Here's the PID hunting in action when the light is not cooled.
Having just received my H600Fc mk IV and performed some measurements, I thought others might be interested to see how Zebralight appears to have tuned their firmware.
At my ambient temperature of 17C (63F) I did not experience any PID throttling on the lowest PID level (the nominally 562 lumen level), so to check for hunting I had to use the second highest level (nominally 980 lumen). I only had an older NCR18650B whose capacity is down to about 2.8Ah, so my times are a bit shorter than others have measured, but still quite long enough to experience throttling. I have no way to measure lumens, so I have normalised to the starting brightness as 100%:
Here is more detail of the first 15 minutes:
Note that there no longer seems to be the dramatic hunting. There is a small amount of ringing between 4 and 11 minutes, but I don't find that distracting or annoying (indeed I don't really notice it unless I am looking for it).
The regulation quality of my H600Fc is remarkable. It draws essentially the same power regardless of the battery voltage. Here is a graph (for the nominally 296 lumen level) showing the power vs battery voltage:
One implication of this is that, if we know the total energy that a battery can deliver at a given power draw, we can calculate the runtime based simply on the power. I have made some guesstimates (based on
https://electricbike.com/forum/foru...asonic-ncr18650ga-lg-mj1-samsung-inr18650-35e) of the energy that a Sanyo NCR18650GA can deliver at the approximate power drawn for each level, and measured the actual power drawn by my H600Fc at each level, and arrived at the following estimates:
Note that for the PID levels these numbers assume perfect cooling; in practice the light will last significantly longer than this because it will throttle down to lower power levels when it gets too hot.