Re: Should cmm programs be controlled?
The following is just a portion of a paragraph from UTC's (Pratt & Whitney, Sikorsky, Hamilton Standard, etc) 8 page procedure for control of non-deliverable software covering just the verification and validation of CMM programs.
"Verification and Validation:
Define the Verification and Validation process.
Test procedure or test description and results shall be documented,
reviewed and retained.
Provide objective evidence that the software performs its required
function.
Trace software to requirements.
Inspection review and approval of software must be performed by
someone acting in an acknowledged product integrity role. Software used
to verify quantitative values (e.g., CMM, etc.) requires an independent
method of validation (i.e., layout inspection, fixture check or comparison
with another CMM program previously verified by an independent
method) and correlation of the two sets of results.
Acceptable correlation requires the difference to be within 10% of the
tolerance for each characteristic. Differences greater than 10% but
not exceeding 25% may be acceptable with documented justification.
Differences greater than 25% are not acceptable.
Variable data shall be recorded and retained."
Customers of aerospace machined components put heavy controls on non-deliverable software (i.e. CMM programs, NC programs, test stand programs, etc.)
AS9100 also requires:
"7.5.1.3 Control of Production Equipment, Tools and Software Programs
Production equipment, tools and software programs used to automate and control/monitor product realization processes, shall be validated prior to release for production and shall be maintained."
Good luck