I'll better reserve XP-G leds to 2xAA lights as boost should work better with a lower Vf. I guess any 18650 light can unleash the potential of an XM-L. I plan to start with something with more space, like an UltraFire C8 or a bigger TR-1200 body. Ordered my insulators, thermal epoxy, 2.8 linear and 3A buck drivers, empty P60 drop-ins, Dremel tools and so on... :)
I did see somewhere on CPF that an XP-G R5 was brighter than an (unstated bin, probably T5) XM-L until you went past 1500mA. But my memory may be at fault. The XM-L is more efficient at any given current, but not necessarily brighter. It is when the XM-L becomes much less efficient that it gets really bright. The same is true for most LEDs
I don't think that's true about the output. Cree determines the bins at 350mA. So a higher bin will always be more efficient and put out more total lumens at pretty much any current than a lower bin. Now the brightness of the LED die itself (and therefore the throw) may not increase for the XM-L until you get past what the XP-G is rated for which is 1500mA. That's more of a lumens/mm2 thing though.
That is not quite true. XM-L's are binned @ 700mA
I went looking for the post at the Other Place and gave up as they've rearranged it in a way I do not find useful. I gave up on searching as a bunch of patients needed seeing. Wonder how many threads "got lost" this time?
I've no idea why but a lot of websites I look at daily have been redesigned in the last few weeks.
Nearly all of them are now worse - except here. Personally I'd prefer slightly smaller text but given that a pinching gesture on the mouse will fix this, it really isn't a problem
Yeah, you're right. That seems inconsistent on Cree's part. So really the XM-L T6 has about the same output at 1500mA as an XP-G S4 would. And the XM-L T5 is about equal to an XP-G S3.
Very informative thread. Thanks!
This ought to make things very interesting once supplies catch up with the inital demand and economies of scale begin to kick in.
Where do the XP-G S3/S4 start to lose efficiency though? The XP-G R5 levels out really fast around 1 amp in the tests I've seen. Nowhere near linear between 1 and 1.5 amps. Did you make that consideration part of your calculation? And where are those test numbers for the S3/S4?
From KD's XM-L T6 drop in, Don was getting:
436 Lumens @ 3.736V battery / 1523mA tail
An XP-G S4 will do that?
You really can't say this as you really can't compare the bins on these two LEDs comfortably. First, as uncajesse pointed out direct multiplication doesn't take in to account decreasing output and losses to heat. It's likely that the XM-L is geared towards accepting a MUCH higher current than the XP-G due to the larger die.
Secondly, S3 bins of Cree emitters are extremely rare in the wild and S4's are pretty well unheard of right now, while XM-L T5s and T6s are shipping. Heck, 4Sevens is claiming that large numbers of S2 bins are hard to get.
Cree has graphs of output vs. current in the product sheets for each LED. I'm using information at the wiki:
http://flashlight-wiki.com/Cree
which in turn is based on a table that Isotco (is that someone here?) posted at DX, which in turn is based on the Cree literature. So the nominal output of the LED is given for 350mA, which in the graph is 100%. Then you can get values for 700mA and 1000mA by figuring out the factor to multiply by. I got the graph below from the Cree datasheet (in black) and added lines: the blue (350mA), green (750mA700mA), and purple (1000mA) lines plus a straight red line to show how the efficiency is tailing off. Anyway, increase the 350mA value (for R5 Cree says 139-148 lumens) to 185% for 700mA and 250% for 1000mA. At 1500mA it looks to be about 330%. This graph is for 25 degrees C, so it doesn't take into account that you would have a lot of heat at higher currents that you need to get rid of in a hurry to maintain that temperature.
Anyway, as long as the tables at the Wiki have the current values you want, you don't have to mess with the graphs, and the XM-L and XP-G both have values shown for 1500mA, so I compared those. And the lumens at 1500mA match up with bins as I said. The S4 doesn't exist, but I think the 350mA values have been extrapolated from the Cree binning guide which go all the way up to S3.
I think you mean 700 mA instead of 750 mA. The same 'mistake' can be found in the Wiki in the Cree article, 2 lines into the XM-L section. In the sceme/diagram the 700 mA is correctly used. 
Well, at least I'm consistent . . . I fixed that. Thanks for signing up. I went ahead and confirmed you so you can fix any more mistakes you find on your own!
Thank you for your confidence, I have thoughts of contributing with an article, perhaps concerning tint. I have a little knowledge as I worked professionally on a 'true color' camera and also possesses a color analyzer that I use for tint measurements.
By the way, I am offering my assistance in measuring tint of emitters for other members if they pay the postage fee. When I buy emitters I wil report the tint in the 'Tint Measurements' thread.
When the manufacturer orders emitters from Cree, the narrowest range of eg. cool white XP-G R5 that can be ordered covers 5300K thru 7000K (called kit no.00H50).
If the manufacturers wishes a narrower range they have to sort the emitters. I don't think they do that for budget lights. Therefore tint is normally a draw of luck.
Where did you get that A20 with no strobes?
Anyone found XMLs in 14mm star in DX, KD or so?
Mine has strobe ... Good thing about the original driver is it starts on high ...
Hi-med-lo-strobe-sos .. turn the light off from lo and then restart in high , if you wish ...
Or you can try a after market driver ...
Oh.. don't you know if a 17mm driver would fit? It would be interesting to make a really pocket rocket with the NANJG 2800mA .. http://budgetlightforum.cz.cc/node/1101
It would also need IMR 16340 or something that can hold that amperage..
It uses a 17mm driver ...
Im currently using a Solarforce V2 RCR123A ...[ Blue and White = new one ]
Previously I have used the Trustfire Flame RCR123A
You can get them from Cutter Electronics. They have some shipping though.
KD has bare PCB's to mount an XM-L on. You'll need to reflow solder the emitter to the board. Some youtube videos show you can do this even with an electric grill :)
From cutter: 9usd (the led) + 11usd of shipping.. :/
I think I will get one from KD in 16mm star and try to cut 1mm of the edges to fit the A20 dropin.. what do you think about doing this?
Shipping isn't much of a problem if you buy in quantities. You can even buy 13 x 14mm T6 Led in several tint options for $107 inc. shipping, which makes $8.2 each. In case of building a 5x XML setup I believe I could buy in such quantities. But it isn't that good an idea for just a couple of leds.
16mm ones might not file down that well. Maybe just 15mm or so. The pads are too close to the borders and I don't know if they would short if they shared the same border with the alu base. (EDIT: doesn't seem to short so I guess you can file it down) But still there can be some way you can make a smaller star. You can cut the PCB roads, remove the isolator on them and solder directly to these. Never tried though. But I believe I've seen one application, filing down to even a smaller diameter (~8-10mm) and putting the led on a pole to get a better focus.
EDIT: Ah, I found the link to this application.