Lume1-FW3X: Constant Current Buck-Boost & FET Driver with Anduril1/2 + RGB Aux

I wanted a set but the $8 shipping killed it for me sadly. For $29.90 I can almost buy another whole FW3a with free shipping which makes it unjustifiable.

My Lume1 driver with aux LED board just arrived! :heart_eyes:

I’ll have to try installing it this weekend.

Also got a few Oslon White 2 boost LEDs. I plan to try them in my Sofirn S11.

Is there a newer datasheet?
https://cdn.shopifycdn.net/s/files/1/0031/0155/6806/files/LUME1-FW3X_Datasheet\_-\_PRELIMINARY.pdf

I installed one of the drivers I received yesterday. It is definitely not the easiest but doable.

1) start by filing down the pcb cut marks.
2) My FW3A had a flat lip on one side of the PCB, but Lume1 did not. If you don’t file it down, the driver does not sit properly. The circuit still works fine, but you will have a 1mm gap once you screw on the tube. I filed it down to the inner ring on the pcb, and then it fit just fine.
3) I used 0.3mm (28awg) enameled wire for RGB wire, marked them ahead with sharpie dots to distinguish later.
4) You absolutely have to have the original mcpcb. I had a noctigon mcpcb with sliced dogfarts, and did not have the right wire routing for secondary PCB to sit properly. So I switched over to original FW3A mcpcb with SW45K flowed.
5) It is a major PITA to make sure that the secondary LED pcb sits properly. It does not need to be flush with the mcpbc (I was not able to), but it needs to be as close as possible. In then end I had about a mm of gap and the optic sat nicely.
6) By the time I got to wiring the secondary LEDs, the sharpie dots were already cleaned out. However, it is ok if that happens - put together the battery tube, and test out the primary LED. Once you verify it works, turn it off and then start probing the secondary led wires with a voltmeter, you will find that one pair has some voltage between them on the order of volts. Between those, the (-) polarity is K on the secondary PCB, and the remaining three wires are RGB. Now, the modes are cyclic, so I honestly think it is not important how you wire between RGB wires, all permutations are valid, they just change the blinking order. I wired them randomly, and the flashlight seems to be working quite nicely so far.

In the end, I have a FW3A with a buck boost driver supplemented with a FET for DD and SW45K. It is the almost the same size as my SW45K SC64C LE (not McBob version, my DIY version, but definitely inspired by him) and a similar efficiency for most lo-mid modes, but with a much more powerful hi-mode. Honestly, this is quite the holy grail for me, thanks lone ocean.

Next, I’ll switch over my FW3C to Lume1 and the sliced dogfarts which are another favorite of mine for when I prefer purer white of sliced LH351D over rosy 219B.

We already went through the pcb variations with Lexel’s boards, even the OEM ones have like 3 versions. If you have to you can scratch off a new pos or neg pad if you really want to.

Thanks for the report. It sounds like the main challenge is fitting the RGB board into a tight space?

Do you think a Noctigon board can be filed or drilled to make room for the RGB wires?

To be honest, probably it can fit, I initially thought that the secondary led needed to be flush with mcpcb so I did the swap.

I ended up with a 1mm gap between them, probably that has enough clearance to route the cables underneath, though you would have to make sure that the solder bridged pcb connections on the mcpcb are not too tall, and that the primary led wires are somewhat flattened.

You know, I always try to salvage the mcpcbs if I lift off one of the pads. I was never successful, maybe I was unlucky. Is there a trick to it? I usually use a sharp blade to drag/scratch the coating

I’ll add a bunch of commentary and tips but here’s album from my build today:

For those of you that have installed this in your lights. Are your green aux leds much brighter than the red and blue leds? I just installed this driver rgb board in my light. The mixed, red-green and blue-green, colors look solid green instead of yellow and teal like my other lights.

The resistor on the green channel has a value of 100ohm. Does anyone know what the correct value should be? I checked some of my other lights and they have 120ohm resistors for the green channel.

I would like to confirm that the triple 219B can be used with Lume1 with the existing/installed firmware. The stock driver (the one with 7135’s) needs the different version of the firmware (less current) for the triple 219B.

219B SW45K with Lume1 is the holy grail for me as well. E21 would be good, too. It should result in the smaller hot spot/less flood.

I believe with a high capacity cell the 219b should be fine. There is a series sense resistor with the FET and that should probably be enough to keep the current in check for those emitters. Just my gut on this though…

Just finished installing my Lume 1 with aux RGB board inside my LM10.

It took me around 3 hours to complete the assembly, but it works.

Some assembly notes:

  • It was necessary to file down the outer edges of the driver board to get it to fit in the socket. Unlike the FW3A, the LM10 doesn’t have a flat edge on the driver so I didn’t need to bother with that.
  • I also had to file down the outside of the aux LED board to enable it to fit in the star’s compartment.
  • Even then, my aux board didn’t fit over the LEDs properly Not enough leeway around the holes in the aux board for the LEDs and they didn’t quite line up. I used a square needle file to make the holes slightly bigger until it fit. I might have been able to omit this step if I removed the screw preventing the star from torquing as it may have been holding the star to one side of its compartment. However, I liked the protection for the wires so didn’t do that.
  • It took me several tries to solder on the driver wires on the star. This was expected as it is very important that they do not overhang their slots in the aux board. If they do, the aux board can bend and break when pressure is applied. I had to desolder and remove excess solder and shorten the wires a couple times before I had it just right.
  • The bondpads on the aux board are absolutely tiny, and they’re all together… separated by maybe 1/10 of a mm each!!! I spent an hour soldering them in place as after each wire I had to pull out my DMM to check to see if it shorted to the pad next to it, which happened often. When it did, then I had to redo and try again. I ended up using 2.5 power magnification reading glasses combined with a fine tip on my soldering iron.
  • Once I finally got the aux board soldered up and working, I then discovered the optic wouldn’t fit since the holes in the optic were not perfectly aligned with the holes in the star. To fix this I took a fresh #10 X-acto knife and thinned down the legs on my Carclo 10507 optic until it.
  • It is VERY important that if things don’t fit, do not force it. Those aux boards are super-delicate. A small amount of pressure the wrong way might cause it to snap in half. And don’t use much pressure when filing it down either.

My impressions:

  • The Aux LEDs are dimmer than that in my Emisar D4v2. Especially the blue. And that applies to both high and low brightness settings. Low brightness mode is barely visible. I think they’re a bit too dim in low brightness mode. I’m not sure they could be used to locate the light at night unless the bezel is pointed right at your face from close range.
  • This is a VERY tough soldering job. The contacts on the aux board are too close together in my opinion. For a hobbyist board they should be spread apart. Or at least have a couple mm gap between them. I suspect most modders aren’t going to be successful in their wiring of the aux board. I found this board far harder to solder on than the aux board in my Emisar D4V2 or Lexel’s aux board in my Emisar D4v1. If you do plan on soldering this I recommend using a DMM that tells you when you have a short. Also, plan on the project taking multiple hours. Then slowly and methodically work your way through it without forcing anything.
  • Ramping mode is not completely smooth. It smoothly ramps up, but then there is a stairstep jump in brightness at the very top of the ramp. This is different from an FW3A that smoothly ramps up all the way except for the 2 blinks when switching channels.
  • I don’t think there’s any chance I’m ever going to change the main LEDs in this light, since if I did I’d have to redo the nightmare soldering job on those tiny bondpads.

PWM of FET is disabled and not recommended with Lume1-FW3X. FET can either be on or off.

This jump only happens when top of the ramp is configured to be 1 click (maximum), the same as turbo. You can configure top of the ramp to be 2 clicks (one less than maximum), and it will be buck-boost only and FET will be available separately when you double click for Turbo.

Glad I didn’t jump in this yet. Does not look like the funnest mod to do.

Other users did report this so far and I can confirm as well. It’s most noticeable on “yellow” I guess. Not sure what the correct values should be in fact I cant even read the values they are so small.

I agree with most of your points but I think this series of notes could be due to your LEDs not having been re-flowed perfectly centered. The dry fit is fine and everything drops into place. In comparison the Lexel boards have a bit more room in the LED openings and a bit less room for LED wires. I assume your LM10 uses that extra thick MCPCB the Ti models use? Something could be slightly off there too.

My best advice for LED wires having done several Lexel boards and now this one (it’s still not easy). Break the suction of your thermal paste just before soldering, use a big blunt tip, use flux paste on both ends, and right before you go to solder touch just a bit of fresh solder on the iron. Do all this while the aux board is flat against the PCB and none should get “under it”.

Yup I thought of that. And thought of possibly reflowing the LEDs on the star. However, when I inspected the main LEDs closely they appeared to be perfectly centered on their markings on the star so any reflow would have been guesswork. I decided it was easier to try to file the aux board down than try randomly reflowing the star.

I have also installed multiple Lexel Boards and a D4v2 aux board after main LED reflow. None of those boards had this issue.

I knew it would take several tries since I wanted to solder on the main LED wires before I installed the aux board into position. I figured this meant it would take me more tries to get the LED driver wires bonded to the star just right, but I wanted to minimize the chances of damaging the aux board.

Still, it didn’t take that many extra tries to get the driver wires to fit properly. An extra 15 minutes maybe.

By far the biggest problem with this install were those 4 tiny pads on the aux board that are almost touching. Those were insanely hard to solder and took the most time. And once I finally got them right there was no way I was going to take it apart to try to recenter the LEDs.

Still, I think I could avoid needing to file down the inside of the aux board and the legs on the optic by doing what you did in your last post:

Before attempting any wiring, remove the star from the light and try dry-assembling the star, aux board and optic. Then try fitting them all in the light without the driver or any wiring installed. That way you can ensure everything fits before starting the difficult wiring job.

Had I done that I might have been able to shave an hour off time spent on the assembly.

Those tiny RGB pads do make this 100x harder. I know nothing about PCB design, I only wonder if maybe the K pad could be relocated and then spread out the RGB pads evenly in the same spot.

I tried to get away with not taking my own advice and pre-tinning a few RBG wires and that was a big mistake. I’m using stranded 30 AWG PTFE wire and it wanted to spread out and fray on me. In my series of pictures I also left the RBG wires too long so they really had to be pushed down to make optic seat.

I think for me the biggest challenge was getting the wires in position to comfortably solder. The actual solder process is so quick and with flux they resist making shorts since it tends to melt straight downwards. I still did redo some that looked bad and ultimately ending up with a “good enough” result.

Hopefully the driver is rock solid, able to be flashed easily, etc and the aux board can just be “icing on the cake” for people who want the additional challenge of defeating the final solder boss.

I would say it’s a matter of risk. The 219B’s on a FET driver will be overdriven relative to Nichia’s specs. I believe there are some people using 219B’s on the stock driver with the stock firmware, on the assumption that because they don’t use turbo much, and use a medium drain battery like the Sanyo 18650GA or Samsung 35E, it is low risk. However I may have misunderstood some of the previous posts.

There is a Nichia-targeted firmware in Toykeeper’s repository that limits the FET channel to 50%. This technique probably shouldn’t be used with the Lume1, because it can not PWM the FET channel. On Lume1, the FET channel is max on only. The Lume1 also has a customized version of Anduril, and I missed if the code for that is shared yet.

If the Lume1 firmware is tweaked to disable the FET channel entirely, then it is entirely safe, but obviously the max output is reduced. However, you still get probably 700-800 lumens, high efficiency, and RGB aux capability, all in super high CRI.

Keep in mind E21A has noticeable tint shift across the beam if using narrow optics. As a result, most people choose the medium beam optic with the E21A, so the spot is not very small.