Senior Director of Product Marketing proteanTecs, Ontario, Canada
When does manufacturing end and system health begin for advanced electronics? For modern electronic systems the line between manufacturing and in-field use is becoming blurred since the goals for understanding the health of the device are converging. Most electronic systems require repair or calibration to enable a working product to ship to the customer, which means that product test becomes a manufacturing step, not just a pass/ fail decision. These methods can then be re-purposed in the field to ensure longer operation of systems by adjusting parameters over time. These adjustments over time can also be applied to determine system health by leveraging advanced data analytics for predictive monitoring. To achieve this with existing data would require product reliability studies and unique models to be developed which would impact time to market for your product. As a result, Smart Manufacturing not only occurs at time zero to ensure a working product, but throughout its lifetime to ensure optimal operation.
This presentation will discuss how Deep Data and Machine Learning will enable smart manufacturing and intelligent devices. This will be accomplished by determining the initial health of the system, optimizing performance at time of manufacturing and then extending the learnings into the field for predictive health of a system. What is Deep Data? It starts with a purposeful on-chip monitoring method called UCTTM, Universal Chip Telemetry, designed to understand material variation, operational conditions, and application stress of a system. UCT agents which are built for analytics generate new parametric measurements at high-coverage and are designed with machine learning algorithms, driving new actionable insights and optimized data flow. UCT data with machine learning can then be applied in the cloud or edge, depending on the level of computation or the immediacy acting on the insight delivered. We will present an application highlighting the benefits of improved visibility that enables smart manufacturing. It will focus on production test where the UCT enables improved binning methodology with a trained machine learning model. This will improve yield understanding that minimizes latent defects as well as reduce tight limits which lead to device overkill.
