Just tried an LED of a fresh charged cell and it went to over 4 amps!
So would need some careful planning.
Anyway, this should be of more interest.
I have finally got round to dismantling the ?20 "10,000" lumens torch off eBay, and also some testing.
To be honest, its junk!
The emitters look like XMLs, but the old original which are ten a penny now. The switch is the common low load micro switch which switches the electronics on a soft control through a chip with no serial number so cant identify it. There are two IC chips, what looks like a MOSFET and a scattering of resistors, on full power its virtually direct drive, 4 parallel cells to 6 parallel emitters. I got 7.35 amps of fresh charge, down to 5 amps at 50% discharged state.
The switch also has no kind of seal at all, neither does the tail cap. A rubber seal inside the tail cap would fix that, and the switch could be uprated to a more robust waterproof one, these are available for a few quid.
So, now to the Cree data sheet
Fresh cells: 7.35/6 = 1.22 amps per LED, Cree data sheet tells us 462 lumens, multiply by 6 = 2688 lumens
Half discharged cells: = 0.83 amps per cell, Cree data sheet gives us 308 lumens, multipy by 6 = 1848 lumens
Of course, this leaves the question hanging in the of why my X3000 with 2 X XM-Ls U2s beats it. Possibly the emitters in the torch aren't even T6s, or arn't even XMLs, my lamp has bigger reflectors, what ever. Thinking about it, I probably understate my output.
Now what I have to say is that the Chinese torch is very highly moddable, loads of room for electronics and it would be easy fit a better switch and series the battery cells to 16V. The issue is numbers, number of cells to emitters, which is probably why the Chinese parallel them. Best design would be a boost driver that boosts the voltage to match the number of LEDs in series.
I will have a play over Christmas. Might come up with something for Stu
All of this Chinese stuff is aluminium die-cast. Tooling is expensive but churning the stuff out by the thousand and flooding the world market means that eventually it costs next to nothing to produce.
By way, the testing was on Sanyo cells, I shudder to think of the results with 'Junkfire' brands.