Week 4

Hello Dear Readers! I am back to give a quick update about the state of my project. As of this moment, I am about 95% of the way to having a final model that will be ready to present on May 17th. In this blog, I will explain what I have done and what is left to do to get my model presentation-ready. 

My Finite Element Model of the Directional Acoustic Sensor consists of two separate models, each simulating the performance of their own component. The two components have been nicknamed the rocker and the bender. 

As of this moment, the model of the bender is fully complete. It resonates in the correct frequency band (750-850 Hz) and the parametric sweep from -180 to 180 degrees shows that the amplitude of oscillation of the bender is proportional to the cosine of the angle at which the sound reaches the bender from the DOA.

As of this moment, the model of the rocker is not fully complete. I started my investigation by checking out the two parameters Dr. Grbovic suggested (substrate thickness and Gap 2). After only a few days of study, I concluded that neither of these parameters would in any way help make the directionality of the rocker sine-aligned. From here, my work became a lot tougher. 

I would not have time to experiment with every parameter, because the investigation of a single parameter would require multiple simulations testing the performance of the model while plugging in different values for that parameter. A single one of these performance tests requires running two different simulations. The first is a test of the natural resonant frequency of the chip with the new parameter, which takes about twenty minutes on average (less or more depending on how much the mesh is compressed or loosened by the given change to the parameter). The second simulation is a parametric sweep from -180 to 180 degrees at the frequency the first test outputs, which takes about thirty minutes on average (again less or more depending on how much the mesh is compressed or loosened). 

My solution to this time constraint would be to be more strategic with which parameters I chose to play around with. Modeling the rocker as a damped spring, you can easily build an Analytical Model of the system and solve a simple differential equation to determine the amplitude, frequency, phase shift, and every other factor of oscillation in the rocker system. Doing this quick math would not tell me how a specific change to the parameter would affect the system it would only give me a general idea of if any change to a parameter in the rocker system would affect the oscillation of the system. This saved me a bunch of time by helping me know in advance which parameters were completely irrelevant to the oscillation of the system and would not affect the directionality of the rocker in any way. 

Once I used this method to narrow down which parameters I had to test, I began the brute force portion of the research process. I would plug in various values for every parameter, making sure to change only one independent variable at a time so I would know for certain once the simulations were complete how a given parameter affected the directionality and frequency band of the rocker.

This is the phase of research I am in currently, finishing up this brute force testing of every relevant parameter. I will be done with this step tonight, and I will have a notebook filled with collected data on how every single relevant parameter affects the performance of the rocker. Tomorrow, I will go into the lab one final time to present my findings to Dr. Grbovic and Dr. Fabio and hear their feedback. Maybe they will have some out-of-the-box ideas about what I could try to play around with, such as using multiple materials or something else out of the norm.

I will make another blog post on Thursday explaining what happened in the lab and what the final results of my project are.

Thank you for reading and see you next time.