diamond thermal grease?

[Tech Geek Deluxe]

Hello:

I'm new here but will jump right in.
I'll intro myself some ... then get into "the project."

I'm an investor and researcher and have overclocked for over 15 years. Geek solved a dozen or so things that only a handful of peeps have done in the world in the Palm, Microsoft and hacking categories. My favorite win OS pre-install boot disks are by Digiwiz and Tech 182 ... found their work duplicated mine ... [but theirs was better.]

My daughter let some of her friends play with her PC and loaded it up with some unfriendly trojans etc ... so whilst cleaning and tuning it [the dual core P3 Tualatin PC] ... I decided to tweak system services a tad tighter than usual ... and I thought I messed her PC up ... so I ordered a new PC setup as described below. [I ended up fixing her PC through a remote Services method] But that left me with a new PC on the way ...

This is where "the project" starts. Well ... ordering parts late at night after some beers and mediocre research had me staging an AMD system. [Stupid, I know ... other PC's in the house are Core 2 and overclock easily.]

I missed the cancel order window and didn't want to pay the 15% return CPU and shipping fee and hassle with shipping and the RMA mobo ... so ... I was committed to an AMD system 5200+ 2.6 GHZ [Theoretical stable overclock of 3GHZ] at $50 net cheaper than a "Quad core ready" mobo based Core 2 system 2.2Ghz Conroe [Theoretical stable overclock of 3GHZ] ... So ... I figured since I give away 1 or 2 computers a year ... that I would just build this one and make it a Christmas present for one of the Kids and send their Old Tech on the way.

Honest ... I would rather "not" have started down this path ... and rather have simply gone conroe. But I enjoy rebates and great deals and free shipping and the thrill of the build etc.

This led me to deciding to remove the IHS [Internal Heat Sink ... on die heatsink] from the AMD CPU and make my own "diamond heatsink" ... or at least try to get a "reality" stable 3 GHZ from the 2.6GHZ 5200+ with lowered voltages and tight specs.

I picked up 25 Carats of .25 Micron Diamond powder for $24 with free shipping. [Now I'm in at only $26 net cheaper] I will make a mix of some Arctic Silver epoxy and another slower curing [Runnier] epoxy and try and get 70-80% diamond powder content.

I will have to fab a form over the cpu die that fits the wafer and has flat surface contact with my Gemini 2 CPU cooler.

Likely I will spread a 200-300 micron film over the outer shell of the Gemini 2's main copper core all the way up the pipes with diamond dust adhered by epoxy.

My theory is that since diamond thermal properties are transfer efficient at the speed of sound ... that the pathway and surface area for heat to travel on are more important than the heat transfer properties of copper and its problems of heat retention. (E.G. ... the larger the surface area and the further the travel for the heat transfer via diamond "speed of sound" heat transfer ... the more efficient the cooling due to "velocity of cooling" transfer ... versus having a dense and thick media [like copper] ... I will work on providing the most "square inches of travel" and allow for the heat pipes and copper to relieve heat loss from the additional heat being transfered across the "skin" by the much faster diamond heat transfer media.

[This phenomena is similar in how speedier electron transfer like electricity finds the path of least resistance across the skin of an object.]

[BTW: Can't wait for Proton based CPU's are here ... much faster than electrons and create far less heat.]

Where non-permanent interface media is required ... I will blend as much diamond content into Arctic 5 as I can ... or take a runny thermal grease and load it up to 70-80% content.

I picked small particle size .25 micron believing that it will provide the highest likelihood of diamond to diamond particle contact in the most uniform way to allow for the "velocity" of heat loss that I'm theorizing is the real issue of the near future for gaining processor headroom.

Hope I don't braingeek bother anyone ... but if others are looking down this road ... I would be happy to provide pics and results ... heck ... I'd be happy to set up a publicly traded corporate shell if the ideas have legs and the right people or there is market capability to corner the market on diamond powders.

I won't have a "Pre-project benchmark" since I'm not interested in proving anything based on non-proveable Pre-IHS and post IHS removal results. I also won't compare post IHS thermal grease to post IHS Diamond.

What I will do is spec the PC and CPU-Z it after I find a suitable stable clock well below its max.

I'm hoping for 34 C Idle and 45 C Full load with normal voltage and a 3GHZ clock. [Custom tweaked in all respects from ram timings to registry hacks.] I'll be satisfied with 42/58 and increased voltage but won't figure all the effort was worth it. I've sen a lot of numbers tossed out there and a few actually seem sound at far lower specs than I'm shooting for ... but I want stable ... actually really stable. I know I can likely clock that CPU this way to perhaps 3.4GHZ [3Ghz without IHS removal and Diamond HS tweak] with my second set of temps and it would appear stable ... but I've overclocked for 15 years and have seen "real" stable as a 10-15% back down from provable stable settings.

• ASUS M2N-SLI Deluxe AM2 NVIDIA nForce 570 SLI MCP ATX AMD Motherboard
• Thermaltake Purepower W0100RU ATX 12V 2.0 500W Power Supply
• COOLER MASTER GeminII RR-CCH-ANU2-GP CPU+ Board Cooler
• 2GB Corsair DDR2 4-4-4-12
• WD Sata 160GB [Not Raid 0] Have to build a non Raid 0 System to find out if BF 2142 has issues with raid 0 systems
• Liteon 20X DVDRW
• Rosewill R5604-BK 0.8mm SECC Screw-less Dual 120mm Fans ATX Mid Tower Computer Case
• AMD x2 5200+

Comments?

[RA1D]

hey there, welcome to HL!

[Tech Geek Deluxe]

Heya: Thanks! ... I hope I can stir up something with my post.

I forgot to mention that I wasn't impressed with the results of previous "Diamond Thermal Grease" compounds or test results ... on the level of diamond content, diamond particle size or the surface area capitalized on for "thermal siphoning."

I think "velocity" of thermal siphoning requires a different mindset than simply grease between two media "old thinking" or CPU to Copper micron surface matching.

Heck ... I would venture to guess that high content small particle diamond media with a wide and large travel pathway would carry heat rapidly away from the core to the slower heatpipe, copper and fin air cooled sink.

BTW: I justified the diamond powder purchase since my Heatsink was free after rebate and I need diamond slurry to mix with oil for honing a sub-micron edge on some high end carpentry tools.

[Stormcrow]

So from what I've gathered you'll be adding a fine layer of diamond dust to the base of a GeminII cooler? While diamond itself is a good conductor, I'm not sure the dust would be. It's not like a solid heatblock where each particle is able to pass on the heat energy to it's adjacent particles due to their compact, dense form. The dust would be scattered, uneven, fragments barely touching or spread to thin in places to conduct heat adequately. Maybe if it was a solid diamond contact plate with the heatpipes attached to that, but the expense wouldn't be worth it. Then again I've never personally worked with diamonds aside from saws and grinders, so I can't tell for sure.
This does have me wonder about the effectiveness of graphite or just plain carbon. I remember hearing about carbon nanotubes as replacements for the standard heatpipe but nothing else after that about wide scale development.

[Tech Geek Deluxe]

The CPU wafer would have a high content solid ... mostly diamond with some arctic silver and epoxy ... replacement IHS tiny block heatsink ... that would have a high diamond content with some arctic silver thermal grease contacting a Gemini 2 Heatsink. The outer skin would have a solid thin film of touching diamond particles perhaps a layer of 200 to 300 microns. [or the potential of 800 to 1200 diamond particles thick ... ~ 1/8 Millimeter thick film.]

[Stormcrow]

Hmm, alright I see what you're saying. Of course you gotta keep us up to date about this project, right? I've never thought of diamond as a worthwile thermal conduct for PC heatsinks but you got me curious now.

[Hitman]

Wow talk about brining a Thread back from the Dead.

[Tech Geek Deluxe]

I'm waiting for the diamond powder ... when I cut off the IHS and fab a diamond IHS and lay a "outer skin" on the copper core and heatpipes ... I'll take a series of pics.

As I see it I will get instantaneous high velocity "thermal siphoning" on boots allowing for higher clocked boots ... plus due to the heat traveling on two paths towards the wind tunnel of the two 120mm fans ... I should have higher sustainable clocks.

It would be like the core having 6 diamond heat relief pipes 12 square inches of diamond heatsink and the volume of copper 6 heat pipes and fins on the Gemini 2 ... so the cores heat would have almost instant transfer over the entire surface area of the copper and along the heatpipe pathway thus making the Gemini 2 super efficient up the limits of the heat retantion properties of the copper and fins being unable to disperse the heat ... but since the transfer takes the majority of the heat farther up the sink towards the 2 120mm fans ... then the lower solid core won't bog down as fast with heat retention.

[gvblake22]

WOW, you're really going all out here! Isn't there a point though where it doesn't matter how fast the heat is transferred because the heatsink fins and heatpipes end up fully "saturated" with heat and the fans can't cool it fast enough? Kindof like the problem with TEC coolers where it can remove the heat very quickly, but it is still very hard to dissipate the removed heat. You've definitely picked a great cooler for this experiment, but I am just wondering if there is any chance of there just not being enough surface area and it being a cooling bottleneck to how fast the diamond grease slurry is "supposed" to move the heat?

[Tech Geek Deluxe]

Agreed ... the theory is sound but the heat removal issue of the HS "keeping up" needs be considered. I think the issues that will give me an edge is the speed which the heat will travel ... the additional cooling surface area and I believe the reduced heat retention of the lower solid core. Test will only show my specs ... since I am not setting out to prove a before and after.

I think the diamond will service heat relief from the core to the Sink in two ways ... but also service heat removal from the sink in 3 ways.

To:
1. Direct CPU to Sink transfer at the contact point.
2. Diamond "skin" film to copper core and pipes.

From:
1. Base of HS should have reduced heat retention since the heat can travel past its lower base faster
2. The diamond path up the pipes will move more heat and "hotter" heat farther away to be relieved earlier by air.
3. The "skin" ads surface to air additional heat removal of "hotter" heat.

I had seen reports of a 12/20 [idle/load] degree Celsius drop in temps with the removal of the IHS and using arctic silver 5 ... those results were well beyond the pack average of 7/12.

I'm hoping to get 10/15 with the diamond based heat transfer IHS, "skin," pipes and thermal grease.

I think using higher diamond content and .25 micron versus 40 micron particles and having direct contact with the cpu die and as much heat travel area for the speed of heat travel that diamond has ... well I think I will get my temp drops and maybe even better.

The Gemini 2 was free after rebate and offers tons of air movement. I may turn my front 120mm fan out versus in ... in order to make the side vents the intake.
As it is now I'm only going to have one 120mm exhaust excluding the power supply exhaust fan [Unless I turn the front intake to exhaust.]