Dead 4870 1gb in 3 days...Was PSU the issue????

[o3srt4me]

Well with my low tech knowledge, I scrummaged up a logical test in which I think I can prove that why my 4870 1gb burned out after 3 days can be attributed to my Antec 500w PSU....


BTW, the story behind all of this can be found here http://forums.anandtech.com/me...=2274637&enterthread=y

The night before the card died, I ran a prime95 stability test at a moderate oc to 3.66ghz (385 frequency at the 9.5x multiplier) and it failed after about 5 mins. Scratching my head wondering as to why I would recieve a hardware failure note with moderately low temps (43c-45c), and nowhear near my max fsb of 1600.....I turned the CPU back to stock clock...

The next day I went to play cod 5 and the game crashed a final time and the computer refused to reboot....After swapping back in the old 8800gt (A way less power demanding card) I ran prime95 again at the OC that the CPU had failed at the night before....The test passed and is now still running far past the point in which it had failed....

Is it safe to say that the removal of the 4870 offered back sufficient power to the CPU that allowed the OC to stabilize??? and can that be translated into saying that my 500w setup was insufficient for that GPU???

 

[Quiksilver]

Well the 4870 1GB does use more power than the 8800GT, so there is a less of a load on the PSU; which I suppose could lead to better voltage regulation on the PSU's part. Which could have played a small factor to your OC. Also let's not forget the the 4870 probably ran way hotter.

As far as whether or not your PSU killed you video card; unlikely. That was the temps; as you said you ran Furmark and the GPU temp was around 90°C or so, well the VRM's were probably way hotter which inevitably killed the card.

If I wanted a little insurance policy (eg. more cushion) I would probably pick up a 600W good quality PSU though.

 

[o3srt4me]

Hmmm but I suppose what im trying to discover is why the card overheated.....The only other viable option is that the card was defective....The case has great airflow and I could hear the fan spinning at almost maximum efficiency....hmmm...keep in mind it was at stock clocks...

 

[Quiksilver]

Poor contact between cores/VRM's and the heatsink possibly.

 

[Jessica69]

Sapphire........'nuff said.

 

[nerp]

Sounds very much like a PSU issue.

 

[beray]

Originally posted by: o3srt4me
Hmmm but I suppose what im trying to discover is why the card overheated.....The only other viable option is that the card was defective....The case has great airflow and I could hear the fan spinning at almost maximum efficiency....hmmm...keep in mind it was at stock clocks...

Take a look at your vid card and visually inspected it for what was/were burned. Typically for things to burn required power, look in areas of high power density concentration -> caps, GPU, VRM...

If you can't see any burn-marks then try the scenting technique, put your nose to the vid card and smell for the burned spot(s). If the smell is overwhelming all over then take your vid card and waft it through the air a few times before scenting again, wafting would get rid of the smell from all over except for the burned spot(s) which helps localizing your scenting detection.

If your power supply was insufficient for the job and the cause of your problem --> The vid card VRM (voltage regulator module) is what normally get burned. However if your vid card VRM are well built/tough and would not burn then it shifted the burning power problem to your GPU. Your PSU is usually the last likely to die as it is made to handle much greater power than the others above even when being insufficient.

Basic voltage regulation fundamental --> POWER = VOLTAGE x CURRENT

When voltage is not enough, current will shoot up toward infinity to maintain the equation, INFINITE current will burn anything at all.

 

[o3srt4me]

Originally posted by: beray

Originally posted by: o3srt4me
Hmmm but I suppose what im trying to discover is why the card overheated.....The only other viable option is that the card was defective....The case has great airflow and I could hear the fan spinning at almost maximum efficiency....hmmm...keep in mind it was at stock clocks...

Take a look at your vid card and visually inspected it for what was/were burned. Typically for things to burn required power, look in areas of high power density concentration -> caps, GPU, VRM...

If you can't see any burn-marks then try the scenting technique, put your nose to the vid card and smell for the burned spot(s). If the smell is overwhelming all over then take your vid card and waft it through the air a few times before scenting again, wafting would get rid of the smell from all over except for the burned spot(s) which helps localizing your scenting detection.

If your power supply was insufficient for the job and the cause of your problem --> The vid card VRM (voltage regulator module) is what normally get burned. However if your vid card VRM are well built/tough and would not burn then it shifted the burning power problem to your GPU. Your PSU is usually the last likely to die as it is made to handle much greater power than the others above even when being insufficient.

Basic voltage regulation fundamental --> POWER = VOLTAGE x CURRENT

When voltage is not enough, current will shoot up toward infinity to maintain the equation, INFINITE current will burn anything at all.

OK well I sniffed around the card a good deal and the only area that smells of burned rubber is the exhaust port under the DVI ports, by sticking my nost to the grill under the DVI I can definately smell burn......Though I dont know what location that translates too...

Also I noticed the green lights on the card were still active though the card wouldnt even render a screen....after it "died" the monitor would just flash no signal.

Although I have on more bit of information to disclose, upon first recieving the card I didnt notice the exhaust port and for the first 2 hours of owning it I didnt remove the back paneling on the case that vented the exhaust out the back. After my case temp got around 38c I realized that I was blocking the vent and immediately fixed it, but the card ran the same as It did until it died 3 days later so Im not sure if that was another factor

 

[VirtualLarry]

Originally posted by: o3srt4me
Although I have on more bit of information to disclose, upon first recieving the card I didnt notice the exhaust port and for the first 2 hours of owning it I didnt remove the back paneling on the case that vented the exhaust out the back. After my case temp got around 38c I realized that I was blocking the vent and immediately fixed it, but the card ran the same as It did until it died 3 days later so Im not sure if that was another factor

That's not good. It could have cooked.

How did you get the card installed, if you didn't remove the bracket cover for that slot?

 

[o3srt4me]

my friend was baffled by that too lol, My case has break-out brackets that you first have to literally push out with your thumbs that are flush with the case....then once you do that it has the thicker slide in type if you needed to close it off again, since this was my first modern card, I didnt know they needed two slots and since it fit just fine lol, I thought I was made....


Yea I guess I have so many unknown variables over this death I just need to buy a new PSU, and a fresh card and once the PSU base is covered ... the...bracket is already removed, and I have a another card. The only viable answer for the next cards death is defective card.

 

[techboie]

If the PSU is old and/or less than Earthwatts/NeoHE then it is highly likely to have been the culprit. Even otherwise even a NeoHE just may handle a 4870, it should actually, provided the rest of the rig doesn't take too much.

 

[o3srt4me]

Originally posted by: techboie
If the PSU is old and/or less than Earthwatts/NeoHE then it is highly likely to have been the culprit. Even otherwise even a NeoHE just may handle a 4870, it should actually, provided the rest of the rig doesn't take too much.

The PSU is an antec 500w earthwatt, bout a year or so old

 

[MBrown]

I posted a thread earlier and they said that my 500w NeoHE can handle a gtx 260 core 216 and a q9400. I am not sure how much of a difference there is between an earthwatts and a NeoHE but based on how much I know, I don't think it would have been the supply that killed it.