CircuitGenome

CircuitGenome is a Python toolkit for analog circuit topology synthesis and recognition, focused on op-amp design. A complete circuit is assembled from independent functional building blocks (differential pair, load, tail current, bias, compensation, and output stage), and by enumerating every valid combination of block implementations it can generate thousands of structurally distinct op-amp netlists for dataset generation, automated design exploration, or topology studies.

It works in both directions. 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.

Initial sizing computes minimum transistor W/L values that satisfy DC specs (gain, GBW, phase margin, slew rate, CMRR) through an OR-Tools CP-SAT solver and a gm/Id flow. It covers one-, two-, and three-stage op-amps, single-ended and fully differential, including nested-Miller and reversed-nested-Miller compensation. New module variants can be added by editing a YAML file, with no code. It installs from PyPI as circuitgenome and needs Python 3.9 or newer.