Issue link: https://iconnect007.uberflip.com/i/1053050
70 DESIGN007 MAGAZINE I NOVEMBER 2018 quency. You don't see those resonance peaks at the corner because the wall of shorting vias is only a quarter of an inch away, forcing the impedance low. Figure 10 displays the same trend. As opposed to the static capacitance—which is the same at the center and corner—the extracted inductance does depend on the location. As you move closer to the wall of shorting vias, the loop inductance gets lower. If you want inductance values representative to the laminate, you must take the induc - tance at the center. The inductance is around 100 pH, which correlates approximately with the spreading inductance estimate of planes, which is 33 pH for each mil of dielectric spac- ing. The inductance curves are relatively flat, but there is a small downslope starting around 10 MHz. You can see the correlation for the 1-mil lam- inate in Figures 20 and 21. It follows the same basic signature; however, the primary differ- ence is the lower inductance. Not only does the entire curve runs lower, but you can also see a more pronounced change of inductance with frequency. The inductance starts to drop at 10 MHz, which is commonly considered the skin corner frequency of 1-ounce copper. Figures 22 and 23 show the 0.5-mil correla- tion. The data follows the same trend—lower Figure 19: Inductance extracted from the imaginary part of impedance of the 4-mil shorted-edge board data at the corner (L) and center (R). Figure 20: Impedance magnitude correlation of the 1-mil laminate data at the corner (L) and center (R).