need opinions on supposed PSU voltage "fluctuations"

Hi all,

 

For the past 2 months I've been trying to find the cause of frequent video artifacts I see when gaming. My card is a GTX 570. Things are fine for a couple days after a cold boot, but eventually artifacts start creeping into games to the point that I can't see the screen, and I finally have to shutdown, turn the power off, then start my PC again. After doing that things are fine for another couple days, but then the cycle repeats.

 

So I've been working with my video card manufacturer to narrow down the problem. I've tried all the normal things (memtest86+, hard-coding the RAM timings into bios, reseating the card, loading the video drivers cleanly several times, plus tons of other stuff). Now they think it might be my PSU (a Corsair AX1200). However they said something I'm not sure I trust, and I'd like to get your reactions to this:

 

TECH SUPPORT: Sometimes a PSU which is otherwise functioning normally may develop voltage fluctuations if left on for more than a day or so. To test, could you try rebooting once a day for a couple days and see if it eliminates the problem? ...

 

TECH SUPPORT: The main type of fluctuations to look for are any fluctuations of the 12v rail that are more than 0.05v. It is not likely for these fluctuations to cause damage to the card, but they can cause instability while they are happening. The 400 and 500 series cards can be more sensitive to these fluctuations.

 

ME: I think you mean 0.5v, not 0.05v, right? 0.05v would mean only a 0.417% fluctuation in voltage! The acceptable variation from a stated power supply flow is + or - 5% for each rail, which would be a 0.6v variance...

 

TECH SUPPORT: The voltage can be within +/-5% but if it fluxulates +/-.05v it could indicate an issue. Due to the difficulty and variances of getting a voltage dead on this is why the voltage can range by 5% but if it fluctuates from where it normally sits then it can indicate a power supply issue possible bad power due to failing components.

 

Sorry for the long quote but I wanted to give you guys the context. What is your reaction to that last paragraph? Is that really true? I did take a quick look at the voltages over time and I've seen it vary from its usual 12.096v down as far as 12.000v a few times. To be certain of course I'll use my multimeter if it comes to that.

I did take a quick look at the voltages over time and I've seen it vary from its usual 12.096v down as far as 12.000v a few times. To be certain of course I'll use my multimeter if it comes to that.

Even at 5% varience you are still not dipping down below 12v. I dont see where that would be an issue. This in no way indicates a failing power supply. If you were down around 11.4 i could see it. But not at 12v

Short answer, you're not crazy, they are.

 

According to authoritative power supply reviews I have seen, the majority of PC power supplies meet the +/- 5% ATX specification. That is while being tested at their full rated output on the 12V rail, and while the other rails are being loaded to an appropriate level. High quality, high power, or "High End" power supples can maintain 12V regulation of 2% - 3%, and some are capable of 1% 12V regulation. That is at their full, rated 12V output. The time frame of sustaining that regulation is unknown to me, but I doubt the tests are maintained for days at a time.

 

Fluctuations of 0.05V on a 12V or any PC power supply rail is very small, as you know. I use a hardware monitoring program that displays all the standard voltages to three decimal places, although I don't know how accurate it really is. I've used it on multiple PCs, with differing power supplies (PS). This program updates it's display every two seconds. Some PSs have random, small fluctuations, up to 0.2V, with the PC at idle. Those fluctuations happen constantly, within the two second update period of the program. Others are much more constant, and only change when under a heavy load, or occasionally at idle.

 

My new S_Sonic X-series PS, which is one of their top of the line models, varies occasionally between two voltages on the 12V rail, at idle. The difference in those voltages is 0.056V according to this program. Your AX1200 is in the same class of PSs as this unit. What is higher quality than these units? Nothing, these are the top models that exist.

 

Requiring a PS to never fluctuate by 0.05V is ridiculous. Given the entire system providing power, from the electric utility to one's house wiring and the PC's power supply, how is that possible? What length of a 0.05V dip in voltage will cause this video card to become unstable? If a PC has been running for days, how does a restart of the PC refresh the PS or give it a rest? A game could be paused for a few minutes, really giving the PS a rest. If you did this, would you still see the artifacts?

 

Fluctuations of 0.05V can't indicate a PS problem, otherwise we all have bad PSs. IMO, the video card company is looking for excuses. Has anyone seen a PS of their's whose voltages do not change over time? When I first purchased expensive PSs, I was impressed by how stable the voltages were, and they are very stable, but never changing by 0.05V is unreasonable and unheard of in any specification and in actual use. If a video card requires that level of voltage consistency, then the manufacture should state that. What PS manufacture does this company recommend? Do they think it is typical for a PS to be unwavering on it's voltage rails?

Thanks for the answers, I appreciate it. Yeah, I thought the +-.05v requirements were way too strict. I've never had to deal with "fluctuation" issues though. If anyone else could report what their normal fluctuations are for their 12v it would be helpful for comparison. My 12v fluctuates between 12.096v to 12.000v, and it hits various values between those 2 extremes. Seems kind of random, though mostly during gaming it goes to the low side.

 

I hope it doesn't come down to a war between the video card maker and Corsair support. I doubt Corsair will want to replace an AX1200 because of a .1v fluctuation!

FWIW, my PSU fluctuates down to 11.89 and i have never had a problem with video or anything else for that fact. And that is with my I-7 920 OC'd to 4.3ghz. If you were dipping way below 12v i could see it.

 

Same goes for my 3.3v. It will dip down to 3.1 ocasionally and it hasn't posed any problem either.

 

Your video card manufacturer just needs to replace the card and get over it

I finally tested the 12v with a multimeter and the actual readings are 12.32-12.33v during cpu and gpu stress testing. The software monitor showed 12.000-12.096v at the same time so obviously software monitoring can't be trusted.

 

This shows that my PSU is NOT fluctuating more than .05v like I thought.

 

Thanks for the feedback guys!

The ATX spec for power supplies is 5% fluctuation on the 12V rail, and the best units that exist can do ~1%, but they are rare. One percent of 12 is 0.12, so 0.05V is less than one half of one percent. If 0.05V changes in the 12V rail indicate PS issues, then every PC power supply in the world has issues.

 

A video cards own voltage regulation must deal with any changes if voltage being that steady is needed.

A video cards own voltage regulation must deal with any changes if voltage being that steady is needed.

 

Thanks for verifying that because I was thinking the same thing. And as we all know the GTX 570 VRMs are CRAP which is why a lot of them are frying.

I've recently been having the same issues. I have the GTX 470 video card of which I'm on my second RMA. I originally thought that it was just a video card issue as I was getting some artifacts during use. I got my replacement yesterday and after twenty minutes of use both of my displays failed to show anything. After contacting support again they mentioned that my power supply, which is a Corsair TX750W, could be to blame. At the bios it fluctuates between 12.302V and 12.239V. Could the power supply be to blame or is it the video card?

How are you measuring your voltages? If via software I don't think you can trust it. If via voltmeter that's better. The best way is via oscillosope (so I've heard). Read my post above where my software shows fluctuation but the actual readings via voltmeter (multimeter) doesn't.

 

btw, are you the same meowmix that plays TF2? :laughing:

I've just been measuring it through my bios. Where did you get your voltmeter? I have TF2, but I was using it for testing purposes of my video card. Did you ever resolve that issue with EVGA?

No that fluctuation is nothing but normal usage. Those voltages are fine if you get the same reading with a multimeter.

Check the 5v and 3.3v rail too.

Everything checks out ok with the multimeter. You guys are right, the bios isn't accurate. I've sent them the video card back so this will be video card number 3. *sigh*

At the bios it fluctuates between 12.302V and 12.239V.

 

According to EVGA if the PSU fluctuates more than .05V it can cause problems with the card. I asked them about that to confirm what they said! They said that it's ok if the card's voltage is anywhere between 11.4V and 12.6V (a 5% variance) but once the PSU is turned on it can't fluctuate from whatever voltage it's at. Make sense?

 

So my PSU is always at 12.33V but sometimes fluctuates to 12.32V. That is only .01V fluctuation which is fine. But your fluctuation is .063V (12.302-12.239) which is more than .05V. However your BIOS readings are probably not accurate. For example, I used software to monitor my readings and the software said my PSU fluctuated between 12.00 and 12.096. So at first I thought I had a problem. So I used a multimeter and tested the actual 12V rail and found it to be only 12.32V to 12.33V.

 

You can buy a voltmeter/multimeter at any hardware store. They sell for less than $10. Just search for multimeter at Amazon to get an idea of what they look like and what people say about them, etc.

Everything checks out ok with the multimeter. You guys are right, the bios isn't accurate. I've sent them the video card back so this will be video card number 3. *sigh*

 

oops, I just now saw this post so my follow-up above is kind of irrelevant now. Thanks for following up. BTW, I'm getting ready to RMA my 3rd EVGA 570 card so I'll be on my FOURTH card soon. sigh :[pouts:

Great. Hopefully I won't have to RMA a fourth card. This time I'm trying the card out on one of my buddies rigs to see if it works with him before I even try it on my computer. If it works there and not for me then I'm going to replace the power supply. I was told by EVGA that if I keep calling them with issues and RMAs then it starts to look fishy. Loving this back and forth with EVGA...

I just started talking with EVGA this morning about another RMA. My first card caught fire, the 2nd card artifacted, now this 3rd card is artifacting. They would like me to try another PSU but I don't have one to try. They said that although my PSU is within spec (12.33-12.34V) the voltage is "higher than they would like to see". I would think if a manufacturer found their cards didn't work with "high" voltages that are still within spec then they'd do a recall of the cards. :-/

OMGosh, so let's see;

 

Don't call so much, it's suspicious (aka, we're just sick of dealing with you.)

 

Your PS voltage is in spec, but too high (aka, lets see how smart this guy is.)

 

My video card caught fire, why? Fans are running to fast, to much oxygen! That's Ok, why do you think it's called Crossfire?

Your PS voltage must never change from it's start up value (aka, by the time he goes through ten PS's, which will never do that anyway, he'll be to broke to call.)

 

And from the OP, +/- 0.05V means instability of your PS, restart your PC so PS is reset and gets a rest, (aka, I hope he buys that excuse, it's lunch time.)

 

Did you notice that EVGA sells a PS that has a control to adjust the 12V rail voltage up or down a bit? Hmm, I wonder if all this talk about the 12V rail being just perfect is related to that? And even better, that PS supports all OS's! (Yes, it actually says that.)

 

And of course the all time classic, That's not an issue, it's a feature.

So I got my third card in the mail today. They were nice enough to bump me up from the 470 to the 480. Worked well initially, but unfortunately I'm still getting the sporadic crashes I had before. I'm beginning to think power fluctuations, although minor are causing the crashes. *sigh*

So I got my third card in the mail today. They were nice enough to bump me up from the 470 to the 480. Worked well initially, but unfortunately I'm still getting the sporadic crashes I had before. I'm beginning to think power fluctuations, although minor are causing the crashes. *sigh*

There is no way. You would be the very first if this was true. There has to be something ellse going on in your sytem.

 

I would take RG up on his offer to RMA the PSU to rule it out. Al though i really think you have something else going on. I actually suspect your MB from everything you have posted so far.

 

Are you getting a BSOD or are games just crashing back to your desktop?

 

If you are getting BSOD's what is the error code?

I'm thinking about the RMA. I reinstalled windows today so I'm going to go home after work and test it out. I never get the BSOD, it just takes me back to the desktop.

I agree with Peanutz, at the very least that it is not the PS, unless it has an issue with it's 12V rail. If it was to low, I imagine the video card itself would shut down, and you'd have no display at all.

 

Think about all the testing and benchmarking that is done on video cards at many hardware review websites. If it was common, or even unique, for one or more cards to crash due to PS voltage fluctuations that are (well) within the ATX PS spec, we would know about it. If a video card, or any PC component, requires it's voltage supply to be stable within +/- 0.1V, which is more than you know who said it should be, then the problem is the product, not the PS. It becomes the products responsibility to provide such a stable voltage, or the manufacture must state what power supplies are approved for usage with the product.

 

Hmm, 32 bit Windows... are you running out of memory, given the 4GB maximum of a 32 bit OS (actually a bit less than 4GB.) Are you using a UPS? The US EVGA website does not show 470 or 480 type cards on their product lists anymore. Am I just lost or what?

No, I'm lost. I'm sorry, I don't know what a UPS is. If you told me I could get back to you on that. My installed memory is 4.00GB of which is says (3.25GB is usable). Processor is AMD Phenom II X4 965 3.40 GHz. Hard drive has a terabyte of memory. My MB is a gigabyte "Ultra Durable" 3 MB GA-890GPA-UD3H. I have 64bit, but I switched to 32bit when all of this started happening. For some reason I told myself 32bit would give me less issues. When I tried to go back this time around it told me my 64bit disc wasn't valid, which is a lie because I bought the OS right off of newegg.

No, I'm lost. I'm sorry, I don't know what a UPS is. If you told me I could get back to you on that.

Uninteruptable Power Supply or battery back-up? :)

 

 

If it was common, or even unique, for one or more cards to crash due to PS voltage fluctuations that are (well) within the ATX PS spec, we would know about it. If a video card, or any PC component, requires it's voltage supply to be stable within +/- 0.1V, which is more than you know who said it should be, then the problem is the product, not the PS..

There also would be no overclocking/overvolting GPU's either! :)