Who should control Part Average Testing?


With ever-increasing demands in the automotive industry, more and more semiconductor companies are interested in monitoring and improving quality and reliability. Outlier Detection, more specifically, Part Average Testing (PAT) is the industry standard for the automotive industry. But, who is responsible for quality?

Historically, OSATs are responsible for this. In the past, once they confirm that they tested it, the batches are signed, sealed and delivered. More and more we are seeing fabless companies who want to take control in-house. It begs the question, “Why on Earth would they want the added responsibility of ensuring quality?”

First of all, we’ll cover the basics:

To understand PAT you need to understand Outlier Detection.

What is Outlier Detection?

Outlier detection is a method of identifying a member in the group that deviates grossly from the norm.

You test semiconductors across different parametrics, such as voltage, current and how it reacts to different stimuli. You calculate the average manually or let your Outlier Detection system crunch the numbers for you. Using an algorithm, outliers that deviate vastly from the norm are found. In other words, you calculate what is average, then exclude what isn’t.

What is Part Average Testing (PAT)?

PAT is one way to apply Outlier Detection. Others include Out of family (OOF), Good Die Bad Neighbourhood (GDBN) Statistical Bin Limits (SBL) and Statistical Yield Limits (SYL).

Who needs to control this?

In the past, the test house was responsible for controlling it, as the action of calculating and applying different limit sets needed to be close to the testers.

With a modern system, there is no need to change test limits. As this is done on a private cloud server, so it can be anywhere in the world. This enables the engineer to control the system at the touch of a button, anywhere. For example, the Product Engineer could be in the USA while manufacturing and testing could be in China. Management is able to see the process too. In some cases, I’ve seen Fabless customers share data with their customers. Outlier Detection is now more transparent, so the Fabless companies can monitor and control the process as they wish. Rather than being forced to hand over control to a test house, PAT is now firmly in the control of the Product Engineers who oversee the manufacturing and product quality.

Another key element of putting the fabless company in control is that all data, the original data and the additional data created during PAT and GDBN processing, is stored in the database on the cloud server. So full traceability, every limit for every test site on every wafer with a record of every die through that limit, is stored under the control of the fabless company. We work with one fabless company who wants their data archived so they can access it years later. The industry has moved from Fabless companies having zero control and relying on contracts, to having full visibility and peace of mind, which they can share with their customers.

Why should fabless companies take control of quality?

Fabless companies who want to work with automotive companies should take responsibility for quality and reliability. Quality is becoming more and more important to automotive companies. Given the choice between working with a fabless company who can guarantee the quality and a company who can’t, the automotive company will choose the Fabless company who ensure quality even if it costs more. As it costs a lot to manufacture cars, they can’t afford defects.

This is true of automotive companies, but we’re seeing more and more consumer electronics (for example, smartphone manufacturers) who demand quality. The reason you need to invest in quality is to win new business and avoid losing existing business. With the fabless company or product group in an IDM, owning the PAT process then it can be set up exactly as needed to give the highest quality for the smallest yield loss. Choose which tests to include in PAT processing and choose which bins to include in GDBN. “Who should control your Part Average Testing?” You should control it, efficiently and from your location. Product quality belongs in the hands of the product owner.