Sign-in / Sign-up
Your question
Solved

Do I need a new PSU?

Tags:
  • Power Supplies
  • Components
Last response: in Components
October 3, 2010 10:22:11 PM

So for a while now I've noticed this sort of real high-pitched whistling coming from my computer...I had always assumed it was my graphics card since it only occurred during the most strenuous gaming. But since I upgraded to my current config (in my sig) from 2x8800GT's and an OC'd Q6600, when the whistling gets louder the system just shuts off, with no blue screen or anything. But it's inconsistent: playing Crysis or Crysis Warhead will make it whistle (crysis Warhead actually has it whistling in just the intro logo videos), and the 3D video benchmarks in the PC-Matic tool make it whistle and then shut off, but I can play Civilization 5 for hours with no problem, as well as Splinter Cell Conviction, and regular usage or just HD movies have no effect either.

So then I bought a power supply tester and everything passed except the +3.3V rail, which displayed as having higher than 4.5V. The power supply is an OCZ GameXstream 850W unit that's 80Plus certified and SLI-ready (I know not to buy cheap PSU's). Anyways, I'm wondering if this is a fatal flaw or if there's something I can do to lock it down to a lower voltage.

More about : psu

a c 165 ) Power supply
October 3, 2010 10:26:28 PM

your 3.3v should be stable and not reaching 4.5v

a little off topic but anyone ever walk into a microcenter and see all those OCZ refurbished PSU for sale?
m
0
l
October 3, 2010 10:36:14 PM

ct1615 said:
your 3.3v should be stable and not reaching 4.5v

a little off topic but anyone ever walk into a microcenter and see all those OCZ refurbished PSU for sale?


Haha, actually yes...wonder if I should've taken that as a sign...but always thought OCZ was a solid brand...anyways, I know it SHOULDN'T be reaching 4.5V, just wondering if there's any way to fix it short of buying a new PSU...really don't wanna have to do that...
m
0
l
Related resources
a c 165 ) Power supply
October 3, 2010 10:41:59 PM

delluser1 or johnnylucky would know more about popping open a PSU, I would ask them.

I think OCZ is a solid brand when it works....they get good reviews but...quality control could be better
m
0
l
a c 144 ) Power supply
October 4, 2010 4:02:57 AM

My objection to OCZ is that they have at least 4 PSU lines. One is pretty good, one is so-so. The other two are excessively mediocre. And I just can't remember which is which.

But before you replace the PSU, borrow a meter and verify that the 3.3 volt output really is out of spec (+/- 5%). At the main power connector, measure between any orange and any black wire.

You are not going to be able to repair the PSU yourself and it will cost you more than the PSU is worth to pay someone else to repair it for you.
m
0
l
a b ) Power supply
October 4, 2010 5:33:36 AM

you can also try a software hardware montior such as hwmonitor and look at the voltages while the system is running to confirm this. If your confident, open up the PSU and look for leaky or bulging capacitors. If its just a bad cap it would be easily repairable if you know how to solder. Better a 50c capacitor than a whole new PSU.
m
0
l
October 4, 2010 6:34:57 AM

iam2thecrowe said:
you can also try a software hardware montior such as hwmonitor and look at the voltages while the system is running to confirm this. If your confident, open up the PSU and look for leaky or bulging capacitors. If its just a bad cap it would be easily repairable if you know how to solder. Better a 50c capacitor than a whole new PSU.


Ugh...HWMonitor reads 3.31V...right on target...now what?
m
0
l
a c 143 ) Power supply
October 4, 2010 6:39:21 AM

Replacing a cap in a motherboard is one thing, but telling someone to replace a cap in a PSU is something else. If its the main cap and he doesn't bleed it first, he'll be in for the shock of his life. Possibly the shock that ends his life. From what I know of PSUs a bad cap won't cause this problem. More likely a failing rectifier(?).
m
0
l
October 4, 2010 7:22:53 AM

Ugh...okay so I just ran 3 DX11 benchmarks: Heaven, AvP, and STALKER Call of Pripyat, all worked fine, no whistling, no crashing. The I ran Prime95: 8 threads, blend mode, again no whistling or crashing. Then Furmark at 1920x1080, 8xMSAA, and Xtreme Burn-in mode, only got 13 fps so I KNOW it was killing the GPU but again, no issues. Then ran Prime95 and Furmark AT THE SAME TIME...which may or may not be healthy for my comp...and again, no issues. But as soon as I fire up Crysis Warhead, just in the logo videos the whistling starts immediately. I start up a new game and the system shuts off before the intro video even finishes...oddly enough in both Crysis and Crysis Warhead, the game crashes before the new game intro video finishes, but on both, I can load up a saved game and play with no sign of whistling and no crashing. I should also mention that when the system shuts off it's instantaneous, no shutting down screen, no blue screen, just...off.
m
0
l
a b ) Power supply
October 4, 2010 7:50:18 AM

That sure sounds like heat or psu, but heat doesn't seem to make sense compared to your P95/Furmark runs.

But I notice you aren't giving us any temperatures.

How about getting some numbers, using CPUID's Hardware Monitor, for core temps and gpu temps during P95/Furmark, and as you launch Crysis. maybe we can conclusively rule out heat.
m
0
l
October 4, 2010 11:56:34 AM

Twoboxer said:
That sure sounds like heat or psu, but heat doesn't seem to make sense compared to your P95/Furmark runs.

But I notice you aren't giving us any temperatures.

How about getting some numbers, using CPUID's Hardware Monitor, for core temps and gpu temps during P95/Furmark, and as you launch Crysis. maybe we can conclusively rule out heat.


CPU reaches 68C, GPU reaches 75C. Reason I hadn't been considering heat is mostly because the shut offs are so reproducible. If I run Crysis or Crysis:WH (from a new game) or that specific benchmark, even right after booting up, the system will crash and there's no way the system will have had time to overheat.
m
0
l
a b ) Power supply
October 4, 2010 7:06:25 PM

Thanks for the info. Sometimes the absence of numbers means the OP has got that under real control, other times it means a wet finger test may have been used.

You have the whistling across two gpu installations (2x8800, 5870); the PC doesn't "whistle", overheat, or shutdown under extreme stress of P95/Furmark; the PC "only" whistles during "extreme gaming", but doesn't shut down; however it shuts down in Crysis from a new game or benchmark, and it will do so from a cold boot.

The psu isn't the best, and a 700W sample from the same line exhibits out of spec voltage ripple. (see link earlier)

The odds still seem heavily in favor of the psu needing replacement, though I suppose it could also be the mobo, which may have been damaged by the psu. I'd replace it ASAP with a Corsair, Seasonic, Antec Earthwatts, or other psu that got a specific passing grade from a qualified reviewer.
m
0
l
October 4, 2010 7:33:07 PM

Twoboxer said:
Thanks for the info. Sometimes the absence of numbers means the OP has got that under real control, other times it means a wet finger test may have been used.

You have the whistling across two gpu installations (2x8800, 5870); the PC doesn't "whistle", overheat, or shutdown under extreme stress of P95/Furmark; the PC "only" whistles during "extreme gaming", but doesn't shut down; however it shuts down in Crysis from a new game or benchmark, and it will do so from a cold boot.

The psu isn't the best, and a 700W sample from the same line exhibits out of spec voltage ripple. (see link earlier)

The odds still seem heavily in favor of the psu needing replacement, though I suppose it could also be the mobo, which may have been damaged by the psu. I'd replace it ASAP with a Corsair, Seasonic, Antec Earthwatts, or other psu that got a specific passing grade from a qualified reviewer.



Actually, there was no whistling when I had the 2x8800GT's, then I got the HD5870 followed by a bigger PSU a few days later...can't really tell you when the whistling started...I've had both for ~6 months.
m
0
l
a b ) Power supply
October 4, 2010 8:00:50 PM

OK.

Do you have that old psu around? If so, make/model? A surprisingly small psu should drive your 5870 just fine - would be a simple test.

For future reference, do you have the 8800's around?
m
0
l
October 4, 2010 11:15:24 PM

Twoboxer said:
OK.

Do you have that old psu around? If so, make/model? A surprisingly small psu should drive your 5870 just fine - would be a simple test.

For future reference, do you have the 8800's around?


No, I'm afraid I sold the 8800s and the PSU is currently residing in my little brother's computer. However, I talked to a tech guy from OCZ and he said something about a feedback loop? Said that somehow the monitor, PSU, and GPU are all sort of synced together and when the frame rates get real high then that sound can be caused by coils vibrating real fast. It actually makes sense since that would explain why the hefty benchmarks don't phase my computer, but during that other bench and the intro videos I'm getting like 4000+ fps. Though he said it still shouldn't be crashing... So I forced on Vsync for everything through the ATI control panel but that doesn't seem to affect the intro videos or that one particular benchmark...anyone know how to set a hard lock on FPS...like around 800fps or so? I ran the Call of Pripyat bench at the lowest settings and 800x600 and it maxed at ~950fps, and that caused the whistling but much fainter than before...so maybe if I could just lock the fps, that would fix it.
m
0
l
a b ) Power supply
October 4, 2010 11:24:01 PM

What the OCZ guy said is interesting for sure, but the crashing and the voltage issue you reported earlier remain unexplained.

Can you grab your bro's psu for an hour?

And here I thought there was no such thing as too much horsepower lol.
m
0
l
October 5, 2010 12:14:42 AM

K, so I grabbed my bro's PSU (guess there are some perks to living somewhat close to my parent's house) and verified that the power supply tester IS working, and there is almost certainly and issue with the PSU, as the other PSU didn't make the whistling noise. I know next time to go with a better brand, but for now I'm gonna RMA this one since it's still under warranty and I'd rather not spend $200, and OCZ apparently offers cross-shipping which is nice. In the meantime, does anyone know if there is a way to permanently lock down your fps? ATI's forcing "wait for vertical refresh" doesn't seem to be working since I ran the Heaven benchmark on low settings and it was way over 150fps (shouldn't it be locked at 60?) and running Fallout 3 with some crazy texture mods was still around 100fps...not sure what the deal is...
m
0
l
a b ) Power supply
October 5, 2010 12:30:13 AM

Pass on my thanks to your Bro . . . if he's aware of his "loan" :) 

Vertical synch should hold stuff at 60 FPS, but as you've seen it doesn't for everything. Sometimes games have their own V-synch . . . I don't have Crysis here to check.

Glad the problem is resolved. Watch the new psu - remember, they have ripple under stress. If you start getting that whistle again, don't risk your mobo's regulators.

I'd advise you post the V-Synch question separately. Questions with lots of replies are often not looked at by the folks you'd most want to see it.
m
0
l
a b ) Power supply
October 5, 2010 2:41:49 AM

So the shut down problem is fixed with the other PSU?? that whistling noise from the psu is components vibrating/ressonating. Normally when you look in a PSU, you will see a whole lot of glue type stuff between capacitors and things. That holds it together so that it can not ressonate and create the noise you are hearing. Chances are, it has cracked or come apart with heat stress.
m
0
l
a b ) Power supply
October 5, 2010 2:52:59 AM

Probably. I don't know if motherboards can whistle, but the voltage ripple from those psus was shown to be out of spec at one time. I like that as the root cause, whichever part was crying out in pain.
m
0
l
October 6, 2010 10:48:17 PM

Best answer selected by bikeracer4487.
m
0
l