French startup tackles data overload in silicon manufacturing and test

October 22, 2012 // By Nick Flaherty
Montpellier startup Qualtera has developed technology to automatically process and analyse the millions of data points on yield and performance for silicon manufacturing and asssembly.

Silicondash is an automated Decision Support System (DSS) for semiconductor test data analysis that runs in a choice of secure data centres. It takes data direct from test equipment in the foundry and assembly plant to produce a real time hierarchy of reports on yield and performance.
Even though there are millions of data points, the aim is to reach the required data in four mouse clicks, with each click taking a maximum of two seconds, says Paul Simon, VP marketing and operations, and co-founder of Qualtera. At the same time all the data is kept secure and encrypted, he says.
The company was founded in 2010 by a team of experienced semiconductor industry professionals from Philips Semiconductors, NXP Semiconductors, PDF Solutions, and Si Automation.
“To collect and extensively analyse manufacturing and test data can require days or even weeks from an entire team, slowing down decision processes and adding cost,” said Simon. “From our team’s background in semiconductor manufacturing, we could see that a different, more efficient, approach was needed; one that speeds up the process, simplifies the analysis, and reveals abnormalities instantly. Silicondash does this and more, by letting people in multiple locations access both the raw data and the statistical analysis results in real-time. Being cloud-based, it’s also much more cost-effective than existing solutions and only requires web access. By using Silicondash, our customers speed up their problem detection and improvement processes and thereby save significant time and money.”
Currently engineers need to compile test data from multiple sources and then add their own analysis, a process than can take hours or even days. By using Silicondash the analysis and reporting processes are to a large extent automated and can be completed in minutes. Very detailed statistical root cause analysis, as well as high-level aggregated summary reports are accessible in those four clicks.
It is designed to handle all types of test data regardless of product type,