I just gave it a try. KR1AA at default step 5/7 in advanced mode. I don’t have fully charged cells right now, but … here’s what I got:
- 2025 Wurkkos 14500 @ 3.79 V: 287 lm
- ~2015 Eneloop @ 1.28 V: 286 lm
That’s about 4.3% lower than Zak’s estimate of “~300 lm”, but within the expected ~7% error margin manufacturers use for binning LEDs. Scaling down his numbers by 4.3% changes the efficiency estimate from 148.8 lm/W to 142.3 lm/W.
BUT
That was in a lumen tube. And I think Zak uses a box.
So next I tried measuring in my lumen box. And there… I got 311 lumens, which is about 3.6% higher than Zak’s estimate, and would be about 154 lm/W.
So, based on that, I think his measurements were probably pretty good. They’re within a few percent of what I got, moments after calibrating both of my lumen devices to a maukka reference light.