The test bed I built previously was an excellent model for sanity checks. However, the structure was too large, heavy, and loosely connected.

Obviously, I have 3D printed all the frames, and everything was connected tightly with bolts and nuts. Why is this all necessary?

It is about the flight controller design. The tailsitter has 2 motors and 2 elevons. Unlike the quadcopter with 4 motors, balancing in the middle of the air is not straightforward. I mean that the flight controller for the quadrotor is relatively trivial to implement. However, for the tailsitter, it is not so.

Therefore, I had to implement a very specific flight controller for the tailsitter. Yet, I do not need to start from scratch. Thanks to the researchers from the University of Toronto, the controller is well-defined.

The Phoenix Drone: An Open-Source Dual-Rotor Tail-Sitter Platform for Research and Education, Phoenix Drone)

The authors have applied a PID controller. Yet, they specifically identified the amount of angular velocity of the propeller and the deflection angle of elevons.

*The Phoenix Drone: An Open-Source Dual-Rotor Tail-Sitter Platform for Research and Education (Wu et al., 2019)*

Yet, the equations are not just the end of the story. As I have different shapes of the wing, the length between the motor and the center of gravity, and so on, I would need to measure all the parameters in the equation, such as the lift coefficient and pitch moment coefficient.

On top of these, the flight controller I'm using is PX4. Unfortunately, the PX4 project does not specify any unit. NO NOTHING! Therefore, the data that correlates the deflection angle and the PWM (Pulse Width Modulation - actuator inputs) need to be identified.

Motor Thrust Measurement