Manufacturing perfectly round, tight fitting heat sinks from stock material is difficult without a lathe. While trying to ‘forge’ a 20 €-Cent piece (high copper alloy) into a smaller diameter - it quickly bent out of shape, didn’t work well - I got the idea of melting and casting metal into round shapes instead.
Casting metal directly into a pill wouldn’t work, the pill is usually from the same material (aluminium) so it would melt or deform; a mold needs to be made. I chose wax for the positive and a 1:1:1 mix of fine, sieved sand and gypsum for the negative, that is the mold into which the metal is poured in the end.
The positive consists of two parts, the one from casting wax into a hollow pill (rested on a CD case to get a flat base, put silicone grease in it first, easier to get the wax out, also put some weight on it so the wax can’t lift the pill and flow through the gap) and the ‘sprue’, which in my case was ~1 inch from an old candle. The first 1/2 inch of the pour will always shrink heavily, hence a sprue, preferably a bit bigger than the primary part, is necessary. Complicated molds also use a well to trap dross and impurities, as well as vents to allow water vapor to escape, risers to indicate the filling of the mold and a gating system. The ends of both pieces are melted together, gaps are closed with more wax.
Different model. This pour failed, the sprue was too narrow. The tissue is unnecessary, just use toothpicks or an old toothbrush and rubbing alcohol to remove wax from the pill.
The wax form needs to be adjusted for shrinkage, ~1.25% for aluminium, 1.9% for copper but casting that is a lot more difficult.
A 19 mm wax round will result in a 18.5-18.6 mm cast (2x2.3 mm since it’s not a one-sided piece) - not a good fit. Adjust for shrinkage by dipping the piece into molten wax, once or multiple times; measure with caliper. This can be done before or after connecting it to the sprue. I recommend to over-adjust slightly and to make multiple molds, they’re small and cheap.
The negative is made from the mix mentioned earlier; gypsum holds it in shape, fine sand prevents cracks while not deteriorating the surface finish too much. My method is more hamfisted than traditional lost wax techniques, instead of dipping the wax positive into the slurry I just wait ~5 minutes for it to solidify a bit, then I stick the positive in. ~200g of this stuff will suffice for a 100 ml container. I used a joghurt cup, it is cut away after a few hours. Then the mold is turned over and baked to remove the wax and any water in the negative.
The latter has the unfriendly characteristic of expanding by a factor of 16 when being hit by molten metal several 100 degrees hotter than itself, that is it will explode and potentially send metal and mold pieces flying. Avoid this at all costs, pre-heat tools, scrap and mold before introducing them to the metal; either in an oven or by resting them on top of your furnace (or just hold them into the flames for a few secs if there’s no better option).
Speaking of baking the mold, this part is a bit tricky. Increase heat slowly to prevent cracks (<100°C/h, min 3h) and don’t go over 230°C if you bake it in your own home; I’m not 100% sure what it is but I suspect the negative absorbs part of the wax as it drips out, later at higher temperatures it burns/vaporizes and emanates ill smelling smoke. The traditional method uses a kiln or large fire to burn out the mold.
Aluminium is easier to melt, it doesn’t oxidize quite as readily and melts at lower temps (i.e. requires less or no additives like fluxes, de-gassing agents); 460-670°C vs copper’s 1085°C (pour temp is 1-200°C higher to keep the stuff sufficiently fluid while flowing through the mold). Copper does however dissolve in molten aluminium, just as salt dissolves in water despite its melting point of 800°C (well not quite, but it’s a similar concept). 10% copper by weight can be added without causing brittleness, I’d not recommend doing that on the first pour due to the oxidation issue mentioned earlier.
A furnace can be constructed from steel tubes of various origin (oil drum, beer barrel, bin, paint bucket) and refractory like fireclay/chamotte, insulating agents like perlite, shredded styrofoam and filler (silica sand) to prevent cracks. Construction, proper firing and usage of a real furnace is better explained elsewhere http://prometheus-foundry.com/tutorial.html.
For small amounts of aluminium a coal starter, fired with coal and additional air from a circular blower or vacuum cleaner is sufficient. Melting more than 1-2 pounds in such a setup is not possible, necessary heat isn’t reached and held due to total lack of insulation. You can improve efficiency slightly by adding an improvised lid from an old cake form or some other fitting shape, just make sure to include a vent hole in the middle.
PVC pipe, coated in the same slurry used for the mold to prevent it from melting, was used as an extension for the blower. It aims at the spot directly below the crucible (cooking pot, also in the pic).
For higher melting metals ceramic crucibles made from graphite and other conductive but highly refractory materials are used. They need pre-firing to prevent cracking and are expensive. Steel crucibles, while oxidizing quickly under furnace conditions and not as long-lived as their ceramic counterparts, are commonly used to melt aluminium, brass, bronze etc. Steel melts at 1500°C+. Steel or iron cups, steel pipe with a base (and for thicker/heavier models lifting lugs) welded to it or cooking pots can be used. If the material is thinner than 1.5 mm you should not use it more than 2-3 times, it will oxidize and break open.
Generally speaking, if your makeshift furnace can get such a crucible to cherry red colour it will melt aluminium.
Put your baked mold onto some sand (for insulation), heat up the furnace, wait for the crucible to reach cherry red colour (occuring layer of oxides will help to slow the aluminium dissolving the steel), put your (well dried/dehydrated!) aluminium scrap, broken into small pieces into the crucible. Once it has melted, wait a few minutes more, then skim of the dross (oxides) from the top (a spoon welded to a piece of rebar works well), then grab your crucible using leather gloves, large pliers or a holder (again, welded rebar) if yours has lugs, hold it close to the mold to reduce exposure to air and pour quickly before it freezes in the pot. Good luck.
Results:
Oxidation on sprue part, likely from remaining wax. Mold wasn’t baked sufficiently long (ran into problems with smoke formation…). Rough surface - maybe due to water vapor, same issue. Nonetheless, surface is plane, polished coin (nordic gold, cut to size and put into the mold before the pour) used in this piece seems to have been fused to the main body as intended. Another factor may have been mold temp, usually >200°C hot molds are used to prevent the metal from cooling too quickly or so I’ve heard. Mine was at room temp.
I also forgot to add groves for the wires. The HS isn’t cut into the right length and the coin needs some more planing/polishing, but apart from that…
… it’s workable! Not as tight a fit as I’d had wanted, ~18.9mm instead of 19. This is with one additional coating of wax. Could fill the small gap with thermal glue.
No finished piece (working FL) yet, bought driver of the wrong diameter 
Image mirror:
http://postimg.org/gallery/2cxlvg3mm/c7bb916f/
