Issue link: https://iconnect007.uberflip.com/i/1522641
86 PCB007 MAGAZINE I JUNE 2024 We may not be able to explain the root cause, but without 100% impedance testing coverage, we would not have known about the issue at all. In this instance, RoBAT's measurement would have rejected 70% of the PCBs on imped- ance alone. Can we afford to rely on the simulation and measurement of just a coupon when the imped- ance of every single track is critical? I think we would all agree that it's certainly not where sig- nal integrity is paramount—although I'm sure Einstein would. Perhaps now, with fast, accurate, high-spec TDR automated test equipment, we can com- plete the feedback loop of PCB design, closely linking manufacturing and test for important impedance KPIs across entire panels so that we can be ready for the next generation of high- speed PCBs and signal integrity challenges. PCB007 Alex Knowles is an R&D manager at RoBAT. Figure 5: We've gone from 94% to 2.5% in a single panel? If we look at three sequential panels there's a panel-by-panel step change down ~6Ω and back up again. Figure 7: It looks like there's more going on across the panel than just that. If we change the heatmap to single panel measurements and look at the high- est of 6σ, that is panel 37. Figure 8: The pattern is now clear: a diagonal band of lower impedance from bottom left to top right across the whole panel. This shows up independent of the large ±6Ω step changes already seen in the other heatmap figures. Figure 6: Without access to the exact process steps that occurred between each of these panels, it is difficult to imagine why there could be such a step change and then a reversal in impedance across the entire panel. We will leave that one to the fab houses to explain.