Search

Micron-scale structures, including but not limited to MEMS devices, are a class of mechanical systems with a wide range of real-world applications. One important type of dynamic properties are modal, or vibrational, characteristics, which can have great effects on performance, reliability, and useful life of a system. This makes the determination of these characteristics an important element in the design and testing of these systems. The research described in this thesis addresses important challenges in experimental modal characterization of micron-scale structures, including difficulties in: applying suitable known excitations; measuring small magnitude response motions; avoiding excessive mass loading; and dealing with high natural frequencies. Two forms of experimental modal analysis are investigated, being output-only and base excitation based methodologies. In the case of output-only, an existing implementation of the Stochastic Subspace Identification algorithm, known as MACEC, was used, while for base excitation, an algorithm based on the complex exponential method was implemented. Several representative structures were tested in this research: a set of micro-cantilever MEMS-based switches, cercal mechanosensory hairs of crickets, and several lengths of fine wire, selected to have first natural frequencies in the range expected for the mechanosensory hairs. The switches and wires were examined using both output-only and base excitation methods, while the mechanosensory hairs were examined using the output-only method alone. In applying excitations, a piezoelectric stack actuator was used as a shaker for applying base excitations, while for output-only identification excitation was provided by way of the integrated electrostatic actuator for the switches, and by moving air with a loudspeaker for the wires and mechanosensory hairs. The micro-cantilever switches were found to have modal characteristics substantially in agreement with those predicted by Euler-Bernoulli beam theory for their design parameters. Notably, no significant effect on modal parameters of actuator position or size was found. For the mechanosensory hairs, behaviours significantly different from those previously reported in the literature were observed, with highly complex modes being seen. The methods used in this research demonstrate usefulness for development of biomimetic sensors, characterization of biological sensing systems, and testing of MEMS devices.