Skip to content

Latest commit

 

History

History
69 lines (51 loc) · 3.12 KB

README.md

File metadata and controls

69 lines (51 loc) · 3.12 KB

Bulbdial Clock Firmware

This is a port of the Arduino sketch for the Bulbdial Clock sold by http://evilmadscience.com.

It is licensed under the same license as the original: the GPL-3, with the exception of src/timer.rs which is (presumed to be) MIT.

This was a learning project -- I wanted to learn both Rust and embedded programming, and I had the clock, so this seemed like a great idea before I began. It was a very steep learning curve.

References

https://wiki.evilmadscientist.com/Bulbdial is the home page for the Bulbdial Clock kit, with links to the original source code, the hardware designs, schematics, and design prose. Definitely read the article https://www.evilmadscientist.com/2010/on-the-design-of-the-bulbdial-clock/ to understand this device and re-read the section on Charlieplexing.

Crossreferencing the original code and the Arduino library code with the ATmega168 Datasheet (PDF) also helped orient myself with the code and schematics.

For type conversion from the Arduino C to Rust integers, I used https://learn.sparkfun.com/tutorials/data-types-in-arduino/all which indicates the size and signedness of Arduino C types.

Thanks to the Embedded Rust book, the regular Rust book, Cliffle's Learn Rust the Dangerous Way, and of course Rahix for the avr-hal crate.

millis() in rtc.rs from https://blog.rahix.de/005-avr-hal-millis/ . Crossreferencing millis() with Arduino Time.h from https://github.com/michaelmargolis/arduino_time/blob/master/Time/ to make sure I understood what was happening and which methods from the original I could discard.

For conversion of the LED manipulation code, https://docs.arduino.cc/hacking/software/PortManipulation and https://docs.arduino.cc/hacking/hardware/PinMapping168 , though I definitely didn't understand either Charlieplexing or the Arduino port bitbanging code until quite late in the game. https://en.wikipedia.org/wiki/Charlieplexing explains the technique used for lighting the LEDs.

I cared deeply about binary size because the device has only 16k of storage, and using panic-serial used up most of this. Debugging this without panic-serial is difficult difficult, but I managed with merely ufmt::uwriteln! in the right places in the end. TODO: Check out the hints in https://jamesmunns.com/blog/fmt-unreasonably-expensive/ and compare with the current state of the world.

I managed to squeeze bytes out by using the let match forms and changing mode varibles into enums, which was very satisfying, and helped refactor the code into something I could understand :D I could get more out by using the explicit wrapped_* arithmetic instead of leaving the bounds-checking default in the dev build.

TODO:

  • There's a flash at the start of each second so I think there is an integer arithmetic error in the fade calculation.
  • Buttons haven't been tested yet. The whole button handling code looks like it wants further refactoring, into a single state enum.