Computer popped


M

Metspitzer

A few months back, a spare computer popped and smoked when it was
turned on. I looked very carefully at the mobo to see if I would find
any trace of smoke. I didn't. That, incorrectly, made me think it
was the Power Supply. I ordered a new PS and the computer was still
broke. It would turn on and play, but video and sound were choppy.

I harvested the hard drive from it and put the case and dead computer
away. Well the Internet was down for around 5 hrs in Georgia today. I
was going to tinker with the dead machine. I pulled the CPU fan off
and was going to remove the chip and I found what looks like a sign of
smoke.

I didn't want to take the CPU out and take a chance on bending pins so
I just snapped a couple of shots. Could the popping sound we heard
cause something like this:

http://imgur.com/a/nO9fH

I was pretty sure it was a dead mobo and I would just order a new one,
but now it looks like there still may be two parts that could be bad.
It could be the mobo or the chip.

Suggestions?
 
Ad

Advertisements

P

Paul

Metspitzer said:
A few months back, a spare computer popped and smoked when it was
turned on. I looked very carefully at the mobo to see if I would find
any trace of smoke. I didn't. That, incorrectly, made me think it
was the Power Supply. I ordered a new PS and the computer was still
broke. It would turn on and play, but video and sound were choppy.

I harvested the hard drive from it and put the case and dead computer
away. Well the Internet was down for around 5 hrs in Georgia today. I
was going to tinker with the dead machine. I pulled the CPU fan off
and was going to remove the chip and I found what looks like a sign of
smoke.

I didn't want to take the CPU out and take a chance on bending pins so
I just snapped a couple of shots. Could the popping sound we heard
cause something like this:

http://imgur.com/a/nO9fH

I was pretty sure it was a dead mobo and I would just order a new one,
but now it looks like there still may be two parts that could be bad.
It could be the mobo or the chip.

Suggestions?

What am I looking for exactly ?

I see dust in the photo. I see a radial pattern of dust deposited from
the cooler you've removed.

I see signs of some thermal paste on the CPU. It looks like the paste
might not have reached to the edge of the CPU, and now there is dust around
the edge of the CPU.

Am I misinterpreting what I'm seeing ?

I don't see any bad caps in the area of the CPU shown in the picture.
None of the stamped lines on the capacitors appear cracked.

http://i.imgur.com/o8e0Uh.jpg

A popping sound, can be arcing around a high voltage capacitor
in the power supply.

A failing power supply, can damage other stuff in a computer.
So assuming just one thing was damaged, might not be it either.

If the computer is working, but is "choppy", well... it's still working.
It means the thing still "computes". And a lot of it has to be
working, for that to be the case.

If you want an alternative test case, boot a Linux LiveCD, then
use the movie player in there and play some movie content. Is it
still choppy or not ? If the playback is smooth, then there is
something going on with the Windows install.

Review the BIOS settings, and make sure it's set up right.

I don't see a reason to be running off and buying stuff, without
at least some hint of what's cooked :) If this was a rush job
(computer must be running by sunset), then by all means, buy
a stack of parts and go to it. But if this is a leisure
activity, run more test cases first and try and narrow it down.

In terms of the picture you've provided, I'd take the PC outside
and blow the dust off it. If there was a carbon trail left on
there somewhere, I'd hope it would remain intact, while the dust
would take a hike. If there is something there you want me to
see, that might improve the contrast ratio.

I've only seen pictures of a few, truly burned PCs. My favorite,
was a plane to plane short around the CPU socket, which causes the
entire area around the CPU to be charred. The VCore regulator
can provide 100 amps at a low voltage, and as long as the
overcurrent doesn't trip on it, it can "cook" a carefully adjusted
short. If the short has just the right resistance value, then
it could sit there for hours cooking, until the PCB develops
a charred appearance.

Other kids of faults, are toroidal coils that are burned so
bad, the core disintegrates when touched. Or, you can have a
MOSFET that is burned black. Those are other kinds of faults,
where the CPU will not POST or boot as a result. Your machine
runs, so by definition, there can't be that kind of damage.

*******

If you want some software to run, try DPCLat, which measures the
service time for Delayed Procedures. It indirectly identifies
if the CPU is not able to service things in a timely manner.

http://www.thesycon.de/deu/latency_check.shtml

That web page has some pictures of the DPCLat window. You should
have low amplitude green blocks, rather than a lot of red spikes.
That program runs in Windows. I don't run that very often (perhaps
once, per system build) as normally it's a check that the BIOS
doesn't have a bug. (A bad BIOS design, can cause some of the
red spikes.)

Paul
 
F

Flasherly

A few months back, a spare computer popped and smoked when it was
turned on. I looked very carefully at the mobo to see if I would find
any trace of smoke. I didn't. That, incorrectly, made me think it
was the Power Supply. I ordered a new PS and the computer was still
broke. It would turn on and play, but video and sound were choppy.

I harvested the hard drive from it and put the case and dead computer
away. Well the Internet was down for around 5 hrs in Georgia today. I
was going to tinker with the dead machine. I pulled the CPU fan off
and was going to remove the chip and I found what looks like a sign of
smoke.

I didn't want to take the CPU out and take a chance on bending pins so
I just snapped a couple of shots. Could the popping sound we heard
cause something like this:

http://imgur.com/a/nO9fH

I was pretty sure it was a dead mobo and I would just order a new one,
but now it looks like there still may be two parts that could be bad.
It could be the mobo or the chip.

Suggestions?

Doubt the CPU is bad, but it's going to be vintage anyway after a few
years. S478? Wrap the CPU in some foam and put it in the bag that
grows with the rest of them. Being able to troubleshoot old stuff
with known working spare parts turns into a privilege at some point as
parts become harder to find;- not sure that was the inverse case on
DDR2 pricing on Ebay when I looked recently. New MB I picked then
also was from the standpoint of longevity -- says on the package it's
built to last 25% longer on MTBW averaged. Fairly priced but not
outstanding in a $200+ scheme of 6 and 8 multicored MBs. Less an ASUS
loyalist and trying crossed-fingers with solid state capacitors these
days.
 
M

Metspitzer

What am I looking for exactly ?
On the right side it looks like it could be burned. I don't see too
many CPUs because they are always covered with the fan.
It looks darker than dust to me. I hope I am wrong.
I see dust in the photo. I see a radial pattern of dust deposited from
the cooler you've removed.

I see signs of some thermal paste on the CPU. It looks like the paste
might not have reached to the edge of the CPU, and now there is dust around
the edge of the CPU.

Am I misinterpreting what I'm seeing ?

I don't see any bad caps in the area of the CPU shown in the picture.
None of the stamped lines on the capacitors appear cracked.

http://i.imgur.com/o8e0Uh.jpg

A popping sound, can be arcing around a high voltage capacitor
in the power supply.
But I replaced the power supply and the problem remained. That rules
out the PS.
A failing power supply, can damage other stuff in a computer.
So assuming just one thing was damaged, might not be it either.

If the computer is working, but is "choppy", well... it's still working.
It means the thing still "computes". And a lot of it has to be
working, for that to be the case.
I left out some details. The computer only stayed on for a few min
and then shut it self down.
If you want an alternative test case, boot a Linux LiveCD, then
use the movie player in there and play some movie content. Is it
still choppy or not ? If the playback is smooth, then there is
something going on with the Windows install.
Since it was a spare machine I took out the hard drive and reformatted
it. I had also removed the mobo because the original thought was to
turn it into a gaming machine for my nephew for Christmas, but he has
another gift in mind over a computer. I was going to just take an old
small spare hard drive and reinstall Windows XP. When I started
assembling it again, I remembered the loud popping sound my nephew
said it made and I decided to just order the same type mobo. That is
when I pulled the fan off and saw what looked like it could have been
smoke.

Something caused a pop and caused the computer to shut down. My
trouble shooting skills are pretty low. Replacing parts is about as
good as I can do. I really thought it was between the mobo and the
PS. I replaced one thing thinking it would be either it or the other.
Now It can be the CPU.

I may still do the Windows install if the CPU just looks like normal
dust to you.

I did notice in the instructions that it does have a " PC Health
Monitoring" feature.
I don't know much about what that could tell, but I wish I had noticed
that before I took everything apart.
Review the BIOS settings, and make sure it's set up right.
I don't change anything in the BIOS I let it use factory settings.
I don't see a reason to be running off and buying stuff, without
at least some hint of what's cooked :) If this was a rush job
(computer must be running by sunset), then by all means, buy
a stack of parts and go to it. But if this is a leisure
activity, run more test cases first and try and narrow it down.

In terms of the picture you've provided, I'd take the PC outside
and blow the dust off it. If there was a carbon trail left on
there somewhere, I'd hope it would remain intact, while the dust
would take a hike. If there is something there you want me to
see, that might improve the contrast ratio.

I've only seen pictures of a few, truly burned PCs. My favorite,
was a plane to plane short around the CPU socket, which causes the
entire area around the CPU to be charred. The VCore regulator
can provide 100 amps at a low voltage, and as long as the
overcurrent doesn't trip on it, it can "cook" a carefully adjusted
short. If the short has just the right resistance value, then
it could sit there for hours cooking, until the PCB develops
a charred appearance.

Other kids of faults, are toroidal coils that are burned so
bad, the core disintegrates when touched. Or, you can have a
MOSFET that is burned black. Those are other kinds of faults,
where the CPU will not POST or boot as a result. Your machine
runs, so by definition, there can't be that kind of damage.

*******

If you want some software to run, try DPCLat, which measures the
service time for Delayed Procedures. It indirectly identifies
if the CPU is not able to service things in a timely manner.

http://www.thesycon.de/deu/latency_check.shtml
I may give that a try too.
That web page has some pictures of the DPCLat window. You should
have low amplitude green blocks, rather than a lot of red spikes.
That program runs in Windows. I don't run that very often (perhaps
once, per system build) as normally it's a check that the BIOS
doesn't have a bug. (A bad BIOS design, can cause some of the
red spikes.)

Paul

Thanks
 
M

Metspitzer

Doubt the CPU is bad, but it's going to be vintage anyway after a few
years. S478? Wrap the CPU in some foam and put it in the bag that
grows with the rest of them. Being able to troubleshoot old stuff
with known working spare parts turns into a privilege at some point as
parts become harder to find;- not sure that was the inverse case on
DDR2 pricing on Ebay when I looked recently. New MB I picked then
also was from the standpoint of longevity -- says on the package it's
built to last 25% longer on MTBW averaged. Fairly priced but not
outstanding in a $200+ scheme of 6 and 8 multicored MBs. Less an ASUS
loyalist and trying crossed-fingers with solid state capacitors these
days.

Since it is just going to be a spare machine that does web surfing,
the original plan was to try to save the memory and CPU and just order
a mobo that would work. I may just order another mobo, memory and
chip because I never have any luck with any spare parts working with
anything else I have. Luckily I don't do this very often.

Thanks
 
P

Paul

On the right side it looks like it could be burned. I don't see too
many CPUs because they are always covered with the fan.
It looks darker than dust to me. I hope I am wrong.
But I replaced the power supply and the problem remained. That rules
out the PS.
I left out some details. The computer only stayed on for a few min
and then shut it self down.
Since it was a spare machine I took out the hard drive and reformatted
it. I had also removed the mobo because the original thought was to
turn it into a gaming machine for my nephew for Christmas, but he has
another gift in mind over a computer. I was going to just take an old
small spare hard drive and reinstall Windows XP. When I started
assembling it again, I remembered the loud popping sound my nephew
said it made and I decided to just order the same type mobo. That is
when I pulled the fan off and saw what looked like it could have been
smoke.

Something caused a pop and caused the computer to shut down. My
trouble shooting skills are pretty low. Replacing parts is about as
good as I can do. I really thought it was between the mobo and the
PS. I replaced one thing thinking it would be either it or the other.
Now It can be the CPU.

I may still do the Windows install if the CPU just looks like normal
dust to you.

I did notice in the instructions that it does have a " PC Health
Monitoring" feature.
I don't know much about what that could tell, but I wish I had noticed
that before I took everything apart.

PC Health is the BIOS hardware monitor. The SuperI/O chip (takes care
of floppy interface, serial ports, parallel port, low speed stuff), it
also has interfaces on it for measuring a few temperatures and voltages.
The CPU temperature is the one you'd be interested in. The CPU temperature
will be displayed in that BIOS screen. It gets updated perhaps once
a second.

You'd switch on the computer, enter the BIOS, go to the PC Health page
in the BIOS, and view the CPU temperature readout. See if the temperature
is gradually rising, until it gets too high and the computer shuts off.
It might get up to around 100C or so, before it's "lights out".

The CPU will have a THERMTRIP signal, which turns off the computer
if the CPU gets too hot. Reasons for the CPU to get too hot are:

1) VCore voltage is incorrect. A defect in the VCore regulator circuit
could do it. Or, on an enthusiast board, setting a "boost" for VCore
too high could do it. On one of the Tomshardware projects, they force
the CPU to draw 200 watts, and stuff gets pretty warm. I don't think
that's the case here. But if something I can't see in the picture,
part of VCore has blown, then that might result in the CPU overheating.
Normally, the CPU would be destroyed by that much of an overvoltage.

2) Heatsink retention levers were loose, and the heatsink
isn't touching the CPU. The thermal paste pattern on the top of
your CPU isn't ideal right now, so when you reassemble the machine,
you can clean off the paste and use fresh thermal paste. You want to
make sure the heatsink is really pressing against the CPU. I've had
one computer, where it was overheating, because the heatsink was
binding on some adjacent metal. It helps to inspect the heatsink from
all four sides, to decide whether it's sitting flat or not.

3) CPU fan stops spinning. The CPU fan speed will be listed in the
PC Health screen as well. Some designs, run the fan at low RPM
(like 300-500 RPM or so). But most of the stuff I have here, is
likely 1800 RPM or higher. Don't forget to plug in the CPU fan
cable, before switching on.

If (2), the thing can turn off in a couple seconds.

If (3), it might take a while for the CPU to heat up the metal
in the heatsink. But without the fan spinning, it may eventually
get so hot, as to trigger THERMTRIP and turn off the PSU.

I tried to look for things that are "burst" in your photo. There are
a few components underneath the retention frame, that I can't see too well.

If an electrolytic actually explodes, the material released looks like
vaporous trails of "soot" floating in the air. So much of it is released,
you'd think it was a black confetti storm. It's pretty gross and messy.
When you take electrical engineering, there's usually at least one
individual who has to see what that looks like, and the rest of us
get to inhale the results :-(

Your "dust" doesn't look exactly the right color, but it's
pretty hard to judge at this distance.

It's also possible, the CPU fan pulled in polluted air, from
some other thing in the machine that popped. Like, a component
on an adjacent video card, or the CPU fan could also have pulled
smoke from the PSU, into the fan blades.

I've spent as long as two hours with a jewelers loupe, or with
a binocular low-power microscope, inspecting printed circuit
boards for defects. For really microscopic faults, that's how
long it takes (try finding a crack in an 0603, when there are
a thousand of them). You have to twist them at an angle sometimes,
to see details off to the side of a component. You can't possibly
hope to capture all the available information with one or two
camera shots. It really takes a pair of eyes right on the scene
to catch every little detail.

I think your VCore may have a few tantalum caps in the area, and
I don't know what color those are when they blow. I've had an
epoxy-dipped tantalum blow in my living room as a kid, and it
bounced off the living room wall like a bullet. There's no place
for gas pressure to go on those. The epoxy makes an air-tight
enclosure around the "sugar cube" of stuff inside. I think it
fractures near the base, and the gas pressure is released. And
of the ones I've had fail, that was the most spectacular.

But your sound was a "pop", and that's either arcing, or
it's an electrolytic (like the rubber bung in the bottom
of the cap, causing the cap to lean like the Leaning Tower
of Pisa). Continue your inspection, and see if some
electrolytic (cylinder) outside of camera range in the
photo, has blown. Pressure is either release on top,
via the stamped pattern in the aluminum shell. Or, the
rubber bung in the bottom of the capacitor, is ejected
from the bottom, and the capacitor leans at an angle
from the vertical.

Not every component failure in a PC, causes it to
stop running. People have bumped SMT components on
motherboards, causing them to detach and fly off, and
the computer continues to run. So not every component
on there, is in the critical path. You might have to lose
just about all the bulk bypass caps, before the computer
would crash for example. And there might be half a dozen
or more of those.

If a cap blows on a VCore circuit, it's relatively easy
to fix. It's when a user ignores something like that, and
"lets it burn for a while", that one of those toroidal
coils or one of the MOSFETs with the three legs on it,
gets burned. And those are harder to find substitutes for.
When a cap fails, it can stress adjacent components that
are part of the same circuit leg, and when those other
components fail, it makes the thing harder to fix.

Paul
 
Ad

Advertisements

F

Flasherly

Since it is just going to be a spare machine that does web surfing,
the original plan was to try to save the memory and CPU and just order
a mobo that would work. I may just order another mobo, memory and
chip because I never have any luck with any spare parts working with
anything else I have. Luckily I don't do this very often.

Thanks

Spare parts should work by cycling them to another MB. Never been for
me as much a problem, but rather the accumulation of old parts that
turn to into a room full of junk. They just get harder to find
matches as the newer stuff leaves them behind. But I don't spend
money on computers as I once did, which makes it a balancing act
between how long I get to run with gear, to get my money's wor6th,
before it's obsolete or broken.

Compgeeks used to be good at that angle but I'm not seeing much now,
like any $17 motherboards w/ free shipping, nor would I recommend them
if issues arise and a working replacement or their money back is
indicated, including shipping, for faulty gear sent in the first
place. Newegg is better than that, only they charge fair enough for
stuff that's not priced for bottom of the barrel. Built a kick-butt
little system from a $17 MSI MB & P4 all shipped free from Compgeeks.
$150 or so when done, but the next one I tried, got a different make
MB that wouldn't hold its BIOS settings and kept switching itself back
to defaults. Had to pay money for a course in hair-pulling. Return
ALL my money - hell, they didn't want to talk to me when I said it was
broken gear in the 1st place.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top