Abstract
Admittance frequency-sweep measurements is an accepted procedure for characterizing transformer terminal behavior for the purpose of frequency-dependent black-box modeling. The errors introduced by the measurement cables is with one existing practice mitigated by removal of the associated shunt capacitance effect. In this paper, it is shown that the accuracy can be greatly improved by using transmission-line representation of the measurement cables with parameters obtained from standard cable data or from S-parameter measurements. The procedure is demonstrated for the modeling of a 45 MVA generator step-up transformer requiring cable lengths of six meters and an upper frequency limit of 10 MHz.