Eye diagram simulation in full-wave simulator

[rodgerwxh]

I am studying eye-diagram while questions raised up in my mind.

To my best understanding, eye-diagram superimpose outputs based on bit period (or Unit Interval). This assumes that a random 0-1 series is set as the input signal, and its output is squeezed into one bit period, right?

With that said, I have two questions:

1. For freq-domain solvers, how is eye-diagram being calculated?

2. What I am more concerned is: For all transient simulators, the initial period is not a steady state. (E.g. devices need to be charged and biased), thus the early outputs are actually not reflecting the system's steady performance. Shall eye-diagram include these initial responses? Or it is only valid for steady-state systems?

 

[jsherman]

I can help out with 2! Never really used a freq-domain solver.

Eye-diagrams are only generated using steady state information for the reasons you mentioned. Eye-diagrams can help measure jitter, over/undershoot, and other distortions of the steady state signal, so including the early outputs when everything is charging up will only interfere and obscure the useful information in the eye diagram!

 

[rodgerwxh]

Thanks Jsherman, that is very clear explanation.

 

[DonnyP]

There are actually two types of eye diagrams: the usual digital eye diagram and there is also a I-Q eye diagram. It sounds like your looking for the digital eye diagram. ADS has some nice tools for this. I'd be interested to hear what other tools can do this.

 

[rodgerwxh]

Thanks DonnyP, I pretty much learned all about digital eye diagram, can you introduce a bit more about I-Q diagram?
Also, one my original question remains unanswered

1. For freq-domain solvers, how is eye-diagram being calculated?

Any inputs?