Reaction Wheel

Final reaction wheel and electronics bay

The reaction wheel electronics contain 6 main elements:

• 3S LiPo Battery
• 6DoF IMU
• Microcontroller
• Electronic Speed Controller
• DC Motor
• External FRAM chip

The microcontroller samples roll data from the sensors at 60Hz. A rotational velocity error is computed that is then fed into a PID controller. This controller determines the velocity and direction the motor must spin to create a counter torque that would null out the rocket's rotational velocity. Following this computation, a signal is sent to the electronic speed controller to spin the motor.

Because the rocket would be flying at speeds exceeding mach 2 and was likely to encounter significant vibration and loads during flight, an FRAM chip was chosen for data storage rather than an SD card. The limited capacity of the chip meant that logging should only start when the rocket begins flight. Because of this, the chip was enabled by the microcontroller once a launch event was detected using the accelerometer.

The controller was tested heuristically on a rocket simulator using the Ziegler-Nichols Tuning method with the following main procedure:

• Ki and Kd were set to small values
• Increased Kp until system oscillated
• Increased Kp until system oscillated
• Decreased Kp by half
• Increased Ki until SS offset is corrected
• Increased Kd until a loop is acceptably quick to reject disturbance

With new knowledge of control systems, I would redo this project by creating a dynamics model of the system and determine a desired response (with respect to peak overshoot, rise time, peak time). From the desired response, I would determine the desired Kp, Ki, and Kd gains as follows:

• $M_p = e^\frac{-\pi\zeta}{\sqrt{1-\zeta^2}}$
• $t_r \approx \frac{1.8}{w_n}$
• $t_s \approx \frac{4.6}{\gamma w_n}$

Second Order Plant with PID Parameterized for Kp, Ki, Kd Gains

After this theoretical calcualtion, there will likely still be an error because all aspects of the system (such as sensor noise) would not be perfectly modeled. Therefore, I would perform additional heuristic testing and tuning to achieve the desired response.

Final software update before payload integration with the rocket

Rocket launch to 47k ft with Reaction Wheel Onboard

Rocket schematic, detailing placement of wheel

During the flight, an anomaly occurred which caused the data logging to fail. It was determined to be a fault of accelerometer for detecting liftoff which enabled the rest of the system. An additional flight is planned in April of 2022 to test the system.