Sort of both. Higher modes tend to be cooler and whiter, while lower modes tend to be warmer and greener. The tint ramping can change the warm/cool balance, but it can’t change the white/colorful balance.
On this diagram, lower current levels tend to make the tint go to the right (warmer) and away from center (farther from the dotted line). Higher current generally makes tint go left (cooler) and toward the dotted line. The tint ramping can adjust things on a left/right axis but not an up/down axis.
For example of how current changes tint, maukka measured the Olight S1 Mini’s tint in different modes. At lower modes, it gets warmer and greener:
To minimize any warm/green shift, the lantern could use PWM at full power to keep the tint as close to white as possible. The lantern could also potentially use 3 emitters per channel instead of 4, to increase the current going to each emitter.
For an example, let’s say the lantern can make 600 lumens and it’s running at 33% brightness on a visually-linear scale. That works out to about 33 lumens total, at ramp level 50 of 150. With a 4000K tint it uses all 8 emitters, at about 4 lm each.
- A constant current driver then delivers only about 10 mA to each emitter. On the Olight chart above, this would be somewhere between “moon” and “low”.
- A PWM driver instead delivers 350mA to each emitter, producing about 150 lm per emitter, but only for a short time. Visual brightness looks the same, but the tint coordinate would be somewhere between “mid” and “high” on the chart above.
However, the chart shown is for a different emitter, and I don’t have data on the tint shift effect on LH351D. It may behave differently. Most LEDs follow the same general pattern though, where the lowest levels are the least white. When we increase efficiency by using those low levels, we also reduce the quality of light coming out.
So the question is whether to prioritize tint or efficiency.
Since the lantern is using high-CRI emitters, I’ve been assuming that tint is the priority, so an AMC7135-based driver would probably make the most sense for that purpose. But if efficiency is more of a concern, it could definitely use a linear FET driver with medium-CRI emitters. It’ll make significantly more lumens per Watt, delivering higher output and longer runtimes.