Figure 1. Pad Y-Location Drawing.
|
This case study is from a manufacturing process that attached a connecting pad to a circuit board. The location of the pad was critical because it was used by the customer. The operation that made this assembly was monitoring the y-location as well as many other features of the assembly. See Figure 1. After many mechanical improvements were made in the process, the process was still performing below expectations, and the General Supervisor asked me to examine the operation, focusing on the pad y-location as a start.
Figure 1. Pad Y-Location Drawing.
The operators in charge of this process had been collecting data on the pad y-location, but were using other means to accept circuit boards and adjust the process. Some of these were visual methods and discussion with the customer. So, the first thing I did was get their data, which was on a control chart. I added control limits to it. Control limits are derived from the process' own history and they tell the natural amount of variation for the process. See Figure 2.
Figure 2. Control chart of the process, when it was performing below expectations.
(Click the charts to view of the full control chart.)
The control chart in Figure 2, shows a number of important things. First, the upper control limits (UCL) and the lower control limits (LCL) on the average chart show the natural boundaries, or maximum amount of variation, that the process should have, when it is acting in a consistent and stable manner. If the process is acting in a consistent and stable manner, all of the points on the average and range chart will stay within the control limits. This process was running wider than the control limits. This process, as shown on the average chart, swings from high to low and often beyond the control limits. The range chart also showed one point out of control. Both charts indicated that something was influencing the process in an unnatural and disturbing manner. I explained this to the operators and began to discuss with them what could cause such variation or noise.
They explained there were many material and batch changes each day, and that the process was very temperamental, and required many adjustments. Specifically, they described how throughout the day they had to continually try to adjust the pad y-location to keep it in specification. The pad y-location was very important to their customer, who had complained about the amount of variation in the pad y-location, and so, they were very interested in keeping the pad y-location in specification.
After listening to their experiences with the process and reviewing the control charts, I suggested a plan for the next day. I suggested in the morning, when they started their shift that they make one adjustment to put the pad y-location on target, the center of the specification, and then make no more adjustments for a few days. They were very concerned about this plan. They reminded me how temperamental the process was and that it may create problems with the customer. I listened to their prospective, but convinced them that we should try it and that we would do extra visual inspection to verify the y-location of the pad in the outgoing material. After much discussion, they reluctantly agreed to try it, but were still very concerned about quality to the customer.
Our plan included,
This last increased frequency was very important to the operators to assure that they would not hurt the customer. Figure 3, shows the results of the first few days of our 'no adjustments' plan.
Figure 3. Control chart of process after 'no adjustments' plan was implemented.
(Click the charts to view of the full control chart.)
Note the significant reduction in variation in both charts. A great deal of new, profound, knowledge was gained with the 'no adjustments' plan of operation. The operators and managers were surprised and pleased with how much more consistent the process output was. After many 'mini-lessons' in the production department, they began to understand that no adjustments was the proper thing to do most of the time. They began to understand that they had been responding to noise, not signal. This was a revolutionary learning and improvement in the operation, that required no new equipment, expenditures, and only a little operator support.
Subgroup 14, marked with 'See note pad,' was a true signal and was the only point (other than the first adjustment to get on target) that the process really needed. It was easier, and more appropriate, to run this process with minimal adjustments. The process was adjusted back to target after subgroup 14 and that information was recorded on their 'note pad.'
The process was easier to run, had better quality and the customer noticed and appreciated the improvement. See Figure 4, for some 'before' and 'after' comparisons. Further work was also done on the pad y-location process to identify and eliminate the assignable causes of the change in the process at subgroup 14. Additionally, the knowledge gained in this application was used on other features of the circuit board on this assembly line to reduce variation and improve quality..
Figure 4. Comparison Table of 'Before' and 'After.' The reduction in variation was 54%, from 0.101-inches to 0.046-inches.
Measure Before After Total Variation (Performance 6 sigma - smaller is better) 0.101 - inches 0.046 - inches Performance (percentage of tolerance used by process - smaller is better) 211% 96% * Statistical Control Many points out control and other patterns of out of control Only one point was out of control * Note that by identifying and eliminating the cause of variation for subgroup 1 and 14, the two high and extreme subgroups, the performance would have been 53%. This would have been a total reduction in variation of 75%.
Not all applications will respond this well. However, many processes will respond in a similar way, when over-adjusted or not properly understood. The operators were intently focused on doing their best, but were making fundamental mistakes by responding to the noise that they saw in the data. Originally, many of the signals in the data were from over-adjustment. The customer was very pleased at the improved consistency and reduced variation in the circuit boards.
This is a good example of how good intentions and hard work are not necessarily enough. It also takes educated and supported managers and operators to achieve the real potential in a good process.
Would you like to know more or have McDonald & Associates work with you? Then Contact Us with your specific training and service needs. Questions or comments on this case are always welcome.
Return to Applications and Case Studies List
About Us Contact Us Classes and Services Home
Copyright © 2001, Michael L. McDonald
