PSU problem?

[lai_cc]

Hi,

My PC is about 3yrs old with AcBel 510 (450W) PSU.
Just changed my graphics card last year to Palit GTX550Ti.

Recently, I have been getting screen freezes and eventually it will crash to blue screen with error on nvlddmkm.sys. When i start to see black squares flickering on the screen, the crash will follow.

Tried updating driver and tested the graphic card on another PC. No help (not a problem with graphic card). I saw someone posted a similar problem and it mentioned about power supply problem. I try to check with HW monitor, it seems that my +12V supply is only 9V!!?! +3.3V and +5V seems normal. Could it be a faulty/aging PSU?

Any advise would be appreciated.

 

[melovelinux]

do a memtest on ur ram first

 

[jtjt00]

TS, high chance is PSU.

 

[lai_cc]

do a memtest on ur ram first

will try memtest when i get the chance.

 

[lai_cc]

TS, high chance is PSU.

Strange thing is CPUID HW Monitor reported 9V for +12V supply but bios claim it is 12V for it. Not sure which one is inaccurate.

Anyway, my PC spec is as follow. Not sure if the PSU is enough.
PSU 450W
CPU i5-2400 (3.1Ghz)
GPU NV GTX550Ti
Mem 2x4GB + 2x8GB
HDD 2xSATA
USB 1wireless adapter
Fan 3x120mm

I dont overclook. This spec seems to be simple to me which don't really consume a lot of power. Correct me if I am wrong.

 

[tuckloong]

Strange thing is CPUID HW Monitor reported 9V for +12V supply but bios claim it is 12V for it. Not sure which one is inaccurate.

Anyway, my PC spec is as follow. Not sure if the PSU is enough.
PSU 450W
CPU i5-2400 (3.1Ghz)
GPU NV GTX550Ti
Mem 2x4GB + 2x8GB
HDD 2xSATA
USB 1wireless adapter
Fan 3x120mm

I dont overclook. This spec seems to be simple to me which don't really consume a lot of power. Correct me if I am wrong.

Do u connect a lot of devices via USB?

 

[lai_cc]

Do u connect a lot of devices via USB?

No, I have only 1 wireless adapter and sometimes usb thumbdrive. Don't think they are heavy on power.

It all seems to get started after i run COH2 for an hour or so.
I suspected overheating issue but the GPU temp is fine.
Since then I have not run COH2, just normal surfing, it still crash after sometime. Is there a chance of permanent overheating which might have damaged the GPU?

 

[deepblue_82]

correct me if i am wrong.
ur "rated 510w psu" is really only a 300+ psu...

 

[lai_cc]

correct me if i am wrong.
ur "rated 510w psu" is really only a 300+ psu...

according to the manufacturer, its output is 450W (max 510W)
AcBel Polytech Inc.

 

[lai_cc]

do a memtest on ur ram first

Did memtest86, no error found. Still getting weird flickering black pixels over the screen after sometime.

 

[Maeda_Toshiie]

Ok,

1. HWMonitor's voltage readings can and are usually wrong. Do not use them.

2. Bios readings can also be wrong. The most accurate is to use a multimeter.

3. PSU - maximum +12V is only 360W. Borderline. Add in aging and that this is not a high quality unit, PSU is not able to supply sufficient power.

4. Note: when the system is heavily loaded (in this case, heavy power draw on the +12V rail(s)), the vdroop on the +12V may be too much for the system to maintain stability.

5. 450W is the maximum continuous, 510W is the maximum peak (over a short period of time).

5. TL;DR - Get a better PSU.

6. I won't use such PSU, not when there is a >= mid range discrete card.

 

[lai_cc]

Ok,

1. HWMonitor's voltage readings can and are usually wrong. Do not use them.

2. Bios readings can also be wrong. The most accurate is to use a multimeter.

3. PSU - maximum +12V is only 360W. Borderline. Add in aging and that this is not a high quality unit, PSU is not able to supply sufficient power.

4. Note: when the system is heavily loaded (in this case, heavy power draw on the +12V rail(s)), the vdroop on the +12V may be too much for the system to maintain stability.

5. 450W is the maximum continuous, 510W is the maximum peak (over a short period of time).

5. TL;DR - Get a better PSU.

6. I won't use such PSU, not when there is a >= mid range discrete card.

yeah I think hwmonitor is not accurate. I tried using the motherboard's monitor and found that 12v supply can drop to 11.7 when playing game. 3v seems constant at 3.28 and 5v at 5+

I think better to get 600+w psu. thinking of cm 650w. any other recommendations?

 

[lai_cc]

anyway I had the impression that gtx550 is considered low range

 

[Maeda_Toshiie]

GTX 550 Ti is not really low range. Closer to mid range. GT 520, 530 are low range.

If you get a quality 500-550W, you don't need 650W, since your CPU is not overclockable and you aren't using a Titan or 780. Your current system will draw less than 400W (though most of it is at the +12V). A good quality PSU 500-550W will do.

 

[lai_cc]

Update

Just to update on my case the series of event for anyone interested:
- few months ago, polygon 'tearing' starts to show up during game. It would get worst till I can hardly see anything on screen. So decided to update 306.97 nv driver to higher version.
- black flickering squares starts to show up during games, sometimes in combination with polygon 'tearing'. If play long enough, crash would happen.
- tried to update display driver when new ones come out. problem persist.
- This leads me to suspect GPU too weak (or old), but some online post suggest PSU aging or memory (RAM) fault
- Eliminated memory fault with memtest and test on another PC
- In the mean time, GPU will crash on me after 1 or 2 hours of normal work (word, internet ... etc) most of the times.
- Tested GPU on another PC for 2 hours, no such issue.
- Tried to gauge the PSU status with voltage measurements from 3rd party software and BIOS. BIOS seems to provide somewhat more accurate readings which indicates signs of unstable 12V supply. At idle load, 12V supply is around 11.7+. In games, i suspect it could go below 11.
- Decided to get a new PSU. End up getting faulty new PSU which complicates my issue (that's another problem).
- Now, finally, my PC seems to be stable again. No more polygon 'tearing' or black flickering squares or BSOD. I hope it stays that way for another 3yrs

 

[tuckloong]

ya finally...happy to hear that

 

[ktee]

Great to hear that!

So in this case, it was the PSU problem. Same things happened to me long ago. It was the RAM (crusial ballistix) fault - memtest - sometimes no problem (even tested at agent), sometimes error. After a kind forumer gave me his Hyper 4gb, it is smooth sailing til now.

So, i a bit phobia of ballistix, the bling bling is tempting but the BSOD is frustrating. Like a beautiful woman with a bad heart. Sadness>happiness.

 

[lai_cc]

Great to hear that!

So in this case, it was the PSU problem. Same things happened to me long ago. It was the RAM (crusial ballistix) fault - memtest - sometimes no problem (even tested at agent), sometimes error. After a kind forumer gave me his Hyper 4gb, it is smooth sailing til now.

So, i a bit phobia of ballistix, the bling bling is tempting but the BSOD is frustrating. Like a beautiful woman with a bad heart. Sadness>happiness.

I do have crucial ram 8GBx2 (not those with blink blink one) they work great. My old PSU still can power up, just that it is not stable. Never encounter this kind of PSU issue having gone through several generations of CPU (P3 > P4 > C2D > i5).

 

[Mythmaker]

Great to hear that!

So in this case, it was the PSU problem. Same things happened to me long ago. It was the RAM (crusial ballistix) fault - memtest - sometimes no problem (even tested at agent), sometimes error. After a kind forumer gave me his Hyper 4gb, it is smooth sailing til now.

So, i a bit phobia of ballistix, the bling bling is tempting but the BSOD is frustrating. Like a beautiful woman with a bad heart. Sadness>happiness.

Agree. Ballistix failure rate is insane. Never trust Crucial rams lol.

 

[melovelinux]

Agree. Ballistix failure rate is insane. Never trust Crucial rams lol.

Shocking leh,