Upcoming Thermal Compound Shootout

[anton]

Yeah guys, I'm sure you are aware of this, but if you are going to do the test with the Coollaboratory product, you have to have a copper base on the heatsink. It is quite specific about not using an aluminium based heatsink.

The stuff is pretty wicked to use... reminds me exactly of mercury. Very cool.

 

[3lfk1ng]

Ugh
Yes....we know.

 

[dasickninja]

This is going to give me a headache to figure out

 

[Surferosa]

Good test- this Im interested in.

You definitely need to include the stock- we want to know how effect they are compared to a pad.

And why cant you do them all? Im sure you have plenty of time at your hands....

 

[dasickninja]

I'll break down how its going to work for you. The Intel boxed TIM will be included, but not in a way that you'll see in any kind of obviousness. As I've been told, or haggled to, depending on my mood, every bar graph needs to be zeroed. But my counter argument is that a graph starting at zero when the results are at 50C and have increments of 2 to 3C up or down is very inefficient when it comes to showing the degree of change between the samples. Therefore I decided that I'd do a test of the Intel TIM and zero the graph with the average from that test.
Meaning if the Intel results are at 50C the graph will start at 50C and all results will be interpreted with that zero point in mind.
Just as a general notice, this isn't open to debate.
As for testing all the TIM's on the list, no, we don't have all that time on our hands. I, personally am out sick but I've got work, school and my other activities to occupy my time as do the others. Also, this is going to be a test between the TIM that overclockers would be most likely to use and whether or not AS5 is really the best stuff for the job.

 

[Surferosa]

Therefore I decided that I'd do a test of the Intel TIM and zero the graph with the average from that test.
Meaning if the Intel results are at 50C the graph will start at 50C and all results will be interpreted with that zero point in mind.
Just as a general notice, this isn't open to debate.

Its your game- you get to choose the rules. No probs there mate Besides- that is the most logical way of doing it.

As for testing all the TIM's on the list, no, we don't have all that time on our hands.

Bah humbug

 

[dasickninja]

Thanks for understanding.

 

[rwaritsdario]

The base on the stock HSF is copper, only the fins are alu.
You should inject some of that stuff up your a$s, maybe itll lower your temps.

 

[dasickninja]

*BAAAASTAAAAGE*
Morning to you to.
I'm at 102.8F toady.. Still a hot Tjunction

We should be fine then. No more complaints about destroying the heatsink plate. Please read, its not a problem anymore.

 

[NovemberWind]

One more thing, might this be done on a test bench? I know having it in a case will simulate real world conditions more, but I can't think of any reason why getting results on a test bench wouldn't be easier for ya'll. Just looking out for the team =)

In the same line as Doughbuy, I'd also recommend simplifying this experiment as much as possible. As in, use a known heat source about the same size as a processor. While processors are certainly what you're interested in, there's no guarantee (or likelihood) that every test is going to behave identically, even at full load each time. A resistance-heated plate (optimally copper for thermal transport since you'd be emphasizing the change in thermal dissipation in the paste rather than the base materials, but steel would be good too) that is known to hold about 70°C, with a steady power supply(DC) and an on/off switch would probably be optimal conditions to test different thermal compounds.

Yeah, pc applications are what you're interested in, but doing the testing in real world conditions is like trying to figure out what car goes the fastest by driving them on back roads. Sure, the roads around your house are where you're going to be driving all the time and in the end what matters, but if you want to figure out which car has the highest top speed, you test at a race track, not Park Rd.

Just a thought.

Best of luck with the testing. I look forward to the results.

 

[dasickninja]

You do realize that we don't get payed to do this right? This is the simplest way we could do it real world. Everything is in a condition that an OCer would encounter.
Thanks for the suggestion though. We'll be sure to remember that when planing a heatsink test.
What the hell, you can't make a test that'll please anyone. We're just doing it the way thats most efficient for us and that'll net us the least amount of hate mail.

 

[anton]

Thanks guys, really looking forward to the test. Good luck.

 

[Doughbuy]

LOL Ninja, no need to throw a hissy fit with small letters, you're straining my eyes here, vision's not what it used to be...

That picture should make you happy again.

Edit: No more pics please, I'm at work, and my bosses are right next to me...

 

[MAK-Daddy32]

Told you so...

ok, just kidding


Seriously, I'm looking forward to seeing the results. I'm about to swap out the stock HSF on my E6400 and would love to know if there's something better than AS5. It would also be nice if the review included a brief note about how difficult it was to remove each type i.e. did it harden etc. but, the straight temp data would be much appreciated.

thanks again.

 

[mdmoy]

Question:

How do you plan to establish repeatability and characterize the potential variability that comes when applying the SAME compound - are you going to try multiple measurements of the same compound, applied 2-3 times, to see if there is much variation WITHIN a sample set, vs across sample sets? I am curious because I wonder how much variation that people have experienced in switching compounds may come from exercising more care in removal and application of new compound as compared to the compounds themselves, or to slight variations in "technique" of applying compound, the amount of forcce applied by the heatsink, etc etc....

Thanks for your work - looking forward to seeing the results.

 

[dasickninja]

The question is aimed at Dario, but I'll take a whack at it before I take a nap. simply put, we're going to do everything humanly possible to replicate parameters through out the test. Will we measure? Yes. Our methods? You'll find out.

 

[Doughbuy]

I doubt they'll redo the same compound because that in itself will add hundred of more hours of work. I think a general basis is enough. Yes, there will always be margins of error, but they can't really do anything about it unless they want to dedicate half their life to testing thermal compounds...

 

[3lfk1ng]

Yes, there will always be margins of error, but they can't really do anything about it unless they want to dedicate half their life to testing thermal compounds...

Yeah...*sigh*... I think I would rather die a horrible death from being CNC milled to peices (Pretty sick huh?), than to spend over half my life analizing the general statistics and characteristics of different thermal pastes.

We definitely wont be applying the paste over and over again to see if there are any changes. It's just going to be a very accurate and precise 1 time thing for the top 5 pastes that win in this poll.

As I mentioned earlier we will be taking lots of pictures thoughout the procedures. One of the things that many other paste reviews lack are pictures of the amount of paste applied to the surface of the processor before the heatsink is seated. These pictures give our readers 100% confidence that our review is 100% legit in every aspect. The pictures will also show how the paste should be correctly applied on a Core 2 Duo processor before installing the heatsink for those that keep asking.

Basically the tests we are trying to recreate are that of a standard application from the average Joe after properly/thoroughly reading the instructions (If we have the time, perhaps we will include the instructions for each paste in the review). The average Joe would follow the directions...put on the paste...put on the heatsink and be done. We aren't doing anything that you yourself wouldn't do during the application process, which is why we chose not to lap the heatsink.

Our reasons for the methods chosen are because we want to replicate, test, and record the exact same paste application that most readers do during the installation of a heatsink.

Please check back at DaClans blog often for the latest on interesting news as well as the most unique product reviews.

~3lfk1ng

 

[Surferosa]

Yes, there will always be margins of error, but they can't really do anything about it unless they want to dedicate half their life to testing thermal compounds...

Nobody asked for half their lives. I'll be happy if they devote all their waking hours for the next couple of months.

You do realize that we don't get payed to do this right?

Ya, but what price could you put on the eternal gratitude of TH forums??? Surely thats priceless :lol:

 

[db101]

We will be taking lots of pictures thoughout the procedures. One of the things that many other paste reviews lack are pictures of the amount of paste applied to the surface of the processor before the heatsink is seated. These pictures give our readers 100% confidence that our review is 100% legit in every aspect. The pictures will also show how the paste should be correctly applied on a Core 2 Duo processor before installing the heatsink for those that keep asking.

Basically the tests we are trying to recreate are that of a standard application from the average Joe after properly/thoroughly reading the instructions (If we have the time, perhaps we will include the instructions for each paste in the review). The average Joe would follow the directions...put on the paste...put on the heatsink and be done. We aren't doing anything that you yourself wouldn't do during the application process, which is why we chose not to lap the heatsink.

Yay, pictures! You guys are awesome to test these products - I'm interested to see how some of 'em stack up against the lauded AS5. This will really help people like myself who don't apply paste often ourselves and are interested in the differences these aftermarket products will give us over stock. I look forward to seeing the results!

 

[mdmoy]

thanks for response. I recognize the effort involved...

 

[audiovoodoo]

I can apreciate that this is turning into a bit of a mission and repeat test on all the compounds is not really and option but perhaps we have the makings of a part 2 review here...

Try this.. From these first and substantial tests you should be able to seperate the good from the bad and produce a reasonably good representation of the better compounds. Why not look to then repeating tests on the best 3 with 3 different coolers. (These could perhaps be from a cooler shootout )

Seriously though I was interested by some earlier comments about some compounds beeing better suited to lapped heatsinks and it would also allow you to show the consitancy of performance across your earlier tests.

 

[Doughbuy]

Pictures are nice, always a helpful thing, especially to those people who don't know how large a grain of rice is...

<.<
>.>
<.<

 

[dasickninja]

CoreTemp text files, CPUz screens, measurements of the dimensions of the applied paste, size of applied layer, graphs of both the line and bar variety, I've got a lot to do. I've gotta remember pictures now. Yep. This is the biggest review yet. Besides the PS3 vs XBox360 vs Wii review, and I wasn't taking reader input on that review.

 

[Doughbuy]

But see, PS3 vs. Wii vs. Xbox 360 is fun just because you get to play around. This, you're just installaing and reinstalling heatsinks... shrug...