Additively Manufactured Waveguide Hybrid Septum Coupler Optimized Using Machine Learning

This paper describes a waveguide septum coupler design having a smooth profile well suited for additive manufacturing. The large aperture of this hybrid coupler is shaped with even-degree Legendre polynomials. Machine learning-assisted global optimization is employed to extend the operating bandwidth of the component. A design in K-band is detailed and a prototype is manufactured and tested. The experimental results confirm an improvement of 19% in operating bandwidth compared to the previously reported design in the same band while keeping all other key properties mostly unchanged, specifically the physical dimensions. The use of additive manufacturing leads to a mechanically simple and lightweight component of interest for the design of integrated microwave devices, such as beamforming networks and compact feed systems.