A few months ago I came to the realization that I would have to abandon my bid for AVC this year. A whole bunch of issues came up in my work and private life that made it impossible to work on the project. I had half considered flying out just to watch the competition, but the expense of it just did not seem to be appropriate. But that does mean I have a full year to get get ready for the next AVC!
And along those lines, I have gotten some time to get back working with my entry. Over the last few days I have been characterizing the magnetometer in the Pololu 9Dof IMU. I built a 2 foot diameter non motorized turntable with a one degree accurate scale. I put the IMU on the turntable and recorded 20 data points at every 10 degrees. Plotting the X and Y data gave a very nice sinusoidal 90 degree out of phase plot. Interestingly the sensitivity of each axis seems a little different and not centered on zero.
The next test was to to the same data collection, but this time with the electronics board on the RC car. Again the data was nicely sinusoidal, but this time with a significant offset in the X axis. The X axis is aligned with the motor and I suspect the permanent magnet in the motor is causing the shift. It is also possible the battery or other ferrous metals are to blame.
The next test will be to collect data with the motor powered up. My guess is that all hell will break lose. Part of it from pure electrical noise, but mostly from all the magnetic flux kicked up by the motor. After this test will be one to determine how far away the IMU needs to be from the motor and will a MuMetal shield have any effect.
I also managed to get down to Boston last weekend and stopped in at You Do It Electronics in Needham. Fun store and I spend two hours browsing and came away $80 poorer. Amazing how a small bag of doodads added up that fast.
Curious to see if your findings are the same as or contrary to mine regarding motor’s effect on magnetometer.
There is sooooooo much to go wrong with the magnetometer! Any ferroud metals, and magnets, any current loops will distort the field. Using the mag in combination with an accelerometer to give you a down vector is also a problem when you may not be moving at a constant speed. So far I have spent lots of time characterizing the magnetometer in a friendly environment. I am about to start the harsh environment now that I can get data off the robot using a Bluetooth transceiver. The real slowdown I am having right now is that I have outgrown the Arduino testbed and have not spooled up on the ARM processor I got to replace it.