SC QUALITY GUILDControl chart constants
SCQUALITYGUILD.COM/CALCULATORS/CONTROL-CHART-CONSTANTS.HTML
d₂, d₃, A₂, D₃ and D₄ derived in your browser from the distribution of the range. Nothing transcribed.
DERIVED ON PAGE LOAD FROM THE DISTRIBUTION OF THE RANGE · FREE · NOTHING YOU TYPE LEAVES YOUR BROWSER
Q-BotClick for an explanation of this calculator.Control chart constants
DERIVED ON PAGE LOAD FROM THE DISTRIBUTION OF THE RANGEThese are not transcribed. When this page loaded, your browser integrated the distribution of the range of n standard normal variables to get d₂ and d₃, then formed the chart factors from them. Check any row against your own table — agreement to four decimals is the point.
| n | d₂ | d₃ | A₂ | D₃ | D₄ | E₂ = 3/d₂ |
|---|
d₂ = E[W] AND d₃ = SD[W], WHERE W IS THE RANGE OF n STANDARD NORMAL VARIABLES. A₂ = 3 / (d₂√n). D₃ = max(0, 1 − 3d₃/d₂). D₄ = 1 + 3d₃/d₂.
For educational use. This page is a study aid. The method behind it has far more depth and detail than one page and one tool can cover, and what you see here is not everything the method offers or everything this site offers. For full understanding and application, use the governing standards and texts, and seek further education, training or qualified advice. Results depend on the data you enter, so check them before relying on them.
Where these constants come from
Take n values from a normal distribution with standard deviation one, and record the
range — the largest minus the smallest. Do it forever. That range has a distribution of its own.
- d₂ is the mean of that distribution. It is the conversion factor between an
average range and a standard deviation:
σ̂ = R̄ / d₂. - d₃ is its standard deviation, which is what the range chart's own limits are built from.
- A₂, D₃ and D₄ are shortcuts assembled from those two so
the limits can be worked out on a shop floor without a calculator — which is exactly why they
exist.
A₂ = 3 / (d₂√n).
None of them are arbitrary, and none of them were measured. They are properties of the normal distribution, and they can be derived rather than looked up.
Derived here, against the published tables
These were computed from the range distribution and set beside the values printed in the standard tables:
| n | d₂ derived | d₂ published | A₂ derived | A₂ published |
|---|---|---|---|---|
| 2 | 1.1284 | 1.128 | 1.880 | 1.880 |
| 3 | 1.6926 | 1.693 | 1.023 | 1.023 |
| 4 | 2.0588 | 2.059 | 0.729 | 0.729 |
| 5 | 2.3259 | 2.326 | 0.577 | 0.577 |
| 6 | 2.5344 | 2.534 | 0.483 | 0.483 |
| 7 | 2.7044 | 2.704 | 0.419 | 0.419 |
| 10 | 3.0775 | 3.078 | 0.308 | 0.308 |
They agree to every digit the tables carry. That is the point of deriving them: a transcribed table can contain a typo and you would never know. A derived one cannot.
What the constants assume
Every one of these numbers assumes the underlying data is normal. The range distribution
they come from is the range distribution of normal variates. Apply R̄/d₂
to a strongly skewed characteristic and the sigma estimate is wrong before the chart is drawn.
Where they get used incorrectly
- The subgroup size changed and the constant did not. If the operator started taking four instead of five, every limit on the chart is now wrong. This is the most common error and the hardest to spot after the fact, because the chart still looks like a chart.
- Range used beyond about n = 10. The range throws away everything between the two extremes, and past roughly ten that waste becomes serious. Use s and the c₄ family instead.
- D₃ treated as a typo when it is zero. For n of six or less the lower range limit genuinely is zero — a subgroup of identical values is not a signal. It is not missing.
- Limits recalculated every period. Control limits describe what the process has been doing. Recomputing them every month guarantees the process is always in control and the chart never tells you anything.
Common questions
What is d2 in a control chart?
d2 is the expected range of a subgroup of n values from a normal distribution, in units of the standard deviation. It converts an average range into a sigma estimate: sigma-hat = R-bar / d2. For a subgroup of 5, d2 is 2.326.
How is A2 calculated?
A2 = 3 / (d2 × √n). The X-bar chart limits are the grand average plus and minus A2 × R-bar, so A2 folds the three-sigma rule, the d2 conversion and the √n for averages into one number. For n = 5, A2 is about 0.577.
What are D3 and D4?
They set the range chart limits: LCL = D3 × R-bar and UCL = D4 × R-bar, where D3 = 1 − 3·d3/d2 (floored at zero) and D4 = 1 + 3·d3/d2. For subgroups of 6 or fewer, D3 is zero, so the range chart has no lower limit. For n = 5, D4 is about 2.114.
Why estimate sigma from R-bar / d2 instead of the overall standard deviation?
The within-subgroup estimate captures only short-term, common-cause variation. The overall standard deviation also includes any shifts between subgroups, which would widen the limits and hide exactly the signals the chart is meant to catch.