Non-Linear Time-Invariant Link Model Based on Separation of Random and Correlated Jitter and Noise by combine Response and DDJ Extraction

摘要

We present a method of modeling both the linear and non-linear behavior of a serial data transmitter and signal path where the model can be built from measurements data, as well as from simulation. Our process separates different sources of impairments present in the measured waveforms as follows: regular linear pattern correlated component, non-linear pattern correlated component, and a data-uncorrelated component which includes random and deterministic components. Of particular interest is that the regular component of the waveform can be used to extract the equivalent bit or transition response with 'asymmetrical' or 'non-LTI' components. From the uncorrelated part of the response we find the distribution of the uncorrelated components. Once correlated and uncorrelated components are analyzed, it is possible to advantageously model the behavior of the system using different types of processing for different impairment components. We demonstrate this separation of random and correlated components and the LTI- and nonlinear-portions of the correlated components. Simulation results based on characterization data are compared to results from measurements on physical DUT.