Thermal Paste Comparison, Part One: Applying Grease And More

[threehosts]

In another article, it was advised to put the thermal paste in an X formation on top of the chip. I want to believe that although the die is underneath only a fraction of the chip surface, it doesn't hurt to ensure that the entire surface is in good thermal contact with the cooler.

 

[Guest]

In my opinion some device makers apply heat sink material like McDonalds does Mayo. The more the merry. I think they could save even more money by applying the correct amount. Which is on the lighter side. You only want to fill in those small imperfections. Not create a barrier and another layer for heat to pass through.

 

[groundrat]

good information, even for the experienced builder.

 

[MANOFKRYPTONAK]

It says that results are in Kelvin on your actual benchmark results. 30 degrees kelvin would be roughly -405 degrees fahrenheit. I think that is a mistake, or I'm buying that thermal paste.

 

[ubercake]

When I got to the end of the article, somehow the term 'blue balls' came to mind.

Can't wait for partie deux.

 

[jtd871]

Various paste application techniques compared here:

http://www.pugetsystems.com/labs/articles/Thermal-Paste-Application-Techniques-170/

 

[WyomingKnott]

"When you cut a CPU in half..."

Okay, that's a beautiful way to start a sentence!

 

[vertexx]

Great info! Bookmarked for reference. Looking forward to the test results!

 

[moulderhere]

Tomshardware should do a survey on its forum members asking what method do they use to apply thermal compound.

Do you cover the entire cpu?
Do you do the "lintel" pea size method?
Do you never add any thermal compound?
Do you use thermal pads?
etc.

 

[clonazepam]

Plenty of GPUs have heat spreaders. My 580s do and they aren't exactly young.

 

[clonazepam]

There's a good video out there about application techniques. He applies the paste in various ways, then takes a piece of plexi-glass on camera, mushes it down evenly, to help illustrate how they spread. It just helps to visualize which methods have a higher potential for creating air pockets, which spread too far, etc.

 

[brendonmc]

Haha...talk about leaving us hanging! This should have been done 15 years ago!!! Great idea and extremely important! Give us the results please!

 

[clonazepam]

http://www.youtube.com/watch?v=-hNgFNH7zhQ -- How to apply thermal paste

 

[clonazepam]

http://www.youtube.com/watch?v=-hNgFNH7zhQ -- How to apply thermal paste

 

[CaedenV]

Great article! Finally puts into words some things that I have had troubble explaining myself.

Most interesting part for me was the bit about the heat pipes on today's coolers, and how only the center pipe (or maybe two) are the only ones that do the bulk of the work at heat disipation. In my own rig I use to have a traditional setup of moving air from the front to the back, but when I eventuially got a bigger case which made use of larger 140mm fans on top and bottom I went with more of a bottom to top air flow. As part of this change I changed the orientation of the CPU cooler (212 evo) to move air in line with the rest of the case and was amazed to see the temps drop ~5*c even though I also replaced the fans with quieter (but lower rpm and cfm) fans. I thought it was merely because I had a more efficient airflow and was working with convection, but in the back of my mind I knew there was no way that could account for that large of a change.
But as it is explained here it looks like there are just more (in my case all) of the heat pipes that are making direct contact with the hot spot on the CPU which is making the cooler more effective in the first place.

 

[Traciatim]

Thanks for that link jtd871, that was some pretty good information. I had recently been watching a video where the pre-spread methods were mentioned as debunked and that you shouldn't do that, yet the spread method performed very well in that test.

Personally I use my finger covered by plastic to spread and incredibly thin layer on both the heat sink surface itself as well as the top of the CPU spreader. I've always had pretty good results with this method. I only recently tried switching to the one middle dot method when I got a heat sink with exposed pipes since I figured the squishing would help get some contact in to the gaps between the pipes and the main heatsink block. This method seemed to work pretty well too but it didn't change my numbers in any measurable way from my finger-spread method I had on there previously.

I think the main point, no matter how you do it, is just to use a small amount. Think filling microscopic gaps, not spreading peanut butter on toast.

 

[InvalidError]

"don’t apply too much. Otherwise, the paste will ooze out on all sides. If your paste is electrically conductive, you can almost be assured of hardware damage."

Tell this to OEM's ever seen a hp, dell, Lenovo, Asus, Acer cpu after you take off the heatsink with thermal paste on it? it's oozed all over he place...

Most metal oxide pastes are non-conductive so it does not really matter if they overflow all over the place aside from making a mess to clean up if you want your CPU to look nice and shiny between re-applications. If you aren't too picky about cleanliness, you clean the IHS and leave the surrounding mess as-is.

 

[InvalidError]

Direct touch heat pipe heatsinks do not have air between the pipes where they contact the cpu. There's the block that they are set into.

The popular Hyper 212+ has sizable gaps between the flattened heatpipes and "fingers" of its aluminum mounting block between each pipe. On the 212EVO which has no aluminum fingers between pipes, the gaps are much thinner but they are still there.

 

[Calculatron]

Oh *thank god*. I was two steps away from drafting a generic post for replies to threads about thermal compounds. Thank you very much for making an article out of this, and I look forward to part two.

Sometimes it is kind of worrisome with how readily people will still recommend Arctic Silver 5. Do not get me wrong, it is good stuff still, but better products do exist now, if only in regards to the lack of cure-in time and being potentially damaging. The same with spreading thermal compound versus some sort of dot or line method.

But I can grumble and complain only so much; I know that, as a general rule, people often prefer to rely on historic/popular memory, rather than exploring to find things out for themselves. (Thank god for article reviews.)

 

[warezme]

A small nipple size dot at the center of the CPU heatsink spread out vertically and horizontally with a credit card has ALWAYS created a nice smooth even coat across the entire surface of the heatsink for me over the years. It doesn't have to be very thick at all. the gaps are microscopic so a thin layer works fine, it just has to be even. That brush technique looks ridiculous and wasteful.

 

[ubercake]

Anyone 'tinting' prior to application? I did this with my i7-930 and have been doing it since. This is in the AS5 application manuals. I've had no problems using this method prior to normal application.

 

[nukemaster]

I do on new coolers(I am betting the microscopic stuff does not come out when you clean a cooler after the first use), but it has never seemed to make a difference.

 

[InvalidError]

A small nipple size dot at the center of the CPU heatsink spread out vertically and horizontally with a credit card has ALWAYS created a nice smooth even coat across the entire surface of the heatsink for me over the years.

It may LOOK smooth but if you use a glass plate to simulate the heatsink or CPU being attached to it and scan the result with a microscope, you will likely find thousands of microscopic air bubbles between the glass and paste caused by fine streaks from the card's edge. With the center pea method, air gets displaced out as the pea flattens.

The "credit card" method is useful for coating unusually rough surfaces. For relatively smooth surfaces, it most likely creates more bubbles than it prevents.

 

[nukemaster]

A small nipple size dot at the center of the CPU heatsink spread out vertically and horizontally with a credit card has ALWAYS created a nice smooth even coat across the entire surface of the heatsink for me over the years.

It may LOOK smooth but if you use a glass plate to simulate the heatsink or CPU being attached to it and scan the result with a microscope, you will likely find thousands of microscopic air bubbles between the glass and paste caused by fine streaks from the card's edge. With the center pea method, air gets displaced out as the pea flattens.

The "credit card" method is useful for coating unusually rough surfaces. For relatively smooth surfaces, it most likely creates more bubbles than it prevents.

For gpus as well? I have always used that for gpus and most times the line method for cpus with heat spreaders.

I have sometimes used the baggie method(same as latex glove method) without any problems and it works in more enclosed spaces well.

I need to try the plate of glass idea as that is just a great way to visualize the spread of paste.

 

[InvalidError]

I have always used that for gpus and most times the line method for cpus with heat spreaders.

The thing is even the worst (yet sensible) methods will still yield 90+% coverage in the critical area with very little effort or care or 95+% if done remotely well so the differences between application methods would be barely measurable unless done exceptionally badly or other variables entered the equation.