Improving First Pass Yield (FPY) with PathWave Manufacturing Analytics (False Failure)
2021-07-30 | 6 min read
Being a Test Development Engineer during my early years of career has given me exposure to manufacturing production environment. And being production engineer or technician, do you ever feel pressured in achieving high First Pass Yield (FPY) while maintain high board quality with high test coverage? I am guessing your answer to be “All the Time”.
Prior to Industry 4.0 transformation, it will take one with experiences to meet this goal. Tasks including retrieving measurement logs from system, crunching the data using scripts or populated into excel, analyzing the results before arriving at some insights. Not to mention the time effort incurred and possible human errors that could happened during these tasks.
We are more fortunate now in this Industry 4.0 transformation where more software applications and platforms are available to perform data analytics on-demand with high performance.
First Pass Yield is directly impacted by False Failure. The higher the False Failure rate, the lower the FPY.
False Failure refers to scenario whereby the failed board is simply being retested again and it passed; without any repair or rework performed.
With PathWave Manufacturing Analytics (PMA), we aim to optimize manufacturing efficiencies and improve product quality. Figure 1 shows False Failure dashboard which perform analysis on failure failures with the objective to provide insight to reduce false failure rate.
The bottom left chart in Figure 1 displays a list of boards with retest pass history. From here, user can tell how many retest cycles the board has went through. This is crucial information for certain product markets like medical and automotive, where quality of boards is of upmost importance. Typically, these markets have maximum retest policy like 3 retest cycles for example. If board does not pass after 3 test cycles, it will be scrapped. With this, one can ensure retest policy is being adhered. You can imagine the severe consequences if these unreliable boards flow out of the factories to the markets. Lives may be lost, and company reputation will be at risk. One must avoid at all cost.
The way PMA determines a false failure is if the board is retested pass within 20 minutes of its previous failed run; assuming that within 20 minutes, no repair or work is performed to this failed board.
[This 20 minute duration is user editable in application setting.]
With these false failure labels, information is further analyzed into the false failure pareto chart (bottom right of Figure 1). User can then focus on the highest contribution test to the false failure rate.
PMA provides user the ability to perform next level drill down to its measurement details and trend (as shown in Figure 2). As the saying goes, a picture speaks a thousand words. Being able to visualize the entire measurement trend in scatter plot provides a whole new dimension of insight. And to top it up, by mouse-over the false failure measurement, its corresponding passing measurement is also highlighted (as shown with the yellow arrows in Figure 2). Users get to know where it failed and where it passed subsequently.
In this scenario, the boards are tested with multiple equipment and fixtures. This is typical scenario in mass production stage where high volume production is needed to meet daily volume target.
Imagining the beauty with 2 mouse-clicks, user get to see side-by-side comparison by equipment and fixture (as shown in Figure 3). This totally eliminate the time, effort and error if user is to compile and produce such comparisons. One can easily tell from the side-by-side comparison that the false failure occurred only one specific equipment and fixture (Tester1 : Fix01), indicating likelihood of hardware related issues. Working with equipment and fixture vendors to fix the issue.
Reducing false failure results in higher FPY. Higher FPY results in better production throughout.
I hope you get to know how PMA performs false failure analysis and provide insights to improve FPY.
Do look out for my next blog where I share other PMA features and analysis for FPY improvement as well.