circuitgenome 0.2.0

A Python toolkit for analog circuit topology synthesis and recognition, focused on op-amp design.

Modular synthesis, recognition, and sizing of analog op-amp circuits.

CircuitGenome is a Python toolkit for analog circuit topology synthesis and recognition, focused on op-amp design. It takes a modular approach: a complete circuit is assembled from independent functional building blocks — differential pair, load, tail current, bias, compensation, and output stage. By enumerating every valid combination of block implementations, the tool can quickly generate thousands of structurally distinct op-amp netlists for dataset generation, automated design exploration, or topology studies.

The toolkit works in both directions of the design problem. Going forward, it constructs and sizes op-amps from building blocks and a performance specification. Going backward, it reads a flat SPICE netlist and recovers its structure — identifying subcircuits (differential pairs, current mirrors, cascode loads, bias generators, CMFB, compensation) and assigning each to its functional role. An end-to-end designer chains these layers together, keeping only the circuits whose ngspice-measured metrics meet the spec.

Key Features

• Topology synthesis — enumerate every valid op-amp from modular building blocks and emit flat or hierarchical SPICE netlists.
• Subcircuit & functional-block recognition — recover structure and slot assignments from a flat SPICE netlist.
• Initial sizing — compute minimum transistor W/L values that satisfy DC specs (gain, GBW, phase margin, slew rate, CMRR) via an OR-Tools CP-SAT solver and a gm/Id flow.
• End-to-end designer — enumerate, size, simulate, and keep only the designs that pass ngspice-measured metrics.
• One-, two-, and three-stage op-amps, single-ended and fully differential, including nested-Miller (NMC) and reversed-nested-Miller (RNMC) compensation.
• Extensible with no code — add new module variants by editing a YAML file.

Have a question? Start a discussion or open an issue on GitHub.

📖 Documentation

Full user guide, API reference, and design theory live in the Sphinx docs:

Installation

pip install circuitgenome

Or install from source:

git clone https://github.com/analog-ml/CircuitGenome.git
cd CircuitGenome
pip install -e .

Requires Python 3.9+. PyYAML and OR-Tools (for the sizer) are installed automatically.

Getting Started

Run the end-to-end designer: enumerate a topology, size each candidate, simulate with ngspice, and export the designs that meet your spec.

circuitgenome design --spec spec_gf180.yaml --topology two_stage_opamp_single_ended \
    --output-dir designs/ --limit 200 --workers 4

For the full CLI reference, Python API, and worked examples, see the documentation.

Contributing

CircuitGenome is an open research project and contributions are very welcome — new module variants, topology templates, recognizer patterns, sizing heuristics, bug fixes, and documentation improvements.

• Browse or open issues: https://github.com/analog-ml/CircuitGenome/issues
• Fork the repo, create a feature branch, and open a pull request against main.
• Run the test suite before submitting: python3 -m pytest tests/ -v.
• Adding a new module variant usually needs no code changes — just edit opamp_modules.yaml (see the docs for details).

References

1. A Data-Driven Analog Circuit Synthesizer with Automatic Topology Selection and Sizing — S. Poddar, A. F. Budak, L. Zhao, C.-H. Hsu, S. Maji, K. Zhu, Y. Jia, D. Z. Pan. Design, Automation & Test in Europe (DATE), 2024.
2. FUBOCO: Structure Synthesis of Basic Op-Amps by FUnctional BlOck COmposition — I. Abel, H. Graeb. ACM Transactions on Design Automation of Electronic Systems (TODAES), 2022.
3. A Functional Block Decomposition Method for Automatic Op-Amp Design — I. Abel, M. Neuner, H. Graeb. Integration, the VLSI Journal (Elsevier), 2022.
4. Constraint-Programmed Initial Sizing of Analog Operational Amplifiers — I. Abel, M. Neuner, H. Graeb. IEEE International Conference on Computer Design (ICCD), 2019.

PDFs of all four papers are in docs/papers/.