Battery Chamber for 3 x 18650 in series.
http://www.manafont.com/product_info.php/flashlight-component-battery-chamber-for-18650-p-6404
Fit with mine, space around 1mm between this Battery Chamber and the tube.
Difference Length around 1mm.
For install new driver. You must have thinner driver pcb board.
Maybe around 3-5mm.
not to poop on everyone’s cornflakes but an xp-G on alu won’t like over three amps too much, your really going to want to try and get a copper mounted xp-G2 in there.
is there not space to use the driver as a contact board and use a dr jones electronic switch nanjg?
I’m thinking, simplest driver upgrade would be to use the same MCU negative output to power a MOSFET instead, which connects LED- to ground.
So only need to add one MOSFET. To do the above requires NPN transistor, right?
@Pok, bookmarked, thanks.
Its stated as an xp-G but the image shows a xm-l in there.
However if I get one with a xp-G it will be swapped out for a xm-L2 u2 (on Nocigon) or even a u3 if it becomes readily available by time I get it, 3~5 weeks to Australia normally, mix in the Lithium battery postage fiasco with the christmass rush, who knows when I will get it :~
Cheers David
I’m not sure how to distinguish between xml and xpg, but the packaging says “1 CREE XPG /1 CREE XM-L T6”. So it could be either even though WB says it is xpg.
oooops, having another look it IS a xp-G in the product pic on WB.
Cheers David, who is now slithering into a deep hole :Sp
Thanks for doing this teardown much appreciated!
Biggest disappointment is that emitter heatsink plate, why oh why do manfacturers get so mill happy when it comes to heatsinking surfaces? It’s perforated to all heck and particularly baffling is the large hole directly below the center of the led mcpcb….why of all places…WHY!?
They must get very good prices for recycling the aluminium swarf? 
Out of curiosity what’s the diameter of the led mcpcb? Looks bigger than a 20mm
Cheers
One thing that just occurred to me…How much does the driver spring actually compress? Because the driver is held in with those two screws, what’s keeping one from getting little aluminum or plastic spacer and longer screws to put it a few more mm’s off the bottom of the pill? That would give one room for whatever driver you want, even that silly thing Flowmotion linked. And that driver needs 8-18v? CNQualityGoods sell 3S 18650 carriers (in the ‘torch host’ section), and they’re as short as possible with just PCB’s at each end.
I noticed that too, but forgot to measure. I’ll check it out tonight.
It compresses a decent amount because the tail spring is rather solid. I think shorter springs could net 5-10mm more room though (Perhaps just a contact board and no driver spring like a T08 iirc)
OK light for the money, looking at it as a very long run light (2 cells, low draw) or as a host (with work needed to push it hard).
Exterior and reflector were pretty much flawless on my sample. A few observations to add to what I’ve read here so far:
- Alum bezel
- O-ring at the front seems to have pulled a little of the reflector finish off its front lip but it’s not visible when re-assembled
- Reflector has a centering piece for XPG/XPE
- LED PCB ~25mm (I didn’t remove, but a 22mm washer is smaller)
- Pill looks to be a screw in mounting plate under the PCB, affixed with a spot of green thread locker, mine has a couple thread on the PCB side in front of the pill
- Only a few mm between + contact plate and pill
- O-ring on body at head touches head but no shelf behind it
- Flat tops did work for me, but carrier design makes it hard to remove any cells (I used a pick); 3 cell carrier could be fit
- Body tube ID 43mm
- Short spring under tail cap button (seems)
- All threads dry
- Draw consistent with OP report, even using Sony IMRs
Question - wonder if and how much parasitic draw exists
Unfortunately, the switch on mine has no effect and was stuck on SOS when I inserted cells. I can change modes or turn it off [EDIT: destroy the wiring] by shorting the charge pin, but SOS overrides all on modes.
EDIT: I should have known better but shorting the charge pin burnt up some of the wiring. The switch shows a .5 on the continuity test and goes to 0 when the button is depressed. Not sure if the switch should be a 1 when unpressed; maybe there is more wiring to be replaced up in the handle? I’d never charge cells in the light, so all the charge plug wiring will go anyway.
Outside all I had was SOS to test, but pretty decent throw for being so underpowered. Typical decent but muted XPG spill.
looks like a 32mm pcb to me.
that driver flomotion linked is 4.2v I’m pretty sure.
your definitely going to want something on copper for that.
Nice! The pill was the worst point for me. (I was not even thinking of using the stock driver)
as soon as my light arrives, I’ll visit my machinist friend with a plate of copper and two beers in my hands
Any ideas on how to get this thing to a reasonable 3amps? could I throw a 105c in there and call it a day?
I don’t really need the charging port, or the long press SOS. High and Low/medium or just High would be fine for me. 3x 18650 cells in parallel.
Well I guess from the long press UI stuff the switch is electronic rather than a clicky?
If that’s the case you’d have to look at a DrJones driver for good electronic switch UI, but otherwise if it fits in there somewhere with some kind of a heatsink path I don’t see why it shoudn’t work fine.
Drjones driver is my option. If there’s no room under the pill, I’ll try to install it in the space between the reflector and the head
I see posts about his drivers, and a bunch of threads with them… where do I go to see prices and possibly buy one?
If it is really 25mm, that’s the same size as the star in the Small Sun T620, so for anyone who has one of those and upgraded to copper that’s a potential drop-in XM-L, albeit a T6-bin.
I did a little bit of maths and that’s just slightly too small for a traditional 4P SRK battery arrangement.
The easiest method in theory, is just to short the empty bridge right next to the resistor. This will practically drive the emitter as much as the controller chip can allow, which based on experience, is about 3A or tad below that.
But remember running the chip at max might just burn it out quickly. You might get lucky, or maybe not 
. That’s why IMO the better method is to add a FET on the ground side (similar to 7135) and use the controller output as gate signal.
My lifetime’s worth of luck ran out years ago.
What is an FET?