Sign in with
Sign up | Sign in
Your question

Thermal Compounds-what, and what method is right?

Last response: in CPUs
Share
January 13, 2008 3:15:08 AM

I am building my new PC this week and am wondering what Thermal Compound to use. Usually everyone says Arctic Silver but I have found a bunch of reviews using google.com that say Cool Silver and perhaps others are better. Anyone have any first hand experience with that? Is the BB sized dab and credit card spread the best? Straight line then Heat sink on and twist 1-2 degrees? THat doesn't seem like it would spread out well enough but the Arctic Silver people say it for dual cores.
a c 265 à CPUs
January 13, 2008 3:43:11 AM

I think most of the compounds are good. It probably matters more how well you mount the heat sink. Also, the as5 is supposed to reach it's best efficiency after a number of start/stop cycles(200?). The application method on the as5 web site(stripe down the middle) is probably as good as any. They should know something about their product.
January 13, 2008 4:20:20 AM

i am right
i am always right
i am always right accept for when i am wrong
i am right when i am wrong since i said i was wrong therefore i am right
hence i am always right even when wrong!

spread it from the center and cover the entire chip - it sometime can make a difference. if the HS is dished after you trowl it evenly add more to the center and move toward the edges so it evenly coats and its slight mounded in the middle!

Dragon method of appling thermal compuond!

or Dmatc
Related resources
January 13, 2008 4:31:14 AM

i always get high temps when using the thermal compound based on the instructions at arctic silver site. Usually for me to get low temps, i cover the whole IHS with the compound and flatten it as evenly as i can with some hard plastic like a credit card or a plastic putty knife.

Others told me to just put a small amount, much like an uncooked grain of rice on the middle. By putting the heatsink on top of the processor spreads the compound evenly, but that advice could be effective about 4 years ago when i still have a processor with a single core. Maybe one time ill try with four small amounts exactly on top of the chip.
January 13, 2008 4:39:17 AM

exactly wbirkin

i learned the method by trial error, artic silver is great stuff but it does not always flow in the right direction

thats why my method "dmatc" works well with every hsf
January 13, 2008 4:44:16 AM

i use AS5 and the rice grain down the middle method, my temps are 25c idle and 35c load on Orthos. Using overclocked x2 3600+ at 2.4ghz(1.9ghz stock) on a zalman cnps9500led. prior to this setup i had an Antec formula 5 thermal compound a gigabyte g-power cooler @ 2ghz and my temps would be 32c idle and 49-52 load Orthos. so in my view AS5 is good, but i did chenge my cooler also.
January 13, 2008 4:45:46 AM

May I suggest Arctic Cooling MX-2? It's non-conductive, does not require burn in time to acquire best temps, and is very easy to apply.

There are numerous reviews which compare MX-2 to AS5 with results giving the MX-2 compound lower temps.

http://aphnetworks.com/reviews/arctic_cooling_mx_2

http://www.mvktech.net/content/view/3568/39/1/3/

http://www.goldfries.com/hardware-reviews/arctic-coolin...

http://www.ocia.net/reviews/acmx2/page1.shtml

http://www.tweaktown.com/reviews/1148/arctic_cooling_mx...

http://www.dragonsteelmods.com/index.php?option=com_con...

http://www.rbmods.com/Articles/Arctic_cooling/Mx2_therm...

In addition, I can vouch for these results as I have used AS5 and MX-2 on my Q6600. The MX-2 giving better results (about 1-3 degrees difference).
January 13, 2008 5:35:56 AM

interesting read =0
MX-2 sounds good to me.
January 13, 2008 7:12:45 AM

Hi Grundalizer

I usually put a very small (!!) pea sized amount of AS5 in the centre of the CPU/GPU/etc. and wear latex/vinyl gloves (cheap disposable kind). Then I spread out the compound evenly over the surface of the chip/heatspreader. This avoids adding natural oils from your hand to the cocktail but still gives you a lot of control over the spreading process...

Bob
February 14, 2008 1:53:58 AM

Typically I use Arctic Silver Ceramiq, put a bit on the center of the chip, and spread it around with a finger.
After that I install the HSF, and wash my hands again. (you can do the same with AS5)
February 14, 2008 2:23:24 PM

AS Ceramique is supposedly the best 'commonly available' thermal compound. I have seen the "Masscool x23" do even better, but it's tough to get ahold of. I personally am using some Antec Silver Thermal compound5 (sounds like a marketing term to get sales from confused people shopping for AS5) but, the specifications for this compound show higher silver content, and supposedly better thermal conductivity. I can't say for sure, but i got it because i split hairs on specs.

For TC distribution on a chip/HSF;

there have been reports that gob/line method is the best, but i don't buy it. i tried it myself after reading a test review on methodology, and i proceeded to get higher/more sporadic heat readings/jumps in temp.

MY method, that in my experience works best, is you place a dab in the middle, and use CARD STOCK to spread it. trimming a 3x5 card in half works excellent. Now, I think the reason why this works so well in my case, is that the card stock is fibrous and porous. it will collect the excess as you spread it, keeping the layer of TC as thin as possible while completely covering the HSF. My method of spreading, is akin to that of a monk performing his religious duties. I honestly think that because of my ....retentive nature, that it very thinly and evenly distributes the compound. If you scrape and spread to the point where one more swath of the card will undo your handy-work, then it is perfect. This method alone i believe has lead me to a 3-5C drop in average temperature readings.
February 14, 2008 2:48:00 PM

When you think about how thermal grease is used, it basically creates a seal, and fills in all the gaps, imperfections on the sureface of the Heat Sink and the IHS (intergrated heat spreader on CPU).

We are talking imperfections that the eye can not see. Once that is done, the heat produced by the CPU doesn't have to pass through any air pockets (the worse way of transferring heat through metal).

So basically you won't need very much. Like the others mentioned, the size of a rice seed in the middle, and spread it out in a thin layer should be sufficient.

Lapping (making the surface as flat as possible on the HS side and CPU side) is another method, but... to me, using thermal grease is suppose to correct that flaw. Although if the surfaces are really concaved, then the thermal grease may not be able to create a good enough seal to help transfer heat efficiently.

Also its helps to understand how the core sits underneath the IHS. So in a sense, its not really important to have the thermal grease all the way to the edge of the IHS. It is generally the center and how far the corners of the core go out wards under the IHS.



That's how a C2D die sits under an IHS. I couldn't find anything on the Athlon64 die, but the AS5 recommends covering the entire surface of the IHS.

Just my 2 cents. :D 
February 14, 2008 4:12:42 PM

If there is any major temp difference when using different brands of grease, then you are using too much of it and/or the surfaces aren't flat enough. It is meant to be used like make-up (smoothing out dings), not putty or spackle (filling huge gaps). Using a "premium" compound to fill a huge gap will never work better than sanding the surfaces flat and using a TINY bit of the cheap white stuff. If you are too lazy to do this, then by all means spend the extra money on the premium goop. Metal will always conduct heat better than the grease. For a joint with no voids, the less grease the better.

Also I laugh when people insist you need to use reagent grade isopropanol to clean off old thermal grease. You could clean it off with spit and it wouldn't matter.
February 14, 2008 4:21:46 PM

Blefuscu said:
Also I laugh when people insist you need to use reagent grade isopropanol to clean off old thermal grease. You could clean it off with spit and it wouldn't matter.


Well.. you could use spit.. but it won't get the old grease out of microscopic cavities that you dont' see.

For example, I did use carburator cleaner on a HS (aluminum) that I used AS5 on. Before I used it, I basically cleaned it with rubbing alcohol, and the paper towel still got gray/black smear on it from left over grease.

And I do believe rubbing alcohol will break grease down better then spit. :lol: 

Too lazy? What.. doing all the lapping? I'm not gonna ruin a perfect CPU/HS just because I COULD get better temps of 3-5C.

But... everyone has the pet peeves. :sweat: 
a c 123 à CPUs
February 14, 2008 4:28:39 PM

Well one thing that is nice is that Zalman includes thermal grease with its CPNS 9700. It comes in a tiny bottle with a little brush so application is easy. This being said yu could possibly ditch the Zalman paste, get some AS5, clean the brush, and use it to apply the AS5 evenly across the IHS.

That worked great for me. I got it all nice and evenly spread out in a thin layer.
February 14, 2008 4:46:43 PM

Grimmy said:
Well.. you could use spit.. but it won't get the old grease out of microscopic cavities that you dont' see.

For example, I did use carburator cleaner on a HS (aluminum) that I used AS5 on. Before I used it, I basically cleaned it with rubbing alcohol, and the paper towel still got gray/black smear on it from left over grease.

And I do believe rubbing alcohol will break grease down better then spit. :lol: 

Too lazy? What.. doing all the lapping? I'm not gonna ruin a perfect CPU/HS just because I COULD get better temps of 3-5C.

But... everyone has the pet peeves. :sweat: 


Actually, as solvents, alcohols and water (spit) are chemically pretty similar. Neither dissolves greases particularly well. Which is why it puzzles me that alcohol is recommended by manufacturers. It is more the wetting and scrubbing action that removes the grease when using these. Carburetor cleaner is a great idea. Now you're talking about a proper solvent. Acetone or maybe hexane would work well too. But both of these could cause problems if dripped onto circuitboards or plastic. Perhaps this is why alcohols/water are recommended.

I'm not trying to imply that everyone who buys the primo grease is lazy, but it is clearly not necessary for flat surfaces. It is actually the manufacturers' fault that the chip heat spreaders and heatsinks are not flat. Computer chips are particularly bad. I don't know why. Power chips such as MOSFETs always have incredibly flat heat spreaders. Manufacturing flat heat spreaders would probably drop CPU temps 2 or 3 C across the board. Perhaps it is just a case of cutting corners.

As for the spreading technique -- if there are huge gaps, then it might matter because you are moving so much stuff around when you press down the heatsink. But if the surfaces are flat then I don't think it makes a big difference as long as you don't use too much.
February 14, 2008 5:05:00 PM

Blefuscu said:
As for the spreading technique -- if there are huge gaps, then it might matter because you are moving so much stuff around when you press down the heatsink. But if the surfaces are flat then I don't think it makes a big difference as long as you don't use too much.


Agreed. :D 
February 14, 2008 9:36:11 PM

Blefuscu said:
...

I'm not trying to imply that everyone who buys the primo grease is lazy, but it is clearly not necessary for flat surfaces. It is actually the manufacturers' fault that the chip heat spreaders and heatsinks are not flat. Computer chips are particularly bad. I don't know why. Power chips such as MOSFETs always have incredibly flat heat spreaders. Manufacturing flat heat spreaders would probably drop CPU temps 2 or 3 C across the board. Perhaps it is just a case of cutting corners.

As for the spreading technique -- if there are huge gaps, then it might matter because you are moving so much stuff around when you press down the heatsink. But if the surfaces are flat then I don't think it makes a big difference as long as you don't use too much.


Here, here... couldn't agree more!!

I had my first go at lapping a dual-core Opteron (939) heatspreader and stock (heat-pipe) cooler recently. The heat spreader on the 2x Opteron had a massive pothole in the middle. With Artic MX2 and lapping (both heatsink and IHS) to 1200 gritt I am getting a load (both cores fully loaded with CPU burn) temperature of 38C (OK only at stock 1.8Ghz) - my lowest ever aircooled CPU load temp.!! That is using the stock Opteron heatsink (copper heatpipe version) with a silent Arctic 12cm cooling fan.

As stated before and because nobody seems to LISTEN on these forums... The best way to apply heatsink compound...

Adminster a small pea shapped amount in the middle of the IHS. Then put on a pair of disposable latex/vinyl gloves and spread the heatsink compound out until till it evenly covers the top of the heatspreader.

No other technique is as quick and easy as this... You also have 0% risk of contaminating the heatsink compound with sweat/grease.

Bob


February 15, 2008 6:15:11 AM

bobwya said:
No other technique is as quick and easy as this... You also have 0% risk of contaminating the heatsink compound with sweat/grease.Bob


Sorry... I couldn't stop laughing at that.

Oh no... I have sweat mix with my thermal compound. Owell.. a few CPU burns and that sweat should be evaporated. :sweat: 

Actually.. its body oils that AS5 stress on. But comeon.. how much body oil is on the tip of yer finger, to cause a catastrophic thermal phailure? :lol: . o O( just use carburator cleaner on yer finger, and your oil problem is gone for... a few mins)

Edit:

Now I did go by the directions.. and on my final use of what was left in my AS5 Tube. I said screw it... I using my finger. And so far I still see no difference in temp... what so ever. :lol: . o O (grease is made from fat or is oil based)
February 15, 2008 6:47:35 PM

Blefuscu said:
Actually, as solvents, alcohols and water (spit) are chemically pretty similar. Neither dissolves greases particularly well. Which is why it puzzles me that alcohol is recommended by manufacturers. It is more the wetting and scrubbing action that removes the grease when using these. Carburetor cleaner is a great idea. Now you're talking about a proper solvent. Acetone or maybe hexane would work well too. But both of these could cause problems if dripped onto circuitboards or plastic. Perhaps this is why alcohols/water are recommended.



Alcohol and water are completely different on a chemical and solvent standpoint. The only thing in common is that they both have a hydroxyl group (-OH). Alcohols are more like detergents than anything else since they have a polar section and a non-polar section.

Saliva isn't comparable to water, it has many enzymes that break down components and act as emulsifiers.

Thermal compounds are oil based FYI.
http://www.arcticsilver.com/PDF/as5msds.pdf

Nail polish remover would be ideal most likely because it is easy to get and use. They are usually acetone or ethyl acetate based but as said above, they will dissolve plastic.

Carb cleaners are good solvents but not good for the health...it usually has a large mixture of solvents like xylenes, toluene, and methanol (not good to inhale).

I would imagine that starting fluid spray would work well too since it is ether / heptane based. But why not front the $6 and get the AS cleaner along with your paste order if you are captain anal?
http://www.newegg.com/Product/Product.aspx?Item=N82E168...


September 26, 2008 8:49:05 AM

Xigmatek 1283 just arrived in the mail, and was wondering if the included paste is good enough to use? Is there a overwhelming reason for me to spend extra on the MX-2, or just make do with the free stuff?
September 26, 2008 9:09:40 AM

The compound that comes with it should work okay.

But it usually doesn't hurt to get more thermal grease of another brand, since you don't know if you will end up taking it back apart if something is wrong.

Over time, some thermal compounds will loose their properties to transfer heat efficiently, namely the reg white stuff. The MX-2 is suppose to last 7 years, if I remember correctly from their site. :lol: . o O (not sure if I can go that long without taking it apart for cleaning or upgrade)

I still use MX-2, works great so far.. basically same temps from moving from 650i chipset to P35 chipset.
September 26, 2008 9:14:14 AM

Thank you Grimmy. That explains a lot! I think I'll pick up some MX2 tomorrow morning. :) 
!