150 steps are smooth
The question is how many of them spent to the 7135 and how many to the FET
Makes a difference
a 14500 single emitter 350mA vs 4A
Needs more ramping steps for 7135
Q8 4x18650 quad emitter 350mA vs 15-20A
Needs less ramping steps for 7135
Ah I think TK has a table or tool to calculate tables for this.
There is no bump during the ramp so optical the ramping steps used work very well 
Over 1300 interested, tells something, doesn´t it :sunglasses:
Its how much time you have in FET ramping, the more steps the longer it cycles through those modes, so more time and easier to handle
2.4 seconds it takes from low to turbo
So like 1 second for 7135 and 1.4 for FET would be good
Where to download that ramping calc tool I asked once and nobody could answer it
heh wanted to respond with “yeah that it must be done right with a good light as result” but that would also be the case when only 10 people wanted it, so it tells me that th effort put into it up to now was not in vain 
I don’t know how up to date the code is:
http://bazaar.launchpad.net/~toykeeper/flashlight-firmware/trunk/view/head:/Tom_E/narsil/Narsil/Narsil.c
Put me on the list for one flashlight .
Thank you
Welcome to BLF, will do!
Lexel, I hope Fritz answered it, or that TK can chime in for I cant help you with the tool or table
All I see is the result and it is like a fast sunrise, smooth and natural feeling.
To address the comment about the spread of ramping level brightness, when I created the ramping tables, I definitely gave preference to the 7135. I do not agree a considerable amount of entries should be given to the FET, just like we do in sets of fixed modes. For example, for a FET+1, you got 5 modes - 3 of them are in the 7135, two in the FET.
I believe the low levels are more useful in having more choices for. Just like I can't find a 70% or 80% mode useful. Highest I would go with a 2nd brightest mode would be 65%, and that's for 7-8 modes in a set. For 6 modes, 40% is the 2nd highest.
I could be totally wrong about the theoretical spreading pure eye perception of brightness, but this is how I always defined mode sets since I started defining my own, over 4 years ago, and really though that's how most of our mode sets were done by the community in general - much better than most stock mode sets.
The good thing is, of course, it's open source, there's a Python app available, written by ToyKeeper, here in the Flashlight Firmware Repository, to generate your own tables and I document the Python command parameters for the tables I generated. Unfortunately it's so not easy to simply download a replacement table - the firmware must be re-compiled and downloaded.
Also I just haven't had the time to experiment with different table settings much. My goal was making it smooth as possible, and weighted more towards the low end.
Just for my understanding, and to kill time, the Q8 driver is considered a dual channel, right?
It uses the single 7135 (350 milliamps?) current regulator for the light levels under 350 milliamp, then the brighter levels switch (maybe the 7135 is still in use?) to using the FET from there all the way to max 20 amps. Is this correct? (This is the FET+1)
To veer away from the Q8 for a second, FET drivers that have 9 of the 7135 would be considered triple channel where a single 7135 chip controls the levels up to 350 mA.
Then the other 8 7135 chips come into play for the levels from 350 mA to 3,150 mA (350×8=2,800+350=3,150).
Then for levels above 3,150 mA the FET takes over.
Am I right? This makes sense to me.
That sounds right to me. Keep up the good work. :+1:
I hadn’t ever looked at “BLF” in Morse code before. It seems to be one of the easiest patterns possible, almost like 1 2 3… kinda neat.
_ . . .
. _ . .
. . _ .
Continuing the pattern, the next letter would be V … but I don’t know what the V would mean.
About the ramp shape, it’s right there in Narsil.c linked a few posts ago.
The tool to calculate ramps is bin/level_calc.py in the firmware repository. It has lots of options (some of which require small code edits), but by default it uses the cube-root model for estimating what would be “visually linear”. To change the ramp shape, edit power() and invpower() at the bottom.
In my testing, I’ve found Narsil’s ramp to be very very smooth. I can’t tell where the FET comes on, and it appears to ramp at a constant speed. It might be a little slower on the FET portion, but mine (SRK w/ Q8 driver) is 3000 lumens instead of 6000, so on the real Q8 I think it would be just about right.
Just ordered a set of HG2 for the Q8 
Ooh nice Thijsco has his as well
5 of 7 arrived
Tom has fixed the bugs!
If somebody is able to help test this latest version that would be awesome!
What is also awesome is that you TK seemed to have read my mind from the future (well when you posted that Morse it was future now it is present) for seeing that pattern I wanted to see what the next was and you already looked it up
For the ramping se are of course looking at perception nicely described in the Weber–Fechner law.
From PMs I get that Lexel mainly talks about the triple channel ramping that does look good and natural just has a very short ramping in the FET region so it is hard to select FET brightness .
This does not apply for the Q8 with its FET+1 for it switches to FET modes above 350mA so a lot of time to choose several FET driven brightness levels.
Yes - the BLF Q8 driver is two channel, but for the new FW3A, the driver will be 3 channels.
Yes - brighter levels include the 7135 full on, but on the max level, brightest setting, it's FET only. We found it's more efficient giving all the amps to the FET for resulting maximum amps to the LED.
Triple channel drivers can have 2 or more 7135's in the bank. We (or I, not sure) call the channel with the multiple 7135's a bank. I've used boards with 3 to 9 7135's in the bank but the bank can go much higher. Texas_Ace has a large selection of exclusively triple channel boards, including SRK versions up to 15 7135's in the bank here: https://budgetlightforum.com/t/-/40540, while Harley's SRK board is versatile enough to use as a triple or 2 channel board here: https://budgetlightforum.com/t/-/40540.
pilotdog68 was the first one to create an open source driver using 3 output channels here, back in Dec 2015: https://budgetlightforum.com/t/-/37080
For ramping in triple channels, when the FET kicks in, the 7135 channels remain turned on full, up til the max output. Sometimes you can notice the slight bump at the max level when the 7135 channels are turned off.
Thanks, I’m slowly learning this stuff.
Is the 7135s voltage drop that cause the less power reaching the led ?
I'd be guessing, but sounds like it could be.
What batteries are recommended for this flashlight?