New-117 (TK35 clone) mod. Turn a TK35 into a TK35UE (Updated with Beamshots 3-jan-2015)

Is this the New-117? Is the PWM of this light really as bad as many had commented on? I am thinking of getting one too to mod with MTG2 :stuck_out_tongue:

the PWM on low and Low-Low is bad indeed, it does not bother me that much but if you hate PWM you will hate the stock driver on the low settings. im planning on switching the uC out with a custom programmed one (so i can have ramping but still use the rest of the driver), this way i can also use higher PWM speed.

My New-117 was forgotten because of a faulty switch. After seeing your post I replaced the switch and changed the emitter to a XP-G2 on noctigon. I’m not super sensitive to PWM, but I can see it on the lowest setting on a white wall up close.

Anyways, great mod you have there!


Thank you for posting this. I think this clone will have a new life of its own now that folks are looking for 2S 18650 hosts to install MT-G2 emitters in.

Now my question(s), I noticed that you are the first one I have seen to by-pass the resistor on other then on Direct Drive, Driver circuits. Everyone else seems to “double”/stack another resistor on top that leaves some resistance in the circuit.

  1. Why did you opt for Jumping the resistor?
  2. Does our clone need some resistance in the circuit?
  3. Can other resistors be jumped/others added to also boost that individual setting for the light?
  4. You jumped R2, what would happen if you (only/also) jumped or added a resistor to R1?
  5. Is this as far as we can push this driver (short of further circuit current loss reductions)
  6. Oh don’t forget we are eager for some outdoor beam shots.

As you can tell I am not an expert but I am working hard to come up to speed.

Oh do you know of any options where we can address the less then desirable PWM this circuit has? you mentioned “switching the uC out with a custom programmed one” do you have a microcontroller in mind?

Thank you in advance


I’ll answer most of your questions:

  • Shorting any or all of the sense resistors has the same effect: preventing the buck circuit from working correctly. The controller will no longer step-down the voltage, so it in effect becomes a crappy DD circuit. There are many explanations on the internet of how a simple buck circuit works. Once you read one of those and then imagine lowering the resistance you should quickly identify what will happen.
  • In this case Vin and Vout are close enough that DD is fine. Rather than shorting out sense resistors there’s really no reason (other than cost!) not to replace the stock guts with a proper DD driver such as the BLF17DD V1.0 or the A17DD-SO8.
  • All the modes in the light are created by PWM of different duty cycles. There is only one resistor bank. This driver is based on the QX9920 controller for the buck section along with a (PIC, probably) microcontroller to produce modes and possibly LVP.

The stock light uses a buck circuit to bring 2s voltage (6.0v to 8.4v) down to 1*XM-L or 1*XM-L2 voltage (~3.0v to 4.0v range). With the higher voltage MT-G2 there is very little headroom for the buck circuit to operate in.


Thank you for help.

I read “Understanding the difference between Linear, Buck, Boost and Direct Drive drivers” at Understanding the difference between Linear, Buck, Boost and Direct Drive drivers

And I think I follow most of it. but I think the new MT-G2 and XHP emitters are all exceptions to his conclusions about design number limitations do to the higher needed voltage.

I looked up “17mm & 20/26/27mm single-sided DD/FET driver release: A17DD-SO8 / A20DD-SO8 / etc” at 17mm & 20/26/27mm single-sided DD/FET driver release: A17DD-SO8 / A20DD-SO8 / etc regarding the “BLF17DD V1.0”

And “BLF17DD Info Thread - Reference” at BLF17DD Info Thread - Reference and frankly I am more then a bit lost.

I know that old-lumens took a tk35 clone and stripped the boards and made it 2p and put new stuff on the board but that kind of work is above my skill set. I liked this host better because unlike the Small Sun zy-t08 this host is already 2S.

I can solder a jump wire or resistor onto a board (I know big deal :slight_smile: . I know that the TK35 Clone is a very different size and configuration then say the typical C8. so how do I get one your recommended drivers screwed into the TK35 clone? each of the two boards is screwed on with four screws. Do they come in this configuration?

From one of the postings I saw you have had your challenges upgrading wiring. If the design goal was to push “HI” to 4-6 Amp range I think (but am not sure) that 22 gauge silicon as shown in the post above should be sufficient. What do you think?

My understanding is that some work has been done to develop a good 6V Driver but I do not know if the design has a consensus yet or if it is available in a screw in configuration for this host, that to me just screams for a MT-G2 emitter to make a very usable all around light.

ANY help is appreciated.



I like New117 because it is relatively compact for 2x18650 host and it is already 2s (no need for ugly hack, eg:ZY-T08).

If I were you, I would start something simple like bridging/jumping the current sense resistor and upgrading wires to the emitter to thicker ones, while still using the stock driver.

Those are what I did. I plan to completely to change the driver someday in the future, if I have the time and grow bored of it. But at this present moment, I find that I’m quite okay with the stock driver UI and the brightness (I can get up to 4A driving MT-G2 with a good pair of batteries).

Just give it a try, and later decide whether next you are going to replace the driver, or not. Aside from the MT-G2 emitter, it almost costs you nothing (eg: only wires vs new driver).

This may call for a slightly deeper understanding of how things work. DD is what you can get away with whenever input and output voltages are close enough to settle down into a… … now where did I already go on about all that? hmmm… . .

OK, take a look at my posts in this thread (Note that I made a typo as Bright_Mike points out, this has now been struck through. Focus on posts #9 through 23): Basic Driver Questions

Also take a look at my posts in this thread (Note that DD ‘drivers’ are NOT part of the discussion. Later in the thread when I say “LED drivers are all constant current” what I really mean is that generally nobody is selling Constant Voltage LED drivers. Focus on posts #7 through 18.): How many amps can a flashlight pull ?

Report back on your findings of course.

Can you link to whatever you are looking at with me & wiring? I’m sure I have plenty of difficulties, but I don’t remember whatever you are referring to. (24AWG can be OK for 6A, depending on other factors. You’ll probably have a better handle on it after scrutinizing the graphs I refer to in those threads.)

Now, as to how you’d fit an AxxDD-SO8 driver… they are very small drivers. You’d just bypass the stock driver and put the new driver in the middle of the sandwich. No problem: PLENTY of space in there.

-Yup common sense strikes again. :slight_smile:


I guess My questions to the greater community are:

  1. What can be done with the existing boards (I believe this is largely answered) but I wanted to leave the door open if there is an idea out there.
  1. - Is there a consensus around “driver” that is available that replaces one or both boards and can be screwed in and wires connected?
  1. If there is no “#2 solution” then what or whom are good sources to get involved to get’er done. :slight_smile:



I have read both links provided -done

Regarding wiring I did not intend it to sound like a dig, I simply saw a bit about 18g wire that you and cereal killer discussed. As far as what I am looking at, it is the 3rd photo of this post (don’t want that to sound like a smarta)

My plan is to buy some 22g or maybe 24g if you say that is not beyond the point of diminishing returns for the mods to this clone, if that’s a bad call let me know.

The AxxDD-SO8 sounds like a good easy solution. do I need to strip out components from the existing boards? or does it really just fit in there?

I know enough to know I don’t want to learn how to design PCB’s, I know that the folks here who do the designing have forgotten more then I will every know.

With that said my gut still tells me “we” should have a MT-G2 emitter (“bolts in”) driver for this light. it just deserves it! :slight_smile:

Oh and this thread “17mm single-sided / DD driver w/ low parasitic drain for e-switch lights: AxxDD-SO8+LDO” started by a very smart guy at 17mm++ single-sided / DD driver w/ low parasitic drain for e-switch lights: AxxDD-SO8+LDO seems to make a good case for a design with no zener diode in it, although this TK35 clone is clearly a clicky light (aka mechanical switch) and as the thread says thus parasitic drag is a non issue, but the moonlight mode argument is a good point.

Yes I can see you were the thread OP. :slight_smile: maybe you are just the guy to come up with a PCB’s / Driver for this itching MT-G2 platform?

After reading both threads you probably see how DD works: between battery voltage sagging and LED current being tightly tied to current, cell performance dictates maximum drive current…. assuming your DD driver is good (Like A17DD-SO8 or BLF17DD, which is just as good.) The performance of the stock driver with resistors bypassed is limited by the baby FET and the inductor (coil).

I did not take the wiring comment as a dig, don’t worry. I often need clarification when people mention things. 18AWG is only for extreme mods. 22AWG will be fine for your purposes I think.

The A17DD-SO8 is very small. Take a look at the pictures in the OP of the driver thread. I’m sure there are more pictures deeper in the thread, but those should suffice. If I was putting it in the stock light I’d strip the bottom PCB. I don’t think the top PCB needs to be stripped. All you really need to get to the new driver is BAT, GND, and SW. All you really need to get to the top/contact PCB is LED- and BAT+.

Unless supply shows up again there doesn’t seem to be much point in making a new driver for an unavailable light.

I wouldn’t bother for this application. If you are interested in very low moonlight levels then one of the new drivers (DD+single-7135) would be a good choice except for the lack of a Zener or LDO. You can still Zener mod them as I mention briefly in this thread: [WIP] 17mm DD+single-7135 driver / single sided / Dual-PWM

The lights are available, It is not like they are out of production (yes the names on model numbers keep changing but it is the same tooling). As single cell lights have been the big mover it has lost is appeal from 2011 or when the tk35 came out. But now with folks showing a growing interest in 2S, MT-G2 mods I bet this will be ever more available as “they” see market demand.

To some degree the TK35 UE and the new Nitecore P36 are cases in point of new supply for similar 2S MT-G2 lights.


Dealsmachine has them, ordered one a few days back. Shipping was 3$ or so.


I have also found that they can be also listed under the following model numbers:

Small Sun ZY-T92
Unique Fire UF-F10B
UniqueFire UF-F10
and some times “mislabeled” as Fenix TK35

I think in my search I found the “current” mfg but I am not sure? The distributor I bought mine from I know is operating under two different names.

I think if the Mfg was smart they would release their own “TK35UE” for the market but maybe they need some help? Which might be one of the benefits of a forum like BLF. So maybe both our community and the Mfg can benefit from each other.

I was also thinking that it sounds like you could whip out a board design pretty fast because as I understand it, it is not really a design challenge other then laying a “smaller” design on a larger board that has the holes for the screws and wires added in the right place. Yes I am sure I have over simplified it, but 3-5 brightness modes or if strobe or sos flickers first does not seem to me a deal breaker. But having the ability to open it up, solder bigger wires coming of the batteries and then screw in the new boards and solder the needed wires in place could put this on the short list of desirable hosts to mod or for a “new new 117” to be built on.

To me one of the questions is, “Is the MT-G2 emitter usage in growth phase or has it peaked and has matured and is now on the decline phase of the its life cycle. for use in Flashlights?”

Or maybe I have flashaholism? and I am not able to make rational decisions? :slight_smile:

this is correct, if i would try and hook up a XM-L (3V vf) the LED would go poof. i went with shorting out the sense resistors just because it was easy and fast. it gave me the exact output i wanted as this light does not have a great amout of heatsinking and would get to hot if it would run on high for a longer time. i am still looking for a propper ramping firmware, so if anyone knows one feel free to PM me.

as im having finals i have not found any time to take some beamshot. as soon as i can (and if i dont forget) ill take some and post them here, but there is no real hot spot when using this light outside. but because of the amount of light it can still throw 200 - 250meters


cereal killer said

“I just did a new-117 for a guy here and really liked it, I’ve got another I’ll be doing for myself when I have time.

I used the factory driver but stripped off the MCU and used a 10F322 air wired on with UFLC which works really well with this switch setup (also resistor modded it for the desired output). It’s a pretty damn good driver once you get rid of the horrid PWM rate, seems to remain in regulation down to about 3.85v (per cell) from what I was able to see.”

I am sure that makes sense to you, I have no idea what a 10F322 or a UFLC does

the 10F322 is a µC:
UFLC is a Firmware: Ultimate FlashLight Code.

My notes on how it was previously sold are/were similar: What is most powerful 2x18650 flashlight in side by side series circuit format? - That said, it’s out of stock from the normal sources. I won’t pay >$30 for this to use as an MT-G2 host. (I have one, the quality does not put it >$30 IMO.) As per tryps Dealsmachine seems like the best source at the moment (if they are telling the truth about stock) but I don’t think highly of their price.

The bottom PCB probably needs at least one component removed for any driver replacement and a custom driver would still need to be assembled… so it’s really on great advantage.

As far as the MT-G2 peaking… who cares? High voltage definitely isn’t peaking, so we’ll continue to see emitters which need at least 2s just to DD.

Thanks tryps!

I always recommend a sense resistor mod rather than a sense resistor bypass. If you aren’t comfortable with higher currents in that host (fair enough) then you won’t be able to drop high drain cells into the light as you have it built now. With a resistor mod you could. A resistor mod would also allow you to use 4*CR123A in a pinch, not that most of us are commonly in the sort of pinch where we have access to CR123A’s, but still. OTOH you’d be forced to settle with a somewhat lower maximum drive current with the resistor mod.

As far as ramping firmwares, take a look at what ToyKeeper and comfychair were working on over in comfychair’s Roche F6 hacking thread. The Roche F6 is also a dual-switch light, but I don’t recall whether these firmwares included memory or what. The firmware called Ferrero_Rocher or maybe “Ramping_UI_table.c”? in ToyKeeper’s repository is probably what you want (it has ramping and other features). I think the development conversation continued elsewhere.

dentillozie is right. Not very interesting to me, the barrier for entry with PIC is slightly too high and I’m already setup for Atmel (ATtiny / Atmega). UFLC has plenty of great functionality but runs only on PIC of course…

EDIT: struck incorrect text.