LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Hardware (http://www.linuxquestions.org/questions/linux-hardware-18/)
-   -   Thermal contact grease between the CPU and cooler? (http://www.linuxquestions.org/questions/linux-hardware-18/thermal-contact-grease-between-the-cpu-and-cooler-900756/)

stf92 09-02-2011 12:34 PM

Thermal contact grease between the CPU and cooler?
 
Hi:

Between a socket 370 Celeron and the cooler ought there be any thermal contact grease? Thanks.

{BBI}Nexus{BBI} 09-02-2011 12:36 PM

Short answer, Yes there should.

screwbottle 09-02-2011 02:27 PM

Lesser processors like a Celeron could also have a thermal rubber type cooling mat between the processor cap and the heatsink, but yes a special thermal paste is applied in the majority of cases. It has a simple process, although complex in how it does this, firstly it fills in the micro uneven surfaces to create a smooth contact between these surfaces and to transfer the heat rapidly between the surfaces. A very thin layer is applied. Hope this is what you are looking for in an answer.

stf92 09-02-2011 04:42 PM

That is, indeed.

About the mat, it is the cooler that seems to have one, though I can tell for certain. I feels different the the sense of touch than the rest of the cooler but maybe it's because of the old grease deposited on it.

I intend to remove the silicon grease anyway with isopropyl alcohol. Thankks.

jefro 09-02-2011 10:19 PM

Less is more in this case. You don't need to remove old grease. It doesn't go bad.

warpuck 09-03-2011 06:41 AM

any non-conductive grease will do
 
Jefro is absolutely correct.
I have replaced cpus in laptops reusing conductive mats. What works for me is isopropanol, mineral oil & artic silver. I mix that all together & use the head of straight pin, once,,, to put that mixture on the middle of the cpu and clamp it down. I dont bother to spread it, the clamping pressure does it for me. This has worked well so far.

stf92 09-03-2011 08:41 AM

I'm happy to see the Linux-Hardware forum honors its name. No conductive mat in this case, I've been looking better after removing old hardened grease. The fact is that the CPU top surface is metallic and that part of the cooler which is in contact with that surface is some aluminium-like mat somehow fixed to the cooler base. Furthermore, it is a bit wrinkled all over its area.

This did not pose a problem to me. I wiped both contact surfaces with isopropyl (you can read Wikipedia article on it and its amazing number of applications in the electronics field) and used a drop of silicon grease, the same used for power transistors.

Its the prettiest thing in the world to see how socket, processor, cooler and fan fit altogether with absolute perfection. Well, thanks for your posts.

markhahn 09-03-2011 11:36 AM

sorta
 
the main goal in applying a thermal transfer compound is to avoid voids, so to speak. bubbles or other gaps form very effective local insulators. so when applying goo, a very effective technique is to "wet" both surfaces by applying just a smear of compound or grease, then put a blob in the center. when the surfaces are mated, parallel, the blob will spread radially and avoid bubbles. something like a hair in the goo is bad news (thought a cleanroom is not absolutely necessary ;)

the so-called thermal pad is normally a piece of waxy phase-change material (ie "it melts"). this is very convenient for vendors. I don't know any reason why either the waxy stuff or the grease-based goo can't be reused, as long as a voidless layer is achieved.

some older types of grease-based goo actually _do_ degrade over years of heat, since they were substances like zinc oxide suspended in a thin, somewhat volatile grease. when you see hard transfer material, you want to remove and not reuse it. the waxy stuff can be reused, though I don't usually bother, since a syringe of good stuff is only a buck or two per application.

cascade9 09-03-2011 11:54 AM

Quote:

Originally Posted by jefro (Post 4460124)
Less is more in this case. You don't need to remove old grease. It doesn't go bad.

Actually, thermal paste can go 'bad'. As the paste dries out, it can slowly lose heat transferring capabilities.

Quote:

Cheap thermal compounds can dry out after a few thousand hours, which causes problems for a few reasons.

First, wet compound that's dried out won't have the density, and thus the thermal conductivity, that it had when it was wet. These compounds aren't a solution, they're a mixture; remove the liquid from them and you end up with lots of little spaces where the liquid was, which is bad.
http://www.dansdata.com/goop.htm

kilgoretrout 09-03-2011 12:12 PM

For any future readers encountering this problem, here's a link to the Arctic Silver instructions recommended by the manufacturer:

http://www.arcticsilver.com/methods.html

Arctic Silver is one of the leading brands of thermal paste for cpu/heatsink. There are others on the market and all the manufacturers have instructions on their websites. There's a lot of folklore out there regarding cpu/heatsink thermal paste but you are probably better off just following the manufacturer's instructions. Note that the instructions vary depending upon the cpu. Also, if your cpu is still under warranty, using a third party thermal paste will void your warranty.

bholland2 09-03-2011 12:45 PM

Actually, going back to the old Celeron 300 and 400 Mhz. days, there is no thermal grease necessary if you don't overclock. Yip, I remember them thar days. Intel used to ship a Celeron and a cooler and there was no thermal grease necessary, the metal to metal contact was all that was necessary to cool that CPU.

I had 2 of them, used them for about 4 years in an office environment then used them for folding@home for a couple of years till they just became so out of date and slow they just weren't worth the electricity to run them. I still have them in the basement and drag one of them up from time to time to test questionable power supplies.

jefro 09-03-2011 02:19 PM

Part of the issue is what side the chip is really on. Original packages had the chip inside where many of the newer ones are on top to give direct contact with the heat sink. Thermal paste is only meant to fill the very small voids and leave direct contact with as much metal as one can. Less is more and the stuff has been silicon based as far back as I ever saw. Silicon and some metal like zinc or maybe copper in some brands.

replica9000 09-03-2011 04:40 PM

I remember building PCs where all I needed was a heatsink, no fan, and thermal compound was optional (though it usually held the heatsink in place), but this was back when 50mhz was fast!

cascade9 09-03-2011 09:02 PM

Quote:

Originally Posted by bholland2 (Post 4460555)
Actually, going back to the old Celeron 300 and 400 Mhz. days, there is no thermal grease necessary if you don't overclock. Yip, I remember them thar days. Intel used to ship a Celeron and a cooler and there was no thermal grease necessary, the metal to metal contact was all that was necessary to cool that CPU.

You might have got away with no thermal grease with fairly low speed celery, but AFAIK you were still meant to use thermal grease.

Even the horrid original celeron 266 and 300 ('Covington') had thermal grease installed 'stock'. They had no L2 cache, so they ran cooler than the later cached celerons.

Quote:

Originally Posted by jefro (Post 4460606)
Silicon and some metal like zinc or maybe copper in some brands.

Silicon is the most common (and cheapest), zinc oxide is around, copper I've only seen once (GeIL). Silver is pretty common now, has been for years.

bholland2 09-03-2011 11:48 PM

Actually it was the first copper Celerons that needed no grease, they just ran that cool. They came with CPU and heat sink from Intel with instructions pointing out that nothing else was needed. OEMs shipped pre-built computers that way too. The Coppermine cores ran a lot cooler than the older Mendocino cores. All older versions used aluminum conductors, the advent of copper really cooled them down. If we were still using aluminum around 1.5 to 2 Ghz would be the speed limit, copper lets us go much faster because of the reduced heat generated.

I was thinking earlier that the switch to Coppermine was at 300 or 400 Mhz, I was wrong, it was at 500 Mhz.

Heat is a matter of both speed and the manufacturing technology used. The last of the older Mendocino cores (500 Mhz)had a max dissipation of about 27 watts, the newer, faster, Coppermine cores (533 Mhz)maxed out at about 11 watts. It doesn't take much cooling to deal with only 11 watts.

Remember also that the Celeron had thermal protection built in so that if it started getting too hot it would just lower its clock till it was "thermally happy". This was never intended to be a high performance CPU.


All times are GMT -5. The time now is 11:13 PM.