Well that's very strange. I do stuff like that all the time and only use pliers, hammer, & bench vise. I've never had to anneal a piece of copper ever in my life. What am I doing wrong?
Dunno…but it fits. 
I’ll take pictures later, I’mma finish soldering the contact board.
I’m still unsure that a driver with this little actual circuitry is as good as everyone proclaims
[EDIT] Umm…eh-heh-heh-heh…I uh…I may have cooked the driver by accident…
Yeah…put the batteries in, screwed on the tailcap. Clicked the switch. One brief, brilliant flash and…nothing. Just the dull blue glow of residual current across the emitters. 
I…I killed the driver. I killed it! :_(
[EDIT2] I slept on it. and I’m at a crossroads with this project. Option 1 is for me to sell this as a host for $10 and buy a real 3T6 from Fasttech. Wash my hands of this project, be done with it and spend less money overall than option 2.
Option 2 is to re-order the DRY…and a few new emitters (one has a scratched dome). If I cut the corner of copper stars and get Al-sunk XM-L2s, I’d be looking at a theoretical 2800-and-change lumens OTF at 4 amps in direct drive. If I went the extra expense of copper stars and connected everything with 18ga wire, direct drive could theoretically give this thing 3800LM OTF. This presents a problem though. I removed the centering rings for the emitters in effort to boost the output, and in the process of cleaning my room, lost the rings.
http://www.cnqualitygoods.com/goods.php?id=1634 Would those work as substitute if I wanted to change the emitters? the ridge doesn’t looks quite deep enough to grab the reflectors and work to center the emitters while the fujik sets.
I would try and test the emitters, one by one, by direct driving them to a battery to see if they’re any good… I have killed two emitters so far where I got the brief light flash then nothing. Driver and emitter were both dead.
If you’re just looking to center the emitters while the glue sets then any number of things can be made to work, but none will be as easy as the original rings.
Sounds like you shorted or overheated something on the driver.
I know I did. I’m sure the emitters are still good as they glow very faintly. I even tried reinstalling the original driver and still just a faint glow, but it’s kinda strange.
+ E1(dim) ~ E2(dimmer) ~ E3(dim) -
Gah…Yeah, the center emitter is toast. Is there a place I can get 16mm-mounted XML2’s at least as hot as 5000k for less than $6.50 each?
Fasttech has them for about $4.35 if you buy 3 IIRC. It’s about $6.75 for copper mounted XM-L2 at IOS.
I already bought one from IOS for my T620 (and a MT-G2 for a X6 copy I might end-up returning, lol). I forgot about Fasttech, kinda inclined to write them off with the decline in their service, but I’ll keep that in mind. So far I’ve had good luck with them, they shipped a 20mm XML2 and a Convoy M1 host within a couple days of payment, should be here this weekend or early next week.
I've had blown emitters still glow faintly. I have no idea what is happening in those magical things. So I can't tell you why.
Lol, it’s way beyond any pay grade I’ll ever have. :party:
I’m also looking-up thermal compounds for securing the stars to the pill. I found one thread here where someone tested it and Fujik is as thermally conductive as toothpaste. 
On the other hand, Arctic Alumina is $14 for 7 grams. Is the difference minimal, should I spring for copper stars, or should I try to mix normal silver paste and fujik the edges of the star?
The biggest heat transfer bottleneck in high current applications seems to be in between the emitter and the mcpcb (star), next would be the MCPCB (star) to pill junction. The most important thing is to get as much metal-to-metal contact as possible and rely less on thermal compound to fill the void (as little as possible). This is why many polish and lap the two surfaces together: more contact area. Fujik seems to work fine if you have good contact, use a thin layer, and use pressure while drying to squeeze out as much Fujik as possible. No doubt the more expensive compounds are better at transferring heat, but in the MCPCB to pill junction if the surfaces mate together well the gains will most likely be marginal.
+1 on what RMM said. If you go high current, go copper stars for sure.
Like RMM said, thermal compound layer should be as thin as possible. So type doesn't seem that important to me. I use this cheap stuff from HD. I only use it when I have to. When possible, I like to reflow my bases to my pills with silver bearing solder. Sometimes I just use normal solder.
You probably said so already, but did you wire your emitters in series?
They came wired in series. The DRY driver is supposed to be made for series-wired LEDs as well. I was going to follow that pattern if/when I used new emitters. Of course, lapping things is not on the list of things I’m capable of/have access to. This pill could use it though, it looks like the surface of a record, just scaled down really fine.
On the other hand, the first time I fired this thing up it got nice and hot…wait, could it be that the LED was dying to begin with, or that I cooked that particular emitter, and that if I replaced the emitters the original driver would work just fine? Ah, screw it, I still want that turbo mode.
Either way, I’m going to end-up spending ~$25. The only decision I need to make is, do I spend $25 the adventurous way or the safe way. 0:)
[EDIT] STAHP!
I jumpered the dead emitter so I just had the two in series. I tried the old driver. The original driver is still worthless but it put out a measurable amount of light. So then I became hopeful. I reinstalled the DRY, and after forgetting the polarity of the output wires (heh-heh…), there was light! Bright, extra-bright, low, medium, and bright, and then there was a pop and the light wasn’t so bright anymore. One of the remaining two LED’s fried.
Long story short, the driver is still good. 
Off to IOS to buy more copper and silicon!
LOL
Sorry to be savoring your killing of the led, but the way you wrote that is really funny!. Good luck with your upgrade.
Glad to hear your driver is still alive. I ordered some of those during that shipping mess and never got them.
You don't have to spend much to be able to do a halfway decent lapping. Sandpaper on a flat surface for the base. For pills I usually use those round, flat dremel grinding stones like pictured on the far right below. I don't use a dremel or drill. I use them by hand. I keep a variety of sizes and grits around. Also, some with out the dremel shaft too. There is a 20mm size that fits perfectly in most pills, but you should only use that at the end as a final check of level. The reason is if you use that stone, the only movement you get is rotating. So the center doesn't get any grinding. Better to use a smaller stone so you can slide the stone around the pill surface including across the center. Its best to do this under running water to reduce clogging of the stone by the soft aluminum. Oil may work even better, but I haven't tried that yet.
Once flat, I then cut sandpaper disks of increasing grit. The dremel stone holds them for fine sanding the pill. The whole process goes pretty quick.
Lapping could be considered if not for the fact that the emitter side of the pill has a raised lip that supports and spaces the reflector off of the emitters.
Can you guess which two emitters ‘popped’?
That said, I think I know why the one emitter died to begin with.
NOT ONE ATOM OF METAL CONTACT! All of the heat from the stars has to get through a solid layer of Fujik before it gets into the pill! :Sp
Furthermore, for some reason I cannot bring myself to spend this light’s initial cost in emitters. I’m going to go with the aluminum-sunk ones from Fasttech for a couple reasons.
1.) I’ve had good service from them so far.
2.) I calculated 2800 lumens OTF on Al with 4 amps. That’s more than enough for me, for now. If I eventually buy…that Lust_e_fire 26650-powered TK-70 clone, or a BTU, then sure, I’ll get all Tim-Taylor “Moar-Power!” on it.
Almost every pill has a lip. If they didn't, I would just use a sandpaper on a a flat surface to lap my pills.
4 amps and aluminum sounds ok to me too, but I'm no expert on such matters.
I’m going off Match’s test curves for emitters. the XM-L2 is already pretty close to the top if its curve at 4 amps.
Seems strange to me that your original setup exceeded the true limits of that one XM-L even with direct drive and that somewhat poor heatsinking. It may just have been a dodgy emitter to begin with. As for the reason why the other two died, well if you ran 3s/12v directly into them without the 3rd emitter in the chain then that’s no surprise! Poor emitters
Best of luck with the new emitters and just approach things methodically. Ideally outside of the light with the emitters mounted to a beefy heatsink to make sure everything works before you start assembling. It may seem like more hassle but it makes tracking down potential problems much easier later on. Also I would figure out a way of accurately measuring the tailcap/emitter current in the system to ensure you’re in the safe range. Direct driving can be rather unpredictable and subject to many variables in the system, reason why I’m not a huge fan of it myself.
Also keep in mind XM-L2s only have 2 bond wires instead of 3 which seems to mean they can’t handle quite as high a drive current as the original XM-L before going poof. Thankfully as a tradeoff they have a higher vF demand and therefore tend to draw less current overall in a direct drive setup compared to the originals. So just be extra careful, and take measurements to verify things along the way. I wouldn’t want to see the disappointment of more emitters going pop!
PS: And yes contrary to what I said earlier in this thread, it might be wise to start with the thin wires. Just to be on the safe side and see where your drive current is at before deciding to upgrade them to thicker ones.
@ Linus, I’m saving the tiny-thin wires that connected the emitters in series to try first. And there’s still one emitter that works. I don’t have anything to sink the emitters on to test the beforehand. I did make some nice, beefy 14ga leads for my DMM, so I should be able to take accurate tailcap readings. More ideal I suppose would have been to order a 3s battery cradle from fasttech with the emitters. I have a 2s one so I can tune the trimpot on my T620, might have been a good idea here.
Yes battery cradles would be handy, and for the test setup maybe you can track down an old aluminium CPU heastsink. That’s what I had laying around, then I drilled and tapped some m3 holes to bolt the emitter pcbs down for a decent thermal path. Again it probably sounds like a hassle to make but once you have it it’ll see plenty of use, testing emitters and drivers outside of a host can save so much hassle.
Okeedokee. Project finished. Done. Over.
…for now… 0:)
Thanks to RMM for sending me some centering rings when an order from IOS didn’t arrive after three weeks. They were the 7mm variety however*, and as such were 1mm too small to work in the 3T6 reflector. I had an idea that seems to have worked-out quite well. I cut the wings off the centering rings and filed them down such that the OD was 8mm, then glued them atop new ‘wings’ made from the scraps of the old ones. Some sanding down later, they just barely engage the openings of the reflector and center the emitters nicely. As they’re aluminum stars I wanted to limit current, so bridging the emitters is the so-small-it-should-be-funny-but-isn’t wire originally from my Trustfire X9.
Fired her up, and with three Sony US18650GR’s charged to 4.15v, I got 2.00 amps at the tail on high and 3.5A on direct drive. Yes, that’s an amp down from what the DRY is advertised as putting out but you know what…it;s enough.
Aaaand then I promptly dropped it on the kitchen floor. Well, I already wanted to buy some Duracoat for the slide on my Kareen Mk-III, this is another reason to actually buy some.
*He said beforehand that they were and I asked for them anyway. Wallbuys carries an assortment that contains an 8mm centering ring in it.
[EDIT]
I have a question about this arrangement. What happens when the collective voltage of the cell stack drops below 9v? Does one of the emitters die? I imagine that if the voltage drops such that there is not enough to forward-bias all the emitters that at least one will go dark, correct? I’m just wondering as since this light cannot use protected cells and is quite clearly a cell-sucker, it would be nice to have some warning when the cells get low.