PID Test Bench

I decided to pursue this project to re-familiarize myself with control theory after a year away from Dr. Altintas' wonderful controls course: MECH 467: Computer Control of Mechatronics System. I wanted to build something that would be able to serve as a fun desk toy once complete, which meant being less than 300mm on any major axis. All the plastic parts were printed on my Prusa Mini.

There are many sources of error on this project which are worth discussing. First, the ultrasonic rangefinder (HC-SR04) is intended for planar targets and the spherical ball will have a diminished and noisier return. Second, the transmitter and receiver of the HC-SR04 are separated by around 20mm, meaning that at close ranges, the transmitter and receiver will be pointing at different hemispheres. This means that the calculated range goes to infinity. Third was the low refresh rate of the HC-SR04. While in MECH 467 we were accustomed to working with encoders with high refresh rates, the recommended refresh rate on the HC-SR04 is 16Hz. I was able to increase this to 100Hz because of the very short distances involved on this project. Fourth, while I used the straightest sections of the straightest dowel rods I could find at Home Depot, there is still warpage in the rods which will exert an unmodelled disturbance on the ball. Fifth was the servo motor. The servo has a 5us deadzone, which translates into +/- 0.45deg. In addition, the Arduino servo library accepts integer inputs. While the mechanical layout converts the +/- 40 degree output of the servo to +/- 5deg, the inherent inaccuracy of the servo made it difficult for the system to settle around the setpoint.

These sources of error combined to make a system that was unstable, despite being marginally stable in theory. With some additional tuning and filtering, I was able to achieve an underdamped but stable system.

Future upgrades include: a variable setpoint using a second HC-SR04, use of a GP2Y0A41SK0F SHARP Infrared Distance Sensor to increase accuracy (or use both sensors with a complementary filter), upgrading the actuator, and using a 2020 extrusion or metal rods for the ball's track.

With a puppy on the way this summer, I expect to be taking a break from my projects for the next little while. When I return, I plan on making a foray into RC airplanes! I am very keen on a P-38 style twin boom plane, mounting a TOPDON TC001 IR camera for SAR purposes...