The official PySpice Home Page is located at http://fabricesalvaire.github.io/PySpice
The latest documentation build from the git repository is available at readthedocs.org
Written by Fabrice Salvaire.
PySpice is a Python 3 library which interplay with Berkeley SPICE, the industrial circuit simulator reference.
SPICE (Simulation Program with Integrated Circuit Emphasis) was developed at the Electronics Research Laboratory of the University of California, Berkeley in 1973 by Laurence Nagel with direction from his research advisor, Prof. Donald Pederson. Then Spice emerged as an industrial standard through its descendants and is still the reference 40 years later.
PySpice is born as a personal project to relearn electronics where circuit simulation is a part of this goal. Since I use the Python language every day, I quickly feel the need to plug SPICE and Python.
The aim of PySpice is to address several needs:
- SPICE language is fine to describe circuits, but it lacks a real language for circuit steering. By contrast Python is a powerful high level, oriented object and dynamic language which is perfectly suited for steering and reusing circuit. But it comes at the price its more general syntax is less fluent than SPICE for circuit description.
- Ngspice provides some extension to Berkeley SPICE for data analysis, but its interactive environment or TCL module are now outdated. By contrast Python has scientific framework like Numpy and Matplotlib that compete with Matlab.
- Ngspice source code is derived from Berkeley SPICE and thus has a very old basis. Moreover the sources are poorly documented. So it is really difficult to understand how it works and modify it. PySpice could serve to easily experiment extension.
As opposite to other SPICE derivatives, PySpice focus on programming and not on graphical user interface. Thus it doesn't feature a schematic capture editor and we cannot pickup a node or an element and plot the associated waveform. Moreover we can notice the Modelica language treats diagrams as annotations. A choice which I consider judicious. Thus we can imagine to follow the same principle and extend PySpice later.
The main features of PySpice are:
- actually PySpice only supports Ngspice
- an oriented-object API to describe circuit in a way similar to SPICE
- a library and model manager that index recursively a directory
- an incomplete SPICE parser (mainly used for the library and model indexer)
- a circuit can be simulated using a subprocess (aka server mode) or using the NgSpice shared library, NgSpice vectors are converted to Numpy array
- the NgSpice shared library permits to interact with the simulator and provides Python callback for external voltage and current source
- some data analysis add-ons
Since PySpice is born with a learning goal, many examples are provided with the sources. These examples could serve as learning materials. A tool to generate an HTML and PDF documentation is included in the tools directory. This tool could be extended to generate IPython Notebook as well.
These features are planned in the future:
- implement a SPICE to Python converted using the parser. It could be used for the following workflow: quick circuit sketching using Kicad > SPICE netlist > spice2python > PySpice which could help for complex circuit.
- implement a basic simulator featuring passive element like resistor, capacitor and inductor.
- implement a Modelica backend. Modelica is a very interesting solution for transient analysis.
Users should be aware of these advertisements.
Warning
The API is quite unstable until now. Some efforts is made to have a smooth API.
Warning
Ngspice and PySpice are provided without any warranty. Thus use it with care for real design. Best is to cross check the simulation using an industrial grade simulator.
Warning
Simulation is a tool and not a perfect representation of the real world.
The installation of PySpice by itself is quite simple. However it will be easier to get the dependencies on a Linux desktop.
PySpice requires the following dependencies:
- Python 3
- Numpy
- Matplotlib
- Ngspice
- CFFI (only required for Ngspice shared)
Also it is recommanded to have these Python modules:
- IPython
- pip
- virtualenv
For development, you will need in addition:
- Sphinx
- circuit_macros and a LaTeX environment
Usually Ngspice is available as a package in the major Linux distributions. But I recommend to check
the compilation options before to use it extensively. For example the Fedora package enables too
many experimental codes that have side effects. The recommended way to compile Ngspice is given in
the manual and the INSTALLATION file. Ngspice is an example of complex software where we should
not enable everything without care.
Warning
For the following, the compilation option --enable-ndev is known to broke the server mode.
PySpice is made available on the PyPI repository at https://pypi.python.org/pypi/PySpice
Run this command to install the last release:
pip install PySpiceThe PySpice source code is hosted at https://github.com/FabriceSalvaire/PySpice
To clone the Git repository, run this command in a terminal:
git clone git@github.com:FabriceSalvaire/PySpice.gitThen to build and install PySpice run these commands:
python setup.py build
python setup.py install