Measure process spread, centering, and stability from sample data today. Review Cp and Cpk instantly. Guide teams toward smarter quality control decisions every cycle.
Enter sample measurements to calculate mean and variation automatically. If the measurement field is empty, the calculator uses the manual mean, standard deviation, and sample size.
Example specification setup: LSL = 49, Target = 50, USL = 51
| Observation | Measurement |
|---|---|
| 1 | 50.10 |
| 2 | 49.90 |
| 3 | 50.30 |
| 4 | 50.00 |
| 5 | 49.80 |
| 6 | 50.20 |
| 7 | 50.10 |
| 8 | 49.70 |
| 9 | 50.00 |
| 10 | 50.20 |
| 11 | 49.90 |
| 12 | 50.10 |
Cp = (USL − LSL) / (6 × σ)
CPU = (USL − Mean) / (3 × σ)
CPL = (Mean − LSL) / (3 × σ)
Cpk = minimum of CPU and CPL
Cpm = (USL − LSL) / [6 × √(σ² + (Mean − Target)²)]
Estimated Yield = 100 − (Total PPM / 10,000)
Cp measures the potential capability if the process is centered. Cpk measures actual capability after considering the process mean location. Cpm adds target alignment, which is useful when project deliverables must meet a preferred performance point, not only stay inside limits.
Cp measures potential capability by comparing specification width with six standard deviations. It assumes the process is centered and only evaluates spread versus tolerance.
Cpk measures actual capability after considering both variation and centering. It drops when the process mean moves closer to either specification limit.
Cp ignores mean location, while Cpk includes it. When the process is off-center, one side has less margin, so Cpk becomes lower than Cp.
Use measurement data when raw observations are available. It reduces manual entry errors and lets the calculator compute the mean, sample size, and standard deviation directly.
A Cpk of 1.33 or higher is commonly treated as capable. A value near 1.00 is marginal, while values below 1.00 indicate higher defect risk.
Cpm accounts for both spread and distance from the target. It is helpful when the ideal result matters, not only staying inside the specification limits.
No. The defect estimates assume the process follows a normal distribution. Real-world results can differ if the data is skewed, unstable, or poorly measured.
It helps project teams quantify delivery consistency, predict defect exposure, justify corrective actions, and support quality planning with measurable process performance indicators.
Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.