Correct calculation and presentation of DFLP with commercial software

摘要

In many cases, S parameters measured by network analyzer are exported as datasets for post analysis and processing of the measurement results with a circuit simulator, MATLAB procedures and even Excel spreadsheet. This approach improves production efficiency, as the measurement equipment is usually very expensive and requires experienced operators. The algorithm often used to extract Deviation from Linear Phase (DFLP) is a Linear Regression algorithm. In most programs this algorithm is provided as an internal function, but it is not the proper algorithm for DFLP since it provides a linear phase function that has the least average error from the investigated phase function, while a proper algorithm for DFLP should provide a linear function that has the least absolute maximum error from the investigated phase function. This discrepancy could (and does) lead to over specifications of products and to rejection of good products. In this paper, an easy and practical way of providing the correct DFLP by using the internal Linear Regression Function is presented.