Issue link: https://iconnect007.uberflip.com/i/942744
54 DESIGN007 MAGAZINE I FEBRUARY 2018 A Practical Example I have taken a range of scenarios to show the interplay between loss drivers and how the key drivers of loss will gain and lose (relative influ- ence) as you vary base material and geome- tries. While we consider the following figures, I would like you to focus on the yellow (copper loss with roughness) and the green (dielectric loss) series in the graphs and consider how each property's influence waxes and wanes, depending on the geometry and material selec- tion. I have used generic "good enough" val- ues to illustrate the point. In Figure 1, the generous line width keeps the copper losses low, and even with a fairly rough surface, the copper loss and roughness loss combined are less than half of the con- tributors to total attenuation. In Figure 2, the line width is slashed to 3 mils and the heights adjusted to maintain a 50-ohm environment. The effects of reduced copper are clear even with a high-loss basic FR-4 material; cutting the line width to 3 mils pushes the losses, with roughness included, to significantly more than the dielec- tric loss. Figure 3 takes the same 3 mil 50-ohm line but substitutes a mid-loss material with a loss tangent of 0.007. Now copper losses are lofted to the vast majority of attenuation and when seeking better performance with this scenario it starts to pay to look at copper profiles. To keep this in perspective, the loss is much lower than with the high-loss material, but any varia- tion is now increasingly likely to come from the copper rather than variations in the base material. Taking this example to the extreme, which is a great way to highlight these kind of sce- narios, I have inserted a very low-loss mate- rial into Figure 4 with a loss tangent of 0.0021. Here you can see the drivers of loss are entirely in the copper; there is no more gain to be made from better laminate, other than having smoother copper. Let's think a little more about Figure 4. Say, for example, you purchase two identical PCBs from different suppliers, both made with the best base material, and the insertion loss of one board is significantly different from that of the other. Where do you look first? A glance at these graphs leads you to suspect that the Figure 1: Generic FR-4 with 8 mil line width and heights set to achieve a 50-ohm line. Figure 2: The same characteristic impedance, and same material, with line width reduced to 3 mils.