Yes - this is definitely a good idea!
the MCPCB you use is pretty useless
with 2 seperate buck circuits you can simply run a normal 6V MCPCB and connect the 2 buck in parallel
if the Buck chips support synchronisation you can 180° phase shift them
MF01 4S—>3S simple connect both outputs in parallel to the MCPCB
Original driver has 4 2S—>6S boost converters in parallel but is still a lot less efficient
One of the less obvious benefits is the fact that the losses are proportional to the current squared. As a result, it is very likely that a pair of 2.4 amp drivers will have smaller losses than a single 4.8 amp driver.
2 x 2.4 x 2.4=11.52
4.8 x 4.8 =23.4
So such a ‘twin’ driver would in theory have only about half the losses of a single larger driver. Less heat, less wasted energy, longer run time.
Well that is assuming two identical circuits with identical resistance components. Of course more large inductors and FETs is better but we don’t have unlimited space.
It sounds like heat dissipation is improved with multiple small components compared with single larger components, possibly because of the higher surface area to volume ratio of the small components.
This driver I see on mountain electronic do 5.5A output as bucking driver in 17mm size!
http://www.mtnelectronics.com/index.php?route=product/product&path=67_115&product_id=554
It say it can take 5-18V input and work for 3, 6 and 12V led.
For heating, I believe usually the led is the main heat element, not the driver, so usually flashlight overheat first because of led not driver?
It’s amazing you post this because I had just recently proposed this thought to someone earlier, and this helps confirm my thought.
It looks like it’s DTP, in which case, not useless.
Thanks for sharing this idea svitlo!
There may be other advantages to a MCPCB like this… for a mounted light, you could make the light “directional” by dimming one channel and brightening the other - like for a stationary vehicle light that still “turns” with the vehicle, or “points” up or down with the terrain… possibly with a cross fader or stacked pot.
This driver is implemented on MAX16820
I tried these drivers. At 5A, they work only for a very short time. All power elements are very hot.
In this version of 17mm, this driver can be operated with currents up to 2.5A
In my version do not use specialized chips with the ability to synchronize. No need to shift the phase 180 ° .
I used two simple non-complex Buck drivers. Project price matters
I have master and slaved two buck drivers together before and tied the outputs. It doubled the power out and doubled the current draw. Both ran in a reasonable range.
It worked like a charm and both where controlled using one mcu. No reason why they couldn’t be used on a single designed board if the host had the room.
Where to buy your driver svitlo?
I want to use the two channels to strobe each half of the XHP70.2 alternately. I'd say, something like 10 Hz, and less than 50% duty cycle, 180° out of phase.
Even more fun if you use two separate strobe frequencies 
think about the thermal stresss in the LED if only one side heating up, also would affect the beam negatively, your directional idea would not work like you think, you can make that with 2 indipendant LEDs and optics
there is no benefit of such a MCPCB it generates only disadvantages
Buying one driver does not make sense.
It needs to be bought with a LED.
Available XHP70.2 90CRI 2channels
Interesting idea. And why the frequency of 10 Hz?
This seem like very good assembly product, but I still am confuse of usefulness.
Driver is small, this mean flashlight is small, this mean thermal capacity and heat sink is small. Because of this, at any ‘high’ drive power, led will heat up fast the most, before thermal regulation is required, then power is throttle to low. Therefore in this use case, it seem to me like there is no advantage of use two separate buck driver if the reason is to improve efficiency, because most energy loss is in led, and run time on highest mode is likely short, at most a few minute.
I only understand this concept if flashlight is bigger, and allow big sustain run time at high power. In that case, then dual buck make sense because you can use cheaper component and parallel them together. However, I don’t think efficiency is actually improve because like EasyB say, you likely use smaller cheaper component if you make two, but certain cost will be reduce.
For mcpcb, I think lexel is right. It is not advisable to run xhp70 half at a time due to thermal stress. This mean the buck need to run at same time and same power, so better to just parallel them and use regular 6V mcpcb. Less wire to led also.
However, it look like you put a lot of effort into this and I like the double stack pcb design. Good job to svitlo!
They do make color-mixing LEDs with four separate dies on one chip that can be individually activated/dimmed; sure they’re designed as such but for something that’s momentary, like a few seconds, I don’t see it as a big deal… but I won’t blow my own $20 LED to find out…
There are also color-mixing optics - hybrid reflector/TIR - tat work with color-mixing chips to give even output regardless of intensity of each die on the chip as well…
Small diameter driver was designed for diving flashlight.
Underwater hunting is popular with us, and powerful light is needed for underwater hunting.
but XHP70.2 are usually driven pretty hard, have seen enough XHP70.2 with 2 dead dies in the past, the flip chip seem to be a bit fragile compared to first generation
on the other hand I can build 20mm diameter drivers with up to 8A output and 95% efficiency at 6A, so why making this big driver that can do only 4A?