Ok, so I’ve found this post from Richard, which answers 1 of my questions…
“I don’t think you’re missing too much! This is not the optimum solution, it is one that was designed to work with existing drivers and still be able to fit into lights with small 17mm driver compartments. With IMR 18350 cells I haven’t seen much any thermal throttling when driving an MT-G2 at around 4.5A when actually installed in a host. With two 20Rs sitting in the open air you see the 7135s begin to thermally regulate (cut current) at less than 30 seconds. The closer the input voltage is to output voltage (look to emitter vF curve) the less heat the 7135s will have to shed, this means that batteries with a little more voltage sag may actually be better for this setup.”
Old-Lumens, regarding my heat sinking question, I’ve found this post from you which answers that question…
“After doing many of these, I feel that the 7135 chips need heat sinking to keep them from getting too hot and dropping out. As to the best way to heat sink them? I do not know. Fujik isn’t enough. It needs to be something like direct to metal, so I use copper and it seems to help, as the light does not dim near as quickly. When I say quickly, I mean that with no heat sinking, pushing the MT-G2 to 9 amps, causes dimming in seconds, where with copper soldered to the outside tabs of the 7135 chips, it lasts for a couple minutes before dimming.”