Prolimatech PK-1, Arctic Silver Matrix, and Cooler Master IceFusion

Intro

This is the second installment of many of the Skinnee Labs TIM Comparison 2011.  For most of the installments, results will be released in two TIM pairs roughly every two weeks but today, again, we have three TIMs for  review.

Our testing methodology has changed for 2011; we’re back down to five mounts per TIM and down to ten hours per mount.  But we’ve added a large new wrinkle: testing at three different ‘contact’ settings.  By testing at “Poor,” “Moderate,” and “Great” contact, we can see how well each TIM performs across a wide range of situations.  To do this, we’re using three Koolance CPU-360s with various modifications and different mounting pressures.

Like our previous TIM reviews, we’ll also show how the TIM cures over time and how it varies across mounts.  Overall we feel the TIM Comparison 2011 will be familiar in format, just as rigorous, but with a lot more depth.

Today’s TIMs for the comparison are Arctic Silver Matrix, Cooler Master IceFusion, and Prolimatech PK-1.  After today, we’ll have six TIMs completed in the 2011 Comparison, but we still have roughly 30 more on the docket.  Lots of testing left to do!

The TIMs

Arctic Silver Matrix is Arctic Silver’s latest and greatest TIM, following up on the success of Ceramique and Arctic Silver 5.  According to Arctic Silver, “Developed in conjunction with TIM Consultants and based on their acclaimed 0098 grease, Matrix Thixotropic Premium Thermal Compound offers optimum performance in demanding moderate to large bond line situations. It is designed for a wide range of applications between modern high-power CPUs and GPUs and high-performance heatsinks or water-cooling solutions.”  Arctic Silver ups the ante for their typical cure-time recommendations to 300 hours plus multiple thermal cycles.  Considering Arctic Silver 5 has been the benchmark other TIMs are measured by, but is also an older TIM that’s no longer near the top of the market in terms of performance, I’m really interested in seeing if Arctic Silver’s latest can bring them back.

Cooler Master IceFusion is a high-volume, budget paste.  It comes in 40g and 200g tubs and at extremely low costs; the 200g tub can be found for less than $20.  It’s a white paste that’s usually pretty thin but separates into two components: a thick white paste and a thin oil.  This means it requires mixing before each use for consistent performance.  In the AMD Installment of our last review we showed it was a poor performer, but that’s why it’s being included in the early portion of our 2011 Comparison: having a reference point for a ‘bad’ or ‘generic’ TIM can be important for people looking to upgrade.

Prolimatech PK-1 is a paste that’s growing in popularity.  In addition to the fact it’s gaining popularity quickly, it’s in this review because it didn’t really have a logical pair to be reviewed in tandem with later.  PK-1 is bundled with Prolimatech heatsinks and Aqua Computer’s cuplex Kryos line, both of which are considered high-end products in their own markets, so there’s promise from that aspect.  It’s a grey, wet paste that has pretty low adhesive properties–a blob would rather stick to the syringe tip than to an IHS–but it also spreads very well.

The Blocks

For all three contact “settings” I use a Koolance CPU-360.  I’ve chosen the CPU-360 due to its great mounting system (although I’ve modified all three blocks’ mounting systems) and because it’s pretty easy to add slight modifications.

At the “Poor” end of the spectrum, I have a stock CPU-360r1.2 with extremely low mounting pressure; the stock CPU360r1.2 has a somewhat irregular base and when paired with low pressure, TIM does not spread into a thin layer particularly well.

For the “Moderate” contact setting, I’ve taken a CPU-360r1.1 and reduced some of the internal structure so that there’s absolutely no bow.  With pressure in the center of the base the block can actually become slightly concave as the o-ring compresses, but with only moderate mounting pressure the base seems to stay perfectly flat.

As for “Great” contact, I might have gone a little overboard; no block on the market has contact this great.  I’ve modified a CPU-360r1.2 to have a thicker midplate with a compressible layer and the result is a pretty extreme bow that flattens with mounting pressure.  And there’s a lot of mounting pressure.  The result is impressive; with low viscosity pastes, it looks like there isn’t even any TIM on the center of the CPU when taking the mount apart.  Even with viscous pastes the resulting layer is extremely thin.

Pages: 1 2 3 4

Comments

Posted On
Mar 17, 2011
Posted By
Batou

I see your point of view. Anyway, i noticed that the CPU test surface seems lapped. For what i know, every stock CPU has a concave surface,
i guess this is the reason because many air and water coolers come with a convex surface. I imagine that the perfect contact between two surfaces
comes when both of them are perfectly flat; but yes, i’m not taking into consideration the contact pressure that could eventually make one of them flexing.
That being said, i still think that if i buy a new CPU, and for example a new supreme HF or 370, i’ll get the best temp result lapping both of them near perfection, using grits till 10000 synthetic and a comparison plate (i don’t know really english word for this, but i’ve got a plate like this to lap: http://www.roderelectronics.com/cta/old_web/it_b0243_piani_riscontro_mitutoyo/902_304.jpg)
Do you sincerely this this kind of looong (and lovely!) work could become useless since the CPU or block surface will flex? I’m not talking about temperature difference between lapped/non lapped surfaces, but am actually satisfied with this kind of work even if i get 1-2 degree less than before.
This is my lapped cpu :
http://i638.photobucket.com/albums/uu110/nethan2/securedownload.jpg
http://i638.photobucket.com/albums/uu110/nethan2/CPUlapped.jpg
I hope i did not spend 3 days to lap my cpu like this for nothing, that’s why i am asking to an expert like you are.
Sorry i am going a bit off-topic, SL open a forum plz!! :)

Posted On
Mar 17, 2011
Posted By
Eric (Vapor) Hassett

My CPU isn’t lapped…I did a few trial mounts with IC Diamond before all the testing began, and it kind of polished it (and thus eliminated itself from the review). All the Core i7 IHS’s I’ve ever seen (dozens) are at least slightly convex, but not perfectly regular/symmetrical.

I’m not against lapping a CPU, but lapping a water block while it’s assembled is a big no-no. The coolers are designed with a certain amount of bow and internal compressibility (somewhat flattening it under pressure). The only way lapping a block is ‘okay’ is if you take the block apart and lap the base when the block is not assembled. Overall, the bow is extraordinarily important (unless you’re using Indigo Xtreme, in which case lapping won’t gain you anything noticeable anyway) and not be messed with…..at least with EK/Koolance/Swiftech/D-Tek blocks. I haven’t been able to assess the quality of contact from Enzotech/AquaComputer/Watercool (Heatkiller)/XSPC/etc blocks yet.

Posted On
Mar 18, 2011
Posted By
Batou

Thank you very much for your reply, i will take your words into consideration for the future.

Posted On
Mar 17, 2011
Posted By
theseeker

Great review!

Posted On
Mar 16, 2011
Posted By
Batou

Cool, finally the second round! (Also very interesting your last article about Koolance 370, nice block.) About the “great” contact, do
you think a “perfectly” lapped surface would give a similar results? Now I’ll wait third round :)

Posted On
Mar 16, 2011
Posted By
Eric (Vapor) Hassett

I think a lapped surface is overrated, especially in watercooling. With air coolers you at least have ~1cm thick of copper as structure to enforce the flatness of the structure of a lapped base, but that’s not the case with waterblocks.

With a waterblock, you have ~2-3mm of copper that’s been machined (with the thinnest points being <1mm thick now) and it’s significantly weaker. If you start with a flat base, with pressure it flexes and goes concave. So you need to start with a convex (bowed) base…with pressure it either flattens or stays a little convex. By being a little bowed, you ensure the thinnest TIM bondline thickness (which means least thermal resistance) is right over the CPU die.

If you look at the pictures of the pastes with "Great" contact, you can see the middle of the IHS has the thinnest amount of TIM (and with IceFusion and PK-1, it's nearly 'bald')…that's what makes it better than the other contact settings. Same with the CPU-370, they added a bit of a bow and that's where most, if not all, the performance improvement is coming from.

Leave a Reply

You must be logged in to post a comment.

Login