* Corresponding author. Tel.: #1-480-965-9068. E-mail address: doug.montgomery@asu.edu (D.C. Montgomery). Robotics and Computer Integrated Manufacturing 16 (2000) 377}382 Optimization of a bonded leads process using statistically designed experiments Alejandro Heredia-Langner , Elvira N. Loredo, Douglas C. Montgomery*, Alan H. Gri$n Department of Industrial Engineering, Arizona State University, Tempe, AZ 85287-5906, USA Durel Corporation, Chandler, AZ 85224-6155, USA Received 22 February 2000; accepted 20 May 2000 Abstract In manufacturing high-technology products using automated manufacturing equipment it is critical to obtain process-operating conditions that simultaneously optimize several output variables of interest. Often these output variables are non-traditional responses such as standard deviations, and there may even be discrete (count or binary) responses. Statistically designed experiments are very useful in optimizing these processes. We illustrate how factorial experiments, statistical models for the responses of interest, and simple optimization techniques can be successfully applied to a bonded leads process.  2000 Published by Elsevier Science Ltd. Keywords: Designed experiments; Response surfaces; Optimization 1. Introduction In the manufacture of certain electrical components, such as those used in communication devices, printed circuits or luminescent lamps, the "nal product is connec- ted to batteries or to other components through bonded metal leads. These leads are secured in place using a high-temperature, high-pressure process to bond the leads to the circuit. The metal leads are covered with tape keeping them in place and insulating them from the environment throughout the useful lifetime of the "nish- ed product. Several problems can arise from the bonding of the leads: defective bonds will prevent the device from functioning properly, extreme or erroneous manufactur- ing conditions may compromise the integrity of the cir- cuit or of the adhesive tape resulting in useless or weak bonds. The wrong combination of factors may lead to cosmetic defects in products where appearance is impor- tant to the consumer. In addition, a bonding process that has not been optimized may be consuming excessive amounts of processing time (especially when high volumes of units are being produced) decreasing overall productivity and needlessly adding to the manufacturing costs. In this paper, we demonstrate the use of statistically designed experiments to optimize a bonded leads opera- tion. The objective of the study is to understand the e!ect of several manufacturing variables on a set of responses with the intent of "nding the optimal operating region for the bonded lead process. 2. Description of the process and experimental design The bonded leads process is comprised of the following steps: A unit containing multiple printed circuits or lamps is put in position and the required number of metal leads placed above it. The assembly machine then "xes the leads in their position covering then with an insulat- ing resin, tape or a plastic case. The strength and appear- ance of the bonds and the integrity of electrical contact between the metal leads and the circuit during normal operating conditions are the decision variables to opti- mize. The manufacturing conditions selected should not adversely a!ect the performance of the "nished product as measured by appropriate reliability tests. Other elec- trical characteristics of the device are not a!ected by 0736-5845/00/$ - see front matter  2000 Published by Elsevier Science Ltd. PII: S 0 7 3 6 - 5 8 4 5 ( 0 0 ) 0 0 0 1 6 - 8