Thanks. Another quick question, are all of the eight pins assignable? It seems like pins 4 + 8 are always ground and positive. Is it the other six pins that are assignable? This has always confused me.
Yes. This is the chip we’ve been using:
Yeah, when I saw that I could not identify anything other than vcc and ground. The rest is Greek. Lol
The parts needed for PWM are OC0A, OC0B, OC1A, and OC1B. So it should be able to do 4 channels. But two of those are on the same pin, so it’s really only three channels.
Vcc and Gnd supply the power, are hardwired to the silicon.
Leaving six configurable pins. Now that we have an e-switch, switch LED, three LED driver banks, and possibly an Aux. LED board to control (though that can run autonomously from its own MCU in Lexel’s proprietary design), we have run out.
Some pins are reconfigured during flashing, which has to be considered, so anything attached to them doesn’t interfere with the process.
The ATiny85 is now a very old design (2005 introduction).
Since then, Microchip took over Atmel (2016) and have developed the platform, also introducing technology from their PIC designs. Tiny leadless SMD packages allow much higher pin counts (though are more challenging for DIY assembly.)
This is where future development will have to go, and from where I would be starting, as a new developer.
Legacy code bases can be ported, and design environments updated. Already happening, e.g. see Adventures in TinyAVR 1-Series
TK, Texas_Ace has posted some worrying information about his experiences with Anduril thermal control on the MF01S, hence his decision to stick with Narsil.
Do you have any comment, considering that the D18 is quite similar in scope ?
She addressed that question here:
Strictly it only needs one PWM channel, since only one thing is PWMed at a time. But because of the way the pins are configured, at least two are actually required.
Mike C uses a different topology, as I understand it only a single 7135 is PWMed. Others are switched in in binary sequence without PWM to provide a smooth ramp and potentially better efficiency, but this requires more pins, hence he has moved on to newer devices.
Led4power has also developed his own design, using a FET operated in linear mode, not using 7135s.
There are many ways and other ideas how to do all this, meanwhile this sort of design, and firmware from a couple of active coders, is what we have.
I agree with the 4500K. It’s buy far the most useful tint in the outdoors. Where can we go for more info on this light?
Hello TK, I did observe PWM on both level 50 and 100, on my D18. And after some experiments, I found that level 65 and 120 were probably the correct levels you mean. Can you help me confirm that?
I believe that is exactly how it is programmed
Turbo by HOLDING the button = 70° temp ceiling
Turbo by Holding the button for 6s, then releasing the button = 50° temp ceiling
the reason behind this being, that as long as the light is in your hand you can react if it gets too hot. If it is locked turbo but not in your hand, you do not want it to be 70° when somebody picks it up.
The D18 hasn’t even started shipping yet. I sent Hank the final firmware just four days ago. Are you thinking of some other model?
Yes, I am familiar with the Meteor’s UI.
This test used UI 3. It has “turbo 70C” while physically holding the button, or “turbo 50C” while not holding the button. I was not holding the button during the 76-minute-long test. So it should have had a limit of 50C, but in reality it was much hotter.
Both lights had the same temperature limit in theory, but one was too hot to hold while the other stayed at a comfortable temperature. I assume this was due to poor calibration on the M43, because that is a known issue with the attiny85 chip and the M43 has no function to work around that issue.
No, I’m (probably) the first one who has received a purchased D18 
. So I did mean D18…
Pics please or it have not happened 
did you do next day shipping from china or something?!
@Tom Tom
@Tom E
Thanks for all the information. I’m quite interested in the topics especially since I understand how many mechanical things work but haven’t delved into electronics before. Simply learning how to do non-complex electrical work was enough for me so far. Lol as you may have predicted most of your responses left me looking like this:
Nevertheless, you also provided some great resources for my drivers 101 course which I’m excited to dig into! Thanks for the help!
Lol When I tell people I’m a member of a flashlight enthusiast forum the looks I get are priceless. Then I start by explaining the experts on here are actually consulted or contracted to design components or entire lights with input from the community and people are like wow a market that takes direct feedback from the consumer? Then when I tell them I’m learning about electronic components, circuit boards & drivers, they look at me like maybe it could make sense and then they laugh and say I have a fetish for flashlights. lol the world will never understand us but I find it fascinating how much I’m learning here.
Interesting.
If it has the non-PWM levels at 65 and 120, that would be the 2019-01-05 firmware… produced without even having much idea what the hardware would be. If you have the ability to do so, it might be worth updating the firmware at some point. I should have the code uploaded soon; was waiting to make sure it’s okay to publish first.
The versions are:
- 2019-01-05: No-PWM levels at 65 and 120. Turbo step-down 120. Super early version meant only for testing during hardware development. (and, apparently, for toobadorz)
- 2019-03-18a: No-PWM levels at 55 and 110. Turbo step-down 120. Testing only.
- 2019-03-18b: No-PWM levels at 50 and 100. Turbo step-down 110. Testing only.
- 2019-03-21: No-PWM levels at 50 and 100. Turbo step-down 125. Intended for production use.
I had a bit of difficulty flashing mine; the attiny85 wouldn’t respond. I’m guessing it may have accidentally had the lock bits set or something… so I removed it and put a fresh chip on to finish development. That shouldn’t be an issue for production lights though; probably just a typo when putting together an early sample unit.
Also, when taking the driver out to reflash, it’s important to be gentle with the switch wires. Unsolder the LED wires, push them through the MCPCB, lift the driver a little on the edge opposite the switch, then flip it upside-down by rotating it around the switch wires. This should make the MCU easy to access, without tugging on the short delicate wires still attached to the switch.
Hmmm. Toobadorz, can you tell us more about how you received your D18 so early? I placed an early pre-order and just received a shipping notice today. I hope Hank is using TK’s newest firmware from 3/21 in all the production models.