DeCiDa Data Analysis and Procedural Simulation Python Library¶
DeCiDa == Device and Circuit Data Analysis.
DeCiDa is a Python Library of functions and classes for device characterization, circuit design and data analysis.
DeCida is written in pure python (2.7), and requires no code-compilation. It is portable to any operating system where python is installed, and runs under MacOS, Windows, Cygwin, and Linux. Many DeCiDa classes and functions require the numpy and Tkinter packages.
This version of DeCiDa adds matplotlib XY-plot rendering, with clearer fonts and a more polished look. To use matplotlib, add the option -mat to dataview and plotter. This is not yet the default, since trace direction (flyback line removal) is not yet implemented for the base-class that calls matplotlib routines.
For circuit analysis, DeCiDa provides a flexible scripting class for performing simulations with various circuit simulation tools, such as (Synopsys) HSpice, (Cadence) Spectre, (Silvaco) SmartSpice, (UC Berkeley) NGspice and others. The Tckt class is used to provide a database of process corners for each project, and to provide netlist templating for performing procedural simulations and post-processing. Scripts using Tckt access the database to obtain the corner conditions, modify the netlist, and allow full python looping structures for running the simulation, viewing and analyzing the simulated data.
For data analysis, DeCiDa provides a Data class for reading-in and analyzing data in a number of formats, including nutmeg (NGspice, Spectre, LTspice), CSDF (HSpice), CSV (comma-separated value), SSV (space-separated value), and others. The Data object can be viewed using the XYplot, Histogram or DataViewx classes. DataViewx has commands via menu entries to manipulate and plot the data in different ways, including frequency versus time, eye- and scope-diagrams and column calculations.
DeCiDa started out as a Tcl/Tk application for analyzing measurements of electron devices for performing routine compact-model parameter extraction. To do this fitting, a least-squares optimization algorithm was used. This python version of DeCiDa has a function LevMar (for Levenberg-Marquardt), based on the mpfit package. It is still under development.
What is in the distribution¶
The decida python library of functions and classes (./decida). This is installed into the site-packages directory.
A test library under decida.test for testing the distribution (./decida/test). This is also installed into the site-packages directory.
Applications in the distribution bin directory (./bin). These are installed into the python bin directory.
HTML documentation of the functions and classes (./doc/html). This is not installed.
Tool-specific scripts (./etc):
simulation tool wrappers (./etc/wrappers)
Circuit simulation tool wrappers that DeCiDa interfaces with. These are installed into ~/.DeCiDa/bin
cython (./etc/cython)
Setup scripts for using cython to compile the Data and XYplotx classes. These are installed into ~/.DeCiDa/cython
dot files (./etc/dot)
Several resource files to be placed in user home directory for Cadence and python. These are installed into ~/.DeCiDa/dot
user local lib directory (./etc/lib)
A place to put user python code. This is set up as ~/.DeCiDa/lib
models (./etc/models)
Case-corners and models for two example technologies from the Predictive Technology Models web-site. These are installed into ~/.DeCiDa/models
projects (./etc/projects)
Two example project simulation directories (bird and trane). These are installed into ~/.DeCiDa/projects
Cadence skill files (./etc/skill)
Several scripts for automatically generating DeCiDa python procedural simulation scripts, and verilog test-bench environments. These are installed into ~/.DeCiDa/skill
stdcell (./etc/stdcell)
Two example standard cell libraries for the two example PTM technologies, from the NangateOpenCell Library. These are installed into ~/.DeCiDa/stdcell
verilog (./etc/verilog)
Files for running Cadence NCsim and viewing the results using SimVision. These are installed into ~/.DeCiDa/verilog
DeCiDa applications¶
All of these should be in the path after installation:
| application: | description: |
|---|---|
| calc | scientific calculator |
| dataview | read, plot and analyze data |
| plotter | plot Cartesian, Parametric, or Polar functions |
| twin | text editor, with additional capability |
| gifimg | create embedded GIF image python class from a GIF image |
| pllss | plot PLL small-signal transfer functions, S-domain and Z-domain |
| pll_phase_noise | plot PLL phase noise components and total phase noise |
| ngsp | UC Berkeley NGspice gui |
| op | read Cadence Spectre operating-point analysis, display node voltages and operating points |
| create_sspice_script | create a Silvaco SmartSpice DeCiDa script |
| simvision_csv2col | convert exported Cadence SimVision csv data to column data |
Simulation tool wrapper scripts¶
These scripts are installed in ~/.DeCiDa/bin
| wrapper script: | description: |
|---|---|
| run_hspice | wrapper to run Synopsys HSpice |
| run_ngspice | wrapper to run UC Berkeley NGspice |
| run_sspice | wrapper to run Silvaco SmartSpice |
| run_spectre | wrapper to run Cadence Spectre |
Thanks to¶
decida/ItclObjectx:
Concepts from [incr Tcl], described in chapter 2, “Object-Oriented Programming with [incr Tcl],” by Michael McLennan, of “Tcl/Tk Tools,” Mark Harrision, 1997, O’Reilly.
decida/FrameNotebook and decida/Balloonhelp:
Adapted from the Tcl/Tk examples in Mark Harrison and Michael McLennan, “Effective Tcl/Tk Programming”, 1997, Addison-Wesley.
decida/Data.read_nutmeg method:
Modified from the read_spice module from Werner Hoch (python_spice-0.0.3).
decida/LevMar:
Modified from the mpfit module from Sergey Koposov, Craig Markwardt and Mark Rivers (mpfit_2013).
bin/gifimg:
Modified from the img2pytk module from Bill Allen (imageEmbedder-1.0).
Example model files:
From Predictive Technology Model from the Nanoscale Integration and Modeling (NIMO) group at Arizona State University.
Example standard cell libraries:
From Si2 openEDA project, Nangate 45nm Open Cell Library, a generic open-source, non-manufacturable standard-cell library.
George Howlett, Michael McLennan, Sani Nassif, Mike Toth and others for developing many of the original concepts which are incorporated in DeCiDa.
Dean Gonzales, Sanquan Song and Phillip Johnson for supplying test data files and test-driving DeCiDa.
MatPlotLib matplotlib.
Known Problems¶
FrameNotebook: (test_FrameNotebook_2)
XYplot window doesn’t display at first
solution: frame_notebook.lift_tab(tab_name), shown in the test
LevMar:
still under development
MatPlotLib rendering (dataview and plotter):
trace direction (flyback line removal) not yet implemented.
solution: use dataview or plotter without -mat option
If plotting on Macbook screen in multiple-monitor environment, the window may shrink to a small size.
solution: make another monitor the primary screen.