BSAC Research Review Individual Project
Download
Report
Transcript BSAC Research Review Individual Project
SUGAR: A MEMS Simulation Tool
http://www-bsac.eecs.berkeley.edu/cadtools/sugar
David Bindel, Jason V. Clark, David Garmire, Babak Jamshidi, Shyam Lakshmin, Jiawang Nie, Ningning Zhou
Alice Agogino, Zhaojun Bai, James Demmel, Sanjay Govindjee, Ming Gu, Kristofer S.J. Pister
Introduction:
Software architecture:
Most MEMS designers use:
Most of the IC fabrication technology
Netlist
Analysis Results
Solvers
uses mumps.net
Little of the simulation technology
transient analysis
param ox=0, oy=0, oz=0
Our goal: Be SPICE to the MEMS world
gap3dV2 p1 [b c D E]
[l=100u w1=5u w2=5u
gap=2u t1=0u t2=500u
Challenges
System Assembly
V1=5 V2=12
Device Models
steady-state analysis
static analysis
sensitivity analysiss
ox=ox oy=oy oz=oz]
Must be fast enough for design (not just
verification)
MATLAB™ Web
Must handle coupled physical effects
User Interfaces
M&MEMS: SUGAR on the Web
SUGAR in action:
Moment
arm
Torsional
hinge
Mirror
Library
Perforated
beams
Cosineshaped
beams
Actuation
direction
Recessed
inner plate
Perforated comb drive array
Torsional micromirror.
MEMS Design by: M. Last, K.S.J. Pister
Might use O(106) continuum finite elements
SUGAR system uses 2621 elements
Described with a few parameterized subnets
Measurement feedback:
Hosted on Berkeley Millennium cluster
Requires only a web browser
Java required for 3D graphics
Used in Fall 2001 MEMS introduction course
Related work:
SPICE and related network simulators
CAD/Simulation Tool
ANSYS, ABAQUS, FEAP, other finite element codes
idea
design
rough draft
NODAS from CMU
mathematic model
-
simulation
evaluation
Coventorware from Coventor
of
model
Production Tools
manufacturing
measurement
sample
-
Characterization Tools
optical measurement
KSJP15
methods
Ongoing and future work:
More and better mechanical (and other) models
prototype
Figure courtesy R. Muller.
Optical measurement facilities provided by
Muller’s Matisse group
Validate simulations by comparison to
measured data
Use measurement to improve models
Contact detection and simulation
Exploiting sparsity and parallelism
Reduced-order modeling
Sensitivity analysis
Design synthesis and optimization
Improved numerical treatment of multi-scale
multiphysics problems