Control Plan Builder

What this tool does

CM8-242 builds a process control plan in the format automotive and general manufacturing quality systems expect: one row per control, stating the part, the process step, the characteristic being controlled, its specification, the gauge that measures it, how many pieces are checked and how often, the control method, and — the part most plans skimp on — the reaction plan that tells the operator exactly what to do when a check fails. One file can hold the plans for several parts; the report prints the largest plan in the classic column layout, alongside a gauge calibration schedule, the control-method hierarchy and the check-frequency mix.

Everything runs inside this single file. There is no account, no upload and no network request of any kind, so your process know-how — tolerances, methods, gauges — never leaves the computer you are using.

Where a control plan sits

In the standard quality-planning chain, the process FMEA works out where the process can fail and what that failure would do — it finds the risks. The control plan is the other half of that pair: for each risk worth controlling, it says how the process is held in check — what is measured, with what, how often, and what happens on a failure. A control plan without an FMEA behind it is a list of checks nobody prioritised; an FMEA without a control plan is a list of worries nobody acted on.

A PPAP submission — the evidence pack a customer reviews before approving production parts — expects both documents, consistent with each other and with the drawing. If a characteristic is special on the drawing, it should carry a high severity in the FMEA and a tight, well-defended row here.

Product and process characteristics

Every row controls one of two kinds of thing:

• Product characteristics are features of the part itself — a bore diameter, a thread depth, a surface finish, a plating thickness. They are what the customer receives, and they can be checked on the part after the operation.
• Process characteristics are settings and conditions of the process — coolant concentration, furnace temperature, torque setting, fixture condition. Nobody ships a coolant concentration, but when it drifts, the surface finish goes with it. Controlling the process characteristic is usually earlier, cheaper and less wasteful than catching the product characteristic after it has gone wrong.

A mature plan has both. A plan that is all product characteristics is pure inspection — it finds scrap after making it. A plan that is all process characteristics is an act of faith that the settings really do determine the part.

Special characteristics

A special characteristic is one where a nonconformance affects safety or regulatory compliance, or critically affects fit, function or the customer's process — brake dimensions, pressure integrity, anything the drawing marks with a special symbol. Customers and auditors treat these rows differently: they expect tighter control methods, tighter frequencies, and an explicit reaction plan, and they will check that the symbol on the drawing, the FMEA line and this row all agree.

This tool enforces the floor of that expectation:

The tiles also count special characteristics protected by neither error-proofing nor SPC — rows where the most important features of the part depend entirely on someone remembering to look. That count is the first thing to drive to zero.

The hierarchy of controls

Not all control methods are equal, and the hierarchy chart ranks them deliberately. Error-proofing (poka-yoke) makes the defect impossible or catches it mechanically every time — a locating pin that stops a part seating backwards, a reject chute with no override. It works at 3am, on the operator's worst day, at full line speed. Automatic gauging is close behind: every piece checked, no attention required. SPC watches the process statistically and signals drift before parts go out of tolerance — it prevents scrap rather than merely finding it, but someone must plot the points and act on the signals. First-off approval, check sheets and visual standards all depend on a person doing a check at the stated time, and human vigilance is the least reliable control there is — not because people are careless, but because sustained attention to a rare event is something humans are genuinely bad at.

The honest economics: error-proofing costs money once, at the fixture or the machine, and then costs nothing per part forever. Inspection costs a little money on every part, every shift, indefinitely — and still lets defects through. Over any real production volume the fixture is cheaper; it just presents its bill up front. The error-proofed share on the tiles is the maturity signal: watch it rise as the plan is revised, and be suspicious of a plan that has stayed all-inspection for years.

Writing the reaction plan

The reaction plan is the row's payoff, because checks exist for the day they fail. Write it for the operator on night shift with no engineer in the building: stop or keep running; how far back the quarantine reaches; who to notify; how the process is corrected; what proves it is right again. The quarantine scope has a standard answer — everything made since the last good check — which is also why frequency matters: an hourly check means an hour's parts in the cage, a weekly check means a week's parts, possibly already shipped. If a reaction plan says only “inform supervisor”, the real plan is whatever the supervisor improvises. The tiles count controls with no reaction plan at all; on an active plan that number should be zero.

Sample size and frequency

Sample size and frequency are a risk decision, not a habit. Two questions set them: how bad is a miss (severity — a special characteristic deserves more looking than a cosmetic one) and how likely is a drift (capability and stability — a process that holds a quarter of the tolerance can be checked less often than one that uses most of it). High severity with marginal capability points to every piece or automatic gauging; low severity with proven capability can justify per batch. Frequency also bounds your exposure: it is the maximum quantity you can make between a problem starting and a check finding it. Set it by asking how many parts you can afford to quarantine — and revisit it with capability evidence, in both directions: tightening after a failure is normal, relaxing after sustained capability is legitimate and saves real money.

Gauge calibration

Every measured control leans on its gauge, and a gauge out of calibration turns the whole column of checks into noise — the readings continue, but they prove nothing, and a calibration failure discovered late can force recall of every part accepted since the last good calibration. Each row carries the gauge's calibration due date; the register flags overdue and due-soon gauges, the calibration horizon chart shows the workload coming, and the schedule table lists each distinct gauge once with everything it is used for. Controls with no calibrated instrument — poka-yoke fixtures, visual standard boards — leave the date blank, though fixtures still deserve a periodic condition check of their own. An overdue date is deliberately not blocked when saving a row: the flag exists to surface the problem, not to stop you recording it.

Keeping the plan alive

A control plan is a living document, and auditors read it that way: the fastest audit finding in manufacturing is a control plan that says one thing while the operator, quite sensibly, does another. Use the revision field so every row states the plan issue it belongs to, and the status field to manage change: draft for controls being introduced, active for the released plan, superseded for rows replaced by a later revision — keep superseded rows rather than deleting them, because they are the history of how the plan evolved and why. Revise the plan whenever the process changes: a new machine, a new gauge, a tool or material change, a customer complaint, or capability evidence that justifies relaxing a check. If the shop floor has a better method than the plan describes, the plan is wrong — change the document, not the behaviour.

FAQ

One plan per part, or one row per characteristic? One row per control — a characteristic at a process step. The same characteristic can appear at two steps (checked at the operation and again at final inspection), and that is two rows. The part field groups rows into plans, so one file can hold several parts.

What goes in the plan — every dimension on the drawing? No. Every characteristic the FMEA says is worth controlling, every special characteristic, and the process settings that determine them. A plan with three hundred rows nobody reads controls less than a plan with thirty rows everyone follows.

Who owns a row? The person or role that performs the check — operator, setter, shift leader, inspector. Ownership by a department means ownership by nobody at 3am.

Can a special characteristic rely on a check sheet? It can, but the tiles will keep counting it as unprotected until it is error-proofed or on SPC — and a customer auditor will ask the same question the tile does.

Prototype, pre-launch and production plans? Some customers expect all three phases, with tighter checking in the earlier ones. Keep them as separate parts (for example “HC-204 — pre-launch”) or separate files, and mark the retired phase superseded when production takes over.

Saving your work

Controls, settings and the report header are written to this browser's local storage as you type, and the toolbar shows the time of the last save. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the register for spreadsheet work. Reset asks twice, then erases everything this tool has stored. There is no undo. A control plan encodes your process know-how — treat exports as confidential.

Accuracy & disclaimer

This tool records what you enter and summarises it faithfully. It cannot know whether the checks are actually performed, whether the gauge named is the gauge used, whether the reaction plan is followed, or whether the characteristics chosen are the ones that matter — those come from your FMEA, your drawing and your process knowledge. Customer-specific requirements for control plans differ between industries and between customers; where a customer format or approval applies, this document supports it but does not replace it. Nothing here is engineering, certification or compliance advice.