Control Chart
What does a point outside the control limits mean?
It indicates the process may have a special cause and should be investigated, because the observed change is unlikely to be just routine random variation.
What is the difference between an X-bar chart and an R chart?
The X-bar chart tracks the subgroup mean to detect centering drift, while the R chart tracks within-subgroup spread to detect changes in consistency or dispersion.
Can a process be 'in control' and still make bad parts?
Yes. Statistical control means stable variation, not guaranteed compliance with engineering tolerances or customer specifications.
A control chart is a Statistical Process Control (SPC) graph that plots a process measure in time order against a centerline and calculated upper/lower control limits to determine whether the process is stable or affected by special-cause variation. It separates common-cause variation (normal process noise) from special-cause variation (assignable shifts like tool wear, vibration, or coolant issues).
In a CNC cell, a control chart is built around one critical feature such as bore diameter or milling height. The operator measures a subgroup at regular intervals, plots the average and range, and watches for signals like points outside limits or non-random trends. This allows drift to be detected before parts go out of spec. In millwork and edgebanding, the same logic applies to panel thickness or edge band overhang, catching drift from cutter wear or setup changes. Crucially, control limits come from process data and judge stability, not conformance to specification limits. A process can be statistically stable yet still produce parts near or outside tolerance if the average is poorly centered. Conversely, a process can momentarily violate a control rule before a part becomes nonconforming. Control charts prevent over-adjustment of stable processes, which would otherwise increase variation.
Tolerance vs Control Limits: Reacting to parts near specification limits as if they were control chart alarms drives unnecessary offsets and amplifies variation rather than correcting it.
Improper Sampling: Inconsistent subgroup selection or timing misreads process behavior, so the chart fails to reflect actual drift and signals become unreliable.
Ignoring Root Cause: A point outside limits is only a symptom; without checking tool wear, vibration, or fixture looseness, the chart becomes a logging exercise instead of control.