You could use a shark/remora combo from the sandwich shoppe. I may try that or just run the XPG parallel with AMC sandwich. Both would obviously work out. I am not sure which would be more efficiant maybe another member that knows the formulas for figuring that can chime in.
The tricky part with a Shark is that it is 0.75" diam. The base of the Moddoo heat sink is 0.75" IIRC. If you make the base larger to fit a bigger driver cavity for the Shark, then you start to run into problems with the heat sink fitting into the flashlight (6P?) neck.
Other Shark considerations:
- the Shark Buck needs about 2V of voltage headroom to run in full regulation. Thus a parallel XP-G triple could have Vf~3.2V, and you'd need over 5V input. That means 2xLi-ion voltage or equivalent.
- the Shark boost delivers a max of 1A, so you'd probably want to wire the triple in series. If we assume a total Vf of 9.6V for the 3 XP-Gs in series, a driver efficiency of 80%, and an input voltage of 5V (e.g., 2x123A), then the tail current is going to be about
Ibatt ~ (9.6V*1A)/(0.8*5V) = 2.4A
So actually, the input voltage will probably be less than 5V. And even if it were 5V under load, 2.4A draw from 123As is a lot. So again, you might want to use 2xLi-ion. And if you use 16340s, make those IMRs.
An AMC sandwich can be very efficient. LDO regulator efficiency is essentially Vf/Vin.
The downside with the AMC approach (presumably 8x7135) is that you ideally should drive the light with 1xLi-ion -- probably an IMR18650 or IMR26500 would be best solution.
In Moddoo's version 1 product, he used two parallel wired SOB1500s to get effectively an SOB3000. The 14mm sandwich fits. You can use potting epoxy to create the necessary mechanical support structure to hold the smaller diam driver sandwich in your heat sink (or machine one specifically for 14mm instead of 17mm).