Issue link: https://iconnect007.uberflip.com/i/241303
article influence of via stub length and antipad size on the insertion loss profile continues Figure 1: Backdrill principle. Best performance would be reached with a signal path without distortion and zero attenuation. In the imperfect reality, the insertion loss of the transmitting structures needs to be as small as possible and should not show large non-linearities. For insertion loss reduction, the dielectric loss needs to be minimized by using low-Df materials. The second important parameter is the copper loss, which is influenced significantly by the roughness of the signal trace. The application of both adequate oxide replacement and copper foil quality is key1, 2. However, there is another element in the transmission channel that needs to be evaluated. The via structure connecting the integrated Figure 2: A board 0.220" in thickness. 44 The PCB Design Magazine • January 2014 circuit or connectors to the traces on the innerlayers of the printed circuit board has a huge negative impact on the insertion loss profile, especially if the via extends significantly beyond the layer that needs to be electrically connected. As discussed in Eric Bogatin's Practical Analysis of Backplane Vias3, the via stub creates a large notch in the insertion loss profile at the "quarter-wave frequency." A commonly used method to reduce the via stub is to backdrill: A second drilling step after electroplating of the through-holes removes the copper in the unused portion of the via (Figure 1). Since this is a mechanical operation, improving the depth accuracy is not simple and very often complicates the process significantly, which in turn increases cost. It is important to understand how much stub is still acceptable in a given application to avoid excessive strengthening of the via stub specification. To get real data on the effect of the via stubs, single ended and differential channels were created with stub lengths varying from practically zero to close to 100 mils. As a second parameter, the sizes of the antipads on the reference layers have been modified. Two-port and 4-port S-parameters were collected on these test structures and an "analysis-of-variance" (ANOVA)4, 5 approach was used to evaluate the effect of these parameters on the magnitude of the insertion loss as well as on the quarter-wave resonance frequency.