Robust Design of a Micro-Electromechanical System in the Presence of Assembly Uncertainty

Three methods of minimizing error between the transient response of a dynamic system with parameter variation and its nominal open-loop dynamics are tested on a second-order example system of a piezoelectric microactuator. The example system is a piezoelectrically actuated silicon flexure intended for use in micro-robotic systems. Polymer and silicon layers are to be stacked on top of the original flexure to increase out-of-plane weight-bearing capacity, but this process is subject to substantial alignment error. The procedures evaluated in this paper seek target stiffness and damping coefficients that minimize error in open-loop actuator motion. The first method is based on simple damping ratio and natural frequency calculations, while the second and third methods are based on state-space and transfer function models, respectively. All three approaches reduce error in transient dynamics compared to nominal designs based solely on static weight-bearing or fabrication considerations, with the state-based being identified as usable to a wide range of systems, although the ability to reduce sensitivity to model variation in purely open-loop operation is limited.