Automating Analysis of Anomalous Products
SigmaQuest’s Automated Maverick Analysis Software
Power-One Inc. products power high-availability infrastructure applications ranging from alternate energy, routers, data storage and servers, wireless communications, optical networking, and medical diagnostics to railway controls and semiconductor test equipment. As a global power systems manufacturer, Power-One, headquartered in Camarillo, California, has operations in Asia, Europe, and the Americas. To more easily identify marginally performing products, in real time, across its entire lineup of products, the company sought to automate analysis of anomalous or “maverick” products.
According to the Joint Electron Device Engineering Council’s JESD50A standard, a maverick product is atypical and exhibits anomalistic characteristics that cause a higher-than-normal level of failure in the end-user application. Causes of a maverick product can include fabrication, assembly, test, packaging, and handling operations. Maverick analysis is the process by which maverick, or marginally performing, products are detected. Traditionally this process has been manual and is extremely time-consuming. As a result, it could only be implemented on a small scale and, in Power-One’s case, at very few locations and only on high-volume products.
In cases where maverick analysis is not performed, affected populations can go undetected until used in a final application, adversely affecting contract manufacturers, field reliability, and end-user satisfaction. Yield-improvement programs, unfortunately, do not necessarily provide protection from maverick products.
As Chris Rodriguez, head of worldwide test engineering at Power-One, explains, “We knew maverick analysis was key to ensuring high-quality products, but it was difficult to implement and even harder to incorporate into the manufacturing process. Manual maverick analysis on a large telecom system, for example, could take us weeks while we waited for all the data to be aggregated and analyzed. That’s just not feasible given our tight deadlines. We would perform statistical process control, then maverick analysis, and finally a root cause analysis to try and figure out why something failed to pass or barely passed.”
The answer, Power-One realized, was an automated maverick analysis, but no such solution existed. That’s when Power-One turned to SigmaQuest Inc. for help. SigmaQuest, headquartered in Sunnyvale, California, had previously provided Power-One with a real-time data collection and analysis solution that allowed the company to see data in unique ways. The solution also had the capability to allow Power-One to see data in familiar ways, prompting Power-One to request that SigmaQuest help create automated business review charts and monthly data-review reports. Automating maverick analysis seemed like a logical next step. Through close collaboration with Power-One, SigmaQuest was able to fulfill that request.
The resulting solution—Automated Maverick Analysis—is an in-depth analysis capability that quickly identifies which products meet the highest-performance criteria and which exhibit anomalistic characteristics. Available as part of SigmaQuest’s SigmaSure 6.1 product performance intelligence application, this capability helps original equipment manufacturers and their suppliers improve and consistently make higher-quality parts. It also enables implementation and management of supplier-related maverick product elimination.
“SigmaQuest’s Automated Maverick Analysis provides tremendous value, allowing us to quickly identify marginal performing products in real time,” says Rodriguez. “With this knowledge, we can make best-case decisions on the individual products that will deliver optimum performance. In essence, this tool provides us a means to measure consistent quality of our people, processes, and materials. Any deviation from the norm generates an alert.”
Automated Maverick Analysis provides median/mean performance deviation information on parameters such as voltage, frequency, current, and resistance for complex devices ranging from integrated circuits, process control blocks, and batteries to power supplies. Parametric/performance data are automatically uploaded into a database format with easy-to-use views, including statistical diagrams, and tables containing detailed metrics. These data can then be utilized to further refine, or narrow, key performance specifications, such as the operating voltage range on a power supply.
“The fact that this time-consuming analysis is now automated makes it more feasible to apply it to a wider scale of products and locations than were possible beforehand,” Rodriguez says. “We find ourselves more proactive, instead of reactive. Our quality teams now spend more time responding to and fixing problems they have been alerted to, rather than spending their time finding data. This cultural change has allowed us to not only improve our products and manufacturing process, but enabled our quality teams to work more strategically and cost effectively as well.”
SigmaQuest’s Automated Maverick
Analysis Software
- Identifies marginally performing products in real time
- Helps original equipment manufacturers/suppliers improve and consistently make higher-quality parts
- Implements and manages supplier-related maverick product elimination
- Provides median/mean performance data for complex devices
- Automatically alerts users to maverick products
www.sigmaquest.com
|
