Antenna pattern comparison using pattern subtraction and statistical analysis

This paper discusses a technique that can be used when comparing two antenna patterns that produces a measure of the difference between the patterns and an associated confidence level for the results that is derived from a statistical analysis of the pattern differences. The first step in the process is to verify that the same coordinate system is used, and that the AUT is precisely aligned to those coordinates in all measurements. Small differences in beam pointing and polarization that correspond to rotations about the three rectangular axes can arise due to AUT alignment differences or due to some measurement errors. These changes in beam pointing can produce apparent pattern differences in the pattern comparison process that can distort the results, and adjustments in the AUT alignment or pattern data should be done before the pattern comparisons are carried out. It is easy to interpolate the pattern or change the pattern centering using standard processing software to try and correct for the angular misalignments, however the interpolation or centering may not completely correct for rotations about all three axes and some effects may remain after software adjustments. The next step in the process is to separate the analysis of pattern differences in the main beam region from those in the sidelobe region. The main beam region is analysed by focusing on beam pointing, peak gain, directivity and beam width comparisons. For sidelobe and cross polarization comparisons, the patterns are normalized to the peak of the main beam for each measurement rather than to the directivity or gain of results. This will separate changes in main beam parameters from sidelobe differences. The normalized patterns are then compared by calculating the difference between the two patterns and from these differences the ratio of an Equivalent Stray Signal (ESS) to the peak of the main beam is derived. The derived ESS will generally have large variations over the angular extent of the patt- - erns. A statistical analysis of this ratio produces a single ESS that is constant over the full angular extent that is a measure of the difference of the patterns along with a confidence level for the ESS. The derived ESS can then be used as an estimate of uncertainty for a measurement error source such as reflections, data point spacing, etc. It can also be used to quantify the difference between measurements on different ranges or using different techniques. Example of both of these applications will be presented.