By Dr. James C. Rautio, President, CEO, and founder, Sonnet Software, Inc.
In part I, I described how I came up with the EM port calibration theory that we have been using very successfully for nearly 30 years. It is so simple. To de-embed a single, lonely port, just EM analyze a 2-port through line of length L, and a second one of length 2L, do some magic math (described in Part I and in  and ), and we have beautiful, fully de-embedded EM analysis data for that lonely port. (If we have additional lonely ports, just repeat the process for those ports, one at a time, as well.)
Central to that calibration and de-embedding procedure is the conversion of S-parameters into ABCD-parameters and back again. We usually measure (or EM analyze) one S-parameter matrix for each frequency of interest. If we are calibrating a single port, our calibration standards (the through lines) are 2-port devices, and our matrix is a complex (i.e., it has real and imaginary parts) 2×2 matrix. We calibrate and de-embed at each frequency, one frequency at a time. Both the S-parameters and the ABCD-parameters that we use for calibration are 2×2 matrices.