rfnetwork

Linear circuit solver for networks of RF components.

Installation

rfnetwork requires Python >= 3.9. Wheels are currently only available on Linux.

   pip install rfnetwork

To build from source on Windows or Linux,

   git clone https://github.com/ricklyon/rfnetwork.git
   cd rfnetwork
   git submodule update --init --recursive
   pip install -e .

If building from source, a C++ compiler must be available on the system. On Windows systems the recommended compiler is the MS C++ Build Tools.
In addition,nvcc must be present if compiling the GPU accelerated solver. The CPU solver will still compile if nvcc is not found, but the GPU solver will not be available.

The GPU accelerated solver is available in the full_gpu branch.

Features

• Simulate networks of RF components with hierarchical building blocks.
• Compute noise figure of multi-port networks.
• Interactive tuning of variable components (i.e. switches, phase shifters, capacitors).
• Supports internal voltage probes inside a network.
• GPU accelerated FDTD solver.

Performance

FDTD solver metrics for the dipole.py example on the following device:

CPU: Intel i7-9700 (8) @ 4.700GHz
Memory Speed: 2666 MT/s
GPU: NVIDIA Quadro P2000

Model is simulated over a time span of 400ps.

Documentation

Documentation can be found here,

https://rfnetwork.readthedocs.io/en/latest/

License

rfnetwork is licensed under the MIT License.