Wow this is really impressive. Being able to reverse through the modes is a great idea
The holes on the Convoy clip are too far apart for the A6
Bummer. I think I’ll replicate this GB with a convoy S8.
I need my clip, bro!
I’d rather this just wrap around to the high end and have the hidden modes accessed some other way.
Glad to see there will be a 4 mode option.
Wow, toykeeper you truly are a genius!!! 
Put me down for another CW 
For those that aren’t as interested in max output, I ran the Panasonic BD cell to see how it performed. 5.29A for 1438.65 lumens. That’s a 3200mAh cell so it’s the best of both worlds… longest run time possible for the highest output available.
The LG HE-2 I tried allowed 6.53A and “only” made 1659.45 lumens. My Efest 35A, or one of many, matched the Sony C5 both in amperage and output. All cells fresh off the charger.
Edit: I also changed the R1 resistor to match the driver ToyKeeper is working with, so now my battery monitor is correct.
Would it be feasible for Toykeeper to investigate doing indirect current regulation in software for the non-turbo modes? In other words, I’m wondering if the MCU could compensate for the falling voltage from a cell by increasing the PWM duty cycle over time?
Now, I’m probably missing something, because if this were feasible, I’d imagine all the FET drivers here would be doing it. Can someone enlighten me as to why they aren’t?
I was going to wait for the answer to that question before deciding whether to order or not, but while I made this reply I was lured in by the brightness. I don’t really care if the lower modes dim over time when the peak brightness is so high.
Count me in for two of the NW if the price is actually low 20s. I’m also willing to pay a bit more for a better emitter bin, if that’s what it comes down to.
Edit: s/your resident hacker/Toykeeper/. You can tell I haven’t read the entire thread. I was under the impression that the updates in the first post were all I needed to know. I might go back and read the whole thread later when I have time.
I think the answer to that one, XP-R, might lie in the capacity of the MCU. The ATTiny13A only holds 1Kb, so I’m sure the ability for the firmware to do as you’re asking would eat up way to much space. As it is, ToyKeeper has worked small miracles in getting this trimmed to fit, and it’s a just barely thing. If I’m not mistaken she claims it’s within 10bits of maxing out.
There are plans around here to port over to the ATTiny25 for double the storage capacity in the same footprint. Maybe then it’ll be possible to do as you’re asking. And then some. 
(Not a code writer, only guesstimating here.)
Short answer: no.
Long answer: It’s possible, and I’ve already done it in a different firmware. However, there are some pretty big issues with it…
- The code for it takes a lot of ROM space, and the attiny13a doesn’t have much space to begin with. I had to do code tetris for several days in order to get everything to fit, and it still lacks a few of the original features I was aiming for. So, soft regulation would mean cutting out other, more important features.
- The soft regulation must be done completely blind. The MCU has no idea how much light is actually coming out, so it would have to guess…
- The “guess” is surprisingly complicated, since it involves matching, inverting, and compensating for half a dozen different variables with different independent response curves… and the attiny is terrible at any math beyond adding, subtracting, or bit-shifting. Floating point is right out, and a single divide operation takes like 15% of the total ROM space. The full equation literally doesn’t fit, even if it’s the only thing the MCU knows how to do.
- No matter how good the guess is it’ll be wrong as soon as someone puts in a different brand of battery.
- Even after compensating as much as possible, the result is still mediocre.
- The low modes (moon, low, med1, med2) are already regulated anyway, thanks to the 7135 chip, so it’s only the higher modes which sag with voltage. The high1/high2 modes also use as much regulated power as possible before resorting to unregulated power. This reduces or eliminates the sag on all modes except turbo.
If you’d like to try it anyway, feel free to flash my “cypreus” firmware onto a clicky-switch light with a FET driver and OTC. It uses PFM-based soft regulation on the moon mode, which can be visually seen each time the light turns on (it self-adjusts for a second or two before settling).
I feel like I am rapidly losing my comprehension of the info in this thread.
Haha…don’t worry. You are not alone.
I had this thought yesterday but didn’t toss it out there as the conversation was spinning away from the topic. BUT, since I just remembered this here goes… we can stack resistors, we can stack regulation chips, we can stack MOSFET’s (see where this is going?) Is it possible to stack an MCU and gain the capacity? Probably wouldn’t gain any pins as they’d all be connected the same anyway, but would the additional ROM be available? Or is this, essentially, porting to the 25?
I don't feel so lonely now. Our own little non understanding club that appreciates the work the very talented few here do.
Nope, it doesn’t work that way.
This one’s perfect now anyway so I was just askin.
More features and I wouldn’t be able to work it. This one is easy, straightforward, and intuitive. I won’t forget how it works. More stuff and I’ll forget.
Will this firmware work on the WIP fet + 7135 driver that is Wight made, or will this driver be different and software not compatible?
What I’m using is a FET + 7135 driver designed by wight, so … yeah, I’m pretty sure it should work.
Currently, STAR_offtime and “starry-offtime” both work on it, plus the firmware I’m making for this particular light. The first two are already published, and the third will be available for modification soon.
This is the driver that we (term used loosely :) ) are working on
https://budgetlightforum.com/t/-/31102
It is not an absolute done deal but that is we we are headed towards. It's purpose is to have a suitable or acceptable moonlight and yet still have FET. ToyKeeper is a genius whereas the first three modes are on that 7135 and there is no voltage sag and the 7135 stays on except at all out turbo to minimize voltage sag on the FET.
This is the same driver PLUS the addition of a 7135. Although there already was dual PWM code available...ToyKeeper wrote one and tweaked the hell out of it :)
If there’s a hidden strobe mode in the final product, I’m in for 2 CW.
Yes, there is a hidden strobe in TK's firmware.