Suspect PSU?

[Martin]

I have a problem with a home-build PC that's been running XP with SP2
flawlessly since Aug 2005.

In the past two or three months it's developed a problem where it
randomly just reboots itself.
To begin with it would only reboots very occasionally so the problem
was not dealt with.
In recent weeks the PC has taken to rebooting very often so i was given
the job of fixing it.

First thing i suspected was CPU overheating - BIOS reporting temps of
approx 65C at idle just minutes (seconds even) after booting from cold.
I removed the CPU's thermal pad, cleaned both CPU and heatsink with
spirits and applied a small blob of thermal paste.
CPU temp now averaging a much more healthy 50C on average - the HSF is
not a performance model, just an el cheapo, so i suspect that's about
the coolest it'll run without a better HSF.
Anyway the reboots continued after the CPU temp treatment so i rules
out CPU overheating.

The reboots happen anytime - either when booted into XP or just in the
BIOS setup mode - so i ruled out any software problem.
(We did successfully clean install XP 7 days ago, this was when the PC
wasn't showing any signs of rebooting - needless to say the problem
continued).

I removed the AGP card, cleaned the contacts with a pencil eraser, and
reinstalled it - still the reboots occur.
(The PC has no PCI cards to inspect).

I removed both memory modules and reseated them - still the reboots
occur.

Only components left to suspect (other than the motherboard) were the
two hard drives and the CD-ROM.
CD-ROM was prime suspect as it's ancient!
CD-ROM has been disconnected (both IDE and power cables unplugged) for
the past 5 days - still the reboots occur.

The motherboard's BIOS is the last but one update available for it -
the latest BIOS update states that it includes a fix for large hard
drives, one of the PC's hard drives is a 200GB drive so i downloaded
the BIOS update and decided to update the BIOS.
The PC has no floppy so i dug out a CDR that i'd made years ago with
the WinME boot floppy as it's image.
With that i could boot to DOS and update the BIOS i hoped.
(The BIOS update and BIOS updater being on a FAT32 partition that i
hoped to be able to access after booting from the WinME boot CD).

I reconnected the CD-ROM inserted the WinME CD (the PC was booted to XP
at the time) and as soon as i closed the CD-ROM tray the PC rebooted.
It then went into a loop rebooting, getting to the part of the BIOS
screen where it scans for IDE devices and then as the CD-ROM spins up
it'd reboot.

I removed the WinME boot CD and it booted ok but after less than 15
minutes started to reboot.
I disconnected the CD-ROM and again reboots happen.

I then disconencted the 200GB drive leaving it to run on it's 40GB boot
drive - within less than 15 moinutes it'd start rebooting - so maybe
the BIOS update with the large hard drive fix wasn't going to help
anyway.

I'm now suspecting the PSU is the cause - a power drain surge as the
CD-ROM spins up may explain them reboots. The fact that the reboots
happen with the CD-ROM disconnected indicate that the CD-ROM itself is
unlikely to be the cause.

The PC is a homebuild with a cheapo case and an el cheapo unbranded
300W PSU.
It's worked perfectly from August last year up to about 10 weeks ago
and the reboots happen more and more.

Here's some technical details about the PC:
Foxconn K7S741GXMG motherboard.
Sempron 2400+ CPU.
2 x 256MB DDR333 DIMMs.
Connect 3D Radeon 9550 AGP card.
200GB and 40GB hard drives.
CD-ROM.

BIOS 'PC Health' screen reports these voltages:
CPU core - 1.58v
+3.3v - between 3.21v and 3.26v averaging about 3.24v
+5v - 5.15v
+12v - 12.32v

What do you all think then?

The PC's owner hopes to add a DVD writer sometime soon so i reasoned
that a new PSU would not only (hopefully) fix the reboots but also
prepare it for another device that'll be more watts of power.

Thanks for any help.

Martin.

 

[kony]

I have a problem with a home-build PC that's been running XP with SP2
flawlessly since Aug 2005.

Ok, but, since this is a hardware group it's always a good
idea to first list your major hardware, and the PSU make,
model, wattage- even better to include the amperage rating
on the most loaded rail which is typically the 12V rail.

I see you did list the system below, but what's this PSU?
Given the symptoms, if this is a generic came-with-case PSU,
it's such a likely failure point that I'd go ahead and
replace it, unless an inspection reveals your motherboard
has failing capacitors.

In the past two or three months it's developed a problem where it
randomly just reboots itself.
To begin with it would only reboots very occasionally so the problem
was not dealt with.
In recent weeks the PC has taken to rebooting very often so i was given
the job of fixing it.

First thing i suspected was CPU overheating - BIOS reporting temps of
approx 65C at idle just minutes (seconds even) after booting from cold.
I removed the CPU's thermal pad, cleaned both CPU and heatsink with
spirits and applied a small blob of thermal paste.
CPU temp now averaging a much more healthy 50C on average - the HSF is
not a performance model, just an el cheapo, so i suspect that's about
the coolest it'll run without a better HSF.
Anyway the reboots continued after the CPU temp treatment so i rules
out CPU overheating.

Not necessarily, we don't need to know the average temp so
much as the max temp. Actually only the max temp matters,
it is useful to run a stress test like Prime95's Torture
Test, noting what the max temp was (after running at least
30 minutes), AND whether during this test, any errors were
reported by Prime95. The error feedback part is important
as many stress tests won't tell you about that problem.

However, after you redid the heatsink, it's not so likely
the system would be overheating from a cold-off state onto
the point of rebooting so soon thereafter. Unless the
heatsinks aren't making contact well at all, it'll take a
few minutes before they get up to normal operating temp as
the chassis also heats up.

The reboots happen anytime - either when booted into XP or just in the
BIOS setup mode - so i ruled out any software problem.

If you have a bios health monitor menu page, have you left
it on that page to observe the voltages?

Do you have a multimeter you might use to check voltages?
If so, leaving it connected might allow comparison with the
bios.

(We did successfully clean install XP 7 days ago, this was when the PC
wasn't showing any signs of rebooting - needless to say the problem
continued).

Is it possible something has changed since then? Had the
system been moved, and that stressed some mechanical
connection, or maybe it was placed next to a heater or HVAC
duct blowing hot air? Might it be on a home AC circuit with
other equipment (like a heater or motorized
something-or-other, maybe a frige)?

I removed the AGP card, cleaned the contacts with a pencil eraser, and
reinstalled it - still the reboots occur.
(The PC has no PCI cards to inspect).

Ideally you should not clean contacts with an eraser,
because practically all erasers are abrasive and the gold
plating on the contacts is very very thin and soft. Even
so, one time shouldn't kill it, though eventually it'll wear
out the gold and the underlying copper will corrode... over
time, so it's best to clean with a non-abrasive evaporating
cleaner like alcohol or official contact cleaner.

I removed both memory modules and reseated them - still the reboots
occur.

Only components left to suspect (other than the motherboard) were the
two hard drives and the CD-ROM.
CD-ROM was prime suspect as it's ancient!
CD-ROM has been disconnected (both IDE and power cables unplugged) for
the past 5 days - still the reboots occur.

The motherboard's BIOS is the last but one update available for it -
the latest BIOS update states that it includes a fix for large hard
drives, one of the PC's hard drives is a 200GB drive so i downloaded
the BIOS update and decided to update the BIOS.
The PC has no floppy so i dug out a CDR that i'd made years ago with
the WinME boot floppy as it's image.
With that i could boot to DOS and update the BIOS i hoped.
(The BIOS update and BIOS updater being on a FAT32 partition that i
hoped to be able to access after booting from the WinME boot CD).

I reconnected the CD-ROM inserted the WinME CD (the PC was booted to XP
at the time) and as soon as i closed the CD-ROM tray the PC rebooted.
It then went into a loop rebooting, getting to the part of the BIOS
screen where it scans for IDE devices and then as the CD-ROM spins up
it'd reboot.

You might be lucky- if the system is that instable you dont'
want to be trying to flash the bios, the odds are too great
of losing it. Although a bios update can fix logical bugs,
seldom would it be to combat this kind of problem- unless
your bios had a glitch that made it incorrectly detect
hardware temperature and then it was tripping a safety
shutoff threshold, BUT generally that shuts the system off,
not just resetting to a continual reboot... it'd be fairly
pointless for the feature to reset but keep running.

I removed the WinME boot CD and it booted ok but after less than 15
minutes started to reboot.
I disconnected the CD-ROM and again reboots happen.

I then disconencted the 200GB drive leaving it to run on it's 40GB boot
drive - within less than 15 moinutes it'd start rebooting - so maybe
the BIOS update with the large hard drive fix wasn't going to help
anyway.

Unquestionably, the large hard drive fix aspect of the bios
update would not help you with this problem. "Sometimes"
bios also include other fixes or changes not enumerated in
the bios notes, but for a problem this significant I'd
expect there to be a primary listing of this correction.

I'm now suspecting the PSU is the cause - a power drain surge as the
CD-ROM spins up may explain them reboots. The fact that the reboots
happen with the CD-ROM disconnected indicate that the CD-ROM itself is
unlikely to be the cause.

Agreed, PSU is the prime suspect.

The PC is a homebuild with a cheapo case and an el cheapo unbranded
300W PSU.

Ok but there must be some kinda label on it?
You might unplug it from AC for a few minutes then open it
and inspect it. I'll randomly guess that it's capacitors
have failed.

It's worked perfectly from August last year up to about 10 weeks ago
and the reboots happen more and more.

Here's some technical details about the PC:
Foxconn K7S741GXMG motherboard.
Sempron 2400+ CPU.
2 x 256MB DDR333 DIMMs.
Connect 3D Radeon 9550 AGP card.
200GB and 40GB hard drives.
CD-ROM.

BIOS 'PC Health' screen reports these voltages:
CPU core - 1.58v
+3.3v - between 3.21v and 3.26v averaging about 3.24v
+5v - 5.15v
+12v - 12.32v

What do you all think then?

The PC's owner hopes to add a DVD writer sometime soon so i reasoned
that a new PSU would not only (hopefully) fix the reboots but also
prepare it for another device that'll be more watts of power.

Since the system already has upgrades planned, it seems a
good idea to replace the PSU even if it isn't the immediate
problem, as it's likely to be later if not yet... but odds
are good that's the problem.

 

[Mike Walsh]

Low power supply voltage or intermittent loss of voltage can cause a reboot. Set the BIOS to have the PC remain off instead of turning on when power is applied. You can then tell if you have a problem with power because the PC will remain off instead of rebooting.

 

[w_tom]

...
BIOS 'PC Health' screen reports these voltages:
CPU core - 1.58v
+3.3v - between 3.21v and 3.26v averaging about 3.24v
+5v - 5.15v
+12v - 12.32v

What do you all think then?
...

Assuming your motherboard monitor has been calibrated, then your 3.3
volts is too low while 5 & 12 are on and should not be so high. If one
voltage is too low and other tend high, then good reason to suspect
that power supply. What do those voltages do as peripherals (including
CD-Rom) are accessed simultaneously? If 3.3 drops even slightly, then
it is defective. To blame, with certainly, the power supply, we would
need an oscilloscope. Even if power supply is good, the 3.3 volt
numbers is too marginal for long term reliable service.

But again, above assumes numbers from motherboard have been
calibrated which is why a 3.5 digit multimeter would be a useful tool -
to calibrate the motherboard and to verify a new supply when installed.
Verified as multiple peripherals are accessed simultaneously as Kony
has suggests using bios health monitor menu page; else get numbers
using multimeter.

Temperature was not a valid suspect. CPU temperature must be
sufficiently low so that when room temperature increases 15 degrees C
(exceeds 100 degrees F), then CPU still does not exceed manufacturer
numbers (typically 90 degree C). Your temperature numbers were so
below as to be irrelevant to the problem.

Another reason to ignore those who recommend using an eraser on
electrical contacts: connectors are self cleaning. Any minimally
acceptable connector will be cleaned simply by 'breaking and remaking'.
Furthermore, design must have sufficient margin to make any 'dirty'
contact irrelevant. If using an eraser does anything useful, then
hardware is 100% defective.

Kony suggests power supply capacitors could be a reason for failure.

From your numbers, that would be capacitors for that unacceptable 3.3

volts output.

 

[Martin]

Thanks for all the replies.

Been in touch with the PC's owner again today and the problem is
getting worse.
PC now unusable due to it's rebooting.

Haven't got a multimeter to test the PC's current PSU but we intend to
replace it anyway.

Martin.

 

[w_tom]

You have a screw driver? If doing anything electrical, then meter is
useful. The tool is sold even in Home Depot, Lowes, Radio Shack,
Sears, and K-mart. $20 standard price; seen on sale for $10. Question
is not what you have. Question is what is needed. Once that power
supply is replaced, calibrate the motherboard monitor AND confirm
problem is really fixed. Do you assume replacing what was a good power
supply with a new power supply automatically fixes the problem? More
reasons why a meter and screw driver are both needed.

 

[Martin]

Well it took us a few days to get some cash together to buy a new PSU
and we fitted it earlier today.

PC ran for an hour and a half faultlessly (so far) - no signs of any
problems.
CD-ROM re-connected and working perfectly too.
Also the BIOS health readings say all voltage lines are getting the
required voltage or (slightly) more.
The 3.3v line now recognising 3.31v from the new PSU.

We hope that's the problem fixed

Martin.