Thermal Paste Application Guidelines

[Morg.]

Alright, second time I read weird stuff about thermal paste application, so here's my question :

- Are those line or w/e patterns any better than the old-style thermal paste "sheet" method ? (like, put some thermal paste, make a flat sheet out of it w/ your credit card, then slam the rad on it with extremely brutal force, then post vid on youtube)

 

[dawp]

I think most have application directions on their websites. I know arctic silver does.

http://www.arcticsilver.com/instructions.htm

Shin-Etsu doesn't.

 

[BoomerD]

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=170&Itemid=38

 

[Morg.]

Well thanks for the links ... however this does not answer my question, except it confirms that the old way of doing it still rocks .. or something.

Does anyone know the specific details of why/if applying thermal paste old-style would/could be worse than any of those examples ?

I'm just trying to rationalize here, are all these "modern" instructions about how to do it in an easier fashion, or a better way.

 

[bmaverick]

What I have tried and has been proven at a few of the OCing sites is applying the thermal paste like candy dots evenly in a grid pattern. of course the height is very marginal. Many of the retail aftermarket air coolers have this pattern too.

 

[Zap]

- Are those line or w/e patterns any better than the old-style thermal paste "sheet" method ?

You basically want your end result to be:

1) Full coverage of any place the heatsink base is supposed to touch the CPU heatspreader.

2) While fulfilling #1, using the least amount possible.

Whatever method you prefer just needs to fulfill #2.

Thinner thermal compounds are usually okay with just using some dots, a line or two, or a small blob in the middle since the pressure will squeeze the stuff out the sides. Of course how anal you are about your TIM can also affect how you apply it. ^_^

 

[Syzygies]

It depends on the heatsink and thermal paste.

I just installed a Noctua NH-D14 cooler onto an i7 2600K. (For air cooling, the Noctua tests second only to the Cogage Arrow, with a much better mounting system, and 2600K's are limited by voltage not heat.)

I installed it several times. Their advice was a single 4mm to 5mm dot, dead center, using their paste. Their mounting system has good pressure, and did an amazing job of spreading this dot out to the entire chip surface.

Their FAQ is quite explicit, an utterly smooth mirror finish is a poor idea, because it offers less grab for the thermal paste, making gaps more likely. Anyone ever convince you why they want the penny to reflect so nicely in their cooler base? Yeah, I didn't think so. It always sounded like one of those ideas with fake authority, a mental tar pit for OCD types.

Of course, every cooler base has a different coefficient of grab, and every paste a different consistency. Either use manufacturer-supplied paste and follow specific directions for your base, or experiment, pulling the cooler back off and looking.

 

[stargazr]

Well thanks for the links ... however this does not answer my question, except it confirms that the old way of doing it still rocks .. or something.

Does anyone know the specific details of why/if applying thermal paste old-style would/could be worse than any of those examples ?

I'm just trying to rationalize here, are all these "modern" instructions about how to do it in an easier fashion, or a better way.

I think there are at least two reasons for different strategies of applying TIM. One is the CPU architecture and where the most heat is generated. When you fire up your computer, the CPU heats up and the material flows or spreads out.

Another factor is the type of cooler, and more to the point the condition of the surface that contacts the CPU face. For example HDT coolers have grooves along the heat pipes. Check out the photos in this article.


http://benchmarkreviews.com/index.p...sk=view&id=150&Itemid=62&limit=1&limitstart=5

They try applying just one blob in the center, then take it apart to observe how the TIM spreads out. It seems like maybe if there isn't enough compound it will all collect in the grooves, and will not spread out further like it does when they apply two blobs evenly spaced.

Also I disagree that a highly polished finish is not good. Ever take a fan off a notebook? The CPU face is highly polished because the cooling environment in a notebook sucks. The purpose of TIM is not to create a complete layer, but only to fill the microscopic voids.

Ideally you want the heat to be able to radiate into the heatsink across the entire mounting face on a molecular level. Theoretically, if you had two perfectly flat smooth surfaces with no tiny voids, TIM would not be necessary.


The only way too good a finish is an issue would be if it was polished incorrectly by hand; by that I mean depending on what method you use, when polishing the tendency is to concentrate on the deeper scratches. By doing so you no longer have a flat surface.

 

[killster1]

I agree with stargazr. Heat pipe direct touch coolers need you to spread paste in the groves then on the cpu. And mirror finish is great. I have run a cpu with no thermal paste for 3 years before. was a 478 with xp90 with mirror finishes. worked perfect and very cool. (i didnt even mean to run it with no paste i just tested it real quick and never redid it after testing)

Really doesnt matter to me anymore about paste and everything, i used to care and buy really expensive pastes and shine everything but now wow 5c more who cares just use what ever ac5 i have laying around or cheap syringe of shin itsu or what ever was on sale at svc.com last time i bought some.. i spread it very thin with my finger or cut up card and call it a day.

after reading the article looks like spreading flat is best anyway unless you use stock cooler with tiny copper circle.