Skip to main content

antenna-cad

Design compiler for antenna and phased-array PCBs: YAML spec in — synthesized layout, KiCad DRC, and openEMS full-wave verification out. Deterministic Python core, agent-ready via MCP.

PyPI Python versions CI License Ruff uv mypy

Compile antenna and phased-array design intent into simulation-ready KiCad PCB layouts.

antenna-cad is a Python design compiler for printed antennas. You describe requirements (frequency, substrate, polarization, gain); it synthesizes geometry, writes a DRC-checkable .kicad_pcb, simulates the layout with openEMS, and reports predicted performance against the requirements. Every step is deterministic and runs headlessly, with no KiCad GUI and no LLM in the loop. Agents can drive the same typed API, but nothing depends on one.

4x4 corporate-fed patch array, generated and DRC-verified by antenna-cad

A 4×4 corporate-fed patch array at 10 GHz: synthesized, routed, DRC-checked, and full-wave simulated from a 15-line spec file. No human drew any copper.

From spec to verified board

design:
  name: array_2x2_10ghz
requirements:
  center_frequency: 10 GHz
  impedance: 50 ohm
  substrate: RO4350B
  substrate_height: 0.508 mm
element:
  type: rectangular_patch
array:
  nx: 2
  ny: 2
  spacing: 0.6 lambda
pip install antenna-cad
antenna-cad report spec.yaml -o build/ --tune

That one command synthesizes the patches and the corporate feed (quarter-wave transformers, phase-compensated mirrored rows), emits a KiCad board, runs real KiCad DRC, simulates the exact layout in openEMS FDTD, iterates a simulate-and-correct tuning loop, and writes a verification report:

Board (KiCad render) Return loss Radiation pattern
2x2 board 2x2 S11 2x2 pattern

Measured for this 2×2 (openEMS, tuned): resonance 10.12 GHz against the 10 GHz target, S11 −14.8 dB, directivity 12.8 dBi, gain 11.7 dBi, single broadside beam.

flowchart LR
    S[spec.yaml] --> SYN[analytic synthesis]
    SYN --> IR[typed design IR]
    IR --> K[KiCad board + DRC]
    IR --> EM[openEMS FDTD]
    EM --> R[report + metrics]
    R -->|simulate-and-correct| SYN
    K --> FAB[Gerbers / STEP]

Verified examples

Example Result (openEMS)
Single patch, 10 GHz f_res 9.84 GHz, S11 −11.5 dB, D 6.8 dBi patch
2×2 corporate-fed array tuned: f_res 10.12 GHz, S11 −14.8 dB, D 12.8 dBi, gain 11.7 dBi 2x2
4×4 corporate-fed array D 16.8 dBi, gain 15.4 dBi, −13 dB E-plane sidelobes; match limited to −8 dB (why) 4x4

Every number above is a real openEMS result from the committed example specs; the figures rebuild from checked-in run data with uv run python figures/make_figures.py.

Design principles

  • Compiler, not file generator. The source of truth is a typed internal representation of the design — geometry, stackup, nets, ports, constraints, units. KiCad, Gerber, and solver files are build artifacts compiled from it.
  • Deterministic core. The same spec produces the same board, byte for byte. Verification runs through real engines: KiCad DRC (kicad-cli) and full-wave FDTD (openEMS).
  • Physical quantities carry units (via pint). 10 GHz and 8.42 mm are values in the API; bare floats cross into backends only at explicit conversion boundaries.
  • Narrow scope by design. This targets the structures printed antennas need (patches, microstrip lines, impedance transformers, feed trees, ground pours, via fences) rather than general PCB autorouting or schematic capture.

Install

pip install antenna-cad            # core: synthesis, IR, KiCad emission
pip install "antenna-cad[pam]"     # + phased-array-modeling lattice interop
pip install "antenna-cad[mcp]"     # + MCP server for agents
pip install "antenna-cad[dxf]"     # + DXF export

KiCad 8+ provides DRC, rendering, and manufacturing export (brew install --cask kicad on macOS; distribution package kicad on Linux). openEMS runs natively where its Python bindings are installed, or reproducibly in Docker:

docker build -t antenna-cad-openems docker/

For agents (MCP)

The same pipeline is exposed as typed MCP tools — synthesize, layout, DRC, simulate, report, export — so an agent can iterate designs against real DRC and full-wave feedback instead of hallucinating copper:

pip install "antenna-cad[mcp]"
antenna-cad mcp serve
{
  "mcpServers": {
    "antenna-cad": {"command": "antenna-cad", "args": ["mcp", "serve"]}
  }
}

Conventions

The MCP surface follows the family convention documented in APAB's mcp-conventions.md: same server layout, tool naming, error contract, and path-safety rules, so an agent connected to both servers sees one house style.

Citing

If antenna-cad contributes to academic work, please cite it (GitHub's "Cite this repository" button uses CITATION.cff; an archival DOI is planned):

@software{hodge2026antennacad,
  author  = {Hodge, John},
  title   = {antenna-cad: a design compiler for antenna and phased-array {PCBs}},
  year    = {2026},
  version = {0.1.1},
  url     = {https://github.com/jman4162/antenna-cad},
  license = {MIT}
}

License

MIT


#antenna #phased-array #kicad #pcb-design #openems #fdtd #microstrip #patch-antenna #rf #electromagnetics #eda #mcp #python

Release files for antenna-cad 0.2.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for antenna-cad 0.2.0
File Size Uploaded
antenna_cad-0.2.0.tar.gz 601.0 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for antenna-cad 0.2.0
File Interpreter ABI Platform
antenna_cad-0.2.0-py3-none-any.whl Python 3 none any Details

Total release size: 672.1 kB

Release files / antenna_cad-0.2.0.tar.gz

Download URL antenna_cad-0.2.0.tar.gz
Size 601.0 kB
Tags Source
SHA-256 checksum
How to use checksums
bdd8a3582d595f23c35d59c06dc1bc50e22653dbc7a0ffb86ac255eb46112aeb
BLAKE2b-256 checksum
How to use checksums
297065d8dc8f39f96149c46100c5acef899b2099c8fc38c13205671d001c2be2
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 12, 2026.

Transparency log

Release files / antenna_cad-0.2.0-py3-none-any.whl

Download URL antenna_cad-0.2.0-py3-none-any.whl
Size 71.1 kB
Tags Python 3
SHA-256 checksum
How to use checksums
697db3975899f5a1d02eca75946cb5c8a95f9d01d5be85acc50ebb49b5aa5d5c
BLAKE2b-256 checksum
How to use checksums
5b1ab728676e91f37b7fadfbd23e951c3270a6c4557e73fc2aeef48789596d39
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 12, 2026.

Transparency log

Release history Release notifications | RSS feed

This release

0.2.0 This release

2 release files

0.1.1

2 release files

0.1.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page