The Vf will automatically double, as long as protection was not added for that. The current will not automatically halve; you will need to change the current sense resistor. However, if it is at 3A already, this is a good start. I drive my MT-G2s at 5A, but I have concerns about this driver delivering that much power.
I like to dry run on a power supply because I can limit the current is something goes wrong. Protected cells would be an alternative.
Anything useful in testing with two XMLs in series on an external 'sink, or just throw it in there?
MT-G2 will have the original MCPCB plugged with copper through the center pad, that should help. 3.3A will be fine to start with.
I need to get with DrJones and/or PPTK about the PIC, I wouldn't know where to start with writing code to save my life...
I tested with Two XM-Ls in series, just to be sure. I also had a current limited power supply. My driver was fin with it, but this was a step down in power. I was driving an SST-90 at 10A (40W), now an MT-G2 at 5A (30W). Since you are going up in power, it’s not as simple. Watch the driver for heat.
Some of the drivers that I’ve connected to an MTG2 end up putting out the same power - ie watts, as they do with an xml. Others draw the same current at the tail as they do w/ an xml…so the power is nearly double. And still others, seem to dry to do the latter, but maybe can’t quite hack it…so they are nearly as much current…
But I’m no electronics expert, and have no idea what your driver will do, I can only relay my own test data in case you decide to go with a different driver.
I’m looking forward to your results, and hopefully beamshots. I share your thoughts on the stock beam - If I were to buy a TN31, it would be to put an MTG2 in it…
Tests with 1, 2, & 3 XM-L bolted to a fat heatsink. Stock S1100 driver. Longer wiring than what's used with the light assembled, so I measured using just one XM-L so that it's using the same length wiring, heat sinking, etc.
With ONE XM-L, stock driver does 3.23 volts, 3.31 amps.
With TWO in series, it does 6.36 volts, 3.25 amps.
THREE in series, 9.56 volts, 3.70 amps.
Interesting. I have no idea how it knows what's on the other end of the wires, but whatever...
should work nicely with mtg2!
+1. Comfy, maybe you want the 9V MT-G2…given that you’re modding the PCB with copper plugs through the thermal pad it should be OK
J)
The 9V MT-G2 running at ~3.7A should draw roughly the same wattage as the 6V version at 5.2A.
The sensible part of me counters by pointing out that the 6V MT-G2 may be a safer choice and will still kick major ask.
LOL, yes, my sensible side is in charge of the S1100 build. Now the other half trying to sort out how to do three MT-G2s in a Skyray King, that dude just ain't right.
probably wouldn’t get much run time - if the buck driver needs 1.5V head room then the light would cut out when the cells were at 3.5V
Oh, knew I forgot something. Is there any significance to how the amperage (I know total power increased, let's pretend to ignore that for a sec) went down from one to two, but up from two to three? The LEDs are the originals from the gutted SRK still on their stock AL MCPCBs, I wonder if it would test different if I used a different pair of the LEDs for the 2x test?
Can’t wait to see where that build goes, though you need need one of these
There's enough room to switch the SRK from 4P to 4S with a drop-in battery carrier fairly easily, but that would still give a supply voltage less than the total Vf of three 6v LEDs in series. Or keeping the 4P layout and using master/slave drivers, but I don't know of any boost drivers that can do that. Something like the H6Flex still requires higher supply V than output, correct? Or are there any drivers that won't barf on 3 MTs in parallel? That would make it a lot easier.
LOL
I don't have pics of the innards but the bare emitter is soldered to a .300" wide x .060" thick x 2" long copper strip with wires soldered direct to the underside, and sitting on a big copper plate that fills up the original recess in the head. Reused the original insulator/centering piece with the opening filed out to a square to fit around the LED base. Works nice but awfully fiddly to put together without one piece or another slipping out of place, so I'm not opening it up again until I have to. Just a sheer stupid amount of light, still throws farther than anything around here I can aim it at.
I just discovered this thread. Like the work you’re doing on this light. It sounds like a beast. I’m not familiar with the model. Do you have a pic of your whole light?
EDIT: I googled it. That is one bad ass looking light. Best of luck going MTG2. Sounds like it could be awesomer.
I’m late to the party, but Id guess the 3S emitter was virtually direct drive. I don’t think the driver was bucking at all. That may be why you got more current than the regulated amount.
That driver is perfect for MT-G2. Are you going to resistor mod for more current? J)
No, not cutting up the driver for now. I just checked current at the tailcap, went up to 2.21A with the MT-G2, versus 1.22A with the XM-L. I didn't check it at the emitter.
The copper bar the LED's on doesn't stay in place because of the stiff teflon wires, I have to take the bezel off and reach down thru the reflector to hold the pieces in place, then hope nothing moves before I can get the bezel screwed back on. If I solder the bar to the copper spacer/shim/baseplate thing there's no way to get to the solder pads under the LED (plus, it's already a nightmare trying to solder to something attached to a chunk of copper). Mounting it with screws would be great but me and cutting threads in copper don't exactly get along.
Why would it go direct at Vf of only ~9.5v? Input was 3s/12.6v (as the cells go flat though I'd agree it probably would drop out). Given that input current roughly doubled from XM-L to MT-G2 with the roughly double Vf, you can extrapolate from there what it would have been at 9.5v output, and it's still easily within what a single one of those cells will put out in single LED/DD config.
This driver may need more headroom than one designed to run with Vout close to Vin. it may not be DD, but I’m guessing the coil is saturating due to the high duty cycle. The switcher is switching, but the coil is saturated, so it’s running in DC (Direct Current) mode. It has to do with the magnetic properties of the coil itself. Once the field saturates, it may not collapse enough during the off cycle, so current stays flowing all the time.
A smarter switcher IC may deal with this better, but this driver was not designed to run like this.
OK, now see what happened? I went and got all technical again. Oops.
Anyway, it doesn’t matter for you since at 6V out it’s regulating fine. Now you can do a resistor mod and get the current where you want it. 
I had it in the car earlier tonight, and held it out the window while driving. It's brighter than both headlights at the same distance. (and also a nicer tint)
and uses less power 
Here's a P60, smooth reflector, de-domed XM-L2 T6 (3C 5000K) at 3.8A, on copper. Ignore the color, that's down to two things: one, the ceiling is far from what you could call white, and second, at these exposure times (1/250(!!!) second, ISO200) my camera isn't all that happy with getting the colors accurate. At longer exposures the color looks more like what it does in reality but then the S1100 (should I call it S2300 now?) pics are just a big blank rectangle of oversaturation.
XM-L2 shown, mouse-over for the MT-G2 @ 3.whatever amps.
lol