DOE WITH ATTRIBUTE AND VARIABLE RESPONSES A CASE STUDY ON MINIMIZING SOLDER DEFECTS
Donald S. Ermer, Ph.D., P.E.
Procter and Gamble Professor in Total Quality
Departments of Industrial and Mechanical Engineering
University of Wisconsin-Madison, USA
Abstract: A randomized, full factorial experimental design with four parameters was performed to determine what the new wave-solder machine settings should be when using a new non-toxic flux in the manufacture of expensive and sophisticated printed circuit boards. The paper describes the practical planning, execution, and joint analysis of multi-responses, including both variable and attribute outcomes. A confirmation experiment produced no solder joint defects, and the amount of flux residue was minimized, leading to the elimination of the final washing stage for a very significant cost savings. The paper is a case study of how the powerful methodology of Design of Experiments (DOE) is applied in a real-life manufacturing environment.
Key Words: Design of Experiments, Attribute Responses, Three-Factor Interactions Significant