Abstract

In this combination of measurement and modeling, the relative effect of process variations present in the cross-section dimensions and material properties of two automotive printed circuit boards (PCBs), with 10-inch striplines is studied. Specifically, the insertion loss (IL) uncertainty up to 20 GHz. The PCBs were horizontally cut into fifteen cross-section samples each. Normal distributions were obtained after measurements of their physical dimensions and for the PCB dielectric properties, a uniform distribution was selected. The measured distributions in combination with assumed stochastic variations for the dielectric properties of the striplines were used to simulate a 2D cross-section via an electromagnetic (EM) 2D-Method of Moments (MoM) solver. The sensitivities due to the input variables were obtained through a second order polynomial chaos expansion (PCE) in two different analysis. The first case focused on the measured physical dimensions variability and showed that the conductor roughness had the biggest effect on the IL uncertainty. The second case analyzed the combination between measured data and the assumed stochastic variation of the dielectric properties; and showed a bigger dependence of the IL uncertainty due to the dielectric properties than the physical dimensions.