Microsystem CAD: From FEM to System Simulation

Microsystem technology is a highly interdisciplinary area. Therefore, a combination of different CAD methods and tools is necessary for supporting microsystem design. Process and device simulation are basic CAD methods but more and more higher levels of abstraction need to be applied in order to analyze microsystems adequately. This paper summarizes several modeling and simulation strategies for system simulation of microsystems on different levels of abstraction: generalized Kirchhoffian networks, black-box models, macromodels, the application of hardware description languages, and simulator coupling. 1. Abstraction levels in microsystem design Computer-aided modeling and simulation are very topical in the design of microsystems. Other powerful CAD tools may be used in the construction process e.g. for handling geometrical data and design rule checking. Only first approaches exist for other tasks like specification [40], synthesis [37], and optimization [24]. Modeling and simulation methods depend strongly on the abstraction level. Basic mathematical description means for heterogeneous systems are partial differential equations (PDE) and ordinary differential equations (ODE) for continuous systems. Discrete-event systems (DES), Finite State Machines (FSM), PETRI nets or other automata models are suitable for digital or time-discrete systems: System level ODE + DES + FSM + ... Subsystem level ODE, DES, FSM, ... Device level PDE Process level PDE Digital systems are not taken into account in this context because they are covered by traditional CAD methods which are not specific to microsystems. Process simulation is not included in this paper either. Well-known programs like SUPREM support this technological design step. On the device level, the numerical solution of PDEs is also the basis of simulation and may be carried out with simulators like ABAQUS, ANSYS, CAPA, FLOTRAN, NASTRAN, SESES, ... . We have to start here with the development of system models. On the next level of abstraction, ODEs are commonly used as mathematical models in continuous subsystem simulation. It depends on the simulator and on the subsystem’s physical domain which of the mathematically equivalent representations of ODE’s will be given preference: • mathematical expressions Mathematica, Maple, Macsyma,... • block diagrams, signal flow graphs Matlab, MatrixX, Dymola, ACSL, Simplorer,.. • multi-body systems ADAMS, NEWEUL, ITI-SIM,... • electrical (or non-electrical) networks SPICE-related electronic simulators