Experiments with a Compass Module
https://www.amazon.com/dp/B01LXY7CM3
https://www.adafruit.com/product/5129
https://github.com/CytronTechnologies/MAKER-PI-RP2040
Writeable filesystem:
https://learn.adafruit.com/circuitpython-essentials/circuitpython-storage
https://www.amazon.com/HiLetgo-LSM303DLHC-Compass-Accelerometer-Magnetometer/dp/B07X3GFKYD
>>> import machine
>>> i2c = machine.I2C(scl=machine.Pin(5), sda=machine.Pin(4))
>>> i2c.scan()
[25, 30]
>>>
Hex 19: 00011001 (Linear acceleration)
Hex 1E: 00011110 (Compass)
https://github.com/adafruit/Adafruit_CircuitPython_LSM303DLH_Mag
https://www.amazon.com/dp/B00EERJDY4
# py -m pip install matplotlib
import json
with open('data.txt') as f:
data = json.load(f)
import matplotlib.pyplot as plt
fig = plt.figure()
ax = plt.axes(projection='3d',xlabel='X',ylabel='Y',zlabel='Z',title='Compass Spin')
ax.scatter3D(0,0,0,c='blue',s=100)
plt.show()
These graphs were made by spinning the compass module in a circle about the gravity vector. The Z axis of the compass is parallel to gravity. The (x, y) plane is parallel to the Earth.
Made with plot_base.py.
The motors have magnets (and iron) in them. How do the motors affect the compass reading? Let's spin the compass/controller/batteries with and without the robot base attached.
Interestingly, the motors bend the readings closer to ideal (zero slope in Z).
Made with plot_mounted.py.
How does a magnet attached to the robot affect the compass readings? How does a chunk of iron attached to the robot affect the compass readings? Let's spin compass/controller/batteries with iron attached.
With no magnet and no iron, the readings (black curve) are offset from center. This is caused by hard and soft iron built into the compass/controller/batteries. Mounting screws and header pins contribute to this offset.
The magnet pushes the readings (red curve) away from the origin. The iron (blue curve) does too, but less noticeably.
Made with plot_externals.py.
How does a magnet placed near the robot (but not attached) affect the compass readings? How does a chunk of iron placed near the robot (but not attached) affect the compass readings?
The iron I used is a square tube stood upright along the Z axis. Notice how the blue curve is distorted along the Z axis.
The magnet wreaks havoc with the readings (red curve).
Made with plot_motors.py.
When a motor is running, the current creates an electromagnet within each motor. Does this affect the compass readings?
Let's run the motors at full and half power both backwards and forwards.
The Z-axises of these curves have been tweaked to separate the curves visually. Just looking at the (x, y) values we find that all of the curves are nearly identical. The readings are affected the same whether the motors are running or not.
Made with calibrate.py
The Z slope is slight. We will ignore the compass readings on the Z axis. The (x, y) offsets are easy to calculate and adjust for. The distortion, or squashing, of the circle by soft-iron is harder to correct for. We'll use the "simple" correction factor because it is computationally cheap.
Notice that the final (black) curve is not a perfect circle. The "simple" correction algorithm leaves a little error, but it is close enough to perfect for our use.
TODO - REPEAT THIS EXPERIMENT. Use the line of LEDs as an indicator of facing north and compare to the compass.
I tried to mount the compass so that forward for the robot is right down the Y axis. Actually, negative on the Y axis because of how the circuit board is laid out.
I pointed the robot to head straight north and took samples. I would expect the X,Y-to-heading math to result in 180 degrees. Instead my calculations say 160, which agrees with the plotted data:
That's close. Is my error in mounting the compass or lining the robot up pointing north. I'll test driving to 160 and to 180 and see which matches the actual compass better.