An investment in automated video measurement helps improve quality assurance.

When UltraSource Inc. (Hollis, NH) experienced a significant increase in parts production, they knew they needed faster and more reliable inspection than what was being used. After looking into various optical inspection options, the quality manager discovered a way to expand quality assurance (QA) productivity while increasing the ability to gage fine tolerances.

UltraSource Inc. is a supplier of thin-film circuit devices and related services. The company produces components for both commercial and military markets using precision sputtering, photolithography and dicing technologies. They supply their customers with a range of thin-film expertise to produce products from prototype samples to high-volume production. These parts are used in everyday products such as cellular and wireless communications, microwave components, fiber-optic, infrared and laser systems, as well as other products and applications.

In their QA operation, UltraSource originally used two high-end, manually operated toolmaker microscopes with video and digital readouts. UltraSource initially assumed its needs could be met by purchasing an additional manual-measuring microscope similar to the one currently in use. After soliciting quotes for new systems, their QA manager quickly discovered that for $12,000 more, an automated video measuring system could be purchased and add enhancements to the data collection system that would further its ability to service customers' needs.

The company looked at automated video systems to solve its two main mechanical inspection problems. The need to automate because of a greater volume of parts, especially those from two high-volume customers' thin-film applications, a tight-

tolerance military part and a large-volume, difficult to inspect, commercial device. Additionally, the engineering group needed a reliable and easy-to-learn video system that could collect data to aid in optimizing the manufacturing processes.

Before launching its search, UltraSource developed five primary requirements that an automated vision measuring system needed to meet:

• A system must increase throughput of final inspection and reduce overall inspection time by automating high-volume inspection applications.

• The system must improve present gage repeatability and reproducibility (GR&R) errors found when operators used the manual microscopes. Studies done in conjunction with the company's largest customer showed UltraSource's GR&R to be as high as 0.000343 inch, which is more than 59% of the total tolerance being consumed in error by the combination of the operator and equipment when using the microscope.

• Data needed to be collected automatically because process improvements that require ongoing CpK analysis require constant data collection. Using the manual microscope, it was necessary to have an operator manually record data and submit it to the quality assurance manager for analysis.

• The system had to be easy to use and understand after programming because semi-skilled users would be the primary operators of any vision system.

• Additionally, the system had to be flexible enough for the skilled operator to perform more complex tasks such as collecting and analyzing data from process improvements.

With these well-defined parameters in place, several video systems were evaluated. In addition to the pre-set parameters, the systems had to meet tolerances of ±0.0005 inch or less and still be affordable. Based on its evaluations, only one system met all of its pre-defined requirements, and more importantly, could accurately detect their fine line traces while inspecting in an automated mode of measurement. The system they chose was the Galileo QC 5000 Automated Vision System from The L.S. Starrett Co. (Athol, MA). It is important to note, that UltraSource found that many vision systems, while meeting most requirements, simply could not detect and inspect UltraSource's type of devices.

GR&R reduced by 85%

The company invested in the video measurement system in August 2003 for the in-process and final inspection of thin-film circuit devices. The fully automated computer numerical control (CNC) system quickly proved its worth.

With the system, total GR&R error was reduced from 0.000343 inch to less than 0.00005 inch, an improvement of more than 85%. Independent studies done by UltraSource's largest customer confirmed that its measurements correlated with its high-end video system. The customer concluded that the combined GR&R errors between the two companies' measuring systems was now less than 0.5 micron.

In addition, real-time statistical data collection is now possible. CpK studies that once took hours to manually collect the data and study can now be done in minutes. Using a feature of the Galileo software with Microsoft Excel, an X Bar and R and short run CpK spreadsheet was designed. Thus, as the devices are measured, the data is automatically pasted to a data file and trended using the spreadsheets they designed. Without using third-party software, real-time statistical process control (SPC), gage studies and process improvements are now possible in minutes after devices have been measured. Further timesavings were made possible by creating forms for their customers in Excel as well and having the vision system data pasted directly into the on-line forms while the program is running. Using this method, data that was manually written and sent to customers, could be saved electronically and printed.

Within days, the lead mechanical inspector and other quality personnel were able to write and design complex programs for the highest-volume customers. Additionally, the software proved flexible enough so that lower volume, semi-repeat jobs could be programmed in a few hours.

Custom work-stage fixture

After training with the software, UltraSource took the productivity advantage one step further by developing a modified glass plate used to accurately position parts and make inspections faster for the two high-volume customers' products.

The unique design of the system's tabletop allowed for a quick modification which greatly enhanced UltraSource's productivity. The tabletop was measured and fitted with several precut 9-inch by 14-inch by 1⁄8-inch clear glass laminated sheets that were purchased from a local glass store. Triangular points were designed for use as positioning stops; each measured device needed three points to act as an inspection datum. Because the tabletop came pre-drilled with M4-tapped threads, four long thumbscrews were purchased to hold the glass and guarantee accurate glass re-positioning when removed. For less than $200 and a few hours, a simple palletized holding system was built that helped speed part inspection.

With the combined productivity features of the video system and new custom-designed fixturing, UltraSource was able to 100% dimensionally inspect product in 50% less time and electronically collect 100% of data on all features measured by the machine. Additionally, in-process inspections on more complex devices that once took three to four hours to complete are now done in less than 90 minutes. Inspection duration is reduced to the time needed to physically load and unload the pallets from the machine.

Sidebar: Tech tips

• The thin-film circuit device supplier was looking to replace two toolmaker microscopes, but research found that video systems would be more appropriate for the company.

• With the video system, total GR&R error was reduced from 0.000343 inch to less than 0.00005 inch, an improvement of more than 85%.

• Statistical data collection is now possible. CpK studies that took hours to manually collect and analyze the data are now done in minutes. Also, measured data can be instantly downloaded to an Excel spreadsheet.