Accuracy of modeling for evaluation of an integrated diversity wireless system on a small PCB

Designing antennas for diversity/MIMO requires a hybrid simulation/measurement approach for real-world situations where the idealized modeling can becomes dubious. Simulations can be accurate for well-defined configurations, but for elements mounted on a complex platform, a typical, simplified numerical configuration behaves differently to the real-world situation. A comparison is presented here for diversity antennas on a printed circuit board (PCB). The PCB is modeled as an unpopulated surface of copper for the numerical experiments, and for the physical experiments it is partially covered with copper tracks and heavily populated with components (chips, connectors, transducers, LEDs, etc.). Simulations are shown to offer only first-cut design guidelines for most metrics. An optimistic interpretation for the feed impedance comparison is a relative frequency shift which varies between from 4% to 13%. For the antenna correlation, the relative difference between the results from the pattern inner product approach and the S-parameter approach can be extremely large, but it is low in absolute terms for low correlations. The use of the pattern frequency correlation function (PFCF) offers a metric for the difference between the approaches and also gives a minimum frequency shift for deploying frequency diversity.