Is Product Quality Getting Lost In The IIoT?

Manufacturing operations have continuously evolved using data capture and management to assess and test production effectiveness on the manufacturing floor. The advent of the Industrial Internet of Things (IIoT) and its anticipated ability to track and manage the factory environment with machine-to-machine process analytics heralds yet another transformation, promising a higher level of data intelligence. “Industry 4.0” will have a substantive impact on manufacturing business models and profitability, which is a fundamental justification for its advancement. Perhaps because of the tangible expectation of increased profit, much of the focus for IIoT investment has been toward manufacturing process improvements, with less emphasis, or acknowledgement of, the resulting effects on the quality of the product that is being manufactured. In this era of burgeoning big data analytics, product quality’s impact on enterprise profitability should not be overlooked.

In all industries, substandard product quality can negatively impact financial performance and company reputation. Scrap material costs and recalls resulting from defective parts circulated into the supply chain can compromise an organization’s profitability. In the semiconductor and electronics industries, each manufacturing step has the potential to affect product quality either positively or negatively.

Capturing Product “DNA”
Data availability within the enterprise is exploding at an unprecedented rate, and when captured and analyzed correctly, provides previously unimagined levels of intelligence for product quality control. One area where product analytics is distinct from current data usage is in its ability to map and track a device’s test DNA (tester and test program info, individual test results, touchdowns, etc.) and correlate that data to other similar devices being manufactured. Every device encounters its own separate life experience in the multitude of production steps it progresses through, and this level of information is simply not captured in process analytics.

Process analytics can optimize each step separately, but not through to the functional quality of the end-device. However, the sourcing of enterprise-wide data combined with the power of big data analytics can enable improved product analytics that can be used to track facets of an individual device that enhance quality from source to end-user performance. Effective analytics is a result of the breadth, relevance and timeliness of the data aggregated. Data abounds on the manufacturing floor. The ability to gather and unify all this complex data from disparate enterprise geographies, supply chain partner companies, and functional disciplines creates opportunity, but also introduces the challenge of effective management of manufacturing test data.

The Challenge of Big Data
Product analytics is practiced successfully by many semiconductor and electronics companies, delivering profound new insights, real-time product quality control and increased profitability. The key is to properly collect data to have actionable impact throughout the enterprise. For product analytics to be more successful, data must be relevant to the goals of the organization. It must also be actionable, defined as:

• Available quickly (“real-time”)
• Processed immediately and automatically
• Accessible to the supply chain

Defining Product Quality with Big Data
If enterprise manufacturing data is well-sourced and effectively managed, there are system-wide benefits unrealized in current production practices. There are also many assumptions in the definition of “quality” based on current data practices that can be expanded upon with product analytics.

Maintaining a record of device-specific test “DNA” enables quality control beyond an individual chip’s absolute performance. A chip may be “good” when combined with a specific chip and “bad” when combined with a different chip. Or it can be good in one customer system and bad in another customer system. Knowledge about different grades of “good” supports smart binning of chips, and can be used for things like smart pairing of chips in a system to improve system performance, and adaptive test time reduction to improve cost. Chip combination testing expands the quality definition to include end-device performance in multiple end-use environments.

Product Analytics Throughout the Supply Chain
With all the business risks associated with releasing poorly performing or faulty devices to consumers, big data will be a necessary tool to cultivate product quality “DNA” from the semiconductor through to the electronic end device. As a device moves from chip to board to sub-system, and from manufacturer to OEM to end user, the complexity of data multiplies, making it challenging and costly to quickly identify the source of failure. With product analytics, this information is quickly traceable along the entire manufacturing process through to the end user. With the advent of the IIoT, it is imperative that product analytics information be combined with process analytics to support the next era of industrial automation.