thermal paste on the CPU, how do you do it?

Hello,
I am about to paste a cpu. I would like to know what the best methode is for pasting and why. I have seen several video´s about it on youtube about the different methodes. I´d love to hear some input

As well any recomendations about thermal paste? What is a good tho cheap one. Im not looking for the best paste just a decent one at a low price.

thanks!

 

Best cheap one imo is Prolimatech PK1 for both desktop and mobile's. Moderately viscous, comes in 1g packages (not syringes). You can heat it up near a lightbulb after cutting the package for easy application.

On cores with heatspreader you put a moderately-sized grain in the mid of the cpu and fasten the heatsink.
On bare dies you have multiple options such as:

Spread on the die using latex gloves (make sure to wipe the extra to not have it smear).
Pour a thin line and spread it using a credit card

 

It depends on the shape of the die. If it's a square, use the dot method. If it's rectangular, make a thin line. You don't want to use too much of the compound, but it needs to cover the surface evenly. After applying, attach the heatsink and tighten the screws in an "x" pattern so pressure is applied evenly.

 

I always thought this was a interesting video.

John.

width='420' height="315"><param name="movie" value="http://www.youtube.com/v/EyXLu1Ms-q4?hl=en_US&amp;version=3&amp;rel=0"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/EyXLu1Ms-q4?hl=en_US&amp;version=3&amp;rel=0" type="application/x-shockwave-flash" width='420' height="315" allowscriptaccess="always" allowfullscreen="true"></embed></object>

 

yeah saw that one. It seems all methodes are all about equal with only the `credit card` methode being not so good because air bubbles arise

 

With Ivy and upcoming Haswell you have to use the line method since the die is long

Edit: I see you have the i5. If that is the CPU you speak off, a dot method might work since the die isnt as long as my 3610QM. But even there you might be better off with a line

 

I have seen thermal pads for gpu`s new ones that melt, supposedly great condutivity, so where are the ones for cpu`s , using paste on a cpu is so hit and miss, unless you do it a lot.

John.

 

You shouldn`t be afraid anyway if you are using a non conductive paste.
I`ve seen people who use liquid metal on the die. You gotta know what youre doing if you are gonna go that route

 

As long as you do it properly there's nothing wrong with the CC approach. I've swapped CPUs on thousands of ThinkPads using it with no issues to speak thereof.

 

I do cover all chip but dont wipe it out with credit card as this produces air bubbles. I just cover all chip and then put heat sink plate and press hard .. and that's it. Never gave me any problems. Note there will be some paste on the side of the chip but that doesnt hurt.

 

I've always used the dot in the center of the processor, simplest method that is hard to screw up (unless you empty an entire tube on). Line is similar to dot, you just end up with more of it on the surfaces.

Spreading it yourself is too hard

 

thanks for the input. I do have an i5 now but i want to upgrade to an i7. That will generate some extra heat so good paste is a must.

 

Depends on the TIM used and the cpu it will be applied on:

See:
Arctic Silver, Inc. - Instructions


Good luck.

 

This bears mentioning: whatever you end up doing will be *vastly* superior to what OEMs do (which is to glob on tons of cheap stuff).

Don't stress too much about tiny differences application methods make. Just don't use too much and you'll be ok.

 

I always just use my finger to spread the paste.

 

interesting link thanks +1!

It seems for mobile processors the line methode is recommended. No comments on Ivy Bridge tho. Thanks for all the input so far it is really apreciated

 

So who turns the heatsink 2mm to the left and right to remove air bubbles after you have put the heatsink on?

John.

 

I never measure those 2mm, but yes, I do shift the heatsink around a bit...

 

square die -> dot
rectangular die -> line

I then apply pressure on the heatsink to make it spread.

 

I hope you have a glove on, i dont want skin oil on my heatsink or cpu core.

John

 

The technique is simple; keep your hands off the die and place a blob in the middle of it. Place the heat sink at perfectly aligned angle and don't move it once in place. Don't tighten the screws beyond their normal hold and tighten them in alternative order moving in diagonal directions. There are several guides posted over at this forum where you could see the pictorial presentation of what I have explained above.

Sent from my GT-N7100 using Tapatalk 2

 

any thoughts on coolermaster premium thermal paste. It doesnt get the best reviews but i can pick it up cheap. Any thougts on it?

What about coolermaster high preformance. That seems to be one of the only kits i can buy localy which will save quite a lot of shiping costs.
http://www.coolermaster.com/product.php?product_id=149
costs 9€

 

The paste job is much more important than the paste itself. Better paste will make a slight difference, but if you need to repaste, any decent paste will do.

 

I agree. I messed around with a few different compounds and found that the proper application is more important than which TIM you use.

Here is an interesting link about random thermal compounds used.

 

Also, thicker paste usually does better if your heatsink has a poor surface, something like Arctic Cooling MX-2/4 or IC Diamond. I got slightly better results on my G73 with ICD than MX-2 but both gave me very good results.

 

thanks for the link, it seems like coolermaster high preformance is really good. I think i´ll go ahead and get that one. Thanks again.

I´ll first try it on my old Pentium 4 and check the difference in temps with and without the paste. Later when i get my i7QM i´ll know how to do it properly. My heatsink is quite odd tho. It isnt attached my screws but by clips

 

I've tried all the methods described and I use another one. I use my own fingers. I get a piece of plastic bag and wrap it around my fingers and then freely spread the paste like I want without any issues or residues. imo it's the easiest, fastest and most efficient method. it works with all kinds of pastes including liquid metals due to body temperature.

 

Seems like a quick method but I wouldn't risk it, there might be dust/dirt on that piece of plastic bag or whatever material that gets in touch with the paste.

 

I use dot method, TIM doesn't need to cover the whole area of the CPU, so less is better.

 

I use and have had nothing but good results with AS5 but would guess different TIMs work best with different methods.

For viscous sticky goopy pastes like AS5 you're causing yourself problems if you work it around. Concave surface on the paste blob is what becomes air bubble against HS as its lowered so just make a nice peak and squish it down flat with the HS.

If you want to know how much to use, do it once, take it off and have a look. Experience comes from trial and error.

I have never needed to repaste due to rising temps within 3 yrs. Other TIMs may be a couple of deg cooler but IMO they dont last.

 

People only think the "credit card method" (I use a razor blade, btw) produces air bubbles and is unsatisfactory because someone in a youtube video made a bogus visual test of it.

If you watch the video, when he is finished with the credit card method, you will see that there are not air bubbles between the piece of plastic and the heatspreader, you will see that it is a full on gap between the heatspreader and piece of plastic. The guy doing it was such a moron, because what he did with the credit card method, he didn't do with the other methods. The heatspreader in his video is concave, and he pushed down on the plastic in the center really hard until the thermal paste all squeezed out from the sides, then when he released pressure, the piece of plastic sprang back up with a huge gap under it.

If you do the credit card method in real life, you will get good results. First of all, our laptops don't have heatspreaders, they have exposed dies that are very flat compared to any heatspreader. Also we use heatsinks made of metal that don't bend so easily and that put pressure down evenly on the core. Years ago I was really into overclocking and I was repasting CPU's, northbridges, southbridges, and GPU's like a madman. I final settled on the spread method because it used the least amount of paste AND it gave the best most even results. Also I always use a razor blade, which stays straighter than a piece of plastic would. For cores without heatspreaders, plastic is fine to spread with, but for heatspreaders you need to be careful and should use something that won't bend along with the heatspreader's uneven surface.

 

interesting post man. I dont have a convential heat sink in my Dell Dimension 8250. It is hold in place by clips rather than screws so i wouldnt know if it will spread as good as with screws. So the credit card methode might not be too bad of an idea. Does anyone have any experience with Dell Dimension 8250 pasting in particulair?

 

I'm also a proponent of the spread method like Qing, at least on exposed dies. With IHS components I use the dot/line method.

Discussed here, here, and here.

In a nutshell, I use spread because I'm not working on a desktop. The componants have have an IHS, the heatsinks don't clamp down as hard as a desktops, we can't put heatsinks down completely vertically, and we don't have much room to wiggle the heatsink around.