A Practical De-embedding Error Analysis Method Based on Statistical Circuit Models of Fixtures

Abstract

De-embedding methods require multiple identical fixtures. However, in reality the fixtures cannot be fabricated perfectly identical due to manufacturing variations. The identical assumptions for de-embedding algorithm will be unavoidably violated, which is going to introduce error due to de-embedding. In this paper, a novel methodology is proposed to estimate the error due to de-embedding for practical testing vehicle measurement. Models of the Thru and Total lines with fixtures are created. Perturbation in the fixtures is introduced based on TDR measurements. The confidence interval of the de-embedded insertion loss is obtained after statistical analysis assuming the variations among fixtures are subjected to Gaussian distribution.