E-switch UI Development / FSM

That looks like a build toolchain issue, like it’s missing the AVR library headers or isn’t configured to find them perhaps.

It’s already massively sped up compared to RampingIOS, and IIRC dimmer too… and only blinks once. It’s basically using the same boot blink I used in Ferrero Rocher, just a quick blip to confirm power is connected… without causing any meaningful delay. But if you’re in a dark place and trying not to be noticed, it’s still a good idea to cover the light before connecting power.

No, I don’t think it’s quite ready for stable release yet. This is all still alpha or beta code. It’s not even merged into trunk yet.

In general, the ramp (including moon level and ramp shape) needs to be calibrated for each combination of driver type and emitter configuration. A FET+7+1 and FET+16+1 have different ramp shapes, a single XP-G2 and a quad XP-L have different ramp shapes, a raptor-claw and failboat 7135 chip have different moon levels, a 1-channel/2-channel/3-channel driver all have different configurations, an old XM-L and new XM-L2 have different moon levels, a 1-cell (or parallel cell) host and 2/3/4-cell serial host have massively different ramp values (and need different methods of measuring voltage), hosts with/without indicator LED need different configs, hosts with/without a tail clicky need different configs, etc. The full set of configurations grows exponentially with each option.

There is the approach of trying to build a .hex file for every possible config, and hosting dozens or hundreds or thousands of them in the repository, but this means an awful lot of completely untested builds and it’s expensive in terms of repository size. It’s mostly just useful as a brute-force way to test builds to make sure each configuration can at least compile. I don’t think I’ll be hosting all those .hex files though.

My preferred solution is to provide sources only, plus a very small number of stable and well-tested .hex files targeted to very specific and popular hardware — like one for the D4, one for the Q8, and one for the FW3A. For hardware produced in smaller quantities, it’s someone else’s job to build and test.

Thanks for explaining.
So if I were to put 4 Nichia 219B LEDs in my Q8, I would have to reconfigure the software on the driver to get a nice ramping?

Yeah, I’m not Mr. Covert, but in a dark house in the middle of the night it would help to have the blip either off or on moonlight. Just one of those functions I don’t require.

dekozn, a 219B quad would probably work with default settings. Those tend to max out at ~2000 to ~2500 lumens, which is reasonably close to the default ramp’s shape.

That is, assuming the driver works the same as the default, which is calibrated for an Emisar D4.

A single 219B likely wouldn’t work well with default settings, since it’ll probably max out at 500-600 lm. The top half of the ramp would go up really slowly. But I wouldn’t recommend a FET for a single 219B anyway; it’s overpowered for the emitter configuration.

anyone tried andruil in a s42s? I’d love to give it a try but changing the driver while keeping the usb-charging seems to be a pain according to lexel…

What are the parameters for “n” values in the config menu?

Are you referring to the Andúril cheat sheet?

’N’ is just a symbol for ’number of times you click the button.’

For example:

  • If you’re setting the interval for beacon mode, clicking 8 times will make the beacon blink every 8 seconds.
  • If you’re setting the ramp ceiling, clicking 20 times will set the ceiling to 130 (150 - 20).
  • If you’re setting the temperature limit, clicking 5 times will set the limit to 35 C (30 + 5).

Does this answer your question?

I got a Q8 in the mail today… and Anduril now has indicator LED support. It’ll probably change a bit in the near future, because I did kind of a quick hack, but it is at least working.

It also adds two more config settings — indicator state for regular operation, and indicator state for lockout mode. Each one can be “high”, “low”, or “off”. At some point I’ll probably also add a fourth state for “blinking”.

To configure these, go into lockout mode. Then click 3 times to step through lockout indicator modes (one step per triple-click), or do “click, click, hold” to select a main indicator mode (release when it’s at the desired level).

By default, it uses high for regular operation and low for lockout. But this could also be low/high, off/off, high/high, or whatever.

The indicator LED does not blink during use, like during battcheck, but it would probably be relatively easy to add if that’s desired.

This is great news btw! Sorry, haven't been following the progress but it sounds all good. Are you using PWM's to control the brightness? If so, is it the same typical rate, ~15K?

Have you considered or looked at external temp sensor support? Are you having the same issue of I/O pin shortages or have you looked at I/O pin sharing as Mike C has done? I really think for us to take the next step, we need to look at the 16 KB Atmel with greater I/O pin counts - this would solve a bunch of road blocks I'm having. We just need to team up with a board designer and come up with an easy way to dnld the firmware to one of these MCU's. Couple options there - not difficult to do, takes some extra real estate, but we also save real estate because of the smaller foot print of these MCU's, if we go with the square 5x5x5x5 configuration.

From this chart: https://en.wikipedia.org/wiki/Atmel_AVR_ATtiny_comparison_chart

best option seems to be the 1634, pricing/package options here: https://www.arrow.com/en/products/search?q=attiny1634

I have used DS18B20 temperature sensors with Arduino projects in the past and had great luck with them. They use the Dallas OneWire protocol to communicate and as such, only require a single I/O pin for the comms bus. Each sensor has a unique serial number so it is possible to control multiple OneWire devices via the same pin. This also means that the code in a production environment needs to be autonomous in discovering and identifying all the devices on the bus. I am not smart enough to know how heavy the library is on memory so I don’t know if it would even be possible in the smaller processors. I have been looking around at various other OneWire devices in an attempt to justify using this method but so far have not found anything terribly promising. The most interesting so far has been the DS2762 Lithiuum battery monitor. It is pretty big real estate wise and the spec sheet appears to show that the current monitor max is 2.5 Amps, but there may be a way around that. It also includes temperature monitoring so would be one device for both temp and battery.

There are also OneWire Real Time Clock chips that would allow much more precise Alarm clock modes.

goshdogit, yes that answers my question:)

TK, that is awesome!

TK might have used the internal pullup resistor.

+1
Tom, if you and TK decide on the next MCU and adapt the firmware, I believe the rest of us will follow and jump on board very quickly. I say just go for it! (On your spare time that is) :wink:

Finally! :partying_face: You had a long wait.

Awesome! I just reflashed three Q8s to play with this evening.

How about a third setting to select indicator high/low/off when the main emitters are on?

What HOLD_TIMEOUT value are you using on your Q8? I’m still using 15 without issue.

Is the low brightness barely visible or it depends on led forward voltage ? Or with a higher resistor on switch board I can lower the brightness more?

No, I set the pin as input for “low” mode. Same trick I used in Ferrero Rocher.

I haven’t really looked into external sensor support, but I’ve tried to structure code in a way which would make it easy to change between internal and external. Pin shortages haven’t been an issue yet but they probably will be in the near future. ROM space hasn’t been an issue yet either, but again, it probably will be.

If I recall correctly, we’re mostly waiting on toolchain support before going to newer MCUs.

I haven’t changed the default HOLD_TIMEOUT. I think the current value is probably pretty close to Goldilocks for the population average, but I’m kinda guessing.

Instead of high/low/off while the main emitters are on, how about a slightly different approach? Set indicator brightness based on main emitter brightness — “low” for moon to 7135, “high” for 7135 to turbo, “off” while main emitter is off? I just tried this on mine, and it seems to work reasonably well. It automatically stays in sync with the main emitters, it eliminates the need for a mid-ramp blink, and it can still be set independently for the two “off” modes (off and lockout).

No, it’s still fairly bright. It’s about 1/3rd as much power.

Hmm I want high as bright as original with narsil and a very low that I don’t see in daylight just in complete darkness. But still great option. Maybe If I change the switch board led resistor to set low mode low enough for me the high will be bright enough for daylight.