It was a pesky problem: the highaccuracy EM analysis of circuits with ports placed internal to the circuit (instead of on the edge). These internal ports would be used to connect surface-mount devices (SMDs) or include transistors inside an RF IC. Much of the time, uncalibrated internal port results are "good enough." When measured and calculated, they line up reasonably close. But there was that pesky little bit of difference that occasionally became large and design-disabling.

We knew why the differences were there. When you build something and make a physical measurement, you have a port discontinuity. It might be an SMA-to-microstrip connector or CPW probe. EM-analysis ports also have discontinuities. It could be a couple tenths of a picoFarad or nanoHenry. Such small discontinuities are not a problem in non-critical situations. In other cases, such as on the input of a power FET, they cause design failure. If accuracy is critical, the ports must be calibrated— even if they are in the middle of the circuit.

If two ports are close together-—say the input and output ports where a transistor will be connected— the stray fringing fields coupling between the ports is not removed from the resulting data. That stray coupling between the ports was causing the problems described above. I had already tried to solve the mathematics of this problem several times.

Then, a customer came across this problem. This customer uses a lot of SMDs including multi-pin narrowband SAW filters—lots of internal ports all close together. The problem is especially bad for high-Q narrowband filters. So I decided either I or the problem was going down. I went into immersion and literally did nothing but eat, sleep, and work the problem. For a full 30 days, I engaged in a holy war with this problem. (Okay, I did take one evening off and went to one of my kid's concerts; Some things are really important!)

At the end of the 30 days, I was victorious. The mathematical aspect of the problem was vanquished. I gave the solution to our development team, which spent a lot of time evaluating a host of issues. The team came up with an easy-to-use interface. I hope that when you use it, you will think, "Gee, that's so easy. Why didn't anyone think of it before?!" That will be a wonderful compliment.

Sonnet Software, Inc., 100 Elwood Davis Rd., North Syracuse, NY 13212; (315) 453-3096, Internet: www.sonnetsoftware.com.