Quality control chart: types, control limits and how to read one

A quality control chart, or control chart, is a time-ordered graph of a process measurement with a center line and an upper and lower control limit calculated from the process data. Points inside the limits show normal variation. A point outside the limits, or an unusual pattern, signals a special cause that the team should find and remove.

Control charts are the core tool of statistical process control, or SPC, and one of the seven basic quality tools. This guide explains the types of control charts, how to calculate control limits, how to read the chart, how it differs from a pre-control chart, and how manufacturing plants use it to keep processes stable.

What is a quality control chart?

A control chart plots a measured value, such as a shaft diameter or a fill weight, or a count, such as defects per batch, in the order the parts were made. It has three lines. The center line is the process average. The upper control limit and the lower control limit mark how far the process normally varies around that average.

Walter A. Shewhart developed the control chart at Bell Telephone Laboratories in the 1920s. His key idea was that every process has two kinds of variation. Common cause variation is the normal noise of a stable process: small differences in material, machine and environment. Special cause variation comes from something specific that changed, for example a worn tool, a new material batch or a wrong setting. The control chart separates the two, so the team reacts only to real signals. The international standard series for control charts is ISO 7870.

Which type of control chart should you use?

The right chart depends on the kind of data you collect. Variable data is measured on a scale, such as length, weight or torque. Attribute data is counted, such as the number of defective parts or the number of defects on one unit.

Variable charts are more sensitive than attribute charts and need fewer samples to detect a change. Use them for critical characteristics. Attribute charts are simpler and work well for visual inspection and final audits.

How do you make a control chart?

You can build a control chart in Excel, in statistical software such as Minitab, or in the SPC module of your MES. The steps are the same:

1. Choose the characteristic
   Pick a characteristic that matters to the customer or to the next process, for example a critical dimension from the control plan.
2. Check the measurement system
   Make sure the gauge is calibrated and the measurement system analysis is acceptable. A poor gauge adds noise to the chart.
3. Define rational subgroups
   Take parts made under the same conditions, for example 5 consecutive parts from one machine. Variation inside a subgroup should be only common cause.
4. Collect the base data
   Collect 20–25 subgroups from a process that runs normally, before you calculate any limits.
5. Calculate the center line and limits
   Calculate the average and the control limits with the formulas for your chart type.
6. Plot and check the base period
   Plot the points. If any point is out of control, find the cause, remove it and recalculate the limits.
7. Monitor and react
   Keep plotting new subgroups against the fixed limits and follow the reaction plan when the chart gives a signal.

How do you calculate control limits?

For most charts the limits are the average plus and minus three standard deviations of the plotted statistic. In practice the standard deviation is estimated from the subgroup ranges and a constant that depends on the subgroup size. For the X-bar and R chart the formulas are:

• X-bar chart: center line = grand average X̿. Upper control limit = X̿ + A2 × R̄. Lower control limit = X̿ − A2 × R̄.
• R chart: center line = average range R̄. Upper control limit = D4 × R̄. Lower control limit = D3 × R̄.
• For a subgroup of 5 parts, the constants are A2 = 0.577, D3 = 0 and D4 = 2.114.

Here is an illustrative example. A machining cell checks a 25 mm diameter in subgroups of 5 parts. After 25 subgroups the grand average is 25.000 mm and the average range is 0.100 mm.

If the drawing allows 25.00 ± 0.15 mm, the control limits sit well inside the specification. That is what a capable process looks like. The numbers are illustrative.

How do you read a control chart?

A process is in statistical control when the points vary randomly around the center line and stay inside the limits. The chart signals a special cause when you see one of these patterns. They are based on the Western Electric and Nelson rules:

• One point outside the upper or lower control limit.
• 7 or more points in a row on the same side of the center line.
• 6 or more points in a row steadily rising or falling.
• 2 out of 3 points in a row close to the same control limit.
• 14 points in a row alternating up and down.
• A repeating cycle or points that hug the center line far too closely.

Agree on which rules your plant uses and write them into the work instruction. Using too many rules at once raises the number of false alarms. Many plants start with the first three.

What should you do when a process is out of control?

A signal on the chart is a request to act, not only to record. A good reaction plan says who stops the process, what happens to the parts made since the last good subgroup, and who looks for the cause. In automotive plants this reaction plan is part of the control plan. See our article on the PFMEA and control plan.

1. Stop or contain: hold the suspect parts and check them against the specification.
2. Find the special cause: what changed in material, machine, method, operator or measurement?
3. Correct it and record the cause and the action directly on the chart.
4. Confirm with the next subgroups that the process is back in control.
5. If the same signal returns, start a root cause analysis with the team.

For the root cause step, see structured problem solving. When many different defects compete for attention, rank them first with a Pareto chart in quality control, then put a control chart on the characteristic that matters most.

What is a pre-control chart and how does it differ from a control chart?

A pre-control chart is a simpler tool that some plants use at the machine. It does not calculate limits from process data. It splits the tolerance into zones. The green zone is the middle half of the tolerance around the target. The yellow zones are the outer quarters of the tolerance. Everything outside the specification limits is red.

To qualify the process, 5 consecutive parts must fall in the green zone. After that, the operator checks 2 consecutive parts at regular intervals:

• Both parts green, or one green and one yellow: continue production.
• Both parts yellow: stop and adjust the process.
• Any part red: stop and adjust the process.
• To restart, 5 parts in a row must again fall in the green zone.

The sampling frequency comes from the stops. The usual rule is to divide the average interval between two stops by six.

The two tools rest on different ideas. Pre-control treats every part inside the specification as good, and a part at the edge of the tolerance is as good as one on target. Control charts follow the Taguchi loss idea. Every deviation from the target creates some loss, because the next steps in the value stream have to absorb that variation. A part made on target is therefore worth more than a part made at the specification limit.

Pre-control is most useful right after a changeover. It helps center the process until there is enough data to calculate control limits and confirm capability. Change the rule “stop and adjust” to “stop and investigate”, so that operators do not over-adjust a process that is only showing normal variation. Used this way, as part of the changeover procedure, pre-control protects the customer until the process is stable, in control and capable.

How do manufacturing plants use control charts?

In automotive and industrial plants, SPC is a standard customer requirement. IATF 16949 and the core tools expect control charts on special characteristics and capability studies before series production. Typical uses are:

• Critical dimensions on machining, molding and stamping lines, with X-bar and R or I-MR charts at the machine.
• Process parameters such as temperature, pressure or torque, often charted automatically from sensors.
• Defect rates at final inspection or customer audits, with p or u charts.
• Launches and changes: proving that a new or changed process is stable before the customer approves it.
• Improvement projects: the Control phase of Six Sigma DMAIC uses control charts to hold the gain.

Many plants now collect the data automatically and see the charts live in the MES or SPC software. That removes manual work, but it does not replace the reaction. The value comes from people who look at the signals and act on them every shift. This is one of the ideas behind the principles of TQM: decisions based on facts, made by the people closest to the process.

How GQ Interim helps

Plotting a control chart is easy. Getting a plant to run SPC that catches problems before the customer does is harder. It takes clean measurement systems, sensible subgroups, clear reaction plans and operators who trust the charts. When your quality team is busy with complaints and audits, an interim manager can set this up during the assignment and hand over a working system.

Since 2011, GQ Interim has completed 1,550+ projects in manufacturing and industrial companies across Central and Eastern Europe, for clients such as Stellantis, Porsche, Siemens, Valeo and Magna. Our database holds 7,500+ verified experts, including interim quality managers who have introduced SPC, passed customer audits and stabilized launches in automotive and industrial plants. We are certified to ISO 9001 and TISAX.

Conclusion

A quality control chart tells you when a process has changed and when it has not. Choose the chart that fits your data, calculate the limits from a stable base period, and agree on clear rules and a reaction plan. Use pre-control only where it helps, for example after changeovers. When operators act on every signal, the chart stops being paperwork and starts preventing defects.

Frequently asked questions