What has happened so far? The capability of the measuring devices has been confirmed and it has been defined what a controllable process is. Now comes the exciting question: Is SPC always the use of quality control charts?
Which processes require the use of quality control charts?
This question is not easy to answer. The counter question is easy to answer, because many processes do not need one. But when do they?
Firstly, you can keep it simple: You have to manage the process with a QCC for the characteristics and their processes if this is required by the customer, the legislator, or if you yourself have the desire to do so.
If you delve into the world of written documentation, you will find a lot of text on the subject of "When to use quality control charts". Basically, however, it is very simple:
"Those processes where significant changes to the process (In terms of variation and location) need to be recognised early to avoid future quality problems as well as the production of scrap, should be managed with a quality control chart".
Great sentence. The author is very proud of it. But the sentence has hidden information:
...changes to the process (in terms of variation and location) recognised early...
So a QCC can tell you when a process changes....
The Q-DAS software looks at "characteristics". Commonly, characteristics are something like lengths, diameters, positions... And then there are the vast number of process parameters. Changing process parameters such as pressure, temperature, time, but also sound are or can be "characteristics". And in many areas, these process parameters should and are also checked with quality control charts. Even if these do not have "specifications", the classic charts ignore the specification limits and only care about the values. And so, by keeping quality control charts of the process parameters, process changes can be recognised even before they occur in the geometric characteristics.
Classic examples here are pressure and temperature in injection moulding. Non-classical examples would be the sound of a gearbox after assembly, the frequency of brake discs in the final inspection when a defined hammer strikes them, and much more.
Sometimes the processes that should be controlled are hidden behind the visible processes.
But what to do if not QCC ?
Especially when new customers want to get started with the topic of "SPC / QCC", the author asks a first counter-question. "How have the individual values been "regulated" so far?".
Unfortunately, the answer is often: "If we are outside the specification, then we do something...". This statement is often accompanied by the question: "All my measured values were within specification, why is my C value not capable?"
In statistic, we take subgroups. Not all components are measured. Based on these "subgroups", we "estimate the population".
This means that a correctly drawn subgroup is important for the process analysis alone. The issue of sample size and correct application therefore affects all characteristics that are to be evaluated with capability.
But for those characteristics whose status does not require SPC, the correct single value control may be sufficient. And yet everything that has been written so far is also valid for characteristics that are not managed with quality control charts. The measurement uncertainty, correctly taken subgroups, and the knowledge of those responsible for their "process".
"Alarms" can be defined for the 5 charact. classes if the measured value exceeds a range of the specification limits. And here, depending on the class, you can also move within the specification,
There are also the aforementioned "warnings" for individual values, which may be based on the determined measurement uncertainty, and where a difference can also be made here in the charact. classes:
Self-defined scrap limits and acceptance limits (often referred to as "production tolerance") can be entered on the masks:
And these are then also alerted:
And if measured values are definitely not plausible, a plausibility limit can be defined:
And to show the user with an alarm that a measured value was completely implausible.
This provides various options for controlling individual values instead of control chart parameters, or simply in preparation for the jump to SPC.