yes I am happy with mine as well. It has given me the confidence to go ahead and order another fireflies flashlight. Probably the E07
[♛ FreemeGB] Fireflies PL47 Gen II 4*XP-L/ Nichia/ SST20 Hi CRI 21700 Right Angle Flashlight - ENDED
17track still has nothing.
< sigh >
Ordered late October. Jack Zhang confirmed the order 30 October. Shipped 26 November.
A tiny bit frustrating when other people have their lights already.
If a CW fan likes this 4000k, it must be really good. 
Did you take any lumen measurements?
Not a CW fan, I like white, as in completely neutral with no tint in either direction.
The angle head doesn’t fit in the 4 1/4” opening of my light box so the readings from it seem skewed, very low. I tried it on the iJoy 21700, numbers are too low for me to believe so I didn’t even write it down. Seems like it was in the upper 1300’s. And yes, I know to double click for Turbo in Anduril…
I think you need this. And maybe this.
Measure a few known points on the ramp, then mess with the calculator parameters until the curve fits your measurements. Reference values are in Anduril’s cfg-*.h files to get you started.
Does the PL47 Anduril still have 255 steps? I though I read that was default for the UI.
I can say with ramp floor set to lvl 10 on my 219B is NOWHERE near 120 lumens. Probably closer to 2 or 3 at most.
?? I don’t understand what you’re disagreeing with, that is what I expected. I remember posting in one of these threads just after you, saying which batteries give more current, according to HKJ, and it agreed with your observations.
These all agree with what HKJ says, except for the second one, which he didn’t test. Interesting to see it’s really good. I expected the Vapcell 2500 to be a rewrap of the Samsung 25S but HKJ shows it gives significantly less current than the VTC5A, so probably the Vapcell isn’t a rewrap of the 25S.
In the same boat as you. I would say more than a tiny bit frustrating…
Ugh. I cant really make sense of the links. Code is outside my frame of reference. And i do not have a way to measure lumens, hence blf user estimated lumens (peak).
But it sounds like what you are suggesting is there is a weird curvey thing at work. In the code stuff i did see “visually linear”. And arent lumens funny that way, such that twice as bright actually is 4x’s technically bright? Which would suggest an exponential curvey thingy, right?.
So what you seem to be saying is my table doesnt work—which i suspected.
I am not really-math-capable, that formula truthfully exceeds my ability level to begin with. That said, it does seem like if the curve parameters are understood, the peak is given and the number of steps are given one should be able to define those steps with math, right?
Is there anyone out there who can do this?
It’s quite easy. It’s a logarithmic curve following our eyes’ perception of brightness.
I got 150 from the rot66 thread, i think. Post 636 and another one— I have 2 screenshots of guys explaining how to configure ramp. But it is unclear to me the exact source. Both suggested 150 steps, but it is now unclear to me if that is complete steps or regulated steps.
That was what i suspected, but seeing it is the easy part. I think the math part is slightly more complicated, no?
At least more complicated than my tables formula.
[Edit: Ha! easy is relative.]
cclight, you said that the 30T doesn’t give that much more current than other cells, when in fact it yields some 50% more than other 21700’s… at least the one’s I have.
Maybe I misunderstood the context or something. . .
It has 150 ramp steps from moon (level 1) to turbo (level 150).
Internally, this is implemented by using PWM with duty cycles of 0 to 255 on multiple power channels. But as far as the user needs to know, it’s 150 ramp steps following something close to a perceptually-linear curve.
It’s not necessary to understand the code… but if you run it, it can calculate the estimated lumen output for you. It generates tables like yours, and lets the user choose which type of curve shape they want and how bright some of the points along the way should be.
It’s a python script, so it’d need a python interpreter installed. Depending on the OS, this can be a little inconvenient or it could already be installed by default.
Anyway, the links are the script used to calculate the ramp, the parameters sent to the script, and another thing to help determine exactly which of the 150 ramp levels it’ll use in stepped mode. The ramp shape it used wasn’t a logarithmic curve in this case; it used a ninth root instead.
Ha. Thank you. But, maybe someone else with like minded enthusiasm will be able to do that, because i just hugged my knees, started crying and i am now rocking back and forth. I hope to be done crying by friday but we’ll see how it turns out.
Seriously though, that seems about a phd away from my level of typey thing aptitude.
I said it doesn’t give twice the current of other high current cells. This is what your data show. It was in response to someone who asked if it’d be safe to use the 30T in the E07. I said it would be. In the same post I said I’d use the 40T in the PL47, due to heat issues, and the 30T in the E07.
Seems we’re in agreement.
So i just tried to acquaint myself with the concept of ninth root as it is entirely foreign to me and i found stuff like this:
“…Any non-zero number, considered as complex number, has n different ”complex roots of degree n” (nth roots), including those with zero imaginary part, i.e. any real roots. The root of 0 is zero for all degrees n, since 0n = 0. In particular, if n is even and x is a positive real number, one of its nth roots is positive, one is negative, and the rest (when n > 2) are complex but not real; if n is even and x is a negative real, none of the nth roots is real. If n is odd and x is real, one nth root is real and has the same sign as x, while the other (n − 1) roots are not real. Finally, if x is not real, then none of its nth roots is real…[wikipedia]”
The rest of the internet made about as much sense to me, didnt even seem like a comprehensible language for me, like i recognized the letters and some of the words but in combination, in context it just sounds like, “barblesnarfing harnkle pop doozelwenkel root plurfle zonkin buffle. Horple n > 2, yintoffen roffer (n - 1) noktory jimpley pinduffel…”
So unless someone else is willing to take up the cause i suspect this query is outside the scope of my abilities.
That said…i will consider the table i generated as a guideline with caveats. The provision being that it is wrong-ish, but ballparkyish right. In other words, good enough—for me—well not really, but no choice.
It basically means the ramp is configured to the shape of “x to the ninth power”… but scaled to fit. Every available ramp formula generates a smooth curve, but different formulas emphasize different parts of the output range. The ninth power thing I used places more of the ramp steps toward the bottom, then escalates quickly at the top.
In theory, a cube / cube-root shape should be visually linear, but I find it really depends on the light. So I made the curve shape a parameter for the calculator, for easier fine-tuning. And I use different shapes for different lights… just depends on what seems appropriate for each model. This one was actually calibrated for the Emisar D4S, and then copied to PL47 because it uses an extremely similar driver and emitter configuration.
The ramp shapes it has so far are: square, cube, fifth, ninth, and log. In order, they get progressively steeper at the top of the ramp and progressively flatter at the bottom of the ramp.
In any case, the higher it goes, the faster it rises. It’s not linear.
See why we love her? 
Beautiful and painfully smart to boot. The beauty behind my brawn (lumenetically speaking, of course) I don’t have to understand it, I just have to run big enough wires to it to light the moon… 