CAD/Behavioral Modeling Group

The Computer Aided Design (CAD) Tools Group of the Mixed-Signal CAD (MSCAD) lab develops automated design software for mixed-signal and behavioral modeling. Such modeling strives not only to reduce the complexity of circuits and systems by describing them in a more general, mathematical form, but also to extend them with new and different effects, such as thermal, optical, mechanical and mixed analog/digital signals. The CAD Tools Group creates tools to facilitate the rapid creation of such models in a language-independent representation. From this format the tools have the ability to generate a multitude of target languages for many different device simulators. The models may be created in a graphical, non-programming fashion (e.g. the Paragon tool or ModLyng), or they may generated automatically from circuit-level descriptions (e.g. the Ascend tool). The complete effect of such tools is to greatly speed model development and to increase the amount of design exploration that may be accomplished with such higher-level descriptions.

This group is currently working on two major projects.

Past Projects

Paragon

NEWS: Paragon technology is currently being commercialized by Lynguent, Inc. Please visit Lynguent for more information (http://www.lynguent.com).

../_images/paragon_block.gif

PARAGON is a modeling package that allows a user to automatically generate model code in several Hardware Description Languages (HDLs) from a graphical interface available for Windows (and UNIX) based systems. The tool facilitates creates of an language independent model file, from which source code can be created in languages such as VHDL-AMS, fREEDA, VTB (Virtual Test Bed) and others. The user begins by creating a model interface: model name, connection points, and model parameters with ranges of validity. Next, the user creates a topology of branches which interweave the connection points and various reference nodes in functional schematic package. Equations for each branch are entered in an equation editor. Finally, a SVG-formatted (XML graphics language) symbol can be generated in the symbol editor.

The result is a package that is simple enough for a beginning modeler to use while remaining functional enough to encompass the scope of analog systems. PARAGON is a tool that should aid modelers of all levels of proficiency.

The PARAGON project is a combination of research grants from Defence Advanced Research Projects Agency (DARPA), the Office of Naval Research (ONR) and Semiconductor Research Corporation (SRC).

PARAGON versions come pre-packaged with many example models for users to browse, From simple resistors to semi-conductor and multi-domain models, these examples strive to illstrate the power of this graphical modeling concept.

The Paragon team very much appreciates any feedback and user created models. Work is currently being performed to allow direct submission of models to the Paragon model library but, for the time being, please forward any feedback and model examples (zipped model directories) to:

About Paragon:

PARAGON uses PyQt, a binding of the multi platform Qt libraries to Python, an extraordinarily powerful and simple scripting language. Paragon is built upon the following packages, libraries of which are included in each built release:

The Paragon Team:

(Developer’s areas of expertise):

  • Omair Abbasi (fREEDA)
  • Anthony Austin (Schematic editing, Statemachine editing)
  • Vivek Chaudhary (Code Generation Lead: Verilog-A, VHDL-AMS)
  • Chandrasekhar (VTB, VHDL-AMS, MAST generator)
  • Matt Francis (Interfaces Lead: Composer interface, Schematic editing)
  • Pinki Mallick (VHDL-AMS, MAST importation, MIE)

Development Tools Utilized:

CVS: Versioning Control System for software development TKCVS: Graphical frontend to CVS WebCVS: Web frontend to CVS SSH: Allows secure remote access to CVS and unix servers Gvim: Graphical Vim (ViImproved) editor Request Tracker: Handles all bug reporting and feature requests for PARAGON HappyDoc: Automatically generates documentation in Python. GUI Programming in Python - by Boudewijn Rempt: Our personally preferred text on programming in PyQt. DDD Utilized using the pydb debugging module. Cygwin Unix emulation for windows Ghost Installer Installation generation

Ascend: Algorithms and Tools for Automatic Behavioral Model Generation of Mixed-Signal Systems-on-a-Chip

As the complexity and performance criteria of analog and mixed-signal systems are steadily increased, it becomes crucial for the design engineers to be able to perform rapid higher-level simulations achieved by using behavioral models. Behavioral models are used both for top-down design and for bottom-up verification. The primary objective is to develop systematic methods for automatic generation of compact behavioral models from circuit netlists. A thrust of this project is to combine the numerical and symbolic approaches into a single, more powerful approach for model generation.
The efforts are outlined below:
  • Bottom-up behavioral model generation and model order reduction
  • Dominant pole-zero identification based equation extraction from circuit netlists
  • Determinant-decision-diagram based symbolic analysis for behavioral model generation
  • Behavioral modeling tools prototype
  • Modeling tool prototype for HDL code generation
  • Modeling tool prototype for model validation and characterization
  • Modeling tool prototype for capturing atypical behaviors (noise, thermal)
  • Modeling tool prototype for model creation and model order reduction
The project impacts include
  • Automate model generation
  • Enhance design insight
  • Reduce design time by impacting design, verification, and test

The class of circuits targeted in this investigation include building blocks in wireless communications and base band analog processing such as operational amplifiers, LNAs, power amplifiers, mixers, and phase-locked loops. Some of the primary behavioral characteristics to be generated include various circuit transfer functions such as DC and AC gains, input and output impedance. Further, nonlinear dynamics will be a major focus of these investigations.

This project is sponsored by DARPA.

Technical Papers

    1. Mantooth and G. G. E. Gielen, ?Guest editorial - Special Issue on Behavioral Modeling and Simulation,? IEEE Trans. On Computer-Aided Design, pp. 121-123, vol. 22, no. 2, Feb. 2003.
    1. Mantooth, L. Ren, X. Huang, Y. Feng and W. Zheng, ?A survey of bottom-up behavioral modeling methods for analog circuits?, IEEE International Symposium on Circuits and Systems, May 25-28, 2003.
  1. Huang, C. Gathercole and H. A. Mantooth, ?Modeling nonlinear dynamics in analog circuits via root localization,? IEEE Trans. on Computer-Aided Design, accepted for publication, Jan. 2003.

C.Gathercole and H. A. Mantooth, ?Pole-zero localization: A behavioral modeling method,? IEEE International Workshop on Behavioral Modeling and Simulation, pp. 59-65, Oct. 2001.

  1. Chaudhary, M. Francis, X. Huang, H. A. Mantooth, ?PARAGON ? A mixed-signal behavioral modeling environment,? IEEE International Conference on Communications, Circuits and Systems (ICCCAS‘2002), June 29-30, 2002, Chengdu, China.
  1. Huang and H.A.Mantooth, “Event-driven electrothermal modeling of mixed-signal circuits”, Proceedings, 2000 IEEE/ACM International Workshop on Behavioral Modeling and Simulation, pp.10~15,Oct. 2000.