Cpk vs Cmk: what is the difference?
Short answer
Cmk (machine capability) judges the machine alone in a short, controlled run of usually 50 consecutive parts. Cpk (process capability) judges the full process, with material, operators, tool wear and shifts, over a longer period. Because the machine study excludes most sources of variation, its requirement is stricter: typically Cmk ≥ 1.67 versus Cpk ≥ 1.33.
Side by side
| Aspect | Cmk (machine capability) | Cpk (process capability) |
|---|---|---|
| What it judges | The machine or equipment alone | The whole process: man, machine, material, method, environment |
| Data | About 50 consecutive parts, one setup, one operator, one batch | 125 or more values in rational subgroups (e.g. 25 × 5) over days or weeks |
| Variation included | Short-term, machine only | Within-subgroup (Cpk) and total (Ppk) process variation |
| Typical requirement | Cmk ≥ 1.67 (often Cm ≥ 2.0) | Cpk ≥ 1.33; Ppk ≥ 1.67 for initial process studies |
| When | Machine acceptance (buy-off), after relocation or major repair | PPAP, serial production, ongoing monitoring with SPC |
| Stability check | Usually not required | Required: control chart must show a stable process |
Why Cmk must be higher than Cpk
A machine study removes most sources of variation on purpose: one material batch, one operator, a warm machine, no tool change. Whatever spread remains is only the machine. In real production every other factor adds variation, so the process spread will always be wider than the machine spread. If the machine already uses most of the tolerance, there is no room left for the rest. That is why customers ask for Cmk 1.67 or even 2.0 while accepting Cpk 1.33 for the process.
Same formula, different data
Both indices use the same idea: the distance from the mean to the nearer specification limit, divided by three standard deviations. The difference is in the data and in the standard deviation used. Cmk uses the overall standard deviation of a short consecutive run; Cpk uses the within-subgroup estimate (R̄/d₂ or s̄/c₄) from a stable process, and Ppk uses the overall standard deviation of the long-term data.
Practical sequence
Run a Type 1 gauge study first to make sure the measurement system is fit. Then do the machine capability study (Cmk) at the machine supplier or at installation. After the line is running, collect subgrouped data for the preliminary process capability (Ppk) and finally monitor the long-term Cpk with control charts.