Ok thanks Tom!
I updated the OP at the top. Trying to help people find things - dunno if it helps much. This thread has gone all over the place, not such a bad thing though... 
I will probably run it on 2s mostly.
I need to get off my butt and do some things outside, its so beautiful down south today. Mid 70s for the high temp today instead of 90s just a few days ago. Great day to be fishing!
I’m thinking about the Y3 and do the MTG2 swap.
Can anyone confirm if the noctigon is a direct fit, or requires a slight raise to press against the reflector?
Thanks
Post #425 here: https://budgetlightforum.com/t/-/27963 has details. Yes, you need to do something - shim the MCPCB or you can follow what aoeu did - see this thread's OP - Post #145 description.
How did it handle the heat as at 4.6 amps it should start getting hot quickly?
Yes it did heat up quick with the MT-G2, not sure how it did over several minutes. I don't have it configured like that anymore - it's back to a XM-L2 U2 1A on a 26mm MaxToch. I fixed the switch problem - the driver fits a little loose so the driver mounted switch can get out of position a little, or pushed back with use. So I jammed in some copper braid on the edge of the driver, opposite side of the switch and seems to be working 100% again. The R100 mod is holding up well. Here's my results:
Single KK4000 cell, 3.30A tail, lumens: 1,112 @start, 1,081 @30 secs, throw: 79 kcd measured at 5m
Two KP 26650's, 2.70A tail, lumens: 1,380 @start, 1,343 @30 secs, throw: 100 kcd measured at 5m
Note: all my lumens and kcd measurements are my own taken on a older LX1330B meter - kcd taken indoors and lumens in a PVC constructed lightbox, accuracy can be freely debated
You weren't happy with the MTG or just experimenting?
Can't recall (I'm bad), think the MT-G2 died? There were so many things goin on here with the R100 resistor - if it works single cell, etc., so decided to go to a MaxToch 26 with XM-L2 in that unit. That MT-G2 LED I've had a long time, and it's been re-flowed a couple of times - just not sure if it was ok. I think the MT-G2 was shorting on the reflector, so not sure if it's fried. Didn't know I should have used kapton tape or something - didn't know what was commonly done to mount these... I pulled it off the Noctigon - but again, totally forgot what the problem was. Think I was gonna xfer it to a MaxToch MCPCB... But I did just get in a new bare MT-G2 from KD ($14.50), so will probably use this with a custom zener mod driver - probably a BLF17DD, and mod another Y3.
I upgraded mine for MT-G2 (6V). I drilled the reflector hole a bit larger as well as the original white plastic spacer that was over the led. I used the white piece as an insulator wit MT-G2. It works well, I did not try to optimize spaser any further because during tighening it seems to get pretty good already.
I replaced the original sense resistors with 3 x R100 in parallel (ie. about 33mOhms). It measures about 3.6-3.8A at the tail with two 26650 cells. I’ve not used it for long periods, but for occational use it is one heck of a light now. The beam is somewhat large but it brightens up quite a long large area now. It does get warm in use, I’m not sure if it would take constant full power in summer temperatures.
(I modded another with noctilux XML2 and 2.45A current at tailcap with two cells which should equal about 4-5A at led.)
@boffid: Am I reading correctly that you pulled both the R150 and R180 and replaced them with three R100’s? Was that for both lights? Any chance you measured current at the LED? I thought some of the other guys were seeing currents similar to your MT-G2 light with the R150 + R180 + 2x R100’s. I’ve got one modded with one R100 added that’s pulling roughly the same current as your XM-L2 (mine has a dedomed XM-L2).
So… are the currents going up with lower total resistance or with more? As I’ve said before, my understanding of modding existing drivers is pretty limited.
Original resistors removed and replaced with 3x0.1R resistors in parallel for my MT-G2 3.6-3.8A drawing mod.
Originals kept (0R150 and 0R180) and R150 or R180 added in parallel for my XM-L2 mod drawing 2.45A or so.
(Both with two cells)
Unmodded version drew about 2,4A with one cell and less with two cells.
Sorry, no measurement on the led but assuming a reasonable efficiency one can do rough estimates. Both are bright enough for my use.
Less sense resistance gives higher output current. I’ve got to write a sticky on this I guess.
In most cases, including this one, the math is very straightforward. First you must calculate the total resistance: use a parallel resistor calculator and remember that the ‘R’ is a decimal point, so R100 is 0.1 Ohms. Second you must plug the total resistance into the appropriate formula - for this QX9920 based driver it’s [ Iout = 0.25 / Rsense ] where Iout is current to the LED and Rsense is the total resistance of the sense resistor bank. And then you’re done: that’s the target current, figured out in 2 steps!
Remember that we are setting the target current. For XM-L2 (& similar voltage) especially affects single-cell applications with this driver since it will have a lower maximum current in single-cell applications when compared to a DD driver. So for currents above 3-4A I’d guess you’ll only see the target current on 2s or more. This same logic applies to MT-G2 on 2s instead of 1s: it will be limited to lower than DD performance on 2s.
Gotcha. Thank you for the explanation — that is very helpful. I’m totally new to modifying buck drivers as of these two Y3 builds… everything I’ve done so far with flashlight drivers has been either stacking 7135s or building BLFDD drivers from bare oshpark boards or following zener mod instructions from RMM. Makes a lot more sense now.
If the driver has Ipeak=0,25/Rsense, my values would result in 7.5A for MT-G2 and 4.5A for my XM-L2. However, peak value is not same as the average value. Average value depends a lot on ripple in current. I would assume my MT-G2 must have a high ripple ie. average value much smaller than the peak value.
8V total cell voltage to a 7V load (MT-G2 at high current) would be only 4.2A with 3.7A current at tail, assuming 100% efficiency… With practical efficiency below 100% the average current should be smaller than this of couse. I think 4.2A could be a reasonable value for a very triangular current waveform with 7.5A peak.
If I was really interested on this, I could desolder some wires and measure it with an oscilloscope. However, I’m not. At this point I’m happy with knowing that the average current is reasonably high, the light is bright enough and the led is not severly overloaded (based on average draw at tail and a rough estimate for led current being slightly above it).
Since you aren’t interested enough to hook it up to a scope, take my word as gold (I have hooked buck drivers up to scopes, including two other QX9920 based drivers last night). Unfortunately your post is mostly incorrect. The formula I posted is for the average, not peak, current. The output waveform is much more like a flat line than a triangular wave.
You are running into the situation I warned about in the last paragraph of my last post in this thread (#239). The driver is unable to reach the current you have set. You could decrease from three R100 to two R100 and see almost no difference (your setup will be unable to make 5A, 7.5A, 10A, 20A, etc - it can only make around 3.8A as you’ve said).
You seem to be absolutely right on that one. I added another cell and with three 26650s it did indeed rise up to 4.2A at tailcap.
That should result about 6-7A for the led with reasonble efficiency assumed. 
No, I did not dare to keep it long for a extended period at that current. I’m not too confident about how much overdrive the driver can handle.
Good sport!
If I were you I’d probably target around 4A with my resistor mod (assuming 4A is what you want!). You should be able to do that using just the resistors you’ve got, and it sounds like you’ll probably want to drop the current for your XM-L2 a little too if you’re worried about overcurrent for that driver (6.1A).
That’s just my suggestion though! (EDIT: if you set the actual desired current you don’t have to worry about how many cells you use in the future)
Changing the resistors for a 4-5A max current crossed my mind as well. I think 4A is quite plenty with a MT-G2. For general use that could actually use a low-low mode in addition to the Hi/med/low it has now.
That being sed, I must first give it a quick try outside just to see the huge amount of light it is able to give. 
I’ll keep the XM-L2 as is for the time being. I really like how it throws and I’m not too conserned it it gets sacraficed for the science.
I really love this host even if it is somewhat large.
[edit] Tested it. A real joy to see that amount of light! I guess it kind of “throws” ie. you can see far but it lights up much larger area than any of my other lights. It seemed to start dimming some at about 30s or so at what point I ended the test. No need to destroy this fun toy. (It warms up pretty fast, not hot at that time though. I guess I should have leveled the pill before glueing on the led). I’ll be returning to 4A level I guess.
The Y3 is a great light thats for sure.
But I wonder if anyone tried to put this LD34 driver into this light, the stock driver has 22ish mm and that driver has 23,6 according to spec (haven’t seen an exact masurement). It has an on board switch that could fit. It has a voltage range up to 8,4V and uses constant current regulation. The UI is good imo(Convoy L4). You might even change the current with the Key-Set method that “Simon” posted somewhere (Has to be checked). I don’t know if it works with MT-G2 LEDs but CC would be a great benefit. Sould give you great efficency in lower modes.
edit: Whoops just noticed that the stock driver is a 23ish mm driver so that makes the task in hand a lot easier. Might be a perfekt fit, the little knob won’t fit anyways so you either have to drill a hole or just file it of.