The reason you lose low voltage warning is the two resistors that the mcu uses to “read” the voltage take their input from the power + pin of the mcu. The Zener fixes that voltage so that it is a constant that doesn’t drop until the combined. Voltage of the two cells dips below the Zener voltage. The resistor values are wrong for two cells and I think you would also have to use a 6V Zener to get close to reading the cells at even 3V each but I could be way off on this.
The over-discharge protection isn’t lost, just effectively disabled by the Zener. I haven’t had too much trouble taking cells down too far in the past, but it is always a good idea to keep a mental check on runtime and amperage level used as a safety margin to help prevent fatal error. Fatal to the cells is the most likely scenario, but potentially fatal in other ways as well.
Pulled cells from my K3 and checked em when I finished my last “adjustments” and found them at around 3.8V low for me, lol! The more lights I get the harder it is to keep up with the cells. My memory is horrible! The worst part of that is that I might remember specific things but give those attributes to the wrong light. And then would swear by it, bet on it, get totally blown away that something isn’t so, when I just swapped “memories”
Sucks getting old! Should’ve never quit drinking…
“I Picked a bad day to quit ( ).”
Just watch the output and be watchful for dimming. Low mode is the most dangerous to cell with this mod and why I never actually used it in a light. I wanted to make a separate battery guage pcb on an 18mm board but didn’t get enough eagle under my belt to pull it off. I figured out a board layout but not the eagle part. I won’t need that for the Elf since it has it’s own monitoring system.
Voltage protection still 'works' after the zener/resistor, in that the MCU is still looking for it, but it's being fed an invalid voltage reading. I guess since the other resistors it uses for that aren't altered it would still trip at 3 volts, but that would be with each cell down to 1.5v (once input voltage falls below the zener's rating actual battery voltage is fed to the MCU). Best to use protected cells, pretty much for the same reason protected cells are best for any series-multi-cell light. Not always convenient to do that, though.
Taking the cells down to ~2.75v each is I guess technically 'safe', but not best for battery life. After all most cells have their advertised capacity measured when taken down to 2.75v, I think some are even rated down to 2.5v. And at 2.75v each the light output will be noticeably low, it's pretty easy to spot.
Well, the hype says that the newer IMR chemistries are safer than the standard Li-ion and work better in multi cell configurations. How does this apply in the real world use of these lights?
And let me ask here, how can my S2200 be showing 6.35Vf on 2.53A at the tail and the K3 is showing the same 6.35Vf on 4.05A at the tail? How can such different amperage readings yield the same forward voltage?
Lck Led, Constant Current LED Driver Board - 3 Modes, 5.5-12V 3.0A.
Measurements at the led running off a power supply at 7.6volts input.
High 3 amps.
Med 1.45 amps.
Low .25 amps.
Used in this light.
Edit. Run time tests.
Edit two. Trustfire 26650’s used in run time tests.
At the start battery voltages were 4.1 volts with a ceiling bounce reading of 68.
20 minutes, ceiling bounce 65.6, temp 47C.
30 minutes, ceiling bounce 70.5, temp 55.2C. It would appear that the current from the driver was increasing slightly as the voltage in the batteries dropped.
45 minutes, ceiling bounce 65.2, temp59.5C.
60 minutes, ceiling bounce 63.6, temp 59.9C, maximum temp reached.
75 minutes, ceiling bounce 55.3, temp 57.2.
Shortly after this the light output dropped off dramatically. It still had 3 modes so I dont think it was low voltage protection kicking in. When the temp measurement was around 55C the light itself was to hot to hold. When the light had cooled down battery voltage was at 3.15 volts/cell.
I’m running that driver in an IOS T10B with an MT-G2 right now. I’m very happy with it. I originally had what seems to be the same driver from IOS, that comes with a contact board, but something went wrong with it after a few hours. I don’t know if I did something or what. I sent that driver to relic to see if he could find the problem. The LCK driver has been working just fine. A nice driver for that host.
I’ve used this boost driver from CNQ:
http://www.cnqualitygoods.com/goods.php?id=1391
for several small triples, and I’ve been trying to “characterize” it, so, out of curiousity (asked RIC, but never responded), I was wondering what it’d do with an MT-G2, so just for fun, I hooked up a 6V MT-G2 with it, outside a light, and with a single TF 26650. I measured both tailcap and emitter current using a clamp meter:
Tailcap: 1.65 amps
Emitter: .65 amps
I was kind of surprised that it actually worked, and actually, the light was kind of blinding :)! I also think that I could get a little better current, because I was using long, relatively high-gauge test leads, plus a tailcap switch that I pulled off of an old light.
It’s not a super-powerful driver, but thought that you all might be interested in something different :).
Ill say up towards 10A with with "22 AWG driver" and the right 26650 batteries. 0:)
Hi just posted here
With Noctigon + MT-G2, Nanjg 105c from FT and 8 extra 380 mA chips with zener/resistor mod (AW22 silicone wire short as poss):
Tailcap readings:
Got 5.84 A high (Actually hit max calced from chips on DMM - approx AW 12 leads!)
Med 2.4 A
Low 0.31 A
7.2 A draw DD on same batteries.
Incredibly bright! Calced run time rather short though with 2 x AW IMR 18350’s! Approx 7 - 8 mins! 
If anyone has any suggestions for a battery extension tube for this POPPAS Id be very grateful! 
Edit: Just thought Id ask has anyone tried the East-92 driver with 2 x Li? I was speaking with Ryansoh and perhaps a similar mod would be possible to protect the MCU if that might be a prob? Cheers
Thanks grainde. Added to the OP.
Zener modded Qlite driver with 12x380mA 7135 in C12 host.
Tailcap Measurements
Moonlight: 0.01 A (or less)
Low: 0.10 A
Medium: 1.25 A
High: 4.49 A
Thanks RMM. Added to the OP.
I had a few spare drivers that were pulled from the J-19’s I modded. It’s the same driver as in the old 3XT6 trustfire I am told. It says 4.8amp on the outside of it. It does a very nice job running a single MT-G2. I can’t say how many amps or lumens exactly but it’s pretty nice. Not as many lumens as the dry driver with two batteries but not too far behind. Plus I de-domed this one so a side by side comparison doesn’t really quite count. However if I had to guess I would say that it’s actually putting out the 4.8amps. Both the DRY driver and the 4.8amp driver were rewired with 20 gauge wire.
It’s in a ZY-T08 with the flat bottom of the reflector taken out, the batteries in series, emitter de-domed, it turned out pretty nice. Obviously it could be pushed harder but I’m not going to mess with it. It outthrows a stock T08 with tons of flood and isn’t heating up bad at all.
Trustfire 4.8amp driver = success.
Thanks 18sixfifty. Added to the OP.
Has anyone done a straight DST with MT-G2 conversion and stock driver? ohaya’s above is listed as “stock” in first post but actually directly connected to cells.
Maybe you could add my X7 MT-G2 mod to your list.
IMO, its the easiest way to get inexpensive high performance from a host that caters to the MT-G2.
- Only mod necessary is the emitter swap and a 20mm copper disk to space it properly into the reflector
- Seems to come stock with a TR-3T6 driver
- Runs on 2 or 3 x 18650’s
- Reflector focuses the light very well
- Around 6.5A to the emitter… nice!
- Host provides great heatsinking for its size
- Cheap!
Thanks for the update FlashPilot. Added to the OP.
Just wanted to post that I used the same Lustefire driver as Relic38 did, the LusteFire 3XU2-X3 and it rocked. I used it to drive a single MT-G2 in a ZY-T08 and all I can say is WOW!!! I had to file it down to get it to fit but it worked perfectly. I don’t know how many amps but it’s enough to pop the protection on two Panasonic 3400’s. It doesn’t do it right away though. It takes about 30 seconds for the protection to turn it off. With samsungs it’s no problem and damn is it bright.
I’m going to ask Wallbuys if they will carry it.