Fault finding need advice

[MCR]

Hi,
I have a P5B motherboard with 4GB of RAM which dual boots with Vista and
Ubuntu. I left the machine on overnight (in ubuntu, I doubt it matters)
seeding a torrent. In the morning I found that the PC was off. I
pushed the power button and it briefly flashed.

I tried the usual to see the cause of the problem. I removed all drives
and tried powering (same), and I took out the RAM sticks. When it was a
board/GFX card and RAM the machine still would flash and powered off but
on the GFX card on of the caps cought fire!

I removed the card (a Nvidia 7900 PCI-E) and checked the board for
physical damage, there was none. I left the GFX card out, put all the
SATA/IDE devices in and powered up and it appeared to boot (although I
didnt get a POST beep, which may be because it is in error status), the
fan came on, I heard the drives start and the green light on the board
was on.

I didnt hear any beep codes, but obviously couldnt see if it was going
to boot. I removed one of the SATA drives (not the OS one) and plugged
it in another PC and it is trouble free,

Now the question

I suspect the board is fine and it is just the GFX card that has gone
bye-bye. However I dont want to spend money on another GFX card only to
find the board did it or the board is damaged.

What is the chance the motherboard is ok?
Is there a way I can tell without getting a board

BTW my other machine is not PCI-E it is AGP so I cannot swap over GFX cards.

Anyone have any ideas or suggestions?

 

[Paul]

Hi,
I have a P5B motherboard with 4GB of RAM which dual boots with Vista and
Ubuntu. I left the machine on overnight (in ubuntu, I doubt it matters)
seeding a torrent. In the morning I found that the PC was off. I
pushed the power button and it briefly flashed.

I tried the usual to see the cause of the problem. I removed all drives
and tried powering (same), and I took out the RAM sticks. When it was a
board/GFX card and RAM the machine still would flash and powered off but
on the GFX card on of the caps cought fire!

I removed the card (a Nvidia 7900 PCI-E) and checked the board for
physical damage, there was none. I left the GFX card out, put all the
SATA/IDE devices in and powered up and it appeared to boot (although I
didnt get a POST beep, which may be because it is in error status), the
fan came on, I heard the drives start and the green light on the board
was on.

I didnt hear any beep codes, but obviously couldnt see if it was going
to boot. I removed one of the SATA drives (not the OS one) and plugged
it in another PC and it is trouble free,

Now the question

I suspect the board is fine and it is just the GFX card that has gone
bye-bye. However I dont want to spend money on another GFX card only to
find the board did it or the board is damaged.

What is the chance the motherboard is ok?
Is there a way I can tell without getting a board

BTW my other machine is not PCI-E it is AGP so I cannot swap over GFX
cards.

Anyone have any ideas or suggestions?

With swapping hardware, as a means of identifying and repairing PCs,
there is always a danger that one part can damage another part. We've
had cases here, where it took a participant three dead parts, to arrive
at the conclusion that some other part was blowing them up. So there
is a cost associated with that kind of debugging. (This is one of the
risks that any store or repair shop takes, when they repair computers.)

The cheapest PCI Express card you could find right now, would be about
$20. This card would be good enough as a "slot checker", before using
an expensive card. (X1550).

http://www.newegg.com/Product/Product.aspx?Item=N82E16814102737

Another example, is the CPU swap. Some people have gone to a friend,
and said "let me borrow your CPU, to test my motherboard". When the
CPU arrives back at the friend's place, in no longer POSTs in the
friend's PC.

So stuff like that happens - not frequently, but enough for it to be
a concern.

With the video card removed, and the computer case speaker connected
to the motherboard, I'd prefer to hear the "missing video" beep.
Try removing the RAM, and see if you can get a beep from it that
way. If there is no beeping, it is also possible no BIOS code is
getting executed. And that might mean that more than one thing
got damaged, for whatever reason.

It could be that the +12V overvolted from the power supply, and fried
the Vcore regulator, and popped an input filter cap on the video
card. Or perhaps I'm just getting carried away...

So BigJim's advice sounds like a good place to start - the PSU.
You could grab a multimeter ($20 to $100 at a hardware store),
set it to 20VDC full scale, and measure the voltages on the main
ATX power connector, with the system powered. The tolerance of the
power supply is about 5%, so the voltage values if the thing is
working, should be pretty close to the nominal value. (You can
measure by sticking the multimeter probe into the backside of the
main connector - the connector voltages per pin are listed in the
motherboard manual.)

The main purpose of doing the measurement, is to see whether the
power supply was the culprit or not, and may have damaged more than
one piece of hardware. For example, it wouldn't take too much
overvoltage on +12V (motor and actuator), to damage the hard drives.

It is going to take more than a 5% error to blow up stuff, but
every component has a different tolerance to overvoltage. Switching
regulators can be designed to accept a wide range of voltages, if
the designer knows that wide excursions are to be expected.
Chips that use a power supply voltage directly, would be
more sensitive to an overvoltage event.

I'd also want to look carefully at the capacitor that caught
fire. I'd check the sleeve, and see what voltage rating is
printed on the cap. The purpose of looking at that cap, would
be to see whether it is an input filter cap that is connected
directly to a power supply voltage. (Just like a canary in a
coal mine.)

Paul

 

[MCR]

try another Power supply first

Thanks!

 

[MCR]

Paul wrote:

Snipped

With swapping hardware, as a means of identifying and repairing PCs,
there is always a danger that one part can damage another part. We've
had cases here, where it took a participant three dead parts, to arrive
at the conclusion that some other part was blowing them up. So there
is a cost associated with that kind of debugging. (This is one of the
risks that any store or repair shop takes, when they repair computers.)

The cheapest PCI Express card you could find right now, would be about
$20. This card would be good enough as a "slot checker", before using
an expensive card. (X1550).

http://www.newegg.com/Product/Product.aspx?Item=N82E16814102737

Another example, is the CPU swap. Some people have gone to a friend,
and said "let me borrow your CPU, to test my motherboard". When the
CPU arrives back at the friend's place, in no longer POSTs in the
friend's PC.

So stuff like that happens - not frequently, but enough for it to be
a concern.

With the video card removed, and the computer case speaker connected
to the motherboard, I'd prefer to hear the "missing video" beep.
Try removing the RAM, and see if you can get a beep from it that
way. If there is no beeping, it is also possible no BIOS code is
getting executed. And that might mean that more than one thing
got damaged, for whatever reason.

It could be that the +12V overvolted from the power supply, and fried
the Vcore regulator, and popped an input filter cap on the video
card. Or perhaps I'm just getting carried away...

So BigJim's advice sounds like a good place to start - the PSU.
You could grab a multimeter ($20 to $100 at a hardware store),
set it to 20VDC full scale, and measure the voltages on the main
ATX power connector, with the system powered. The tolerance of the
power supply is about 5%, so the voltage values if the thing is
working, should be pretty close to the nominal value. (You can
measure by sticking the multimeter probe into the backside of the
main connector - the connector voltages per pin are listed in the
motherboard manual.)

The main purpose of doing the measurement, is to see whether the
power supply was the culprit or not, and may have damaged more than
one piece of hardware. For example, it wouldn't take too much
overvoltage on +12V (motor and actuator), to damage the hard drives.

It is going to take more than a 5% error to blow up stuff, but
every component has a different tolerance to overvoltage. Switching
regulators can be designed to accept a wide range of voltages, if
the designer knows that wide excursions are to be expected.
Chips that use a power supply voltage directly, would be
more sensitive to an overvoltage event.

I'd also want to look carefully at the capacitor that caught
fire. I'd check the sleeve, and see what voltage rating is
printed on the cap. The purpose of looking at that cap, would
be to see whether it is an input filter cap that is connected
directly to a power supply voltage. (Just like a canary in a
coal mine.)

Paul

Thanks for the advice, I'll grab a PSU too...

 

[spodosaurus]

Hi,
I have a P5B motherboard with 4GB of RAM which dual boots with Vista and
Ubuntu. I left the machine on overnight (in ubuntu, I doubt it matters)
seeding a torrent. In the morning I found that the PC was off. I
pushed the power button and it briefly flashed.

I tried the usual to see the cause of the problem. I removed all drives
and tried powering (same), and I took out the RAM sticks. When it was a
board/GFX card and RAM the machine still would flash and powered off but
on the GFX card on of the caps cought fire!

I removed the card (a Nvidia 7900 PCI-E) and checked the board for
physical damage, there was none. I left the GFX card out, put all the
SATA/IDE devices in and powered up and it appeared to boot (although I
didnt get a POST beep, which may be because it is in error status), the
fan came on, I heard the drives start and the green light on the board
was on.

I didnt hear any beep codes, but obviously couldnt see if it was going
to boot. I removed one of the SATA drives (not the OS one) and plugged
it in another PC and it is trouble free,

Now the question

I suspect the board is fine and it is just the GFX card that has gone
bye-bye. However I dont want to spend money on another GFX card only to
find the board did it or the board is damaged.

What is the chance the motherboard is ok?
Is there a way I can tell without getting a board

BTW my other machine is not PCI-E it is AGP so I cannot swap over GFX
cards.

Anyone have any ideas or suggestions?

Boot is when the PC boots an operating system. POST is Power On Self
Test, which comes before boot. If you get nothing on the screen, it has
failed to POST and will not get to the boot stage. You need to test your
PSU with a multimeter. Restarting a PC with a dodgy PSU is a good way to
kill all of the components, starting with the motherboard. You've
already had one catastrophic hardware failure with your video card, and
I doubt it went by itself.

Ari

--
spammage trappage: remove the underscores to reply
Many people around the world are waiting for a marrow transplant. Please
volunteer to be a marrow donor and literally save someone's life:
http://www.abmdr.org.au/
http://www.marrow.org/