Analysis of a Single Control Factor Experiment in the Presence of Several Strong Noise Factors

A case study of an experiment on a copper electroplating process in a printed circuit board manufacturing facility is presented. The experiment's objective is to determine the effects of noise factors including product type plated, product position in plating cell, side of product plated, and time variation, precludes the use of an analysis of variance, the favored data analysis tool at this manufacturing location. The advantage of using a paired comparison t test rather than an analysis of variance for the data in this experiment is shown. Making comparisons in matched pairs eliminates the effects of the noise factors and, thereby,significantly increases the power of the analysis