overheating CPU

[Guest]

I am hoping someone can tell me what a "normal" RPM for a CPU fan might be,
or why else my CPU is overheating.

1) I do not overclock
2) CPU fan speed is 2556RPM
3) The average CPU temperature lately is 90C
4) I just went into BIOS right now, the CPU temp was 77 degrees, which is
the lowest i've seen in a few days, but i sat there in BIOS for only about
3-4 minutes and the CPU temp went from 77 to 90 in those few minutes when all
i was doing was looking at BIOS. During those few minutes, the CPU fan speed
was 2518-2556 RPM.
5) The other box fans are working, and the side is off, so there should be
plenty of airflow. The heat fins are clean, vents and fans are clean.

PC is only about 2 years old, XP Pro SP2, 512 ram. I don't even use this
computer for gaming, music, etc. Just mainly email (i never open attachments
or allow active x)and browsing. It's behind a firewall, router, switch,
Norton AV, MS AntiSpyware, and routinely updated, so a virus or worm is out
of the question.

Thanks!

 

[Guest]

Hi Greg,

May I know what CPU you are using? Maybe I could help.

Regards,
Beelee

 

[Guest]

Sorry, it is a Pentium 4 2.4Ghz

Hi Greg,

May I know what CPU you are using? Maybe I could help.

Regards,
Beelee

 

[Bob I]

That may be caused by insufficient thermal compound between the CPU and
heatsink.

 

[frodo]

2600 rpm for cpu fan is reasonable, tho 90 degree C is WAY hot. Since
it's an older 2.4 GHZ, I assume it's not a Prescott core, so it should be
down around 45-50 degrees C (under load, 40 w/ no load). Prescotts do run
hotter. Are you (or the bios) running any kind of fan speed controller?
Disable it and check fan speed again.

What is case intenal temp - if it's way high then that explains it.
Check for dust bunnies, good airflow, etc.

Are you SURE the temp reading is actually correct? What does the bios say,
and do they jive? Get the proper utility from the motherboard maker, or
the proper version of Motherboard Montior, if they have one that supports
your motherboard.

If everything checks outs, I too would guess the interface between the
cooler and the cpu top is bad. VERY CAREFULLY remove the heatsink (it can
stick to the cpu, be very careful to remove it gently, without yanking cpu
from socket - use a GENTLE twist/rock motion). Clean both surfaces and
reapply a good heat transfer compound like Artic Silver. Follow directions
cafefully, a little dab will do ya.

[Intel original heat compound is a double-sided-tape like strip, and it's
hard to remove. Be gentle and patient, it will come off with care. Also,
consider a better heatsink+fan unit, such as a Zalmann.]

Good Luck.

 

[cquirke (MVP Windows shell/user)]

That may be caused by insufficient thermal compound between the CPU and
heatsink.

Or too much.

Or an inaccurate temp sensor, or a case that can't dump air out of the
case as fast as the CPU fan pulls the heat off the heatsink.

Greg wrote:

Prescott generation Pentium 4 or Celeron?

Looks OK

90C !! Or 90F? Over 70C is nasty, under 50C is OK

Windows XP has a true idle loop, that allows the CPU to "cool off" by
parking it in logic that reduces current flow. DOS mode and BIOS do
not have this, and AFAIK neither does MemTest86 - so that's why CPU
can heat up there, even though there's seemingly not a lot going on.

When I started building Prescott systems, I used nice little full-ATX
cases that had the PSU orientated vertically, at the same level as the
processor. The motherboards (Intel Bayfield ATX S478 i865G at the
time) would report high Zone 2 temeratures, and got unstable when this
went over 42C, even though the CPU temp was OK. They were noisy, too.

Adding fans did not help; the problem only went away when I switched
to larger cases that had the PSU "flat", and above the processor, so
there was no pocket of "dead" air at the top of the case to cook up.

Another change I made that reduced fan speed and noise, was to disable
fan speed control in CMOS setup. With that enabled, the fans would
tick over when the PC was started up, allowing it to get hotter
faster, before speeding up to try to cool it down again. With the
"feature" disabled, the fans work at a good moderate pace from the
start, so that the system takes longer to warm up into trouble and
noise. This is in a 15C-35C ambient environment; YMMV if < 0C etc.

Sounds old for a Prescott. AMD or Intel CPU? Boxed or add-on CPU
fan? How many other fans? SVGA card/chipset/fan?

Make sure the fan mounting hasn't come adrift, as can happen with the
P4-generation clip-on fan cages, which can sometimes break at one of
the 4 clips and let the heat sink angle off into bad contact. That's
a Socket 478 thing; the new Socket 775s seem to be OK on that.

I read somewhere that unlike most "passive" conductors, semiconductors
(i.e. transistors and diodes, as in chips) drop in resistance as they
heat up - it's called the "cascade effect". Whereas a tungstun wire
heats up until it glows, but increases resistance so current is
limited and it doesn't fry, the cascade effect means that once
semiconductors start getting hot, they pass more current and get even
hotter, until they run away and fry. Positive feedback loops suck :-(

------------ ----- ---- --- -- - - - -

The most accurate diagnostic instrument
in medicine is the Retrospectoscope

 

[Dick Mahar]

Or too much.

Or an inaccurate temp sensor, or a case that can't dump air out of the
case as fast as the CPU fan pulls the heat off the heatsink.


Prescott generation Pentium 4 or Celeron?


Looks OK


90C !! Or 90F? Over 70C is nasty, under 50C is OK


Windows XP has a true idle loop, that allows the CPU to "cool off" by
parking it in logic that reduces current flow. DOS mode and BIOS do
not have this, and AFAIK neither does MemTest86 - so that's why CPU
can heat up there, even though there's seemingly not a lot going on.


When I started building Prescott systems, I used nice little full-ATX
cases that had the PSU orientated vertically, at the same level as the
processor. The motherboards (Intel Bayfield ATX S478 i865G at the
time) would report high Zone 2 temeratures, and got unstable when this
went over 42C, even though the CPU temp was OK. They were noisy, too.

Adding fans did not help; the problem only went away when I switched
to larger cases that had the PSU "flat", and above the processor, so
there was no pocket of "dead" air at the top of the case to cook up.

Another change I made that reduced fan speed and noise, was to disable
fan speed control in CMOS setup. With that enabled, the fans would
tick over when the PC was started up, allowing it to get hotter
faster, before speeding up to try to cool it down again. With the
"feature" disabled, the fans work at a good moderate pace from the
start, so that the system takes longer to warm up into trouble and
noise. This is in a 15C-35C ambient environment; YMMV if < 0C etc.


Sounds old for a Prescott. AMD or Intel CPU? Boxed or add-on CPU
fan? How many other fans? SVGA card/chipset/fan?

Make sure the fan mounting hasn't come adrift, as can happen with the
P4-generation clip-on fan cages, which can sometimes break at one of
the 4 clips and let the heat sink angle off into bad contact. That's
a Socket 478 thing; the new Socket 775s seem to be OK on that.

I read somewhere that unlike most "passive" conductors, semiconductors
(i.e. transistors and diodes, as in chips) drop in resistance as they
heat up - it's called the "cascade effect". Whereas a tungstun wire
heats up until it glows, but increases resistance so current is
limited and it doesn't fry, the cascade effect means that once
semiconductors start getting hot, they pass more current and get even
hotter, until they run away and fry. Positive feedback loops suck :-(

Fan speed is not as definitive as air delivery, expressed in CFM. Cubic Feet
Per Minute, which more accurately defines how much air is actually moving, as
established by fan speed, and pitch and size of the fan blades. In general, fan
speed and pitch and size of blades also determines the noise produced, another
tradeoff.

 

[Guest]

This is exactly what happened. I have never touched the fan, or mounting, or
heat sink, etc, so i never thought to check it before.

It is broken, both the top bracket (on top of fan) and bottom bracket
(attached right to motherbaord) has broken into many pieces and the heat fins
were not sitting flat on the CPU. I have the 478 socket.

I have it cooling correctly now and it's been at 45 degrees for the last 2
days with no reboots. Thank you everyone for taking the time to help me, i
appreciate it