Any indicator will be a voltage indicator whether it’s 0-4 or digits/tenths, percent capacity has to be inferred for either one. If the blinks are fast but still distinct then d/t would be ok, I just don’t want to wait long for something this routine.
Yes, probably not a good idea with a FET driver.
It does work OK with a 6* or 8* 7135 driver.
Something to try with the real-time indicator is to do the filtering at a frequency at least 2x the PWM frequency (Nyquist-Shannon sampling theorem).
Might have to speed up the ADC for this.
No, I’m saying it seems you are failing to consider most probably use this function most commonly to see if its “pretty full” to full and do I need to get another battery when starting out, as is said in varying ways by several posters already in the thread. For that, Whole/tenths is equally as fast and yet still most detailed for those who do care.
Yup thats what I was trying to say, it has an advantage in this (it was late and its late yet again…hardly any time on here anymore).
I think that it is just as fast for the most common usage, and by saying its not fast over and over more will just not vote for it 
Even though they will generally just use it to see if they have a full battery when going out. I find most users hardly use the light very long ever, I don’t even know someone who uses full batteries up like I do, or uses them forever like you do.
Yup I understand that, but I think its an easy way to estimate for the “muggle” that is pretty close to the same accuracy as 25/50/75/100 flashes, and in many cases more accurate when they are at either end of those ranks…
Not only that but to make it more complicated, if you use the cell at 5A vs if you use it at 1A, the same measure is really a different percent used, and both whole/tenths and 0-5 blinks would be very off target on some discharge rates, dependent on the battery, so they are equal here.
Yes I understood that part too, but that is the part that makes the 0-5 less useful than most think, since most batteries voltage drops, so if 4.2V is “full” and your battery drops to 4.19V or 4.18 or 4.17 etc, you will probably never get a 5th blink to know your light is “full”, which is currently annoying in the A6 
It still does ameliorate this some since you can detect where the error lies. I can only guestimate what the A6 is doing. But if it was whole vs tenths, I could see “oh, when I put a 4.1V battery in, it is registering at 4.0” (or 3.8 or 4.3) and make adjustments for this. The bigger the error, the more the 0-5 is off and becomes unusable, whereas whole vs tenths is still somewhat usable.
Yup I understand and thanks for considering. I do know those who would use this only in their lights and will never get a volt meter (pretty much everyone I’ve given or to give a light to, sadly ). I really think most haven’t thought about this issue, and will superficially choose the method billed “simplest/quickest” method without thinking about it, but in practice wouldn’t really care if its whole/tenth vs 0-5 is all since 99% of the time it will be used for “did I put a full battery in here or leave the one from last time in?”. That’s exactly what I’d use it for too, but I’d want to know how full exactly as a preference. Additionally I want it to let a couple people I know have a crude voltage meter to make sure they don’t run their batteries too low since they sometimes forget and refuse to buy either a voltmeter or a protected battery 
Goodnight for now
Yup, but the 0-4 is more off at either end of its 25% range than the whole/tenths method. Every time your light nears a margin with the 0-4 method, it is off by up to 24%
In other words for example:
Real capacity…0-4 method shows…Whole/tenths method shows
99…………….75………………….4.2V (We all know this one)
74…………….50………………….3.9V (you may not know , but you prob know its > 50)
26…………….50………………….3.4V (you prob know this is less than half)
What this shows is the 0-5 method creates situations where if you really do care and it matters how much battery you have, you are mislead, which can be…disastrous if it really mattered, most likely severely annoying in most cases. Also importantly, even most people who know little about voltage curves, still know enough to approximate closer to reality than the 0-5 is giving them. And can still learn a tiny bit to understand this if they dont.
OK really going now
I’m confident that whatever TK settles on some will like it, some will not, and some won’t understand it. I’ll get used to it either way so I’m not going to sweat it. I like chocolate and that’s final.
I’m going to be very different, “none of the above.” I’m new to the Attiny13a and programming it so I’m sure I’ll be biting off more than I think but it wouldn’t be the first time and since I’m part Bull dog I’ll just keep growling and chewing on it. LOL
My plan is to use an open port (or two) and drive a bicolor LED mounted flush on the side of the head monitoring all load condition voltages. It would indicate at turn on, if triggered by the user and if the voltage drops below a set point with several blink rates and intensities to indicate the urgency.
Very different indeed. Somehow you managed to get zero posts while obviously having at least one. Neat trick!
First time I’ve seen someone with zero posts.
Welcome to BLF… from the future.
Anddd ToyKeeper beat me!
Er, it’s not supposed to give 5 blinks for a full cell. That’s used for > 4.20V, meaning over-charged (or a 4.35V cell).
In any case, I prefer more precision. I’d probably go for 8 percent-based blinks or even a binary representation of volts+tenths, but neither of those seems to be very popular. The 4-blink style was chosen based on its popularity in other places, like for phone signals.
It’s clear from the survey results that there is a more variety than consensus on this. So, I plan to implement a few styles and make it a compile-time option.
And for my next trick . . Perhaps the handle of Alchemist has deeper possibilities. LOL
Do posts in polls not count toward one's post count?
I voted for the 2-4,0-9 for me.I love it on the Nitecores.For gifting etc,I think the coarse 4/5 blink system is the easiest and still accurate enough;2 blinks or more…take the dog for a walk , 1 blink…charge or change the cell,same as your phone.
So what I’m saying is,5 blink standard,with (relatively)easy deletion and cut and paste the 2-4,0-9,or whichever, firmware changes for those who want more precision.Break out the AVRUSB,AVRdude etc
In the last year,it seems software modding has become as important as hardware.If you want a Courui to throw 1000 metres,you have got to mod it,If you want 0.1 volt measurement resolution,same thing.
TK,that sounds like the direction you are heading anyway.
Just a suggestion TK but if you were doing one program for one person you might stand a 50/50 chance of them liking what they said they wanted. More than that and the odds odds go south so since it sounds to me like you are the expert here just go with your preference. There will be always be another ui.
Ahhh - there is always a solution. Implement both popular methods, let the user choose what they want 
. That's how it's look'n so far, about even split with options #1 and #4.
BTW, something similar is already implemented in the Ferrero_Rocher/*.c code if you’d like a reference point to start from. It doesn’t blink, but it does at least drive a pair of red and green voltage indicators.
Yes, that’s what I’ll probably do. On attiny13, it’ll have to be a compile-time option, but on tiny25 and bigger it can probably be a runtime option.
On attiny25 I got 4-blink, 8-blink, and volts+tenths styles all working. It’s a compile-time option for now.
On attiny13 (blf-a6), it at least allows 4 or 8 blinks… but volts+tenths didn’t fit. I’d need to come up with something far more clever (and probably less accurate) to make it fit.
In either case, 6 or 12 blinks is really easy to add as a compile-time thing if people want it.
Both methods sounds great if doable.
With my posts I was just hoping to make people aware, I think most really haven’t even thought about this very much, I know I didn’t until recently. And when I did, I realized that if you tell someone “this method is fast (nothing else), this one is slow but detailed, which do you want?”, usually fast wins, hands down, and the one being labeled “slow” was useful to me. For the most common use (is it relatively full or do I need to put another battery in before taking it out?) its equally fast, so I was hoping to get that point across. In fact I’ll probably most commonly only wait for 3 vs 4 blinks in practice to know if I need a different battery in it going out the door.
0-4 is full of so many unknowns as to be useless to me, you pretty much know within 25% where your battery is I’d say if you remember your past use at all, or by the heat it generates on turbo if you have a high output light. Unless you use your battery for a very long time between recharges, and can get by in your next activity on 25% of your battery, then it becomes useful. I think is what is going on here: high output users preference vs low output users preference.
Heck, I didn’t even know I cared about a battery indicator until I realized that if it was whole+tenths I would have a use for it myself and then more importantly that others I know could use it who didn’t have voltmeters, instead of me trying to get them to buy one or have me measure their battery. I guess I’m now sort of excited about using a battery level indicator, flashoholics are pretty silly eh? 
Well, it’s working now if you’d like to mod a driver with a tiny25 and try it.
I’ll have to use a different approach to make it work on the tiny13 version of BLF-A6 though… it’s just too big. I guess I could save a fair amount of room by sacrificing accuracy, but that might defeat the point. Or go with the 12-blink thing (tenths above 3.0V), which fits and is just as accurate, but doesn’t look as nice and doesn’t measure below 3V. Or maybe something clever I haven’t thought of.
Hmm, well, 12blink is also useful, but maybe reversed so its “fast” too: 1 blink = 4.2, 2=4.1, 3=4.0 etc. That way the “I just want to see if my battery is full” use is really the fastest out there (I think most people’s usages will fit in that category), its simple since it can be inaccurately but simply thought of as “twelfths of usage hits on the battery” or simply “alot of flashing = recharge” and its accurate for those who really want to know detail of charge above 3.0V by just subtracting flashes as tenths from 4.2.
I do have the mod a driver in “queue”…if I can just get through my never ending work faster 