Applying Thermal Compound

[Draconian]

Just figured might be some great info out there from fellow crunchers.
I'll post what works for me - and you do the same. There seems to be so many "right" ways to do it!

Heatsink - clean - strongest rubbing alcohol - not the normal stuff - that you can find. Same with CPU - clean it up. Don't waste money on "special" cleaners - pretty much the same stuff.

Put a plastic bag on your finger - rub a very small amount of TC into the contact area of the HS. Lightly remove with a paper towel (it should be hazy looking).
For the CPU - pretty much the same process - but do not remove with paper towel. Going for full coverage and thin - make sure on both the HS and CPU to rub it in nice and good. Don't worry about it being "too thin" - if it is rubbed in and fully covered - you are fine. Too thick is NOT good!

Assemble - correctly. Reading some of the instructions they give is...a challenge to say the least. If it doesn't make sense or the CPU doesn't seem to be making good, solid contact - revisit - do not power the system up if there are doubts!

Use quality components, of course. I use Zalman heatsinks, but there are a lot of them out there that are good as well. I still use AS5 - have for years.

Also - well - some of the manufacturers will say to just put a pile of TC in the middle of the proc and clamp it down. Really? Too much - and it spews out - and it can be conductive. Too little - and I am buying a new proce...wait...I figured out why they say that - never mind!

Please post your method and component selection. This is still applicable to those using water - except you are cheating.. <laughing>

[zombie67]

I have always used Arctic Silver 5, and follow their instructions. No problems to date.

They have instructions with pictures, by specific CPU model. I think what you describe is the "surface spread".

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

[Fire$torm]

I use Arctic Silver 5 and cover the entire surface of a CPU/GPU using a small plastic spatula. I make it as thin as possible. I then attach the coldplate starting on one edge and rotate the cooler downward, think of walking, as in heel to toe movement. This prevents air pockets. Before bolting down the cooler I give it a twist, just a quarter turn and back with light pressure to aid seating the coldplate. Never had an issue with this method.

[Draconian]

I have always used Arctic Silver 5, and follow their instructions. No problems to date.

They have instructions with pictures, by specific CPU model. I think what you describe is the "surface spread".

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

Just reviewed the document - pretty similar. They have a person use a credit card to spread and push the TC into the processor and HS - tinting. I just rub it in with a plastic bag and a fingertip. They also have a person use a line of paste and rely on the proc to spread it out - no doubt that will likely work fine, but I prefer to do it myself.

Either way. Just so a person is careful and does it proper - the method is likely not all that critical. Just...don't do it wrong (had a friend that bought a new HS - had clear plastic on the mating surface to protect it - well - he didn't take it off...)

AS5 always here as I have mentioned. There MAY be something better out there now - but - I KNOW AS5 works very well.

[Draconian]

I use Arctic Silver 5 and cover the entire surface of a CPU/GPU using a small plastic spatula. I make it as thin as possible. I then attach the coldplate starting on one edge and rotate the cooler downward, think of walking, as in heel to toe movement. This prevents air pockets. Before bolting down the cooler I give it a twist, just a quarter turn and back with light pressure to aid seating the coldplate. Never had an issue with this method.

Yep, good point about walking the cooler and seating it with a twist. Anything that prevents air pockets.

[zombie67]

I think the point of the different AS application methods: we see a square of metal, and assume that the heat from that square comes out evenly. It reality, the CPU is a small chunk underneath that square. The shape of that chunk, and the kind of heat it gives off varies from CPU to CPU. So one method of thermal grease across the top of the big square is not appropriate for all CPUs. It needs to be tailored to the specific CPU. For example, all 1155 socket CPUs look the same. Yet AS recommends different applications depending on the specific CPU.

[kmanley57]

I have learned/used both methods described, and if done right both work. But you must make sure to remove the air pockets between! First thing taught in H.S. assembly 101! :p

[Fire$torm]

I think the point of the different AS application methods: we see a square of metal, and assume that the heat from that square comes out evenly. It reality, the CPU is a small chunk underneath that square. The shape of that chunk, and the kind of heat it gives off varies from CPU to CPU. So one method of thermal grease across the top of the big square is not appropriate for all CPUs. It needs to be tailored to the specific CPU. For example, all 1155 socket CPUs look the same. Yet AS recommends different applications depending on the specific CPU.


I have learned/used both methods described, and if done right both work. But you must make sure to remove the air pockets between! First thing taught in H.S. assembly 101! :p

True, true.

@Zombie: Not trying to pick a fight (honestly). I understand that perspective and no doubt it has merit. Its just that I look at from the standpoint of thermal conductivity. Basic rule: The larger the thermal conductive area, the more heat energy that can be transferred in a given unit of time. So this also holds true for total contact area between cooler coldplate and CPU/GPU. It is true that CPU/GPU coolers are not uniformly conductive across the entire plain of contact. But, since virtually none of the 3rd party coolers are designed only for one specific CPU/GPU, it greatly diminishes the potential of Arctic's methodology (as described in that document) across all 3rd party coolers on the market. At least that is how I see it.

Of course I could still be wrong....... :P

[zombie67]

All true. The best thing to know:

No air bubbles! Once you touch one to the other, never pull apart at all, and then try to push together. Ever. Clean it all off and start over.

[STE\/E]

I use the Arctic Silver 5 too, I just slap a dab on, smooth it out with my finger, lick off the excess, taste pretty good really lol & slap the heat sink on. Unless I see a abnormal reading or a big difference in Temps between the Cores I go with it. I've never had to pull one back off yet & redo it & the Q6600's I put together ran for over 7 years with my slap happy method, overclocked too for about six of those years ... :D

[Mumps]

The other key thing to remember is that metal-to-metal contact will always conduct better than metal-to-something-else-to-metal. The whole idea behind thermal paste is that it's simply more conductive than air, not because it's more conductive than metal-to-metal contact. So you're simply trying to fill in those air pockets that are created because the CPU and heatsink coldplate will always have microscopic variations in the surface that lead to gaps between the two. That's also one of the reasons the really insane OC's will spend the time to polish the surfaces in attempts to get as much metal to metal contact as possible.

[Draconian]

The other key thing to remember is that metal-to-metal contact will always conduct better than metal-to-something-else-to-metal. The whole idea behind thermal paste is that it's simply more conductive than air, not because it's more conductive than metal-to-metal contact. So you're simply trying to fill in those air pockets that are created because the CPU and heatsink coldplate will always have microscopic variations in the surface that lead to gaps between the two. That's also one of the reasons the really insane OC's will spend the time to polish the surfaces in attempts to get as much metal to metal contact as possible.

I agree. TC is also known as "thermal interface material" as in - an interface. Pure metal to metal will always be best. Fortunately, the HS manufacturers have been, for the most part, paying attention and making contact surfaces pretty smooth (for mass production anyway). I would like to develop a melt once TC - pretty much like solder - it would melt when your system hits a temp - then solidify. To melt again would require a higher temp (which would likely never be seen). I know - that's pretty much the idea of that bubble gum crap - except..it's for sh*t.
There are other solutions available - but we always want to be able to separate the HS from the proc - if that were not the case, a pretty much perfect solution could be found.

[Draconian]

I use the Arctic Silver 5 too, I just slap a dab on, smooth it out with my finger, lick off the excess, taste pretty good really lol & slap the heat sink on. Unless I see a abnormal reading or a big difference in Temps between the Cores I go with it. I've never had to pull one back off yet & redo it & the Q6600's I put together ran for over 7 years with my slap happy method, overclocked too for about six of those years ... :D

What do you consider a big difference between the cores?
And..I am thinking..everyone...uses AS5....

[Draconian]

True, true.

@Zombie: Not trying to pick a fight (honestly). I understand that perspective and no doubt it has merit. Its just that I look at from the standpoint of thermal conductivity. Basic rule: The larger the thermal conductive area, the more heat energy that can be transferred in a given unit of time. So this also holds true for total contact area between cooler coldplate and CPU/GPU. It is true that CPU/GPU coolers are not uniformly conductive across the entire plain of contact. But, since virtually none of the 3rd party coolers are designed only for one specific CPU/GPU, it greatly diminishes the potential of Arctic's methodology (as described in that document) across all 3rd party coolers on the market. At least that is how I see it.

Of course I could still be wrong....... :P

Wherever it is warm, it should make good contact with the HS. True - directly over the core is most critical - but the rest of the package heats up as well. Any heat that can be removed from the proc package is a good thing. Simply why I make sure to cover the entire proc package manually. After all - there are NO heatsinks made that are the size of the core (tiny). Back in the day before the current packaging, of course, we applied TC directly to the proc core - now..we apply it..to metal packaging. (at least on my 3930k)

[STE\/E]

What do you consider a big difference between the cores?
And..I am thinking..everyone...uses AS5....

5c isn't abnormal because of how hard each core might be working, 10c might be especially if your having problems like shut downs or BSOD. I hardly ever check core temps anyway, only if I start having problems with a Box might I think of checking for the CPU overheating ...

[John P. Myers]

I have learned throughout my life that i do things differently than most anyone else, and applying TIM doesn't seem to be any different :p

My method: for dual cores, i run a line of AS5 (yeah i use it too) from top to bottom down the center. For quad cores, i run 2 lines from top to bottom each about 0.75cm to the left and right of center. Then i do the twisty thing like F$ does to get it all spread out.

And there is some stuff better than AS5 and it's also made by AS, but i can't remember what it's called. I do remember it's more expensive though.