P4P800 on caught on fire?

[ron]

9 Months ago I helped my younger cousin build a computer. Everything
went together just fine (I have built several machines AMD & Intel
wihtout issue)

Antec case & power
Asus P4P800
P4 2.8
1GB of Memory (Kingston I think)
Saphire ATI 9800 Pro
Generic DVD/CD drive
TEAC 3.5 disk drive
Western Digital HD

Nothing was overclocked...all at default/stock settings.

Today, I get a call from his uncle saying that the computer caught on
fire last night. Not raging flames engulfing the case, but that it
was smoking and when he looked inside the case there were flames. He
stated that when he unplugged the computer, the flames stopped (isnt
the power supply designed to shut-off in this situation?)

I have never had a computer catch on fire....any ideas on where to
start my investigation? According to my uncle, there are burn marks
on the motherboard near the center. I will be swinging by later
tonight to see exactly what part of the motherboard he is referring
to.

Anyone had a similar problem or have any ideas on this one?

TIA
Ron

 

[Paul]

9 Months ago I helped my younger cousin build a computer. Everything
went together just fine (I have built several machines AMD & Intel
wihtout issue)

Antec case & power
Asus P4P800
P4 2.8
1GB of Memory (Kingston I think)
Saphire ATI 9800 Pro
Generic DVD/CD drive
TEAC 3.5 disk drive
Western Digital HD

Nothing was overclocked...all at default/stock settings.

Today, I get a call from his uncle saying that the computer caught on
fire last night. Not raging flames engulfing the case, but that it
was smoking and when he looked inside the case there were flames. He
stated that when he unplugged the computer, the flames stopped (isnt
the power supply designed to shut-off in this situation?)

I have never had a computer catch on fire....any ideas on where to
start my investigation? According to my uncle, there are burn marks
on the motherboard near the center. I will be swinging by later
tonight to see exactly what part of the motherboard he is referring
to.

Anyone had a similar problem or have any ideas on this one?

TIA
Ron

The only thing I see near the center of a picture of that
motherboard, are some linear regulators based on MOSFETs.
Asus does some of their onboard regulation, with quad op
amps and MOSFETs, and if that is what failed, then some chip
they are connected to must have failed, overloading the
circuit. For example, you have an AGP slot near there
as well, and perhaps something on the video card shorted
and caused the power dissipation in the MOSFET to go up.

I haven't traced the circuits down for those, but it is
quite likely those linear regulatioo circuits are not
protected against overload. Ondinary linear regulator
circuits (three or five pin regulators), are protected
against overheat, but when Asus builds their own onboard
regulation using low integration devices, it costs them
money to protect against every fault, so those MOSFETs
could fry.

If anything was to fail on the motherboard, I would have
guessed "Vcore circuit", and that is in the upper left hand
corner of the motherboard, between the I/O connectors and
the processor socket. The Vcore regulator chips usually can
sense an overcurrent condition and shut down, but that is only
on their output side. There will be some fault conditions
(like a MOSFET failure), where power coming from +12V could
be what burns up the circuit. But that would be in the upper
left hand corner of the board.

When it comes to PSUs, they have an inlet fuse (for power
coming from the wall), but their overvoltage and overcurrent
settings on output, even if they exist, are so sloppy as to
be useless. According to the ATX spec for power supplies,
no flames are supposed to shoot out of the PSU in the event
of a fault, but like the linear regulator circuit on the
motherboard, there are going to be some faults (where the
fault doesn't draw enough current to trip a protection device)
that will cause a part of the PSU to overheat and smoke/burn.

And for all you case modders out there, this is why you don't
want a computer with a plexiglas window, or a computer that is
completely made of plastic, as if the computer catches fire,
there is a risk that a non-steel case could catch fire as
well. A plain, beige colored steel case may be boring, but
is more likely to at least contain the flames, if something
like this happens. For most users, the smoke damage will be
the most annoying part. And I have read accounts of flames
shooting out of the fan hole of the PSU, so that does happen
occasionally as well (by specification, it is not supposed
to!).

Obviously Ron, be careful with any components you recycle
from that computer. For example, if the video card has a
short on it, you could fry whatever motherboard you plug
the video card into next. As you are planning to do, try
to trace down what circuit has failed, as that will aid
in determining what parts of the computer are safe to
keep. For example, if the Vcore circuit in the upper left
hand corner is intact, and no signs of damage, I would be
tempted to reuse the processor. For the video card, pull
it and inspect pins and copper traces, for any sign of a
short. If the source of the failure was actually a MOSFET
failing short on the motherboard, whatever part that got
power from it, could now be damaged too.

Paul

 

[DaveW]

One possible scenario that occurs to me is that the CPU's fan failed and the
CPU really overheated.

 

[ron]

I had a chance to do a brief inspection of the motherboard tonight.
The fire was based around USB coneectors 5&6 which were connected to
the front USB ports on the Antec Case. These ports were not used
according to my cousin, but I believe they worked when I finished
builing it (I did have problems with the front audio connectors
because the Sonata case doesnt use the intel standard...I left the
wiring for the audio unplugged).

So the question remains....how does a computer catch on fire at the
USB pins/connector? I will double check to make sure that the USB
wires were connected properly, but I doubt that's the issue (it would
have either not worked and/or I would have seen a problem long before
6 months (I originally thought it was 9 months ago, but my receipts
tell me otherwise).

Any theories? I have heard of Antec mis-wiring the front ports on the
Sonata case, but would this casue a fire....I thought it would just
make things not work. If the power/ground wires got connected
incorrectly, wouldn't that cause a problem up front....not 6 months
down the road?

Thanks for the input.

Ron

 

[Leythos]

So the question remains....how does a computer catch on fire at the
USB pins/connector?

I would guess that you shorted the power pins through something being
connected improperly.

 

[ron]

Dave:

Based on where the burn marks are.....I'd have to say no. Also, I
think the ASUS board will shut off if CPU gets too hot.

 

[WoofWoof]

I am not an expert (IANAE ?) but surely these things are designed so
that there isn't sufficient power available at the USB connectors that
a short would start a fire. How would they get UL approval (and
others) if that were so?

In fact the power available at a USB connector for the external device
is generally pretty limited, such that many devices need an external
supply. It's possible that MB manufacturers have been increasing it
lately because of that, but surely not to that extent?

 

[Leythos]

I am not an expert (IANAE ?) but surely these things are designed so
that there isn't sufficient power available at the USB connectors that
a short would start a fire. How would they get UL approval (and
others) if that were so?

In fact the power available at a USB connector for the external device
is generally pretty limited, such that many devices need an external
supply. It's possible that MB manufacturers have been increasing it
lately because of that, but surely not to that extent?

The output can provide "a minimum of 500ma" according to this article:
http://www.micrel.com/product-info/products/mic2010.shtml

..5 Amps is enough to melt small computer wires, and other things, if
directly shorted, without proper fusing.

Here is another from the same site that shows "short circuit current at
1.25 Amps max - well higher than needed to fry 28ga conductors and some
foil traces.

http://www.micrel.com/product-info/products/mic2525.shtml

 

[John]

Hi, I read an article somewhere, I can't remember were, about graphics cards
becoming unseated and then catching fire.

Thanks, John

 

[WoofWoof]

The output can provide "a minimum of 500ma" according to this article:
http://www.micrel.com/product-info/products/mic2010.shtml

.5 Amps is enough to melt small computer wires, and other things, if
directly shorted, without proper fusing.

Here is another from the same site that shows "short circuit current at
1.25 Amps max - well higher than needed to fry 28ga conductors and some
foil traces.

http://www.micrel.com/product-info/products/mic2525.shtml

Interesting .... then, unless things have changed significantly in the
last few years, UL would require a fuse in those power lines if
brought to an external connector (such as a USB connector).

 

[Cathy & Frank]

I had a similar problem once, a long time ago
plugged a serial port on the joystick port and the cable went on fire when
turned computer on(voltage to serial is bad)...

maybe a similar problem with the USB connector could happen when USB port
was plugged...

 

[Leythos]

Interesting .... then, unless things have changed significantly in the
last few years, UL would require a fuse in those power lines if
brought to an external connector (such as a USB connector).

So, are you suggesting that the board was fully UL listed - a board not
made in the USA?

Are you also suggesting, after looking at the board schematic, that
there was a fuse?

Having designed board and many circuits over my years, it seems highly
likely that there was a short that caused the fire. Short (pun intended)
of a electrical fault there is little in a system that can cause a
fire.

 

[Paul]

I had a chance to do a brief inspection of the motherboard tonight.
The fire was based around USB coneectors 5&6 which were connected to
the front USB ports on the Antec Case. These ports were not used
according to my cousin, but I believe they worked when I finished
builing it (I did have problems with the front audio connectors
because the Sonata case doesnt use the intel standard...I left the
wiring for the audio unplugged).

So the question remains....how does a computer catch on fire at the
USB pins/connector? I will double check to make sure that the USB
wires were connected properly, but I doubt that's the issue (it would
have either not worked and/or I would have seen a problem long before
6 months (I originally thought it was 9 months ago, but my receipts
tell me otherwise).

Any theories? I have heard of Antec mis-wiring the front ports on the
Sonata case, but would this casue a fire....I thought it would just
make things not work. If the power/ground wires got connected
incorrectly, wouldn't that cause a problem up front....not 6 months
down the road?

Thanks for the input.

Ron

The USB is protected by a polyfuse. If you look at the picture
in the manual, there are the two 2x5 headers, and there is a fuse
next to each USB header. The 2x3 block of pins (USBPW56, USBPW78)
selects between +5V and +5VSB. The fuse comes after the power
selection and before the power reaches the header.

The fuse is supposed to automatically recover. The material
inside the fuse undergoes a phase change of some sort, and
changes from a conducting to non-conducting state based on
temperature. When it cools down enough, it starts conducting
again. The temperature the fuse opens at, will be low enough
that is won't catch fire.

That is, unless there is a solder splash underneath it, and
the fuse was shorted out by solder. But, if you get out a
soldering iron, and pull the fuse off the board, you could
heat the solder splash underneath it, and remove the evidence
as well. You could try using an ohmmeter, but it would be
hard to tell the difference between the normal Polyfuse
resistance, and the resistance of a solder splash.

I would say this takes two events - a fault around the fuse
area, and your cousin jamming a USB connector into the
front of the case, backwards. The +5V and GND on the USB
connector are the end pins, so reversing the connector
would reverse power inside the USB device. It could
possibly conduct a lot of current that way. If the fuse
is bridged with solder, then it won't open, so a lot of current
flows in the copper trace leading from the 2x5 header, to the
2x3 USBPWxx block. If the USB was powered by +5V, there could
be a fat copper trace inside the board, which would run cooler
than the short trace between the 2x5 and the 2x3 header block.
So, maybe a trace burned up, due to the fuse not doing its job.
I would look for damage on the USB connectors themselves, to
see if a USB device was connected at the time.

HTH,
Paul

 

[WoofWoof]

So, are you suggesting that the board was fully UL listed - a board not
made in the USA?

I was making the assumption that it should be UL listed. (Whether or
not a product is made in the USA, of course, is not what determines
whether UL certification (or equivalent) is required ... but whether
it is sold there). It possibly wouldn't require that, actually, to be
simply retailed as a MB on its own, but any significant OEM using it
in a product would need to get his end product certified which would
either mean certifying at that product level or using subcomponents
(such as the MB) which are already certified. Since the OEM in that
case has little control over the MB, it's generally easier for them to
insist that the MB supplier get the certification.

Of course, if no OEM's are actually using the board, the point is
moot.

(This whole thing struck a chord because - a good few years ago now -
the company I then worked for was forced to include a fuse in the 5V
line to the keyboard of a workstation they were designing ... unique
design, not PC based. The power available was minor and the
requirement was made known rather late in the schedule causing some
considerable embarassment for all. I presume that the UL requirements
haven't changed in this respect).

Are you also suggesting, after looking at the board schematic, that
there was a fuse?

No, rather the opposite actually. I don't have a schematic anyway and
I assume (since it in fact did catch fire) that there was no fuse. I
was just pointing out that in my experience UL, had they examined the
design, should have required such fuse.

Which perhaps is another way of saying that these (and presumably
other manufacturers') MB's are not UL certified. A case like this
makes you wonder whether it should be required. Scary. It would drive
up costs of course.

I think we've probably hijacked the OP's thread for long enough -)

 

[Leythos]

Which perhaps is another way of saying that these (and presumably
other manufacturers') MB's are not UL certified. A case like this
makes you wonder whether it should be required. Scary. It would drive
up costs of course.

I can remember the days when they keyboard connectors were not fused,
was a PITA when you had a bad keyboard - most people can desolder and
replace the surface mount devices without damaging the boards.

I don't know if the board is UL listed either, but I've seen a lot of
devices produced outside the US (and many inside) that are used in
varied applications that are not UL listed and are not fused.

Later