Well, I took some interest in this unusual FR4 a guy on eBay has:
http://cgi.ebay.com/10-shts-Copper-...Electronic_Components?_trksid=p3286.m20.l1116
Now there are 2 remarkable things here that might make useful thermal boards. The FR4 is only 0.018" thick instead of the more common 0.062". With this thinner board, even thought the thermal conductivity of FR4 is only 0.25W/mK, this works out to 2.835 C/W per sq in of conduction through the thin FR4 IF the heat is evenly distributed across the top and bottom.
Well he board is 4oz, which is 4x the normal copper thickness. This is important because normal 1oz copper has a significant lateral resistance which means the heat does not spread far. With 4oz, I get that heat conducts down a 1mm wide trace with a resistance of 17.5 C/W per mm.
How good of a heat spreader is that? Well, depends on your package, actually the perimeter of the thermal pad is most important. Unfortunately without finite point analysis software this is difficult to solve for with anything other than rough estimates. Luxeon Rebel's tiny pad has 11.1mm of perimeter area. The area immediately around the pad has to conduct through an 11.1mm ring around the pad but that ring's width gets much wider as we go further out, providing more copper conduction per mm of increased distance. At 1/2" radius the added resistance per mm is 0.219C/W because the circle's circumference is 80mm wide.
Maybe someone with software could work this out, but it seems like this board *could* make a custom thermal board with reasonably low resistance at home and work with solderable pad devices. Flatness may be an issue. 0.018" FR4 is paper-thin. I expect you'll be able to bend the board by hand and it'll stay there because the copper is providing all its stiffness not the FR4 and copper deforms. That causes major trouble mating with a heatsink. Also etching deep-copper boards is its own difficulty. CNC pcb engraving, also very difficult because the FR4 is so thin any engraving into the FR4 could cut right through it.
http://cgi.ebay.com/10-shts-Copper-...Electronic_Components?_trksid=p3286.m20.l1116
Now there are 2 remarkable things here that might make useful thermal boards. The FR4 is only 0.018" thick instead of the more common 0.062". With this thinner board, even thought the thermal conductivity of FR4 is only 0.25W/mK, this works out to 2.835 C/W per sq in of conduction through the thin FR4 IF the heat is evenly distributed across the top and bottom.
Well he board is 4oz, which is 4x the normal copper thickness. This is important because normal 1oz copper has a significant lateral resistance which means the heat does not spread far. With 4oz, I get that heat conducts down a 1mm wide trace with a resistance of 17.5 C/W per mm.
How good of a heat spreader is that? Well, depends on your package, actually the perimeter of the thermal pad is most important. Unfortunately without finite point analysis software this is difficult to solve for with anything other than rough estimates. Luxeon Rebel's tiny pad has 11.1mm of perimeter area. The area immediately around the pad has to conduct through an 11.1mm ring around the pad but that ring's width gets much wider as we go further out, providing more copper conduction per mm of increased distance. At 1/2" radius the added resistance per mm is 0.219C/W because the circle's circumference is 80mm wide.
Maybe someone with software could work this out, but it seems like this board *could* make a custom thermal board with reasonably low resistance at home and work with solderable pad devices. Flatness may be an issue. 0.018" FR4 is paper-thin. I expect you'll be able to bend the board by hand and it'll stay there because the copper is providing all its stiffness not the FR4 and copper deforms. That causes major trouble mating with a heatsink. Also etching deep-copper boards is its own difficulty. CNC pcb engraving, also very difficult because the FR4 is so thin any engraving into the FR4 could cut right through it.