Complete removal after being in a light for an extended period is not an easy task. Pull off what you can by hand and then us an alcohol to clean off the residue. For all the little nooks and crannies, I use plenty of alcohol and a tooth brush. I recommend you don't press this potting mix into all the nooks and crannies if you think you may be removing it later.
Silicone Recipe (Haven't tried yet):
Ground Silicon Carbide (Both Medium and Fine Grit)
Thanks for the ideas guys (except for Garry ). I would like to use something that fully surrounds all the components. I could use thermal paste, but I really don't like the invisible residue it leaves. Very hard to completely get rid of and it interferes with future soldering.
Yep, it's just a release agent. Cool thing about it is that you can lay it on thick where you want to prevent the epoxy from getting into, like under the corners of chips or under exposed leads, etc. It doesn't affect the epoxy curing, either. Just scrub everything clean with an old toothbrush and shampoo afterwards and it's all gone.
I'm planning to use that method for the difficult IOS 9A driver in a big Maglite pill. Works neat for boards where the hot things aren't on the same level, this gives them all a common surface that will have a nice flat surface for contact to the ID of the pill, and even has good contact to the PCB. Any pinch points that make it fit too tight can be opened up with a tiny engraving bit in a Dremel, I haven't found JB Weld to do anything strange electrically, but it really only needs to make good contact with the flat areas on top of chip packages and whatever.
Reminds me of the arguments many years ago about aluminum 'heat spreaders' glued to PC DDR RAM. RAMBUS needed them, as the first chip was worked harder than the others, and thus ran hotter, and spreading the heat from a hot area to a cooler area actually did some good. Then somebody had the bright idea to stick them onto regular DDR RAM which ran all the chips in parallel and so they were all the same temp, with no thought put into it other than 'they look neat so they must work', and all the little aluminum shields did was to keep everything insulated and away from any airflow that might do some actual good.
Only thing that insulates better then air... many pockets of air1. Try to think of a material that doesn't rely on pockets of air. Best insulator in the world? Aerogel, 99.98% air by volume, record lowest-density solid.
Plasti-Dip and copper and JB Weld are all better than air. Is Plasti-Dip enough easier to work with that it makes the lower performance acceptable? I don't think so. 'Better than' doesn't mean it'll work well enough to make a real-world improvement.
Liquid sodium metal is sometimes used to cool automobile valves. I don’t know if a driver is hot enough to liquify it, anyway it reacts violently with water or air. Next best is water, but it tends to pick up ions and hydrolize. I tried alcohol, but it softened the epoxy. Oil has been used in consumer led products. Freon has been used in computers, including the CDC 7600 and the Cray II.
Bare ram vs heat spreaders is convection, for a driver its conduction.
With ram and heat spreaders you are running into where the bare surface of the ram chips provided better convection (transfer to air) then the spreaders. Spreaders are just flat colored metal thats trying to look cool. No fins to improve convection.
When you have a driver enclosed inside an airtight host you need to conduct heat to the host before it can be shed.
Yep, nearly anything is better then air for conducting heat away from your driver.
As long as you don’t use wool, cotton, sawdust, sprayfoam or anything else that comes with tiny pockets of air.
No, you're missing the point here. It's the same amount of effort to use Plasti-Dip and JB Weld, yet they have vastly different heat transfer abilities. Does it make sense to use the material that doesn't work as well, simply because it's 'better than nothing'?