Control Chart Calculator

Sample-based estimates of variability (like the average subgroup range or standard deviation) are biased estimators of the true process sigma, and the amount of bias depends on how many observations are in each subgroup. The A2, D3, D4, B3, and B4 constants this calculator looks up correct for that bias at your specific subgroup size, which is why control limits are calculated from process mean/std dev plus a sample-size-specific constant rather than a flat ±3-sigma band.

For subgroup sizes of 6 or fewer, the D3 constant this calculator looks up is zero, which means LCL = D3 × center line collapses to zero. This isn't a calculation error — it reflects that with small subgroups, a range close to zero isn't statistically unusual, so an R chart at these sample sizes only has a meaningful upper limit to watch for excessive variability.

The zone boundaries divide the distance from the center line to UCL (and LCL) into thirds, supporting Western Electric-style zone rules that catch problems before a point actually breaches the outer limits — for example, flagging two of three consecutive points beyond the 2σ boundary as an early warning of a shift or trend. A single point outside UCL/LCL is the most obvious out-of-control signal, but the zone rules are more sensitive to gradual drift.

It means the number of individual observations within each subgroup — for example, if you measure 5 parts every hour, your sample size is 5, regardless of how many hourly subgroups you've collected in total. Getting this input right matters because the A2/D2/D3/D4/B3/B4 constants are looked up specifically by this per-subgroup count, not by the total number of subgroups in your dataset.