Do You Know Why SPC Alone Just Isn't Enough?
By Joe Ventimiglio, Sciemetric
From Sciemetric's Signature Newsletter. Subscribe now.
Manufacturers are always looking for ways to improve processes to get better yield. The thing is, very few have the right information and/or tools to do it. All too often I hear “but we already have SPC” (Statistical Process Control) when I try to explain the benefits of QualityWorX, our quality operating system, to someone who is experiencing quality issues on the production line. The perception is that SPC is the right tool to solve the problem but the reality is that it doesn’t give you a complete picture: it tells you there’s an issue but not what the actual problem is or how to fix it. That’s where process signature technology comes in.
Some Background on SPC
SPC has been around since the 1920’s, and is a widely accepted methodology for identifying issues in manufacturing processes. It’s based on point-data analysis and uses statistical tools to monitor trends in production parameters to spot deviations that may eventually result in rejects. This helps production engineers to identify potential problems early, even before yields are affected. However, while the various statistical charts and analyses can point them in the right general direction, it does not provide much insight into what is actually causing the trend in the first place. In short, SPC is great for identifying that a production problem exists, but falls short when it comes to actually fixing it.
So, how do you build on what you learn from SPC?
What I tell my customers is to combine SPC with Process Signature Technology. SPC will help them monitor and track the health of a production line, while signature analysis will enable them to identify the “how” and “why” of yield decreases, and allow them to fix the source of the problem. This same type of visibility and insight can also be u sed to optimize tests or processes to ensure the best possible product quality, at the lowest manufacturing costs.
Here’s how it works. While SPC deals exclusively with point data such as “max force” or “max leak rate”, signature analysis looks at the entire waveform generated from the entire process. The primary advantage of a waveform over a single, discrete data point is that it provides a complete picture of the behaviour of the manufacturing process, not just a single characteristic. Even when multiple feature points are taken from a single waveform, it only provides information for a single point in time. For example, if the maximum value recorded is in a force vs. distance measurement, you will only know that single value. You do not know where the problem occurred, and if the value was reached gradually or was a spike in the process. Often how a result was arrived at is just as important as achieving it. This is why the entire process, and not just a single point, is monitored.
The histogram in Figure 1 shows the flow rate measurement for a leak test at the end of the cycle. A determination is made of “good” or “bad” part based on this value being between the specified limits. What the histogram does not show are consistent patterns in the failures . It indicates that there are some clear failures but provides little guidance as to whether the failures are the result of test malfunctions or actual defective parts.
Now let’s take a look at the process signatures from these same parts, shown in Figure 2 below. We can immediately see there are two different failure modes, labelled “A” and “B” in the figure. “A” has a constant flow rate at approximately 180 ccm that indicates a failure in the test setup, not a part failure. “B” failures have a flow rate that does not decrease as far as the “good” parts. This is likely a part-specific failure.
Another interesting point is observed: some parts that passed the test exhibit a slow steady decline in flow (these are labeled “C” in Figure 2). This is not consistent with the majority of conforming parts, which display a very rapid initial decline, followed by a slow stabilization of the flow rate. This may indicate an undetected test problem or maybe a different part behaviour that needs to be investigated further.
Closer examination of the waveforms in Figure 2 reveals that the analysis could have been done 2 seconds earlier in the test with no change in the accuracy or repeatability of the test. This simple change would represent a 10% reduction in cycle time.
Minimizing cycle time increases throughput of the test station, which can reduce the amount of equipment required. This is especially true for the end of line functional tests that are typically a bottleneck in the manufacturing line. Thoroughly examining the process signatures allows you to determine if a test has been optimized or contains unnecessary delays.
The more data you have, the better the decisions you can make. The extra level of detail that waveforms provide is critical to understanding how your part is behaving during the assembly process. If you’re only using SPC, you’re not getting the full picture – which means you’re not getting the most out of your production line.
For more information contact Joe Ventimiglio, Director of sales, automotive/industrial at Sciemetric. He can be reached at firstname.lastname@example.org