Site icon Creative Electron – The X-ray People

A New Standard for Standards – From Data to Information

A New Standard for Standards – From Data to Information

Dr. Glen Thomas

The main challenge we have today with our manufacturing standards is that they are deterministic. For example, for X-ray inspection of BGAs, the standard is a 30% maximum void per ball. This standard determines if a BGA assembly passes or fails. Now, there’s nothing to say that a ball with 29% void could not fail and there’s also nothing to say that a 31% void in a single ball could not work. It’s understandable that for a manufacturing process you need to set some clear parameters to define what’s a pass and what’s a fail, otherwise it becomes very complicated to set quality standards for a manufacturing process.

Now, it’s important to keep in mind that these deterministic thresholds, for example 30% maximum void per ball on BGAs, were developed in a time when collecting data from instruments in a manufacturing line was very costly. And further, contextualization of that data, meaning the process to input all the data that’s collected from the individual instruments and inserting that into context, was even more expensive. In some cases, it was technically infeasible.

So, what has changed with Industry 4.0 and other initiatives from equipment manufacturers? The answer is simple: data is more available. Individual instruments are already collecting a lot of data. But that data has not been used or contextualized. Data is getting cheaper. However, data out of context has another name, it’s called noise. Contextualized data, on the other hand, is called information.

At Creative Electron, we’re proposing a new perspective on setting pass and fail thresholds in the manufacturing line, based on actual performance and test data, rather than an arbitrarily set number. Instead of setting deterministic numbers, like the 30% void, we would collect data and determine what’s a pass and what’s a failure. Thus, scrap is minimized, since you only rework or scrap the parts you really need to reject because you know they’re likely to fail. What’s more, you gain a greater understanding of where to set the pass or fail parameters.

This, in a way, is what is being done now for QFNs. There’s no clear directive for voiding on QFN. So, when asked by our customers, we advise they follow the guidelines of the component manufacturer, based on what works for that specific component and application.

How Does this New Solution Work?

Take the example of a new board with a BGA. With the normal NPI (New Product Introduction) process, you would assemble several boards and would X-ray the BGAs and measure the voids on each one of the balls and check if you are below the 25% or 30% void in each of the balls, depending on the class of product that you are developing.

We propose still collecting the data, but instead of using a simple pass or fail threshold, we suggest testing and using the test data to determine the threshold. So, we collect more and more data to set a dynamic threshold that can move up or down depending on the results we have from real tests. This way we use the actual test data that’s available to fine tune our manufacturing thresholds.

There are several test parameters we can use to determine pass or fail in high volume. Going back to our QFN example, if there’s a temperature band that the component is supposed to work at, we can place the component and measure with an X-ray system how much voiding we have on the QFM, and using an infrared or laser thermometer, we can determine if the temperature guidance is being followed.

Smart factory solutions are not just about collecting data, they are about using that data intelligently to make faster and better decisions. We think this is a great example of using the data derived throughout the line to create a more efficient and more effective use of X-ray and rework resources.

Exit mobile version