I think you learn a lot about the PID effect by figuring out the basic efficiency of the emitter-battery combination, and then looking at the times for the highest outputs.
First, run some very rough numbers for the mid-range outputs x times, to see how many lumen-hours this emitter and battery combo is generally capable of:
100 x 12.5 = 1250 lumen-hours
34 x 36 = 1224 lumen-hours
19 x 2.6 days x 24 hrs = 1185 lumen-hours
10.4 x 4.6 days x 24 hrs = 1148
There will be some drop-off in efficiency on the low-end and the high-end, but the basic picture is pretty clear: this emitter-battery combo puts out around 1200 lumen-hours in its most efficient band.
Use that figure to calculate a maximum output at each of the high outputs:
1200 lumen-hours/ 1500 lumens = 0.8 hrs
1200 lumen-hours/ 666 lumens = 1.8 hrs
1200 lumen-hours/ 429 lumens = 2.8 hrs
1200 lumen-hours/ 270 lumens = 4.4 hrs
What that tells me is that in the 270 and 429 settings, you really will get that output throughout nearly all of the advertised run-time, without the output being reduced by PID.
But for the top two outputs (i.e. 1500 and 666), most of the advertised run-time is going to be a much lower levels. PID will kick in and throttle it down to something in the 400s or so for the vast majority of the run-time.
- High: H1 1500 Lm (PID, approx. 2.3 hr) or H2 666 Lm (PID, approx. 2.7 hrs) / 429 Lm (PID, approx. 2.9 hrs) / 270 Lm (4.1 hrs)
That's why the times for 1500 and 666 don't look that much shorter than the times for 429: they are going to be putting out 429 (more or less) for most of the period, and putting out their top lumens only briefly at the beginning.