Joel Neff and Dr. Brian D. Jensen, Mechanical Engineering
The idea for our project was to use vibrations from the human vocal folds to harvest energy during speech. This energy would then be used to power a wireless throat microphone, which would acquire a speech signal and then transmit it to a fixed receiver for communications systems. The measurements we took estimated that human vocal folds could produce vibrations with amplitudes on the order of 1-5 m/s2 at the skin surface on the throat. One research study done by Shad Roundy at UC-Berkeley demonstrated that a device capable of producing 300 µW of power could be driven off of vibrations of only 2.25 m/s2 at 80 Hz. The device occupied a space less than 1 cm3.
The study also showed that the amount of energy that could be generated was proportional to the size of the device. Based on this assumption, a low power wireless microphone could potentially be powered from the vibrations of the throat during speech, if the size was large enough and the energy from the throat could be effectively converted without causing discomfort or injury.
The first part of the project involved developing an analytical model of the system. Specifically, we wanted to know how much energy was generated in the human vocal folds as they vibrated, and how much energy could potentially be converted through a piezoelectric device. After the analytical model was developed, we could then design hardware that was capable of harvesting the energy, and then design a circuit that would efficiently transmit the speech signal. The wireless throat microphones we had available to us used a piezoelectric disc on a membrane to generate a speech signal directly from the throat, so the microphone technology was not a major issue. The issues were energy conversion and signal transmission. Testing of the different concepts would hopefully lead to an optimized design that could be used to demonstrate the feasibility of such a product.
The development of the analytical model was the first challenge that we faced. The constitutive equations of piezoelectrics needed to be combined with a mechanical system to predict a power output. Unfortunately, the nature of the mechanical system greatly affected the effectiveness of the piezoelectric energy conversion. Because piezoelectric materials exhibit different properties based on the type of material, the direction that they are stressed, and the frequency at which they are stressed, it is impossible to predict power based on an arbitrary geometry. Only with a specific design to analyze will any predictions be useful. Therefore, we decided it would be useful to develop an analytical model for a specific design. Therefore, we decided it would be useful to develop an analytical model for a specific design. The design that we chose was a cantilever beam design. The “beam” was actually a piezoelectric bimorph, or a thin brass shim sandwiched by two layers or PZT piezoceramic. A small mass was attached to the end of the beam and the other end securely attached to a vibration source. This design was used in a study by Shad Roundy at UC-Berkeley. Our objective was to put together a model of this system and then test it to see if it would be accurate. Roundy derived an analytical model that gave a predicted power output for the design with parameters such as beam width, piezo thickness, acceleration amplitude, etc. However, it also has variables that were difficult to determine, such as damping. Therefore, we set out to try and re-derive the power equation in order to understand the logic behind it. An effort was first made to verify the values that Roundy predicted in his model. This resulted in frustration as we could not get our answers to match with his. Based on the information he gave in his book and papers on the subject, we predicted power outputs significantly different than his. Although he gave specific dimensions and information about his setup, some values such as damping were based on assumptions. However, even inaccuracy in our assumptions could not explain the large discrepancy in the predicted values.
We spent several weeks troubleshooting, although not much was found. We finally went back to the analytical model, and spent more time trying to derive the equations that Roundy got. We made progress in this area, in part thanks to our familiarity with the project due to our experiments and the time we had spent on it. However, we still had many unanswered questions and simply not enough experience with piezoelectrics. Rather than spend more time being frustrated and making little progress, we decided to scrap the project for the time being and focus on another project.
The piezoelectric energy generation was the main focus of our research. Without power, the wireless microphone was simply not possible. Therefore, the rest of our project depended on the success of the power source. However, we never really looked into the best way to attach a vibration-based generator to the throat, or which data transmission to use. Another problem was trying to vibrate a piezo generator at resonance when human speech has a very large range. All of these were problems we just never got to. In theory, the power generated by a piezo device should be sufficient to transmit data, as shown by Roundy’s project, and the amplitude of vibration in the human throat should be sufficient as a vibration source. However, making that piezo device work on the throat at a wide range of frequencies is still quite a challenge.
If work on this project continues, it is recommended that more study be done on the theory of piezoelectricity, so that the researchers can understand what will maximize power in piezo generators. It is also recommended that more research be done on how to attach the power generator to the throat to maximize vibration amplitude. Further, some sort of active tuning system will probably be necessary if the device is to work at a wide range of tuning.