems" fine for literally years to another system and it fails a year later and takes out video cards, then you know it is bad.

Key point I am making is, unless you want to risk frying out more equipment, replace the power supply with something quality (such as a Corsair TX950, or AX1200) and have the peace of mind that the problem is taken care of.
 
To be honest, an AX 850W would do for that system. 1KW is a little overkill for two 220W cards.
 
1 said:
I kept the video cards @ stock superclock speeds....

I think it might be my psu's fault....the point being, if it is my psu's fault it would be nice to know so I can replace it before getting the 285 replaced and reinstalled. I am running on an old 8800gt right now, but eventually the two 285s gotta go back in....

If its just back luck, I am cool with that. But I would like to be sure, or at least do my best to eliminate any risk factors I may have missed.

Is there any chance the beta driver I was running in Ubuntu could have had a hand in this?
Click to expand...

more like shit happens. some faulty part finely gave way. check your PSU but I would file this under "shit happens category"
 
MrXROK said:
Were the VRM cooled?

Modern video cards need this cooled.
Click to expand...

This most definitely. If your case is pure watercooled and has no airflow, then the components on your video card are probably overheated.

Under normal circumstances, the GPU PCB and all the components attached act as their own heat exchanger - they're not as efficient as a metal heatsink, but they do help with cooling those components. This is why on certain cards the memory chips and VRMs often have no attached cooling - the PCB conducts the heat away and dissipates it in the moving air.

So, the question is: does your watercool setup have air moving inside the case?
 
Well the EK blocks are FULL cover blocks. In fact, if you look at the pictures you can see what part fried and that it was sitting under a section of the water block. You can even see where the burn marks on the thermal tape....so they were cooled.

In fact the whole water block is fairly cool to touch on any part, under full load for an hour of gaming.

To answer the question, yes there is air flow through the case. The memory, cpu and all other parts are air cooled. I have a large (180 or 220 mm) fan in the front of the case, and a 120 for exhaust. The top of the case is pretty open for air to flow out...

Overall the case is well vented and I have never had any heat issues (that I know of.) The case is a Lancool (Lian Li off brand).

http://www.google.com/products/cata...og_result&ct=result&resnum=5&ved=0CDYQ8wIwBA#
 
Faulty VRMs let go then, I think. Seeing the pictures I'm unconvinced anything caused the problem but the card itself.
 
dbphelps said:
After having lost at least 3 or 4 video cards and 2 motherboards over the past few years, I can say, if I were in your shoes, I would replace the power supply. No question about it.

In all cases, the voltage regulators are what popped on the video cards and the motherboards I lost. After you move a power supply that "seems" fine for literally years to another system and it fails a year later and takes out video cards, then you know it is bad.

Key point I am making is, unless you want to risk frying out more equipment, replace the power supply with something quality (such as a Corsair TX950, or AX1200) and have the peace of mind that the problem is taken care of.
Click to expand...

Perhaps you are doing something wrong to lose that many video cards.

I've lost 1 video card out of 20+ years of computing. That was my fault which my hand slipped with a solder iron when I was voltage modding my card
 
Long term faults, none to report. Short-term faults, DOAs and the likes, I've had four. All non-reference low quality crap. Sticking with reference cards I've never had any issues.
I'll also point out that 8800s and 9800s basically have a shelf life of 2-3 years. Very few of them last any longer than that before something fails on them. I don't see the same trends emerging for the GTX2 series, but it's possible that it could happen, I don't know if nvidia improved the pretty shoddy build standard of G80/92 when they made GT200.
 
nvidia VRM chips (which you burnt up) are notorius for self destruction with almost ANY voltage mods. Fact is you probably got a card with weak VRMs and the stock OC did them in.

Have you checked with evga lifetime warranty coverage?
 
MrXROK said:
Perhaps you are doing something wrong to lose that many video cards.

I've lost 1 video card out of 20+ years of computing. That was my fault which my hand slipped with a solder iron when I was voltage modding my card
Click to expand...

Nope, it was a couple of bad power supplies (one was a BFG ES-800 that had too high of ripple)... I run almost always SLI/Crossfire setups, so it is usually more than one card in the machine at a time when things go pop...

Haven't had a single problem since I started buying well researched and well constructed power supplies... And I have 7 major rigs running, all with multiple video cards and none with a power supply less than 900w... My main rig here is dual GTX480s and has a Corsair AX1200 power supply on it...
 
dbphelps said:
Nope, it was a couple of bad power supplies (one was a BFG ES-800 that had too high of ripple)... I run almost always SLI/Crossfire setups, so it is usually more than one card in the machine at a time when things go pop...

Haven't had a single problem since I started buying well researched and well constructed power supplies... And I have 7 major rigs running, all with multiple video cards and none with a power supply less than 900w... My main rig here is dual GTX480s and has a Corsair AX1200 power supply on it...
Click to expand...

That is interesting to me....

I have NEVER had a video card failure until this one popped.....keep in mind my first video card was a voodoo add in graphics card and I have been running cards ever since. I would say that is about 15 years or so of never having a problem until now...

I am trying evga's lifetime warranty, but the card was not registered within 30 days so all I can do is hope they will let this one through. I have bought 4 evga cards and never needed warranty service until now.

In any case I think I am going to buy a new power supply before putting these back in. It could very well be that this video card had weak VRMs and just popped while loaded up, or it could be a shoddy power supply. I just don't feel comfortable taking chances....

Anyone have any recommendations for my setup? I am thinking I could get away with 800 to 850 watts....but I am open to suggestions.
 
samuelmorris said:
Given that the rest of the system still works with the same PSU and nothing else has failed, I'm somewhat reluctant to believe it's the PSU at fault, rather just a fairly typical mid-term graphics card failure, but I do agree with the sentiment of llose's post, those BFG PSUs are pretty naff, and if you don't have problems now, you may do not long from now.
Click to expand...

Multi-rail PSU? PCIe Connectors on one rail and system on the other?

Don't be so certain it isn't the PSU or you could hook up new cards and pop them as well...

(May have quoted the wrong person...If so, disregard...but the PSU does sound suspect tbh)
 
1 said:
That is interesting to me....

I have NEVER had a video card failure until this one popped.....keep in mind my first video card was a voodoo add in graphics card and I have been running cards ever since. I would say that is about 15 years or so of never having a problem until now...

I am trying evga's lifetime warranty, but the card was not registered within 30 days so all I can do is hope they will let this one through. I have bought 4 evga cards and never needed warranty service until now.

In any case I think I am going to buy a new power supply before putting these back in. It could very well be that this video card had weak VRMs and just popped while loaded up, or it could be a shoddy power supply. I just don't feel comfortable taking chances....

Anyone have any recommendations for my setup? I am thinking I could get away with 800 to 850 watts....but I am open to suggestions.
Click to expand...

If they give you a hard time about not registering the card for it's warranty, take the matter to consumer mediation, at your local DA's office; it took about a month of back and forth with EVGA and the county, but it worked for my friend who never registered his 8800 that frazzled out.

Corsair HX, or AX series my TX has been running good as gold since purchase, that was over two years ago, AX, and HX not only being modular, have better power regulation.
 
For what it's worth. my BFG ES-800 just bit the dust after over two years. Before it went, I noticed my GPU became unstable. It's not completely dead, but I can't game at high settings now. Thankfully, the PSU didn't take anything else out.
 
Corsair AX 750W will be ample for the test system as it was before one of the GTX285s went.
 
Evga approved the rma and I dropped the video card off at the UPS station today. It should get to Evga by Friday.

Right now my concern is what I will get back for a card. I need another gtx 285 with revision 1 pcb for the EK block to fit properly. Depending on what I get back from Evga I might need to spend some time in the F/S area looking for trades.

As far as the power supply goes, I went with a new OCZ
http://www.hardocp.com/article/2008/04/07/ocz_elitexstream_1000w/

I realize 1kw is perhaps a bit of overkill, but need 100% stability for up to several weeks of continuous load on the whole system. I will also likely be adding in a raid 5 array and so I feel a solid 1kw psu is best for my purposes.
 
OCZ PSUs are relatively mediocre but thankfully that is one of the better ones out there. I hope it wasn't too expensive :p
 
samuelmorris said:
OCZ PSUs are relatively mediocre but thankfully that is one of the better ones out there. I hope it wasn't too expensive :p
Click to expand...

It bought it from a friend who used it maybe a month so its basically new. I paid 90.00 for it so I feel it was a good deal for a solid psu.
 
Assuming that's $90 then yeah that's a good deal, although did you ask the friend the reason for him selling it? :)
 
dbphelps said:
Nope, it was a couple of bad power supplies (one was a BFG ES-800 that had too high of ripple)... I run almost always SLI/Crossfire setups, so it is usually more than one card in the machine at a time when things go pop...

Haven't had a single problem since I started buying well researched and well constructed power supplies... And I have 7 major rigs running, all with multiple video cards and none with a power supply less than 900w... My main rig here is dual GTX480s and has a Corsair AX1200 power supply on it...
Click to expand...

I run mid range power supplies with all my systems. Never had a single hickup.

Then again I avoid running SLI or crossfire as it's more problematic.
 
1 said:
Yeah...its $90 USD...lol

How do you know my friend is a he? :p I didn't ask....but I trust my friend.
Click to expand...

sorry, it's sexist I know, but I tend to assume he with this particular area, because it is pretty male-dominated.
 
You must log in or register to reply here.
Back
Top