This repository contains the software designed to run a wireless telemetry system for the VU Motorsports team's formula car.
The hardware consists of a BeagleBone embedded Linux development board connected to a wireless router and various sensors.
Currently, the only sensors fully implemented are IR temperature sensors for measuring tire temperature, but the system has been designed to be easily extensible. Instructions on adding sensors can be found later in this document.
Before starting, the router should be powered up and the BeagleBone connected via Ethernet to one of its four switch inputs.
The first step is to power on the BeagleBone. This can be done either from the labelled 5V power adapter or from the USB (mini B) port.
Once the green power LED lights up on the board, you know it is recieving power. Soon afterwards, the green user LEDs on the opposite side of the Ethernet jack should begin blinking, signaling that the board is in the boot process. The board should boot within about one minute, so give it some time before proceeding to the next step.
From time to time, the board will show that it has power, but not boot correctly (the user LEDs will not begin to blink) This is easily fixed by cycling power while the small reset button next to the Ethernet jack is depressed.
After booting successfully, the next step is to access the device via a terminal (Terminal.app on OS X, for example). The router is configured to create a wireless network with the SSID "telemetry" and give the BeagleBone the static IP address of 192.168.1.101 on that network. Therefore, from a computer connected to that wireless network, running the following terminal command:
and hitting Enter at the password prompt should successfully log you in to
the BeagleBone.
Once you have console access, running the following command will start the telemetry server:
node server.js
Once the server is running, the telemetry dashboard should be accessible to any computer connected to the telemetry wireless network by typing "192.168.1.101:8080" into a web browser's address bar.
Please note that an up-to-date version of Google Chrome is the recommended browser to make sure that the page displays correctly.
Above is an abbreviated version of the current schematic. The wires at the bottom continue to the other two attached temperature sensors and make up the entirety of the I2C bus and power connections for the sensors.
The connection to the BeagleBone is as shown below, with the yellow wire being SCL and the green wire SDA:
All of the pin definitions can be found in the official BeagleBone System Reference Manual.
The attached temperature sensors are MLX90614 digital infrared thermometers. Their datasheet can be found here.
They speak I2C natively, so they simply need to be hooked up to 5V, ground, and the two bus lines, each of which can be connected directly from the BeagleBone. The SDA and SCL lines also require 4.7K pullup resistors, as shown in the schematic. Below is their pinout:
One important thing to note is that each of the sensors came form the factory with the same I2C slave address, so they had to be manually reprogrammed with different addresses. When adding devices to the bus, it is important to check their default addresses and make sure there is no current device on the bus with that same address. The current sensors are 5a, 5b, 5c, and 5d, but there are some system devices sharing the bus as well.
The software running on the BeagleBone consists of an application software with two distinct components, a static file server and a WebSocket publish/subscribe server.
The static file server is based off of the node-static module and handles serving all of the files in the public folder, including the HTML, CSS, and Javascipt files used to build the dashboard page.
The publish/subscribe server is based on Faye and uses WebSockets to allow pushing data in realtime to the web page.
Our application server loops over the array of available sensors (described next) at a given frequency, reading their values and publishing the data to over the appropriate channel.
The client web page uses Javascript to subscribe to these data updates, adding data to the relevant graphs in realtime, without any need to refresh in the browser. The graphs are made with help from the Smoothie.js charting library.
Available sensors are defined in the sensors.js file. Each one has a name, system command that returns the desired value, parser that parses the output of the system command into the desired value, and a units suffix.
The fact that the value comes from a system command means that sensors are very loosely coupled from the server, so separate programs can be written and called that gather the data, without ever adding anything to the server.
All that needs to be added is the relevant sensor information, the appropriate HTML in index.html to provide a cavas for drawing the graph, and the declarations of the graph and timeseries in public/js/script.js. The existing sensors should provide a very easy-to-follow template for any future sensors.
- 1 BeagleBone
- 4 MLX90614 temp sensors
- 2 4.7K resistors