Ph.D. Dissertation Defence (June 4th, 2007) - Manu Agarwal
amanu at stanford.edu
Thu May 31 09:41:53 PDT 2007
Stanford University Oral Exam: Manu Agarwal,
Department of Electrical Engineering
Date/Time: June 4th 2007, 9:30AM
Location: Packard 101
Thesis title: Impact of Nonlinearities and External Accelerations in Electrostatic MEMS Resonators
Electrostatically transduced Micro-Electro-Mechanical-System (MEMS) resonators are being viewed as a viable replacement for quartz crystal resonator technology for the multi-billion dollar timing and frequency reference market. Single-crystal-Si MEMS resonators have been shown to exhibit frequency stability over temperature and over long time periods (long term stability or aging) that is comparable to mass produced state-of-the-art quartz crystal resonators. Several efforts to commercialize this technology are already underway. In this work, we have investigated some of the mechanisms that limit the phase/frequency noise performance of these resonators.
Sensitivity of resonant frequency to acceleration leads to frequency noise, as random environmental vibrations (random accelerations) are always present in real world applications. We have investigated the acceleration sensitivity in a double ended tuning fork resonator and have found this sensitivity for our design to be comparable to quartz.
The impact of the noise from the electronics has also been investigated. One of the biggest challenges of this technology, compared to quartz, is to achieve large signal current handling capability. The signal current handling is limited by force nonlinearities in the resonator. This limitation causes reduction in the achievable signal-to-noise-ratio (SNR), thereby limiting performance. Several mechanisms are responsible for force nonlinearities, such as structural stiffening and electrostatic softening. We have developed lumped nonlinear analytical models for these resonators, and have verified them experimentally. Using these models, optimization of the resonator parameters to increase the SNR has been shown. Impact and tradeoffs in scaling of physical parameters like size, frequency and parallel plate gap size have been investigated. We find that several conclusions are at departure from recommendations that come from linear modeling of the device, and these discrepancies are discussed. The developed nonlinear models help better understand the interaction of the mechanical devices with electronic circuits, and aid in developing high precision MEMS-based frequency reference oscillators.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the labmembers