That’s thanks to RMM and JonnyC, they cooked this up. RMM posts about the motivation over here: (#1447) Oshpark Projects
I haven’t played around with it. I don’t think that it will help at all with that problem. That’s really part of what the dual PWM output is intended to help with I think.
I should have at least one driver soon with dual PWM, and am planning to add the ability to control PWM for each channel for each mode. This should enable things like 1x7135 at full power for a more efficient medium mode, and stable moon/low/med on 7135 while still keeping high/turbo on a more-powerful FET.
For drivers with 8x7135 (one split off for moon), it could still use all 8 for the highest mode, but should be able to use just 1 for the lowest few modes.
Richard flew only a select few of us out to his compound to have a discussion, complete with cigars and strippers. Actually Richard came up with this idea, asked if I could do it, so I just added it.
I haven't done any testing to see if you can get lower lows on 7135's with phase correct and if they are consistent across variations of 7135's, but I do know you can use lower values to get the same output compared to fast PWM. To be honest I don't do any tweaking of moonlight modes though, as it's not a feature I ever add to my lights.
Yes, you can. At least, on the very small few I’ve tried it on.
In general, it’s not necessarily lower, but just different. Like, on one 7135-based driver, I found the following available modes:
So, I went with phase PWM=4 for moon mode. The 0.1lm mode was just a little too dim, and the 1.1lm mode was too bright.
This is something which needs calibration for each individual light though, or at least each driver+LED combination. The lowest and most ideal level for moon varies with the hardware.
ToyKeeper, can you elaborate on how many 7135s the driver in your example had?
My favourite Med mode gradually went down from 750mA until I’m at 350 now… It’s PWM 6/11/40/255 which (at 18kHz/fastPWM) gives me measured 1/36/342/2830 mA. A very nice mode spacing. This screams for
1x7135 pwm255 + 7x7135 pwm255
1x7135 pwm255
1x7135 pwm~low
1x7135 pwm~very-low
Imagine how looking at the new STAR firmware with ALT_PWM made me smile. I thought about how this might work (using mode_idx for voltage stepdown) but I’m sure you’ll be quicker and your results are more reliable.
By the way, I still love the bi-directional mode change with short and medium press. I’ll never ever use anything else anymore. It’s fantastic to change direction without having to go through all modes, especially the high ones.
I should have guessed…
Moon has truly grown on me. It’s nice for a very dark environment and I use it all the time to look after the kids when they’re asleep or woke up at night.
I’m very sure her results are for 8x7135.
-This pretty much is the same that I measured (and saw) in my little writeup about PWM values at different frequencies
Especially the values where the lights start to light up.
But I’m not sure whether this is a matter of fastPWM vs PhaseCorrect, I believe this is a matter of the resulting frequency, as the comparison with 150 Hz suggests.
EDIT: Skip my answer. Any 7135 driver would light up at these PWM values, regardless how many of them are on it. Arghhhh. Just the lumens are different.
EDIT 2:
I just flashed a stock nanjg 105c with most recent STAR Offtime.
- Moon can’t (and does not) light up as it defaults to PWM-3 at the moment.
Oh, that one was the 4x7135 driver in the CNQG brass 18650 light. It uses 350mA chips.
I have one with 5x380mA chips, and its best moon mode is fast PWM=6, for about 0.25 lm.
So, I just check a few settings on each light and pick whichever one I like the best.
I finally got around to trying out STAR offtime. I like it. Feels pretty intuitive.
What kind of behavior could be achieved with offtime “no memory”? I think Ideally the driver would always be next mode when the light is on, but from off it would always start on LOW.
That's correct. Or it can always start on high, if the mode order is reversed .
Offtime-no-memory is my favorite configuration. 
Basically, I set it up so that a short press goes to the next mode, a medium press goes to the previous mode, and a long press goes back to the first mode. It thus always starts in the first mode, and can navigate in two directions. If the user goes backward from the first mode, it enters some semi-hidden “negative” modes like turbo and battery check.
Just tried offtime-no-memory. It is exactly what I am after. The general public would probably appreciate that simplicity, I figure.
I’m using a low of “6” with fast pwm. Seems to work pretty good. Is there a limit to how low I can make low before there is some kind of issue?
The lower limit depends on hardware. Driver, LED, and [potentially] input voltage.
In general, for a nanjg driver with current-gen LED the lowest is about 6 (fast) or 4 (phase). For a FET driver the lowest is generally about 1 (phase) or 0 (fast, with some caveats).
It also depends greatly on the number of 7135s (the higher the number, the higher the stable PWM value), the batch of 7135s (they vary from batch to batch, and even within batches), and whether the 7135 is a 350mA or 380mA (350mA can generally run at a lower stable value).
Is reverse polarity protection something that is physically built into the circuit or something that is handled via firmware? Doing a S2 build for my wife with a 105c and STAR and wanted to make sure that it stayed in.
Thanks
It’s physically built in. It’s a diode.
Many thanks.
Is this version posted in your repository? Think I went look'n - not sure I could locate it. So... Are you saying you are using 3 level values for reading the cap to determine the short/med/long press's? If so -- is this reliable? Standard value OFF time cap used, etc.? If it's reliable, wouldn't JC incorporate it into this? Maybe he doesn't feel confident, or maybe just no time to do it yet, or maybe he doesn't care for this feature?
If it's a mod of JC's OFFTime, is it based on his latest 1.3 version?
This sort of seems like it works like the "werner" style e-switch firmware. I make the long hold do a strobe typically. I think the OFF time style med and long press's though would be a little more tricky/finicky then a simple button hold for an e-switch, maybe?
Only a couple of questions.... 
I haven’t used it, but I just can’t see the short/long press navigation being half as user friendly as an e-switch. With the e-switch you do a long press by holding the button until something happens. You do a short press by hitting it and letting go. Both pretty straightforward actions. Now when it comes to “how do I turn this thing off?”… the clicky may have an advantage. 
I’ll still give this business a try, I want to see what it’s all about.