|CMU MEMS Laboratory Publication Abstract|
|in M.S. Thesis, May 1999, Carnegie Mellon University, Pittsburgh, PA.|
| Surface micromachined structures are composed of atomic elements like anchors, beams, and fingers, which can further be grouped into functional elements like springs, comb drives and plates. Automatic recognition of these atomic and functional elements is crucial for a structured design methodology for MicroElectroMechanical Systems (MEMS). In the structured design methodology, the schematic design is followed by a transcription of the design into a layout description. Ensuring that the layout description is a correct spatial realization of the schematic requires the extraction of the atomic micromechanical elements. Furthermore, efficient layout verification requires MEMS functional element extraction. An extraction module has been developed which begins with a layout description file and generates the netlist of the schematic corresponding to the layout. An ordinary differential equation solver combined with models of atomic and functional elements can then be used for efficient behavioral verification of the layout by simulating the extracted netlist.
Atomic elements are recognized on the basis of their shape, size and position and are classified into anchors, plate masses, beams, cantilever beams (fingers), joints and holes. This is followed by the extraction of functional elements such as springs, and electromechanical comb sensors and actuators. Comb drives are extracted using similarity in shape, inter-finger gap, electrical connectivity and locality of region of fingers. Springs are detected using a finite state machine type algorithm. A library of springs is written in a library file which is used to generate the graphs needed to match groups of beams and joints in order to recognize a spring. The utility of the extractor is demonstrated for a variety of MEMS devices composed of different types of springs and electro-static actuators and sensors. Simulation time for the extracted netlist decreased by a factor of 10 when functional element extraction and functional element models were used compared to a netlist of only atomic elements.
|© 1999 Carnegie Mellon University, Department of Electrical and Computer Engineering.|
|(opens in new window).|
|This page was generated in 0.021427 seconds at 01:28:32 pm EST on 24 Nov 2017.|
overview | projects | people | publications | intranet | resources © 1998-2009 Carnegie Mellon