My system keeps shutting off when idle - which component is failing?

[moeburn]

OS: Windows 8.1

CPU: AMD Athlon II X4 620 2.6ghz quad core

RAM: 4GB (2x2GB) DDR3

System drive: Crucial M4 128GB SSD

Motherboard: 2nd-hand Acer FRS780M OEM board

GPU: XFX Radeon HD 5770 1GB DDR5

PSU: Thermaltake TR2 500W (don't worry, it's not the infamous terrible Thermaltake model)

 

This has started happening more and more often.  It used to be once a month, now it's more like once a week.  I typically leave my computer on 24/7 (I would let it go to sleep when idle but for some reason it won't, something is resetting the idle timer), but sometimes I will come in and find my computer turned off - when I boot it up, there are no "your system has recovered from a serious error" messages, although I don't know if Windows 8 even does that.  When I start up my browser it says it wasn't closed properly and asks me if I want to restore my open tabs.  It almost never happens while I am using the computer, I think it only ever happened once while I was using it.  I always come to find it turned off as though there was a power failure in the house - but there wasn't, otherwise my clocks would be reset.  It just happened again a few minutes ago, but this time, my desktop icon layout was scrambled, like what happens when the resolution changes back and forth.  Now I have to go back and rearrange them all.

 

My system is pretty old and crappy, I know.  I haven't upgraded it because it still lets me play almost all the latest AAA games at med-high settings at 1080p at 30-60fps.  But I really don't think this is the power supply.  The PSU is less than a year old, I bought it recently because my last one failed.  But I'm guessing that something is starting to fail.  It probably doesn't help that this computer runs on the floor of a sometimes-damp, very dusty basement.  I do vacuum it very often though, I never let the fans collect much dust.  But my point is that something could be corroding, maybe on the motherboard.  Does anyone know what I should be looking for, or what component other than the PSU failing that could cause this to happen?  Or at least what I can rule out?  Burst capacitors on the motherboard, or corroded traces?  The GPU?  The CPU itself failing?  Should I do a RAM scan?  My SSD?

[conna]

Turn off the sleep modes and hibernation. Don't let it go to low power mode and see if you problem stops.  

[Circaflex]

Check the motherboard for blown or extended caps

 

disable ulps (amds power saving mode for graphics cards)

 

 

I had a similar issue where it seemed when my system (server) was idling it would freeze or randomly reboot or shut down. I eliminated heat as the culprit and in the end it was my psu. Any chance a friend or family member has an old psu you can at least try?

[techbeck]

Processor failing could cause this.  Or a fan not working properly.  Sounds like it is over heating and shutting down.

[Circaflex]

  On 14/07/2014 at 20:13, techbeck said:

Processor failing could cause this.  Or a fan not working properly.  Sounds like it is over heating and shutting down.

i thought heat as well, but he states it rarely happens when he is using the pc/playing games which would generate more heat than idle. It wouldnt hurt to double check your thermal paste and heatsink though

[techbeck]

  On 14/07/2014 at 20:15, Circaflex said:

i thought heat as well, but he states it rarely happens when he is using the pc/playing games which would generate more heat than idle. It wouldnt hurt to double check your thermal paste and heatsink though

 

Still could be failing processor.  Or memory.  I doubt it is software related since that normally results in a BSOD or recovery message.  Probably hardware most likely since the system just shuts off.  I would start disconnecting any external components, and any internal you can live without, one by one to see if you can narrow it down to hardware.

[Circaflex]

  On 14/07/2014 at 20:19, techbeck said:

Still could be failing processor.  Or memory.  I doubt it is software related since that normally results in a BSOD or recovery message.  Probably hardware most likely since the system just shuts off.  I would start disconnecting any external components, and any internal you can live without, one by one to see if you can narrow it down to hardware.

 

I have to agree, I would start testing all components individually:

 

Run memtest overnight

Test your Hard Drives via MFG diagnostics

PC Check is pretty decent for testing CPU

 

 

check your board for the blown caps though or extended caps

[moeburn]

  On 14/07/2014 at 20:09, Circaflex said:

Check the motherboard for blown or extended caps

 

disable ulps (amds power saving mode for graphics cards)

 

 

I had a similar issue where it seemed when my system (server) was idling it would freeze or randomly reboot or shut down. I eliminated heat as the culprit and in the end it was my psu. Any chance a friend or family member has an old psu you can at least try?

 

I've never heard of ULPS - disabling it wont shorten the life of my card, will it?  it's already 5 years old :p

 

  On 14/07/2014 at 20:13, techbeck said:

Processor failing could cause this.  Or a fan not working properly.  Sounds like it is over heating and shutting down.

 

Definitely not a heat issue, unless none of my thermometers are registering properly.  I run OpenHardwareMonitor and graph all my temps with min and max:

IDLE: CPU - 37?C, GPU - 47?C

MAX LOAD: CPU - 47?C, GPU - 86?C

 

I'm pretty sure those are well below tolerable limits. But the CPU is kinda old, it could be failing otherwise.

 

  On 14/07/2014 at 20:39, Circaflex said:

I have to agree, I would start testing all components individually:

 

Run memtest overnight

Test your Hard Drives via MFG diagnostics

PC Check is pretty decent for testing CPU

 

 

check your board for the blown caps though or extended caps

 

Allright I'll run some mem tests, can't see any obvious blown caps but the bulging ones can be hard to identify sometimes.  

 

I'm also running AMD Cool-n-Quiet which automatically throttles the CPU clock down to 700mhz and in between depending on demand, I guess I could try turning that off too.  The trouble is that this happens so infrequently that eliminating components is gonna take a while...

[Circaflex]

nope disabling ULPS wont shorten the lifespan.

 

Cool n Quiet could also be causing this, maybe try turning it off for a week.

 

 

Does anything pop out within event viewer after these crashes?

[badb0y]

Random shutdown are also caused by RAM. Check it as well.

[farmeunit]

Even if you don't "think" it's the PSU, I would check it.  What is the brand?  That one area it's not a good idea to go cheap.

 

ANY part can fail within a year, so don't rule it out.  If it's completely off, then I'd say PSU.  Memory or other things will cause BSoD and typically loop unless you set it to NOT reboot, then it'll stay on the BSoD screen.  Same with HD errors.

[techbeck]

  On 14/07/2014 at 20:46, moeburn said:

I've never heard of ULPS - disabling it wont shorten the life of my card, will it?  it's already 5 years old :p

 

 

Definitely not a heat issue, unless none of my thermometers are registering properly.  I run OpenHardwareMonitor and graph all my temps with min and max:

IDLE: CPU - 37?C, GPU - 47?C

MAX LOAD: CPU - 47?C, GPU - 86?C

 

I'm pretty sure those are well below tolerable limits. But the CPU is kinda old, it could be failing otherwise.

 

 

Allright I'll run some mem tests, can't see any obvious blown caps but the bulging ones can be hard to identify sometimes.  

 

I'm also running AMD Cool-n-Quiet which automatically throttles the CPU clock down to 700mhz and in between depending on demand, I guess I could try turning that off too.  The trouble is that this happens so infrequently that eliminating components is gonna take a while...

 

Then I would lean more towards something else over heater or just failing.   Would look at the processor or mobo.  Also, disconnect components internal and external to see if you can narrow it down that way.  And did you run a memory check?  How often does this happen in a day?

[techbeck]

  On 14/07/2014 at 21:01, farmeunit said:

 

ANY part can fail within a year, so don't rule it out.  If it's completely off, then I'd say PSU.  Memory or other things will cause BSoD and typically loop unless you set it to NOT reboot, then it'll stay on the BSoD screen.  Same with HD errors.

 

Not necessarily.  I have seen PCs refuse to boot, or just shutoff, with bad sticks of memory.  I have even run in to laptops that would not boot if there was an external USB drive plugged in.  Chances are it probably isnt memory, but it is the process of elimination at this point.  Dont rule anything out but I am leaning more towards CPU/mobo personally.  Ran in to a few laptops that would turn on for a few minutes...or couple hours...and just turn off.  Ended up being one of the two.

[T3X4S]

If the PSU says Thermaltake on it -- its definately suspect.

1.) Check error logs
2.) memtest
3.) replace PSU, even if its not the problem - it needs to be replaced.  Besides you can always use it on your next computer (BTW - you need a new computer)

[Circaflex]

Only because I have seen a few of OPs posts, flat out telling him to replace something isnt always the easiest, money doesnt grow on trees for everyone. What might be a cheap part replacement for you, might not be for the next.

 

 

Do you have any retail stores near you that you can purchase a temporary PSU just to rule it out? If it is the PSU keep it if possible, if it isnt return it for a full refund?

[techbeck]

  On 15/07/2014 at 00:00, Circaflex said:

Only because I have seen a few of OPs posts, flat out telling him to replace something isnt always the easiest, money doesnt grow on trees for everyone. What might be a cheap part replacement for you, might not be for the next.

 

 

Do you have any retail stores near you that you can purchase a temporary PSU just to rule it out? If it is the PSU keep it if possible, if it isnt return it for a full refund?

 

Can always test the PSU with a multimeter as well.  But not everyone has those laying around. :)

[moeburn]

  On 14/07/2014 at 22:59, T3X4S said:

If the PSU says Thermaltake on it -- its definately suspect.

1.) Check error logs

2.) memtest

3.) replace PSU, even if its not the problem - it needs to be replaced.  Besides you can always use it on your next computer (BTW - you need a new computer)

 

I know Thermaltake god a bad reputation from a particular manufacturer's model that was pretty much guaranteed to fail, but not all of them are bad.  The Thermaltake's made by Channel Well are quite good.  That being said, I have no idea if mine is made by Channel Well.  According to this article, my Thermaltake is "E303666 - OEM Thermaltake - Quite good usually".  The fact is that, in the budget range, a power supply's brand has almost nothing to do with their quality, because most brands mix and match manufacturers.

 

  On 15/07/2014 at 00:15, techbeck said:

Can always test the PSU with a multimeter as well.  But not everyone has those laying around. :)

 

I have a multimeter that can measure up to 10A and a whole lotta volts, and I know how to use it, although I would know how to test a PSU or what I would be looking for.  I do know that according to my OpenHardwareMonitor, I get some pretty big voltage changes during load.  But I don't have a 12v rail measurement on the motherboard, and I hear that's the important one.

[binaryzero]

It's an old system, so replace the PSU anyway. You'll probably find this is the problem if everything in software is configured correctly.

[farmeunit]

  On 15/07/2014 at 00:00, Circaflex said:

Only because I have seen a few of OPs posts, flat out telling him to replace something isnt always the easiest, money doesnt grow on trees for everyone. What might be a cheap part replacement for you, might not be for the next.

 

 

Do you have any retail stores near you that you can purchase a temporary PSU just to rule it out? If it is the PSU keep it if possible, if it isnt return it for a full refund?

That's understandable, but sometimes that is the only way to test things. I ALWAYS have a PSU laying around somewhere, just for this purpose. Of course I'm sort of a pack-rat as well and I don't throw anything away.

Even if a PSU tests good, they'll still cause random issues. I had a computer that would run fine usually, but it would blue-screen every day at some point. In the logs it would show a graphics driver crash every time for the onboard video. Changed the PSU and never had that issue again. That is the ONLY thing I changed.

I had another that when you turned the computer on, all fans would spin and lights would come on, but it wouldn't POST. Changed out the PSU and everything worked as intended.

The memory tests and such won't cost him anything, so he can feel free to run those. If they don't show anything, it's either replace something to fix it, or don't and he'll have the same issue. No harm no foul. I don't see the problem with suggestions, especially since he said he didn't think the PSU was bad. Just because it's only a year old doesn't mean it's not functioning correctly. That's all anyone is saying. Ultimately, it's up to him to either do it, or not. The OP might even be able to borrow one temporarily, just to test.

Don't always count on a multimeter. You could have voltage spikes or drops at random times. In my example above, the PSU tested just fine, but still caused issues.

[moeburn]

So I looked in my event log, looking around the time of my OP, because I knew I had started this thread just minutes after a failure.  The last message was 4 hours before the shut down, and it was an unrelated information message.  Here's what it says after the shutdown:

 

Kernel-General - Information -  7/12/2014 2:43:22 AM - The operating system started at system time ?2014?-?07?-?12T06:43:22.499163600Z.

Kernel-Boot - Information -  7/12/2014 2:43:22 AM - The last shutdown's success status was true. The last boot's success status was true.

The boot type was 0x0.

There are 0x1 boot options on this system.

The bootmgr spent 0 ms waiting for user input.

EventLog - ERROR - 7/12/2014 2:43:33 AM - The previous system shutdown at 2:38:27 AM on ?7/?12/?2014 was unexpected.

 

So, because it knows the exact time of the shutdown, does that mean that it was a "soft" shutdown?  Or would this be the same message that would be displayed in the event of a power failure?

[T3X4S]

  On 15/07/2014 at 00:00, Circaflex said:

Only because I have seen a few of OPs posts, flat out telling him to replace something isnt always the easiest, money doesnt grow on trees for everyone. What might be a cheap part replacement for you, might not be for the next.

 

 

Do you have any retail stores near you that you can purchase a temporary PSU just to rule it out? If it is the PSU keep it if possible, if it isnt return it for a full refund?

Yeah my bad.  I get used to telling people to just replace stuff and stop wasting time wondering what the problem is - when I get online I never think someone might consider these things are major purchases.  Very close-minded of me.

[T3X4S]

  On 15/07/2014 at 01:49, moeburn said:

So I looked in my event log, looking around the time of my OP, because I knew I had started this thread just minutes after a failure.  The last message was 4 hours before the shut down, and it was an unrelated information message.  Here's what it says after the shutdown:

 

Kernel-General - Information -  7/12/2014 2:43:22 AM - The operating system started at system time ?2014?-?07?-?12T06:43:22.499163600Z.

Kernel-Boot - Information -  7/12/2014 2:43:22 AM - The last shutdown's success status was true. The last boot's success status was true.

The boot type was 0x0.

There are 0x1 boot options on this system.

The bootmgr spent 0 ms waiting for user input.

EventLog - ERROR - 7/12/2014 2:43:33 AM - The previous system shutdown at 2:38:27 AM on ?7/?12/?2014 was unexpected.

 

So, because it knows the exact time of the shutdown, does that mean that it was a "soft" shutdown?  Or would this be the same message that would be displayed in the event of a power failure?

You would get this message if it was a sudden loss of power as well.  The reason it has an exact time is because it is constantly posting about every 2 seconds, so when it doesnt get the little blip from the system, it assumes there was a shut down.

I have also noticed there are a lot of manufacturers who make PSU for other manufacturers/brands.  For the longest time I swore by PC Power & Cooling - anything from them was the best there was.  Then I find out their cheaper Silencer line was made by Seasonic.  Seasonic also makes ones for antec. (not that Seasonic is bad, but I was shocked to hear that the company known throughout the industry for making the best PSUs on the market, weren't really making some of the stuff in the first place.  The Turbo Cool line was always done in house initially, not sure after they were purchased by OCZ

You are right, @ the market segment you are considering - there it not much assurance when relying on brand name.  I am always telling people brand name is just as important as wattage claims. :blush: