Issue link: https://iconnect007.uberflip.com/i/241303
article influence of via stub length and antipad size on the insertion loss profile continues Figure 23: Screen shot of differential insertion loss testing. tion of the display shows the amplitude of the differential insertion loss SDD21. The orange trace is for a coupon with nearly optimum backdrilling (minimum stub), with the yellow trace showing SDD21 for a differential pair with long via stubs. The bottom portion of the screenshot displays the amplitude (left) and phase (right) of all 16 mixed mode S-parameters. For the differential testing, mixed mode Sparameters were measured on 10 different stub lengths, two different antipad sizes and five panels, testing from 10 MHz to 40 GHz with 2048 points. Similar as with the single ended data, the magnitude of the differential insertion loss was plotted for all 100 differential pairs to check for unusual readings (Figure 24). The av- erages over the five panels were plotted for the 10 stub lengths and two antipad sizes, to visualize the impact of the parameters (Figure 25). Figure 25 clearly demonstrates the increase of the resonant frequency for shorter stub lengths and also some smaller changes caused by the antipad size. To get the full picture on the influence of the panel, the antipad size and the stub length, an analysis-of-variance on the magnitude of SDD21 was conducted for various frequencies. Figure 26 shows the main effect plot of this ANOVA for a frequency of 5 GHz. The main effect plot confirms a very small panel-to-panel variation. The effect of the antipad size is slightly larger, but the main influence clearly is the stub length. To get quantitaJanuary 2014 • The PCB Design Magazine 57