| "With the statistical charting and teamwork Haddadin and Ermer introduced, we enhanced our product and educated our workforce." |
In the business of manufacturing high-performance computer components, even small mistakes can cause big expense. Scrapping just one of certain kinds of printed circuit boards can cost more than $25,000. In an effort to avoid such costly mistakes, Madison's Heurikon Corporation, a division of Computer Products Inc., collects data for each printed circuit board assembly it produces. Computers collect information such as the circuit board's assembly serial number, the type and location of defects and comments from the operator.
While many companies find they don't have enough information to correct flaws in the manufacturing process, Heurikon ran into the opposite case. Too much data prevented them from pinpointing the source of flaws. Enter Donald S. Ermer and Russ Haddadin. Ermer is Procter & Gamble/Bascom Professor in Total Quality for the College of Engineering's Departments of Industrial and Systems Engineering and Mechanical Engineering. Haddadin was a process engineer at Heurikon Corporation and a teaching assistant with the college. (He is now a quality engineer at Honeywell in Illinois.) Haddadin used his expertise and Ermer's guidance with techniques known as Pareto Diagrams and Statistical Process Control (SPC) to transform data from a variety of products into something manageable and useful. In fact, Heurikon used the data to reduce defects in one production line by 80 percent.
"Every time we have used these techniques with an industrial partner we have made an improvement," Ermer said. "If it's a quality problem, we can do something about it. It takes time, planning and commitment on the part of industry, but if they stick with it, they will see results."
Heurikon's data acquisition system grouped together all information from all products. Aggregating and analyzing the data as if it were one product generated a number of quality improvement issues. This method of data analysis decreases the detail of the information and can have a detrimental effect on a process improvement program.
When the data is aggregated, a misleading picture of process capability is presented. Imagine a manufacturing environment where data is continuously entered into the database and the product variety numbers between 10 and 30 products per week with varying levels of complexity. The chances of confusing information related to products and processes is high. Thus, it is important to highlight the difference between the perceived versus the true process capability for the various products.
Procter & Gamble Bascom Professor in Total Quality Donald S. Ermer (right) and process engineer Russ Haddadin used Pareto Diagrams & Statistical Process Control to improve Heurikon's production lines. |
One of the first steps in solving this problem was segregating the defect data by product using a Pareto diagram, said Ermer. By ranking the quality performance of products, attention is focused on those with the most defects and the potential gain of reducing those defects is identified. As products with the most process defects are improved, others emerge as more critical and signal that attention needs to be shifted to them. This creates a perpetual loop of continuous improvement.
"Another significant step was getting the operators on the line to believe in the data collection system again," said Haddadin. "We needed them to understand that the information they entered was indeed being analyzed and would make a difference. That encouraged them to record defects. It was really enlightening for the operators."
Concurrently, the creation of control charts aided in identifying where defects occur and the process changes that take place. The control for defects, or C-chart, adds the time/order dimension to process improvement and control. It plots the history of the product's quality and signals any change in the process.
With this new outlook on data analysis, a number of opportunities surfaced. Heurikon can now monitor and improve the production process and quantify improvements that were otherwise camouflaged. In addition, many of the improvements implemented on one product can be applied to others. For example, the team found that reducing the pad dimensions on one product resulted in a reduction of solder bridges between pins. This pad dimension has been added to manufacturing specifications. All board layouts will now use this updated standard.
"With the statistical charting and teamwork Russ Haddadin and Don Ermer introduced, we enhanced our product and educated our workforce," said Heurikon's manager of engineering manufacturing Chris Byrne. " It also gave Russ an application to put these techniques to work. It's definitely something our process improvement and quality engineers will continue."
A by-product of the new method of data-driven problem solving has led to the introduction of product focus teams. Product focus teams analyze the data collected, identify root causes of defects, and generate a number of options that could solve the quality problems identified. Once options have been generated, an estimate of the expected improvement is determined and results from the next production run are monitored to see if this improvement goal was attained. In addition, information that previously fell through communication gaps now flows throughout the organization. Knowledge of quality issues that previously never left a department or factory floor is now communicated to others for additional improvement.
Ermer and Haddadin said the results of this analysis are already apparent. For 35 boards of a new product production run, the team was able to identify a number of defects that were either consistent or that occurred periodically in the same locations. Options for modifying the process and board design were generated by the team, and a list was then given to manufacturing engineering for review. Within a week, the board layout was updated, and wave soldering parameters were modified for the product. This lead to significant improvements including cycle-time reduction and elimination of expensive defects.
--By Jim Beal--
| For further information, please contact: |
Donald S. Ermer, 608/262-2557
ermer@engr.wisc.edu
Copyright 1997 The Board of Regents of the University of Wisconsin System
Content by perspective@engr.wisc.edu
Markup by webmaster@engr.wisc.edu
Date last modified: Wednesday, 19-Mar-1997 12:00:00 CST
Date created: 19-Mar-1997
