Issue link: https://iconnect007.uberflip.com/i/1053050
NOVEMBER 2018 I DESIGN007 MAGAZINE 67 impedances with ED or RA copper. Also, you don't see the effect of copper weight on these plots; it will show up in the data of shorted- edge boards. Now that I showed a good correlation between measured and simulated data, the comparison of different laminates with the clean simulated data from PowerSI can be done. Figure 15 shows the impedance mag- nitude, and Figure 16 shows the capacitance comparison. On the comparison plots, you can see that as the laminate gets thinner, the peak-valley ratio of modal resonances goes down, which makes it less likely that noise at the peaks would get too big and cause power or signal integrity issues and is a clear signal integrity benefit of thin laminates. You can also see that there may be different resonant frequencies depending on the location where you look at the board. In the center, the reflected waves coming back from the open edges cancel for the first two resonant peaks below 1 gigahertz, and there- fore, do not see those peaks at the center. Also note that the capacitance curves have very little dependence on location: at low fre- quencies the static plane capacitance is the same at every location. However, approaching Figure 15: Comparison of PowerSI simulated impedance magnitude with different laminate thicknesses at the center (L) and corner (R) of boards. Figure 16: Comparison of PowerSI simulated capacitance with different laminate thicknesses at the center (L) and corner (R) of boards.