Measurement Error Detection for Calibration Standards Based on Cluster Analysis

This paper presents an automatic measurement error detection method for calibration standards in the vector network analyzer (VNA) calibration process using Density-Based Spatial Clustering of Applications with Noise (DBSCAN). Measurement errors of calibration standards occurring during the VNA calibration process are challenging to detect artificially, which can significantly deteriorate the final calibration results. The proposed method combines the 16-term error model calibration algorithm with DBSCAN, which can automatically identify the calibration standards with measurement errors. The calculation results of the 16-term error model using different calibration standard combinations are regarded as cluster samples. Calibration standards with measurement errors are detected by identifying the noisy outlier samples. The effectiveness and reliability of this method are validated by on-wafer S-parameter measurements.