This is what you should expect. Some emitters are more efficient than others in the same bin. Since we calibrate to lumen output, this will make runtime vary, however this list is a minimum run time.
Got it, looks like my numbers are generally within spec. Has there been any update to the firmware that would have changed the way the stepdowns occur? Perhaps I was overly optimistic with my expectations of the runtime...I understand the choice to calibrate to output vs. calibrating to power draw, but if in the future I do manage to get my hands on another rotary HDS, would you be able and willing to calibrate/flash it to have constant draw per respective rotary level with no stepdowns unless low voltage, i.e. 2.8v, then quickly steps through the levels down to whatever level drives at 50 lumens, then lower until it cuts off at 2.5v?
**added for clarity:
16340 900mAh @4.2v,
Level 23 for N min @0.8A, cell to 2.8V
Stepdown--
Level 22 For 30sec @0.6A,
Stepdown--
Level 20 for 30sec @0.4A,
Stepdown--
"" "" "" "" ""
Then Level X @50 Lumens for x minutes, cell voltage 2.6V,
Stepdown--
Limp mode, until cell cuts at 2.5V
Same logical process for any of the lower modes, except they would last longer on the initial mode until stepping down, where anything arriving at the 2.8V threshold would immediately start stepping down, and 2.8>x>2.6v starts at 50 lumens, and 2.6>x>2.5 starts in limp mode, and 2.5v doesnt activate at all.
For my purposes, the above calibration would be amazing...my brain just works better with known power outputs rather than lumens at unknown power outputs...
Also, I'm translating ideal/practical numbers with real world runtimes of another manufacturer...
16340 ~=3/2V of 2xNiMH serial
16340 @900mAh~= 47% mAh(round to 50% for calculation) of 2nimh
In this other light, 80 lumens for flat 6hrs and change before steep drop and cell failure. Assuming typical HDS is 250 lumens, use 240 lumens for easier calculations, bar inefficiencies for now, that gives 1/3 the runtime, so the 2x NiMH would give 2 hours at 240lm. 2hrs ×3/2 gives 3, then × 0.5 gives 1.5, multiply by 0.85 for losses in circuit, you get 1.275 hours, or roughly 1h15m on max till sudden failure.
If the above ramping is applied, we could have an hour at or near max, then extend that remaining 15m of max into lower, longer running modes for the same total runtime as the existing rating, and the same tactical runtime with more max (or user selected level).
The stepdowns of the 2015 model I had just gave the impression of an inefficient driver...NO slight intended, just feedback delivered 10 years late...
If the user can manually select the level using the rotary ring, there would be no reason to introduce an arbitrary stepdown (besides cell safety) just to extend runtime...