There is no good single cell driver for 3x xml setup, only amc sandwich can do something.
How about something like this?
www.dealextreme.com/p/fandyfire-hd2010-xm-l-t6-5-mode-1800lm-3-led-white-flashlight-w-strap-black-1x18650-1x26650-120677
1x18650 (or even 1x26650) can only give so much.
In other words, it will either give low output, or give very short runtime, or both.
A single Li-Ion simply cannot provide enough power to properly light up 3x XM-L.
It would be better to properly drive a single XM-L than badly drive 3.
Tecmo, I have this flashlight, in spite of my hard work I can't make it very bright, it is a flooder, but not very bright, especially with originall driver sucking only 2.4A on the tailcap. And it is a drop in module with a loooot of air between dropin and body :(
Before making statements like that have a little read of Matchs tests and see how, for example, 3 XMLs at 1amp would compare outputwise to 1 XML at 3amp ;)
yeah really. like efficiency goes up over 1A or something 
Sorry to hear that. A shame since the battery could handle 2.5A to each emitter pretty easily.
Yeah, I drive it with amc sandwich 5.6A now, but losses on switch and dropin are so great, that it goes barely 4.6A into leds, I have to make new "screw in" dropin as there is place for it, but it is a cost, you have to rebuild whole flash almost. Leds are not very high bin in my opinion too. I was really interested in this light, but it is just hard to make bright... Bright in my scale of course, on other ,much smaller 1x 26650 light (Shadow JM07pro) I have 3x xml module on ledil cute driven by 5.6A amc sandwitch and there is no comparision... Not to mention well driven 25-30W 2 cell, 3 xml flashlights.
Ok, strictly speaking, 3x1A gives 1200 lumens vs 1x3A giving ~900 lumens. You can run 3x XM-L in parallel, but you won't have much too runtime from a single Li-Ion. Not even KK 26650 can run that for long. Running it at 3x1.5A (parallel) gives even more light, and shortens the runtume even more.
Well, to each his own, I always say. I respect that running this on a single Li-Ion can be interesting. The gain IS there. ;)
I know I'd never be satisfied with such a big flashlight that only runs for 30 minutes (or less) on high, no matter how bright it is.
Viktor
I'm still curious of the possibilities. The reason for my curiosity began with my dad's uses for his flashlights. He primarily wants a big boy to shine from the house (2nd floor) out 80 yards or so to the garden and surrounding area to check on critters. He only uses it in short bursts. This rules out HID but it also means he's going to be charging it every few days either. For him, output definitely outweighs battery life. He would rather have more batteries on standby than have the ones in the light itself last longer at the expense of output. It will always be in the house and almost always sitting in the exact same spot.
For this type of use, capacity is not a high priority. I'm thinking mostly of battery safety and and simplicity. A guy like him just doesn't have the desire to monitor multiple cells. He could work on his antiques from dawn til dusk with a smile on his face but flashlight maintenance is only an annoyance to him. I can appeal to this. Sometimes we should think of the industry and market as a whole and not just our niche flashaholic community. Well, that's how I think anyway :)
ok, 3 xml at same current and same battery has higher output than 1 xml
likewise, at same total output, 3 xml has longer run time than 1 xml off the same battery.
there are reasons to go w/ 1 xml - beam profile (ie throw) is the first that comes to mind, and the second is what we've discussed here - an almost total lack of budget class circuits suited to 3 xml (at 3.6V nom.).
Are there any non-budget? As far as I know there is nothing well made to drive leds on 1 cell, even if you want to pay for this.