Methods for real-time estimation of parameters for high Q microsystems

High Q microdevices are used in many applications with a promise of improved stability and resolution. Often, these sensors require closed-loop control to ensure stability and minimize large overshoot so as not to compromise the benefits of a high Q element. With the miniaturization of these sensors, they are also subject to higher fabrication process variations and inauspicious degradation profiles thereby jeopardizing the stable operation of the deployed control system. This thesis presents techniques to detect changes in dynamic response characteristics of high Q micro-sensors to estimate real-time system parameters for the optimal performance of the control system. The sensor is excited using a swept-sine stimulus to estimate the system dynamic loop parameters. Due to the control requirement, the test signal energy is spread over the system bandwidth with higher weights placed around the resonance frequency for accurate quality factor estimation. Digital direct synthesis is used to generate on-chip excitation signals with system responses being resolved using a digital phase-sensitive measurement, thereby eliminating the need for additional complex circuitry while providing the added advantage of improved noise performance. Reference-based parametric techniques are employed to estimate microsystem dynamic properties with the aim of developing a reliable and optimal control system. The proposed technique can generate non-parametric and parametric system models with high noise immunity for high-quality factors. Sensor parameters are estimated within acceptable error levels, over pressure levels atmospheric pressure to about 100mTorr. Comparisons were made with mainstream methods, with the bandwidth method used as a reference. A compact, standalone digital interface is developed on a 50mm × 50mm PCB for the proposed analysis.