Cause of motherboard failure?

  • Thread starter Thread starter Jem Berkes
  • Start date Start date
J

Jem Berkes

I'm trying to figure out what might have led to the smelly destruction of
an ECS K7SEM running on a server computer. It typically runs 24 hours a
day, and had been doing so for a couple years. Hard drive and fans are OK,
but recently the motherboard started smoking quite badly.

When I opened it up I found that the row of (what I'm pretty sure are)
voltage regulators on the board are cracked. What might have caused this?
No indications that the power supply is acting up, it is now running a
second motherboard and peripherals without problems.
 
I'm trying to figure out what might have led to the smelly destruction of
an ECS K7SEM running on a server computer. It typically runs 24 hours a
day, and had been doing so for a couple years. Hard drive and fans are OK,
but recently the motherboard started smoking quite badly.

When I opened it up I found that the row of (what I'm pretty sure are)
voltage regulators on the board are cracked. What might have caused this?
No indications that the power supply is acting up, it is now running a
second motherboard and peripherals without problems.
Click to expand...

Lelon
Luxon
Tayeh
Jackcon
I,Q
JPCON
Chhsi


If it has any of those potentially defective capacitors,
this may be the problem. That era of board, especially from
ECS, is quite likely to have one of the above cap brands on
it. You might examine them for signs of failure- domed
tops, venting, swollen, residue on top or bottom.

IMHO, a server, especially one running 24/7, should not have
such a low-end board in it.
 
Lelon
Luxon
Tayeh
Jackcon
I,Q
JPCON
Chhsi


If it has any of those potentially defective capacitors,
this may be the problem. That era of board, especially from
ECS, is quite likely to have one of the above cap brands on
it. You might examine them for signs of failure- domed
tops, venting, swollen, residue on top or bottom.
Click to expand...

Yes but it's not the capacitors that were smoking, it was the voltage
regulators that were visibly cracked. Unless a capacitor failure might have
somehow lead to a voltage regulator braeaking?
IMHO, a server, especially one running 24/7, should not have
such a low-end board in it.
Click to expand...

It's now got a nicer ASUS a7v266e board :)
 
Yes but it's not the capacitors that were smoking, it was the voltage
regulators that were visibly cracked. Unless a capacitor failure might have
somehow lead to a voltage regulator braeaking?
Click to expand...

Since PSU (and CPU?) is claimed to work properly still, that
seems the next most likely problem, unless the controller
failed but I suspect it's demise would have opposite effect,
no power, not overheating.

FWIW, I often put heatsinks on the regulators, but it's not
needed when using any CPU qualified for ("supported") the
board, rather it's a measure taken when overclocking.

I suppose it's possible the regulators were made with a bad
batch of epoxy, but i've never heard of this happening on PC
board regulators.
 
Jem said:
Yes but it's not the capacitors that were smoking, it was the
voltage regulators that were visibly cracked. Unless a capacitor
failure might have somehow lead to a voltage regulator braeaking?
Click to expand...

Certainly could. A sufficiently leaky cap draws excess current,
thus overloading the regulators, etc. etc.
 
kony said:
FWIW, I often put heatsinks on the regulators, but it's not
needed when using any CPU qualified for ("supported") the
board, rather it's a measure taken when overclocking.
Click to expand...

I did this once myself but it didn't make any difference with the stability
of the board. I'm inclined to think the stability problems were due to
something else. I used normal epoxy resin (Araldite) to attach them. I made
them by cuting 12mm squares of an old P1-era passive HS I had that had nine
12mm spikes on each piece.

I'm currently using two MS-6163 Pro's with Slot-T's and Tualatins. The FETs
get hot so with both boards I have a fan blowing directly onto them. I was
wondering about putting HS on them as well. My problem is, I'm not sure that
the epoxy wasn't acting as an insulator or if it was in fact beneficial
having the 'sinks on the FETs. I used a very thin smear of epoxy and made
sure it was spread evenly, like I would thermal compound on a CPU.

What do you think Dave? Is Araldite good for this type of application? I
can't afford proper thermal epoxy (sad but true) so am considering using
normal epoxy. Your opinion please kind sir.

Cheers,
 
Yes but it's not the capacitors that were smoking, it was the
Certainly could. A sufficiently leaky cap draws excess current,
thus overloading the regulators, etc. etc.
Click to expand...

Yes, now I think that this is what happened.

Taking a closer look at the caps I noticed that the ones by the voltage
regulators have domed (raised) tops, i.e. something blew up inside! Caps
are generally used to absorb power fluctuations in voltage regulation, so
failed caps could certainly shift power handling burden to the regulators
which then cracked I guess.

Boo for ECS.
 
Sounds like you have an excessive heat problem. Add heatsinks if you can
bit do not put expoxy between the sink and the chip. Try and get some
thermal compound in there if at all possible or you gonna be ooking at
this problem again. If the parts are too hot too hold a finger on without
burning they are probably too hot.
 
I did this once myself but it didn't make any difference with the stability
of the board. I'm inclined to think the stability problems were due to
something else. I used normal epoxy resin (Araldite) to attach them. I made
them by cuting 12mm squares of an old P1-era passive HS I had that had nine
12mm spikes on each piece.
Click to expand...

It may not improve stability of the board enough to notice,
if at all, but what it may do is keep the heat uniform and
lower to reduce cracking or overheat-shutdown of the
regulators. I don't know anything about Araldite's temp
range though, ideally the epoxy should be rated at least
past 100C.

I'm currently using two MS-6163 Pro's with Slot-T's and Tualatins. The FETs
get hot so with both boards I have a fan blowing directly onto them. I was
wondering about putting HS on them as well. My problem is, I'm not sure that
the epoxy wasn't acting as an insulator or if it was in fact beneficial
having the 'sinks on the FETs. I used a very thin smear of epoxy and made
sure it was spread evenly, like I would thermal compound on a CPU.

What do you think Dave? Is Araldite good for this type of application? I
can't afford proper thermal epoxy (sad but true) so am considering using
normal epoxy. Your opinion please kind sir.
Click to expand...

If you can put your finger on it and not lose some skin,
it's probably cool enough. I can't vouch for any particular
epoxy except arctic alumina/silver, and some other
less-known brand I forget the name of at the moment, but it
was also spec'd for thermal transfer. I would be hesitant
to use epoxy that wasn't at least rated up to 80C, or one
with no thermal rating at all. If it gets too hot it might
even get very brittle and the 'sink could fall off after
many thermal cycles.

As with any 'sink, how well it works may depend on the
thickness of the bond. If the bond is very thin it should
help some even if the epoxy isn't particularly good at
transfer, but to get the very best bond it is often
necessary to put one 'sink per regulator, not bridged across
two or more (unfortunately even IF the regulators are same
thickness, there may be irregularities in mounting that
result in different height).
 
Sounds like you have an excessive heat problem. Add heatsinks if you can
bit do not put expoxy between the sink and the chip. Try and get some
thermal compound in there if at all possible or you gonna be ooking at
this problem again. If the parts are too hot too hold a finger on without
burning they are probably too hot.
Click to expand...


The capacitor failure is likely to be from an instable
electrolyte that broke down regardless of the temp. At
least, that was often seen with potentially defective
brands.
 
kony said:
It may not improve stability of the board enough to notice,
if at all, but what it may do is keep the heat uniform and
lower to reduce cracking or overheat-shutdown of the
regulators. I don't know anything about Araldite's temp
range though, ideally the epoxy should be rated at least
past 100C.



If you can put your finger on it and not lose some skin,
it's probably cool enough. I can't vouch for any particular
epoxy except arctic alumina/silver, and some other
less-known brand I forget the name of at the moment, but it
was also spec'd for thermal transfer. I would be hesitant
to use epoxy that wasn't at least rated up to 80C, or one
with no thermal rating at all. If it gets too hot it might
even get very brittle and the 'sink could fall off after
many thermal cycles.

As with any 'sink, how well it works may depend on the
thickness of the bond. If the bond is very thin it should
help some even if the epoxy isn't particularly good at
transfer, but to get the very best bond it is often
necessary to put one 'sink per regulator, not bridged across
two or more (unfortunately even IF the regulators are same
thickness, there may be irregularities in mounting that
result in different height).
Click to expand...

Thanks for your imput.
 
Thanks for your imput.
Click to expand...

One other thing I failed to mention is that often I see
people putting biggest 'sink possible on the 'fet, while
it's primary 'sink path is to the copper on the board, so
choosing a 'sink that doesn't block a lot of the airflow
from the board surface is better than one that does.
 
Back
Top