Wideband quadrature error correction (using SVD) for stepped-frequency radar receivers

We present a new technique which corrects the wideband quadrature errors associated with homodyne stepped frequency radar receivers. The correction algorithm is derived using singular value decomposition (SVD) which diagonalizes and scales the covariance matrix of a test signal while preserving the coherent phasor alignment across all frequency steps of the homodyne receiver. Using our technique, the wideband quadrature errors of a 1-2 GHz stepped-frequency radar receiver were corrected to within a 0.1 dB gain imbalance, and to within a 0.5 deg phase imbalance. The Hermitian images were shown to be suppressed by more than 50 dB below the main target peaks.