IC-Diamond WATER cooling thread

frostdragon · 26 Mar 2014 at 22:16

hey there

Depends on the size of the IHS. if it is a naked core gpu (like in the ati model above) then it would be too much.However a larger IHS will need more.

Edit just looked at the core.....nah looks like you have too much on there. For a IHS that size a blob in the centre will do.

IC Diamond · 5 Apr 2014 at 21:31

Been awhile. The OCUK tests was the first group that I had out of 20 or so giveaways up to that date that had a significant sample of water cooling tests for comparison, most likely reflecting increased water cooling usage over the years. Prior to that I only had one or two per giveaway.

what was noticeable in the OCUK test was the statistical cluster of under performers of the WC systems marked in red.

This is part and parcel for our troubleshooting and customer support so we brought frostdragon on board for his WC expertise and purchased 5 blocks for him to investigate.

This was a direct contact and pressure (C/P) comparison vs thermal and as expected about a 3 C spread between best and worst C/P. This of course does not reflect whether one block was any better than another only that there was a progression in improving temperatures related to improving C/P.

So with these initial results it was decided involve end users in the process for analysis and feedback as there is quite a broad (and variable) range of blocks and systems to also facilitate our understanding.

frostdragon is setting up with the necessary equipment and I suppose will provide a progress report within the week or so.

frostdragon · 6 Apr 2014 at 15:11

IC Diamond said:
Been awhile. The OCUK tests was the first group that I had out of 20 or so giveaways up to that date that had a significant sample of water cooling tests for comparison, most likely reflecting increased water cooling usage over the years. Prior to that I only had one or two per giveaway.

what was noticeable in the OCUK test was the statistical cluster of under performers of the WC systems marked in red.

This is part and parcel for our troubleshooting and customer support so we brought frostdragon on board for his WC expertise and purchased 5 blocks for him to investigate.

This was a direct contact and pressure (C/P) comparison vs thermal and as expected about a 3 C spread between best and worst C/P. This of course does not reflect whether one block was any better than another only that there was a progression in improving temperatures related to improving C/P.

So with these initial results it was decided involve end users in the process for analysis and feedback as there is quite a broad (and variable) range of blocks and systems to also facilitate our understanding.

frostdragon is setting up with the necessary equipment and I suppose will provide a progress report within the week or so.

As well as the promised 2011 pin CPU test results, we shall be starting the give-away soon. Once that is underway we shall do some motherboard tests. That is how much more efficient a mainboard is AFTER ICD has been installed.

frostdragon · 8 Apr 2014 at 12:41

IC DIAMOND at the GBT Overclocking championships.

Every fancied what our staff-members actually DO with our IC-Diamond thermal paste?

Well here is one..
CUE Gigabyte Overclocking (GOOC) 2010 (UK OC championships)

The scene is set (the venue at milton keynes before everyone else arrived)

Our staff member pouring in the DICE pellets.

fizzing away..

the setup.

Note at this time (back in 2010) it was our staff-members first time at sub zero overclocking..But still he took second place.

Obviously he used our IC-DIAMOND 24 carat on his setup

Damo667 · 9 Apr 2014 at 00:44

I was running 2 x asus hd7770 in xfire and in heaven 4.0 they ran at 106 deg c and 94 deg c. I removed both sets of fans and the tim was like the first gpu pic rock hard like concrete. I cleaned them up, applied ic diamond and they then ran at 78 deg c and 74 deg c a massive difference.

I'm using it on my cpu and current gtx770 waterblock too. Only thing is it's quite thick and doesn't hardly spread well at all.

IC Diamond · 9 Apr 2014 at 15:57

Damo667 said:
I was running 2 x asus hd7770 in xfire and in heaven 4.0 they ran at 106 deg c and 94 deg c. I removed both sets of fans and the tim was like the first gpu pic rock hard like concrete. I cleaned them up, applied ic diamond and they then ran at 78 deg c and 74 deg c a massive difference.

I'm using it on my cpu and current gtx770 waterblock too. Only thing is it's quite thick and doesn't hardly spread well at all.

#1 Cause of compound failure is excess liquid in the mixture in the mix, while creamy and easy to spread when the liquid dissipates it leaves voids leading to early failure of the compound/components.

Our high viscosity(thickness) or high bulk loading is by design as the liquid dissipates compound shrinkage is virtually non existent so no voids are created.

The example below only lasted 400 hours at about 40C before Temps deteriorated enough it so that the user pulled it apart to see what was going on.

With IC Diamond you will never see this type of failure.

While the focus is all on performance we are introducing the idea of quality as it relates to thermal compound. Quality in any product is generally accepted as performance and durability.

Low viscosity is the Achilles heel of your typical retail compound and liquid loss with unavoidable heat and thermal cycling, it is just the nature of the beast.

For application nothing is simpler than our recommended compression spread. Just apply bead or X and let the heat sink mount compression spread the compound, skip the manual spread as you run the risk of introducing air into the final bond line.

http://innovationcooling.com/application.html

Damo667 · 9 Apr 2014 at 16:29

Ok thanks for that. On my cpu I tried to spread it evenly over the ihs but I ended up wiping it off and just applying a line straight down the middle. The gpus as they are small just a blob in the centre. I have to recommend ic diamond to anyone it's the only one I'll be using.

IC Diamond · 9 Apr 2014 at 18:58

For OEM application with silk screening or ink rolling we supply it with a solvent to thin it out you could thin it with an acetone or similar solvent and let it completely dry for 20 min. if you were so inclined.

I just use compression myself as being the simplest.

frostdragon · 9 Apr 2014 at 19:50

Damo667 said:
Ok thanks for that. On my cpu I tried to spread it evenly over the ihs but I ended up wiping it off and just applying a line straight down the middle. The gpus as they are small just a blob in the centre. I have to recommend ic diamond to anyone it's the only one I'll be using.

Indeed the compression method is the best. As ICD (or as i like to say 'the boss in the office

said there is no need to make extra work for oneself. Simply do a bead or my personal favourite the X mark with blobs and bobs your uncle.

frostdragon · 9 Apr 2014 at 19:57

Right folks next project is that we feel that ICD has been limited to CPU's and GPU's to a lesser extent. So as a result we are in the process of setting up some motherboards to demonstrate to you how effective they are with ICD-diamond installed as opposed to the standard TIM.

Obviously we are unable to put ICD on mosfet's as they need a physical pad. However on ICH,MCH and PCH' chipsets this is fair game.

So with that in mind WE WILL DEMONSTRATE IC-diamond and how much of difference on motherboards it makes also.
Motherboards we are planning in test

DFI LANPARTY P45 T3RS UT
DFI LANPARTY P55 MINI ITX (old 1156 motherboard)
DFI LANAPRTY X58 TE3H8 UT
ASUS RAMPAGE 3 FORMULA 1366

MSI X79A GD65 8D
MSI X79A XPOWER 2 BIG BANG

OLDPHART · 10 Apr 2014 at 13:00

Just looked at the amount of TIM on the GPU's and could not believe it what a waste.

When l do my CPU and GPU's for my custom loop and apply the TIM > CPU a small pea size blob, GPU's large pea sized blob and use a lint free cotton bud to evenly spread it and just stop short of the edge of the core's.

Then as l place the block on and just move it from side to side slightly before l evenly tighten the blocks down.

Have tried a few make's of TIM and settled with AC MX-4 as l think its the best of the three > MX-2/3/4, as it gives good temps easy to spread and clean off.

SavellM · 10 Apr 2014 at 15:12

I am running my CPU without the IHS.

I am using IC Diamond on the die itself, connected straight to the EK Supremacy WB using the EK Naked screw fittings.

CPU is so cool its amazing!

Can see a few Pics in this post:
http://forums.overclockers.co.uk/showpost.php?p=25368385&postcount=396

Or in my build log.

I love this stuff and recommend it at work too.

frostdragon · 10 Apr 2014 at 18:58

OLDPHART said:
Just looked at the amount of TIM on the GPU's and could not believe it what a waste.

When l do my CPU and GPU's for my custom loop and apply the TIM > CPU a small pea size blob, GPU's large pea sized blob and use a lint free cotton bud to evenly spread it and just stop short of the edge of the core's.

Then as l place the block on and just move it from side to side slightly before l evenly tighten the blocks down.

Have tried a few make's of TIM and settled with AC MX-4 as l think its the best of the three > MX-2/3/4, as it gives good temps easy to spread and clean off.

With ICD there is no need to manually spread the paste.ever

If you do spread it you will actually introduce oxygen bubbles into the TIM layer and thus reduce performance. Small IHS AND GPU IHS pea-sized blob will do, large ihs and gpu the 'X' pattern is the best

frostdragon · 10 Apr 2014 at 19:03

SavellM said:
I am running my CPU without the IHS.

I am using IC Diamond on the die itself, connected straight to the EK Supremacy WB using the EK Naked screw fittings.

CPU is so cool its amazing!

Can see a few Pics in this post:
http://forums.overclockers.co.uk/showpost.php?p=25368385&postcount=396

Or in my build log.

I love this stuff and recommend it at work too.

It is people like the above gentleman is what makes my job fun and interesting. I love creative solutions, now if you have a image with the ICD actually used on the chip die itself, you win yourself a tube of ic7.

Well done!

Can you please show a before and after temps with your setup using icd? If so and if you can still show a photo with icd installed ill upgrade you to a nice IC24 tube.

Once again well done

Frost Dragon

frostdragon · 10 Apr 2014 at 19:47

SOCKET 1156 CONTACT Impressions.

As promised here are some more contact impressions that we have undertaken, however this time using a different motherboard and CPU.

Below is a photo of the motherboard used and do not worry x79 owners 2011 pin CPU impressions will come soon.

As usual we performed two mounts per test, with the holes (inlet and outlet) parallel to the main board and then a remount with holes across.

Swiftech Apogee HD

EK-Supremecy full nickel.

Koolance 370.

So now we have two motherboards and different CPU'S of which are having the SAME concave IHS issue.

The point of why we are doing this is that PERFORMANCE (temps) is only one part of the equation! If this was the only concern then quite simply ANYONE with a computer and google can learn how to make their own TIM. However the other half of the aforementioned equation is RELIABILITY. Better contact/pressure with your CPU promotes the latter-mentioned ergo less chance of a TIM FAIL. This rule applies on all paste and not just ICD (if that helps)

I have seen many on here using TIM where every so often they have to re-apply their compound ever so often; even without a CPU or block change. With ICD there is no need to do this as this is built not only for performance but reliability too.

Better contact and pressure promotes higher TIM reliability on ANY make of compound. However only ICD is designed from the ground-up for a high heat- high stress environment.

Anyway 2011 pin CPU tests next.

OLDPHART · 11 Apr 2014 at 19:20

frostdragon said:
With ICD there is no need to manually spread the paste.ever If you do spread it you will actually introduce oxygen bubbles into the TIM layer and thus reduce performance. Small IHS AND GPU IHS pea-sized blob will do, large ihs and gpu the 'X' pattern is the best

Don't think l can complain about my temps at a room ambient of 20'c my CPU:4.6GHz 1.344v via CPU-Z core's 16'c to 23'c +/- 1'c when idling.

Gaming playing Titan fall CPU mid 40's Single GPU 670 as SLI is not suppoted at the moment About 38'c, if using SLI 670 temps on 2nd GPU are normally 3 to 4'c higher.

IC Diamond · 12 Apr 2014 at 17:08

For comparison purposes here is a C/P impression of my 1" Test die

And for those that like the grit here is a copy of the manual for interpreting results

http://innovationcooling.com/images/Prescale - Color Correlation Manual.pdf

SavellM · 14 Apr 2014 at 15:30

frostdragon said:
It is people like the above gentleman is what makes my job fun and interesting. I love creative solutions, now if you have a image with the ICD actually used on the chip die itself, you win yourself a tube of ic7.

Well done!

Can you please show a before and after temps with your setup using icd? If so and if you can still show a photo with icd installed ill upgrade you to a nice IC24 tube.

Once again well done

Frost Dragon

OOOOO I'll have to take it apart to show ya

Unfortunately I dont have any before and after temps

As I was using the stock intel CPU cooler before, and now its under a full WC loop so didnt bother, as there is no comparison.

But lets just say the difference is HUGE as there is no IHS between the die and the cpu block, just IC Diamond

I'll get you a pic soon!

frostdragon · 15 Apr 2014 at 20:00

As promised 2011 pin results :0

Right as I promised here are SOME of the 2011 pin x79 contact pressure testing results.

First up.

Mayhems briefly produced a water block known as the Mayhems 'DT5NOZ'

So here it is:

I have only completed holes parallel to the board as this block has been connected to my wc system already.

as one can see there is REASONABLE contact in the centre but the pressure is not all that great. With little or no contact anywhere else.

We are trying to point out that THIS information is important and DIRECTLY IMPACTS your temps!

The below realtemp results bare this out when one examines the diffierence between all the cores.

Test on a 3960x

So more contact/pressure lower temps across the board AND they should be of a similar temperature to each other. Obviously with contact being just in the centre and little else will produce the results shown above.

EK Supremecy full nickel, holes parallel and across the board.

Overall great contact but pressure is poor. Also the second remount has overall less pressure (pink indicating less pressure than the red on the left)

Lastly the Koolance 370

More contact overall on the right, however a great deal less pressure than the EK block.

Bare in mind that the same back plate has been employed throughout all tests as x79 mounting system is the same on any water block.

So what have we learned on the 2011 pin tests? Great contact and pressure not only increases the reliability of your TIM but DIRECTLY IMPACTS CPU temps!

Hope you all find this helpful and sorry for the marketing spiel here. We are the ONLY TIM manufacturer that seeks to educate our customers and explains the information in a direct and open manner.

PS: Still have to perform the swiftech apogee HD tests.

Regards

Frost Dragon

IC Diamond · 17 Apr 2014 at 16:15

Good stuff

After the first round of tests of the contact profile results I suspected that there was some bowing in the system with the high edge contact and light pressure on center so I made a quick model of the assy.

I took a 30mm sq. copper slug as a cpu, a piece of glass for clamping and support along with a 0.50 sheet of copper to simulate the sink base.

With a couple of pony clamps to simulate the mounting pressure I positioned it with a back light and you can see the identical contact profile that you see in the C/P impressions, basically the sink is pivoting with the pressure on the edges and bowing in the centre.

my initial thought was that the sinks were bowing under pressure to produce that contact profile however from the second round of contact tests they are fine and so eliminates the sinks as the issue.

So my correspondence with Frost Dragon and his analysis.

Essentially then, the built in plate being integral with the board does not allow the MB to flex while the after market ones move or give under stress to deflect the CPU from from contact with the sink - few '000's would do it and give us the results we have seen to date except for the last round of 2011 tests

Just want to be sure I got it complete.

Frost Dragon response

It does........

However a back plate was invented to keep the motherboard MORE straight and allow the cpu block to put more pressure onto the IHS without the motherboard bowing.

What i am saying is that EACH of these blocks have a differing back plate, so this is why i think the back plate solution ALSO comes into play.

The 2011 pin motherboard with its own back plate, eliminates the contributing factor of the legion of different back plates as its the same one for all. This THEN shows by the following reasoning.

IF the cpu is the same
the motherboard is the same
back plate is the same
the MOUNTING mechanism is the same

then the ONLY factor is the BLOCK itself, if it produces different results.

There is also another thing we learned. In general contact and pressure is MUCH better than the other of sockets.

cheers