Abstract
We benchmark two different approaches in high frequency modeling of multiconductor cables for EMI system level simulations. Our models are obtained by fitting simulated or measured data of the cable's port admittance matrix to a lumped network. The first approach is based on simulated p.u.l. parameters, whereas the second one relies on measured s-parameters. We find that for electrically long cables the measurement based model outperforms the p.u.l. based one, particularly in predicting crosstalks. This fact is supported by comparing simulations with time domain measurements for the case when the cable is loaded by a resistor as well as for the case when it is connected to a motor. © Copyright 2013 IEEE - All rights reserved