jb weld cure is highly dependent on temperature. if it is too cold - never. If it is spread out at room temperature, 16 -24 hours. If you pour it in a cup, then the heat that it generates (once two components are mixed) will get the hole thing hot and it will be just a few hours.
edit: the stuff is STRONG, and handles higher temperatures than you'll see in an LED light. Higher than anything else I've found actually, including devcon aluminum putty.
How about this?
being it conductive doesent bother me at all or should?!?
I wonder how good or bad the thermal conductivity could be. It seems usefull stuff anyway.
nice find! that's as cheap as I've seen for conductive adhesive.
probably higher resistance that the stuff w/ silver in it, but i ordered some to test it.
I thought of trying copper gasket sealer. Googled it a bit, I'm not the first person to wonder, but in the end I ordered this and it showed up in my mailbox 2 days later
http://www.newegg.com/Product/Product.aspx?Item=N82E16835100013
Mix artic two part epoxy with a little ceramic paste for less hold.someone suggested once , mix the epoxy then add and mix in x amount of ceramic paste. A third or so, have to experiment a bit to see what works for you.
There are two good options for what you're asking about. Both are around $8.
Semi-Permanent (Hexatherm Tape) - http://www.ledsupply.com/hexatherm.php
Permanent - (Arctic Silver Alumina Thermal Epoxy) - http://www.ebay.com/itm/Arctic-Silver-AATA-5G-Alumina-Adhesive-Thermal-Epoxy-5g-/170708546413?pt=LH_DefaultDomain_0&hash=item27bf05b36d
[quote=Rezolution]
There are two good options for what you're asking about. Both are around $8.
Semi-Permanent (Hexatherm Tape) - http://www.ledsupply.com/hexatherm.php
Permanent - (Arctic Silver Alumina Thermal Epoxy) - http://www.ebay.com/itm/Arctic-Silver-AATA-5G-Alumina-Adhesive-Thermal-Epoxy-5g-/170708546413?pt=LH_DefaultDomain_0&hash=item27bf05b36d
[/quote
This might be similar, alot chaper but you have to cut it yourself.
A new test. Unfortunately I don't have premium stuff as artic alumina, only from DX a "star" thermal grease and fujik. You will find an interesting material on the graph. :)
I used an old ssc p7 led at constant 3A, and measured the temperature of the star.
here is the test configuration:
and the result:
nothing means: there is nothing between the star and the heatshink
that's a much better test than the previous. I think it shows that what you should seek is something that has the other properties required (price, setup time, how permanent, electrically conductive, etc.), and then make sure what is applied to the thermal interface is done so properly (thin and even).
personally, next time I need to do one, I will use arctic ceramique under the star, then compress it with a clamp and put the arctic epoxy just on the star points, from the top, so that I can remove it easier down the road.
Nice test, Viffer750. Cool heatpipe setup, too.
dthrckt - thanks for the idea of using the copper sealant.
Budgeteer - that's the cheapest I've seen for thermal tape, thanks.
OMG!!!!!!!!!!!!!! (sorry to bring up such a old thread but I just saw this. The reason I'm freaking out is I thought the Fujik was supposed to be a fairly good thermal compound)
Basically the test shows Fujik is crap. Using toothpaste was just as good thermal paste. I'm totally re-doing all my star to pills with something better than toothpaste!
What the hell!
Fujik is best used for potting, it is no where close to as good as Artic Alumina or a solid metal-metal contact.
Agree. I've been using Arctic Alumina Thermal Adhesive - would love to see so called quality stuff on that graph. Thought other postings (maybe on CPF?) have shown Fujik as not so good (crap?), so not surprised here. Maybe crap should be added as well - thermal characteristics dependent on freshness... 
Cheap Trick: mix epoxy and thermal grease 1:1 and you got your own cheapo thermal adhesive.
Also note to use it as thin as possible, this ensures that the bond is strong and that it’s conductive and no isolator.
This will likely yield results about as good, or perhaps a little worse than toothpaste. I actually doubt it will be any better than epoxy by itself.
PPtk
The advantage Fujik has is removability. Cured, it can be cut with a knife and withstands the temperatures necessary. I use it for potting or fastening LEDs temporarily for testing. Apply and maintain pressure until cured. This could be several hours for mcpcb’s or a few days for potting. But otherwise it really is just glorified caulk.
It's probably electronics grade silicone, similar to caulk minus the corrosive curing agent. It's not bad for potting and, for practical purposes, not too bad for heatsinking flashlight LEDs either. Consider that most LED flashlights need to be upgraded after 3-4 years anyway and I don't think upgrading from Fujik to a more "premium" thermal compound is worth it. The performance gained by upgrading to easy-to-find "premium" compounds like Arctic Silver aren't that dramatic. And the really good stuff is hard to find and is prohibitively expensive. Fujik, for budget lights and if it is applied correctly, is not a bad choice.
In several rather well done tests of thermal compounds used in CPU/heatsink applications, the differences between the best and worst performers were minimal (a rather few degrees C). I suspect that JB Weld will perform as well as AA. And RTV will work as well as Fujik. Just keep the layer as thin as possible. Flexible stuff like RTV/Fujik tends to survive thermal cycling better than rigid stuff.
For potting and thermal adhesives I have used silicon carbide in epoxy or silicone rubber. It has MUCH better thermal conductivity that aluminum oxide. See: High Performance Potting Epoxy - Interested? for some discussion.
I have also used diamond dust in my thermal adhesives/greases. It is not very expensive and nothing conducts heat better. For the carrier in thermal greases I use high vacuum grease or oil.
Well, the thermal adhesives you can buy are actually the same 
Thermal glue or paste is always "crap" compared to lapped surfaces. But its better than nothing and they all perform similar. I dont feel bad about using Fujik.