Spring 2022
Self-Playing Flute
Introduction
The goal of the project was to construct a self-playing flute, with the capability to perform the fingering of nine fundamental notes taught to a beginning flautist, while simultaneously providing the necessary airflow for each note.
For this project I was responsible for managing a team of three engineering students, and was principally responsible for the core concepts, primary mechanisms, electronics, and the final assembly.
Mechanism Ideation
My initial concept for the mechanical design revolved around the use of several linear actuators, relying on the electromagnetic force which occurs between air gaps in magnetic circuits. The motivation for this particular choice stemmed from the desire of having a quick turnaround time between note requests and my personal fascination with this particular form of actuation, as I have in the past had to research inner workings of this electromechanical device. For the air supply we originally envisioned to go with a series of electronically controlled valves connected to a canister of compressed CO2. We also toyed around with the prospect of creating a custom head for the flute, though this idea was never pursued past the ideation phase.
Design Considerations & Iterations - Finger Mechanism
The approach to the finger mechanisms went through the most iterations. The original approach was to attach the plunger of each solenoid to a universal 3D printed frame which would then be connected to custom “fingertips”. It was thought early in the development that precision of the contact point on the keys was critical. Later testing showed that enough adjustability was already present in the mounting brackets for us to use frames that featured static “fingertips.”
The system had to perform for an extended period of time. During our durability testing we found that the solenoids heated up enough to cause the 3D printed brackets to creep and shift the solenoids from the desired position, leading to an inability to close the valves of the flute. To address this concern, the finger brackets were dramatically simplified and nylon washers were inserted between to minimize the contact with the solenoid thus limiting the conducted heat transfer. We also included a number of small heat sinks placed directly on the solenoids.
Design Consideration & Iterations - Air Supply
For the air supply we ultimately incorporating a 24V blower fan which was controlled using PWM motor controller connected to the microcontroller. The PWM unit allowed us to vary the volumetric flow to obtain different notes. Our initial design had the blower directly exhausting on the lip plate. During testing it became clear that the excess air which was hitting the outer edge of the lip plate was producing vibrations that were canceling out the desired acoustic tones. The final iteration included a custom nozzle designed with a diverter channel to move the excess air away from the path of the desired air channel, and produced tones with an improved consistency.
Final Design & Areas for Improvement
The final design features a demo button, which can perform a preprogramed song, along with a user interface which allows one to play one of nine possible notes. When a note is requested the microcontroller sends a signal to the relay switches associated with the requested key and changes the PWM value to one that has been experimentally verified to produce the necessary tones.
If I had to attempt this project a second time I would only change two things.
1. I would begin by moving away from the fan and attempt it with the original plan of using a series of valves and compressed CO2. This would ideally fix one of the biggest shortcomings of the final product, as the time needed for the fan to adjust to the new PWM value is not negligible. This effectively undercuts the quick response time achieved by using solenoids for the fingers. Arguably this is the most necessary improvement to pursue.
2. I would pursue the variably supply current control scheme for the solenoids. Reflecting on the high thermal costs identified during durability tests, and knowing that the "gap force" is proportional to the power consumption, this would be the most interesting improvement to pursue.