How reliable are motherboard monitors?

After working fine for over a year, this SOYO KT-600 & AMD 3400
(Sempron?) started rebooting spontaneously. Once in a while. When I
swapped in 2x1GB memory in place of 2x512MB, it would reboot constantly,
until I backed off the overclock, back to stock ratings.

The motherboard monitor shows this:

http://www.cycleorings.com/soyovolts.jpg

Sometimes the +12V drops to 11.51. I'm guessing the power supply is on
its last legs, so I bought a Corsair 550W unit (hasn't arrived yet).

The question is, how reliable is the information these motherboards put
out? The CPU temp seems to stay at 187F no matter what. The +3.3 seems
too low to even function.

Robert Barr wrote:
After working fine for over a year, this SOYO KT-600 & AMD 3400
(Sempron?) started rebooting spontaneously. Once in a while. When I
swapped in 2x1GB memory in place of 2x512MB, it would reboot constantly,
until I backed off the overclock, back to stock ratings.

The motherboard monitor shows this:

http://www.cycleorings.com/soyovolts.jpg

Sometimes the +12V drops to 11.51. I'm guessing the power supply is on
its last legs, so I bought a Corsair 550W unit (hasn't arrived yet).

The question is, how reliable is the information these motherboards put
out? The CPU temp seems to stay at 187F no matter what. The +3.3 seems
too low to even function.

The labels should probably be switched on the AGP item and
the the +3.3V item. That would make your 3.3V, at a reading
of 3.13V, right on the edge of -5%. The 12V is still passing,
as -5% would yield 11.4V and you've got a reading of 11.64V.

The only way to really know about these, is to use a multimeter.
My multimeter isn't an expensive one, and is only accurate to
about 1.5% on some ranges. Yet, it reads my power supply rail
voltages as being very close to nominal, and not matching
the hardware monitor values. My hardware monitor always
shows my +12V on the high side - the meter shows the value
is perfectly normal.

So the only benefit of the hardware monitor, is as a
reminder to buy a multimeter :-)

Your temp of 187F is also a bit high (86C). If the temperature
is diode based (diode inside CPU), that would be harder to make
sense of with a multimeter. It is pretty hard to fit anything
that will give an accurate CPU temp, but a lack of stability
on your processor should tell you that the overheat is for real.

Take off the heatsink, clean off the interface material currently
present, then apply some fresh thermal paste. Then take another
reading. If the interface is currently "dry", that would
explain it.

http://www.newegg.com/Product/Produc...82E16835100009

Paul

Take off the heatsink, clean off the interface material currently
present, then apply some fresh thermal paste. Then take another
reading. If the interface is currently "dry", that would
explain it.

http://www.newegg.com/Product/Produc...82E16835100009

Paul

Thanks for the reply.

When I did that (clean surfaces and reapply heatsink compound) the
system wouldn't POST anymore. I assumed (ooh, bad word!) that I didn't
exercise enough ESD care and killed the CPU. So it's off to ebay to
pick up a couple of CPU's (one 3300+ and one mobile, for hobby purposes).

Well, it wouldn't POST with either of the 'new' CPU's either. The only
guess remaining was a dead motherboard. Removed same from case and
tried it breadboarded with a new PSU. No POST with any of the CPU's.

(I'm now able to install these blindfolded).

Just for yucks, before I put the motherboard in the trash, I pull a
memory stick. No change. I pull the 2nd, and reinsert it in the other
slot.

It POSTs just fine.

WTF?

I install the second memory stick. Still POSTS fine. Reinstall the
original 'dead' CPU. POSTs just fine.

Reinstall motherboard into case. Notice that of the 5 screw mounting
spots, only 4 contact the motherboard tray. Hmmm? Install these 4
screws. 5th screwhole is now about 3/8" above the mounting hole/pylon.
WTF?

Hooked it up and it POSTs fine, runs fine. Figure I'll just leave that
4th screw out -- everything seems to be OK without it.

Apparently the motherboard is warped. I'm guessing the warpage is what
caused some poor connection somewhere. Who knows?

Last step: Order new motherboard, CPU & memory, all one generation
newer, and sell all the pieces on ebay.

Robert Barr wrote:

Take off the heatsink, clean off the interface material currently
present, then apply some fresh thermal paste. Then take another
reading. If the interface is currently "dry", that would
explain it.

http://www.newegg.com/Product/Produc...82E16835100009

Paul

Thanks for the reply.

When I did that (clean surfaces and reapply heatsink compound) the
system wouldn't POST anymore. I assumed (ooh, bad word!) that I didn't
exercise enough ESD care and killed the CPU. So it's off to ebay to
pick up a couple of CPU's (one 3300+ and one mobile, for hobby purposes).

Well, it wouldn't POST with either of the 'new' CPU's either. The only
guess remaining was a dead motherboard. Removed same from case and
tried it breadboarded with a new PSU. No POST with any of the CPU's.

(I'm now able to install these blindfolded).

Just for yucks, before I put the motherboard in the trash, I pull a
memory stick. No change. I pull the 2nd, and reinsert it in the other
slot.

It POSTs just fine.

WTF?

I install the second memory stick. Still POSTS fine. Reinstall the
original 'dead' CPU. POSTs just fine.

Reinstall motherboard into case. Notice that of the 5 screw mounting
spots, only 4 contact the motherboard tray. Hmmm? Install these 4
screws. 5th screwhole is now about 3/8" above the mounting hole/pylon.
WTF?

Hooked it up and it POSTs fine, runs fine. Figure I'll just leave that
4th screw out -- everything seems to be OK without it.

Apparently the motherboard is warped. I'm guessing the warpage is what
caused some poor connection somewhere. Who knows?

Last step: Order new motherboard, CPU & memory, all one generation
newer, and sell all the pieces on ebay.

Sometimes, a failure to POST can be caused by having a standoff installed,
that doesn't have a matching plated hole on the motherboard. There was
an Nforce2 board, where if you leave a standoff installed on the computer
case, it shorts out one channel of sound on Lineout. On other motherboards,
there have been cases where a power supply rail gets shorted. So standard advice,
is to only install as many standoffs, as there are plated holes. A full sized
motherboard might have 9 holes, and since some are narrower than standard,
those have 6 holes. Install standoffs to match the holes.

The plated holes are at ground potential. The standoffs are at ground potential.
That is why it is safe for a metal standoff to touch a plated hole.

PCB warpage is bad, in the sense that when the motherboard is flattened against
the tray, the solder joints will be placed under stress. It is surprising that
computer PCBs take the abuse that they do, without more trouble.

Paul