The board supports are in one of the trays I used. They're not part of the tray itself, just some standoffs I affixed to the pan. This worked quite well. I have seen a Power Macintosh G5 board curve under extreme heat, which I consider bad enough. I don't intend to let it happen to one of these otherwise valuable MacBook Pro boards, that's for certain. There is no downward clamping force, only elevation at six key points of the board (one each corner and two larger supports in the centre). I find it's not only necessary to prevent warping of the board, but also to get the heat moving evenly around both sides of the board.
The addition of a liquid flux to the mix is next on the order of experimentation with the nVidia 8600M GT BGA. I do need to find one that can flow in under the chip package and be removed by chemical means as I'm not removing the GPU from the boards at this stage. The reflows at this temperature and time, allowed to slow cool, do seem to be relatively stable however. I performed one with an Early 2008 board about 24 months ago and it's still ticking over.
At the moment I'm focused on experimenting with dispersing the heat faster and drawing the heat away from the chip with fewer restrictions. More efficient ceramic based thermal compounds with greater longevity than the factory silicone based material that tends to dry out and go brittle, machine surfacing of the heatsink contact points and greater surface area on the fin stack are all ideas I have in mind.
The addition of a liquid flux to the mix is next on the order of experimentation with the nVidia 8600M GT BGA. I do need to find one that can flow in under the chip package and be removed by chemical means as I'm not removing the GPU from the boards at this stage. The reflows at this temperature and time, allowed to slow cool, do seem to be relatively stable however. I performed one with an Early 2008 board about 24 months ago and it's still ticking over.
At the moment I'm focused on experimenting with dispersing the heat faster and drawing the heat away from the chip with fewer restrictions. More efficient ceramic based thermal compounds with greater longevity than the factory silicone based material that tends to dry out and go brittle, machine surfacing of the heatsink contact points and greater surface area on the fin stack are all ideas I have in mind.