OCUK Official IC Diamond/ Perihelion Test Results

WingZero30 said:
Ok just received the kit today:D. Thanks for the extra IC Diamond tube:cool:

I have two packs A and B. A contains 3 brighter papers and B contains 3 slightly darker papers.

Now how will I conduct the test exactly? What software program do I need to use to display pressure values and contact pattern? I was thinking may be the kit will be similar to gauge equipment but I don't see any wires?:confused:

How will the pressure and contact results be displayed on the screen?:confused:
Click to expand...

Take one of both A and B and place glossy sides to the IHS and sink with the "Frosted" sides facing each other.

One is the donor and the other is the "Print"

Mount your sink as you normally would with out paste.

Dismount sink remove paper - throw away blank piece

Sensorproducts.com has a pressure interpretation sheet that you can use as a general guideline.

You have to return the print to me which I send to sensor products for analysis. - sorry I forgot to leave instructions for a return mail envelope with my mailing person.

Raw image will tell you a lot out of the gate and you can compare to the review and if you post the i9mage I can give you some feedback on it till you get the analysis.

Can be used to improve contact and performance -progressive lapping etc.
 
WingZero30 said:
First of all here are is the spread pattern of IC Diamond which wasn't previously available after the testing has been conducted as I never took off the heatsink after carrying out the tests on IC Diamond. The pics are clickable:)



IC DIAMOND spread pattern after tests completion

As seen IC Diamond spreads out much better than both IC Perihelion and MX-4 See Post 291 for more details). It is a nice smooth layer. However I found that IC Diamond has slightly 'dried up' just like in the case of MX-4 whereas IC Perihelion seems to be slightly wet even after one week application . Is it due to viscosity?



IC DIAMOND spread pattern on IFX-14 base

However for each thermal compound I tested The bulk of thermal compound seemed to be attached to IFX-14 base as seen in the 2nd pic. So I think what we are seeing on cpu IHS is the very bottom layer of thermal compound which understandably is more drier as it is closest to cpu cores.


However now for the main part of this post:

Lol:p

ifx14basecontactpattern.png


(Apology. Digital Camera is rubbish)



I knew it!!. The Thermalright IFX-14 base is convex:p. The paper pic is almost actual real size but slightly bigger. Basically this is as you are reading the cpu info on the IHS. So it is either top-bottom or bottom to top pattern (or vertical line pattern). As you follow the central contact line to each end you meet the IFX-14 heatpipes.

I am disappointed with manufacturer base machining or as stated before this is meant to be their design philosophy. However in this case it seems pressure/contact decreases in the centre. Judging by the pic only 25-30% of the contact area is actually utilised.

Then again it could be Q6600 IHS itself which could be concave.

Now I know why I seem to get higher temps as I push for further overclocks. Its a brilliant cooler but with poor base.

Hmm I can't reach conclusion other than to resort to lapping.

What do you reckon IC Diamond and fellow ocukers?:D
Click to expand...

rawimgb.jpg
rawimga.jpg


The "I" pattern is interesting to me as it shows the bulk of the load carried on the top & bottom edges as in the ICD vs AS5 review which would seem to be a typical pattern with just a couple of samples.

Makes me wonder whether with the TR sinks with exceptional pressure in such a small area is deflecting the IHS on that axis and causing the top and bottom edge contact?

Might be interesting to draw a contour map + with a straight edge on both IHS and sink.

In the ICD vs AS5 tests lapping not only produced a compound/contact improvement but overall improvement was 6 C nontrivial and and would seem worth the effort.

From my understanding the water block manufacturers bend their blocks to a tight tolerance. - however on a couple of forums some have raised the possibility that the bend on many of these heat pipe heat sinks the bow or bend is due to heat warping on when soldering the HP's to the base. Some thing to think about

With the TR high pressure mounting system it seems a waste to not utilize increased contact area for an edge in performance.
 
Last edited:
The Asgard said:
May be there is a market for these contacts kits?
I know I have used and lapped CPU's in the past and had minimal gains. Knowing you have a poor spreader or HS base would allow you to sort any heat problems out.

Sounds like a winner to me?
Click to expand...

Not sure whether there is a market as earlier I offered the water cooling guys free kits and had no takers.

I use it in the lab for troubleshooting and lately as a counter check in conjunction with a torqe wrench.

once you use it a couple of times you get a good feel or sense that carries over to the paste patterns & from just the raw image as to what is going on contact wise.

The more people refine and optimize their mounts and the more they know about the process the better ICD works and that's a plus for IC

The cognoscenti here on the forums all know that good pressure and contact are essential for good performance and none of it is rocket science but the process needs a spotlight occasionally when you are operating in the rarefied performance arena as we get overwhelmed with the details.

I tried carbon typing paper as a poor man's solution with no success. colored wax would probably work to highlight highs and lows but leave a residue.

Of course if you were committed to lapping you could the machine shop bluing stuff as a highlighter/guide - couple of strokes on the sand paper and the highs and lows become readily apparent - some magic marker ink would have the same effect.

for lapping I use a belt sander with a variac, files, diamond DMT stones 200 to an 8,000 grit.
 
WingZero30 - One thing with lapping is that by doing so, wouldn't it decrease the overall pressure between heatsink Base and CPU IHS to a certain degree as Force applied by heatsink mounting system will spread out onto more CPU IHS surface area (Pressure = Force/Area ) as a result of lapping? This can also be seen in the unlapped vs lapped pic you have quoted where the red colour in lapped pic is slightly lighter?

I have seen on some forums including in this one where some people placed 1 or 2 washers underneath their heatsink mounting systems after lapping.
Click to expand...

Not so with the TR sink in the review looks better to me. What I like about the TR sink is it's pressure capability

statisticsb.jpg
rawimgb.jpg


statisticsa.jpg
rawimga.jpg


Not familiar with the TR sink, if the bend/bow is severe enough I can imagine a shim being required,

Email sent
 
Latest update - any errors or omissions on my part let me know and they will be corrected. I have done a couple data sorts and may have mixed things.

Green bars on the chart are the Corsair water results and the red the other water cooling test results. Small sample but the Corsair seems a little more robust mounting wise.

Looking at the water block/systems they are excellent designs on the whole. The blocks @ .025 C/W are pretty much at/near the material limits for copper. The radiators single 120mm fan have a thermal resistance of about .06 C/W dual fans .03 C/W.

So being pretty much at the design/material limits each degree improvement here on out is huge, 4 120mm fans would only net 1.5C on a 100W If I am understanding things correctly. So on a thermal cascade the thermal compound/contact/pressure is one of the few avenues available for improvement.

So looking for some user feedback here - What are the limits on water block design? One thing I am picturing is that the block have thin walls and or bases and being made of soft copper so high pressure might be prohibitive as the block may bend? Some I have looked at seem to be only .125 thick

Fasteners- most seem to be spring +screw with thumb screws common. How do you tighten them down? screw driver or by hand?

Bent base for pressure - most I have looked at have the feature, are there any that do not? Do the sink bases fit inside the outside edges of the IHS? when mounting are they hard to balance or are they toleranced tight enough that it is not a problem? Seems to me might be awkward otherwise would have a tendency to list starboard or port.

I am not schooled in in the finer points of water so any feedback here would be much appreciated



april22ocukupdatechart.png


ocuk%20april%2022%202011%20list%20update.png
 
Only chance I get to do this is late night (1:30 am) so prone to some error. I was feeling good about the accuracy on this one though.... Thanks for the corrections keeps us honest.
 
EVH said:
The proof is in the pudding as they say..


Idle - stock paste (Shin Etsu)................................ Prime 95 - stock paste (Shin Etsu)


Idle - IC Diamond 24............................................ Prime 95 - IC Diamond 24


i7 2600k, stock speeds, with a Corsair H70 cooler, and Akasa Viper fans in push/pull (instead of the corsair stock cooler fans).

Both load readings were taken first, to avoid the artificially "cool" temp from a freshly started PC. I ran Prime 95 (small FFTs) for 3 hrs+, and took a load screenshot, then left to sit idle and cool for 1hr, and took an idle screenshot.

Very happy, that's a 5C reduction at idle, and 3C reduction at load on the hottest core!!

I'm now considering doing my GPU :)
Click to expand...

The Diamond seems to be favored mounting wise on the Corsair water coolers.

What makes the mounting different from the others?
 
Updated the results with corrections - Recently switched from Windows to Ubuntu mostly pretty good but had a devil of a time converting the spreadsheet to a graphic.

may42011ocukupdate.png
 
WingZero30 said:
Nice:)

Since moving from my old Thermaltake Big Typhoon cooler to Thermalright IFX-14, the base of the latter felt different. Big Typhoon seemed to have more of a flat base whereas mounting of IFX-14 felt a bit more awkward as it tended to produce slightly more movement sideways.

The convex base of IFX-14 has now been confimed by the pressure and contact test. It will be very interesting to know the average pressure exerted by unlapped IFX-14.



Judging from the contact pattern, I would say that unlapped contact pattern or the 'I' pattern isn't making a good contact at the centre as colour seems to become lighter in that area.

The 4 cores in Q6600 are aligned horizontally as shown below numbered 1 -4 with contact pattern superimposed roughly (paint work:D) :

[

As seen above the light pink colour is emphasising lighter contact in the centre as seen in the original picture aswell. It is interesting to note that unlapped IFX-14 seems to only partially cover core 2 and core 3 areas while no contact is made with core 1 and core 4 areas.

So more heat is accumulated faster at Q6600 IHS than can be transferred by IFX-14 via heat pipes resulting in higher cpu core temperatures overall. May be this another reason why there is 6-8C difference between hottest and coolest cores on my Q6600.


To me this suggests that unlapped IFX-14 base or even unlapped Q6600 IHS are a poor heat transfer mechanisms

I am very glad that I took part in this thread as I have learnt something further about thermal compounds and mating contacts between heatsinks and cpus. Plus it has been an eye opener into the world of PC cooling:):cool:

It will be interesting to see how the contact and pressure results along with thermal compound results turn out when I lap both the IFX-14 and Q6600 in near future:)
Click to expand...

Nice bit of analysis you have there - looks that even on a rotation the cores would be off center.

What other sinks mount in that orientation?
 
Looking at the analysis you only have .55 inches of contact area out of around 2.6 inches and the bulk of that is biased towards the edges of the heat spreader in the zone of diminishing returns in sectioned areas 2&3 and 8&9.

From the results I am seeing I think the doctrine of convex bases has some shortcomings on the contact/pressure trade off side of the equation as the heat spreaders do work and after crossing a limit point of perhaps <50% contact vs high pressure temps go up.

So WingZero30 are you going to belt sand it flat or hand lap all the way?
 
Devil is in the details

Faster way is to use a belt sander, I use a variac to control belt speed so the pieces do not get to hot. It gets me flat to within .001 the edges are a little rounded but that's ok for this application. Machine's, files work and and save time, that is what they do in the factory.

Then for finish I use Dmt sharpening stones to 8000 grit
http://www.dmtsharp.com/products/diasharp.htm

p85566b.jpg


They vary in cost and quality I bought a 4 sided block 200 - 800 grit for $10 USD and works fine and I think cheaper than sandpaper at least in the long run if you opt for the more expensive blocks.

I rarely spend more than an hour

flat is best and going beyond a 600 - 800 grit yields little improvement as most particle sizes are to large to fill the micro scratches at this level.
 
From OCUK IC Diamond Review

rawimgb.jpg
rawimga.jpg


statisticsb.jpg
statisticsa.jpg


WingZero30 Results

rawimage.png
lapping2.png


pressurestatistics.png


These are two great tests for the reason that initial contact patterns are the same and in both tests the variable of pressure has been fixed at more or less at the upper limit.

In the review result overall improvement was 5C from initial AS5 to IC Diamond with lapping with an extra 2C improvement with the added thermal contact/conductivity of IC Diamond-WingZero30 Result is 6.9C from initial MX4 test to IC diamond with lapping.

WingZero30 Results of greater improvement than the review test could be in part due the approx 18% less contact area at the start than the review, curve gets pretty steep fast under a inch.


Too bad there is no follow on with the MX4 as I believe the ICD/MX4 delta margins would have increased as with AS5 in the in the review - Poor contact and or pressure tends to minimize differences(homogenize) between compounds.


EniGmA1981-Nice results! I so need to get my paws on an IFX-14!! lol

I only ever lapped one cooler, it took hours (as you said) and made no real difference at the time lol (was a CoolerMaster HHC-001 fitted onto an Athlon XP2000) temps didn't seem to change, maybe a degree... So have never thought about it since - after this though, the TRUE might be getting a shave
Click to expand...

Perhaps the compound?



My thinking here on the bowed base concept is that it is either a manufacturer cost/convenience or not a fully developed idea.

For instance from zero and the contact points provided from the test you could draw a curve that would show the optimum pressure/contact combination which is probably 2x -3x of what we have seen in any contact image. Along with other design considerations a simple alternative would be to machine a flat step that would fit those optimum parameters.

Competitively it would make sense as high end sinks and water blocks margins are relatively close, the margin increase noted in these two tests would put them ahead of the pack performance wise.

BTW Great Job WingZero30!
 
As for Q6600, it's IHS surface wasn't flat either as seen in the photos and this definitely contributed towards less contact area aswell.

I believe manufacturers of both CPUs and Heatsinks should pay more attention to the importance of having flat surface and to facilitate large contact area.
Click to expand...

To be fair manufacturers have to figure cost/benefit into the mix for the broader market, for Intel +/- 5C is no big deal and tighter tolerances will drive up cost.

for the sink people it looks to be that machining is done before assembly and the soldering process may engender the "bend" at which point machining the fully assembled piece by my guess would be awkward at best and probably falls under the same cost benefit consideration.

ifx14duringlapping.png


How do you clamp this for volume production machining ? also interesting are the outer heat pipes probably were under utilized having no contact at all with the IHS in that area.

I will say that your contact Image looks to be in the top 5% of those in my contact/pressure image library
 
Last edited:
WingZero30 said:
IC Diamond

I was thinking would it be ok if I also posted the lapped contact pattern to you and then results can be obtained from sensor products to give further insight?
Click to expand...

It's all about contact and pressure in case nobody got the message..

Send it off and I will forward to Sensor Products

I know you had some issues but any chance we could get a before compound test on the lapped set up?
 
You must log in or register to reply here.
Back
Top Bottom