Power supply heat problem?

[-]

My friends' PC has MBM5 installed and the power supply thermometer is
showing peaks of 70-76C from time to time lasting for around a minute, which
is classed as a warning in the program. Is this really "dangerous" or just
problematic in the short / medium / long term ie being symptomatic of wear
and tear etc. The PSU is unknown for now and is about 2 years old. The fan
is spinning freely, but is blowing out some not overly warm air. I know its
best to keep cool, but with peaks of these temperatures, I can't help
thinking its best not to leave the machine unattended! They have the
automatic shutdown linked in to that SHDN program for real unattended
"emergencies" over 76C, which has proved to work very well.

Is this a considerable problem - should we keep a close eye on it for the
next few days or is this relatively normal? The PC is a 1.3 Athlon / 512MB
RAM, 60G HD with 2 ROMs, floppy with poor ventilation design in the case.

Thanks for any advice.

 

[philo]

My friends' PC has MBM5 installed and the power supply thermometer is
showing peaks of 70-76C from time to time lasting for around a minute, which
is classed as a warning in the program. Is this really "dangerous" or just
problematic in the short / medium / long term ie being symptomatic of wear
and tear etc. The PSU is unknown for now and is about 2 years old. The fan
is spinning freely, but is blowing out some not overly warm air. I know its
best to keep cool, but with peaks of these temperatures, I can't help
thinking its best not to leave the machine unattended! They have the
automatic shutdown linked in to that SHDN program for real unattended
"emergencies" over 76C, which has proved to work very well.

Is this a considerable problem - should we keep a close eye on it for the
next few days or is this relatively normal? The PC is a 1.3 Athlon / 512MB
RAM, 60G HD with 2 ROMs, floppy with poor ventilation design in the case.

Thanks for any advice.

76C is pretty hot
i'd add some more ventilation

 

[kony]

My friends' PC has MBM5 installed and the power supply thermometer is
showing peaks of 70-76C from time to time lasting for around a minute, which
is classed as a warning in the program. Is this really "dangerous" or just
problematic in the short / medium / long term ie being symptomatic of wear
and tear etc. The PSU is unknown for now and is about 2 years old. The fan
is spinning freely, but is blowing out some not overly warm air. I know its
best to keep cool, but with peaks of these temperatures, I can't help
thinking its best not to leave the machine unattended! They have the
automatic shutdown linked in to that SHDN program for real unattended
"emergencies" over 76C, which has proved to work very well.

Is this a considerable problem - should we keep a close eye on it for the
next few days or is this relatively normal? The PC is a 1.3 Athlon / 512MB
RAM, 60G HD with 2 ROMs, floppy with poor ventilation design in the case.

Thanks for any advice.

With air blowing, a fan turning, it's very unlikely that 70C reading
is accurate. If the power supply is even close to 70C, the casing on
it will feel very warm. You don't even have to open the system to
feel that, just put your hand on it where it screws to the case on the
back.

Are you sure the power supply even has a temp sensor lead going to it?
Most don't. I suspect the MBM installation isn't set up properly for
the system... this is the problem with a generic-type monitoring
utility, you MUST configure every last item specifically for the
system it's used in, including temperature offset... even if the power
supply has a temp sensor lead going to a motherboard temp sensor pins,
the offset must be correctly calibrated for the correct reading to
result.

Odds are pretty good you can just ignore it, though if the case has
poor ventilation as you put it, that should be rectified regardless of
the MBM's incorrect power supply reading.


Dave

 

[Lane Lewis]

My friends' PC has MBM5 installed and the power supply thermometer is
showing peaks of 70-76C from time to time lasting for around a minute, which
is classed as a warning in the program. Is this really "dangerous" or just
problematic in the short / medium / long term ie being symptomatic of wear
and tear etc. The PSU is unknown for now and is about 2 years old. The fan
is spinning freely, but is blowing out some not overly warm air. I know its
best to keep cool, but with peaks of these temperatures, I can't help
thinking its best not to leave the machine unattended! They have the
automatic shutdown linked in to that SHDN program for real unattended
"emergencies" over 76C, which has proved to work very well.

Is this a considerable problem - should we keep a close eye on it for the
next few days or is this relatively normal? The PC is a 1.3 Athlon / 512MB
RAM, 60G HD with 2 ROMs, floppy with poor ventilation design in the case.

Thanks for any advice.

Its not normal but those sensors in the motherboard are not the most
reliable.

Check the voltage going to the computer and make sure the PSU fan is well
oiled if a sleeve type. In fact I would replace it to be sure. Make sure all
connections are secure and not hot.

Make sure MBM is configured properly.

Mine reads 35C which would mean that yours is probably on fire.

Lane

 

[-]

If the power supply is even close to 70C, the casing on

it will feel very warm. You don't even have to open the system to
feel that, just put your hand on it where it screws to the case on the
back.

Its warmish, not boiling hot or anything though.

Are you sure the power supply even has a temp sensor lead going to it?

Haven't a clue at the moment - its my friends machine.

I suspect the MBM installation isn't set up properly for
the system... this is the problem with a generic-type monitoring
utility, you MUST configure every last item specifically for the
system it's used in, including temperature offset... even if the power
supply has a temp sensor lead going to a motherboard temp sensor pins,
the offset must be correctly calibrated for the correct reading to
result.

Hmm, how do you calculate the offset?
The MBM setup has been configured, although as I'm not 100% sure on what's
inside, there's no telling if its correct. However, the power supply
temperatures seem to be verified by Sandra and Aida32 - unless they all
share the same (possibly erroneous) detection method.

Odds are pretty good you can just ignore it

Well the system is stable I guess, but its just a little disquieting to see
those kind of temps.

Thanks.

 

[-]

i'd add some more ventilation

I agree - although how do you go about adding ventilation when there's only
one hole in the back of the case (where the chassis fan is blowing out),
aside from the PSU fan outlet, when the rest of the case is that off-beige
painted coloured metal stuff? Its just not pratical and its not time for an
upgrade yet!

 

[kony]

Its warmish, not boiling hot or anything though.

Take care of the other problem you mentioned, the case ventilation in
general, then see if anything changes, the reported temp or case temp.
The casing may be mildly warm but that' s a far cry from 70C, would
probably have it's temp sensor in contact with a heatsink to read that
high. Many are almost in contact with a heatsink but have an
aluminumn shell around them, some silicone or other epoxy layer gluing
them down or just an air-gap. Regardless, a reasonable air movement
is all that's needed. If the fan is working properly then just focus
on air intake, that the system has adequate air intake
holes/vents/gaps/etc.

Haven't a clue at the moment - its my friends machine.

It is an important detail.

Hmm, how do you calculate the offset?

You don't, unless you have an accurate temperature sensor equipment or
have a different software configured by the motherboard manufacturer
to report the correct temps, then you could compare it's readings to
those from MBM in the same system state... you'd be assuming it's at
the same temp and adjusting MBM accordingly.

In those situations you might also check the BIOS Health page, if it
reports temp, though on some systems the BIOS temp for items like the
CPU may be higher than in Windows due to power-conservation measures
supported by the OS, like idle-HALT cooling of CPU.

The MBM setup has been configured, although as I'm not 100% sure on what's
inside, there's no telling if its correct. However, the power supply
temperatures seem to be verified by Sandra and Aida32 - unless they all
share the same (possibly erroneous) detection method.

It is important to determine where this reading is coming from, if the
power supply even has a temp sensor being read by the motherboard.

Some motherboards have chips with the ability to take more
temperatures than sensors to report the temps, so one or two temp
readings may represent the temps of a monitored item but the rest may
be ghost numbers, arbitrary based on the level of the signal internal
to the chip. With no temp sensor to pull the level down it might
indeed read high.

Well the system is stable I guess, but its just a little disquieting to see
those kind of temps.

I would certainly investigate further though, just to be sure it isn't
a cooling problem. You already mentioned the case ventilation, as did
I above... that should be fixed if it actually is a problem (check the
other temps and touch-test some parts) even if the system is stable,
so long term heat stress doesn't shorten the system's lifespan.


Dave

 

[-]

It is important to determine where this reading is coming from, if the

power supply even has a temp sensor being read by the motherboard.
Agreed!

Some motherboards have chips with the ability to take more
temperatures than sensors to report the temps, so one or two temp
readings may represent the temps of a monitored item but the rest may
be ghost numbers, arbitrary based on the level of the signal internal
to the chip. With no temp sensor to pull the level down it might
indeed read high.

I see...

I would certainly investigate further though, just to be sure it isn't
a cooling problem. You already mentioned the case ventilation, as did
I above... that should be fixed if it actually is a problem (check the
other temps and touch-test some parts) even if the system is stable,
so long term heat stress doesn't shorten the system's lifespan.

CPU hovers between 46-55C (its an Athlon, so that's fine, but these temps
I've no doubt add to the overall case temp, which is being registered at
around a stable 34C).

I guess you are right about sorting this out, even just to extend the
longevity of the components in the long run. I'll have to investigate easy
ways to ventilate these crappy metal cases we get given as standard in the
UK (ie 1 outlet at the back, no other inlets / outlets except for the PSU
fan).

Thanks for the insights.

 

[*Vanguard*]

I agree - although how do you go about adding ventilation when
there's only one hole in the back of the case (where the chassis fan
is blowing out), aside from the PSU fan outlet, when the rest of the
case is that off-beige painted coloured metal stuff? Its just not
pratical and its not time for an upgrade yet!

The power supply generates heat. So do the hard drives, memory, CPU,
cards, etc., that are inside the case but not inside the power supply.
By having the warm air from inside the case also go through the power
supply then you preheat the air that is trying to also cool off the
power supply.

Do you have any rear-side ventilation holes/grills where you could mount
an outward blowing fan so the case air can get expelled out without it
all having to go through the power supply? Do you have any slots open
where you could install a slot fan?

Does the power supply have 2 fans? That is, besides the backside fan
blowing out, is there a underside fan sucking air into the power supply?
Does it spin? Is it thermally controlled or always on? Is the backside
fan thermally controlled? For many 2-fan power supplies, the underside
one always spins and the backside one is thermally controlled. Maybe
you're sucking so much power out of the power supply that it is normally
generating a lot of heat but with both fans spinning then it gets cooled
down okay. But if it thinks the temp has lowered then maybe the
backside thermally controlled fan is slowing down, the heat rises, and
then the backside fan kicks in again. How many watts is your power
supply rated for, any how much is getting used?

 

[Timothy Daniels]

I agree - although how do you go about adding ventilation when there's only
one hole in the back of the case (where the chassis fan is blowing out),
aside from the PSU fan outlet, when the rest of the case is that off-beige
painted coloured metal stuff? Its just not pratical and its not time for an
upgrade yet!

One easy way to increase ventilation is to remove the grills that cover
the exhaust fans. The wire grills can be removed with just the removal
of a few screws, and the ones that are part of the case can be snipped
away with a pair of electricians' wire cutters. The air resistance that they
impose can be felt quite easily, and their removal allows greater airflow.
Don't worry about losing finger tips in them because the angle of the
blades of the fans push fingers out instead of snipping them off. (But,
YMMV.)

*TimDaniels*

 

[Guest]

My friends' PC has MBM5 installed and the power supply thermometer
is showing peaks of 70-76C from time to time lasting for around a
minute, which is classed as a warning in the program. Is this
really "dangerous" or just problematic in the short / medium /
long term ie being symptomatic of wear and tear etc. The PSU is
unknown for now and is about 2 years old. The fan is spinning
freely, but is blowing out some not overly warm air. I know its
best to keep cool, but with peaks of these temperatures,

How hot is this "overly warm" air, and what makes you think it's
overly warm?
Never rely on feel alone because people differ greatly in temperature
perception, and I've known of people refer to even 120F as "scorching
hot."
How hot is the top of the power supply? If often reaches 110-120F
(about 45-50C) under normal operation with a fast CPU and video card.

Power supplies don't vary quickly in temperature, except once, just
before the protection circuitry shuts them down or they fail. I
suspect that MBM5 isn't configured right for the monitoring hardware,
but the temperature sensor inside the supply could be touching one of
the large heatsinks or semiconductors (transistors, diodes), which
often reach 70C during normal operation. But do NOT try to measure
the heatsinks because many internal components have high voltage
riding on them, including the heatsinks in some designs. Actually
it's fairly safe if you unplug the power cord (don't merely turn off
the switch on any power strip -- pull the plug), wait 30 seconds, and
measure the heatsinks.

 

[Guest]

One easy way to increase ventilation is to remove the grills
that cover the exhaust fans. The wire grills can be removed
with just the removal of a few screws, and the ones that are
part of the case can be snipped away with a pair of electricians'
wire cutters. The air resistance that they impose can be felt
quite easily, and their removal allows greater airflow.
Don't worry about losing finger tips in them because the angle
of the blades of the fans push fingers out instead of snipping
them off.

This is dumb, unhelpful, and unnecessary.

The most important purpose of the grills is to prevent electric shock,
and there's usually exposed high voltage almost immediately on the
other side of the fan. I doubt that power supplies could receive UL,
CSA, or TUV safety certification without those grills.

Many people worry far too much about heat where it doesn't matter and
not enough about it where it does matter.