I vote the same… no blinks except for bottom and top of ramp.
I really like how the D4 resets to a fixed mode after you loosen and tighten the tail cap. It came up several pages back, but I think its a great option. I always figured the D4’s fixed level was 100% output on the 7135, and the FET turned off. (I may be wrong about that)
LED tint rises with current. More power = more blue.
With a 2-channel driver, it can only run at two power levels, like 350mA and 10A, which gives it two tints.
In the video, I used PWM to control the brightness, using only one channel at a time, to show the difference in tint. But in actual production firmware I use both channels simultaneously to help blend from one color to the other. Despite attempts to blend though, it still has a distinct tint “elbow” which is visible during use on a smooth-ramping UI.
With three channels, it’s the same thing… only with three colors instead of two. The more distinct power levels it can run at, the closer the curve gets to a true constant current result.
It’s running Anduril, with some features being decided by discussion and votes. Here’s the most recent UI diagram showing most of how it works:
When the PWM is used to get lower levels from a MOSFET, it’s still getting full power just pulsed, so the emitter isn’t in it’s prime efficiency range. With the single 7135 chip the emitter IS inside it’s efficiency range and so it’s more “relaxed”, giving it’s true tint. Pushed harder, even slightly past the 7135 chip level, the MOSFET gives it a more intense color being overdriven.
This is the reason the blink at the junction of where the MOSFET starts coming into play is important, it will greatly influence the output once the MOSFET is utilized. It might even be prudent to have a stop instead of a blink. Ramp the full 7135, then ramp the combined 7135/FET. That would give a more pure use of the efficient lower channel for those wishing to conserve cell life. With the full ramp taking only ~2 seconds to complete, the blink is passed by pretty quickly. Perhaps a lean towards the full 7135 to create a shelf might make it easier to hit the mark, say if 130 in the 0-255 is the mark to hit, perhaps have 5-10 130 levels so the ramp has a plateau in that spot?
In the default configuration, the ramp stops at the highest regulated level and the only way to go higher is by double-clicking for turbo. The user can change this in the config though.
I love the look of this light. Is there any reason that a double-walled design was chosen instead of putting the driver board in the tail of the light? It seems like that would make the construction simpler and cheaper, as well as making it easy to make a version for an 18350 or 18500 cell, but I could be missing something obvious.
The thermal sensor is built into the tiny85, so it needs to be as close to the emitters as possible.
It would need a double-walled design (or an extra wire) either way. The driver needs access to both BAT+ and BAT-, even when the LED power is switched off.
It’s kind of standard for flashlights to have the driver in the head end instead of the tail.