Empirical Emission Eigenmodes of Printed Circuit Boards

It is shown that radiated emissions from circuits can be efficiently decomposed and represented using principal component analysis (PCA). A hierarchical set of orthonormal spatial eigenmodes and their associated amplitudes are extracted from data for an emitted field that is scanned and measured across a planar area and wide frequency band. This decomposition is applied to the magnetic tangential field intensity for emissions from an analog transmission-line based circuit. The inhomogeneity of the autocorrelation matrix driving the PCA algorithm is experimentally demonstrated. Eigenmodes, loadings, and scores of the principal components are calculated and interpreted physically. Across the selected frequency band, PCA achieves efficiency savings of one to two orders or magnitude for the representation of emitted radiation.