Fireflies E07 preview

Oh niiice.

I’m probably going to do something quite special with the light.

Since LH351Ds 90CRI have about the same VF as SST-20s 95CRI, I’m going to be mixing some emitters in :slight_smile:

A few pics of my raw aluminum E07. The aux are purple but look blue in pics. Plastic cap over the tail cap so I can hold it in my mouth at work and not chip my teeth. Tint of these XPL HI 5000k are gorgeous. I’m surprised how low the moonlight is even with 7 leds on. I’m ready for my gray one to get here before I scuff this thing up anymore. I’ve already dropped it several times.

The E07 has the led it shipped with on the side of the box in that triangle.

My LED's are definitely SST-20's (pulled the optics to check) and best I can tell from the scribble on the box is they are 4000K. The beam pattern looks so smooth and silky.

The 5000K's have the lower CRI but a lot more output and throw. I got some on order to see how they do - appears you should get more throw than XPL HI's.

Lockout in this light has nothing to do with ano, the pcb in the tail cap has to touch the tube, it’s not about the threads at all. :wink:

For those of you with the E07 and the D4S, is the D4S significantly more throwy? I dont find my E07 throwy.

The E07 is quite a bit throwier than the E07.



thank you!

Dale - true, I see the PCB is isolated from the tailcap. The E07 sure does have a big hot spot. The Amutorch X9 is much tighter, but of course it's got 3 reflectors, bout the same width optics/glass.

Were you trying to be funny or forgot to replace one word with the other? XD

Battcheck on my E07 isn’t working correct…It says that the voltage is 0.2V higher than it is in reality.
So, if the batterie is 3.91V, battcheck tells me it is 4.1v(4 flashes—>pause—>1flash), tested it with different batteries
Battcheck on my PL47 is working correct

Yea, I have a meter, battery charger, and light that tell me the voltage and they all give me slightly different numbers.

There's some significant variation from driver to driver (MCU and diode), so each light should be individually calibrated for it's voltage reading to get it precise. I don't see it in the Anduril manual here, not sure I missed it. For the most part, a 0.2V difference won't make a big difference in LVP - LVP will still work fine, just on yours it will kick in on 2.6V instead of 2.8V which is still safe for the battery.

If we used a voltage divider in the design of the driver, it would be more accurate from piece to piece but a higher parasitic drain and we lose a very valuable I/O pin.

I got a unique custom configured diode voltage drop setting per flashlight I mod with NarsilM, adjusted to 1/100th of a volt. I can get it pretty accurate by sampling with a bench PS or using a few batteries at different levels.

Calibrating temp and voltage is something I really need to get in to the NarsilM UI, or Anduril at this point. It's a pain to re-flash for these adjustments.

I’ve just got 21700 batteries I ordered a long time ago.

https://www.aliexpress.com/item/Vapcell-21700-4500mAh-20A-Rechargeable-Li-Ion-red-high-capacity-Rechargeable-battery/32963202094.html

red Vapcell 4500mAh 20A

Flat top and button top
Width is ok
Button top is way too long, as expected
Flat top works, but gets slightly crushed

I received my 21700 50E today too. It would be too long on the vanilla E07, but fits perfectly after the copper wire trick to extend the tube by 1.5 mm (described in a video posted in this thread).

In my case it seems to be the opposite, the E07 underestimates the battery voltage by 0.2V compared to the XSTAR charger. I haven’t tested multiple batteries yet though.

There isn't much left to a battery under 3.2V, so the firmware uses 2.8V as the LVP mark, so if you are under 0.3 or over by 0.4, not much of a difference. Personally I find it annoying that the reading is off though, so prefer it being calibrated tight. I just wish I knew back when that the voltage divider method was so much more accurate and reliable than the internal reference source we use now, but saving that I/O pin enables us to do all the neat stuff we do with AUX LED's and having 3 input channels to control: FET+bank+1 for better regulation.

I think the greatest source of variance is the diode we use to protect against reverse polarity. The MCU may be partially responsible as well.

The thing is though, not to get too disturbed by that small variation in the battery check… the light will be dimming long before those tenths matter and you’ll already know you need to charge the cell so in the long run it isn’t important.

I don’t use the battery check, I mean, once in a while I’ll run it just playing with the modes but I rely on my cells being fully charged and so will stick a new one in or recharge the one in it before use. I just don’t ever really take a cell down to the cut off point, not how I use a light.

On my personal lights, I used to calibrate the voltage on a per-light basis. However, I haven’t bothered with that lately because I find it normally doesn’t matter to me.

If really accurate calibration is desired though, maybe we could add something? I’ve considered putting the values into eeprom instead of hard-coding them, but then it would need a way to edit the value. This seems like a good use case for the optic nerve — directly editing values in eeprom. However, I haven’t really messed with that lately, and it uses an I/O pin, and it’s not physically implemented on most lights.

So… how would you ideally want it to work?

Was think'n when you are in the mode of displaying the voltage, maybe something like you do now, not sure, but a press&hold in this state or something would trigger a config mode, then simply press 4 clicks, then 18 clicks for 4.18V for example. Now 3.98V would be a total pain. Not sure if you go to 100ths, but I do in NarsilM. Again, I'm probably overly precise (i.e. anal) about the accuracy but it's nice knowing. I don't blink out in 100ths but measure and calibrate to 100ths. My DMM's all go to 100ths. Bad/weak batteries under load will drop, but in normal voltage blink out mode, I find the battery measurement to be accurate.

Same could be done for calibrating temp - when blinking out the temp, same sequence in the UI would trigger config mode - I would optimize clicks though, break them down per digit, so 24C entry is 2 clicks, then 4 clicks, instead of 42 clicks. A pause would be the trigger between digits, then we blink, then the user starts the next digit.

Maybe the digit trick could be used for the 3 digits of voltage, not sure. I kind of like using a cell that somewhere between 4.00 and 4.20 though so it's a max of 20 clicks. Maybe for consistency, per digit, or is more consistent for both temp and voltage to be 2 sets of #'s? Dunno.

I tell you though, when you get the calib dead-on, and see your blinks follow a rounded value of voltage -- 4.15V blinks 4-2, then 4.14V blinks 4-1, its pretty cool. Yea, I know, anal, but I just got that perfection in me - math over creative, 750 in math SAT, 540 in English SAT, German heritage. I really think somewhere in my ancestry I had a watchmaker or mason who spent their entire life trying to achieve perfection, died poor... Well, that might be the story of my grandfather as a house painter in Manhattan - high end work, never made decent money at it though