Yeah, I got the light to behave thermally, so I think it should be almost ready to buy. Here’s an example of a thermal test with the maximum temperature set pretty low to make things harder for the thermal algorithm:
Given the style of power channels used, this result is pretty close to optimal.
Which emitter is it using? The sustainable lumens seems low if it is xpl hi or sst 20 70cri. The heat shedding of this version is significantly improved by TA from the original MF01 so should do better than this.
I think this is SST-20 4000K 95CRI, which runs hotter and less bright than the lower-CRI versions. Just checked again, and it’s definitely around 4000K, so it’s not the 5000K 70CRI version.
This high-CRI version is only supposed to make about 10,000 or 11,000 lumens, but I measured it at 13,900 lm at start with a set of just-charged 30Q cells.
This was measured with an integrating tube, so it’s not as accurate as a sphere… but it’s pretty close. I used some biased diffuser sheets to improve the integration qualities, then calibrated it to match reference lights from maukka. The sheets force the light to spread out and bounce around inside, so it’s less favorable to throwy lights. It also acts as sort of like one-way valves, which reduces the impact of the entry point and initial bend, so it gets more consistent readings for a wider variety of lights. But it’s still not as good as a sphere.
Anyway, I think it actually could make 14,000 lumens with some high-amp cells like 25R. And more, when using lower-CRI emitters. Just not for very long, because it gets hot fast.