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

Thanks! There are currently no plans for a group run though, but we'll see..

In this case, the RGB channels are driven directly from the MCU and are only meant for dim aux lighting; again the original intent was based on the use case presented in ToyKeeper's Anduril and similar firmwares. It's not difficult to use those pins to control something else (small fet and resistor, or it's own 7135 etc), but this will probably not be the main focus of the lume driver boards. Design will be open sourced and you will be free to modify it to your own purposes, though you will need to adapt your own firmware too :).

Thanks, but there are no plans to add a jumper for this specific board, since this board was optimized for the FW3x, and may not fit many other flashlights. Other E-switch hosts will do well with a different board design, perhaps in the future. Design will be open sourced, and anyone will be able to design their own for their custom use :).

Thanks! However, Buck-boost here is no more complicated than a buck, does not use any additional components in this case, and in the targeted operating parameters, only very slightly less efficient for a buck with similar resources (but more efficient than cheaper non-synchronous bucks for example, or cheaper synchronous bucks). Buck-boost is a necessary however, and should always be done in this case as good engineering design. In most scenarios, this BB will be operating in Buck mode, but take for example the scenario when temperatures are sub freezing, battery is a little old, and isn't completely full - this is a use case for me such as when I'm camping... boost mode operation is required then.

It is true that there are many more dedicated buck regulators with small footprint on the market with high current capabilities, however it's still a little bit of a conundrum, as the higher the input current increases, the lower the battery voltage will be, necessitating the need for a boost even more. For higher power requirements, I'll refer to the GXB series of drivers. Finally, there are also many inefficiencies in flashlight systems, especially as flashlights get older (contact resistance being very significant), which Buck-Boost operation will hopefully help mitigate.

Wow, there was longing for a driver capable of both high outputs and high efficiency while having BLF UI. And your driver delivers that. I’m sure there will be many people interested, especially once it’s available in a more standard format as well.

The lack of RGB aux has kept me from buying FW3A - you just added this while improving efficiency. Great. :slight_smile:
Though for me the sense resistor in series with FET is actual misfeature, I have a light that steps down in less than 10 seconds and I could easily go with even less than that. And 200 mV drop at 10A is a lot.

I hope these will be more available than the previous drivers…

This. Nice work again. :beer:

Regarding the sense resistor, true, it's not something I'm too happy about either; perhaps I'll figure out a better solution without adding another big FET (which I guess is possible, but not very elegant..). However, I think considering how fast the FW3A begins stepping down at Turbo, perhaps with a less bright turbo, it may be possible to have the light running at this lower turbo mode, but for a longer period of time. Maybe less wow, but could actually be more useful? I think some very low V_fwd LEDs may benefit from this too. I hear that typically running some Nichia 219 LEDs require a change in the FW3A firmware to run turbo at 50% duty, or for the user to use a lower-drain battery to mitigate.

I’d definitely be interested in a couple of these for my FW3A’s, and also some in other sizes if they get made, SP40, A2S etc.

Some users want dimmer but longer turbos, some want them brighter but shorter. There is no duration that’s good for all. Myself I’m one of the more extreme bunch, for me even 5s would be enough while many consider sub-10-second turbos to be too short. Some don’t want turbo at all and want the highest mode to be so low to last until the battery is empty.

I don’t believe that increasing or decreasing flashlight power output makes it more useful - it will always be more useful to some yet less useful to others.
And that’s why I think that firmware is a better place to manage that. Because it’s easier for users to adjust it to suit their needs. As of now it still requires re-flashing. Programming pads make it easier but it’s still not for everyone.
Maybe one day it will be possible to just configure Turbo output like it is possible to change ramp max now.

May I ask what buck boost IC you are using? When I was looking through Mouser for a design of my own I didn’t find anyone that could do 3A.

Edit: No worries, I found it. I looked at that one but within the first three lines of the datasheet it stated output current 2.5V to 3.3V as 2A. I didn’t read the part under the description where is stated “output currents up to 3A are supported”. I ended up selecting a different IC with somewhat similar ratings for mine.

Great Loneoceans!! It is always a pleasure having news of your work.

Very interested in this for skilhunt H03/04s and also other lamps.

You mentioned ultra low moonlight in the previous thread, how ultra ? current zebralight (~0.07lm) or past zebralight ultra (~0.01lm) ? I’m also curious about the finesse of adjustment of the moonlight levels (a firmware thing ?)

Could charging be integrated ? e.g. for H03/04R with USB magnetic charging port.

If particular attention could be payed to the driver being 100% silent that would be great, electronic noob here but it’s both a design and quality of components thing right ?

Any estimate on how long until this becomes available?

This is look'n real good! Sorry if I missed this, but does it fully support smooth ramping? We usually use a 150 level table. If I understand this, the first 3 amps is regulated, then after that the FET kicks in. This is similar to how we treat a bank of 7135's. The linear regulated driver in the new Noctigon K1 works similar, from what I understand. Up til now, I don't think TK released any of the 1634 Anduril config files. All I know of is 2 Anduril 1634 lights - the D4SV2 and the K1.

Since it's a OSHPark design, I'm hoping we can build up our own, but that depends maybe on how small you are going with parts. Think there are some 0402's there. Not sure what the FW3A driver size is, but scaling up the size should be easy - just keep the parts as-is and expand the outer space - we've done this before.

So are you planning on developing the hwdef and cfg files for Anduril?

Thanks everyone for the feedback; I appreciate it! There is no specific ultra-low moonlight for this driver at the moment, mostly to keep things simple. I hope the lowest mode setting should be low enough; I'll take more measurements soon. However, it will be easy to add in hardware to enable truly very low modes if desired.

I also understand the concern about the flashlight being silent - I didn't notice any issues on my dev board across different power levels, but need a little more evaluation on the form factor board. Typical sounds come from a PFM or burst frequency, which can cause some sounds if it's below ultrasonic and if the devices (typically inductor) vibrates too much. There are no plans to add integrated charging; it can be done, but I think is beyond the scope for this particular design. I really wanted to keep this as simple as possible and to keep the BOM cost down without sacrificing on the engineering.

Not yet; please follow this thread and I'll update when I can, thank you!

The FW3x has roughly 22mm driver cavity. The smallest parts on this board are the 0402 passives and a SOT723 transistor (along with the other QFN parts). It will be challenging to put together if you don't have some sort of magnification, but should be pretty easy with some experience and hot air. At the moment, the lume1 board for the FW3x has only components on one side (with only the spring being on the other side) - hopefully that helps.

Yes it supports smooth ramping. Also, yes the first 3 amps are regulated; the FET is for unregulated Turbo, and you are correct, similar to 7135s. I have a primitive firmware I wrote to test the hardware and it serves as a simple firmware for now, but I am currently working on the config files for Anduril... hopefully I get it right!

Here's a preview of things to come:

For those concerned, I did put some tape around the flashlight so as not to have it scratched by the clamp. I think this PCB colourway works quite well with the FW3C.

This looks great. I am looking forward to hearing the results of your testing.

If there is enough interest, are there any resources accessible to the community for building a batch of fully assembled boards, even if such a group buy would need to be run by someone else?

Yeah. that PCB color is cool!

Thank you thank you thank you. That’s spot on real for me.

I accidentally stuffed 1 of my FW3A’s drivers. This is exactly the replacement driver it needs. And maybe even the rest of my shelf queen FW lights… Great work loneoceans!

Wow another new driver from loneoceans! This one looks very nice, I like the clean design. I wish it is complete soon! Unfortunately I do not have fw3 light yet, but may have to buy one. Do you think it make sense for non-esiwtch flashlight also?

I am so excited to see this board! It will deal with my biggest wish for the FW3A - replacing the linear topology with a buck boost which will improve the efficiency of the light at the most commonly used power levels!
Secondly, aux LED support is awesome!

Keep up the good work!

definitely interested in a few of these :+1:

Hello all,

Just a quick progress update. I'm happy to say that we have first light with the lume1 driver.

For the flashlight, I used a FW3C bead-blasted host previously purchased from Nealsgadgets. The stock 1+7+FET driver was removed from the flashlight. This is quite easy, but does require desoldering wires from the LED MCPCB first. After that, it's a simple procedure with no glue involved.

The lume1 driver was then dropped into place and installed. As a quick note, the FW3 flashlight has a small screw hole drilled through the MCPCB shelf, which acts as a locating pin for the LED MCPCB to prevent it from rotating when the top cover is screwed on. This screw hole had a few sharp edges on the driver side, which I deburred away.

I think this PCB colourway works quite well with the FW3C. The lume1 driver fits! No issues with the inductor hitting the driver cavity, as designed. Programming was done using a HQ pogopin which I put together myself, and an AVRISP2. You can obviously use whatever programmer you want.

The original emitters were SST-20 4000K 90CRI. They seem to be quite popular, but I never really liked the tint of the SST20s, since they are very green especially at the lower power levels, and never really getting better unless driven at Turbo.

I thought this would be a good opportunity to use some of the fairly rare but extremely nice Nichia 219B 9080 4500K 'rose tint' LEDs I had in the past. These were reflowed onto the original MCPCB. The thermal paste was cleaned up and replaced with Arctic Silver 5. Finally, a TurboGlow glow gasket was added between the optic and the LEDs, just for fun.

Fun Question - if you had to build your own flashlight with this driver, which LED would you choose and why?

But does it work?

I'm happy to say that it does!

Anduril Firmware

With the support from ToyKeeper, I was able to get Anduril working on this flashlight. Fortunately ToyKeeper had already done a massive amount of work with Anduril and it had all sorts of capabilities built in to allow easy support of this driver. Since I work on a Windows machine, I used Atmel Studio for the firmware builds and flashing - I had a few requests about this surprisingly, so I thought it would be better to get something going with Atmel Studio, since most people don't have a unix environment.

There's still a little bit of work left to do for me to evaluate all functionalities of Anduril, but all the basics are working, including:

  • Ramping functionality (in this case, 1024 levels of brightness with 10-bit PWM, similar to Noctigon K1), all regulated with Buck Boost up to 3A
  • Turbo with FET
  • RGB aux LEDs verified working (on my dev board; the FW3x does not have a LED aux board for it, yet!)
  • Standard strobe modes all working (candlelight, bicycle, party, tactical, and thunderstorm)
  • Utility functions working (e.g. voltage reading, lockout modes, aux LED control etc)
  • [Update: External temperature sensor is now working and much more accurate, likely no user calibration required]

There are still some functions on the lume1 board not taken advantage of yet including:

  • 4 solder jumper pads are connected to MCU but no firmware written to take advantage of them yet

Once most of the bugs are sorted out, the firmware will be posted up on TK's repository.

Finally, I took a quick video to illustrate some of the basic functionalities so far:

Please keep in mind that this is a video of preliminary hardware and firmware, and work on it is still in progress.

Regardless, this one is shaping up to be my new go-to flashlight already. It's not as bright as my GXB drivers but I think it's a lot more useful! I'm even thinking of swapping the battery out for a Samsung 35E for longer runtimes since the point of this driver is to be more efficienct. I'm also glad to say that there isn't any issue with noise from the switching regulator at all power levels. And finally, it's not obvious in the video but the Nichia 219B tint is absolutely beautiful, redder than blackbody!

Hopefully this will be a driver which people will find useful. Feedback, comments and suggestions much welcomed.

Thanks for reading.