Win7 home-built HTPC shuts down without an error message, why?
February 22, 2012 10:46 AM Subscribe
My home-made HTPC running Windows 7 shuts down without warning, first after prolonged use playing HD (720) videos with VLC. Then it stopped twenty minutes into an episode, and switching to Media Player Classic seemed to be the answer. Then the computer shut down after about 20 minutes of simply being idle. And there's never a "sudden shutdown" error message when I reboot. How do I diagnose this, or what could be wrong?
The computer is built with new computer parts in an old tuner case, with USB ports and a power button cut into the front panel, and a PC rear panel, including exhaust fan, in the rear. This is run from a wooden entertainment center with low overhead clearance and a solid back, but ample space on the sides of the case for air circulation.
Still, I thought temperature was an issues, so I thought additional airflow could help. I rotated the case ~90 degrees, with the vent fan closer to the front of the shelf, but the computer still shut down while running.
Every time the computer turns off and we reboot it, it starts back up as if there was never any problem. No option to start into safe mode, no message in Windows that the computer unexpectedly shut down.
What's the best way to gauge the temperature of a Win7 system, and check for the maximum temperature safety level? (I'm sure there's a better term for this, but the words elude me).
Or am I looking the wrong way, and should I be running malware and virii checks? We don't use this computer to browse the internet much, and the only things we have downloaded are freeware apps from the developers' websites (specifically VLC, Media Player Classic, and InfranView, plus Firefox and Chrome), so up to now I've assumed it's a temperature trigger of some sort.
I'm writing this question without direct access to the HTPC, but I'll post the full specs later today.
Thanks!
The computer is built with new computer parts in an old tuner case, with USB ports and a power button cut into the front panel, and a PC rear panel, including exhaust fan, in the rear. This is run from a wooden entertainment center with low overhead clearance and a solid back, but ample space on the sides of the case for air circulation.
Still, I thought temperature was an issues, so I thought additional airflow could help. I rotated the case ~90 degrees, with the vent fan closer to the front of the shelf, but the computer still shut down while running.
Every time the computer turns off and we reboot it, it starts back up as if there was never any problem. No option to start into safe mode, no message in Windows that the computer unexpectedly shut down.
What's the best way to gauge the temperature of a Win7 system, and check for the maximum temperature safety level? (I'm sure there's a better term for this, but the words elude me).
Or am I looking the wrong way, and should I be running malware and virii checks? We don't use this computer to browse the internet much, and the only things we have downloaded are freeware apps from the developers' websites (specifically VLC, Media Player Classic, and InfranView, plus Firefox and Chrome), so up to now I've assumed it's a temperature trigger of some sort.
I'm writing this question without direct access to the HTPC, but I'll post the full specs later today.
Thanks!
I stopped in to question the power supply... how does the rating of your power supply compare to the parts you have plugged in?
posted by machinecraig at 10:56 AM on February 22, 2012
posted by machinecraig at 10:56 AM on February 22, 2012
Looks a lot like a bad PS to me too. First thing I'd check.
posted by bonehead at 11:10 AM on February 22, 2012
posted by bonehead at 11:10 AM on February 22, 2012
Also: Speedfan or RealTemp or similar. Watch your temperature over time.
posted by bonehead at 11:13 AM on February 22, 2012
posted by bonehead at 11:13 AM on February 22, 2012
Response by poster: Would a faulty power supply result in the computer turning off without an error message being logged anywhere? That's what is confusing me, to be honest. I'm used to computer crashes being followed by rebooting to reminders that something went wrong, and the possibility for me to diagnose issues by digging through log files.
posted by filthy light thief at 11:15 AM on February 22, 2012
posted by filthy light thief at 11:15 AM on February 22, 2012
Yes, a power supply shutdown won't necessarily cause a windows fault. It's just like a power outage.. as long as it didn't have something engaged in memory running that was in the middle of a write or some other I/O action, it wouldn't register as a 'crash' to Windows.
Run the usual malware, etc checks, but I'm going with the prevailing winds on the power supply. It's happened to me in the past and it took me forever to figure it out, as power supply wasn't even on my radar as a potential issue then.
posted by rich at 11:27 AM on February 22, 2012
Run the usual malware, etc checks, but I'm going with the prevailing winds on the power supply. It's happened to me in the past and it took me forever to figure it out, as power supply wasn't even on my radar as a potential issue then.
posted by rich at 11:27 AM on February 22, 2012
Would a faulty power supply result in the computer turning off without an error message being logged anywhere?
Yes. Error messages are close to useless when it's a hardware issue, IMHO.
Seconding heat or power. Use your heat sensors to confirm that it's not a temperature issue, and compare your video card's requirements to your power supply.
posted by Stagger Lee at 11:28 AM on February 22, 2012
In an old HTPC I installed speedfan which allowed control/monitoring of the system fans & temperature. I believe it would produce a plot of the internal temperature sensors of most motherboards - it also showed the motherboard voltages if I remember correctly. Might be an easy step to diagnose...
Many BIOS' have an option for automatic shutdown / signal on overheat. If you have the option make sure the audio signal option is set. That might give you a hint...
posted by NoDef at 11:34 AM on February 22, 2012
Many BIOS' have an option for automatic shutdown / signal on overheat. If you have the option make sure the audio signal option is set. That might give you a hint...
posted by NoDef at 11:34 AM on February 22, 2012
Yeah, a heat problem or a faulty power supply are your two primary culprits here. You don't say how old your board is, or what chip you're using, but if you're on an AMD platform, older motherboards had an emergency heat-sensor poweroff in the CPU socket, as AMD chips of that generation would slag themselves if there was a cooling problem.
Modern AMD chips now have a thermal limiter like Intel always has, so they won't blow themselves up anymore, but I know that the sensors postdated AMD's thermal limiters by some time. I'm not sure how long. With one of those sensors, sudden poweroff was the symptom that they'd tripped.
I've also seen people having similar issues with video card heat -- it appears that either the hardware or the video drivers may be able to cause the computer to shut down if the video card gets too hot. I haven't seen this documented anywhere, and it could also be a hot video card driving the power supply into emergency shutdown. (hot cards pull more power).
You can monitor the CPU temperatures with CoreTemp, and your video drivers should already have a method of tracking chip temperatures. If you don't see high temperatures (over 75C on the CPU, over 90C on the video card), then it's probably the power supply. If one of those things is too hot, then it either needs the heat sink remounted (CPU) or possible replacement (video card). If BOTH things are too hot, then you'd look at the power supply again (is it sufficient for the load you're putting on it?) and case airflow.
I really would expect, though, that you'd see something in the System log about 'the last shutdown was unexpected'. You're not seeing anything of that type?
posted by Malor at 11:52 AM on February 22, 2012 [1 favorite]
Modern AMD chips now have a thermal limiter like Intel always has, so they won't blow themselves up anymore, but I know that the sensors postdated AMD's thermal limiters by some time. I'm not sure how long. With one of those sensors, sudden poweroff was the symptom that they'd tripped.
I've also seen people having similar issues with video card heat -- it appears that either the hardware or the video drivers may be able to cause the computer to shut down if the video card gets too hot. I haven't seen this documented anywhere, and it could also be a hot video card driving the power supply into emergency shutdown. (hot cards pull more power).
You can monitor the CPU temperatures with CoreTemp, and your video drivers should already have a method of tracking chip temperatures. If you don't see high temperatures (over 75C on the CPU, over 90C on the video card), then it's probably the power supply. If one of those things is too hot, then it either needs the heat sink remounted (CPU) or possible replacement (video card). If BOTH things are too hot, then you'd look at the power supply again (is it sufficient for the load you're putting on it?) and case airflow.
I really would expect, though, that you'd see something in the System log about 'the last shutdown was unexpected'. You're not seeing anything of that type?
posted by Malor at 11:52 AM on February 22, 2012 [1 favorite]
Just to clarify - the computer is shutting down as if the power was cut, and not doing a normal Windows shutdown, correct? If so, like others have said, the problem is either heat, power supply or bad RAM.
posted by cnc at 12:15 PM on February 22, 2012
posted by cnc at 12:15 PM on February 22, 2012
One of my old PCs used to shut down without warning. I took a look inside and saw that both the fan and heatsink for the graphics card were clogged with dust bunnies. I gently cleaned them out with some compressed air and the problem was solved.
posted by shino-boy at 12:26 PM on February 22, 2012
posted by shino-boy at 12:26 PM on February 22, 2012
I'm actually going to go against the crowd here and suggest that it might not be a cooling or power supply problem, and that it actually might be a video card driver problem, or at worse, an actual mobo problem. I've had both cause issues with a few HTPC builds.
What is your system specs - mobo, video, cpu, etc?
posted by Old'n'Busted at 12:48 PM on February 22, 2012
What is your system specs - mobo, video, cpu, etc?
posted by Old'n'Busted at 12:48 PM on February 22, 2012
Also, as part of your troubleshooting, since you mentioned the case, check to see if you have a hypersensitive power button or if the power button wires are loose.
posted by jabberjaw at 1:30 PM on February 22, 2012
posted by jabberjaw at 1:30 PM on February 22, 2012
You already described here the symptoms of a computer that appears to be failing on thermals: at first it shuts down under high load, and then it shuts down under progressively lighter loads, finally shutting down under seemingly light loads.
When that happens, I open up the computer and vacuum out all the dust, using a light brush to knock it loose.
If it still happens, get a tube of thermal paste (a buck at the local computer hobby shop), open up the computer, and add a little bit under the CPU heat sink. You'll have to remove the heat sink, and reinstall it carefully, and generally be real damn careful while doing it. Heed the advice.
In the past year I've had two computers that needed new thermal paste. I keep my computers a long time, and run them hot (yay SETI) so the paste tends to dry out after a couple years.
posted by intermod at 1:30 PM on February 22, 2012
When that happens, I open up the computer and vacuum out all the dust, using a light brush to knock it loose.
If it still happens, get a tube of thermal paste (a buck at the local computer hobby shop), open up the computer, and add a little bit under the CPU heat sink. You'll have to remove the heat sink, and reinstall it carefully, and generally be real damn careful while doing it. Heed the advice.
In the past year I've had two computers that needed new thermal paste. I keep my computers a long time, and run them hot (yay SETI) so the paste tends to dry out after a couple years.
posted by intermod at 1:30 PM on February 22, 2012
I had random reboots on my win7 pc (repurposed HTPC hardware) for close to a year. Nearly impossible to diagnose for the same reason. Turned out to be a stick of memory was failing, and only when it failed completely did I figure it out, due to the error LEDs on my motherboard. For the record, I thought the integrated video out had failed, since it wasn't outputting anything to my monitor. When I opened it up and looked up the LED code, I was quite surprised that it wasn't the video that had failed.
Another data point is my work PC (also win7), which started freezing randomly a few weeks back. I didn't want to take it to the laptop center because the Dell techs nearly always default to "wipe the harddrive" as a solution. I'm honestly not sure what they get paid for. Anyway, using Speedfan, I found that the video card temp (not CPU) skyrocketed right before it froze. I installed newer video drivers and have not had a lockup since.
So beside using something like Speedfan to check temperatures, you may want to disable "Automatic Restart on System Failure" to see if you can get any sort of BSOD screen to appear. Since the reboots are random and don't seem to result in a "frozen" state, it would likely be difficult to check error LEDs on the motherboard, but that is another source of information to keep in mind. Memory diagnostics that are often built in may or may not be useful.
Either way, I'm with Old'n'Busted - it's not necessarily a power supply or thermal (hardware) problem.
posted by mysterpigg at 1:45 PM on February 22, 2012
Another data point is my work PC (also win7), which started freezing randomly a few weeks back. I didn't want to take it to the laptop center because the Dell techs nearly always default to "wipe the harddrive" as a solution. I'm honestly not sure what they get paid for. Anyway, using Speedfan, I found that the video card temp (not CPU) skyrocketed right before it froze. I installed newer video drivers and have not had a lockup since.
So beside using something like Speedfan to check temperatures, you may want to disable "Automatic Restart on System Failure" to see if you can get any sort of BSOD screen to appear. Since the reboots are random and don't seem to result in a "frozen" state, it would likely be difficult to check error LEDs on the motherboard, but that is another source of information to keep in mind. Memory diagnostics that are often built in may or may not be useful.
Either way, I'm with Old'n'Busted - it's not necessarily a power supply or thermal (hardware) problem.
posted by mysterpigg at 1:45 PM on February 22, 2012
Response by poster: Thanks for all the thoughts and suggestions so far.
cnc: Just to clarify - the computer is shutting down as if the power was cut, and not doing a normal Windows shutdown, correct? If so, like others have said, the problem is either heat, power supply or bad RAM.
Yes, the shutdowns are abrupt, without any warning.
Old'n'Busted: What is your system specs - mobo, video, cpu, etc?
I'm not near my computer at the moment, but I'll post these later in the day.
jabberjaw: Also, as part of your troubleshooting, since you mentioned the case, check to see if you have a hypersensitive power button or if the power button wires are loose.
Good point, I'll check the connections, but the button is a modified doorbell, so it requires a significant push to turn on or off.
posted by filthy light thief at 1:51 PM on February 22, 2012
cnc: Just to clarify - the computer is shutting down as if the power was cut, and not doing a normal Windows shutdown, correct? If so, like others have said, the problem is either heat, power supply or bad RAM.
Yes, the shutdowns are abrupt, without any warning.
Old'n'Busted: What is your system specs - mobo, video, cpu, etc?
I'm not near my computer at the moment, but I'll post these later in the day.
jabberjaw: Also, as part of your troubleshooting, since you mentioned the case, check to see if you have a hypersensitive power button or if the power button wires are loose.
Good point, I'll check the connections, but the button is a modified doorbell, so it requires a significant push to turn on or off.
posted by filthy light thief at 1:51 PM on February 22, 2012
I'm not sure I agree that an improper shut down does not always generate an error. I was under the impression that there was a "running" flag once Windows 7 is booted that doesn't get cleared until a proper shutdown is performed (the bad shutdown is then caught on the next boot).
But that being said:
- What do your system logs actually look like?
- Are you seeing any errors or any sort?
- If not errors, are you seeing a lot of service state changes around the times of the shutdown?
- What about your application logs?
- What are your power settings set to? Have you tried disabling power management on everything under "change advanced power settings?"
Just to rule out malware, run a malwarebytes and tdsskiller scan against the PC. For tdsskiller in particular, go under "change parameters" and check for the tdsfs and unsigned drivers. Create a restore point and remove anything it finds.
For registry and device driver cleanup:
- Download and install ccleaner and run the registry clean 2-3 times until it shows up empty. (create a restore point beforehand to be safe)
- Download and run DeviceRemover and change the view to "SHOW HIDDEN/DETACHED" devices only. Remove anything it finds. (be sure that DeviceRemover is also the last thing you run before a reboot, just to be on the safe side)
For checking temps and faulty hardware:
- CPUid for temp readings
- Memtest86 for testing memory (Hiren's Boot CD might also come in handy)
posted by samsara at 1:54 PM on February 22, 2012
But that being said:
- What do your system logs actually look like?
- Are you seeing any errors or any sort?
- If not errors, are you seeing a lot of service state changes around the times of the shutdown?
- What about your application logs?
- What are your power settings set to? Have you tried disabling power management on everything under "change advanced power settings?"
Just to rule out malware, run a malwarebytes and tdsskiller scan against the PC. For tdsskiller in particular, go under "change parameters" and check for the tdsfs and unsigned drivers. Create a restore point and remove anything it finds.
For registry and device driver cleanup:
- Download and install ccleaner and run the registry clean 2-3 times until it shows up empty. (create a restore point beforehand to be safe)
- Download and run DeviceRemover and change the view to "SHOW HIDDEN/DETACHED" devices only. Remove anything it finds. (be sure that DeviceRemover is also the last thing you run before a reboot, just to be on the safe side)
For checking temps and faulty hardware:
- CPUid for temp readings
- Memtest86 for testing memory (Hiren's Boot CD might also come in handy)
posted by samsara at 1:54 PM on February 22, 2012
Good point, I'll check the connections, but the button is a modified doorbell
Just saw this too...just for simplification, try running the system without this button (or any other non standard pieces) attached to the MB.
posted by samsara at 1:58 PM on February 22, 2012
Just saw this too...just for simplification, try running the system without this button (or any other non standard pieces) attached to the MB.
posted by samsara at 1:58 PM on February 22, 2012
« Older What is this campy horror flick??? | Thief high centered my Jeep on a boulder. What now... Newer »
This thread is closed to new comments.
posted by Confess, Fletch at 10:54 AM on February 22, 2012 [3 favorites]