Did I just hear PflexPro volunteer to do everybody’s re-flows? 
…did I mention I’m moving to Zimbabwe next week?
Ndicha tumira iwe pakedyi.
Tenda iwe.
What he said.
all I can figure out is something about a package -right?
Let me know when you get there.
I’ll send you a package.
Thank You.
Dale, what do you do when you want to replace a LED? Is there any tips for removing stars which have been glued by AA adhesive?
Thank you very much pflexpro. I always like to see some test and some information being confirmed. 
So as a conclusion Fujik is still not bad for flashlight application?
Lothar, since I always use copper stars removing one that’s been AA’d really isn’t a problem, I hold the soldering iron to the edge of the star for a moment while putting pressure on the star with a small screwdriver and when it heats up a bit it pops loose. I don’t find that Arctic Alumina Thermal Adhesive is “permanent” like they say it is…I’ve potted a driver with it before and removed it fairly easily later when there was an issue. Just holding the soldering iron close while prying the adhesive loose with the tip of an XActo knife works pretty well. It doesn’t set up rock hard like epoxy does, but much more so than the silicone type thermal adhesives. The heat causes it to become flexible enough to pry it off, even if in pieces.
Can’t see the reason to use Fujik under the star. Use Fujik around the edge to fix the position but a decent thermal compound under the star.
Thank you very much for this test. Looks like I'll be using AA thermal adhesive instead, although I wish I didn't have to do the two part mix and have it set up so fast.
I wonder, what if a little Fujik were added to the Arctic Alumina, would it slow down the cure and give a little easier removal if/when necessary without compromising too much on the thermal abilities of the AA?
Sometimes, actually most of the time, I just use Arctic Alumina Thermal Paste, instead of the adhesive, particularly when the emitter might need to be able to shift in order to center.
Difficult lights, like the 3 emitter SupFire M6, I glue down to keep placement…just have to plan ahead and work fast.
When I need to remove a star that is epoxied with AA, (usually I have the pill out and driver out), so I just heat from underneath, with a torch, for a few seconds and when the pad gets hot, the star will come off. Doesn’t say a lot for AS and AA when heat degrades it.
I'm not convinced such a small difference in Fahrenheit really does enough to warrant all that work. Soldering a star to the pill is about the best heat transfer. Screwing them down is good too. I just don't worry about them any more. In all the mods I have done, I haven't found any with Copper stars be of any problem, no matter the way they were adhered. Paste, Adhesive, screwed down, it's all in how much mass behind them, in the long run and most of the time, the batteries sag before there are any real issues.
Unless it's a Maglite Solitaire...
I always have a habit of being late to the party, and this carries on with that tradition.
I ended up finding this thread after I did a similar sort of halfass test on some materials for my own purposes. I did both thermal and mechanical strength tests on both STARS-922 and Permatex UG (and other stuff)
http://epicbeardquest.blogspot.com/2015/06/measuring-thermal-resistance.html
http://epicbeardquest.blogspot.com/2015/07/revisiting-stars-922-and-other-stuff.html
I don’t know that my thermal figures are 100% accurate (questionable setup and small sample size), but I really don’t trust any of the specs that are provided from sellers either. For STARS-922, I’ve seen conductivity figures all the way from 0.6 W/m-K to 1.22 W/m-K.
For what it’s worth, UG is not that bad. The 922 had a much better conductivity in testing, but I might prefer the UG for things where I want a bit more assurance of mechanical strength (especially over time). The key with UG is to get the film thickness down. Because of its high viscosity, it can be a bit difficult for large areas.
From what can be gleaned from Permatex MSDS literature (they don’t like to be revealing), Ultra-Grey RTV is the product with the highest filler content. Assuming the mix ratios and particular conductivities of constituent fillers aren’t going to be as good an indicator of thermal conductivity as density alone, I don’t see much need to consider the Ultra Black or other products. I’m also not going to test JB Weld or epoxy; I want to be able to re-use my test fixture!
I never had much luck using spacer threads in my case. I kept getting bubbles trapped in the interface because the thread tends to channel the paste. I ended up wasting a lot of compound that way, and the results weren’t really useful. Of course, a lot of my test methodologies could use revision…
The reason I used monofilament fishing line as a spacer was because I wanted to be sure of consistent material thickness and I think the fishing line did a good job of maintaining the thickness. I didn’t mount the test plates without the monofilament spacer because of the different viscosities of the test materials. If I had not used the .5mm line for an absolute thickness, then I’m sure the RTV would have been much thicker than the others. Since I was doing a comparative test, it didn’t matter the actual thickness of the test materials…just that they were the same. If I had used pressure to set the material thickness, I’m sure they would have been all over the board. Was it a perfect test…absolutely not, but it answered the questions I had about the materials.
Comfychair suggested not using a spacer, but because of the minute variations of the test plates, one may have 10% metal to metal contact and the next may have 15% metal to metal contact. A difference of 5% metal to metal contact would have a big difference on the test results.
By the way, I really like your test setup…what I did was just quick and dirty.
In my writeup, I mentioned that I had pretty much tried to do the same thing you were doing to get volumetric conductivity figures (W/m-K), but I ran into a few hassles and didn’t bother to work out the bugs in the shimming/spacing method.
The reason I did things the way I did was because I was after typical specific thermal resistance (K-m^2/W) figures for things like screw-mounted semiconductors. Specific (per-area) resistance is convenient because it includes quantities that are hard to quantify like film thickness or resistance that occurs at the boundary between the thermal material and the substrate (take the case of a Sil-pad). In my case, I felt that it made made life easier to have the typical film thickness built into the figures. If I want to know the resistance of an interface, all I then need is the area.
Given my methods, the higher viscosity products are measured with a thicker film — specifically because that’s the film thickness that would be seen under typical application pressure (like I said, screw-mount). You mention the problems in keeping the low-viscosity products in the interface gap. While you’ve demonstrated that their bulk conductivities are comparable, in reality, that runny product will produce a much thinner film when pressed in place by hand or in any practical fitup. Its typical resistance when applied by hand is probably lower even though the volumetric conductivities are similar.
I don’t know that I ever got any of the RTV products to “bottom out” so to speak. Even the thinnest films that I could measure were in the 10-30 um range. In my mind, knowing the uniformity of the film thickness or metal-contact is only useful in terms of the test itself; in application, it’s only important that film thickness isn’t excessive. There is no minimum allowable thickness unless you’re using it as an electrical shim (you might need to, idk).
And I don’t think I mentioned it in the writeup, but the RTV products (including 922) were tested under pressure and with the pressure removed. Once film thickness is established by the spring pressure and a bit of twisting, the fixturing pressure doesn’t change the resistance to any degree that I could measure. For practical reasons, I let it at least cure under pressure. I didn’t want to bump it and tear a partially-congealed sample.
Finally, despite my talk of screw-mounting and all that, the forces I used aren’t unreasonable for a hand-fit interface. If you figure the area of a 20mm star, a simple hand force of about 6.5 lbs and a wiggle or two should put the film thickness in the same ballpark.