Can’t wait to see this.
Prelim results on the XinTD XM-L2 U2 on SinkPAD is so-so. can't seem to get the full amps of 3.85A, best I can get is 3.57A on a fresh Pana 3400. On a fresh Pana 2900, get 3.47A. Weird. With the Pana3400, lumens readings are 1142 @start to 1101 @30 secs. Definitely dissapointed there, but it may be the seating of the SinkPAD. Was getting a lot of upward pressure from the teflon silver tinned 22 gauge wires on the SinkPAD because of the tight squeeze - pushing limits all around. I know the independent driver test measured exactly 3.85A as it should when done on a mounted T6 for test purposes.
So, thinking I don't have a good bond for the star, so I can enlarge the pass-thru holes/edges of the star to eliminate some or all pressure, and maybe add epoxy in the center to secure it down well. Still can't explain the low amp measurements though... However, I've seen some strange lower amp readings on these XM-L2's that I couldn't explain also... Maybe there is an issue there with over-driving them.
If the emitter is resisting current somehow, the Vf has to go up or VIn has to drop (Ohm’s law prevails). I doubt the LED is jumping it’s Vf suddenly based on current or heat, since heat usually makes Vf drop slightly.
I’d look at the switch spring and if there is a driver spring. Those steel springs add a lot of resistance to the path. I’m not familiar with this particular light, so I’m just throwing out the areas I usually look at first.
Dang, Tom, that’s an awesome upgrade, mate 
Good job! 
Looking at the datasheets, it appears the XM-L2 may operate at a slightly higher forward voltage than the original XM-L. Anecdotes from P60 builders on CPF apparently supports this—their direct drive XM-L2 dropins apparently draw lower current than a DD XM-L. That withstanding, I find it hard to believe that its Vf is so high that 12*AMC7135 cannot regulate current from a freshly charged 18650.
I don’t find it at all difficult to believe. Extrapolating from the XM-L2 Datasheet, it looks like a typical emitter will be at about 3.45 or 3.5V Vf in order to drive 3.9 Amps through it. The AMC7135 Chips drop about 120mV, so we’re up to 3.62V that the battery has to overcome.
Per HKJ’s testing of the Panasonic 3400’s - within 1 minute of being discharged at 4A, battery voltage would be down to the 3.8 to 3.9V Range, and dropping quickly. This only leaves us with somewhere between .18 and .28V of headroom for voltage drop in the rest of the circuit (Spring, wires, tailcap, body, etc). At 4A of draw, that means the entire loop can’t have more than 70 MilliOhms of resistance. That’s not much! And once Battery Voltage falls below 3.7, it’s game over - The AMC7135’s will no longer be regulating and Emitter Current will start to fall.
And these are all TYPICAL numbers - if you happen to have an emitter with a slightly higher Vf or some AMC7135s with a slightly higher voltage drop, it gets even worse, and fast.
PPtk
Bam. Thanks PPTK; well said.
In this case, it’s no so much that the XM-L2 has a high forward voltage compared to current, it’s that the original XM-L had an unusually low forward voltage compared to current. The process used to make the original yields lower forward voltages as die size increases. It would seem that the new SiC process doesn’t necessarily yield that same trend.
PPtk
Thanks Guys, PilotPTK, for the input. Crap - maybe you're right by the specs. I can easily re-test the pill back to the U3 original, see if I get 3.85A once again. If so, I'll just stay with 3.5A (10*7135). Now I gotta re-check the amps on the 7G9 and HD 2010 - think I was ok there, but like your saying, I may be on the hairy edge with the 7135's and emitters. Oh also, the 7G9 has 3 18650's and the HD2010 is using a KK 4000. Could back off both to a single 18650 and test. More fun later this eve, on to work...
. They don't seem to want to pay my consulting rates when I'm play'n with flashlights!
Ok - a little more info. Even on the HD2010 with a ICR KK from KumaBear, battery is measured at 4.07v (no load), amps reading is now 3.60A (dropped), lumens is now 1309@start. So, you guys are onto something here. Also I swapped batts in the 7G9 from Pana 2900's to EagleTac 3100's (protected) and tailcap measure is 3.45A (note: EagleTac's are known under-performers). Summary of HD2010 battery tests:
KK ICR: 4.07v, 3.60A, 1309 lumens @start
KK INR: 4.09v, 3.60A, 1196 lumens @start
TF 5000: 4.09v, 3.2A, 1088 lumens @start
KK ICR: 4.16v, 3.78A, 1326 lumens @start (same battery, charged a bit more)
So this means it's super sensitive to the battery, obviously. the 7135 amp regulation is breaking down, pretty easily at these high amps with the XM-L2. I think this is what PilotPTK is saying, though I'm not sure I understand it all yet.
So, will a better battery do the trick and or just delay the inevitable a little bit? Didn't look up HJK's tests yet, but I think the AW IMR's are the best. The best battery I got is the Panasonic 3400 unpro and the KeepPower 3400 (pana cell) from KumaBear.
Or do we give up on this venture of driving XM-L2's in the above 3.5A range? Is the hottest setup a XM-L U3 at 3.85A/4.2A or an XM-L2 U2 at 3.15A or 3.5A? All these options are of course over-driven, so assuming SinkPAD and good heat paths, larger pills, larger hosts, etc.
Interesting because ThruNite I believe advertises an XM-L U2, I thought at 3.5A, but I've been hearing it's really driven higher, like at 3.8 - 4.0A, so, did they see the same issue with the XM-L2? What's their next generation going to be?
Also, would multi-battery, parallel configurations solve this, or at least last much longer to keep the amps up? Guess I could do further tests with 3 fully charged Panasonic in the 7G9.
On the XinTD, took the pill out, and got a KK ICR up to full charge on the i4 at 4.18v, then was able to see the full 3.86A on the XinTD pill. So, the XinTD can do it's full amp capacity with a fully charged, high qual 26650 - useless of course...
No idea the KK INR's suck so bad compared to the ICR's - KumaBear is right, and think HJK's tests confirm that - got to look those tests up again.
The point PilotPTK makes is also mentioned in other forums, I stumbled over this intel about the higher forward voltage of the XM-L2 a few days ago at german TLF.
As far as I understand, it’s possible to drive an XM-L2 harder than 3.5A.
[theory] - Let’s say you drive an XM-L at 4.5A, the result with an XM-L2 would be at 4A then, because of the higher vf.
I ask myself if it is possible to lower the resistance of the whole circuit and so gain better results ?
Changing the wires for thicker ones (driver -> emitter) ?
Solder a piece of wire to the tailcap spring (seen that somewhere on BLF I think) ?
Well, I'm doing tailcap measurements with the tailcap off. Also I'm using 22 gauge silver tinned teflon coated wires - pretty thick, high quality, and at the shortest length possible - no coiling the wire in the pill. I'm also looking in a practical sense - how to do the best we can with a 7135 regulated driver. I'm really not sure what's going on when you have 3.85A regulated with 7135's but are only getting 3.5A or so with a fresh high quality battery - is it amp regulated like that or not? Is it better to make the regulation at 3.5A (10 chips) then?
I understand power = volts * amps, so is the problem the driver is asking the battery for more power than it can deliver? I don't think HJK's tests vary voltage from what I've seen. Maybe I hould check voltage the same way I check amps at the tailcap and see what's happening.
I’m not sure I understand the Vf issue yet - but maybe, if maximum (sustained) output is the goal, xm-l will prove better for single cell lights, and xml2 for multicell, particularly serial cells?
how about some runtime tests so we know what happens when those batteries are nearly discharged…even if you need to use a fan, or the refigerator to cool those pocket rockets 
dthrckt - yep, good suggestion! I'm thinking even in a parallel battery setup, you got a lot more amps capability - I should have kept Panasonics in it and did longer timed tests - the 7G9 will stay pretty cool - actually I did run it for 6-10 minutes I thought, gets warm, but that's all - it's a tank. Hoping PilotPTK can offer more help too - he sure understands this stuff, and this seems to be new territory for everyone, or no one else is talking about it much yet. I'm getting shell shocked with the ups and downs of this project so far...
it is very valuable information. definitely appreciated.
BLF is definitely on the bleeding edge
A few questions here, I’ll try to catch them all…
It’s not being regulated by the 7135s at that moment, but the 7135’s are still doing what you’re asking them to - They’re limiting current to a maximum of 3.85A. They can’t ‘force’ current through an emitter - only voltage can do that, but they will regulate to 3.85A if enough voltage is available to push that much or more current through the eimtter(s).
Not sure. Is it? This is a personal choice. Where do you want your maximum set point to be? That’s all any linear regulator can do - LIMIT maximum current when excess battery voltage is available.
The driver isn’t (and doesn’t) ‘ask’ the battery for anything. The battery gives what it has, all the time. That’s why the regulator is there - to LIMIT how much power the battery stuffs through the emitter. The reason current isn’t matching your expectation is that the battery isn’t pushing hard enough - it has nothing to do with the regulators. When current is less than expected, the regulators are doing NOTHING, in fact. They just sit idly by waiting for a chance to do something. The short answer is that you need more battery voltage when under 4A of load in order to drive an XM-L2 to 4A. I’m sure your cells are more than capable of delivering a constant 4A of current, so that’s not the issue - it’s purely a voltage problem.
PPtk
Ouch! Thanks PilotPTK, I think... I did measure voltage over the battery when hooked up to a XML pill and a XM-L2 pill. What I read for the XML: 3.55v, and for the XM-L2: 3.7v. Both had the same driver setup of 7135's at 3.85A. So there's a difference, but small. Do you think the emitter wants over 3.7v but the battery max's at 3.7v?
Unrelated -- just 10 mins ago installed a boost 7x LED driver from I-O into my J18, mainly because the stock driver was quite flaky - amps + brightness drops, etc., also interested in getting more amps. Think I created a beast - getting 5.8A amps at the tail with 3 TF 26650 5000's, lumens of 4930 @start, 4726 @30 secs!! The batteries aren't even full charged - I think I'm afraid to run it more than 30 secs because it's warm already... 
ive got the similar driver in my j12… theyre great right? i bet you would see even more improvement if you swapped the tiny wires for teflon wires and used IMR 26650
im scared to think of what a smiliar driver would do in one of the new 12xml shower head like **fires
I wouldn’t take it like that, he was just being thorough.
I view voltage/current/resistance rather simplistically (it’s the only way for me!) as a river - voltage = height the river drops; current = volume of water flowing; resistance = width of river. So to get a certain amount of water flowing, you need a certain height drop (as long as the width of the river is wide enough. In your case, you don’t have enough push so you won’t get the current you want.
I think I’ll stop with my analogy, as I’m starting to confuse myself (what if there’s a waterfall? How about rivers with large catch basins? etc :laughing: