Evaluation of the VNA verification process based on the normalized errors

Abstract

Verification kits are used as common tools for demonstrating measurement traceability of a Vector Network Analyzer (VNA), and for verifying the process control of the routine calibrations or inter-laboratory comparison. Such measurements performed by a national metrology lab or a secondary cal lab are very accurate due to the tremendous advancement of modern vector network analyzers. An inter-laboratory comparison of verification kits was established and is currently managed by the ARFTG. A robust statistical tool was used to improve the accuracy of the comparison by excluding outliers from the cumulative data of various participants. A statistical bound based on the median of absolute deviation (MAD) provide an evaluation method for the users in determining the pass or fail of their measurements. In this paper, a normalized error method and also z-score procedure are both discussed for their applicability as pass/fail criteria. Finally examples using the normalized error method, which is highly utilized by the metrology community, will be presented. A numeric metric encompassing the interpretation of results will be provided.