DIY IoT xmas tree lights

It is holiday season, and I see a pattern with several of my friends. They have started tuning their xmas tree lights. Thus, challenge accepted. Now, I am not going for excessive amount of lights; 240 RGB-pixels are plenty for me,instead I’ll go for the tech in it.

xmas tree lights – blinke’ty blink, so what

Why would anyone have a remote for xmas tree lights? Because we can, and it is an excellent way to demonstrate Internet-of-things -aspect of these devices – I mean, everyone is expecting fridges and toasters to be in the net, but nobody expects xmas tree lights (nor Spanish inquisition) to be there. All the more reason to do it.

Also, I saw FastLED library had a nice implementation and default color palettes in the latest version, and I wanted to see how they work – the words Ocean, Forest, Cloud, Lava, Party sound pretty promising color spaces for a xmas tree light, instead of just blinking red, green, blue, yellow..younameit.

Building blocks

By an accident I just happen to have on the shelf all the basic components:

  • Particle.io Photon board – essentially an internet-of-things enabled Arduino (albeit some 10x more performing) + self-made PCB for interconnecting it with various hardware
  • A PCB for Photon with logic voltage converter for connecting with LED strip
  • Some four meters of APA 102c LED pixel string and 5V/10A power supply
  • Self-made small Cordova framework for making simple remote UI  with jQuery and jQuery mobile via particle.io cloud

Architecture

Building blocks of xTree

Building blocks of xTree

Making a IoT device that only connects to remote controller is trivial as it has virtually no scalability requirements in the back-end. Also, the back end can only handle few connections at a time, with limited amounts of data in them.

The in-use action flow is simple: user presses button, javascript sends a message to xTree. xTree responds ok, or with error code. (Note that gray areas were something I had to give away due to small  disruptions in the household – I had to leave more interesting bits away, arr.. )

Client “remote” app

Program selector

Color palette selector UI

I wrote a small library to take care of trivial setup operations: login to the messaging cloud,  and the device selector (in case user has several devices connecting to cloud – I have as they are 19 USD a piece), when those actions are complete, the client has OAuth access token to keep with the session.

Note that client misses few critical parts: I haven’t written neither the logic to carry out user registration to the cloud, nor the logic to introduce a new Photon to network. (softAP for entering network credentials and registering the device with user account) As this is prototype, I have accepted those limitations. As a workaround I can manually introduce devices during first use, with Particle’s mobile app.

The app logic is absolutely straightforward, without any clutter – html button launches a javascript function that passed a number for cloud and end device. All detailing – CSS, any extraordinary support for responsive layout have been left out. The core essence of this app is to physically show whether remote control makes sense for these lights.

The whole resulting mess is nicely encapsulated to a mobile app by Apache Cordova.

xTree device – hardware

Device contains only logic level shifter

xTree device contains MCU with only logic level shifter as external components. Supply voltage 5V is provided through LED connector. 3.3V voltage is provided by Photon.

One thing I like about this piece of hardware is it’s size, it can be housed in a match box – and it gets it’s power through the same connector as LED strip. Also the number of external components is very manageable.

The hardware is about as simple as it gets. As Photon MCU logic works on 3.3V and led strips require 5V logic, I used a simple FET-based logic level shifter circuit to drive them properly. In breadboard testing I found out that there are number of factors that may cause disturbances to those when driven with higher speeds – so even if they may work with 3.3V, there is less debugging with proper voltage matching to be expected.

xTree device – software

The embedded software structure is very basic:

  • setup – to set up libraries and communication
  • loop – to refresh LED strip some 100 times per second, and to show new pixels
  • a callback function for network communication – to show indication when color palette is changed, and to change the actual color palette

The whole device end totals less than 100 lines of code – with proper library support the actual code becomes beginner level stuff.

Recap

Making remote controlled IoT devices is not complicated. There are number of areas one needs to master to make it work: It does not hurt to know at least few programming languages, have experience on a few frameworks, understand basics of micro controllers, electronics, and physical design too. The main problem for me is not the number of lines of code it takes, for me it is all the interesting things I could make there, and the number of context switches I need to make to put it all together: Once comfortably coding in embedded C, it takes mental effort to do javascript.

Also, now I got to be green: Instead of buying new LED lights, I made them and I will probably reuse the design – with new software next year.D

Source code here – feel free to take a look, or re-use if you feel brave: https://github.com/takkulan/xtree