For over 20 years, this global software and engineering services company has been helping semiconductor manufacturers increase yield and quality. It provides a big data analytics platform to optimize the quality and reliability of semiconductors going to assembly and test houses for use in automobiles, cell phones, and consumer and industrial products.

"We're the only company that has end-to-end insight into the manufacturing process and the entire product lifecycle," said a product manager for the company’s analytics platform. "Data is collected from design all the way through to final test."

Turning high-volume, high-velocity streaming data into insights to improve manufacturing outcomes is the company's core value for customers. However, that hasn't always been the case.

Before implementing a data and analytics solution from TIBCO, the company struggled to ensure data quality, integrity, and diversity, especially for equipment sensor-trace data provided at sub-second sample rates. Customers needed to process and understand massive streams of unstructured data from hundreds of sensors per tool and thousands of tools per factory to show the quality of wafers being manufactured.

Initially the company developed and maintained its own analytics engine, but over time, faced significant R&D and maintenance costs, as well as the lost opportunity of not focusing on its core competency: semiconductor-specific solutions. As the product manager explains, "We couldn't solve customer problems as fast as we wanted."

The company wanted a solution that would work within its platform and supply equal parts interactive analytics and sophisticated visualizations. The team found that TIBCO offered the best approach to its data analysis challenges. Since its initial adoption 10 years ago, it has seen TIBCO's analytics solutions evolve to stay cutting-edge with an open architecture and APIs that enable the team to build solutions for its customers. Today, the services company can offer customers an integrated visualization platform they can build into their own systems.

With TIBCO's help, the company created a smart manufacturing data platform that ingests information, analyzes data streams, and displays data so customers can generate insights and make quick decisions to uphold quality. "We're moving data from traditional organizational silos, exposing it across the manufacturing lifecycle, and building models to really understand what’s happening in our customers' products and help them optimize quality and reliability," said the product manager.

Digital twin software models represent manufacturing processes, showing what's going on in foundries, assembly and test houses, and customer and subcontractor factories. These digital models, coupled with global and secure availability of data in the cloud, allow machine learning models to actually control processes. As the product manager states, "Having global access to data is really a huge shift from only a couple of years ago."

The speed of insight and faster decision-making is extremely important to the company's customers. Data can be used to quickly predict parts output that before took months to attain. Additionally, the company can help customers minimize adverse effects and control its entire production line to meet product fulfillment goals and launch dates. Data analysis occurs in near real time.

The company invented a smart testing solution to determine test resources needed to keep parts output within quality specifications. "We identified variability in certain manufacturing tools to see how they affect the final part," said the product manager. "That was extremely valuable for our customers. We could tell them how to control processing tools to have a direct impact on the quality and performance of the parts manufactured."

Test burn-in, an expensive operation where parts are tested under adverse conditions to try to make them fail, was made more efficient. With more intelligence throughout the process, the company's customers achieve better ROI. When parts only fail after the prescribed period of time for acceptable quality, customers can have confidence and save significant testing time and money while meeting product launch dates.

"By analyzing all the data, we identify parts that are particularly risky and those that are less so," said the product manager. "Then I can determine whether to devote more testing resources to make sure that that part won't fail, or if it’s perhaps a lower quality part, maybe not spend any more resources on testing because it's not worth my time."

For this use case, the services company used semantic modeling to structure data and deploy machine learning to production. Risk factors are assigned to a part and the test flow can be controlled so risky parts can be thoroughly investigated and identified as potential failures sooner. Less risky or higher quality parts can also be identified and treated appropriately.

With a platform that’s been proven over more than a decade, the company knows it’s empowered with the right analytics to solve manufacturing problems for its customers. Now R&D solves the most difficult problems using interactivity, automation, reporting, and openness of the analytics platform to provide customer value instead of maintaining a homegrown tool.

"Partnering with TIBCO really enabled us to focus on our customers' problems and understand what they were trying to solve in their manufacturing environments," said the product manager. "We were able to take that learning and develop our platform to make them successful."

A partnership with TIBCO enabled this company to equip its customers with factory capabilities for the Industry 4.0 era. This semiconductor consultant is enabling connected and smart manufacturing for operational excellence for customers around the world.