Learning Forum

[Jonghee Park]

After modeling connectors, cables, and PCBs using SI wave, Q2D, and HFSS respectively, I extracted S-parameter data from each simulator and used AEDT to generate an integrated channel's eye diagram. However, when I compared this with the eye diagram extracted from the actual model measured using a network analyzer, I found significant discrepancies. The modeled eye diagram seems to lack consideration for reflections, transmissions, mismatches, etc., making it unrealistic. I'm curious about methods to obtain an eye diagram that closely matches the measured values.

[Dan Dv]

When you extracted the various models, were they all at nominal conditions, or are you simulating across manufacturing process corners? Even if you just consider the effects of manufacturing on characteristic impedances, a single eye diagram simulation with all models at their nominal impedance is going to be something of a best case scenario. You may want to consider taking a design of experiments approach where you have multiple versions of each model  to account for that process cornering and set up a series of sweeps to include all of the potetnial scenarios.