Any buck driver rated for the voltage you will be supplying should work fine. The closer the supply voltage to the vf the better. The 6v mgt2 vf climbs as the current climbs. So more than 7v if going over 3 amps. I have used that cheap 3T6 from Manafont with great success. Drives around 4.5amps stock. That was with 2S 18650’s.
Right now, I’m using just straight 7135’s with manual switches to control current and have 6S nimh cells. So 7.2v total. The cells are 4/3AF’s. Works great.
Don't forget the HBFlex... the only 3-amp-capable boost driver I know of - but the low end of the input voltage range is too high for a single cell (too high for two cells, really, though you might be able to get by), which is a shame.
Does anybody know if the TaskLED drivers, specifically the H6Flex, can survive with LEDs in parallel? I know that's the 'wrong' way to do it, but it usually works fine in the non-theoretical real world.
the driver won’t care about the LEDs wired in parallel, although the risk of an imbalance in Vf and current draw between the paralleled LEDs increases with current I believe. That’s just theoretical though, I’ve no practical experience of it.
So, how do you calculate total/individual amperage with LEDs in parallel run off a common output? For example if I used the H6Flex in the triple MTG2 build with 4S cells and LEDs in parallel, vs. the original plan which was HBFlex boost driver with 4S cells & series LEDs.
(numbers with the HB boost: cell voltage 12-16.8v in, total series Vf of ~18.9v; w/H6 buck: 12-16.8v in, parallel LED Vf of 6.3v)
I have no measurements at all as I have only just got my pill together with driver and led but can confirm that this driver will run the MTG-2 hooked up to a couple off trustfire 26650 at 3.8 volts with no load on them. The led is direct to copper and it is the quickest I have ever had a pill get hot. The mode changes still has that pause between them as reported elsewhere.
Cells in series or parallel? I have one of those drivers on the way, too.
Series.
Crap! :( Was hoping for some actual single cell boost action. Was it you or somebody else who reported that driver didn't like running off a single cell?
I just tested with 2 and 3 cell and now you want 1 cell tested? Ok, you sorta asked nicely. Be back soon. It is cold out there. 
This is some really quick testing on the above driver with new Trustfire Flame protected 26650. Good news first.
3 cells 12.48v No load voltage 2.55 amps at the batterys current draw
2 cells 8.34v No load voltage 3.2amps at the batterys current draw
1 cell. No load voltage 4.15. On high battery voltage dropped straight away to 3.74v. Current reading I managed to get was 4.2 amps. Driver immediately dropped to a current draw off .4 amps with a voltage at the battery off 4.11v.
I would suggest this driver for some reason with one cell hits the low voltage cut out on the driver and drops to a lower mode. I dont believe l have enough time to get the correct current reading as this happens very quickly.
This also happened on an XM-L but just not this quick. The experts can chime in off which I am far from but this driver does not run well if used on high with a single led full stop.
If someone can work it out from this information it would be nice to know what current this may be getting at the led.
On an XM-L2 this driver on higher at the tailcap was running 1.39 amps with 4 TF 26650 which should be about 5.56 less efficiencies at the led.
Oh well. I suppose it would've been a minor miracle if it worked. I wonder if it works in boost mode at all (not asking you to run more tests, just thinking out loud!! lol) and under what circumstances. 8.4v in, total Vf of ~9v, 12v?
I'll have one to torture myself someday, that is if SingPost ever decides to look around for my package that must've got dropped behind a shelf somewhere and actually put the damn thing on the airplane... grrrr
I’d like to build an MT-G2 light that runs off two 18350s. Need advice, and a driver.
Which cells, & how much drive current?
Direct drive from 2x 16340s I got 3.6A, but it falls to 2.5A after only about 30 seconds (and keeps falling, and falling...). The little cells really don't like it...
I guess if you could keep the current down to like 1.2-1.4A it would give you more runtime before the driver throws in the towel, but what's the fun in that? A hard driven XML on a single cell would be brighter and run longer. If we ever get ~2000mAh 18350s this'll get a lot easier. :(
That’s what I wanted to know…if it was worth the bother, or a waste of money and time.
Do you have any IMR 16340 or 18350s, comfy? I wonder if two of them could hold their voltage long enough to be of some use with a buck driver. Also, how is your test setup resistance-wise?
for LEDs in parallel, the current per LED is simply the current provided by the driver divided by the number of LEDs. So if you had 2 XM-L in parallel and wanted each to get 3A, you’d have to set the driver to 6A. As long as you fulfill the input voltage requirements of the driver (eg. h6flex needs no. cells in series = no. of 3V LEDs in series + 1, HBflex I’m not as familiar with), that will always hold true. What you may find though is that if the battery you’re using can’t provide that much current, its voltage will sag so much that you’ll get a much shorter run time than you expect. It’s a tricky one - usually it would be better to run the LEDs in series with a boost driver, but then you run into the same voltage sag problem as the current draw from the battery is higher than that seen by the LEDs. You’d have to run the numbers to see which driver type works best for your battery set up.
the above however is for taskled drivers - the ones from China seem to be more variable in their output with respect to input voltage.
Test 'rig' has just been rebuilt, it's a 4P 18650 box, with triple 22AWG wires attached to each end, neg side includes a typical forward clickie. Each cable is around 10" long. I've pulled 15 amps out of it testing other stuff so far...
I've built a p60 pill, MTG2, brainless zombie single mode 105C with only 4 380mA chips, just to see if the 16340s are up to it. The pill is assembled and functional, I just have to modify a reflector & make some kind of insulator/spacer, after that it'll be up and running in a L2P.
Does 2.21A at the tailcap at switch-on, falls to 2A at around 30 seconds (likely the fault of these particular cells, will get some more soon to see if they deal with it better). The 4 7135s are all on the backside of the board, MCU removed, jumper between Vcc & PB1 pads.
None of the reflectors I have are happy with it, any combination of emitter hole size and spacer thickness leaves a yellow area in the middle of the hotspot, which from other MTG2 experiments means the positioning isn't right yet... just can't get there with the parts I have currently. I need some SST50 and/or MC-E p60 reflectors, if such things exist....