Fast macromodeling of large-scale multiports with guaranteed stability

This contribution introduces a novel approach for generating guaranteed stable macromodels of large multiport structures in a completely automated and efficient manner. The presented method is based on the Adaptive Antoulas-Anderson (AAA) algorithm for rational fitting of scalar transfer functions. We propose a computationally cheap multi-input multi-output extension of the AAA, and we combine the resulting algorithm with a novel post-processing stability enforcement step that is formulated in terms of a small-size convex program. Applying the resulting framework to a large Power Delivery Network (PDN), we show a significant computational cost reduction with respect to commonly employed state-of-the-art methods. The proposed scheme fits naturally as a bridge between electromagnetic and circuit simulation, enabling the representation of high-frequency phenomena and parasitics as low-order equivalent circuits synthesized from the computed macromodels.