A Broadband and Parametric Model of Differential Via Holes Using Space-Mapping Neural Network

This letter presents a novel broadband and completely parametric model of differential via holes by virtue of the space-mapping neural network technique. This model consists of a neural network and an equivalent circuit that is utilized to account for various EM effects of differential via holes. The neural network is trained to learn the multi-dimensional mapping between the geometrical variables and the values of independent circuit elements in the equivalent circuit. Once trained with the EM data, this model provides accurate and fast prediction of the EM behavior of differential via holes with geometry parameters as variables. Experiments in comparison with measurement data and EM simulations are included to demonstrate the merits of this new model in both the frequency and time domains.