Project Logs

Collapse

• Using node-red for advanced automation

  Danny Havenith • 06/25/2017 at 22:46 • 0 comments

  How weird, things are getting useful. This project is all about integrating powerful components of others and Node-RED is a good example of such a component:

  Read more »

• Connecting Sensors and Switches

  Danny Havenith • 06/11/2017 at 22:04 • 0 comments

  Today, two spider devices communicated with each other for the first time, and it worked! In this setup, an AVR with IR sensor detects motion at the front door and signals this to an MQTT broker. A Spider device listens to the message and switches on a light inside the house using an RF socket switch. Schematically:

  The following video shows the system in action.

  Read more »

• Switching lights with MQTT

  Danny Havenith • 06/10/2017 at 22:44 • 0 comments

  MQTT is working!

  Spider can now successfully subscribe to an MQTT topic and switch RF wall switches.

  There are still some hard-coded constants that need to become configurable in EEPROM in the future, but in essence switching the lights works as intended.

  Read more »

• esp-link MQTT software

  Danny Havenith • 05/28/2017 at 19:46 • 0 comments

  Spider uses esp-link for its wifi connectivity. This custom esp8266 firmware offers many functions in addition to plain network connectivity. The one feature that was very useful from day one was the AVR firmware update: An avr with a suitable boot loader can receive firmware updates over WiFi via the esp.

  Spider uses the optiboot boot loader. This means it's flashable through its serial port. It should even be possible to flash a Spider board from the Arduino IDE, although most programs will probably not work as expected due to the 8 Mhz clock.
  

  Another nice feature of esp-link is that it can take care of maintaining a connection with an MQTT-broker. It allows the AVR to publish messages and to subscribe to topics. The AVR and esp-link must use the SLIP protocol in order to have both plain serial communications and in-band esp-link commands on the same serial port.

  For Arduino, there is already an esp-link client library that takes care of the SLIP communications and the format of the command packets that are sent from and to the esp-link. Stubbornly, I decided that this library could not be used for Spider:
  

  • The Arduino library depends on the Arduino environment and, for instance, the Arduino serial port. This would mean that in order to use the library, I'd have to use the Arduino environment and somehow adapt for the 8Mhz clock. All in all not impossible, but quite an undertaking. Additionally I have some fears that depending on Arduino libraries would cost me more program space and RAM than I have available (Spider uses an Atmega328 or Atmega88).
  • Although the esp-link client library is well-written, it is written in C++'03. C++11 offers features that make programming for microcontrollers a lot easier and I thought it would be a good exercise for both the modern C++ muscles and the microcontroller C++ muscles to write my own version.

  In other words: I'm a walking example of the NIH syndrome.

  The library is still a work-in-progress. As always, writing my own version of this library took way more time than predicted. It has also taught me a lot about how the esp-link firmware works though, so in the end I don't consider it a complete waste of time.
  

• Hardware revisions

  Danny Havenith • 05/20/2017 at 21:53 • 0 comments

  The PCB has undergone some revisions since I've started the project:

  Version 1 was the first board ordered from oshpark:

  This board works OK, but had a few issues:

  • I had forgotten to pull down GPIO15 on the ESP. This means the module won't boot properly and this had to be fixed with a patch wire
  • I used a TSOP2438 footprint for the IR receiver, but I found out I had a HS0038. This meant I had Vcc and GND reversed. This was easily fixed by bending the pins on the receiver and adding some heat shrink to prevent shorts.
  • The micro-USB-B connector is supposed to stick out about 1mm from the board, but I had aligned it with the edge. This meant that the connector is pushed up a little at the edge.
  • The 433 Mhz transmitter was powered from 3.3V regulator. Presumably 5V would give better signal and from previous experiments I knew I could power it at 5V and still signal it from 3.3V.
  • Some weird conflict occurred between the ams1117 on my PCB and the one on the USBasp AVR programmer, which broke the 3.3V power on the USBasp. All of this happened with no power applied to the spider board.

  Read more »

View all 5 project logs