EM De-Embedding Magic – Part 3: What Could Possibly Go Wrong?

By James C. Rautio

In Part 1, I described the EM analysis port calibration algorithm that I developed almost 30 years ago and has been in use in Sonnet ever since. At first glance, it might seem like the technique can be applied only to “lonely” ports — ports that have no interaction or coupling with any other nearby ports. In Part 2, I showed how the algorithm can be easily extended to calibrate and de-embed as many tightly coupled ports as needed. Yes, a couple equations are needed, but once you get familiar with them, the process is really easy to do. Everything works so well, one might think that nothing could possibly go wrong.

Of course not. No matter what, something can go wrong. In fact, that is what we as engineers find most interesting. This information is useful to anyone who uses EM analysis data that has been de-embedded. Failure is our enemy, and we must know our enemy well if we are to succeed.