Notes on control, tooling and evidence

Triggering die PM on produced parts is the settled half. The number is where toolrooms stall. Anchor it to the build spec, let recorded failure counts argue it down, and write the count into every service so the interval can learn.

Tool trackers built for construction crews and CMMS built for pumps both miss what a die record has to hold. The software question is really a records question: repairs, services, counts and spares as one story that reads back on demand.

Cp and Pp are not two names for one thing. One asks how good the process could be, the other how good it actually was. Report the one you were not asked for and you answer a question nobody put to you.

A variable Gage R&R needs numbers to work on, and a go/no-go gauge only gives you pass or fail. So you measure agreement, not repeatability in millimetres. What an attribute MSA checks, and the known-standard parts it depends on.

An automated vision system has no operators, so the three-appraiser Gage R&R template measures nothing. What an MSA on a machine that makes its own pass/fail decision actually tests, and why it comes down to the boundary samples you feed it.

The empty shelf on breakdown day and keeping two of everything are both failures of policy. A minimum is built from lead time, real usage and what a stockout costs, and wear parts and failure parts are stocked on different logic.

A failed normality test does not mean bad parts. Standard Cpk assumes a bell curve, and flatness, position and runout are non-normal by nature. How to tell an inherent shape from an unstable process, and what to do about each.

A die does not wear by the week. It wears by the hits it makes. Service it on a calendar and you either strip a healthy die too early or lose a good one mid run. The trigger has to match how the tool actually ages.

An auditor asking for maintenance records is rarely counting whether you have them. They are checking whether the maintenance you scheduled is the maintenance that actually happened. Why PM adherence is the real test, and what a record has to show to pass it.

Everyone quotes 1.33, and some customers want 1.67. Both are real, and neither is the whole answer. What sets the target, and why a Cpk on an unstable process is a number about nothing.

Auditors test which control plan revision governed the work on a past date, not just that a plan exists today. The revision and release records that answer the question, and why the current plan is not enough.

Proving critical spares are a controlled system, not a shoebox. The minimum-stock, movement and consumption records an auditor wants to see for a die or line, and why a full shelf is not evidence.

When an auditor asks for the full maintenance history of a die, the last PM sheet is only part of the answer. The job cards, repairs, PM against produced-part counts, spare usage and release history all belong on the die record together.

An out-of-spec reading is not the evidence. The response is. What an auditor wants to see after a measurement fails, and the record chain from the failed value to the action that closed it out.

Every plant has them. The corner behind the press line, the cell inside a steel enclosure, the spot where the tablet loses signal. If your quality capture dies with the wifi, the data ends up on paper. It does not have to.

A Cpk number in a spreadsheet is not process capability evidence. Without the characteristic, the specification limits, the sample window and the revision behind it, the number cannot answer the question an auditor actually asks.

When the control plan and the inspection record are separate documents, the link between requirement and measurement has to be rebuilt manually. That rebuilding is where audit evidence breaks down.

The auditor question is not "where is your folder?" It is "show me evidence that this control was in place." VoraAudit is built to answer that question from source records, not from a parallel file cabinet.

When the auditor asks for the full service history of a tool, a folder of maintenance reports is only part of the answer. The job cards, repairs, PM records, spares usage and release history all belong together.

The records existed. The work was done. But when the auditor arrives, teams spend days reconstructing evidence that should have been automatic.

The spreadsheet has the number. But it does not know which tool made the part, which revision was active, who captured the measurement, or what happened when it went out of spec.