Pre-Control chart: Know when to adjust. And when not to.
Pre-Control is a simple process control method: the tolerance is cut into green, yellow and red zones, and two consecutive parts tell the operator whether to keep running, adjust or stop. Developed at Rath & Strong in 1954 and championed by Dorian Shainin, it needs no statistics on the shop floor — just a gauge, the rules, and the ability to adjust the process.
What is a pre-control chart?
A pre-control chart draws the tolerance of a characteristic as three colored zones and plots each reading against them. The zones come from the specification, not from the process, so the only calculation is dividing the tolerance into quarters.
Green zone: keep running
The middle half of the tolerance, centered on the target. A capable process puts almost all of its parts here.
Yellow zones: a warning
The outer quarters of the tolerance, one on each side. Parts here are still in specification, but the process is approaching its limit.
Red zones: stop
Anything outside the specification. A single red reading stops the process: it is making bad parts, or the set-up is wrong.
One-sided tolerances get their own layout — the quarters above assume a target in the middle of the tolerance. How? Talk to an expert
The pre-control rules
Pre-Control runs in two modes. A set-up is first qualified with five consecutive green parts; then production is checked with pairs of parts at a fixed interval. Every pattern has one prescribed action, so two operators reading the same parts make the same decision.
Set-up rules: qualify before you run
- Five greens in a rowStart production and switch to the running rules.
- Two yellows on the same sideAdjust the process toward the target, then start the count again.
- Two yellows on opposite sidesStop and fix: the spread is too wide for the tolerance. Adjusting the setting will not help.
- Any redStop and fix the process before it makes another part.
Running rules: two parts per check
- First part greenContinue. The second part is not even needed.
- First yellow, second greenContinue.
- Two yellows on the same sideAdjust the process, then requalify with five greens.
- Two yellows on opposite sidesStop and fix: the process has become too variable.
- Any redStop and fix.
After any adjustment or fix, the set-up rules apply again: five greens in a row before production continues.
How often to sample
Two parts at a fixed interval. A common rule of thumb sets the interval so that about six pairs are checked between adjustments: often enough to catch a drift before it reaches a limit, rarely enough that measuring does not get in the way of production. Start with a conservative interval and lengthen it as the chart stays green.
Why Pre-Control only works on a capable process
The zones are fixed by the tolerance, so how often Pre-Control speaks depends on how much of the tolerance the process uses. On a process that runs on target with a capability of Cp = 1.0, most readings are green, yet about 2% of pairs signal Adjust or Stop although nothing has changed. At Cp = 1.33 those false signals almost disappear, while a real drift is still caught within a few pairs.
87%of readings fall in the green zone when a process with Cp = 1.0 runs on target
0.2%of pairs give a false Adjust or Stop signal at Cp = 1.33: about two per thousand
4pairs, on average, until a drift of a quarter of the tolerance is caught at Cp = 1.33
The further the process sits from the zone boundaries, the quieter Pre-Control stays, and the sooner a real shift pushes readings into yellow. That is why it suits capable processes that drift, and why on a marginal process it causes tampering: adjustments triggered by noise, which add variation instead of removing it.
Calculated for a normal distribution; Cp = (USL − LSL) / 6σ. Pre-Control gives no estimate of Cp or Cpk: establish capability first.
Set the capability and the drift, and watch what a pair of parts says.
Share of readings
- Green
- Yellow
- Red
What a pair of parts says
Continue
Adjust
Stop
Where Pre-Control fits, and where it doesn’t
Pre-Control does not find causes and does not measure capability. It starts with a process that is already capable and merely drifts, and it tells the operator when to steer. In Shainin’s Red X® problem solving it comes after the cause has been found: once the dominant cause of variation is known and its tolerance set, a simple control like Pre-Control keeps the gain on the shop floor.
Check the gauge first
The measurement must tell the zones apart. If repeat readings of the same part land in different zones, the chart reads the gauge, not the process.
Cut the zones from the tolerance
Green is the middle half, yellows are the outer quarters, red is outside the limits. A one-sided tolerance needs a different cut.
Qualify, then run
Five greens in a row qualify a set-up; then two parts per interval. Write the action for each pattern on the chart, so nobody has to remember it.
Listen to the adjustments
If the chart keeps asking for adjustment, the process is not capable enough for its tolerance: reduce the variation, find the Red X®, instead of steering more often.
Pre-Control holds the gain; it does not create it. The variation reduction comes first, from a root cause analysis that finds and proves the dominant cause.
Pre-Control vs. SPC control charts
Pre-Control and Shewhart control charts answer different questions. A control chart asks whether the process is stable and uses limits computed from process data; Pre-Control asks whether the parts are safely inside the tolerance and uses limits taken from the specification. Each has its place.
| Method | Limits come from | Sample | Best for |
|---|---|---|---|
| Pre-Control | The tolerance: zones at ¼, ½ and ¼ | Five parts to qualify, then two parts per check | Capable processes that drift, short runs, operator-run control with minimal training |
| X̄ and R chart | Process data: ±3σ of subgroup averages | Subgroups of 3–5 parts; 20–25 subgroups to set the limits | Detecting loss of stability, studying a process, long runs |
| Individuals chart (I-MR) | Process data: the moving range | One part per check | Slow processes and expensive measurements |
| Go/no-go gauging | The tolerance | Every part, or a sample | Sorting good parts from bad; no warning before the first defect |
Dorian and Peter Shainin argued in 1989 that Pre-Control delivers immediate improvement where control charts deliver slow learning; critics such as Ledolter and Swersey (1997) answered that it is no substitute for control charts and causes tampering on weakly capable processes. Both sides agree on one point: it belongs on capable processes.
Worked example
Setting up a pre-control chart in five minutes
A turned shaft has a diameter tolerance of 20.00 ± 0.10 mm. The tolerance is 0.20 mm, so the zones are cut at quarters of 0.05 mm: green from 19.95 to 20.05, yellow from 19.90 to 19.95 and from 20.05 to 20.10, red outside 19.90–20.10. The operator qualifies each set-up with five greens, then measures two shafts every 30 minutes. Readings of 20.07 and 20.08, both in the upper yellow zone, mean: adjust toward 20.00 and requalify. Readings of 19.93 and 20.07, opposite yellows, mean: stop, the spread is too wide for the tolerance.
0.05mm wide: each quarter of the 0.20 mm tolerance
19.95–20.05mm: the green zone, centered on the 20.00 target
Pre-control chart FAQ
What is a pre-control chart?
A pre-control chart, also called a stoplight chart, is a process control chart whose zones are cut from the tolerance: green for the middle half, yellow for the outer quarters, red outside the specification. Operators read two consecutive parts against the zones to decide whether to keep running, adjust or stop.
How do you set pre-control limits?
Divide the tolerance into quarters. The green zone is the middle two quarters, 50% of the tolerance centered on the target; each yellow zone is one outer quarter; the red zones lie outside the specification limits. No process data are needed: the limits come from the drawing.
What are the pre-control rules?
Set-up: five consecutive parts in the green zone qualify the set-up; two yellows on the same side mean adjust, two on opposite sides or any red mean stop and fix. Running: measure two parts at each interval. A green first part, or a yellow followed by a green, means continue; two yellows on the same side mean adjust; opposite yellows or any red mean stop. After any adjustment, requalify with five greens.
Who invented pre-control?
Frank Satterthwaite, a statistician at the consulting firm Rath & Strong, developed it in 1954. His colleague Dorian Shainin carried it onto the shop floor and promoted it for decades; in 1989 Dorian and Peter Shainin argued in Quality Engineering that it beats X̄ and R charts for immediate quality improvement.
Pre-Control vs. SPC: what is the difference?
A Shewhart control chart compares the process with itself: its limits are computed from process data, and it detects a loss of statistical control. Pre-Control compares the parts with the tolerance: its limits come from the specification, and it tells the operator when to adjust. Control charts suit process studies and long runs; Pre-Control suits capable processes, short runs and operators without statistical training.
When should you not use pre-control?
When the process is not capable: in practice, below a Cp of about 1.3 the yellow zones fill with normal variation and the rules trigger adjustments that only add variation, which is called tampering. Also when you need to learn about the process rather than steer it, when the output is not a measured value, and before the capability has been established, because Pre-Control gives no estimate of Cp or Cpk.
How often do you take samples?
Two consecutive parts at a fixed interval. A common rule of thumb: choose the interval so that about six pairs are checked between process adjustments, and lengthen it as the chart stays green. If the chart keeps calling for adjustment, the problem is the variation, not the interval.
Does pre-control work with one-sided tolerances?
Yes, with a different layout. The two-sided quarters assume a target in the middle of the tolerance, which a one-sided characteristic such as flatness or runout does not have. Cutting the zones for a one-sided tolerance is one of the things a Shainin expert sets up with you: talk to an expert.
Is pre-control statistically valid?
Its zones assume a roughly normal process centered on the target. For such a process with Cp = 1.0, about 87% of readings are green and about 2% of pairs give a false signal; at Cp = 1.33 the false signals fall to about two per thousand pairs. Academic reviews, Ledolter and Swersey (1997) and Steiner (1997), conclude that it works for capable processes and small lots and is no substitute for control charts elsewhere.
How does Pre-Control fit into Shainin’s Red X® problem solving?
As the last step, not the first. Red X® problem solving finds and proves the dominant cause of variation and sets its tolerance; a simple control such as Pre-Control then keeps the controlled input and the Green Y® output in the green zone on the shop floor. If a process still needs frequent adjustment, the dominant cause has not been fully controlled.
Sources: F. E. Satterthwaite, “A simple, effective process control method”, Report 54-1, Rath & Strong, 1954; D. Shainin and P. Shainin, “Pre-Control versus X̄ & R charting: continuous or immediate quality improvement?”, Quality Engineering 1(4), 1989; J. Ledolter and A. Swersey, “An evaluation of pre-control”, Journal of Quality Technology 29(2), 1997; S. H. Steiner, “Pre-control and some simple alternatives”, Quality Engineering 10(1), 1997; Quality magazine, “Dorian Shainin’s influence on quality professionals, part IV”, 2018.
Variation you can’t keep in the green?
Shainin’s Red X® experts help manufacturers find the dominant cause of variation — and keep the gain on the shop floor.