How this calculator works
Near-end crosstalk grows while the aggressor edge occupies the coupled route, then saturates when round-trip propagation exceeds rise time. Measured-coefficient mode is preferred when a field solver or test coupon supplies the saturated coupling. Geometry mode is a rough spacing screen, not a controlled-impedance or signal-integrity prediction. It uses a published spacing-to-height coupling approximation. Stripline combines each trace clearance to both reference planes and includes vertical trace separation. The finite-length multiplier is continuous at saturation. It does not predict far-end crosstalk, reflections or simultaneous aggressors.
Formula
Vnext = Vstep K min(1, 2 delay/rise time). Microstrip screening K = 1/[1+(S/H)²]. Stripline effective heights are hi(B-hi)/B, effective separation squared is S²+(h2-h1)², and K = 1/[1+Seff²/(h1eff h2eff)].
Worked example
For a measured saturated coefficient of 5%, a 3.3 V step with a fully saturated route gives 0.165 V of near-end noise. If the finite-length multiplier is 0.5, the estimate falls to 0.0825 V.
Advertisement
Frequently asked questions
Does this include far-end crosstalk?
No. Far-end coupling depends on the difference between capacitive and inductive coupling and requires a more detailed model.
Why does length stop increasing the near-end amplitude?
Once the round-trip propagation interval exceeds edge rise time, the near-end step coupling reaches its saturated level.
How accurate is geometry mode?
It is a first spacing screen. Real coupled traces depend on width, stackup, reference returns and terminations; use a measured or field-solved coefficient when accuracy matters.