Application of JMP® Statistical Analysis Tools for Process Characterization


Pradipta Kumar Das, Process Development Engineer, Medtronic

This paper discusses the application of tools available in JMP to plan, execute and analyze a combined process characterization of two consecutive processes. Historical knowledge and assumptions of effects between upstream and downstream processes were used in experimental planning, execution and data analysis. Five experimental factors studied included one hard-to-change factor (F1), and four other factors (F2 through F5) across the processes studied. A custom experimental split-plot design with a main effect, quadratic, three-factor interaction model was selected with the purpose of performing a propagation of error analysis across process steps A, B and C. Prediction variance, fraction of design space and power analysis resulted in a DOE consisting of 12 unique runs of F2-F5 combinations for each of the five whole F1 plots, for a total of 60 experimental runs. Eight responses across processes B and C were analyzed. Data analysis included using Graph Builder, Distribution Analysis and Fit Y by X to check data integrity. Then Fit Model with iterative term reduction was used to create the models. Optimization of defect rates was performed with the profiler and simulator tools. Finally, within and between runs variation was assessed using the residual by row plots.

Article Tags