MUMPS Extraction Module
Extract, click here.
micromachined microelectromechanical systems (MEMS) integrating
multi-domain sensors and actuators with conventional microelectronic batch
fabrication processes are becoming increasingly complex. In order to
design systems with large numbers of multi-domain components, we need to
use a hierarchical structured design approach, with design at the
schematic level instead of the traditional layout representation used in
MEMS design. However, since fabrication can only be done from a layout
representation, an automatic or manual layout generation from schematic is
necessary. It is essential to be able to translate from the layout
representation back to the schematic to reason about layout correctness in
meeting the schematic's function as well as to extract geometric
parameters for functional simulation. An extraction module is developed
which reads in the geometric description of the layout structure and
reconstructs the corresponding schematic. This schematic can then be fed
to an ordinary differential equation solver or can be compared with the
design schematic to validate the correctness of the designed layout. The
extraction module also minimizes the number of nodes required to represent
the schematic as a netlist.
Our approach to MEMS extraction involves three steps. In the first step we read in the layout and create a unique representation (called canonical representation). We then consider each element in the canonical representation, and on the basis of size, shape, and interconnectivity, determine its function. Using these elements as seeds, we iteratively consider the remaining elements that have not yet been identified. Eventually, we merge neighboring elements that share the same function to minimize the number of elements and nodes in the extracted MEMS layout. This is then translated into a netlist, and can be solved via lumped parameter simulation.
Listed below are few cif files which can be used for demonstration.
overview | projects | people | publications | intranet | resources © 1998-2009 Carnegie Mellon