INCONCLUSIVE, BUT PROMISING
DIFFICULT APPLICATION
The CK4800 medium has a consistency somewhere between AS3 and AS5. At the moment, I don't have a Mettler balance to weigh the diamond abrasive and mix it by weight, so I attempted to mix equal volumes of the medium and the diamond particles.
People have noted that AS5 is "just a little bit" hard to spread. A 50-50 mixture (by volume) of the particulate diamond and CK4800 is too much like mixing cement without enough water, or a beurre manie or gravy roux with too much flour and too little butter.
It would be acceptable if the stuff was just "hard to spread," but it doesn't stick. It seems more to want to stick to the spreader-paddle than to the processor cap. Then, I would add a quarter-grain of rice-worth of CK4800, in attempt to make the stuff more of a paste and less of a mess.
Finally, I just dumped a very small amount -- maybe a fifth of a coke-spoon's worth -- on the processor cap with the "badly mixed cement," and I attempted to spread it around a bit by chopping at it (as if I were making "lines" on a mirror with people in nosey anticipation behind me. [Just joking -- never tried the stuff -- watch too many TV crime-dramas . . . ] At that point, I decided to drop the Ultra-120-Extreme on the processor-cap hoping to squish the stuff around enough to get it to spread evenly.
Once I got the heatsink squarely on "the mess," and tightened the screws, I twisted the heatsink to spread the mix around more evenly.
Also, my testing at this point leaves something to be desired. I should turn on CoreTemp's logging, first of all. So I'm the first to admit at this point that I myself am afraid of biassing the results. What I had done before, and what I did this time, was to return to the machine every ten minutes and watch the temperature values for a three-minute period. At first, I watched the temperatures for a solid 16 minutes:
Idle Start: #1 = 22C #2=21C [room ambient 73F]
Observed LOAD VALUES, 0 to 16 minutes: between 34C and 40C -- both cores #1 and #2
Observed LOAD VALUES at times between 17 minutes and 45 minutes:
#1 #2
min 34 32
max 45 45
The high values tended more to stay within the 41C to 44C range than with "CONTROL #2." Instances where temperatures reached 45C seemed shorter and less frequent. Further, temperatures for both cores stayed within the range of 34C to 40C during the first 16 minutes and the last 15 minutes of the ORTHOS 1-hour test period. This also "seemed" to show lower temperatures more prevalent for a longer time than for the Control #2.
I think before I add the "Silver Eagle" to this mix, I will run this same test again and log temperatures with CoreTemp. Then, there will be a basis for a solid statistical average.
The local coin-dealer says I can use his Mettler balance, which has an accuracy of +/- 0.1 gram. I have a feeling that this diamond powder needs a medium with properties that allows it to spread and stick, and that othewise, we're not going to get a 70% to 90% diamond mix by weight.
In the meantime, I suggest laying down a rice-grain's-worth of CK4800, and sprinkling (oh so carefully) enough diamond dust on it in a uniform layer, then pop on the heatsink, twist a few times, and let-er-rip.
I'll be back with a re-test in the next few days, and then "on to the Silver Eagle." At least, if the coin-trick degrades performance and temperatures instead of improving them, I can stick in my popsicle sticks and remove the heatsink without taking out the motherboard.
Also -- anyone worried about the coin "dropping out" and causing problems needs to understand that the particulate diamond is an abrasive, and offers the same resistance to movement that occurs with some 320 or 400-grit wet-or-dry sandpaper without water.
There are also open options about other "mediums." One could use a silicon-aluminum_oxide-boron_nitride paste such as can be found bundled with heatsinks. We also considered using "Cool-laboratory's "liquid metal." I have misgivings about this latter idea -- profound ones.
The diamond dust clings to the vial in which it was shipped. The stuff is non-conductive, and opinions have been offered that it cannot become statically charged. However, the Cool-laboratory product is hard enough to spread because it behaves like mercury, and one wonders what might happen if it would cling to diamond particles, or if those particles might then become airborne.
Since diamond is preferred because it is electrically non-conductive but thermally highly conductive, I won't even bother mixing this stuff with AS3 -- certainly not AS5. The CK4800 has about the same consistency as AS3, or only slightly thicker.