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

Strange bout the big lights better with tailswitch - heavy lights are easy one handed operation with a properly placed side switch - tail is awkward at best, worse requires 2 hands.

Ooops, by AUX LED I mean same thing as switch LED. Again, 'AUX' was a term we used to indicate an auxiliary LED, where-ever it's mounted.

A side switch LED can be seen with the light off, whether it's standing on it's head or tail.

Totally agree bout the deep carry clip - I just don't get it...

Let me clarify that the ‘long’ was still within the realm of EDC lights. So things like a Convoy M2 or S21A at the largest. FW1A is the lower limit in length for me to distinctly prefer a tailswitch. Anything longer than a “tactical” ~150mm or fatter than a single cell is back to side switch for sure. I’d really like a lighted side switch with charge indicator and adjustable beacon brightness.

And yes, “AUX” technically could mean any location. Thanks for clearing up what you meant.

FET/DD may not get you an “A” on your EE project :smiley: , but it’s inexpensive, compact, and reliable even for very high power builds (as long as the emitter/battery combo is chosen well). IMO that makes it an elegant solution.

I mostly agree…though for me to call it elegant it should have a linear mode. :wink:

loneoceans, would a future design for about 6A regulated output with the buck-boost be possible? I know you mentioned size constraints, and for this version it is single-sided as well. I mainly ask because of a few popular emitters (Luxeon V, SST-40, white flats) reaching peak output around there, but still having pretty low forward voltage at that point. So they don’t make sense with a full cell on a FET, but by the time the cell gets low you’ve lost your turbo output.

For higher currents we have only one alternative at that time which can be powered from single cell and thats is only from Analog Devices, LTC3119. It is capable of around of 5A from 6V. There is no others alternatives of which I know at moment with integrated mosfet switches. Also LTC3119 is very expensive around 17$ for one IC. For over 100pcs is it about 10$. Another way is to be used buck-boost PWM controller with external Mosfets but at moment it is possible only with bigger PCBs and design will be more complicated. Maybe another possible solution in the future it to be used boost-buck topology and the transistor to be driven directly from MCU. I found on fonarevka.ru similar boost LED drivers driven from MCU without external controller. In that case the disadvantage will be low working frequency which means higher inductor.
For reference something like these: LTC7813 Datasheet and Product Info | Analog Devices

Could you please link to it?

http://forum.fonarevka.ru/showthread.php?t=20143
http://forum.fonarevka.ru/showthread.php?t=20348

They used very clever design to drive MOSFET with analog switch IC which act as driver.

Aslo here another great LED driver from Tamagotchi but it is not open source which I think is also boost driver:
http://forum.fonarevka.ru/showthread.php?t=31892

Thanks icpart. :slight_smile:

That's not a problem; for larger boards or two sided, should be reasonable to build a 10A buck-boost or similar; any more and I think it's probably time to use 2S input. I don't have plans to make one yet, though, focusing on this driver first and we'll see how it goes.

icpart and Agro, yes any larger BB will use external switches. I like the idea those drivers have in using the MCU do the main work, though using an Attiny is not quite optimal for various reasons, including the ones you pointed out... but it can be improved with other MCUs. In fact, I've built a bunch of switchmode systems using a PSoC. The PSoC5 especially with its analog capability is pretty amazing. Then again, while I always had the thought to use 32bit MCUs or PSoC for flashlights, they're really a bit of an overkill and there's a bit too much software development time required than I am able to spend right now. The use of the analog-switch as a gate driver seems to me like a clever use of parts to reduce cost. The switch looks like it'll work just fine for this application with a marginally lower cost than a gate driver, and generally low Iq... though I'm not sure why they didn't just use a co-packed fet pair. Together with just a basic Attiny, overall quite an efficient solution in terms of parts count.

How many amps it’s capable of from 6V is some what irrelevant for single cell designs. How much is it capable of in buck mode from say 3.7V or 4V? I took a quick look at the datasheet and couldn’t find a quick answer. I’m not sure how good of an alternative it really is for single cell designs, but I probably missed something in the datasheet. Can anyone point it out?

I didn’t find anything specific either. A page 30 in the datasheet there’s a graph for voltage in vs current out at 12v out. It looks like with a voltage out close to the same as the voltage in gives about 5 amps pulsed, 4.2 amps continous. In a flashlight the current out would probably depend on the led vf at that current compared with the cell voltage of say 3.8v. As long as the cell voltage is above the led vf at 5 amps it should buck at 5 amps pulsed, about 4.2 continous from what it looks like to me. I dont find any graph or info that relates to that besides the one on page 30 with 12v output.

I found one more reference at page 28 in the bottom graph. For Vout it says 3.3v at 5 amps, >4v.
Its probably going to be led vf specific vs cell voltage.

You certainly found more than me. Thanks for the info!

The LTC3119 is a very nice part and was initially on the list for consideration for this driver, but it's unfortunately extremely expensive as icpart pointed out.

For a reliable driver, I think it's important to look at the absolute limits of the system to ensure it stays within regulation, either via software or hardware. From a hardware perspective, the greatest stress would be when the cell is at highest load (greatest output) at low capacity or cold, with a high V_fwd LED - in that scenario, the driver would need to operate in boost mode. It's possible to design the flashlight as a system with a designated battery and LED so the driver runs only in buck mode, but then it'd make sense to just develop a buck-only driver, of which there seem to be many on sale with very good power capabilities. I wanted the lume1 driver to be fairly flexible in the choice of batteries and LEDs and to be able to operate well near the limits, which helped guide the design choices. As mentioned, if I was developing one for a larger flashlight, I'd go for external FETs. Maybe this will be done in the future for a suitable candidate... maybe a 21700 flashlight.

As a side note, I'm hoping to send out a small number proto drivers to forum members to test since I may have missed something and it would be good to have some fresh eyes to take a look ... Then once we have a stable design, I'll make a poll to see interest for a run of boards. I like the idea of using a Google Sheets - this will come later. Meanwhile, anyone who has experience with electronics, the Anduril FW, and is able to do their own measurements, EE debugging, firmware flashing and compilation, and would like to help beta test, do drop me a PM and we can work something out. Meanwhile, I'm in a process of doing some full characterization and comparison of the system.

Thanks everyone and stay safe in the emerging worldwide health situation!

Subbed

Of course, completely understood. I just got excited. It hurts me seeing some of the newer LEDs thrown on FETs just to get past ~3A

Subbed.

Hello all,

I hope everyone is doing well and staying safe in the worldwide outbreak.

Here's a quick update with some developments for the lume1 project. I received boards for Aux LEDs for the FW3A flashlight and their variants. Unlike regular AUX boards, this one takes advantage of the full RGB capability of Anduril, allowing for a host of different features and capabilities.

Here's the LED board. A total of 18 LEDs for board, 6 of each R, G and B LEDs.

I had some other colour LEDs so I populated them as follows - pink as red, warm-white as green, and cyan as blue. The LED board is a little thick in this iteration, and 0.8mm PCBs should be better, but this works as well. LEDs used were 0603 LEDs on large 0402 pads. I had ordered 0402 LEDs but they haven't yet arrived.

AUX LED control is the same as the one implemented in Anduril for 1634 MCUs such as in the Emisar D4V2 and the Noctigon K1. There are several different modes such as different colours (red, green blue, yellow, cyan, magenta, white), as well as 'rainbow' and battery voltage display. There is also high and low brightness control.

Recommend LED wiring is 24-20AWG depending on what you have. For the LED wires, I used 30AWG.

After soldering on the LED wires to the AUX board above.

And finally, here's the Aux board in action, compared to a glow-gasket with my FW21. In this photo, I had previously used a green-yellow LED for the green channel. I since replaced that with warm white since I thought it came out better.

More to come soon.

Stay safe and healthy everyone!

It would be perfect, if you make variant to work with different color leds. As example, we could repower Sofirn IF25 with efficient Anduril driver. And any such flashlights in future.

Anduril has support for 3 aux LED channels, typically named R, G and B, but you can use whatever LED colour you like. Or am I missing something you're saying?