thanks for the nice review!
sorry in case i missed it or if this is a stupid question:
how to switch from smooth ramping to stepped modes?
or isn’t that possible at the m44 which differs a little from other anduril lights?
many thanks in advance for clearing it up!
The short answer: 6C
The long answer: Older models used 3C, but on multi-channel lights, 3C changes the channel. So ramp shape (smooth/stepped) changed to 6C. However, if you turn off all the channels except one, 3C reverts to its original function of changing the ramp shape instead of the channel. The 3C action tries to do whatever is most relevant to the specific light being used, depending on that light’s features and configuration… and channel changes take priority over ramp shape changes. The idea is that, if someone bothers to buy a multi-channel light, they probably want those channels to be relatively easy to use… so it goes on a relatively prominent button mapping.
Similar story for 3H. It adjusts the current channel’s “tint” when possible, and when not, it reverts to the older mapping of momentary turbo. However, 4H works for momentary turbo in either case.
Does anybody know if the M44 measures in the high risk part of the Opple “PWM” graph at all outputs and single and blended channels? Or is it limited to lower “moon” levels only?
Opple LM4, 2700k/5700k 519A M44.
2700k moon:
5700k/2700k 50/50 moon:
2700k mid-ramp:
5700k/2700k 50/50 mid-ramp:
2700k turbo:
5700k/2700k 50/50 turbo:
Thank you, exactly what I was looking for!
Because of the M44, I’ve had an item on my todo list to go through the regulated driver hwdefs and double the pulse frequency when possible. Lights had been using phase-correct PWM for so long, I didn’t even think about using fast PWM once it became an option.
Phase-correct is necessary on lights with PWM only, and no enable pin to turn the regulator on/off, because fast PWM never fully shuts off. The phase-correct style also greatly reduces sensitivity to minor variations in hardware, making low-mode outputs more consistent and more tolerant of different hardware. Like, 7135 chips don’t have a super quick rise time. Underclocking the MCU at low levels also helps, by slowing down the pulses even further.
But for a regulator like the M44 uses, pulses can be much shorter without causing issues, and there’s a separate pin to turn it off. So it could easily double the PWM rate without causing problems. This would shift the Opple “risk” dot down and to the right.
However, Opple’s measurement omits the 3rd axis of the graph, and that axis is pretty important. Total brightness is a big factor in what kind of oscillations look okay. At low brightness, bigger / slower oscillations are fine… while at higher brightness, that stuff is more visible. So I think Opple makes the moon ripple seem worse than it is.
Interesting. Would that improve PWM on the LT1S Pro? Part of the ramp of mine is unusable due to noise.
Got any good sources on the flicker assessment standard? I don’t fully understand how it works.
The LT1S Pro has slow pulses for a different reason. The regulator it uses on the red channel misbehaves when the frequency is too fast.
For the M44 though, I rewrote how it works and it should greatly reduce flicker. Could you check it with the latest changes?
I know this is a little off-topic, but there is a ton of great information about PWM in this thread, and I believe this applies to the M44 as well.
With many of my Hanklights (maybe even all of them), I am able to measure flicker (whether it’s PWM or not) when not on one of the highest 2 or 3 brightness levels; however, with my Fireflylite E12R and E07X Pro (both Anduril lights), there is no flicker at all – even on the absolute lowest moonlight brightnesses. Why is that? Do those lights just have better drivers?
I imagine you mean the newer version with low moonlight since the older version flickers at level 1-2 (5th cause)
The newer version controls the output with the Attiny1616’s DAC, as long as the DAC level is above 2 (pulsing at 2 and below), there is no flickering due to the control signal (4th cause, which is present on Emisar’s driver due to dynamic PWM).
The 5th cause is eliminated thanks to the multiple sense resistors.
But as a side effect, there should be measurable (e.g opple) flicker at very low output (not necessarily visible if a load resistor has been added), if there isn’t then the buck regulator is probably operating at constant frequency, in that case the minimum current draw should be in the 10s of mA due to the larger power consumption of the regulator, as oposed to ~1.5mA. It would be interesting to check the current draw at min output.
Thanks for the quick reply. A lot of that went over my head, but really the intention of my last reply was more rhetorical than actually wanting to truly understand it from an engineering perspective. As a consumer, I just know the other two Anduril lights I have don’t flicker (or at least don’t have any measurable flicker) at any brightness level, and that is ideal for all flashlights. My Acebeam lights do not flicker as well. (For the most part, ZebraLight is the same, but there are outliers there, unfortunately.) After I realized all of this, I haven’t purchased any Emisar/Noctigon lights, and it is the main reason I didn’t pull the trigger on the M44.
I ended up throwing my LT1S Pro in a storage bin in my closet because of the PWM (and/or flicker) issues, and although ToyKeeper did a truly excellent job of explaining it, I just feel that there shouldn’t be any measurable flicker (or PWM) in modern flashlights. I apologize if I am missing something or come off as rude, but at the end of the day, some of my flashlights flicker and some don’t, and that’s all I can be sure of.
I tested the new ramp today, and the PWM is significantly improved. I’d say at levels 1-4 it’s visible, and on the very bottom of the ramp (levels 1-2) then channel ramping still has a low range of perceptible adjustment, but if you increase your floor to level 4, it ramps much more smoothly than it used to, and there is no visible PWM to my eyes.
Great work @ToyKeeper - for anyone with an M44, I’d recommend updating, and if anyone was on the fence about it due to the PWM at the lows, it is significantly improved now.
Moon is definitely significantly lower as well - I’ll take some beamshots when it’s dark.
@kerplunk the LT1S Pro has a new ramp as well now - did you try updating it?
I haven’t doubled the PWM speed on LT1S yet, but it should be possible on the white-only channels without any significant downsides.
Red, however… behaves badly enough at higher speeds that I plan to keep using the slower pulses when red is on.
I wonder if Hank will eventually do a similar light with 3 X 21700 (or 2 X 26800.) This model looks awfully short for my XXL hands, and I’d rather have the improved runtime over the 18650s as well as some extra length.
46950 
I’ve EDCd the M44 on my belt today to try and find the limit. It worked surprisingly well, even for the car ride. It got some puzzled looks (isn’t that the point anyway?) and is tough to hold in my mouth compared to the D4V2. I’m hestitant to attempt MF01 next.
I should think so!
Managed though!
Yeah, once a light gets past 1x18650 it’s not really ‘EDC’ for me. I could put it in the work backpack but that’s not an ‘everyday’ trek and the backpack doesn’t always travel with me in the building - i.e. my DM1.12 will sit awkwardly in a shirt pocket for walking the mutts but it’s a task light at that point in time.
I find the D4K works well as an EDC light, but that’s about the biggest I’m willing to carry around.
AA / 14500 makes the best overall hosts IMO, with 16340 being a close second, but these both have pretty low capacity so I have to use lower brightness levels and recharge them more often.
18650 seems like a nice sweet spot for size and shape and capacity, as long as the host is compact.
21700 is do-able if the host is small enough, and the extra capacity is really nice.
The M44 seems much too big and heavy for EDC. So mine doesn’t get used much, aside from development purposes. But in my case that’s fine… that’s why I have it.
