CCITE
Creating fruitful and sustainable links between
innovative organisations committed to the improvement
of technological education for young people.
The Centre for Innovation in Technological Education in Cambridge http://ccite.org
Founder: Emeritus Professor Adrian Oldknow MA, MTech, CEng, CITP, CMath, CSci, FBCS, FBIS, FIMA, FRSA
The BBC micro:bit for Telemetry – modelling acceleration data from a bouncing elephant
Adrian Oldknow
[email protected]
20th November 2016
Ever since I learned that the BBC micro:bit had both sensors and Bluetooth I have been yearning to put them
into action for telemetry. I wanted the micro:bit to send data wirelessly in real-time to a device with a
decent display and memory to plot graphs and store a decent amount of data. Over the past couple of days I
have been finding my way around the solution invented by Martin Woolley which I read about in a regular
email update from Kitronik. I have bought several great accessories for the micro:bit from them, and have
signed up for their newsletter – which you can do on their home page. There are several sources of very
useful information which are worth regular visits. Graham’s has his own blog called That IoT Thing. This
contains a welter of information about the Bluetooth capabilities of the micro:bit including his new Bitty
Data Logger App. For the techies, the Micro:Bit Foundation has its own Bluetooth blog. Graham’s Bitty
Software site has the links for the Apple IoS and Android Apps for the Data Logger. I am using the software
with my Apple iMac Pro and Samsung S6 Android phone.
Ellie is a heavy wooden elephant mobile suspended in a doorway on a long spring. She is carrying a BBC
micro:bit attached by an elastic band. As she bounces up and down her accelerometer data is being
recorded by the sensors on the micro:bit.
The data is being sent by Bluetooth in real-time to my iPad running the Bitty Data capture App.
First you need to set up your micro:bit to use the Bluetooth to communicate with your device. I used the
editor found here on my Windows laptop in Chrome: https://pxt.microbit.org/.
When it opens a fresh
screen you will see that
there is a Radio block,
but not a Bluetooth
one.
Use the Add Package
choice from the toolkit
menu (the icon with
three bars).
Now you can swap in
the Bluetooth package
in place of the Radio
one.
The code is short and simple. It is really only just one line! The last line is “Bluetooth accelerometer service”
which tries to establish the communication with a paired device. The five lines at the start are just to give
visual feedback that the micro:bit is ready to communicate with your device.
Enter the program above. Now you can flash this to the micro:bit through the USB cable with the green
Download button. Disconnect the USB cable and connect the battery pack. (For convenience I have posted
the microbit-Acc-test.hex file here.)
Now you can use the Apple micro:bit App from Lawrence Rogers to pair your micro:bit with an Apple device ,
such as my iPad Pro, or the Android App from Samsung to pair with an Android device like my Galaxy S6. It is
probably a good idea to go the Bluetooth settings and remove any already paired micro:bits first. My
micro:bit is called ZAZIG. When you start the Bitty Data Logger App you need to Scan for a paired micro:bit.
When it has been detected its name appears, and you click on it to establish the connection. If this is
successful then the App shows “Connected” and the micro:bit displays the letter “C”.
If you check the Bluetooth devices in settings it should appear as “BBC micro:bit Connected”.
As a first test it is a good idea to record some data while the system is at rest. The image on the left shows
that the y-accelerometer, in blue, is reading a steady value of about 1g. The x- and y-accelerometers are also
showing a steady value with x (red) about 0 and z (yellow) a little below zero. This means that the micro:bit
is not completely vertical and needs its position adjusting slightly. After the correction we can start Ellie
bouncing gently and see that we get a repeated pattern in the blue y-graph, and that the red and yellow
graphs are constant at zero.
When you have collected your data you can save it on your device as CSV file, and also upload it to the cloud.
It gives you a link such as: https://file.io/OhdldZ .
If you open the link in a browser on a laptop or PC it
automatically copies it to the Downloads folder in
Windows. Double-clicking on it opens it in Excel. As
the motion is in one direction only, we just need to
study the y-accelerations. We can tidy up the sheet
by removing the top few rows and the unwanted
columns for x- and z-accelerations.
We can also insert a first column for the
corresponding times. The Bitty App has a default
sampling rate of 20ms, which can be adjusted in the
Settings option. In cell E1 I have entered the formula
“=20/1000” which produces the sample rate as 0.02s.
Right-click on cell E1 and select the option to name it.
I’ve named as “dt”. Now enter a zero in cell A2 and
create the formula “A3=A2+dt” in cell A3.
Dragging this formula down you can create the table
of time and y-accelerations, and use them to create
the scattergraph, which should look like a sine wave.
Well you can see that this isn’t quite what we’ve got
displayed. At the moment I am not sure why we
have the apparent gaps in the data collection, but at
least we can see that we are picking up sections of
the sine wave!
Excel has the facility to fit a
“trendline” to a graphed set of
data. These include quite an
impressive array of functions, but
not a sinusoid. So we could
certainly mount a micro:bit on a
model car and release it to roll
down a slope and record the
relationship between the angle of
the slope and the various accelerations – an experiment which Galileo performed.
I am a devotee of the powerful, free GeoGebra software which has extensive tools for
data analysis. You can install GeoGebra from here or run it through a browser. You
can copy the data from columns A and B. Then you can paste the data into the
Spreadsheet View in GeoGebra. The captions “t” and “y” will appear in cells A1 and
B1. Right-click on the first row and select “delete”. Click on “A” and shift-click on “B”
to highlight both columns of data. From the icon bar select the second one, and pull it
down to select “Two Variable Regression Analysis”.
If you right-click on the red sine-graph you have the option to “Copy to Graphics View”.
The final result has three views. The one on the left is the Algebra View, which show the equation for the
red sine graph which GeoGebra has computed. We would expect this to have the correct frequency and
phase shift, but it looks as we need to fiddle with amplitude and mean value. So I have used the Input bar to
enter a similar function with the mean value as the expected 1 for 1g and a smaller amplitude. This looks a
pretty good fit.
GeoGebra will perform symbolic manipulation including calculus. So we could integrate the function h(x)
once to obtain predicted velocities, and then again for predicted displacements. By taking a video clip of the
motion at the same time as data-logging accelerations we could use the free Tracker software to extract the
position data for Ellie and compare it with our predictions based on acceleration data. You will find an online
guide to using Tracker and GeoGebra here.