Mod: My latest retarded super bright light ( <15000 LUMEN of NW tint mixed light and 200+ kcd)

How often do you put on a light that makes you laugh due to its massive output?

Just happened to me, and Im well used to to having a ton of lumens..

About 10 months ago I made a tint mixed 100W+ 9 XM-L light. Thread. That was pretty retarded in terms of brightness and heat. 100W really needs a fan..

I obviously forgot about the heat issues and decided to make an even brighter light. J) With better tint, better efficiency, more runtime (more batteries), better UI, more compact (shorter), not much heavier and with more throw.

I used the OEM 12 XM-L light as a host. Pictured all the way to the right. (BTU left, TK61 middle).

You can probably find the same flashlight labeled kinfire, ultrafire etc..


Emitter setup:

4x XM-L2 U2 2C

4x XM-L2 U2 1A dedomed

4x XM-L2 U2 7D4 in the middle

Some Noctigons got some excessive cutting, due to bad planning. I drilled two more holes for the emitter wires.

Work in progress...

The heart of all this madness is a little 17mm zener modded BLF FET driver that I put together with one extra FET.

With 6 batteries and 12 emitters in direct drive. I was expecting some serious amps. So I needed a serious switch. I decided on a "twisty switch".

Early design before things was put together.

O-ring on the brushed aluminium plate makes it waterproof.

The longer this stayed on my desk, the more I kept adding. :D lol.. On the positive side, looks kinda artistic, and can carry lots of amps.

Copper plates in the middle.

Wired up the emittes with 22awg wires (two and two in series until all twelve were done). All those wires then connected with some 18 awg wires.

Made a contact plate.

Brass bit on top of it.

This will then be connected to the brass plate where the batteries touch.

Done, looks just like stock..

My centering of the XM-L2s turned out quite bad. I used Artic silver 5 in the middle of them, and fujick on the sides in order to make them sit tight when doing the emitter wiring. They were nicely centered when I attached them, but not when I was done putting the head together after the wiring. hmmh.. Not sure If ill bother to redo the emitters. Not a big deal in a 12 emitter light. If loosen up the reflector a little bit I should be able to improve it by twising the reflector half a millimeter. Put kapton tape on the backside of the reflector. Did not use stock centering rings due to some fitment issues. Also, I like seeing red Noctigon clearly visible peeking through. :)

As you can see. Nice head size on this light compared to a typical 9 XM-L light. Many 12 XM-L lights are closer to the 9 XM-L in size as far as I know...

Switch on the toilet plunger. Twisty tail cap on the "shorty".

At the moment I don't have modes. Maybe because I decided to move the FETs a bit too much to the side (in order to easily reprogram the MCU in the future)? I did something wrong after it was tested or in the wiring process. I knew I did not have modes before I fired the light up with batteries, but could not care. I have to take out the driver in the future to reprogram with a more suitable UI than the whiny phase correct STAR firmware. This light really needs Turbo that goes down below 50% in about 60 seconds. I just wanted to see how it worked now. :)

The 9 XM-L light is 916g with 3 x26650batteries.

The 12 XM-L weight 1197g with 6x18650 batteries.

Not bad considering all the advantages I have with the 12 XM-L light. :)

Some (peak) numbers:

Tailcap amps: 30A (measured at the copper bits on the "twisty switch" maybe a bit more if measured with clamp meter and thick wire)

Emitter amps: 5amp to each of the 12 emitters (5Ax12emitters=60 amp if you like to look at it that way)

Watt: Should mean its about 220W.

Lumens at start: About 14300 and sinking fast below 13000 (others might measure 15000+. Remember, this is NW light with some de-domed emitters. The emitter setup is not even made for max lumens. Its more oriented towards tint and a bit increase in throw. No AR lens either.)

About 227kcd after 20-30seconds. ! Wait what? 227 kcd in a "flooder". Yupp. Easily outthrows a stock TK61. Actually, It can play with a "stage 1" resistor modified TK61.

Heat: No way im letting it run on high for more than 2 min. The heat it makes in just 60 seconds. Its ridiculous. Certainly just a dangerous show toy when used on high/turbo/max with high discharge cells. A very fun toy I must add! ;)

Cells used: Either 6x Samsung 20R@ or 6x LG HE2

Where are the beamshots?

Might take some in the autumn when I have some other interesting lights to compare it with.

Did I say this makes me laugh and smile just by turning it on....

Giggles like a little schoolgirl... :D

The light output and heat is quite retarded when its on full power. :D

I believe in every word you wrote in that last part.

What a build, crazy stuff and maybe a little bit retarded, who cares about modes and if the emitters aligned 100%.
I couldn’t care less with this thing in my hand standing on a pitch black field at night.

I guess, this makes me retarded too, please count me in to the giggling schoolgirl party……….

holy $%^#$

Lets get retarded in here..


Ill improve on the centering and fix the UI when I open it up again. For now, just need to get a feel of what batteries I will be using it with, the heat, and how the UI/modes needs to be. I just had to get her up and running, and then I just had to post about it too before doing the finishing touches.

Such madness. Once again you've modded a light I want to mod.

Racer, that is amazing! I'd be giddy as a school girl after her first kiss with that light too!

Only one 17mm fet driver is running the whole show? What does adding the second fet do? I needs to know. :D

Wow! I love it R86.
That tail twisty is way past my skill set. It makes my brain hurt just just trying to figure it out. Two minutes of insanity. It’s the kind of light you only turn on if there are people around to see it.
Mine is still in the box. I haven’t had the energy to deal with twelve emitters, and I didn’t even think about what would happen to the switch.
What does stacking an extra FET accomplish?


And Im already looking forward to my next big mod. Hopefully it will be a success. Im making a nice TRIO of powerful lights. The BTU shown in OP might not make that TRIO. ;) I think you might like the lights I end up with.

Im usually doing lots of small lights in between working on larger mods mods, so it might take some time before Ive got my full trio done. Still some weeks until its properly pitch black dark outside too.

Thanks JM and Ouchy. Im not sure how healthy it is for one of those FETs to have 30A flowing through it continuous. So I added another one so that they could share the load. I tried to find the correct datasheet, but not sure If I was looking at the right one. Maybe someone can tell me if that was necessary and how much the typical Vishay 70N02 are rated for.

I knew others had done around 15A through them in flashlights. So with 30A divided on two, I should be looking at around 15A too, which I believe should be fine. Two FETs also means a tad less resistance than one. As seen by lots of the wiring and tailcap. I tried to make everything capable of high amps with little resistance.

What do you think the current would be without the stacked FET?

All I know is that I have the FET driver in my hand that’s going to replace the one in my 2xMTG + 2xXML light that I cooked the MCU in. I’m going to take the FET off the old one and stack it on top of the FET in the new one. It can’t hurt, and I’m curious see if the current goes up.

Comfy or someone that are into these FETs or more into electronics can probably tell.

As long as the FET can handle 30A continuous, Im guessing barely any difference. There is not much resistance in them. As mentioned, I was not sure what its rated for. I prefer to have some "headroom", and hopefully I got it.

Halving the resistance in the FET by adding another one in parallel doesn't gain much, since the resistance is so low already. If they were crappy 30 milliohm DTU FETs you would see a big gain, but the better the FET is the less you get by adding more. Diminishing returns approaching the zero bound or whatever...

What's the battery layout? 2S3P?

Its simple. Ive seen what you can do. You do way more tricky stuff than that.

Extra wires just go from the springs to one copper plate. Then from another copper plate to the screws that attaches to the body. Once the tailcap is properly tightened the two copper plates gets connected.

I could have used 18awg wire instead of two 22 awg as an example. But after it was supposed to be finished I decided to add more.

Its nice to not have to worry about the resistance in the tail.


What are the FETs rated for? (continous amp)

Okay. I think I’ve got it. All those wires lake it look confusing. Things are always easier to figure out when you’ve got the actual part in your hand. Actually, I may experiment with that on another light.

A lot. I don't know enough to be able to translate those specs given at different voltages than what we use, but I don't think a single FET would be in any danger.

Wow, just wow! I think the double FET is a good idea, though maybe not much of a difference in output as comfy says. I've been wanting to try that. Maybe do the cap trick of batt+ to ground as comfy found, if you didn't already. Having the 6 cells is a big plus for this host, as opposed to the J20, though the J20 has that massive head.

My 7k+ lumens J18's also throw incredible, not as much as this, but still amaz'n for a flooder.

I find these FETs in the drill packs at the recycle bins. Check the amp draw and the low drain-source resistance

Total gate charge = 160nC, unable to be driven directly by an attiny13. Gate charge has to be around 40 or less. This is the problem with ultra-high performance FETs, if they can do huge current at low resistance, they can't be driven directly without adding other parts, like high-power gate drivers. FETs that are easy to drive have lower performance. There isn't anything out there that has a better balance between those two things than the Vishay 70N02, not by a long shot.

The gate load presented by two FETs in parallel could also be an issue with RaceR86's driver not working correctly (though there are other things to check first, reverting to a single FET for testing wouldn't be a bad idea).