CM Extreme2 475 PSU failing?

[HalfAHertz]

The last 2 or so weeks, I've experienced some strange pc behavior.

When I turn on the pc, the lights and fans turn on, then after 2-3 seconds everything turns off and after about 2 seconds starts back up and the PC boots up as normal.
The second strange thing is that sometimes, when waking from sleep, the pc immediately restarts.
Everything else runs normally without any hick ups.
I am starting to suspect the PSU because it is a pretty low-end model. Any ideas how to test it?

The below is a reading of the rails with hwmon after some gaming:

The cpu has been running undervolted since I built the system a couple of years ago.

 

[EarthDog]

With a digital multi meter to check and see if the rails are still in order...

 

[thebluebumblebee]

Your PSU is the "weakest link" in your system, IMHO, but I am more inclined to think that the problem is coming from the keyboard, USB connections, or a BIOS corruption.

 

[Bill_Bright]

You cannot properly test a PSU with a multimeter. For one, the PSU must be under a variety of proper loads. For another, most meters don't test for ripple or other anomalies that affect stability. Therefore, to conclusively test a PSU, it must be done using an oscilloscope under a variety of expected loads and/or with a power supply analyzer.

So the best way for most users is to simply swap in a known good spare PSU and see if the problem continues.

Note according to the Power Supply Design Guide for Desktop Platform Form Factors, Revision 1.31, April 2013, Table 5 on page 14, your +5VDC voltages are out of tolerance with the maximum allowed of 5.25V.

 

[EarthDog]

You are absolutely right Bill, however, A consumer can only do so much. REVIEWERS have load testers and scopes, not an average consumer.

TO me, it is the first step to check the voltages at both idle then on load if possible to make sure it is within spec (5% is ATX spec on all rails). You check with a DMM because software is notoriously off. NEVER take software readings as The Gospel! So he may be out of spec on that rail, but a DMM will be needed to confirm it because of software inconsistency. NEVER rely only on software readings!



But I agree with swapping it out with a known good to confirm!

 

[Bill_Bright]

You are absolutely right Bill, however, A consumer can only do so much. REVIEWERS have load testers and scopes, not an average consumer.

Agreed. But the average consumer doesn't have a multimeter either. And even if they did, they may not know how to preset it, where to safely put the sharp, high-conductive probes, interpret the readings or even know if the readings are correct and within required tolerances. And again, most multimeters don't measure ripple either.

However, swapping out a PSU is actually easy as far as technical tasks go - especially if you pay attention to the cables as you disconnect the old ones. And most users have access to another PSU via another computer in the house and/or through a friend, neighbor or relative. And then there's no guesswork. The 2nd PSU either fixes the problem, or it doesn't.

If you have a good multi-meter and you know how to use it, then fine. But if you don't know how to use it, I don't recommend jamming sharp, highly conductive meter probes into live electronics.

I would much rather users buy a good PSU Tester and use that instead of a multimeter unless truly qualified to use a multimeter. One like this one that has a LCD readout is preferred over the cheaper models that just use an LED light as these can help tell you if the rail is within tolerance. That said, these don't put a full load on the supplies and they don't test for ripple either.

There's no shame in taking the PSU to a shop either as most shops will test a PSU for a nominal fee, or even free.

 

[EarthDog]

More would have a DMM than a scope!!!!

A simple google link handles that how to for a DMM...but I hear ya.

Swap is easiest, then snag a $20 DMM and that PSU tester to check things.

Cheers.

 

[Bill_Bright]

More would have a DMM than a scope!!!!

No doubt. That's why "conclusive" testing by "normal" users is not possible, except by swapping out with a known good PSU.

 

[95Viper]

One of my systems was doing the same as the OP's.

PSU tested fine.

So, I removed all the components and cleaned the connections; then, re-installed everything, as, doing a rebuild... this was two months ago.
Figured it must have been bad connection or something related.
But, I have had no problems since.

 

[HalfAHertz]

Thanks again for all the great suggestion guys.

I'll try disconnecting all the peripherals tonight( mouse,speakers,camera, keyboard, and 1 usb drive) and testing if that helps.
Later during the weekend I'll dismantle everything and clean the connectors. If that doesn't help, I'll try to borrow a PSU from my friend or from work.

 

[HalfAHertz]

Hi,
I tried all the great recommendations in this thread - disconnected everything, then tried starting the pc with just the ssd and one stick of ram with no success.
I borrowed a seasonic psu from a friend and tested the pc with that and got the exact same results.

My next culprit is the motherboard. I took it out, cleaned it thoroughly and inspected it for any damage. There isn't anything that stands out. I added a thick sheet of cardboard paper between it and the pc case just in case. However the problem persists. I think I'll look for a new mobo next week.
Here's a video of the problem:

I think I might have made things worse now because sometimes it takes two or even three "cycles" before the pc boots up normally.

 

[Caring1]

Try disconnecting the SSD drive and see if it reacts the same (minus booting into Windows)

 

[HalfAHertz]

Same thing keeps happening even with the ssd completely detached.( both power and sata detached)

 

[95Viper]

Have you tried replacing the CMOS battery?
And, tried starting up with the reset and power buttons disconnected?

 

[HalfAHertz]

An update on this. I tried removing all the buttons and hdd led/power led plugs and shorting the mobo unsuccessfully. Then I connected just the reset button to the power input and got the same result.
In the end I got a 2nd hand Gigabyte motherboard because nobody seems to be selling new 1150 boards anymore and things seem to be back to normal for now.
Once again thank you all for your input on the matter.