I put a lantern on my test bench and what I can say the current overshoots very badly during tint mixing
with 5 AMCs the max current peaks up to 2.4A
I used here 5V from a bench power supply with my battery simulator board to compensate all conduction losses in the test leads completely
Statistics do not lie much @ 250 measurements per second the PWM should not affect readings too much without a filtering capacitor added
I know TK is interested in some sort of data to help her change the parameter(s) used for tint shifting. She didn’t elaborate on what exactly she needed for data, so I took the data in the post above as it’s all I can really do. I’m sure she will check in and provide more feedback, and eventually an update to the firmware. I’m guessing the emitter used in the prototypes and the production units are some how different, can’t imagine how else this happens. Likely other’s imaginations are better than mine though. 
Just finished my second run test, 100% NW at level 4. 11h2m before the first step down. This is very much in line with what would be predicted based on the current draw table from above. It’s hard to calculate this one, as I didn’t see the first step down. But if it happened at 6h15min, the prediction for level 4 would be 375x1.756=10h 59 minutes. I don’t see a reason to run more tests at this time, at least until the firmware is changed.
GreenCamper, thanks for updating the OP with your results for levels 5 and 4. It looks like for level 5, 300 minutes when the first significant step down happens. for level 4, I would call that roughly 530 minutes. The measurements I made for current at default for level 5 vs level 4 would predict 1.756 times longer at level 4, or about 527 minutes. I love it when math works. 
A usual thing that can happen is that the AMCs fall out of regulation because the leads losses are too high to get for example over 2A in a test, they simply have some DD transient then that flatness the max current
If anything below 4.9V on my setup I would have measured too small currents
Thanks for all the data. I think this is probably enough to take a stab at adjusting the correction factor. It seems there was a pretty big difference in the amc7135 response curve between the prototype and the production models. The production model is much closer to a linear response, which is an upgrade from how it was before. This means a much smaller correction factor can be used.
For anyone with the ability to reflash the firmware, could you try this and check if the response is more flat?
http://toykeeper.net/torches/fsm/anduril.2019-11-29.blf-lantern.hex
New data with build 2019-11-29.blf-lantern.hex
| Level | ( WW ) | ( NW ) | ( Default ) |
| 1 | (0.016) ) | (0.016) ) | (0.016 ) |
| 2 | (0.158) ) | (0.158) ) | (0.119 ) |
| 3 | (0.490 ) | (0.492) ) | (0.358 ) |
| 4 | (1.01) ) | (1.01) ) | (0.84) |
| 5 | (1.80) ) | (1.80) ) | (1.75) |
very close at the highest level now. levels 2-4 are about 20% lower, but I’m guessing getting it flat all the way across all possibilities is next to impossible, or a least you would likely need different factors for ranges of levels. Run times at the default tint will be longer than the 100% tint ends of the tint ramp now.
Well done ToyKeeper 
AND this build contains the highly coveted SOS function as well. 
It sounds like it may have been reduced a bit too far. But yeah, getting it flat across all brightness and tint settings at all voltages would require rather a lot of data and a very detailed correction matrix. That’s probably not feasible. It’s a 3-dimensional space.
But we could at least get it sort of close using the simple method used now.
This build increases the value a bit, and hopefully it’s closer to flat (ish):
http://toykeeper.net/torches/fsm/anduril.2019-11-29b.blf-lantern.hex
Also, what voltage are you testing at?
I’m using a 30Q cell at 3.8V. Just what I had laying around, took of our of my Q8. I’ve compared a Sofirn cell at 4.2V and the currents look pretty close. 1.74 volts at default on level 5 with the 30Q, 1.81A from the Sofirn at 4.15V. So a bit more current at higher voltages, but I’m not thinking of adding that variable into this testing.
Do you expect the “b” build to be flatter across the various levels? Level 5 is really close now it seems, and level 1 is always flat.
I just retook data for the default level on the five levels with 2019-11-29b.blf-lantern.hex, its even closer. the data for the WW and NW columns is just a copy, I didn’t re-verify it, as it seems like that shouldn’t change.
| Level | ( WW ) | ( NW ) | ( Default ) |
| 1 | (0.016) ) | (0.016) ) | (0.016 ) |
| 2 | (0.158) ) | (0.158) ) | (0.132 ) |
| 3 | (0.490 ) | (0.492) ) | (0.379 ) |
| 4 | (1.01) ) | (1.01) ) | (0.922) |
| 5 | (1.80) ) | (1.80) ) | (1.84) |
That seems like it’s probably close enough. I may try to adjust the formula again sometime after receiving a production model, but it seems like what it has now should work.
Is there a post with instructions on how flash the Anduril firmware on the LT1? I’d love to play testing out new firmware or try putting my own custom firmware on it.
There is not a set of instructions specific for the LT1, but I will point you toward a helpful thread. First a few comments.
There exists a device I believe you can purchase that will interface with the programming pads visible at about 3 o’clock in the picture below. not a great pic, but I think you’ll get the idea. Lexel sells them. I’ve bought a D4 aux light board from him, he sent it quickly, and his prices are reasonable. I just haven’t ever motivated to get one of the programming keys, maybe some day. First the programming key pic, then the driver board with the pads at 3 o’clock.
I’m noticing more pins on the key than pads on the board, guessing you can/would figure that out, or Lexel could tell you.
In the event you don’t want to go that route, you will need an 8 pin chip clip to interface to the Attiny85 chip. This thread has lots of info to get you going. I used it when I began to learn to flash my D4. The only SW you need to load is the AVR Tools and the USBasp driver, unless you want to write/modify firmware, then I think ATMEL studio is useful (I’m guessing here).
Hope this helps, ask more questions as desired.
Here’s the command if you want to flash the 27b version firmware Toykeeper posted above:
avrdude -c usbasp -p t85 -u -Uflash:w:anduril.2019-11-27b.blf-lantern.hex
Would the d4v2 flashing kit work on the LT1? They look similar, not sure if the pin are in the same place.
Not quite. The pin layout is slightly different.
If you were to add one more pin to the Emisar kit though, and move a couple wires, then it would work.
Thanks! I can’t wait to play with the firmware on the lantern now. Good thing I ordered 2 of them so can experiment with the other when I get it
Does changing the firmware to make the default tint ramping current lower change the brightness in any way?
First, I’ll note that eventually I did reverify that the WW and NW levels did not change in the 29b build.
Here is what I shared with ToyKeeper regarding the change of brightness. First paragraph are notes on the 2019-11-29 build,
then the second paragraph notes on the 2019-11-29b build:
The intensity dip is pretty minor, but perceptible.
So what I never tested or commented on is how much brighter the stock build is for levels 2, 4 and 5 on the default tint setting.
In my mind the real advantage of the 29b build vs the as delivered build is it has SOS mode, and it doesn’t draw more current (shorter run times) on the default tint setting. In the big picture minor stuff.
So there’s a slight dip in brightness but not really noticeable , Thank you for the time & effort to do this
I added a chart for level 3 at the default tint to the original post.
Nice job! 16hrs at I guess would be the medium setting.
Great work GreenCampfire! :+1:
Are you still testing other levels? I can a link to this topic on the main LT1 topic so others can see your test results if you like.