METROLOGY AND STATISTICAL QUALITY CONTROL Automated SQC through the integration of measurement tools is easy with appropriate software.

BY JOEL D. FREEMAN Why make measurements in the first place? DeWayne Sharp, retired former manager of IBM's Standards Laboratory, believes the fundamental question is as much practical as it is philosophical. "The real thing that people need to do is to make decisions. If you're going to make the correct decision, you need to make that on numerical data. And if you're going to have valid numerical data on which to make your correct decision, they ought to be based on some kind of measurement," Sharp says. He takes the argument a step further: Measurement data becomes incrementally more meaningful with the combined application of statistical quality control (SQC) and tools of metrology. In fact, application of quality programs such as SQC has become essential for modem manufacturing survival. For many, however, learning and applying new technologies and techniques can prove daunting. The solution is automated SQC through the integration of measurement tools with appropriate software. Because as Jeffery Cawley, founder of Northwest Analytical, Inc., (NWA), says, "What's easy gets done." "Modern measurement systems," he says, "have made the SQC process almost free. Manual collection of data is eliminated. Therefore, it costs a corporation relatively less to automate, and doesn't impede the primary job of making a product. We need to make measurement and data analysis as unobtrusive as possible to increase the user's gain. By design, SQC is a problem-solving tool and should operate as such." Cawley adds that automated data collection allows for more critical analysis, "which is what the human does best," not number crunching. "So tying in software with measurement is pretty fundamental to the widespread acceptance of SQC. It reduces the burden and keeps you focused on the parts of the business that benefit most from your attention." Integration of SQC software with tools of metrology can occur several ways. The simplest version converts tool output to data through the use of physical or software buffers. However, a growing trend is to include data stations--complete PCs, actually--with modern measurement tools. More complex systems feed tool-gathered data to network databases for later processing at workstations or PCs. SQC software selection is critical. It must be adaptable to both automated function and integration for interoperability. Many tool OEMs and systems developers have chosen NWA Quality Analyst for its configurability, versatility, and compatibility with a variety of hardware and software configurations.

Data collection and analysis Standalone. Coordinate measuring machines (CMMs) "look" at an object in three dimensions, giving several critical measurements quickly and conveniently. LK USA, Inc., a wholly owned subsidiary of Cincinnati Milacron, has increased that convenience by including NWA Quality Analyst in its instruments' data stations. LK designs and manufactures a wide variety of tools used by the "big three" auto makers plus other users of large CMMs, says regional sales manager Joe Stokell. "We have five lines of measurmg machines. All

of the lines that we have are either horizontal arm style or vertical bridge style." The model 380C bridgestyle CMM is the premier NWA Quality Analyst implementation. CMM users need SQC, says Stokell, because "everybody out there now is after everybody else's market share. We do most of our business with automotive people. It is essential for them to have a statistical database and a historical database.. to do any subcontract work for Ford, GM, or Chrysler." A command from the 680C data station initiates real-time data collection, data extraction, and chart production. "It can either go into the disk drive of the system that we're using or across a network into a PC that basically does nothing but host all the database and analysis data," Stokell notes. He stresses that SQC chart use is not limited to engineers. "Most of the use is by the guy who' sup to his elbows in work trying to get the parts inspected. The factory engineer can relate to it as well because statistics are statistics, once you generate the data."

Automated International industrial giant Siemens manufactures the only x-ray fluorescence spectrometer with integrated SQC software, giving users automated production of charts. Such automation is essential says John E. Martin, Siemens product manager for XRF, "to eliminate the typical typographical errors that can be expected when producing hundreds of charts per day. If charts are produced for 10 elements per material with five materials, the result is 50 X-bar and R charts. Modern steel facility controls are usually run once per shift (three times a day)." Manual chart making would overwhelm the process control lab. Element intensity and concentration data are automatically exported to NWA Quality Analyst. A macro from the spectrometer's host software then initiates production of SQC charts. NWA Quality Analyst allows chart specifications to be customized by each user, providing the flexibility required for the different stages in a production environment. When used with multitasking computer hardware, the Siemens instrument can run additional samples while NWA Quality Analyst simultaneously analyzes the data. The charts, Martin says, display information in a format understandable to lab, production, and management personnel. Automated data transfer allows the spectrometer to be integrated into plantwide networks and control systems. "This is a tremendous tool in providing information to solve overall production problems," Martin adds, "especially when the instrument's variation is shown in the same format as production variation."

Software interface Many analytical tools, though sophisticated, cannot feed into PCs or other computers without some sort of buffer. In addition, large corporations or manufacturing environments may use instruments from multiple vendors. Data storage, extraction, transfer, and analysis can then present significant challenges to product improvement systems. To meet such concerns, Canadian metals producer Alcan developed Al-OES, a proprietary PC-based software for metals analysis and quality control. Al-OES has proved so successful that Alcan anticipates using it with some 65 instruments within a year. Though the software was developed for in-house use, Alcan staff programmers recently received corporate permission to commercialize it as a third-party product for sale to other foundries and metals laboratories. Al-OES is compatible with several manufacturers' instruments, including ARL and Spectro, creating a bridge for consistency in data output, as well. According to Frank Kimmerle, chief analytical chemist for Alcan, Al-OES provides a PC interface for data collection while controlling the physical function of the spectrometer. Its facilities include configurable

preset calibration methods, analytical processing, data storage and retrieval, data export and transmission, a test package that provides a performance profile of the instrument, and data selection and extraction functions that interface to NWA Quality Analyst. The "miniexpert system" measures spectrometer performance and analyzes the data gathered. It then performs a series of go/no-go checks and, in the case of negative reports, shuts down the process or alerts the user to make appropriate adjustments. Alcan's Oswego, NY, foundry produces 1 billion pounds of aluminum sheet products annually. Its finished product is principally used as beverage container stock. The foundry analyzes alloys with a Fisons optical emission spectrophotometer, also known as a quantometer. Alcan's Dave Lagoe has automated analysis of monthly production data at Oswego through enhanced use of AlOES. By using NWA Quality Analyst's macrolike scripting language, Lagoe extracts alloy data, analyzes it, and prints charts in just three steps, performing as many as 10 analytical functions. A monthly data file contains some 1,500 rows with typically 78 columns each, Lagoe says. Only 25 or so columns contain useful data. File conversion, he says, takes 5 to 10 minutes--"not my time, but on the computer. I not only extract for alloys, I also extract by casting center--by piece of machinery it was produced on." File extraction takes up to 45 minutes on the computer. "It takes me one minute--just run the file," he explains. "I have a master script file that has all the alloys I want to produce charts for." He leaves that task for the end of the day because printing can take up to three hours. "The script file capability is the magic in the software, for sure. "SPC tools have helped us quite a bit in fine tuning our operation," he adds. "Quality Analyst has just made them much easier to use and enabled us to use them to a much greater extent."

Metrology and CIM Complex products and processes require sophisticated measurement and data processing systems. At the forefront of CIM technologies are larger data systems such as MESA (manufacturing execution system application) by Camstar Systems, Inc. The AS/400-based networked system for manufacturing floor operations provides a plantwide view of the manufacturing environment while creating a link between MRP-II (manufacturing resource planning) and machine control systems. Among MESA's many capabilities are data collection and analysis for continuous process improvement. Parametric, yield, and quality data are collected in a relational database on the AS/400. Virtually any kind of measurement instrument can feed into the database. MESA performs quality analysis two ways: user-initiated batch mode or automated, when the system detects that a product or process is out of statistical control. Though the system has some statistical analysis capabilities, it still needs a chart server to present that information in a graphical format on factory floor PCs. To fill that need, Camstar recommends NWA Quality Analyst to many MESA users. "We probably have the tightest integration with NWA Quality Analyst," says Greg Sowle, Camstar technical marketing consultant "It was easier to interface and was more user-friendly." MESA user International Microelectronic Products (IMP), San Jose, CA, designs and manufactures application specific integrated circuits (ASICs). The 11-year old company's products are used in the microcomputer, computer peripherals, and communications markets, among others.

IMP uses the MESA system to collect manufacturing information plus parametrics, concentrating on operations keyed for quality control, says Philip W. Frey, CIM coordinator for IMP. "We are running 128 charts in NWA Quality Analyst and adding more weekly." "When there's a violation of a trend or [a process goes] out of the limit, it causes an alarm file to be written. The operator can then push a button, have the file automatically download to their PC and come up as a NWA Quality Analyst control chart." The operator can then determine the proper correction, if any. In addition, Frey says, engineers can routinely check critical plantwide operations without leaving their desks. "They can do a review of the alarm list on their PCs. They can filter the list in terms of time, or piece ofequipment or the actual spec. They can just go down [the list], and by pushing the number bring up a graph of the data that caused an alarm to be issued," he adds. "Many customers want monthly reports of what their products are doing, or what the process their product is running on is doing," Frey says. IMP automates historical data analysis by combining a MESA batch facility with NWA Quality Analyst's scripting language. The analysis can be defined by any combination of attribute, variable, and product. "You run a query on the AS/400 database and download the data into a Quality Analyst file. The header file is already written with the relevant specs. The engineer is asked what dates he wants. Then, he pushes a button and gets back his control charts. You can get all the data you want." The impact for IMP extends beyond improved product. "The big benefit is that we've made the data input the same across all departments for the operators, and we've made the data extraction the same for different groups of engineers. Another benefit Frey and IMP gained is automatic file updating. MESA's master filing facility combined with NWA Quality Analyst's data extraction capabilities lets Frey automatically update files on networked PCs. "We have these 15 workstations, and I need to only update the shared folder in the AS/400. I'll know that everyone is going to be updated as soon as they log in." Adds Frey, "It's really nice, especially dealing with a fab environment." With floppy disks, "it would be a full-time job to keep 15 PCs updated." Frey found automated data transfer easy to master. "I'm not a programmer, and I do it, so that's proof. I've been able to learn it."