MAX COOLING without THERMAL PASTE
THERMAL PASTE ALTERNATIVE FOR MAX COOLING.
From an engineer, who is not an overclocker
CPU= Basically an, assy of silicon wafer / board / terminals, topped of with a metal cover and filled with resin. And then plug it in and get it gets hot.
With multi core processors the ‘hot spot’ can move, I guess a thermal map of the metal cover would look like something off the 5-day weather forecast. During this heating the metal cover ‘et-all’ is not going to thermally expand evenly. Try heating up a biscuit tin lid with a candle (boi’ng). To some extent the base of the heatsink will also want to do its own thing.
So You have lapped the mating CPU / heatsink surfaces so flat and finely that you could shave in them (nice one!), then add some thermal paste, screw it down nicely, and then set about revving the nuts of it. (Ever thought about hot lapping?).
The variable amounts of thermal expansion, due to the ‘thermal map’ and dissimilar metals are going to alter the contact pressure and geometry. Perhaps even create a localised gap in the mated surfaces, a gap less than a tenth the thickness of a cigarette paper is a chasm for heat conduction, (will the thermal paste flow to take up the void). If it wasn’t for the delicate electronics I would braze them together, (ok) So no heating above CPU max,T.-- How about fluxed liquid metal say mercury, it might work, but would be dodgey and problematic. So what’s left regular solder is to hot. Why not try Woods or Fields metal (wiki) you can melt them in a cup of tea, or 62 – 70 deg C to be precise, these alloys are a couple from a whole group of Low Melt Alloys, how about LMA.
How to apply ‘LMA Technology’
Small amount of LMA, Suitable non-corrosive flux, A couple of thyristors calibrated to LMA melting point. A hairdryer.
Flux and ‘Tin’ and hot wipe the heatsink and CPU mating surfaces (note: aluminium probably bad /copper probably good). Remove any excess LMA from edges.
Install CPU and mount the small thyristors on side of metal CPU case and test by warming case. Ok that works.
Assemble heatsink with CPU applying a thin smear of flux on both tinned surfaces, but don’t fully tighten down heatsink.
Warm up heatsink with hairdryer until the thyristor readings show that the LMA melting point has been exceeded by a couple of degrees or so. Torque up heatsink and then cool things down (now for the science bit, use the hairdryer without any heat).
It would be an idea to trial run the heatsink warm-up before tinning as you will already have the calibrated thyristors hooked up to your multimeter. To calibrate thyristors flatten a drinking straw and seal up one end, insert hooked up thyristors into straw and tape straw to an accurate mercury thermometer the thyristors beside the thermometer’s bulb, and suspend in a glass or jug of hot water and take readings as the water slowly cools.
What Flux?? suggestions!.
Note- if CPU operational temp went above LMA melt point and oxidation was not present the LMA would re-flow, as long as excess LMA is not present this could be a good thing, settling thing in?