No doubt… Would be a ton of fun, though!
1 Thank
Ah yes the 4xAA hefty pill vial formfactor (rolls own example of Mk41 over in hand a few times) - the proportions are good in the hand.
But then you realize why this isn’t a popular enthusiast formfactor. Generally speaking:
- 4x AA NiMH similar energy storage to 18650 with worse power capability
- 4x 14500 similar energy storage to 21700 with worse power capability
1 Thank
From a post on Reddit it seems the Lume X1 driver is available in the D4V2 now.
@Hank_Wang can you confirm if this is the standard boost drive option or a special request?
Is it available in other 18650 lights?
Also I’m happy to confirm the order I am expecting is out for delivery via FedEx Ground just over a week after it shipped.
1 Thank
Sure thats the boost driver upgrade option that has been there for years. Doesn’t say anything about it being LumeX1, which is why I am asking.
Screenshot of D4K page:
Sorry, I thought it’s the same upgrade.
Wasn’t the issue with the X1 driver in the 18650 lights was that it was physically unable to fit?
I had something like that a few years ago, I forget the make, well known brand.
Edit: Nitecore EA41.
Hi, sorry it took me so long… my ability to get distracted is legendary.
Anyway, after some trial and error, I’ve found that roughly half a millisecond is as short as I can make the preflash absorber without getting an actual preflash. Could you try this and see if it helps?
https://toykeeper.net/torches/fsm/anduril2/anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex
In my testing, this has no preflash and virtually no dip with a li-ion cell, and with AA there’s no preflash and a more visible dip. The optimal timing seems to depend on the power source voltage… so I picked the smallest value which prevents preflash for all supported battery types.
4 Thanks
Was never able to source one, always wanted. Very nice!
I saw some mention they had some back in stock a while back, I hadn’t gotten my hands on an XHP35 light by that point and it’s a bit funky so I had to snag it.
I just wish the engineering mode was a bit more like I’ve heard other Mankers are - this one has the max & min outputs tied together for some weird reason. Set it all the way low and you have a severely reduced max output… Set it all the way high and there’s no real moonlight anymore.
But I just run it with NiMH’s anyway so it’s not exactly a hot rod. And the size & weight are unique in my collection.
installed D3AA2025-05-22
https://toykeeper.net/torches/fsm/anduril2/anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex
Confirmed No Preflash on my D3AA with both Eneloop and LiIon
This is excellent!
There is still a dim blip when using Smooth Ramping going up past the 10 lumens mark, on both battery types…
That stutter blip also happens when turning on at any output between 10 and 30 lumens. This stutter only happens when Smooth Steps is Enabled. If Smooth Steps is off, instead of a stutter theres a bit of a flash when the light turns on.
Getting rid of the Preflash when turning on at Firefly outputs, is a Really Nice Update… definite yes to reflashing w D3AA2025-05-22
Thank You!
7 Thanks
That blip seems closer to 15-20 lumens, at least on my d3aa. I REALLY gotta get off my butt and start learning the code LOL. I have a feeling as to what is causing it, like when a car’s next gear ratio is off…TK mentioned before that the lowest lows have thousands of submodes, lol…but the next subset has substantially less
Hi, we have many plans, but they all take long time to be completed.
4 Thanks
Hi, D4V2/KR1/KR4 version Lume X1 drivers are not available for now due to the components shortage, so, the “boost driver upgrade” on the site is still the old version boost driver.
Thanks for confirming that the package was delivered by FedEx, instead of USPS.
4 Thanks
It is indeed caused by “gear” ratios.
Part way up the ramp, it changes “gears” by switching from one sense resistor to another. This works pretty well, but if the resistors have, for example, a 5% error margin in manufacturing, and the ramp steps are each just 3% brighter than the step before… that means the ramp will be totally smooth on some lights while having a “blip” or others. It depends on the random variance in the resistor.
It’s not very noticeable when the step is larger than usual, like if the ramp steps are 3%, 3%, 7%, 3%, 3% … it looks slightly “off” but isn’t very eye-catching. But if the step is negative and makes a zig-zag, it’s very visible. A pattern like 3%, 3%, -1%, 3%, 3% really stands out.
So to create the ramp values, I generally try to compensate for this variance by making that particular ramp step bigger. That way, even though it may be an extra large step sometimes, at least it won’t be a zig-zag.
However, given that I typically have a sample size of 1 (or maybe 2 if I’m lucky) when doing the calibration, I can’t really measure the variance. So a lot of the values are a guess.
To fix it on an individual light, you’d need to recalculate the ramp with a different step size at that specific position… which is unfortunately more complicated than it sounds. The math is pretty finicky and I usually have to trial-and-error it quite a bit to get a set of values which satisfy all the constraints.
7 Thanks
KR3AA would become the absolute best 14500/AA tailswitch flashlight!
3 Thanks
Absolutely no problem, TK – on the contrary, thank you very much for your efforts! 
my ability to get distracted is legendary.
Yours and mine both, sister – yours and mine both (and I suspect, most of our friends and colleagues here in BLF and in the flashlight enthusiast community).
Thank you again! Unfortunately I just tested it and the preflash is still there: https://durval.com/xfer-only/VID_20250523_142528.mp4
(haven’t tested the “dip” as it’s actually not a problem for me, or at least not one as serious as the preflash).
Maybe I’m doing something wrong? Here’s what I did:
- Downloaded the HEX from the URL you indicated above:
$ wget https://toykeeper.net/torches/fsm/anduril2/anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex
--2025-05-23 13:34:33-- https://toykeeper.net/torches/fsm/anduril2/anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex
Resolving toykeeper.net (toykeeper.net)... 74.95.113.201
Connecting to toykeeper.net (toykeeper.net)|74.95.113.201|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 32279 (32K) [application/octet-stream]
Saving to: ‘anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex’
anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex 100%[=====================================================================================================================>] 31.52K --.-KB/s in 0.02s
2025-05-23 13:34:34 (1.94 MB/s) - ‘anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex’ saved [32279/32279]
- Flashed it to my D3AA:
$ sudo avrdude -p avr32dd20 -c serialupdi -P /dev/ttyUSB0 -Uflash:w:anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex
avrdude: AVR device initialized and ready to accept instructions
avrdude: device signature = 0x1e953a (probably avr32dd20)
avrdude: Note: flash memory has been specified, an erase cycle will be performed.
To disable this feature, specify the -D option.
avrdude: erasing chip
avrdude: reading input file anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex for flash
with 11466 bytes in 1 section within [0, 0x2cc9]
using 23 pages and 310 pad bytes
avrdude: writing 11466 bytes flash ...
Writing | ################################################## | 100% 2.13 s
avrdude: 11466 bytes of flash written
avrdude: verifying flash memory against anduril.2025-05-22.hank-emisar-d3aa.0.6ms.hex
Reading | ################################################## | 100% 1.80 s
avrdude: 11466 bytes of flash verified
avrdude done. Thank you.
- Did a hard reset (unscrew tailcap, press and hold switch, screw the cap back until the big flash and then release the switch);
- 10H from off to set Anduril to Advanced mode (in order to be able to achieve minimum moonlight);
- 1C to turn on the light, then 2H to turn its brightness to the minimum;
- turned off the light with 1C, then turned it back on (again with 1C) to check for the preflash (and there it was);
- Repeated the last step above a couple more times to make sure (and to make the video showing the issue, which I just linked above).
Am I doing anything wrong? Please confirm, specially since @jon_slider reported it cured the issue on his D3AA.
1 Thank
With all due respect to @Hank_Wang and to you and all the other Hanklight fans, I don’t think so: my trusty old TS10v1 is still my favorite 14500 light, tail-switched or not.
Granted, it doesn’t have AA capabilty (it’s 14500 Li-Ion only) but its miraculously good CSP LEDs almost match the 519As in my D3AA, it cost a fraction of the D3AA price, and it excels in most other aspects (the more important being, no damned preflash when turned on at its lowest moonlight brightness).
1 Thank
Soooo…with all the hardware already there, if there were some way to swap the driver for a freeman driver, would that essentially be a kr3aa?