Empirical statistical analysis of planar transmission lines on PCBs accounting for manufacturing variations

The authors address the problem of random manufacturing variations in PCB (printed circuit board) structures and describes an empirical technique to perform a statistical analysis of PCB lines. The technique is designed to improve transmission-line characterization by enabling a worst-case analysis of transmission-line effects to be done. This will allow the designer to minimize postprototype problems associated with high-speed PCBs concerning inconsistencies in a manufacturing process. The description of PCB structures is obtained by measuring the scattering parameters and then converting them to frequency-domain characteristic impedance (Z0) and propagation constant (gamma). A deviation in complex Z0 and gamma can be used to model line and discontinuity variations. It is concluded that with the use of this technique, a better representation of Z0 and gamma can be obtained than by previous techniques