+100
if you haven’t - try lapping a star to a pill.
I’ve found that it works so well I have to put slide something between the two in order to solder leads well.
After lapping I apply AA, work it around a bit then scrape off all excess w/ something flat.
If it HAS to be secured with adhesive, I clean the pill around the star w/ qtip and solvent then add arctic epoxy or jb weld around the edge, and often, even over my solder joints.
So basically:
Thanks got the input guys. Great discussion and apologize for my rant above.
I'm confused. I thought I've seen Fujik being bashed elsewhere, and how someone had temp stats showing high quality epoxys used for mounting the emitter star were superior in overall temperature distribution (or temperature reduction on the star) - I though it was in several degs C. Don't mean to diminish viffer750's tests or his test setup because it looks quite impressive and appreciate the effort in posting all the info and pics.
The value of "potting" could be debated for heat management - it's also been debated elsewhere and not sure if there were any scientific results one way or the other.
If I get a chance this weekend, I have to find the source postings.
but it’s another to make sure there is an even firm pressure applied much less even contact. Without that, the minutest air-gap between surfaces will yield not so great results. Plus whenever ya got heat and cooling gaps will change without pressure being there to compensate - much less even flexible stuff like Fujik. The point being is that most of us can’t lap perfectly flat surfaces due to machine deficiencies, tooling, etc. So ya need a ‘filler’ to take up the difference and some way short of welding or soldering than can apply pressure to get the most out of it. A silver-based thermal conductor compound is prolly the best, along with glue or grease-like compounds containing diamond particles.
Either way, ya still need some compound flexible and fluid enough to fill-in those micro-pores and valleys to get the best results, and constant pressure really helps too. Intuitively, those torch designs where the reflector applies pressure to the star should give optimum results along with those that use a screw-down arrangement.
The retaining nylon rings ya see fairly often are not enough by themselves to get the job done right without said reflector pressure or other method of applying pressure. This is where fiddling around with different shim thickness materials between them and reflector bases that come up a wee tidge short can pay dividends with regards to minimalizing effects of thermal-caused led output sags. And screwing the little basturds down nice and even is another approach if a reflector-base solution isn’t too swift or ya don’t have a retaining ring.
Bottomline is that the best metal-to-metal contact even with thermal compounds will still always require pressure for optimal results. Personally given my druthers I’m not too crazy about thermal ‘glues’ that harden either ’cause typically they’re a bi*tch to get off if ya need to someday - plus I can see where there could be a propensity for cracking and eventual disintegration at certain points. Hence pressure applicatiion combined foremost with thermal grease and secondarily flexible glues like Fujik wherever possible.
This is one engineering advantage where pressure-applying plastic reflectors can ‘shine’ without worry of shorts, all other alternate solutions considered. Although if you’re gonna go plastic reflector and the led is high powered enough (heat) you’re gonna have to have some kind of insulator between the reflector base and the emitter or else it’s prolly eventual melt and warp time. 
I sometimes make natural delrin washers parted off on the lathe for that kinda deal. I like delrin. :bigsmile:
For me, I still prefer the stiffness of aluminum reflectors wherever possible though given a choice due to potential ‘warps’ I can envision to a plastic reflective surface from either uneven bezel pressures applied to the reflector top-lip or other flex induced aberrations. Prolly not really an issue though for what we do with these things. After all, we’re not doing Lasik surgery here. 8)
Ps. You don’t necessarily need a lathe either for these insulator washers - but of course ya can always buy them too from the Kaidomains of the world. Ace Hardware, Home Depot, Lowes, etc., usually have a pretty good selection of different thickness/dimension nylon-type washers with donut holes already cut out for ya. Heck some may even be made outta delrin for all I know too.
Don’t use toothpaste, it just works till it is hardened!
If I glue a star on a pill with arctic alumina I press it wit a nut above the led in a bench vise. This works very well.
lumatic makes a good point.
the screws that hold together ide/sata hard drives work pretty well for screwing down stars - when they won’t hit the reflector.
if you drill the right size hole (you need a set of numbered bits, not fractionals), they’ll self tap into aluminum. copper/brass not so much.
but i must say, lapping isn’t about getting something flat - it is about mating parts.
i desolder the emitter, hot glue a punch to the empty star and lap the star to the pill. they’re not flat, but there’s a very high % of contact. when you START lapping, you can see (as metal is removed) that there is about 10% contact (on average…I had a few that were about 2%…just the tips of the star on 20mm pcb).
perpendicularly true you can get two mating surfaces relative to one another the better. This way potential minimizing mis-alignment issues from head, to star, to emitter on through reflector surface the better. IOW centering to me not only involves getting it in the center of the reflector port but also perpendicularly-centered. Especially with the smaller thrower dies. That’s where ya can see real gains in perfecting the beam and optimizing the reflector’s potential.
So if things were ‘cocked’ a bit because of now ‘mated’ surfaces that originally weren’t as ‘flat’ as they could be then ya got a different deficiency. Ya got potentially the pill surface ‘flatness’ itself and/or the mating surface of the pill edges onto the body of the head, and if a screwed-in pill the concentricity of the threads, plus the relative flatness or perpendicularity of the star mating surface to everything else. Conceivably if you had a crappy uneven flow-solder job ‘cocking’ issues with the emitter as well.
Basically though this is where good CNC machining at the factory really pays off. And as you prolly know by now, at least for me, I can immediately tell whether good machining was done or not. You’ll see it right off the bat for instance on how a threaded pill (or non threaded pill mates up) to a machined head surface concentricity-wise.
So you could ‘hone’ to get perfectly ‘mated’ surfaces that are now so to speak indeed perfectly-mated to one another but still be cocked at an angle as an integrated unit to everything else. Perpendicular or concentric finished products at least for me is what I try and aim for when it comes to quality.
Hope this makes sense. H)
Ps. Dthrckt………do ya get those screws on-line?
no, i take them out of dead hard drives.
before I moved I had at least 30 (from work), so occassionally I take them apart and scavenge the magnets and screws. Didn’t feel like moving them though, but I’m still in the same line of work so I bet I can scrounge more 
they’re stainless torx head screws, with wide head that’s flat on both sides . there are usually two different thread pitches in one drive (if you take it all the way apart). the fine thread don’t self tap very well, but the others do.
I use Fujik and will keep doing so because I like the ability to separate the components if need be. I also lap my stars then apply a small dab of Fujik before I clamp them extremely tight. When I have had to separate the two I've found the thinnest of this layers of fujik underneath and this is what you need to have. Just pushing the star into the pill and working it around to even out the adhesive isn't enough. You have to push out all the adhesive as possible under pressure until it cures leaving a literal film to fill in any surface imperfections and eliminate air pockets.
EXACTLY! I learned that the hard way some years ago. If you want an INSULATOR…. that is the best! Ha! One of the first lights I did needed bedding… so I just mixed up a gob of 5 minute and away we went. Epoxy is Epoxy… Right…. NOT!!! Ha! Can you say POOF! That little curl of smoke was attractive though!!
Dan.
Anyone know Fujik’s thermal conductivity ?
may be following adhesives helpful
http://www.acc-silicones.com/products/adhesives/thermallyconductiveadhesive.ashx
For me the value of potting is not in thermal conductivity but in making the wire connections more robust by adding a strain relief. Close contact with metal using a thin layer of fujik as an adhesive for all of the hot bits. Allow that to cure and then pot away.
OK guys, so what’s the consensus here?
Can someone link me to something I can buy that’s better than crappy Fujik but cheaper than Artic Alumina?
I'm not finding compelling evidence one way or the other. I did just get in Fujik grease, not glue, and found the 20g container about 1/3 to 1/4 filled - total waste of $$$, and for that money, I would hve been better off buying an Arctic product. For the epoxy, I went with the Arctic Alumina because of the cost/availability (cheaper than Arctic Silver) and I believe the quality is there - it should last longer than tooth-paste 
, and if I make it a little too thick, I know it will perform better. The evidence is that in thin layers, it doesn't make much of a difference, but in thicker applications, the quality stuff does perform better. I've seen BIGWOOD's posting for help on choosing and the answers over at CPF were not direct - things get so far off topic and theoretical, it's depressing...
Seeing how long a little bit lasts (I've done 8 or so stars, plenty more left to go), I probably would have bought the Arctic Silver Thermal Adhesive if I were to do it over again, $8 vs $15 (NewEgg.com) for slightly more volume (5g vs 7g)? The cost per star assembly should be pretty small - I'm guessing 50-75 cents at most? I think the Arctic Silver products, or similar quality, are a good choice for us DIY hobbyists who just want to buy/have something and get it done, and knowing it's pretty good quality, easy to apply - good instructions, lot of DIY'ers are using it, etc.
I'm not sure where I bought mine, but not from NewEgg - think I google'd around and found slightly better price.
How do you lap the surface of the pill? I have lapped led stars and cpu processors and heatsinks using fine sandpaper on a flat glass, but the pill has the ridge on the edge that prevents direct contact with the surface to be lapped, please share your technique and pics if possible, Thanks!
@AlexGT: I used sandpaper glued to a cork or something round. Just polishing with the Dremel also works pretty well. On one of my last builds I just removed the outer lip and that worked as good.. so there are many possibilities.
Biggest thing with epoxy or silicon is less is more...a incredibly thin layer, you want it just to fill in the spaces between the metals and adhere, otherwise too much it becomes a thermal insulating "gasket"
If you are really worried about it get some http://www.arcticsilver.com/ta.htm
These guys show silicon (probably the element, not the squish tube of stiny goo) is better than epoxy thermally
Also heatsink compound is supposed to be applied ULTRA thin, just enough to fill the microscopic gaps between the surfaces NOT a gasket
Fujik is crap.
The stuff in the tube is silicone, with an 'e', a class of silicon compounds.