Intentional CPU Overheating

[JohnO]

Engineering hardware question for those so inclined.

Working on an experiment for the computer-repair classroom, I'm looking at
pulling the plug on a CPU fan just to watch what happens. I've done it a
couple times with a test system, an Intel DG965SS mobo with a Pentium D 820.
The temp creeps up steadily, the other fans try to make up the difference,
then the system shuts down in about 2-3 minutes.

Here's where the question comes. The normal op temp according to the Intel
Desktop Utils Hardware Monitor is about 51 C. Shutdown happens at about 90
C.

So, my intuition (and my wise engineer-assistant) tells me I'm shortening
the life of the CPU every time it gets that hot. What do you guys and gals
think about that? This is not a system that will go under a desk...it's a
classroom tool used intermittantly throughout a typical school year, so
shortening its life from 100,000 hours to 50,000 hours is irrelevant, if
real.

Last Q... why does the BIOS hardware monitor read 10 degrees hotter than the
Windows tool?

-John O

 

[Bob I]

Engineering hardware question for those so inclined.

Working on an experiment for the computer-repair classroom, I'm looking at
pulling the plug on a CPU fan just to watch what happens. I've done it a
couple times with a test system, an Intel DG965SS mobo with a Pentium D 820.
The temp creeps up steadily, the other fans try to make up the difference,
then the system shuts down in about 2-3 minutes.

Here's where the question comes. The normal op temp according to the Intel
Desktop Utils Hardware Monitor is about 51 C. Shutdown happens at about 90
C.

So, my intuition (and my wise engineer-assistant) tells me I'm shortening
the life of the CPU every time it gets that hot. What do you guys and gals
think about that? This is not a system that will go under a desk...it's a
classroom tool used intermittantly throughout a typical school year, so
shortening its life from 100,000 hours to 50,000 hours is irrelevant, if
real.

Yep, you're shortening it's life expectancy. It will quit when it dies.
Maybe 50,000, maybe 5000, maybe it just won't restart the next time.

Last Q... why does the BIOS hardware monitor read 10 degrees hotter than the
Windows tool?

hard to say, likely guess is the "conversion" factor used. Just use it
as a relative indicator

 

[Eric]

Engineering hardware question for those so inclined.

Working on an experiment for the computer-repair classroom, I'm looking at
pulling the plug on a CPU fan just to watch what happens. I've done it a
couple times with a test system, an Intel DG965SS mobo with a Pentium D
820. The temp creeps up steadily, the other fans try to make up the
difference, then the system shuts down in about 2-3 minutes.

Here's where the question comes. The normal op temp according to the Intel
Desktop Utils Hardware Monitor is about 51 C. Shutdown happens at about 90
C.

Still not a question.

So, my intuition (and my wise engineer-assistant) tells me I'm shortening
the life of the CPU every time it gets that hot. What do you guys and gals
think about that? This is not a system that will go under a desk...it's a
classroom tool used intermittantly throughout a typical school year, so
shortening its life from 100,000 hours to 50,000 hours is irrelevant, if
real.

Last Q... why does the BIOS hardware monitor read 10 degrees hotter than
the Windows tool?

-John O

I'm no expert on that, but I would guess overheating should not shorten the
life of the CPU. The failsafe that turns it off should kick in before it's
damaged.

Which Windows tool are you referring to?
Did you try measuring the temperature yourself?

 

[JohnO]

Which Windows tool are you referring to?

The Intel Desktop Utils hardware monitor.

Did you try measuring the temperature yourself?

No, I don't have a tool to do that at the moment. What's the best tool for
this? Does it require the removal of the heatsink?

-John O

 

[Ron Martell]

Engineering hardware question for those so inclined.

Working on an experiment for the computer-repair classroom, I'm looking at
pulling the plug on a CPU fan just to watch what happens. I've done it a
couple times with a test system, an Intel DG965SS mobo with a Pentium D 820.
The temp creeps up steadily, the other fans try to make up the difference,
then the system shuts down in about 2-3 minutes.

Here's where the question comes. The normal op temp according to the Intel
Desktop Utils Hardware Monitor is about 51 C. Shutdown happens at about 90
C.

So, my intuition (and my wise engineer-assistant) tells me I'm shortening
the life of the CPU every time it gets that hot. What do you guys and gals
think about that? This is not a system that will go under a desk...it's a
classroom tool used intermittantly throughout a typical school year, so
shortening its life from 100,000 hours to 50,000 hours is irrelevant, if
real.

Last Q... why does the BIOS hardware monitor read 10 degrees hotter than the
Windows tool?

Yes, you are shortening the life of the CPU.

There is a very slow chemical interaction between the various metallic
and semi-conductor material layers inside the CPU chip (and all other
semiconductors as well). This is similar to the galvanic corrosion
effects that can occur when two dissimilar metals are in contact.

In electronic components this chemical interaction is often referred
to as aging.

Because this is a chemical process the rate at which it occurs does
vary with temperature, which means that the hotter a component gets
the more rapidy this aging effect will occur and the sooner it will
fail.
Ron Martell Duncan B.C. Canada
--
Microsoft MVP (1997 - 2006)
On-Line Help Computer Service
http://onlinehelp.bc.ca
Syberfix Remote Computer Repair

"Anyone who thinks that they are too small to make a difference
has never been in bed with a mosquito."

 

[Eric]

The Intel Desktop Utils hardware monitor.

That is Intel software, not part of Windows. I have an Intel machine with
XP (from IBM) and it did not come with that.
Most people go by what the BIOS screen shows.

No, I don't have a tool to do that at the moment. What's the best tool for
this? Does it require the removal of the heatsink?

-John O

What sort of classroom is this? college? high school?
Do they have a science (chemistry?) class? I would think you might check
there for temperature testing tools.
Otherwise your guess is as good as mine, check the web for "temperature
tools", and make sure you find one that tests as quickly as possible and
covers more than the min/max temperature you expect to see. Surely you need
something more technical than the standard household thermometer.

I would try it without removing the heat sink and see what you get.
Surely a bad or missing heat sink will kill your CPU much faster than the
missing fan.

 

[JohnO]

What sort of classroom is this? college? high school?

Both, but mostly HS.

Do they have a science (chemistry?) class? I would think you might check
there for temperature testing tools.
Otherwise your guess is as good as mine, check the web for "temperature
tools", and make sure you find one that tests as quickly as possible and
covers more than the min/max temperature you expect to see. Surely you
need something more technical than the standard household thermometer.

I would try it without removing the heat sink and see what you get.
Surely a bad or missing heat sink will kill your CPU much faster than the
missing fan.

The only way (that I know of) to get an accurate measurement is to get as
close as possible to the center of the die itself, or know the diff between
the die temp and the spot on the heatsink being measured. For my purposes
relative measurements are fine, so I'm not going to worry about it much.

-John O

 

[JohnO]

Yes, you are shortening the life of the CPU.

There is a very slow chemical interaction between the various metallic
and semi-conductor material layers inside the CPU chip (and all other
semiconductors as well). This is similar to the galvanic corrosion
effects that can occur when two dissimilar metals are in contact.

In electronic components this chemical interaction is often referred
to as aging.

Because this is a chemical process the rate at which it occurs does
vary with temperature, which means that the hotter a component gets
the more rapidy this aging effect will occur and the sooner it will
fail.

Thanks...but...I wonder if the shutdown temp is within normal operational
parameters or prior to a point where aging starts. I suppose only an Intel
engineer knows for sure.

....I wish this mobo (DG965SS) had a setting for shutdown temp. The BIOS is
great, but it's very poorly documented.

-John O

 

[Bob I]

Thanks...but...I wonder if the shutdown temp is within normal operational
parameters or prior to a point where aging starts. I suppose only an Intel
engineer knows for sure.

...I wish this mobo (DG965SS) had a setting for shutdown temp. The BIOS is
great, but it's very poorly documented.

-John O

"Aging" occurs constantly, the higher the temperature, the more rapidly
it occurs. (unless you could take it to absolute zero ;-) ) Just expect
to replace it if it fails, no big deal. That's the same game the
overclockers play.