PC suddenly shuts down

[Peter Sinclair]

Hi,

The PC my family uses at home suddenly powers off for no apparent
reason. When the power button is hit again, it stays at a black screen
with the hard drive light on the front LED panel glowing.

If the hard drive is disconnected, then the system gets past the black
screen, loads up the BIOS welcome screen and of course reports that
there is a disk boot failure. The same temporary fix is achieved by
powering off the switch at the back of the PSU for about a minute.

We have tried the obvious, which is to switch hard drives. Three
different unique hard drives were tested; 2 IDEs and 1 SATA hard
drive, with fresh Windows installations on each. The problem persists.

The PCI cards on the computer are a Gigabit ethernet card, SB Audigy,
XFX GeForce 9600 GSO. The motherboard is an Asus P5ND2-sli with a dual
core intel CPU. The drives are 1 optical and 1 hard drive.

The system was previously overclocked and had overheating problems,
but it is being run at stock speeds now with no further heat related
problems. The system powers down even during non-power hungry times
like just being idle in Windows or whilst surfing the net.

Can anyone suggest what the problem might be?

Thanks.

Peter

 

[Paul]

Hi,

The PC my family uses at home suddenly powers off for no apparent
reason. When the power button is hit again, it stays at a black screen
with the hard drive light on the front LED panel glowing.

If the hard drive is disconnected, then the system gets past the black
screen, loads up the BIOS welcome screen and of course reports that
there is a disk boot failure. The same temporary fix is achieved by
powering off the switch at the back of the PSU for about a minute.

We have tried the obvious, which is to switch hard drives. Three
different unique hard drives were tested; 2 IDEs and 1 SATA hard
drive, with fresh Windows installations on each. The problem persists.

The PCI cards on the computer are a Gigabit ethernet card, SB Audigy,
XFX GeForce 9600 GSO. The motherboard is an Asus P5ND2-sli with a dual
core intel CPU. The drives are 1 optical and 1 hard drive.

The system was previously overclocked and had overheating problems,
but it is being run at stock speeds now with no further heat related
problems. The system powers down even during non-power hungry times
like just being idle in Windows or whilst surfing the net.

Can anyone suggest what the problem might be?

Thanks.

Peter

There are a few reasons for power off.

The power supply itself may have detected an internal problem, and
decided to power off. That may be accompanied by latch-off, meaning
the supply won't function again, until you turn the switch on the
back to the OFF position, then back ON again.

The processor has a signal called THERMTRIP, which can be used to
turn off the computer. On a modern processor, the THERMTRIP temperature
is 20C above the throttle temperature. (So throttle and loss of performance
start at 70C and then perhaps THERMTRIP is at 90C.) You can use a program
like SpeedFan (almico.com) to monitor the hardware temperature readings, and
get some idea what your temperatures are currently.

Another latchoff style event, is a Vcore switching regulator problem.
If the Vcore regulator thinks there is a problem with the MOSFETS,
like an overload, it may shut off without warning. That may require toggling
the power switch on the back of the machine as well.

You could strip some of the hardware off the machine and retest. And gradually
add it back, to see if there is any change in symptoms. For example, you
could use a Linux LiveCD like Knoppix, download Prime95 for Linux from
mersenne.org/freesoft , and do a stress test while no hard drives, or
PCI cards are present. The current version of Knoppix, even supports making
a USB flash stick version, so you could even unplug the CDROM and just
use USB flash to boot the system.

I've had one Antec power supply fail here, and it had leaking capacitors inside
it. The supply made a muffled "arcing" sound at startup, which is when
I started having suspicions. I didn't keep the supply in there to the
point of failure, and removed it one day, when it crashed in the BIOS.
Since the supply was out of warranty, I was able to open the supply for
a quick look, and there were four caps with brownish-orange goo on
the top. You can also look for bulging or leaking caps, on the
motherboard itself, as they're part of the Vcore regulator.

http://en.wikipedia.org/wiki/Capacitor_plague

Paul

 

[Peter Sinclair]

There are a few reasons for power off.

The power supply itself may have detected an internal problem, and
decided to power off. That may be accompanied by latch-off, meaning
the supply won't function again, until you turn the switch on the
back to the OFF position, then back ON again.

The processor has a signal called THERMTRIP, which can be used to
turn off the computer. On a modern processor, the THERMTRIP temperature
is 20C above the throttle temperature. (So throttle and loss of performance
start at 70C and then perhaps THERMTRIP is at 90C.) You can use a program
like SpeedFan (almico.com) to monitor the hardware temperature readings, and
get some idea what your temperatures are currently.

Another latchoff style event, is a Vcore switching regulator problem.
If the Vcore regulator thinks there is a problem with the MOSFETS,
like an overload, it may shut off without warning. That may require toggling
the power switch on the back of the machine as well.

You could strip some of the hardware off the machine and retest. And gradually
add it back, to see if there is any change in symptoms. For example, you
could use a Linux LiveCD like Knoppix, download Prime95 for Linux from
mersenne.org/freesoft , and do a stress test while no hard drives, or
PCI cards are present. The current version of Knoppix, even supports making
a USB flash stick version, so you could even unplug the CDROM and just
use USB flash to boot the system.

I've had one Antec power supply fail here, and it had leaking capacitors inside
it. The supply made a muffled "arcing" sound at startup, which is when
I started having suspicions. I didn't keep the supply in there to the
point of failure, and removed it one day, when it crashed in the BIOS.
Since the supply was out of warranty, I was able to open the supply for
a quick look, and there were four caps with brownish-orange goo on
the top. You can also look for bulging or leaking caps, on the
motherboard itself, as they're part of the Vcore regulator.

http://en.wikipedia.org/wiki/Capacitor_plague

Paul

Hi Paul,

Thanks for your advice.

Recently, we had quite a few bad brownouts due to severe weather
conditions. The lights would flicker and the power would go out for
about a second. It played havoc with the computer, reset the clocks in
the house and I distinctly remember my son complaining about how it
was affecting his online games.

My son has disassembled the computer PSU and opened it. Apart from
lots of dust, everything looks normal and there are no bulging caps.
Ditto the motherboard.

With just the PSU, motherboard, hard drive and graphics card (the
motherboard doesn't have onboard graphics), the computer now isn't
getting to the BIOS at all. We've tried unplugging the hard drive and
also leaving the PSU off for a few minutes. We shall try replacing the
graphics card with any spare PCI-e one we can find from the old
computers in the garage and see if that is the cause.

What other possibilities could it be?
CPU overheat is out because now it isn't getting into the BIOS even
from a cold start.
Can't be hard or optical drives.
Looks like it is either the PSU, motherboard or graphics card.

Thanks.

Peter

 

[David W. Hodgins]

The PC my family uses at home suddenly powers off for no apparent
reason. When the power button is hit again, it stays at a black screen
with the hard drive light on the front LED panel glowing.

I had a similar problem last August with a custom built new
system. Turned out to be a problem with the power supply.

The 500W power supply worked properly, but only provided
24 amps on the +12v rail. The nVidia GeForce 9600 GT video
card recommended a power supply with a minimum of 26 amps.

The system would run ok with minimal devices connected, but
would reboot after sometime (even when idle) if the dvd drive
or multiple usb devices were connected.

Replacing the power supply with a Thermaltake 650W supply,
with up to 52 amps on the +12v rail fixed the problem.

Regards, Dave Hodgins

 

[Peter Sinclair]

Try powering on the system without the video card to see if
you get beeps. If not, it's probably not the video card -
or at least that's not the primary problem, if your power
fluctuations included some power surges there could be
multiple parts that have damage.

We swapped the high powered XFX Geforce 9600 GSO card with a lower
spec 6200 or something card. It still didn't even get to the BIOS.

Also tried a different power socket in the house. So things we've
tried:
Son's PC:
9600 GSO graphics card with a 6200 card
Removed all non essential PCI cards
Removed hard & optical drives
Results: Black screen, still won't get to BIOS

Daughter's PC:
6600 GT graphics card
No other PCI cards
Removed hard & optical drives
Results: Black screen, still won't get to BIOS

Both computers use Sansun PSUs, although they were bought about 2+
years apart.

Failing PSU seems most likely, though since it still seems
to stay turned on when you attempt to start the system,
swapping in a lower powered video card could temporarily
seem to fix the problem... which could surface again as the
PSU continues to degrade.

Curiously while we were testing things and swapping parts around, we
found that another computer in my daughter's bedroom that is less
used, is also displaying similar problems of a black screen and not
getting into the BIOS. Both my daughter and son say their computers
were on and affected by the brownouts about two weeks ago - they had
to switch them on again many times.

I checked the output voltages of both PSUs with my multimeter, and
they seemed fine.

I don't mean to steer you totally towards the PSU, could be
anything intermittently shorting out and the PSU is working
properly to shut down when it senses an overcurrent
condition, but it would not explain the times the system
turns on but does not POST unless the power was also
instable.

I've ordered a 2 600VA uninterruptible power supplies and 2 new Artic
Silver PSUs for my son and daughter's computers. Let's hope this sorts
out the problem.

http://www.ebuyer.com/product/159615
http://www.ebuyer.com/product/124922

Are these manufacturers reputable? I don't want to spend too much
money on my kids computers as they have been asking for an upgrade,
but by the same token I don't want them to conk out too soon either.

May I ask if laptops which I use for work, even if connected to the
mains through their rechargers, are less susceptible as desktops to
power spikes?

Thanks.

Peter

 

[larry moe 'n curly]

Son's PC:
9600 GSO graphics card with a 6200 card
Removed all non essential PCI cards
Removed hard & optical drives
Results: Black screen, still won't get to BIOS

Daughter's PC:
6600 GT graphics card
No other PCI cards
Removed hard & optical drives
Results: Black screen, still won't get to BIOS

Both computers use Sansun PSUs, although they were bought about 2+
years apart.

I checked the output voltages of both PSUs with my multimeter, and
they seemed fine.


I've ordered a 2 600VA uninterruptible power supplies and 2 new Artic
Silver PSUs for my son and daughter's computers. Let's hope this sorts
out the problem.

http://www.ebuyer.com/product/159615
http://www.ebuyer.com/product/124922

Are these manufacturers reputable?

CyberPower is probably the second most common backup power supply in
the US retail market, APC being the leader. I can't find any reviews
of CyberPower backups and have never looked inside one, but I noticed
that Belkins made by two different companies (one was Delta) had a lot
fewer parts than products from APC did. Also those Belkins failed the
laser printer test: plug the backup and a laser printer into the same
wall outlet, and turn on the printer. With the Belkin, the computer
plugged into the backup locked up, and the backup alarm went off.
With every APC or Conext I tried, nothing bad happened. Conext is or
was another brand of APC but isn't sold in the US any more.

I couldn't find any legitimate reviews of Sansun or Artic Silver PSUs,
but HardwareSecrets.com reviewed an ArCtic Fusion 550RF power supply:

http://www.hardwaresecrets.com/article/713

They said this PSU was made by Seasonic (very good) and sold by them
as a 450W model. Arctic labels it a 550W model, but that rating
refers to the maximum the PSU can put out for just ONE second:

http://www.hardwaresecrets.com/fullimage.php?image=15721

Some companies get their PSUs from several manufacturers, so it's not
always possile to generalize about any one brand.

Don't buy bad PSUs. I've never spent more than $20 US (includes $5 on
new capacitors) on a PSU but have never gotten junk, except for a
couple of free-after-rebate models. Try to stick with the best
manufactures: Delta/Newton, Enhance, Lite-On, Seasonic, Win-tact,
NMB, Fortron-Source/Sparkle, Etasis. You can usually find the actual
manufacturer from the UL file number (listed as Exxxxxx) from www.ul.com
or CSA registration number (Lxxxxx) from www.csa.ca. If the PSU
doesn't have a UL or CSA number, then avoid it.

 

[Peter Sinclair]

Don't buy bad PSUs. I've never spent more than $20 US (includes $5 on
new capacitors) on a PSU but have never gotten junk, except for a
couple of free-after-rebate models. Try to stick with the best
manufactures: Delta/Newton, Enhance, Lite-On, Seasonic, Win-tact,
NMB, Fortron-Source/Sparkle, Etasis. You can usually find the actual
manufacturer from the UL file number (listed as Exxxxxx) from www.ul.com
or CSA registration number (Lxxxxx) from www.csa.ca. If the PSU
doesn't have a UL or CSA number, then avoid it.

Thanks for the advice everyone. Replacing the PSUs fixed my son's
computer but my daughter's computer still didn't post.

In the process of cannibalising it for parts, we found out that one of
the RAM memory modules was faulty. Removing it allowed the computer to
work again, albeit with less RAM. I'm surprised a brownout could cause
a PSU and RAM to go faulty; every other part on the computer appears
to be fine so far.

The RAM that failed was Crucial memory with a lifetime warranty.
Anyone know if they typically honor the warranty and would replace
faulty 512MB of DDR400 memory with the same?

I will be buying new computers for them soon (have to see how a big
contract I am bidding on turns out). I told my son and daughter any
money they get from selling their computers off Ebay, once upgraded,
is theirs. I'm trying to teach them to be enterprising!

Peter