PSU Dying?

S

SlvrDragon50

Limp Gawd
Joined
Apr 28, 2013
Messages
346
How do I tell if my PSU is dying? It's a pretty old Antec Neopower 550W (5+ years old) powering a 7950 and 4.3 GHz i7 2600k.

I'm having weird boot issues where my computer can't boot while I have a Dell 24" monitor connected via DVI-D simultaneously with a Vizio 32" TV connected via HDMI.

My USB powered speakers also make this buzzing sound during boot up sometimes as well.

My keyboard and mouse also sometimes lose power during the boot up process until the login screen pops up.

And sometimes the login screen won't even pop up.

Is it something that I can't really tell until I replace the PSU?
 
It's something you can't tell for certain until you start replacing parts with known good ones. I'm more inclined to think bad motherboard or RAM. How does it perform under load (i.e. OCCT PSU test, simultaneous furmark and IBT)?
 
I've never had any problems gaming, but every now and then I get random BSOD. Usually driver related.

Is OCCT PSU the only one I need to run or do I need to do the Furmark and IBT as well?

I think I ran Furmark and Heaven when I was doing my overclocking.
 
1. OCCT PSU or 2. Furmark + IBT/P95. One of the two. Either way, the point is to load up your system as much as possible to fully stress out the PSU. If the PSU has no problems doing this for an extended period of time (think overnight), then it is unlikely your boot issues (which demand much less power) are PSU related.

Furmark and Heaven stress out the same thing: the GPU. They both do not fully load up a system.
 
It sounds like you might just have an overloaded 5v rail, or failing capacitors. Ripple on the 5v rail would be a good culprit for the buzzing on your speakers.

Test with a known good PSU and go from there.
 
Ok. I just read that someone's computer bricked after running OCCT PSU.

I placed an order for a 40 dollar EVGA Supernova G2... I have a feeling it won't ship since the seller has a 61% rating, but it's through Amazon....

Is there a high chance of OCCT PSU bricking the computer or do you think it was just a fluke?
 
I've run OCCT PSU tester hundreds of times on several different computers. I've had one burst into flames, and two pop the PSU. Other then them, it's either been PSUs safely going into protection, or the system running fine.

It does what it says it's going to do, put the system on absolute maximum load to test the PSUs capability to keep up.
 
Awesome thanks. I'll run the test then! Worst case I'll pick up a PSU off someone here and be without my computer for a week or so.

Should I just do 20 minutes or infinite? Overnight seems a little overkill?

Just ran for like 40 minutes, absolutely no problems.
 
Last edited:
SlvrDragon50 said:
Awesome thanks. I'll run the test then! Worst case I'll pick up a PSU off someone here and be without my computer for a week or so.

Should I just do 20 minutes or infinite? Overnight seems a little overkill?

Just ran for like 40 minutes, absolutely no problems.
Click to expand...

40 min is long enough to prove the power supply seems to be ok.....what are your reading on the different rails before/after and during the stress test would pretty much rule out the supply 99% if they all stay pretty much dead on the hole time (1% deviation would be ok)
 
primetime said:
40 min is long enough to prove the power supply seems to be ok.....what are your reading on the different rails before/after and during the stress test would pretty much rule out the supply 99% if they all stay pretty much dead on the hole time (1% deviation would be ok)
Click to expand...

Uh... which photos are those?
RgRDx29.png

uCkXaqb.png

awWSLPq.png


Or is it the VIN?
 
Software readings tend to be inaccurate. Pull out a molex or sata plug, and test the 12v and 5v lines (3.3v as well with SATA) using a multimeter. 12v is yellow, 5v is red, and 3.3v is orange. You can use any of the black wires as ground. If you do test SATA, make sure you look up which pin is which. Use that to confirm your voltage readings are within ATX spec.
 
HWMonitor and HWiNFO64 are both pretty good at reading the voltages as well.....i suppose hooking a volt meter and checking the molex plugs could be done as well, but imo the software is usually pretty accurate
 
These readings are through HWInfo. Unfortunately I don't have my multimeter with me.

So are those photos good?
 
primetime said:
HWMonitor and HWiNFO64 are both pretty good at reading the voltages as well.....i suppose hooking a volt meter and checking the molex plugs could be done as well, but imo the software is usually pretty accurate
Click to expand...

Software readings might sometimes be accurate but are known for being inaccurate.
 
SlvrDragon50 said:
These readings are through HWInfo. Unfortunately I don't have my multimeter with me.

So are those photos good?
Click to expand...

from what voltages are listed yes.....but i don't see the 5volt or 12volt buss listed.:eek:

could try out hwino64

Capture_zps7253c764.png
[/URL][/IMG]
 
Yea, I don't have any pictures of those voltages from OCCP.

Guess I have to run OCCP and HWInfo simultaneously.

edit: I believe the numbers look good.

I1HH1kI.png
 
Last edited:
SlvrDragon50 said:
Yea, I don't have any pictures of those voltages from OCCP.

Guess I have to run OCCP and HWInfo simultaneously.

edit: I believe the numbers look good.

I1HH1kI.png
Click to expand...

if you were running cpu and gpu near 100% usage for a while and that your numbers, then by all means it appears your power supply is fine. Yes software can report inaccurate numbers but its more rare than "tends to" these days. That and the fact you can also double check readings in your bios from time to time just to confirm there lining up (while unstressed) with what appears to be accurate reading
I for one have a multi meter and have never noticed any inaccurate reading from my board reporting the voltages levels...does it happen yes but its never happened on any board i have owned.
Im my mind there are some app here and there that don't report as accurate as others and thats where software apps get a bad rap from.
you can also use HWMonitor PRO just to be 100 certain your getting accurate readings especially if it lines up with HWiNFO64 readings....doing these things you be ok without a meter
 
primetime said:
if you were running cpu and gpu near 100% usage for a while and that your numbers, then by all means it appears your power supply is fine. Yes software can report inaccurate numbers but its more rare than "tends to" these days. That and the fact you can also double check readings in your bios from time to time just to confirm there lining up (while unstressed) with what appears to be accurate reading
I for one have a multi meter and have never noticed any inaccurate reading from my board reporting the voltages levels...does it happen yes but its never happened on any board i have owned.
Im my mind there are some app here and there that don't report as accurate as others and thats where software apps get a bad rap from.
you can also use HWMonitor PRO just to be 100 certain your getting accurate readings especially if it lines up with HWiNFO64 readings....doing these things you be ok without a meter
Click to expand...

As I have literally hunderds of PCs going though my hands, I'm going to disagree with you. The norm is the software readings ARE inaccurate. Asus seems to be better then most, but I love the Gigabytes that report -3v on the 12v rail and 1.4 on the 5v.

That being said, they still tend to show fluctuations, and the OPs voltages (Even if possibly inaccurate) seem stable.
 
Hmm... so my MB is probably the culprit now? That's gonna suck because I really don't want to reinstall my computer :|
 
SlvrDragon50 said:
Hmm... so my MB is probably the culprit now? That's gonna suck because I really don't want to reinstall my computer :|
Click to expand...

Do a proper cmos reset buy unplugging the pc, take the battery out, put the jumper on clear.....and push the on button on the case for a few minuets to drain any caps....pull it all back and redo all the cmos settings and report back if still having the same issues.
My bios can get glitched from time to time...mainly do to power failures and or other reasons....only takes a few minuets to see if it fixes the problem:D

Also it would be helpful if you listed all your components (mb,cpu,ram ect.)

just for shits and giggles i figured i run a few test on mine....rails seem pretty stable on mine as well
Capture_zpsc4fa74da.png
[/URL][/IMG]
 
Last edited:
I'd actually start looking at peripherals. Hook up only one monitor see if tis good, then only the other. Only a mouse, then only the keyboard. Unhook the USB speakers, etc etc etc.
 
Well, it usually boots fine with one of the monitors unplugged. How is a monitor causing this problem though?

The boot error I get is that the monitor doesn't support the resolution being outputted by my video card, but I usually don't have this issue when I wake the computer from sleep.
 
SlvrDragon50 said:
Well, it usually boots fine with one of the monitors unplugged. How is a monitor causing this problem though?

The boot error I get is that the monitor doesn't support the resolution being outputted by my video card, but I usually don't have this issue when I wake the computer from sleep.
Click to expand...

well if you have multiple cards and multiple displays then of course that card without a display running is drawing near zero power that is until you power on the display hooked to it...so you tell me is this the case, because that would possibly point to an overloaded or dying power supply if that were the case. modern cards will draw close to zero power until it detects a powered on display.....just saying

now in regards to software reporting inaccurate readings, of course its possible but on mine it actually reports very close to what it is in actual as i can show here, but like was mentioned asus is know to be better at getting these readings correct vs some of the others as seen here fully loaded gpu and cpu.

12volt_zps3b180ae2.jpg
[/URL][/IMG]

5volt_zps96392880.jpg
[/URL][/IMG]

the reason it shows higher than reported is because the motherboard is measuring circuits that are currently have a load on then which is almost impossible to check with a multi meter since the wire you hook to have zero power draw on then and don't have any voltage drop because of it...but it does point out that the power supply is most likely working as it should anyway
 
Okay. I might pick up an EVGA Supernova 650W G1. Hopefully it's not as noisy as the older generation or as the NewEgg reviews are saying. Seems like there might be some QC issues, but I'm just not looking to spend 80-90 bucks on a PSU.
 
ran boinc seti all night since my room was a bit nippy and this app is very good at running max cpu and gpu for hours on end. Of course we cant test for ripple as mentioned earlier but my rails are doing better than even i expected with a thermal take power supply (im not a fan of thermal take but i needed a replacement asap one day when my power bird hard ocp gold award supply caught fire one day.)

all in all im impressed with the rails holding steady and of course it hasn't caught fire either lol
Capture_zpsa573bec6.png
[/URL][/IMG] the most noteworthy numbers are actual voltage drops ( as in very little)
 
You must log in or register to reply here.
Back
Top