System identification and dynamic tuning of a tunneling accelerometer and MEMS vibratory gyroscope

This dissertation presents two experimental researches on the HRL Laboratories' micro-cantilever accelerometer and the JPL-Boeing MEMS gyroscope. The HRL accelerometer utilizes the quantum tunneling effect to measure the sensor's structural deflection. To measure a small tunneling current in the order of nano-amperes, a low-noise, wide-bandwidth signal conditioning circuit has been designed and tested successfully. The signal conditioning circuit permits measurements of the tunneling current noise spectrum beyond a 100 kHz bandwidth. The packaged sensor, however, exhibits excessive parasitic capacitive coupling at the sensing pick-off, which prevents any meaningful input/output based identification at high frequencies and limits the closed-loop bandwidth. To overcome these limitations, a coupling reduction scheme with a feed-forward filter has been developed. By producing a cancellation signal via another electrode, the coupling is suppressed by more than an order of magnitude at high frequencies and an unbiased sensor response is successfully recovered.; The JPL-Boeing MEMS gyro is an electrostatically-actuated vibratory rate sensor whose optimal performance is achieved when the sensor's two Coriolis-coupled modes have equal resonant frequencies. The primary object is to develop a system atic method for tuning the sensor dynamics. The frequency split is estimated by fitting a parametric model to empirical frequency response data via a convex optimization. A steepest descent algorithm is initially developed and successfully implemented to the microgyro. Although the steepest descent algorithm is effective for tuning, the algorithm suffers from a long test duration from performing a large number of frequency response experiments for obtaining descent directions and conducting the line search. To overcome these inefficiencies, another systematic approach is developed. The other method is based on explicit modeling of the dependence of the sensor dynamics on the bias electrode potentials. This model-based algorithm facilitates tuning since the analysis of the identified model enables a direct computation of the bias potentials which yield degenerate modal frequencies. With enough freedom in the bias electrode configuration, additional criteria such as tuning to a target frequency or tuning with the smallest maximum bias potential are satisfied. The algorithm has been successfully applied to the JPL-Boeing post-resonator gyro as well.