Physics-based fitting to improve PCB loss measurement accuracy

Measuring and controlling printed-circuit-board (PCB) loss has become an ever-critical task in high speed interconnect design. However, insertion loss measurements are often plagued with errors, especially at high frequencies, due to various sources during measurement and de-embedding procedures. This issue only gets worse in low cost measurement houses. One approach to improve data fidelity is to utilize data smoothing algorithms, but traditional brute-force algorithms have obvious flaws. In this paper, we propose a novel physics-based data post-processing method, denoted as ARO (Advance Root-Omega) fitting. This new method can report insertion loss values that are more error-resistant while also providing an indicator of measurement quality. This method has been tested on a library of over 1000 real PCB measurements, and has proven to be reliable and robust for industry adoption. The algorithm has been distributed to our customers for implementation in their production lines.