Let's see how it does on my estimated lumen scale:
Overall, my lumen estimates are pretty close to Zebralights specs. The lowest output modes seem lower than the specs would suggest, but that could be due in part to sample variation.
As always, note my lumen estimation method is just that – an estimation based on the calibration of my lightbox. As such, don't get caught up on the
absolute difference between lights or levels, as I can't guarantee absolute accuracy. Focus instead on the
relative differences, as that is internally consistent across all my reviews.
Output/Runtime Graphs:
Let's start with a comparison of the Hi modes, under my standard testing conditions (i.e., a cooling fan is used)
All levels show a clear step-down pattern as the batteries near exhaustion. This is convenient, as it means you won't be stranded in the dark without warning.
But you can also see some evidence of the PID circuit feature on the max mode (Hi1). Although the overall pattern is similar to the drop-off seen on many lights as the batteries/circuit head up, there are actually a lot of very finely discriminated steps in output happening. This will not be apparent to you in practice (i.e., they are too subtle to see), but my lightbox can indeed detect the precise thermal regulation introduced by the PID.
You can read up more about PID (Proportional-Integral-Differential) controllers online, but basically they are a more sophisticated mechanism to regulate circuit function in regards to temperature than the typical step-down features we are used to with flashlights. The controller has to be specifically customized to the build in question, and can be thought of a refinement/replacement of the typical current-control regulation we are used to.
According to Zebralight, both the Hi1 and H2A levels use PID. This presumably explains why they are unable to provide typical runtime estimates (i.e., depends a lot on ambient temperature conditions). Note that I don't see any thermal step-down on the Hi2B runtime above - likely due to my use of cooling fan during testing.
To explore this in more detail, here is a close-up of the H1 and H2A modes with and without a cooling fan:
Now you can really see the PID in action. On the Hi1 mode, there's no difference in the runtime pattern due to cooling over the first 5 mins or so. Beyond that point though, the non-cooled runtime shows a progressive drop in output – down to below the Hi2A level. Of course, this also means the light can run for a lot longer, as you are not draining the cells as quickly.
A similar (though less pronounced) effect of the PID can also be seen on the Hi2A run. Output again remains initially consistent between cooling and non-cooling runs, but eventually becomes divergent. By ~25 mins into the non-cooled run, there is a definite drop off in output, although it is not as pronounced as the Hi1 (in relative terms).
That said, the sustained output level of both Hi1 and Hi2A is remarkably consistent once step-down occurs. At the 30 min mark, I would estimate the PID-reduced output on the non-cooled runs to be ~600 lumens on both levels. By 80 mins, it has consistently dropped to ~560 estimated lumens.
Once the pronounced step-downs occur (due to declining battery voltage), performance on the non-cooled runs looks the same as the cooled runs.
Here's a comparison of my standard AW 2200mAh cells to a NCR18650A cell (Xtar 3100mAh)
Finally, let's do a brief examination of surface temperature relative to output. This was done by simultaneously logging data from a temperature probe taped to the head (on the opposite side from the switch, near the base of the head). There is no cooling applied, and I am using the same Xtar 3100mAh cell as above:
As you can see, the PID keeps the temperature within a remarkably consistent range. Again, don't worry about the absolute temp value (that just reflects the specific probe placement).
Let's see how the fan-cooled SC600-II L2 compares to the competition (all on AW protected 2200mAh):