Is it dying? Confusing behaviour.

First, yeah, I know, dont cheap out on a PSU... but a cheap PSU at the moment is better than no PSU. It is a Topower TOP-650PM... not really stellar reviews.

Anyways.

I keep getting restarts, then a warning from the motherboard at boot saying the system was shutdown to prevent damage from a power surge. Then occasionally the system totally shuts down cold, and will not restart unless I unplug the PSU from the motherboard for a sec then plug it back in.

Im just wondering if this is the PSU or motherboard? This is my second motherboard like this one, the first died, it may have been the PSU that killed it, as that PSU died at the same time. But this is a different PSU, new mobo... only thing in common is the motherboard make and my surge protector. I have moved since these problems started, so it is even a different house.


I am using a good surge protector from the wall, and I have another PSU coming, but in the mean time, any thoughts?

is the motherboard mounted well in the case? or does it have any signs of a short circuit like bad caps or loose componets. and are you using the same ram as the old setup?

I mounted the mobo the same as I always have... not sure how you would mount it wrong, it is not loose.

It is the same ram. Sadly, I dont have access to any other DDR3 to try that.

I will have to take a detailed look for loose capacitors, I dont see any evidence of short circuits.

just saying you could run a memtest on your current ram to see if you get any errors. bad ram can cause the same problems. If you dont care about the warranty on your current psu i would crack that open and look for any potential shorts/problems with it.

and a loose motherboard can cause all kinds of funky stuff thats what i asked.

I will run Memtest86+

The PSU isnt worth the shipping back for a warranty :/ I probably will crack it open to have a look.

Im pretty confident I mounted the motherboard nice and tight.


Thinking about it, this behaviour comes and goes too, it will be fine for a week or two, then starts doing it again.

I let Memtest86+ run through 4 times, no problems.

Could this issue be from the PSU being over taxed? I did have another HDD hooked up that I was testing, and so I pulled it. I havent had a restart since, but that could just the PSU going into a stable phase.

I dont think my set-up would push a 650w psu...

You're assuming it really is a 650w PSU. Arguably the largest issue with cheap units is that some outfits take a 200w PSU and slap a 650w label on it.

Run a load and check with a multimeter. Black wires are ground.

Yellow should have 11.5 to 12.5v.
Red should have 4.75 to 5.25v.
Orange should have 3.15 to 3.45v.

yeah bobs right 650 is a stretch for alot of companies. max power isnt the same as actual power but there allowed to label there stuff that way.

What i think is crazy is the fact that you dont have that many components connected to the psu and your having problems. Must be very cheap power supply unit.

Actually I know the brand and they make some respectable models, actually there's one in the recommended psu list/sticky. But it does sound like a psu issue though.
Have you tried running single channel ram?

Bobnova said:

You're assuming it really is a 650w PSU. Arguably the largest issue with cheap units is that some outfits take a 200w PSU and slap a 650w label on it.

Run a load and check with a multimeter. Black wires are ground.

Yellow should have 11.5 to 12.5v.
Red should have 4.75 to 5.25v.
Orange should have 3.15 to 3.45v.

Click to expand...

Yeah, the thought that it might not actually be a 650w crossed my mind.

My multimeter is pretty basic, and analogue gauge, and I never really learned to use it properly, but...


Yellow = looks a smidge over the 10v mark (in the 250v setting), it is off the gauge in the 10v setting.
Red = 5.25v
Orange = 3.25v



I will pull a stick of ram tonight if I get the chance.

Since I pulled the extra HDD, I may have had 1 crash, but im not 100% sure on it as I came back, my computer was off, restarted, just asked if it wanted to boot windows normally.

No crash today...

Oroka Sempai said:

No crash today...

Click to expand...

Using single channel ram?

Okay, so I got a new case (Cooler Master Elite 430 Black), moved all my gear to my new case, crashes stopped, so I was thinking that maybe the PSU wasnt grounded properly for some reason.

Now, I have been burning through SSDs and HDDs like crazy... and Im starting to think my PSU has something to do with it. I had a OCZ Vertex 2 80GB as my OS drive, and a 1TB seagate barracuda for storage.

The Vertex 2 was garbage... worst reviewed item I have ever seen, but I bought it new locally and didnt read the reviews first... I regretted it. It has died and was RMA'd 3 times, the last time they sent me an Agility 3 90GB, and 2 days ago it died too.

My barracuda first died in December, I RMA'd it in Janurary, got a referb back, referb died a month later. I just got another referb back today, and I am scared to death Im going to kill it... I still have to send the Agility 3 back :/

With having problems in the past with my PSU, I still feel it may be an issue, but with the crap OCZ was hawking, and the HDD replacement being a referb, I have a small thought that it might be a bad run on drives too.

Any thoughts? Normally I would just buy a new PSU, but im running low on funds at the moment and I really dont want to risk frying another drive.

Hmmmm, I've never had any issues with my 3.5" Vertex II. Works perfectly. With as many failures as you've had I'm surprised you didn't look at replacing the PSU earlier than this. Silverstone make some excellent mid priced units as do Corsair. The Antec Earthwatts range is pretty decent too.

I bought the Vertex 2 from a guy who bought 2, both died, so he sold one of the replacements to me, the replacement he kept also died. If you do a search on customer reviews on newegg for the Vertex 2 80GB (the one I had), of 10 reviews, 8 were dead drives, of those 8, 4 were multiple failures (replacements died too).

Seagate... I read the original Barricudas were not so reliable, and when I first got it, it was a bit buggy but seemed to have sorted itself out. It died, I got a 'repaired' drive back, it died... For years I was a die hard Western Digital user, every WD drive I ever bought still works, where ever they ended up (given to family). I get Seagate, problems from the get go...


I dont like the idea of replacing the PSU just cause maybe perhaps it is the problem... might be the flux capacitor too if I am guessing. I also moved my RAM to the first 2 slots rather than slot 1 and 3 when I changed cases... since I changed the case I havent gotten any random shut downs or surge protection warnings from my mother board.


Shouldnt there be a way to truly test the PSU to be sure one way or another? I just hate swapping parts just on random guesses, especially when I have to shell out money for it.

There's Yee olde Paper clip test. Beyond that you need a full on load tester and oscilloscope.

Monitor the rails with a DMM.

Do so:
-during boot
-at idle
-at full cpu load
-at full gpu load
-at full gpu and cpu load.

See how much the rails drop and how stable they are. Minor fluctuation is okay, but anything over 10% means that it's time for a new PS.

EDIT: I see you've already tried a multimeter with no luck reading the 12v line accurately. Do you have a friend close by that has a DMM that you can borrow?

corruption said:

Monitor the rails with a DMM.

Do so:
-during boot
-at idle
-at full cpu load
-at full gpu load
-at full gpu and cpu load.

See how much the rails drop and how stable they are. Minor fluctuation is okay, but anything over 10% means that it's time for a new PS.

Click to expand...

I just tried while idling, and I really dont know how to use multimeters properly, just the basic concept, I will have a more in depth look at it tomorrow, or shortly. I fracking hate taking shots in the dark trying to find problems, especially when it might just be a bad run on drives.

::EDIT::
What is a good program to achieve full cpu, gpu, and both... SuperPi for the CPU I would think. It has been a few years since I stress tested anything.

Okay, here are my results using BurnINTest V7 64bit


Yellow idle=12.13 CPU=12.07 GPU=12.09 BOTH=12.06
Red idle=5.14 CPU=5.15 GPU=5.14 BOTH=5.15
Orange idle=3.34 CPU=3.38 GPU=3.39 BOTH=3.40


Now, I only ran these tests for about 90 seconds, but when they settled in to a voltage, they stayed there, very minor fluctuations (+/-.01v) if any at all.

Looks good to me.