Glossary · Statistical Process Control

SPC

Statistical Process Control (SPC) watches the variation in measured values statistically so that a drifting process is caught before it produces defects.

How it connects

  1. Measurements

    Collect process data

  2. Control chart

    Variation over time

  3. Signal

    Limits and patterns

  4. Investigation

    Review conditions and improve

Definition

Definition

Statistical Process Control (SPC) watches the variation in measured values statistically so that a drifting process is caught before it produces defects.

In more detail

Every process varies. SPC separates that variation into common cause — the normal state of the process — and special cause, which is a signal to intervene.

In practice this is done with control charts. A point outside the control limits, or a run of points biased to one side or trending even while inside the limits, is treated as special cause.

The most common misunderstanding is the difference between control limits and specification limits. Specifications come from the customer; control limits come from what the process actually produces. They are different numbers, and drawing specification lines on a control chart defeats the purpose.

How it is used on the floor

  • Measure critical dimensions or process conditions on a cycle and monitor them on control charts.
  • Catch process drift and act before parts fall outside specification.
  • Compute process capability (Cp, Cpk) to see the margin against specification.

Control limits are not specification limits

Specification limits come from drawings and customer requirements; control limits are calculated from process data. A part inside specification but outside the control limits means the process has changed — and left alone it will eventually breach specification too.

Related terms

Terms that make more sense read together.

Related products

What IYULAB provides in this area.

Quick Answers

SPC — frequently asked

Questions that come up when evaluating SPC.

Can we use SPC with little measurement data?

Calculating control limits needs a reasonable sample. Early on, collect data and watch the trend, then derive limits once the process settles.

Should SPC cover every characteristic?

No. Concentrating on the few characteristics that decide quality is more effective. Too many characteristics inflate measurement effort and bury the signal.

Do we need automated gauging?

Manual measurement works, provided the interval and method stay consistent and something guards against entry error.

How would this apply to your plant?

A look at your equipment and current workflow is usually enough to scope it.

Talk to an engineer
KR KO