Update....Thermal paste..best way to apply?

[silicon]

After viewing some videos on how to apply thermal paste where most of them apply the paste as a little blob in the middle of the cpu it occurred to me this is not the best way. I thought that it may be better to spread the paste around on the cpu so all parts of the cpu are covered. Or on the bottom of the heatsink? Does this make sense?

 

[nickbits]

The die is in the middle and it spreads out enough when you lock it in place.
There might be a slight difference using your technique but it wouldn't be measurable in most cases.

 

[RaistlinZ]

Most people recommend the blob in the middle. I use the spread method since I've always had good results with it, but I acknowledge the logic in the blob method.

Just make sure you don't use too much or too little and you can't go wrong either way.

 

[serpretetsky]

https://www.youtube.com/watch?v=EyXLu1Ms-q4

 

[silicon]

https://www.youtube.com/watch?v=EyXLu1Ms-q4

thats an interesting video. Helps to explain the propagation of thermal paste on the heat sink!

 

[PhIlLy ChEeSe]

I use the rice bead method, I have seen some board where it looks like the TIM was slathered on. You DO NOT want paste in places like the CPU socket, or on the hold down. When it gets on your fingers and you then touch the board,(unless its non conductive)you just shorted out your mother board.

I watched that video, I will try the X as it seemed to cover more surface then the rice bead.

 

[ElFenix]

I usually do rice bead method, but the last heatsink (hyper 212) I pulled off only had 2 of the 4 direct contact heatpipes with paste on them. The gaps between the pipes and the base are pretty large, and the paste didn't migrate across them. so I guess we should take that sort of thing into consideration

 

[BonzaiDuck]

I generally take a few more pains with this task than most, but always less than a half-hour's time. This is especially so, because I don't bother with conventional formulations: IC Diamond has been very good, and I have few grams of industrial micronized diamond powder for which those other formulations can serve as a base.

Everybody knows that ICD spreads like pavement cement. I'll drop a 1/4" bead on the center of the IHS, spread it from the center outward with a razor-blade, put a smaller bead on the bottom of the heatsink and do the same.

 

[crashtech]

One thing the video doesn't take into account is that the heat produced while operating reduces the viscosity of the compound, allowing to spread out much farther than when it is cold.

 

[master_shake_]

I usually do rice bead method, but the last heatsink (hyper 212) I pulled off only had 2 of the 4 direct contact heatpipes with paste on them. The gaps between the pipes and the base are pretty large, and the paste didn't migrate across them. so I guess we should take that sort of thing into consideration

you've got to fill the gaps first.

https://www.youtube.com/watch?v=IG_GfSTKQ98

 

[DrMrLordX]

Depends on the paste and the cooler. Your standard HSF with a solid base with your standard grey/white TIM works best with the grain o' rice method. If you start using non-standard stuff (like the TIM referenced by BonzaiDuck), or HDT HSFs, then the instructions change a little.

OP, if you could tell us which TIM you're using and which HSF you're using, perhaps we could give you better advice.

 

[PhIlLy ChEeSe]

I usually do rice bead method, but the last heatsink (hyper 212) I pulled off only had 2 of the 4 direct contact heatpipes with paste on them. The gaps between the pipes and the base are pretty large, and the paste didn't migrate across them. so I guess we should take that sort of thing into consideration

So its a good ideal to remove and check the spread of TIM(advice given and taken), as I am always learning..............

 

[thescreensavers]

A good article on this

http://www.hardwaresecrets.com/article/What-is-the-Best-Way-to-Apply-Thermal-Grease-Part-1/1303/1

 

[BonzaiDuck]

Either the pressure of the heatsink mount or a thorough razor-blade application is needed for the diamond paste, and I wouldn't rely only on the former. It's not going to spread more pervasively with higher temperatures. I think the Great De-Lidding experiments proved that it doesn't "pump out." The experiments also concluded with a less pleasant discovery that micronized diamond on bare silicon die had certain risks.

But that's a long stretch from the OP's needs and questions.

 

[DrDoug]

BC (Before Computing), I used the spread method on other electronic components that required a heat sink paste and never thought about it. It wasn't a big thing as long as you had a good bond and little excess, which was easy to see. So when I got into computing I did the same thing when heat sinks eventually came along for our procs. Same thing, I never even thought about it. As I got into overclocking and processors were getting hotter and hotter, I experimented and found that air was getting trapped in between the sink and proc. I tested the spread and 'pea' methods of application using two 1" square x 1/2" thick polished aluminum plates, a small spring clamp and a bench vise.

Using the spread method between the two metal plates, I used the spring clamp to squeeze the two metal plates together (simulating a clamped heat sink/CPU). I then removed the spring clamp and clamped the plates in the vise and gave it a good crank. You could clearly hear the 'popping' sound of air leaking out from between the plates as I applied pressure. I pulled the plates out, cleaned them up and tried the spread method again, this time using a hair dryer to heat the plates up while being compressed by the spring clamp. I got them good and hot and then let them cool back down before doing the vise test. Same thing, the popping of air escaping from between the plates.

I did the same thing using the pea method (without heat) and I heard nothing at all, nice and quiet. I tore the plates apart, cleaned them and tried the pea method again but with heat from the hair dryer as in the second test above. Still, no sound at all of air escaping! The reason I tried again with heat was because when I tore the two plates apart it was clear that the pea method was skipping the corners of the plates and I wanted to see if the paste would spread out with heat. It did but not enough to reach the corners. It wouldn't do any good to use more paste as it was reaching the sides fine with just a trace squeezing out from between them.

I decided to try the test again, this time with a small pile of paste roughly in the shape of the square plates rather than a small round pile. Bingo, no air and good coverage! Ever since then I shape the pile in the shape of the IHS (or die) and have been happy with the results.

Another result of the test is that I now use a hair dryer to 'break in' a new paste job before firing it up.

 

[rchunter]

Lately i've been using AS5. They have specific directions on their website. I just follow those.

 

[Aikouka]

It's important to keep in mind why we use thermal paste to begin with. Thermal paste isn't some magical heat transfer medium; it's actually worse than the metals in your heatsink. That's the reason why using too much thermal paste in that link posted above caused an increase in temperatures. The point of thermal paste is to make up for imperfections in manufacturing (the metal on your heatsink will not be perfectly flat). If you read some heatsink reviews, you tend to see them look for "mirror shines" on the bottom, but it's not because they needed to fix their hair. A mirror shine usually indicates that lapping or smoothing was part of the manufacturing process.

So, in short, you want enough thermal paste to fill in any imperfections in the metal to avoid any air gaps, but if you put too much, you're going to hinder the heat transfer.

 

[aigomorla]

It's important to keep in mind why we use thermal paste to begin with. Thermal paste isn't some magical heat transfer medium; it's actually worse than the metals in your heatsink. That's the reason why using too much thermal paste in that link posted above caused an increase in temperatures. The point of thermal paste is to make up for imperfections in manufacturing (the metal on your heatsink will not be perfectly flat). If you read some heatsink reviews, you tend to see them look for "mirror shines" on the bottom, but it's not because they needed to fix their hair. A mirror shine usually indicates that lapping or smoothing was part of the manufacturing process.

So, in short, you want enough thermal paste to fill in any imperfections in the metal to avoid any air gaps, but if you put too much, you're going to hinder the heat transfer.

too much thermal paste is no good..
too little thermal paste causes air pockets...
its about finding a middle ground.


And i dont agree with that heat pipe method..
You dont need all the entire surface to touch on a heat pipe.. just the actual heat pipe itself.

If you ask me how i do it... i draw a smiley face on my IHS.. i use the rice grain method... a dab in the middle and then compress and give it a small rotation then mount....

If its on a heat pipe cooler with exposed heat pipe.. i put a dab in the middle of each heat pipe, and then compress..

 

[DrMrLordX]

Oh yeah? Well I paint mine on with little paint brushes that have had most of the brush cut off with scissors (to prevent the brush from soaking up too much of the TIM). That's what happens to you when you're a CLU addict.

 

[BonzaiDuck]

Oh yeah? Well I paint mine on with little paint brushes that have had most of the brush cut off with scissors (to prevent the brush from soaking up too much of the TIM). That's what happens to you when you're a CLU addict.

Definitely! "Little paint brushes" trimmed to a crew-cut for CLU.

CLU is good.

Diamond paste is a close second. With diamond paste, it doesn't much matter if you put on too much. When it's clamped, it may squeeze out the excess in a bead around the processor, but it will leave a thin layer of immoveable micronized diamond all over the metal surfaces. If course, it's better not to use "too much," because ICD costs a bit more than your mainstream thermal formulations.

 

[BroHamBone]

Man, so many different ways! I have done rice bead in the middle, four rice beads (like the #4 on a dice), line horizontally, spread evenly over the whole chip (which was probably the worst).

Im probably going to redo mine in the fall. GC extreme

 

[PliotronX]

It's important to keep in mind why we use thermal paste to begin with. Thermal paste isn't some magical heat transfer medium; it's actually worse than the metals in your heatsink. That's the reason why using too much thermal paste in that link posted above caused an increase in temperatures. The point of thermal paste is to make up for imperfections in manufacturing (the metal on your heatsink will not be perfectly flat). If you read some heatsink reviews, you tend to see them look for "mirror shines" on the bottom, but it's not because they needed to fix their hair. A mirror shine usually indicates that lapping or smoothing was part of the manufacturing process.

So, in short, you want enough thermal paste to fill in any imperfections in the metal to avoid any air gaps, but if you put too much, you're going to hinder the heat transfer.

It makes sense as in the ideal world we would have full metal contact and no need for a TIM. In the real world, the dollop method, whether rice, line, X, along the HDT, has always been superior to spreading particularly with noobshields in place. The air pockets left by spreading manually end up acting as insulators. This is all demonstrated by the Youtube videos nicely and reflects temperatures I have seen since applying TIM in the 90s.

 

[BonzaiDuck]

It makes sense as in the ideal world we would have full metal contact and no need for a TIM. In the real world, the dollop method, whether rice, line, X, along the HDT, has always been superior to spreading particularly with noobshields in place. The air pockets left by spreading manually end up acting as insulators. This is all demonstrated by the Youtube videos nicely and reflects temperatures I have seen since applying TIM in the 90s.

I always lap my HSF-base and IHS with two objectives: make them absolutely flat, and eliminate the nickel-plate. Even I have become more reticent about lapping the IHS, with any thought of reselling a processor. But if I have even a temptation to OC the system later -- I do it.

 

[LV3]

I use tek syndicates 'metal method'

https://www.youtube.com/watch?v=-hNgFNH7zhQ