Reducing CPU temps on X24/X23/X22/X21/X20's w/Arctic Silver

Message

#1 Post by **underclocker** » Wed Jul 09, 2008 7:43 am

Just wanted to share the following pictures from a recent minor operation on an X24. There have been several posts about how hot or cool the X2x's run. Well here is a method that will reduce the running temperature several degrees Celcius, no matter which model you have.

Basically, the X2x's are not particularly hot running machines, however, there didn't seem to be much thought given to keeping the CPU cool on these machines. I've opened up several and always found the same thing. A very, very thin, worn and crusty metalic (I think) pad between the fan and CPU. NO thermal grease of any kind. (FYI, a similar setup is used in R30's, they benefit from this treatment, as well.)

What I always do to the CPU fan is scrape off the old metalic pad, then sand down the residual debris. Then I clean the CPU with an alcohol swab, let it dry and apply Artic Silver 5 (AS5) directly to the CPU. As always, I use less than 1/2 a grain of rice size drop of AS5 and spread it out as thin as possible, while still covering the entire top of the CPU top.

I have large picture links below. Enjoy.

Underside of fan with crusty old pad before removal -> http://www.nikkoaudio.com/pix/X24-Fan-C ... before.jpg

Underside of fan after scraping off the old pad with the tip of a thin screwdriver -> http://www.nikkoaudio.com/pix/X24-Fan-C ... craped.jpg

Underside of fan after light sanding with fine sandpaper -> http://www.nikkoaudio.com/pix/X24-Fan-C ... sanded.jpg

CPU after removal of fan (notice, it's almost perfectly clean! -> http://www.nikkoaudio.com/pix/X24-Fan-C ... before.jpg

CPU after light cleaning and Arctic Silver 5 application -> http://www.nikkoaudio.com/pix/X24-Fan-C ... Silver.jpg

Cleaned up fan, installed and resting happily in X24 -> http://www.nikkoaudio.com/pix/X24-Fan-C ... talled.jpg

weepy · #2 Post by **weepy** » Wed Jul 09, 2008 9:05 am

nice job, good idea, my x41 can get a bit hot, this is worth trying

.

j-dawg · #3 Post by **j-dawg** » Wed Jul 09, 2008 11:20 am

I had been thinking of doing this to my X24 but I didn't know if it would make any difference. Good to have confirmation that it will help; I think I'll give this a shot. Thanks!

graywolf · #4 Post by **graywolf** » Thu Jul 10, 2008 11:02 pm

I do not see any before and after numbers there. What I see is "I think this makes my cpu run cooler". Typically the difference between the factory sticky pad and lapping and using the most expensive thermal paste is about 3C. I can think of better things to do with my time.

#5 Post by **underclocker** » Fri Jul 11, 2008 5:54 am

I knew someone would want to see actual numbers. I purposely didn't want to post my results because, as with investing, results will vary. Plus, people use various power schemes and different hard drives run at different tempurature which has a very big effect in such a small laptop case - the hard drive is within a quarter of an inch of the CPU fan. Plus, internal wifi will cause higher temps. However, I will say that the difference, whether idling or under heavy CPU useage was generally between 5 and 9 degrees Celcius different for me.

This was enough of a difference to affect 1) the heat felt on the bottom of the machine, and 2) the fan speed.

(Right now with my X24 on battery, power scheme with adaptive setting for the CPU, with a Travelstar 60GB 7200rpm drive, 640MB RAM, internal Intel B/G wifi, I'm showing 41 degrees Celcius and the fan is on the lowest setting - inaudible.)

#6 Post by **gator** » Fri Jul 11, 2008 12:15 pm

graywolf wrote:I do not see any before and after numbers there. What I see is "I think this makes my cpu run cooler". Typically the difference between the factory sticky pad and lapping and using the most expensive thermal paste is about 3C.

That is true for new or fairly new laptops only. I have seen "sticky pads" in older thinkpads that were more of adhesive residue than thermal compound. I have never done a AS5 mod myself because I did not know to and I was afraid of damaging the CPU, but now I am confident that I can do it if need be after seeing underclocker's detailed pics/documentation. I am positive that we will see a reduction in temp.s if this is done right and it will not be just in my mind.

graywolf wrote: I can think of better things to do with my time.

That was something rude to say to someone who took the pains of doing the mod, taking great pics, and documenting it just to help others. What you consider better might not be the same for others, and vice versa. Please follow forum rules - criticize the ideas, not the people.

gongo2k1 · #7 Post by **gongo2k1** » Sat Jul 12, 2008 9:57 am

fyi, i did this on my x41 and made the mistake of removing the TIM from the northbridge... the die doesn't make contact with the heatsink without it.

luckily there wasn't any damage done, the machine just overheated and turned off... but now it's all in working order again!

kunfuchopsticks · #8 Post by **kunfuchopsticks** » Sat Jul 12, 2008 10:49 am

I have a lot of experience in thermal compounds.

The absolute KEY here is to apply the smallest amount possible. If you think you applied enough, you have applied too much.

The point of the compound is to cover microscopic bumps and holes on die and sink so to provide most efficient heat transfer. it's not suppose to be a layer of grease medium in between die and sink, but only to fill up the imperfections.

#9 Post by **gator** » Sat Jul 12, 2008 11:42 am

kunfuchopsticks wrote: The point of the compound is to cover microscopic bumps and holes on die and sink so to provide most efficient heat transfer. it's not suppose to be a layer of grease medium in between die and sink, but only to fill up the imperfections.

Absolutely. That's why I think the pink thermal pad is a bad idea for the processor (too thick). For the GPU/northbridge, it makes sense since there is a difference in height between the heatsink and the surface of the chips, but again, it would be ideal to redesign the heatsink such that it "touches" the chips directly.

j-dawg · #10 Post by **j-dawg** » Sat Jul 12, 2008 12:49 pm

The pink thermal pads are probably used because it's much cheaper for mass assembly.

Underclocker, I have one question: what grit sandpaper did you use to clean off your heatsink after scraping the pad off? I'm a little worried about scraping the pad off with a screwdriver; I might try some acetone to remove the pad's remnants and then use some 2000 grit paper and some high-grit polishing sticks to get the rest out.

Another interesting thing to note: on the desktop Core 2 processors, which have very big heat spreaders over the dies, Arctic Silver's instructions recommend applying a long, thin line of compound on the spreader, which gets squashed out and evenly spread when you install the heat sink. It's much easier than the application procedure on these processors, with their tiny dies.

#11 Post by **underclocker** » Sat Jul 12, 2008 1:58 pm

I agree there may be a better procedure for cleaning the fan (than using a screwdriver tip and 3 or 4 hundred grit sandpaper), but I used what I had available. The beauty is that apparently anything is better than IBM's solution!

Perhaps someone, with even more spare time than I, can test 300 hundred grit versus 4000 grit sandpaper for reducing CPU temps.

j-dawg · #12 Post by **j-dawg** » Sat Jul 12, 2008 4:36 pm

Hmm...well, I usually wet-sand, but a heatsink/fan combo is obviously not the same as a plastic model car. I'm also not interested in "lapping" the heatsink like some maniacs do because it will get me a very small return for a very large investment of effort (and also I've only got one heatsink/fan assembly, so I can't mess it up).

Well, I'll give this a shot anyway. Will post pictures if I can.

j-dawg · #13 Post by **j-dawg** » Sat Jul 12, 2008 4:48 pm

Well, it looks like the only small screwdriver I have is just too fat to turn the silver screws on the heatsink.

The HMM seems to indicate that an X24 needs only one screw removed, but looking at your pictures and the way the thing is assembled, I need to take out the silver ones around the CPU too, right?

I'll pick up a set of jeweler's screwdrivers...someday. I've been telling myself that I need to do that but I've just never got around to it.

clayton.bigsby · #14 Post by **clayton.bigsby** » Wed Jul 16, 2008 10:25 am

Does anyone know what kind of thermal medium is shipped from the factory on an X31? Is it a pad or ordinary white goop or perhaps fancy silver compound?

My T21 started showing signs of overheating (before upgrading to 1GHz) so I cleaned up the factory white goop and applied Arctic Silver and it ran like a champ.

proaudioguy · #15 Post by **proaudioguy** » Mon Jul 21, 2008 8:57 pm

This is interesting because my T20 had a pad of some sort originally. I figured the T21 would be the same. Eventually I removed it and worked it like play dough and reapplied it. That worked for a couple years, then I removed it because it was all dried up and the heat sink no longer made contact. I used some layers of normal tin foil with Arctic Silver 5 between layers to shim it. It took a few tries to get it right but it's ran like a champ since. This was originally a 700MHz unit that had been factory updated to a 900MHz. I suppose the heat sink couldn't keep up.

My X32 had normal heat sink grease which I replaced with AS5, as well as my A31s. It made a huge difference in the A31s, (7-10C) not sure why. It has made a less dramatic improvement in the X32. When I opened it, the old grease was around the CPU so they may have applied too much when the system board was replaced. Just removing the excess may account for the difference.

kunfuchopsticks · #16 Post by **kunfuchopsticks** » Mon Jul 21, 2008 9:30 pm

how much difference you get in temperature. With the way the X41 is setup, I'd have to take out the mobo to access it. Don't want to waste time with no results.

proaudioguy · #17 Post by **proaudioguy** » Tue Jul 22, 2008 7:02 pm

kunfuchopsticks wrote:With the way the X41 is setup, I'd have to take out the mobo to access it.

OUCH! Just another reason I love my X32. 4 Keyboard screws get me there, 5 screws on the heat sink and I'm wiping off the old grease.

gongo2k1 · #18 Post by **gongo2k1** » Tue Jul 22, 2008 9:15 pm

yeah, it's a pain to get to the hsf on the x41, i had to disassemble the entire lower chassis. i don't particularly monitor my temps, but i notice that the fan still spins about the same amount and the air coming out doesn't feel appreciably cooler. also add that to the fact that you won't be able to put as5 on the north bridge...

i only did it because i was already in there anyway, re-routing my wifi aerials, so i had to take down a good portion of the x41 to begin with.

not to be a nay-sayer here, but if you're not having problems with high cpu temps, i wouldn't bother on the x41.

proaudioguy · #19 Post by **proaudioguy** » Tue Jul 22, 2008 10:56 pm

I agree. Anything under 50C and you are doing just fine. I think once it's 3 years old it might get kind of dried up and be much more worth it. It's nice to clean the fan out sometimes too.

kunfuchopsticks · #20 Post by **kunfuchopsticks** » Wed Jul 23, 2008 11:33 am

gongo2k1 wrote:i don't particularly monitor my temps, but i notice that the fan still spins about the same amount and the air coming out doesn't feel appreciably cooler.

If it was doing a better job, that means the air coming out should be hotter.

You want hotter air coming out.

j-dawg · #21 Post by **j-dawg** » Wed Jul 23, 2008 12:59 pm

Not really. Think of heat as a quantifiable thing that determines temperature and you will see why. Hotter air coming out means the source is putting out more heat. An effective cooler will keep the temperatures down to begin with; thus, the more effective cooler will put out cooler air. Hot air (as an indicator) is only desirable if it eventually cools down, indicating that the excess of heat that built up on the processor has been drawn away.

In other news, I have acquired a set of jeweler's screwdrivers. Back to the surgery table.

j-dawg · #22 Post by **j-dawg** » Wed Jul 23, 2008 8:44 pm

My surgery does not appear to have helped, but I suspect it is because I altered the procedure. The metallic pad is held in place by a clear sticker with a hole for the CPU die. I took this off because of a buildup of Thermal Stuff which was gross. I think removing and replacing that pad made heat dispersion worse.

As promised, I documented the process with photos:

img]http://img133.imageshack.us/img133/2218 ... em4.th.jpg[/img]
The CPU. As we saw earlier, it is devoid of anything that would suggest thermal interface material.

img]http://img78.imageshack.us/img78/858/cimg2828xc6.th.jpg[/img]
The CPU again.

img]http://img57.imageshack.us/img57/4311/c ... ul8.th.jpg[/img]
This is the heatsink. You will notice in the lower-right corner a good deal of buildup of all that goop. This is why I peeled off the plastic adhesive holding on the pad.

img]http://img133.imageshack.us/img133/1362 ... cd2.th.jpg[/img]
From another angle. It really was a big buildup. I was concerned that I wouldn't be able to line it up properly again; I apparently forgot that the heatsink is held in by screws, which pretty much guarantees that it'll be placed identically when reinstalled.

img]http://img515.imageshack.us/img515/2117 ... ls7.th.jpg[/img]
Everything removed. As you can see, the heatsink itself has a near-mirror finish.

img]http://img155.imageshack.us/img155/9349 ... nd5.th.jpg[/img]
The other side of the pad had a little bit of dried-up thermal stuff on it, but much less.

img]http://img65.imageshack.us/img65/2828/c ... ro8.th.jpg[/img]
Cleaned up and reinstalled. I lined it up by the impression left on the sticker.

img]http://img235.imageshack.us/img235/787/ ... ny0.th.jpg[/img]
The die.

img]http://img507.imageshack.us/img507/5816 ... ev3.th.jpg[/img]
Installed again.

img]http://img92.imageshack.us/img92/6118/c ... il7.th.jpg[/img]
After I removed it the second time. I was concerned that I might have lined up the plastic sticker holding the pad on inaccurately, which would explain the poor performance, but judging by the thermal paste on the pad I seem to have done it right.

After all this I put a tiny bit of thermal paste between the metallic pad and the heatsink, but this does not appear to have helped matters.

All this begs the question: what is the purpose of the metallic pad? I'm guessing it was the thermal interface material of choice when the computer was assembled because it was easy to apply and not messy. But I'm willing to put in a bit more time than a busy shift worker in Mexico. Can I just remove the pad and stick some AS5 between the die and the heatsink? The only purpose the metallic pad is serving now is to increase pressure (not to mention thermal resistance) between the heatsink and the die, but as far as I can tell, the heatsink clamps down with enough force to generate sufficient contact without the pad.

kunfuchopsticks · #23 Post by **kunfuchopsticks** » Wed Jul 23, 2008 9:50 pm

Thanks for pics.

I mean, the reason i would want to replace thermal grease is because the old stuff is old and lost it's capacity to conduct heat efficiently- so only warmer air is piped out.
If re-greased so it conducts at 100% efficiency, then hotter air should pipe out.

gongo2k1 · #24 Post by **gongo2k1** » Wed Jul 23, 2008 11:16 pm

not quite... you're forgetting that a heatsink will always have a higher specific heat than air, so the air is only going to get hotter if
a) you slow down the fan (decrease air flow, but this means that you run the risk of not pushing it fast enough and the saturated air sits there too long),
b) you increase the total surface area of the heatsink while maintaining the same air volume (which you can't, there's no room),
c) you change the material of the heatsink to have a lower specific heat so that it lets go of heat sooner into the air (but this means that you'll have to increase air volume since it will become saturated sooner and eventually so does the heatsink, which is why they choose a higher specific heat in the first place), or
d) the cpu die is actually putting out more heat (=something is wrong).
the amount of heat going from cpu to heatsink will always be the same, the thermal interface material (in this case, arctic silver) only facilitates the rate at which it is transfered to the heatsink, it doesn't increase or decrease the energy that we're sinking away from the cpu.

ultimately we can't feasibly control the interface for heat exchange between the sink and the surrounding air.
furthermore, since the fan in the laptop is presumably controlled via a temp sensor that starts/stops the fan, there isn't too much point in discussing the temperature of the air.

technically, if we always had the same volume of air flowing over the sink, but we changed the TIM to allow for better die-sink heat transfer, the largest observable difference ought to be a faster cool off period when going from full load to idle.

proaudioguy · #25 Post by **proaudioguy** » Thu Jul 24, 2008 1:40 am

j-dawg wrote: All this begs the question: what is the purpose of the metallic pad? I'm guessing it was the thermal interface material of choice when the computer was assembled because it was easy to apply and not messy. But I'm willing to put in a bit more time than a busy shift worker in Mexico. Can I just remove the pad and stick some AS5 between the die and the heatsink? The only purpose the metallic pad is serving now is to increase pressure (not to mention thermal resistance) between the heatsink and the die, but as far as I can tell, the heatsink clamps down with enough force to generate sufficient contact without the pad.

I read the entire post up to this point scratching my head as to why you left that nasty pad in there. If there is flat pressure between the heat sink and the die, get rid of the pad. I'm certain you will see a big improvement.

On the newer heat sinks (not sure about yours) there is a spring loaded washer in each of the screw holes of the heat sink. When you tighten the screws all the way down it compresses the springs a bit. The heat sink is floating on the die. There is no change of breaking the die. I don't think your machine has this system in place. You must be careful to just barely snug each screw a little at a time until it's tight enough. If you screw 1 screw down all the way and then another you could crack the die. I think more likely though is that there will be no contact with the die after you remove the pad. This is how it was in my T20. What I did was make my own pad that was precisely cut strips of household aluminum foil with a tiny layer of AS5 between each layer of foil. I built up the size until it was too big, then started over. it was trial and error and took about 4 tries to get it the right mils thick. Before this, with the original heat sink putty or whatever it was, the machine would not run at full speed without shutting down. After I got rid of that and used the AS5 it was hit and miss. I had a THICK layer in there. When I realized there was no contact, I shimmed it as I just described, and for the first time since I bought this computer for my dad back in 2001, it ran full speed with no issues. In fact it's been running that way for 2 years. With the addition of a 5400 60 gig HD to replace the old 12 gig 4200, this computer has a new lease on life. It's actually usable and my wife doesn't want me to sell it. Before it would barely run windows 2000, after AS5, AL foil, and an improved HD, it runs a full install of Win XP just fine.

Sorry for the long winded response.

j-dawg · #26 Post by **j-dawg** » Thu Jul 24, 2008 11:04 am

No, that's very helpful. I have at least one voice urging me to get rid of the pad. The reason I left it in was because I saw underclocker had done so, but his thermal goop buildup wasn't as nasty as mine.

I'm going to try it with nothing at all between the die and the heatsink, tightening the screws slowly in an alternating X-pattern like they recommend for desktop heatsinks. To be honest I'm not really all that worried about cracking the die, nor am I too worried about contact; though I've been working on it with the pad in, the heatsink spring has enough tension in it to let me lose that half-millimeter without any problems. If not, I'll try your aluminum foil method.

#27 Post by **underclocker** » Thu Jul 24, 2008 9:11 pm

j-dawg, you did a great job on the disassembly. I didn't even notice that there was a pad between the fan and CPU. I thought I was scraping the fan clean! Thanks for the pix.

I suppose that I might go back in now and remove the pad completely. I recall that the fan seemed to be spring-loaded to some degree. Tension did increase as you tightened the screws.

A good test would be to remove the pad, apply thermal grease, install fan, remove fan and look to see if there is thermal grease residue on the fan. If there seems to be a reasonable impression, we're good to go. (Although, from your pix, the pad looks rather thick. What metal is that, anyway?!)

Obviously, I never even checked the reverse side of the metal pad, so who knows how much junk is there.

Let's try to figure out a repeatable way to lower temps on these machines.

kunfuchopsticks · #28 Post by **kunfuchopsticks** » Thu Jul 24, 2008 11:07 pm

Be sure to report back with before and after temps using same temp utility. Thanks.

proaudioguy · #29 Post by **proaudioguy** » Fri Jul 25, 2008 12:01 am

You could put a single sliver of al foil in between with the end sticking out (no grease). Tighten up the screws and pull on the foil. If it slides out, I'd say there is no contact. Basically a sacrificial feeler gauge.

#30 Post by **underclocker** » Sat Jul 26, 2008 12:26 pm

Good idea. I will attempt this.

Reducing CPU temps on X24/X23/X22/X21/X20's w/Arctic Silver

Reducing CPU temps on X24/X23/X22/X21/X20's w/Arctic Silver

hmmm

Who is online