I worked on the programming of the MARG. We use Ultimate Sensor Fusion Solution designed by Pesky Product. It is made up of an EM7180 Sensor Hub coupled with an MPU9250 IMU. The EM7180 outputs quaternions. We tested with the Lucas’ simulation.
We judged this model will be enough for us and bouLED. So we will definitely use it for the design of our PCB.
After several tries we finally printed a triangle with the right size for the led strip.
I also used a laser cutter but the led strip could not fit within holes again. To use the laser cutter we need someone trained to insure the safety so it was simpler for us to use the 3D printing. The main issue was the fact that we didn’t have accurate enough data. I measured with caliper and re-written the placement of holes algorithme on openscad. After several tries I printed a triangle with the right size and the data were to be very accurate (~20 micron). I took some times because each printing took 2 hours.
Lucas soldered the led strip, tied the triangle and the led strip together and we have our prototype.
Now I am working on the programming of the AHRS. We use Ultimate Sensor Fusion Solution designed by Pesky Product. It is made up of EM7180 Sensor Hub coupled with the MPU9250 IMU. The EM7180 uses a Kalman filter and provides in output quaternions.
After we made a model of our icosahedron’s one equilateral triangle using openscad. This is the first time I used 3D printer. After two hours, the first print has finished.
As you can see, the main issue was the led strip can not fit within holes because widening holes I didn’t change the spacing’s dimension between two LEDs and two rows. So the alignment doesn’t match with our LED strip. I fixed it and I tried a new print. After two other hours, the second version is here.
The alignment is more appropriate, but it’s still not perfect. The reason remains to be determined I think we should to have more accurate data (below millimeter) and if it’s possible with 3D printer available. Tomorrow I will fix and retry another 3D printed triangle.
For bouLED, we will need to know its orientation. The orientation can be represented as a quaternion but we haven’t implemented Madgwick’s algorithm yet. So, I wrote a python script to simulate a rotation. The script generates and writes a quaternion on the script standard output.
In fact the sphere will be an icosahedron. It‘s the regular polyhedron made of 20 equilateral triangles. So I drew a equilateral triangle on openscad in order to print it in 3D. Then we will stick on this triangle a LED strip, and we will judge if the resolution is high enough.