Using the accelerometer


Unless you want to do some very fine precision soldering, you better buy the evaluation board with the accelerometer on it. This will have pins with a DC voltage input, ground, output pins for each axis your accelerometer measures and a self test pin which you don't really need to worry about. You will also need to make something to hold the evaluation board so you can attach it to whatever you're studying. You can buy accelerometers which are encased and ready to plug in, but it will cost you an extra $200 or so. I made my holder with plexi-glass and hot-glued the board to it. This allowed the kiddies to actually see the accelerometer.

You need a power supply for the accelerometer board. The power supply limits the portability of the sensor as it must be directly connected even if you are using wireless technology. I used a 9V battery and then used a step down converter to change it to 5V. The accelerometer I used will accept anywhere from 2.4 to 6V input. The output from the sensor will be proportional to the input (ie. +/-18g over 5V means that 0g will be somewhere around 2.5V). The easiest way to calibrate the sensor is to find +1g and -1g on each axis by turning the sensor around relative to the earth's gravitational pull. From there you can produce a formula which will convert the voltage to m/s^2 or g's. An axis will read zero g's only if it is in the plane perpendicular to gravity. You can also use this property to measure tilt. Besides the power supply you will also have to find a way of logging the data. I used hardware by picoscope that connects to the computer directly, essentially creating an oscilliscope and data logger within the computer. Unfortunately the picoscope I used is only sensitive up to 2.5V DC so I had to add some resistors to bring down the output voltage. I also had to solder and assemble shielded cables to connect the computer, power source, and accelerometer.

If you want to find the velocity and displacement from the accelerometer, the easiest way is to use the basic kinematic equations for velocity and displacement (V=Vo + at), ( X = Xo + Vot + 1/2at^2 ). All you have to do is apply each acceleration reading sequentially and add it up. This is quite easy using excel. Of course, this is an approximation, and you cannot know the initial conditions so you must begin with Xo and Vo equal to zero. For very short time periods we can assume that the acceleration does not change. For small data sets, this works perfectly well. I made an excel sheet into which I can simply copy the data (time in ms and acceleration in m/s^2). From there, one can see trends in the error and correct accordingly, or selectively choose the most consistent data.