Page 276 - The Mechatronics Handbook

Statistical techniques to model process variation have been included, for example, in the APLAC
tool [22], which supports object-oriented design and simulation for analog circuits. Modeling and
simulation methods, which incorporate probabilistic models, will become increasingly important
as nanoscale devices become more common and as new technologies depending on quantum
effects and biology-based computing are developed. Several current efforts, for example, are aimed
at developing a “BIOSPICE” simulator, which would incorporate more stochastic system behavior
[23].
• Is there a well-developed educational infrastructure and prototyping facilities? All the organiza-
tions, which support education and prototyping in the digital domain [3–7], provide similar
support for analog and mixed-signal design.
• Are encapsulation and abstraction widely employed? In the past few years, a great deal of progress
has been made in incorporating these concepts into analog and mixed-signal design systems. The
wide availability of very powerful computers, which can perform the necessary design and simu-
lation tasks in reasonable amounts of time, has helped to make this progress possible. In [24], for
example, top-down, constraint-driven methods are described, and in [25] a rapid prototyping
method for synthesizing analog and mixed signal systems, based on the tool suite VASE (VHDL-
AMS Synthesis Environment), is demonstrated. These methods rely on classifications similar to
those given for digital systems in Fig. 13.2(a).
• Are there well-developed models, mature tools, and integrated development systems which are widely
available? In the analog domain, there is still much more to be done in these areas than in the digital
domain, but prototypes do exist. In particular, the VHDL and Verilog languages have been extended
to allow for analog and mixed-signal components. The VHDL extension, e.g., VHDL-AMS [14], will
allow the inclusion of any algebraic or ordinary differential equation in a simulation. However, there
does not exist a completely functional VHDL-AMS simulator, although a public domain version,
incorporating many useful features, is available at [26] and many commercial versions are under
development (e.g., [27]). Thus, at present, expanded versions of MAGIC and SPICE are still the most
widely-used design and simulation tools. While there have been some attempts to develop design
systems with configurable devices similar to the digital devices shown in Fig. 13.3, these have not so
far been very successful. Currently, more attention is being focused on component-based develop-
ment with design reuse for SOC (systems on a chip) through initiatives such as [28].

13.4 Basic Techniques and Available Tools for MEMS
Modeling and Simulation

Before trying to answer the above questions for MEMS, we need to look specifically at the tools and
techniques the MEMS designer has available for the modeling and simulation tasks. As pointed out in
[29,30], the bottom line is, in any simulator, all models are not created equal. The developer must be
very clear about what parameters are of greatest interest and then must choose the models and simulation
techniques (including implementation in a tool or tools) that are most likely to give the most accurate
values for those parameters in the least amount of simulation time. For example, the model used to
determine static behavior may be different from the model needed for an adequate determination of
dynamic behavior. Thus, it is useful to have a range of models and techniques available.

Basic Modeling and Simulation Techniques
We need to make the following choices:

• What kind of behavior are we interested in? IC simulators, for example, typically support DC
operating analysis, DC sweep analysis (stepping current or voltage source values) and transient
sweep analysis (stepping time values), along with several other types of transient analysis [30].

©2002 CRC Press LLC