Automated extraction of multi-energy domain reduced-order models demonstrated on capacitive MEMS microphones

We present a methodology to systematically extract efficient and physically-based reduced-order models based on a mixed-level simulation approach and demonstrate its practicality for the design of a capacitive MEMS microphone. The method has been implemented in a MATLAB toolbox which starting from a FEM discretization, enables the automated generation of mixed-level VHDL-AMS based macro-models, which can be straightforwardly fed into a standard circuit simulator. The extracted models are highly efficient and moreover, in contrast to other equivalent network approaches, physically-based, thus allowing for the predictive simulation of microstructures with complex geometry.