The chart separates two kinds of variation. Common-cause variation is the everyday wobble built into the machine, material and method; it produces points scattering randomly inside the limits. Special-cause variation comes from something specific and changeable: a new reel of board, a loose guide, an untrained relief operator. Signals point to special causes. Random scatter inside the limits points to common causes only.
The correct response differs completely. For a signal, stop, investigate, find the cause, remove it, and record what was found. For random scatter inside the limits, do nothing to the setting. Nudging the machine every time a point lands a little above or below centre is called over-adjusting, and it makes things worse: each correction chases noise and adds a new source of variation on top of the old one.
Control limits are recalculated only when the process itself has genuinely changed and the change has been verified, such as after a rebuilt feed section. Collect a fresh set of 20 to 25 subgroups under the new conditions and compute new centrelines and limits. Do not recalculate just because points fell outside; that would hide the very signals the chart exists to show.
Standard Procedure
- 1Classify what you see: a defined signal, or random scatter inside the limits.
- 2For a signal, stop the line, look for a specific recent change, and correct it.
- 3For scatter inside limits, leave the setting alone and keep plotting.
- 4Log every signal and the cause found so repeat problems become visible.
- 5Recalculate limits only after a verified process change, using fresh baseline data.
Lesson Summary
- ✓Signals call for investigation; scatter inside the limits calls for patience.
- ✓Over-adjusting a stable process adds variation. Recalculate limits only after a real, verified change.