Random Reboots of PC

[Fairy]

My homebuilt PC has been running fine for approx five years but just
recently has started rebooting itself at random intervals. Sometimes
it will stay up for two weeks without rebooting but other times it
won't even manage to complete the boot up process (following an
unexpected reboot) before it reboots again. Nothing has changed in
terms of hardware or software installed. I've opened the case and
blew it out with a can of air and I've also removed the PSU and done
the same with that. It isn't an overheat problem as far as I can
tell, the CPU temp never reaches above 40 degrees Celsius (I've got a
good few fans in there). I even sat in the BIOS setup on the health
status page monitoring the temps, voltages and fan speed and even
though all values looked fine, it even rebooted itself from there!

In my own mind I've narrowed it down to either the PSU or the
motherboard. I've found an article which explains how to monitor the
voltages coming out of the PSU with a multimeter but if the problem
doesn't happen for up to two weeks, how can I sit there watching a
multimeter for that amount of time? One other thing I read said it
could be the battery backup for the BIOS but the machine is maintaining
the BIOS settings and date/time when powered off for several days so I
don't think it could be that.

I can't really afford to start buying new parts unless I'm sure
that the original is faulty, so is there anything else I can do to help
me narrow down the cause or has anyone else experienced anything
similar?

 

[Tomcat (Tom)]

I pulled my hair out a few months ago with a system doing this until
finally rebuilding 90 percent of my system before the problem went
away.

I'd first suspect memory problems. After that the PSU or HD. Then
perhaps MB. CPU failure is pretty unlikely. Some viruses can also
cause reboots. It just takes a lot of trial and error to figure out.

 

[DaveW]

Basically, you either have to try replacing parts (and I would first try
replacing the PSU, since that's easier than the motherboard), or you can
take the computer into a shop and pay to have it repaired.

 

[Rod Speed]

My homebuilt PC has been running fine for approx five years but just
recently has started rebooting itself at random intervals. Sometimes
it will stay up for two weeks without rebooting but other times it
won't even manage to complete the boot up process (following an
unexpected reboot) before it reboots again. Nothing has changed in
terms of hardware or software installed. I've opened the case and
blew it out with a can of air and I've also removed the PSU and done
the same with that. It isn't an overheat problem as far as I can
tell, the CPU temp never reaches above 40 degrees Celsius (I've got
a good few fans in there). I even sat in the BIOS setup on the health
status page monitoring the temps, voltages and fan speed and even
though all values looked fine, it even rebooted itself from there!

Yeah, pretty conclusive that its not overheating thats the problem.

In my own mind I've narrowed it down to either the PSU or
the motherboard. I've found an article which explains how to
monitor the voltages coming out of the PSU with a multimeter
but if the problem doesn't happen for up to two weeks, how
can I sit there watching a multimeter for that amount of time?

You need a max min multimeter for that, but
they cost more than a new power supply.

One other thing I read said it could be the battery backup for the
BIOS but the machine is maintaining the BIOS settings and date/time
when powered off for several days so I don't think it could be that.

Yeah, that wont usually cause a reboot. Cheap to try a new battery tho.

I can't really afford to start buying new parts unless I'm sure that the original
is faulty, so is there anything else I can do to help me narrow down the cause

Yes, check the XP logs to see if there is anything
useful there on what is producing the reboot.

Check for bad caps on the motherboard. Those are the usually blue or
black plastic covered post like things that stick up vertically from the
motherboard. The tops should be flat. If any of them have bulged or
have leaked, thats a bad cap and is likely whats producing the reboots.

or has anyone else experienced anything similar?

Yep, plenty have, with a variety of causes. Its more often the PSU
than anything else, but it can certainly be a bad motherboard.

Can even be a faulty reset switch but thats very
rare, they dont usually conduct intermittently.

 

[Paul]

My homebuilt PC has been running fine for approx five years but just
recently has started rebooting itself at random intervals. Sometimes
it will stay up for two weeks without rebooting but other times it
won't even manage to complete the boot up process (following an
unexpected reboot) before it reboots again. Nothing has changed in
terms of hardware or software installed. I've opened the case and
blew it out with a can of air and I've also removed the PSU and done
the same with that. It isn't an overheat problem as far as I can
tell, the CPU temp never reaches above 40 degrees Celsius (I've got a
good few fans in there). I even sat in the BIOS setup on the health
status page monitoring the temps, voltages and fan speed and even
though all values looked fine, it even rebooted itself from there!

In my own mind I've narrowed it down to either the PSU or the
motherboard. I've found an article which explains how to monitor the
voltages coming out of the PSU with a multimeter but if the problem
doesn't happen for up to two weeks, how can I sit there watching a
multimeter for that amount of time? One other thing I read said it
could be the battery backup for the BIOS but the machine is maintaining
the BIOS settings and date/time when powered off for several days so I
don't think it could be that.

I can't really afford to start buying new parts unless I'm sure
that the original is faulty, so is there anything else I can do to help
me narrow down the cause or has anyone else experienced anything
similar?

What OS are you using ?

If the OS has an Event Viewer, check to see if any error codes or
messages are showing up there for each reboot.

If you have the option, disabled "automatic reboot" in one of the
control panels. If there is an error condition, then you'll get
a nice blue screen (BSOD) with error information. Sometimes the
name of the driver experiencing the error, gives a hint as to
what is wrong.

If disabling automatic reboots doesn't stop it, and it reboots,
then something is resetting the system. You could start by
disconnecting the reset button and cable, in case the switch is
flaky.

It could be Vcore is dropping out (bad motherboard regulation)
or it could be the power supply. A voltmeter won't catch all
fault types, if this is a glitch.

Other possibilities, are flaky connections somewhere. Some people
put an extra standoff underneath the system board, and that spells
trouble if the standoff contacts a copper track. Vibration or
a shock to the system, could cause the standoff to make contact
with the copper track.

The problem might disappear if you strip the system down and
reassemble it. For the standoffs, make sure each standoff installed
in the tray, mates with a tin plated ring on the bottom of the motherboard.

In terms of other tests you can run, you can use memtest86+ from
memtest.org . That will tell you if there is a "bad spot" in the
memory. I had a stick, where a couple memory locations consistently
errored. Memtest86+ does not detect all memory errors, and only
seems to catch the worst (stuck at) type. (It is a very good
program, but doesn't seem to be as effective as Prime95 overall.)

For system level testing, Prime95 from mersenne.org has a Torture
Test option. That carries out a calculation with a known answer.
If that program errors out in a few seconds, then you know that
the CPU, Northbridge, or memory had a hand in it.

For video card, you can try 3DMark. 3DMark2001SE build 330 has
a nice demo loop, that you can leave running indefinitely. It also
has a benchmark mode. See how long the system will run with a
test like that.

For disks, you can go to the disk drive manufacturer web site, and
get a test program. Some test programs have a non-destructive mode
where they just read stuff. See if the test program passes the
disks or not.

Gather as many additional symptoms as you can, and post the
results back. Flaky computers are not easy to debug, so you
need the results of as many tests as possible, including
any error messages you might find along the way.

Paul

 

[Sylvain VAN DER WALDE]

Yeah, pretty conclusive that its not overheating thats the problem.


You need a max min multimeter for that, but
they cost more than a new power supply.

Asus PC Probe has a "history" function which you could use to check the
voltages over a period of time.

Sylvain.

 

[w_tom]

Generally, power supply failures don't happen suddenly. As the power
supply is failing, computer may intermittently crash. But numbers for
a 'marginal' problem can be measured at any time.

Power supply is the foundation of a computer. If you household doors
start sticking, do you plane the doors - or first check for a slowly
failing foundation? Same with a power supply.

Best way to find such a problem is using the 3.5 digit multimeter
with computer running with tasks accessing as many peripherals as
possible - to fully load power supply. In your case, measure voltage
on any one of purple, red, yellow, and orange 'power supply to
motherboard' wires. Voltage numbers must exceed 3.23, 4.87, and 11.7.
Furthermore information may be obtained by reporting those numbers
here.

If in spec, then power supply apparently is not a problem; move on to
other suspects.

Then use meter to calibrate that motherboard voltage monitor. Those
motherboard monitors are good for detecting changes, but need be
calibrated before accurate voltages are reported.

A second powerful diagnostic is heat. Many fix computers by
installing 'more fans' - thereby curing symptoms of defective hardware.
Heating selective regions with a hairdryer on high in combination with
a comprehensive diagnostics can locate intermittents that will
eventually become hard failures later. Responsible manufacturers
provide comprehensive hardware diagnostics for free. If not, then
download diagnostics from component manufacturers or third party
diagnostics. All components inside a computer must work fine when
uncomfortable to touch but will not burn skin. Any component that does
not is probably failing - may get worse in the future.

Third information is in OS system logs. What does the system (event)
log report? What does device manager report?

Battery would not cause crashes. However, using an essential tool -
multimeter - measure battery voltage. Long before battery causes
problems, it will be obvious from meter numbers whether battery need be
replaced. Again, no reason for 'swapping this and swapping that'
nonsense. Only the naive do 'shotgunning'. Generally, problems can
be traced down to a suspect before replacing anything. That $20
multimeter is a tool as essential as a screwdriver.

 

[Rod Speed]

Asus PC Probe has a "history" function which you could use to check the voltages over a period of
time.

Trouble is that that is pathetic compared with a real multimeter and if the
problem is spikes that are out of spec, it may well not even notice those.

And since the history isnt kept on the hard drive, you wont
be able to inspect if after the system has rebooted anyway.

 

[Rod Speed]

Generally, power supply failures don't happen suddenly.

Oh bullshit.

As the power supply is failing, computer may intermittently crash.
But numbers for a 'marginal' problem can be measured at any time.

Not if they are very intermittent like every couple of weeks.

Power supply is the foundation of a computer. If you household
doors start sticking, do you plane the doors - or first check for
a slowly failing foundation? Same with a power supply.

Wota ****ing wanker.

Best way to find such a problem is using the 3.5 digit
multimeter with computer running with tasks accessing
as many peripherals as possible - to fully load power supply.

Useless if the problem is a dry joint or cracked trace
in the power supply. You're very unlikely to be able to
catch it with a fault that can be fine for a couple of weeks.

You need a proper max min multimeter for that.

In your case, measure voltage on any one of purple, red, yellow,
and orange 'power supply to motherboard' wires. Voltage numbers
must exceed 3.23, 4.87, and 11.7. Furthermore information may
be obtained by reporting those numbers here.

If in spec, then power supply apparently is not a problem;

Proves nothing if its in spec when measured,
but goes out of spec when you arent measuring it.

move on to other suspects.

Then use meter to calibrate that motherboard voltage monitor.

The problem with those aint just calibration, stupid.

Those motherboard monitors are good for detecting changes,

Nope, not if they are short term spikes. And they often do show short
term variations in voltage that arent even there when measured properly.

but need be calibrated before accurate voltages are reported.

Not a clue. As always.

A second powerful diagnostic is heat. Many fix computers by
installing 'more fans' - thereby curing symptoms of defective
hardware. Heating selective regions with a hairdryer on high in
combination with a comprehensive diagnostics can locate intermittents
that will eventually become hard failures later. Responsible
manufacturers provide comprehensive hardware diagnostics for free.

Not a ****ing clue, as always.

If not, then download diagnostics from component
manufacturers or third party diagnostics.

Fat lot of use that is with a fault that can show up only every couple of weeks.

All components inside a computer must work fine when
uncomfortable to touch but will not burn skin. Any component
that does not is probably failing - may get worse in the future.

Not a ****ing clue, as always.

Third information is in OS system logs. What does the system
(event) log report? What does device manager report?

The device manager is irrelevant with an intermittent fault like that.

Battery would not cause crashes. However, using an essential tool -
multimeter - measure battery voltage. Long before battery causes
problems, it will be obvious from meter numbers whether battery need
be replaced. Again, no reason for 'swapping this and swapping that'
nonsense. Only the naive do 'shotgunning'.

Only pig ignorant fools like you would try to find a
fault that can stay away for weeks with a multimeter.

Generally, problems can be traced down to a suspect before replacing anything.

Pity about intermittent faults which generally cant.

That $20 multimeter is a tool as essential as a screwdriver.

Pity its useless for an intermittent fault that can take weeks to show up again.

 

[RussellS]

Yeah, pretty conclusive that its not overheating thats the problem.


You need a max min multimeter for that, but
they cost more than a new power supply.


Yeah, that wont usually cause a reboot. Cheap to try a new battery tho.


Yes, check the XP logs to see if there is anything
useful there on what is producing the reboot.

Check for bad caps on the motherboard. Those are the usually blue or
black plastic covered post like things that stick up vertically from the
motherboard. The tops should be flat. If any of them have bulged or
have leaked, thats a bad cap and is likely whats producing the reboots.


Yep, plenty have, with a variety of causes. Its more often the PSU
than anything else, but it can certainly be a bad motherboard.

Can even be a faulty reset switch but thats very
rare, they dont usually conduct intermittently.

--------------------------------------------------------------
Very good hardware troubleshooting advice from Rod Speed.

In addition, updated/corrupt drivers, particularly for a graphics card, can
cause reboots/boot failure symptoms. Have you recently updated graphics or
sound drivers, or do you remember that the issue started up after receiving
some sort of OS or driver update? If your OS avails restore points, you
could try reverting to a point in time when you know everything was working
properly and/or uninstalling drivers, then reinstalling the latest versions
after a reboot.

Russell
http://tastycomputers.com

 

[Timothy Daniels]

"Fairy" taled:

My homebuilt PC has been running fine for approx five years
but just recently has started rebooting itself at random intervals.

Take a look at your "Add or Remove Programs" list in the Control
Panel. If you see "Yahoo! toolbar" listed, uninstall the PoS.

*TimDaniels*

 

[Timothy Daniels]

"Fairy" taled:

Take a look at your "Add or Remove Programs" list in the Control
Panel. If you see "Yahoo! toolbar" listed, uninstall the PoS.

Scratch that. I had intermittent BSODs during startup after the
OS was loaded, and they disappeared after I found and removed
a drive-by installation of Yahoo! Toolbar from my system.
But now they're back, so it's something else.

*TimDaniels*