Best way cpu cooler paste?

  • Thread starter Thread starter jaster
  • Start date Start date
J

jaster

What is the best way of applying cpu cooler paste?

Are the adhesive strips better than the manufacturer's paste?

I replaced my whinning Thermaltake with a recommended Coolermaster
X-Dream III which is quieter but running hotter (59C, +6C). I assume
because I have too much paste. I'd like to get that paper thin
spread of manufacturer applied paste. Thanks to everyone for all of the
cpu cooler recommendations.

TIA
 
Being that you're worried about the proper amount to apply I'd suggest
getting the a thermal pad strip so you don't lose sleep. If it really comes
down to not being as fearful as it seems from your writing, I'd just take a
thin dab of paste and cover the face of the chip evenly and thinly. You
don't want it thick so as t ooze out. When you think of the pressure that
will be generated by the spring clips that hold the HSF onto the socket you
realize that there really won't be very much paste between the two surfaces.
 
Appreciate how thermal conductivity works. Less medium
changes means better conductivity. A good heatsink is
properly machined so that most all the heatsink makes direct
contact with the CPU. Left are microscopic holes that contain
air. Heat from CPU to air to heatsink is a less conductive
path. So we replace air with a more conductive thermal
paste. IOW the past must be so thin as to only fill those
microscopic holes and not impeded the CPU to heatsink
contacts.

This, of course, assumes the heatsink was properly
machined. Some immediately assume the heatsink must be lapped
perfectly smooth. Wrong. But heatsinks that are properly
machined will also tend to include the most important
parameter for heatsink selection: a number called "degrees C
per Watt".

A heatsink properly machined with thermal compound will
typically see a less than 10 degree reduction in CPU
temperature compared to the same heatsink without. IOW the
number of microscopic holes are not everywhere. However many
heatsinks sold to the naive (meaning they don't provide the
"degree C per watt" parameter) are so poorly machined that the
thermal compound causes a greater reduction in CPU
temperature.

Thermal tapes are how we sell inferior machined heatsinks;
then use the tape to mask that bad machine job. Many will
hype Arctic Silver routinely without even providing the
numbers. Most every thermal compound is equivalent to Arctic
Silver except costs many times less money.

Apply so little compound that it does not spread to the
edges. Almost no heat transfers at those edges anyway. But
if compound gets outside the CPU to heatsink interface, then
it can cause other problems including intermittent electrical
type problems. And if thermal compound is applied so heavy as
to squeeze out from that CPU / heatsink boundary, then many
locations that should have been heatsink to CPU direct contact
now have less thermal conductivity - thermal compound is
impeding heat flow.
 
Appreciate how thermal conductivity works. Less medium
changes means better conductivity. A good heatsink is properly machined
so that most all the heatsink makes direct contact with the CPU. Left are
microscopic holes that contain air. Heat from CPU to air to heatsink is a
less conductive path. So we replace air with a more conductive thermal
paste. IOW the past must be so thin as to only fill those microscopic
holes and not impeded the CPU to heatsink contacts.

This, of course, assumes the heatsink was properly
machined. Some immediately assume the heatsink must be lapped perfectly
smooth. Wrong. But heatsinks that are properly machined will also tend
to include the most important parameter for heatsink selection: a number
called "degrees C per Watt".

A heatsink properly machined with thermal compound will
typically see a less than 10 degree reduction in CPU temperature compared
to the same heatsink without. IOW the number of microscopic holes are not
everywhere. However many heatsinks sold to the naive (meaning they don't
provide the "degree C per watt" parameter) are so poorly machined that the
thermal compound causes a greater reduction in CPU temperature.

Thermal tapes are how we sell inferior machined heatsinks;
then use the tape to mask that bad machine job. Many will hype Arctic
Silver routinely without even providing the numbers. Most every thermal
compound is equivalent to Arctic Silver except costs many times less
money.

Apply so little compound that it does not spread to the
edges. Almost no heat transfers at those edges anyway. But if compound
gets outside the CPU to heatsink interface, then it can cause other
problems including intermittent electrical type problems. And if thermal
compound is applied so heavy as to squeeze out from that CPU / heatsink
boundary, then many locations that should have been heatsink to CPU direct
contact now have less thermal conductivity - thermal compound is impeding
heat flow.
Click to expand...

Thanks. Sounds like tape is the answer.
 
Being that you're worried about the proper amount to apply I'd suggest
getting the a thermal pad strip so you don't lose sleep. If it really
comes down to not being as fearful as it seems from your writing, I'd just
take a thin dab of paste and cover the face of the chip evenly and thinly.
You don't want it thick so as t ooze out. When you think of the pressure
that will be generated by the spring clips that hold the HSF onto the
socket you realize that there really won't be very much paste between the
two surfaces.
Click to expand...

Thanks. I'll go for some tape .
 
Appreciate how thermal conductivity works. Less medium
changes means better conductivity. A good heatsink is properly machined
so that most all the heatsink makes direct contact with the CPU. Left are
microscopic holes that contain air. Heat from CPU to air to heatsink is a
less conductive path. So we replace air with a more conductive thermal
paste. IOW the past must be so thin as to only fill those microscopic
holes and not impeded the CPU to heatsink contacts.

This, of course, assumes the heatsink was properly
machined. Some immediately assume the heatsink must be lapped perfectly
smooth. Wrong. But heatsinks that are properly machined will also tend
to include the most important parameter for heatsink selection: a number
called "degrees C per Watt".

A heatsink properly machined with thermal compound will
typically see a less than 10 degree reduction in CPU temperature compared
to the same heatsink without. IOW the number of microscopic holes are not
everywhere. However many heatsinks sold to the naive (meaning they don't
provide the "degree C per watt" parameter) are so poorly machined that the
thermal compound causes a greater reduction in CPU temperature.

Thermal tapes are how we sell inferior machined heatsinks;
then use the tape to mask that bad machine job. Many will hype Arctic
Silver routinely without even providing the numbers. Most every thermal
compound is equivalent to Arctic Silver except costs many times less
money.

Apply so little compound that it does not spread to the
edges. Almost no heat transfers at those edges anyway. But if compound
gets outside the CPU to heatsink interface, then it can cause other
problems including intermittent electrical type problems. And if thermal
compound is applied so heavy as to squeeze out from that CPU / heatsink
boundary, then many locations that should have been heatsink to CPU direct
contact now have less thermal conductivity - thermal compound is impeding
heat flow.
Click to expand...
Thanks for the details on cooling. I get some tape, Now I'm pretty sure
the glue is overlapping, too thick and not doing the job.
 
Thanks for the details on cooling. I get some tape, Now I'm pretty sure
the glue is overlapping, too thick and not doing the job.
Click to expand...

I prefer paste to tape but as he said you have to apply it very
sparingly and thin. I just apply a small dab (size of a grain of rice)
in the middle then spread it thinly with a credit card.
 
Please go back and reread that post again. It says in the
beginning that less medium changes means better conductivity.
So you put tape on the heatsink eliminating all direct from
CPU to heatsink connection? You therefore increase the number
of medium changes? That means less thermal conductivity -
what you don't want. Most heat transferring from CPU direct
to heatsink without tape or thermal compound means better
thermal conductivity. Clearly tape is a completely wrong
answer.

But and again, also mentioned is the first thing you wanted
to address - "degrees C per watt" parameter.
 
Please go back and reread that post again. It says in the
beginning that less medium changes means better conductivity. So you put
tape on the heatsink eliminating all direct from CPU to heatsink
connection? You therefore increase the number of medium changes? That
means less thermal conductivity - what you don't want. Most heat
transferring from CPU direct to heatsink without tape or thermal compound
means better thermal conductivity. Clearly tape is a completely wrong
answer.

But and again, also mentioned is the first thing you wanted
to address - "degrees C per watt" parameter.
Click to expand...

Thanks again.

I used tape on my P3 system and I never had a heat issue.
This XP2000 is the hottest CPU around lots of heat.

I tried bare contact but the pc got stuck booting on reboot bios showed
62C. Cpu information is stamped on the chip so paste is necessary.
I re-applied the paste using a plastic edge. So far the temps is
55C. Maybe a case mod to add a large exhaust fan.

Thanks
 
I prefer paste to tape but as he said you have to apply it very sparingly
and thin. I just apply a small dab (size of a grain of rice) in the middle
then spread it thinly with a credit card.
Click to expand...

Thanks credit card works.
 
55 and 62 degrees C are what was predicted in that previous
post for CPU with and without thermal compound. IOW thermal
compound makes a minor improvement in thermal conductivity.
This temperature improvement (which is essentially same
temperature to CPU) suggests the heatsink is properly
machined. Both temperatures are well below maximum CPU
temperature of 90 degree C; temperature would have to be that
high to cause CPU to go fail-safe. At 55 and 62 degrees, the
CPU is way cool enough.

Don't get caught with myths about more chassis fans. One
chassis fan is more than enough to cool a system. Second fan
(running in parallel) would only result in maybe a 5 degree
difference - irrelevant. However many don't do the numbers.
They just know it must need more fans - numbers be damned.

Since fans are a least reliable function, then some will
install a second fan so that if power supply fan fails, this
second fan will maintain sufficient airflow (fans running in
series). But more chassis fans to reduce computer temperature
is promoted by those who never do the numbers.

A computer with one chassis fan and a properly installed
heatsink will work just fine even when computer operates in a
room of 100+ degrees F.

BTW, P3 systems did not even require a CPU heatsink fan.
The heat output from a PC could be sufficiently removed is
power supply fan causes a gentle airflow across CPU heatsink.
P4s are the first Intel CPUs that really needed that heatsink
fan due to their higher energy consumption.
 
55 and 62 degrees C are what was predicted in that previous
post for CPU with and without thermal compound. IOW thermal compound
makes a minor improvement in thermal conductivity. This temperature
improvement (which is essentially same temperature to CPU) suggests the
heatsink is properly machined. Both temperatures are well below maximum
CPU temperature of 90 degree C; temperature would have to be that high
to cause CPU to go fail-safe. At 55 and 62 degrees, the CPU is way cool
enough.
Click to expand...
Yes I found an old THG report showing AMD cpu fans for XP2000+ to be 55C
idle and 57C at load. So I'm getting that and it's a lot quieter than the
volcano or with an AMD supplied fan.
Don't get caught with myths about more chassis fans. One
chassis fan is more than enough to cool a system. Second fan (running
in parallel) would only result in maybe a 5 degree difference -
irrelevant. However many don't do the numbers. They just know it must
need more fans - numbers be damned.

Since fans are a least reliable function, then some will
install a second fan so that if power supply fan fails, this second fan
will maintain sufficient airflow (fans running in series). But more
chassis fans to reduce computer temperature is promoted by those who
never do the numbers.

A computer with one chassis fan and a properly installed
heatsink will work just fine even when computer operates in a room of
100+ degrees F.
Click to expand...

Room temp issue. The room temp is raised by the XP2000+. It's like
having an iron or hot plate turned on in the room. In winter the extra
heat's okay and the cpu temp drops 10C. Thank goodness the AMD and Intel
have gotten the message and are making lower temp and quieter cpus.

When I got the XP2000 I bought the Antec 3700 case which has a 120mm
bottom intake and a 120mm exhaust fan below the PSU.

Thanks again.

BTW, P3 systems did not even require a CPU heatsink fan.
The heat output from a PC could be sufficiently removed is power supply
fan causes a gentle airflow across CPU heatsink. P4s are the first Intel
CPUs that really needed that heatsink fan due to their higher energy
consumption.
Click to expand...
Room temp issue. Thank goodness the AMD and Intel have gotten the message
and are making lower temp and quieter cpus.

Thanks again.
 
w_tom said:
55 and 62 degrees C are what was predicted in that previous
post for CPU with and without thermal compound. IOW thermal
compound makes a minor improvement in thermal conductivity.
This temperature improvement (which is essentially same
temperature to CPU) suggests the heatsink is properly
machined. Both temperatures are well below maximum CPU
temperature of 90 degree C; temperature would have to be that
high to cause CPU to go fail-safe. At 55 and 62 degrees, the
CPU is way cool enough.

Don't get caught with myths about more chassis fans. One
chassis fan is more than enough to cool a system. Second fan
(running in parallel) would only result in maybe a 5 degree
difference - irrelevant. However many don't do the numbers.
They just know it must need more fans - numbers be damned.
Click to expand...
Agree. As example, I added a second, front intake fan to my Antec
SLK3700BQE case. Temps went UP 1C. Repeated the installation,
uninstallation of the fan 3 times. Each time, temps went up 1C.
 
w_tom said:
55 and 62 degrees C are what was predicted in that previous
post for CPU with and without thermal compound. IOW thermal
compound makes a minor improvement in thermal conductivity.
This temperature improvement (which is essentially same
temperature to CPU) suggests the heatsink is properly
machined. Both temperatures are well below maximum CPU
temperature of 90 degree C; temperature would have to be that
high to cause CPU to go fail-safe. At 55 and 62 degrees, the
CPU is way cool enough.
Click to expand...

If those are idle temps, I would look at airflow and proper installation of
HS/Fan. Under load, those temps can rise easily to temps that could cause
instability. If they are "under load" temps, I would agree.
Don't get caught with myths about more chassis fans. One
chassis fan is more than enough to cool a system. Second fan
(running in parallel) would only result in maybe a 5 degree
difference - irrelevant. However many don't do the numbers.
They just know it must need more fans - numbers be damned.
Click to expand...

If one fan works, I agree. There are many variables though, including the
case itself, proper airflow across NB and CPU and many other things. In a
perfect world, I agree.
Since fans are a least reliable function, then some will
install a second fan so that if power supply fan fails, this
second fan will maintain sufficient airflow (fans running in
series). But more chassis fans to reduce computer temperature
is promoted by those who never do the numbers.
Click to expand...

Fans are the most important part of an air-cooled system. Yes, people do go
overboard sometimes. Many people today run 4 or more HDDs in a couple of
raid arrays and a lot of other peripherals. These can hinder proper airflow
to the CPU area and one fan is not always enough. Every system must be
looked at individually and proper airflow may need an extra fan or two. If
you have a perfectly unobstructed airflow from the bottom-front to the
HS/Fan, one exhaust fan MAY be all you need. IOW, optimum cooling may take
different routes with different systems. You can use all the numbers you
want.
A computer with one chassis fan and a properly installed
heatsink will work just fine even when computer operates in a
room of 100+ degrees F.
Click to expand...

If the proper airflow is achieved with one fan, I would agree.
BTW, P3 systems did not even require a CPU heatsink fan.
The heat output from a PC could be sufficiently removed is
power supply fan causes a gentle airflow across CPU heatsink.
P4s are the first Intel CPUs that really needed that heatsink
fan due to their higher energy consumption.
Click to expand...

P3s would begin shutting down pretty fast without a CPU fan. I doubt if many
would even post. I think you must be talking about the early Pentiums and
x86 systems. Those of us who dabbled in overclocking those early x86/Pentium
systems added fans to the HSs and/or fashioned more efficient HSs for them.
The P2s and P3s are what brought about the industry of all the aftermarket
heatsinks and fan combos. You would have needed quite a powerful PSU fan to
cool a P3 HS.
 
Back
Top