pcb-crossing-optimizer 0.6.0

PCB crossing minimization, component placement, and power trace validation for SKiDL-generated netlists

PCB Crossing Optimizer

Multi-layer crossing minimization for PCB connector pin assignment. Detects trace crossings in SKiDL-generated KiCad netlists and computes optimal pin orderings for single-layer routing using Sugiyama-style barycenter sweep.

Installation

pip install pcb-crossing-optimizer

After installation, the pcb-crossing-optimizer command is available on your PATH.

Usage

The tool has two subcommands: analyze (default) and plan-footprint.

analyze (default)

Analyze crossings between component layers. This is the original behavior and remains the default when no subcommand is given.

pcb-crossing-optimizer [analyze] <netlist.net> --layers "L0_refs | L1_refs | L2_refs" --reorderable REF [REF ...] [--exclude REF:PIN ...] [--json] [--quiet]

Analyze the microSD breakout with passives in the middle layer:

pcb-crossing-optimizer examples/microsd_breakout.net --layers "J1 | R1,R2,C1 | J2" --reorderable J2 --exclude J1:SH

Simple two-connector case (just use two layers):

pcb-crossing-optimizer examples/i2c_breakout.net --layers "J1 | J2" --reorderable J2

JSON output for tooling integration:

pcb-crossing-optimizer examples/i2c_breakout.net --layers "J1 | J2" --reorderable J2 --json

Quiet mode for CI (exit code 0 = no crossings, 1 = crossings remain):

pcb-crossing-optimizer examples/i2c_breakout.net --layers "J1 | J2" --reorderable J2 --quiet

plan-footprint

Compute an optimal pin map for a custom footprint. Given a target component (whose footprint you are designing), fixed anchor components, and optional pin locks, it runs the crossing sweep and produces a complete pin-to-net assignment proposal.

pcb-crossing-optimizer plan-footprint <netlist.net> \
    --target J1 \
    --anchors "J2,U1 | R1,R2,R3,C1" \
    --lock 1=NC 2=NC 3=GND_EARLY_A \
    --unmatched end \
    --exclude-nets GND \
    [--json] [--quiet]

Options:

• --target REF: Component whose footprint is being designed
• --anchors "...": Fixed components organized in layers separated by |
• --lock PIN=NET: Lock specific pins to nets (use PIN=NC for no-connect)
• --unmatched start|end|split: Where to place pins with no anchor connections (default: end)
• --exclude-nets NET [NET ...]: Nets to exclude from analysis (e.g. ground pours)

Nets spanning non-adjacent layers (e.g., J1 to J2 through a passive layer) are handled automatically via virtual pass-through nodes.

How it works

1. Parses a KiCad .net file (S-expression format generated by SKiDL)
2. Extracts pin-to-net connectivity for specified components
3. Sugiyama-style barycenter sweep across all layers:
   • Inserts virtual nodes for long edges spanning non-adjacent layers
   • Reorders both pins within reorderable components and components within layers
   • Iterates forward/backward sweeps until convergence
4. Reports current crossings, recommended reordering, and remaining crossings

Project structure

src/crossing_analyzer.py        Core logic + CLI
tests/                      pytest tests
examples/                   Sample netlists for testing
.github/workflows/          CI / PyPI publishing

Running tests

pip install pytest
pytest -v

Python import

The PyPI package name is pcb-crossing-optimizer, but the importable module is crossing_analyzer:

from crossing_analyzer import sweep_optimize, PinColumn, format_multilayer_report

Publishing to PyPI

Releases are published automatically via GitHub Actions trusted publishing (OIDC — no API tokens needed).

Setup (one-time)

1. On pypi.org, go to your project → Publishing → Add a new publisher
2. Configure the trusted publisher:
   • Owner: ashergarland
   • Repository: pcb-crossing-optimizer
   • Workflow: publish.yml
   • Environment: pypi
3. On GitHub, go to Settings → Environments and create an environment named pypi

Releasing a new version

1. Update the version in pyproject.toml
2. Commit and push
3. Tag and push the tag:

   git tag v0.5.0
   git push origin v0.5.0
   

4. The publish.yml workflow triggers on v* tags and publishes to PyPI automatically

Manual publishing (fallback)

pip install build twine
python -m build
twine upload dist/*

Roadmap

• Passive placement optimizer (assign positions to resistors/caps in routing channel)
• F.Cu/B.Cu layer assignment for unavoidable crossings