Issue link: https://iconnect007.uberflip.com/i/461203
February 2015 • The PCB Design Magazine 31 I 3 : InCIDEnT, InSTAnTAnEOUS, IMPEDAnCE continues the pulse the instantaneous characteristic impedance of high-frequency cables by removing the resistive elements through a technique called launch point extrapolation (LPE) [4] . IPC is now looking at applying this to PCB traces for all the reasons described above. The benefits of applying LPE to PCB trace impedance tests are clear. Linear extrapolation applied to an undisturbed sec- tion of the trace and projected back in time to a point close to or at the transition between the probe and the start of the test trace or coupon should remove most of the resistive effects and allow a closer correlation between the modelled impedance and the measured instantaneous impedance. As a precaution and to ensure the copper is not too thin or the line overetched, sometimes a second point is also measured at an imaginary point at the far end of the trace. This ensures that other effects such as trace ta- per or a higher-than-expected copper resistivity are ruled out. You might be thinking that "loss" is catered for when measuring insertion loss in the fre- quency domain, and this is true. But the types of scenario I am referring to in this particular column are those where the lines are operat- ing at "mid" frequencies (i.e., up to one, two or three GHz, where the dielectric losses are still small enough to ignore). The resistive losses in copper appear at these frequencies because of a couple of reasons: one is the use of thinner copper and the other is that line widths are shrinking too. This means that there are starting to be small but significant losses from DC upwards. To add to that, the on- set of skin effect in copper starts to take effect at lower frequencies than need to be considered in the case of dielectric losses. So it is in this "middle ground" of thin copper/narrow traces and low GHz operating speeds that fabricators and designers start to see impedance traces which rise over time, i.e., distance, with the cumulative build up of resis- tive effects. This effectively means you need to look at the operating frequency band and the trace geometry and the combination of these to choose the most appropriate method for mea- surement. With low-frequency wide traces, trace ca- pacitance is the predominant issue. With medium-frequency 1 or 2 GHz wide traces, lossless impedance: Figure 2: skin depth.