I know. But it won’t cover all my buck-boost needs. 
I was just thinking - a universal driver that can support 1S input and 1S or 2S output would be valuable on its own.
Agro and thefreeman, that's correct. The Lume1 driver in the FW3 was designed to drive single series LED output, which was why the buck-boost topology was chosen.
It's possible to design a more 'general purpose' driver like what you described, but it will place significant compromises in its cost/performance ratio. As a result, I think it makes more practically to have two separate drivers, each optimized for different outputs. This way the cost will be lower for each driver, and the performance higher for each application as well. I just purchased my first FW1A and I'm taking a look at it to see if it is feasible to develop a LumeX1 for the FW3A and drive a 2 or 4S LED (e.g. XHP70 or similar) with it. The tricky part is that the driver cavity is both small and shallow, which will limit the passive components I can use and thus its performance..
How much is the height of the cavity, 2.5mm ? How about using 0.8mm PCB and a spacer to be able fit a 3mm height inductor ?
With a 0.8mm pcb I’d want to pot the cavity for an EDC. I’d be afraid of a bezel first impact stressing/cracking the driver or it’s components.
I didn’t thought of that and indeed that may be a concern.
I’d love to see a high current 6v boost driver for lights like the FW21 Pro/Imalent MS03. Something that can do close to 100w but doesn’t cost too much. Maybe have two separate boards so you can fit larger/more components.
I see.
My thought was: if a driver can support a wide range of LEDs and is easy to reflash with different firmware, then there’s little incentive to swap it.
So for a stock light, using a custom driver format would no longer be a large issue.
But then for a stock light price tends to be critical. Especially that relatively very few lights get modded.
So I think my idea was not really good.
Thanks for your input. 
Reworked Lume1 design for 20mm and up diameter and added RPP .
I ques ill make ” lite” version without DD , temp sensor , with bigger 5.5mm coil , 0603 passives and maybe attiny1616 ;))
This is awesome. thanks!
Hi Quadrupel. What software you used for your reworked version of lume1?
Gerber viewer ,sprint layout and 20 hours of tracing . It looks similar but all pads and tracks retraced manually.
OSH Park
Original
Interesting. I didn’t know that you can import gerber files in Sprint layout and retrace the pcb. That is very cool if you want to change quickly any PCB without to make new one.
Lite version.
PCB 1 Sided, DD Nfet changed to RPP Pfet, all passives 0402 changed to 0603.
Looking at it I don’t see the point of R5, only two resistors and a capacitor are needed for the voltage divider and RC filter, R4, R13 and C6.
R5 connected to R13 and C6
I do not understand circuit around opamp its kind off sensitive and critical component so i left as is. This LTC2066 OpAmp good, but rare and expensive. I wonder if it is possible to change it to something cheap like OPA333
The components between the MCU PWM out and the op-amp inverting input are just needed to divide the voltage from 2.5V(Logic voltage) to 60mV (sense voltage), and filter the PWM signal.
For the choice of op-amp, I imagine Loneoceans chose this one because it’s even more precise (very low off-set voltage) than OPA333, in the hope that it would improve very low outpout performance, in the end it’s not really the case as it seems that the limit is not the off-set voltage but rather noise (probably)
So yes I think the opamp could be replaced with OPA333 or TLV333 which is nearly identical but cheaper. You would need to rewire the enable signal because on the lume1 the op-amp is always powered and is enabled via its enable pin, opa/tlv333 don’t have an enable pin to you would need to enable it on V+, directly powering it from the MCU.
If you add another one capacitor parallel to R4 we will have two stage rc filter. Very good alternative of LTC2066 is MAX9619. It is cheaper and faster compared to OPA333 or LTC2066. I think to used it in some future linear driver.