A low Cpk has two possible root causes, and the Cp versus Cpk comparison tells you which one you have. If Cp is healthy but Cpk is low, the spread is fine and the average is in the wrong place. On a CNC lathe that is usually a tool offset or a program nominal that was set to the edge of the tolerance. The fix is cheap: adjust the offset, re-verify, and recalculate.
If Cp and Cpk are both low, centring will not rescue you. The spread itself is too wide for the tolerance. Now you are looking at variation reduction: worn spindle bearings, inconsistent bar stock, chatter, a dull insert changed too late, or a gauge that is adding noise of its own. These fixes take longer and often need a structured problem-solving effort rather than a single adjustment.
Whatever you change, prove it worked. Run a fresh set of subgroups after the change, confirm the control chart is still stable, and recalculate both indices. Do not report a new Cpk from old data with a corrected average pencilled in. The customer is buying parts, not arithmetic.
Standard Procedure
- 1Compare Cp and Cpk to decide whether the issue is centring or spread.
- 2For a centring problem, adjust the tool offset or program nominal toward the tolerance midpoint.
- 3For a spread problem, list variation sources (tooling, material, machine, gauge) and attack the largest first.
- 4Run new subgroups, confirm stability on the control chart, and recalculate Cp and Cpk.
- 5Record the change and the new indices so the improvement is traceable.
Lesson Summary
- ✓Centring problems are usually fast to fix on a CNC machine; spread problems need variation reduction.
- ✓Always verify improvement with fresh data, not with a corrected version of the old data.