| Microelectromechanical system (MEMS) technology seeks to leverage and extend the integrated circuit paradigm by integrating components that operate in multiple energy domains on a single die. Many MEMS devices incorporate electrostatic “parallel-plate” actuators (PPAs) to implement transport or force-feedback functions. PPAs generate a high force, but they have a limited range of stable motion when driven in open loop. This unstable response severely limits their performance and use. In this thesis, we show that a PPA can be stabilized past its open-loop instability limit and can be operated with controlled stiffness and damping by use of feedback control. We present a position feedback control scheme, as well as a control scheme that feeds back the measured actuator capacitance. For each control scheme, we analyze the performance achieved with an ideal proportional-derivative controller, and we derive suitable closed-form design equations. Furthermore, we analyze the effects of a finite controller gain-bandwidth product on the closed-loop response of a PPA motor.; As part of this work, we introduce the underlying inertial plant formalism in order to derive simple stability metrics for a linearized PPA motor, and we propose to linearize the PPA motor response about the motor's static equilibrium position. These two analytical techniques yield a general PPA motor description that allows the motor's non-linear closed-loop performance to be expressed in terms of high-level metrics, and does not require strong assumptions or simplifications regarding the actuator design.; Furthermore, we discuss architectural and implementation issues which strongly constrain the design of a control circuit for a real PPA motor application. We present a MEMS fabrication technology which offers multiple-conductor mechanical structures, tight integration of mechanics and CMOS electronics, and narrow gaps for electrostatic actuators. We derive the lowest achievable noise floor for a MEMS PPA motor fabricated in that technology and used as force sensor for a data storage application. Finally, we propose two resonance loading methods that use the dynamics of a PPA motor to load the motor without requiring a large drive voltage. |
