What SPC actually is
SPC (Statistical Process Control) is the use of statistical methods — principally control charts and capability indices — to monitor a process, distinguish normal random variation from a real change, and keep the process centred and stable over time. It is one of the five automotive core tools, and its guiding idea is prevention: catch a drift as a trend on a chart before it becomes a rejected lot, rather than sorting bad parts at final inspection.
SPC reads the same specification limits the control plan uses — the nominal, upper specification limit (USL) and lower specification limit (LSL) — and compares them to what the process is actually doing. That is only meaningful if the readings come from a measurement system that passed Gauge R&R; SPC on untrustworthy data charts the gauge, not the process.
Control charts and their types
A control chart plots a process statistic over time against a centre line and upper and lower control limits set at ±3 standard deviations. The right chart depends on the data type:
| Data type | Chart | Use |
|---|---|---|
| Variables (measured) | X-bar & R | Subgroup average and range — the workhorse for dimensions |
| Variables (measured) | X-bar & S | Average and standard deviation for larger subgroups |
| Variables (measured) | I-MR (individuals) | Single readings where subgrouping is impractical |
| Attributes (counted) | p / np | Proportion or number of defective units |
| Attributes (counted) | c / u | Count or rate of defects per unit |
Control limits vs specification limits
The single most misunderstood point in SPC is that control limits are not specification limits. Specification limits (USL/LSL) come from the customer and the drawing — they define what is acceptable. Control limits are calculated from the process’s own data at ±3σ — they describe what the process naturally does. A process can be in perfect statistical control (all points within control limits) and still produce out-of-spec parts if it is not capable, and vice versa. Control limits tell you if the process is stable; capability indices tell you if it is good enough.
Common cause vs special cause
SPC classifies variation into two kinds. Common-cause variation is the natural, random noise inherent in a stable process — you do not chase individual points. Special-cause variation is an assignable event — a tool change, a new batch of material, a machine fault — that shows up as a point outside the control limits or as a non-random pattern. Detection rules such as the Western Electric or Nelson rules flag these patterns: a point beyond 3σ, several consecutive points on one side of the centre line, a run of steadily increasing points, and so on. The discipline is to react to special causes and leave common cause alone — over-adjusting a stable process (tampering) actually increases variation.
Cp, Cpk, Pp and Ppk
Capability indices compare the process spread to the tolerance. There are four, and the distinction between them matters:
| Index | Formula | What it tells you |
|---|---|---|
| Cp | (USL − LSL) / 6σwithin | Potential capability — spread vs tolerance, ignoring centring |
| Cpk | min[(USL − x̄)/3σwithin, (x̄ − LSL)/3σwithin] | Actual capability — spread and centring |
| Pp | (USL − LSL) / 6σoverall | Long-term potential performance |
| Ppk | min[(USL − x̄)/3σoverall, (x̄ − LSL)/3σoverall] | Long-term actual performance |
The key differences: Cp/Cpk use the within-subgroup (short-term) standard deviation and describe the process’s potential when only common cause is present; Pp/Ppk use the overall (long-term) standard deviation and describe actual sustained performance including drift between subgroups. And Cp/Pp ignore centring while Cpk/Ppk penalise a process that is off-centre — which is why Cpk is always less than or equal to Cp. A high Cp with a low Cpk is the signature of a tight but mis-centred process.
Capability targets
Automotive customers typically require Cpk ≥ 1.33 for an ongoing, stable process, and 1.67 or higher for new processes or safety-related special characteristics. Initial process studies at PPAP are usually judged on Ppk ≥ 1.67. A Cpk of 1.33 corresponds to roughly 63 parts-per-million out of spec on a centred process; 1.67 is tighter still. These are the numbers a PPAP initial process study must hit.
How Fast Quality Software runs SPC
Fast Quality Software drives SPC off the same specification limits the control plan uses, evaluating inspection readings against nominal, USL and LSL and rendering trend and capability views.
Because inspection, the control plan and SPC share one specification master, the Cp/Cpk you report come from the same limits the control plan controls and the same readings inspection dispositions — and a characteristic drifting toward its limit is visible as a trend before it becomes an NCR. See it applied for casting and plastics process parameters and precision machining, or review pricing.
Frequently asked questions
What is the difference between Cp and Cpk?
Cp measures potential capability — the process spread against the tolerance width — but ignores where the process is centred. Cpk measures actual capability by accounting for both spread and centring, taking the worse of the distances from the mean to each specification limit. Because of this, Cpk is always less than or equal to Cp; a high Cp with a low Cpk indicates a tight but off-centre process.
What is the difference between Cpk and Ppk?
Cpk and Ppk use different measures of variation. Cpk uses the within-subgroup (short-term) standard deviation and describes the process's potential when only common-cause variation is present. Ppk uses the overall (long-term) standard deviation and describes actual sustained performance, including drift between subgroups. PPAP initial process studies are usually judged on Ppk, while ongoing production is monitored with Cpk.
What is the difference between control limits and specification limits?
Specification limits (USL and LSL) come from the customer and the drawing and define what is acceptable. Control limits are calculated from the process's own data at plus or minus three standard deviations and describe what the process naturally does. Control limits tell you whether the process is stable; capability indices such as Cpk tell you whether it is good enough against the specification. The two are independent.
What is a good Cpk value?
Automotive customers typically require a Cpk of at least 1.33 for an ongoing, stable process and 1.67 or higher for new processes or safety-related special characteristics. Initial process studies at PPAP are usually judged on a Ppk of at least 1.67. A Cpk of 1.33 corresponds to roughly 63 parts per million outside specification on a centred process.
What is the difference between common cause and special cause variation?
Common-cause variation is the natural, random noise inherent in a stable process; you leave it alone rather than reacting to individual points. Special-cause variation is an assignable event such as a tool change, a material batch or a machine fault, which appears as a point outside the control limits or a non-random pattern. SPC discipline is to investigate and remove special causes while not tampering with a process that shows only common cause.