How to Correctly Apply Thermal Grease? (CPU)

Hi all,

I just started taking my Asus M51Sn apart today, in order to re-paste Thermal grease on the CPU. I generally do know what I am doing, I just want to be sure.

1. Take off cooling assembly; radiator and heatsink.
2. Remove old thermal grease off heatsink. (Remove it off CPU core as well I am assuming?)
3. Apply some Arctic Silver 5 compound thermal paste on top of CPU core. (Thin layer!)
4. Carefully reattach heatsink assembly, and screw in place. (As well as other cooling components...)

Also, how important is it to remove every last bit of the old paste? I still have a little bit remaining, and was wondering if it mattered too much.

If anybody more experienced with this, could add a little more detail to help me do it just even the tiniest bit better would mean a lot. Thanks!

PS: If anybody is interested, the CPU in question is a T9300. I run it overclocked at 2.8Ghz (from 2.5Ghz stock) and under load (gameplay) it reaches 70 degrees celsius with maximum cooling (as well as cooling pad.)

My main objective is just to lower that by a few degrees, if possible.

 

You sound like you know what you are doing. Just go for it. If you don't see the improvement you were hoping for, try again. Also try to keep that thermal past on the CPU and not let it go onto other parts.

 

You seem on the the ball. Yes, make sure to thoroughly clean both CPU die and heatsink surfaces. You want it nice and shiny.

Qing Dao and I are also proponents of the spread method on bare dies.

And while you're in there, might as well blow out the radiator.

 

Still dont recommand the spread method, those two guys are skilled and they can handle it well. However, if u are doing this first time, the dot method is safer and much easier to apply.

 

Doubtful, even them two have air bubbles. Would not recommend that method, ever. Dot method guaranteed pretty much to not have air bubbles. Spreading guarantees it...

Sent from my G73JH using KeyBoard

 

Incontro, there was a recent thread about this, read it here.

Read the arguments and decide for yourself.

 

Just be careful you don't use too much. You don't want the arctic silver to squeeze out and get anywhere. It takes a lot less than you think.

You really have no clue. The person who made that youtube video was clearly either a nitwit or trying to fake the results. And those were not air bubbles, it was a gap between the clear plastic and the heatspreader.

 

Or use something that's very easy to spread (and not capacitive, a big plus)like Gelid GC-Extreme. Smoothest TIM I've used, ever.

 

Never Spread It, Ever!!!!!

How Thermal Compound Spreads - YouTube

How Thermal Compound Spreads - YouTube

< width='640' height="480" src="http://www.youtube.com/embed/EyXLu1Ms-q4?rel=0" frameborder='0' allowfullscreen></>

EDIT: that is not fake and is not gap, you have never lapped an IHS before it's bowl shape not a spotted and dotted, those ARE air bubbles.

EDIT EDIT: found my write up, i suck at writting but, i tried

[quote name="tpi2007" url="/t/1261817/purchased-a-used-q9550/10#post_17336103"]
What on earth are you talking about ? You don't address any of my points (except to say nr. 3 doesn't make sense and is incoherent, without explaining) and just talk prejudice against the method ? LOL
The rice grain is the method where you can actually make a mess if you are not careful, not the credit card method. You got it completely wrong.
I would spend the extra time and effort because that way I know exactly where the TIM went, because I put it there with the thinness I want, and it won't spill over, unlike the rice grain method, which is especially important if you are applying to a bare die GPU, for example, that has small capacitors nearby. If you use AS5, for example, which is capacitive, you can cause a short circuit and damage the card.
Putting a real heatsink is not the same as putting a piece of transparent glass / plastic, where you can see where the paste is going and apply pressure where necessary to compensate. LOL
For your information, IHS stands for Integrated Heat Spreader. It spreads the heat, so why on earth won't I take advantage of all that surface area AND the surface area of my heatsink in order to maximize heat transfer ? If your argument had any logic whatsoever, CPU heatsink makers would make the contact surface of their heatsinks way smaller as the CPU die is smaller than current and past CPU heatsinks.
And yes, number 3 makes perfect sense, when you are securing a heatsink to a CPU socket for example, while you are securing on one side, it may, even if just slightly, lift on the other while you don't secure that side, and air gets in. It's simple.
Do you have any idea what you are talking about or you just decide to come here and talk prejudice and speaking in an insulting way (and leave religion out of this!) about a method you obviously have never used ? I've been using this method since the Pentium 4 days, on both CPUs and GPUs and it's not like I do it every day, so I can very well afford the extra time it takes. LOL It's not like people on this forum don't spend hours sleeving PSU cables. Go criticize that, will, you ?[/quote]

 

I just have this to say:

Can we please stop using desktop CPUs with IHSs as examples? We're dealing with the notebook world, with bare dies, heatsink insertion angles, and lower mounting pressure.

 

If it's easier to take the heat sink off just try both methods and check the temps, I would say the dropping a small amount and letting the pressure of the assembly spread method works fine for me.

 

I've tried both. I actually found the blob method to be slightly better. Just make sure both mating surfaces are 100% clean (use alcohol or something similar) and make sure there is no dust.

 

cdoublejj, yes I have lapped heat spreaders before, but I prefer to remove them. Those are not air bubbles trapped between the two surfaces. If you look closely, you will see that at first there is no air in between as the person presses down. Then you will see that the person releases pressure, and the plastic piece comes up a little bit. The pattern of the thermal paste is not air bubbles, but how it is when you lift a heatsink off. So the plastic piece is left resting on the edges of the heatspreader and the thermal paste was squeezed out too much so that it can no longer fill the gap, which is replaced by air.

If that was how the spread method played out in real life, at all, it would not be usable. It is trickier to apply, especially on a processor with a heatspreader. But the results are at least as good as the dot method.

I agree with everything you say about thermal paste and that stuff except that the spread method traps air.

 

Thanks for all the replies guys. There is no harm in having many differing expert opinions.

After reading this thread, and the linked thread, AS WELL AS the Arctic Cooling Website, I have decided to go with the spread method. I have a small plastic knife (like the ones in takeaway shops) and I will use that as my spreading tool.

I have actually used the spread method for Arctic Silver Thermal adhesive before, and did a decent job of it. I actually quite liked the look of the dot method on the attached YouTube link, but a lot of sources claimed that the spread method was more suited to laptops.

The Arctic Silver link is here if anyone is interested, I have a Core 2 Duo T9300. But you all probably have seen this before.

Thanks again.

 

I've always spread the paste, and always had good results.

 

Try it one way, see results, if it satisfies, then leave it, if not or doesn't seem right, do it again another way. I've repasted my own machines multiple times when things just don't seem right.

 

Well i'll tell you this much, if you do the rice grain method properly the heat sink will be stuck the CPU with out any screws or mounting and with the spread method the heat sink falls/comes right off. I have also noticed the spread method yields thicker spreads, TIM is good but, not too much.

I have also noticed the TIM separates differently when removing the heat sink with rice grain versus spread.

I'm not talking about 4 or even 10 heat sinks i'm talking so many i don't even count any more.

I've also noted petrified air bubbles in the really thick spreads so think and bad that virtually NO tim is deposited on the heat sink or CPU and it is all stuck on one or the other and there are VERY NOTABLE pits in the TIM.

You make a good point how is it usable and practice, though i can prove it is un practice. wth waste the the extra effort for the same or less temps?

I think it may traps air bubbles it may not be enough to completely hinder temps. In the video if you look at the spread method, it overs the entire IHS versus the center (drop/rice grain) so even with the air bubble there is just enough surface area contact to conduct enough heat but, not all of it is being used. That is one NASTY looking spread.

EDIT: though i'm not sure how well they spread but, either way in my experience it certainly doesn't age well and is a waste of extra effort.

You could side step the hole debate and get torch super heat some solder so that it stay molten for a usable time and after removing the IHS just solder the heat sink straight on. If you get it hot enough it will stay molten for a good 20/30 seconds.

 

LOL. And I would NOT recommend this to ANYONE. Although could be fun. Maybe will try with an older desktop PC I have.

 

OK, just an update on the situation.

I have applied the paste recently, and everything is done.

I am monitoring my temps closely, and so far my CPU temps seem very similar to what they were before, too hard to notice any difference. I know there is a "break in period", so I will wait longer until making any more assumptions about the CPU temperatures.

However, I have never monitored my GPU temp before, and now idling is says 68C. Is that too hot? Also, is it safe to game? After some stress, I want to keep monitoring GPU temperatures.

CPU temperatures are fine now; 45-50 Idle. Will report back when under stress.

EDIT: I did not touch the GPU paste, maybe I should have done that too?

 

You might as well go back in there and do the GPU if it is not too difficult.

 

Another update...

So far I am pretty sure that:
9500M GS - 68C - Idle
9500M GS - 90-95C - (Around) Full Load

Is it really worth opening it up again? And if yes, do I just follow the exact same procedure I used for the CPU, or is it something entirely different? (IMO the load temps seem a bit high.)

Regards.

 

I would say yes, go for it, but then again I'm a nut for opening machines up to make them cooler.

And process is the same except... Hmm. When you're opening it up you might notice thermal pads on the memory blocks. Leave those be.