Drawing & Document Register

What this tool does

CM8-276 is a controlled register of drawings, specifications, work instructions, procedures, datasheets and reports. You record one row per revision, with who drew it, who checked it, who approved it, where the controlled copies went and what changed. The tool then tells you which revision of each document is currently in force, which superseded copies are still out on the floor, which issues were released without an approver, and which documents have run past their review date.

Everything runs inside this single file. There is no account, no upload and no network request of any kind, so part numbers, supplier names and the internal detail of your product never leave the computer you are using.

The only question that matters

Document control sounds like an administrative discipline. It is not. Strip away the vocabulary and there is exactly one question underneath all of it:

Everything else — the numbering scheme, the approval signatures, the master list, the archive — exists only to make the answer to that question reliably "yes". If a setter is working from a laminated card that was superseded eight months ago, you do not have a document control problem, you have a scrap problem, a warranty problem and possibly a safety problem, and the paperwork behind it will not save you.

Read the rest of this help with that question in mind. Where a rule seems fussy, ask what it prevents. Most of them prevent the same thing.

A register is history, not a list

The common mistake is to keep a spreadsheet with one line per document and edit the revision letter in place. That gives you today's answer and destroys every previous one. This register works the other way: one row per revision, and superseded rows stay.

That matters more than it sounds. When a customer asks which revision was in force when their unit was built eighteen months ago, a list cannot answer. A register can: find the document number, look at the revision dates, and the answer is there in black and white. The same applies when an investigation asks whether the operator had the current instruction, when a warranty claim turns on whether a part was made to the old material spec, or when an auditor asks to see the change history for a controlled procedure.

It also makes the change itself legible. A row that says "rev C, material changed from mild steel to 304 stainless following the corrosion failures at the coastal site" is a piece of engineering knowledge. Writing "general update" in the change summary throws that away, and future-you will be the one who needs it.

Revisions: letters, numbers and when to change one

Two conventions dominate, and either is fine as long as you use one consistently:

• Letters — A, B, C. Traditional on drawings. Letters that look like numbers are usually skipped (I, O, Q) because handwriting makes them ambiguous.
• Numbers, sometimes with a decimal — 1, 2, 3, or 1.0, 1.1, 2.0. Common on procedures and specifications, where a decimal is often used for an editorial change and a whole number for a change of substance.

The revision field here is free text precisely so you can use whichever your organisation already uses, including mixed schemes across document types. What matters is the rule for when it changes:

The failure mode this rule exists to stop is universal and it has a name on every shop floor: pen and initials. A dimension is wrong, somebody crosses it out on the copy pinned to the board, writes the right one and initials it. The job runs correctly. Then a second copy of the same drawing — the unmarked one, in the QC folder or on the supplier's screen — gets used next month, and the wrong dimension is back. Marking up a copy is not a revision; it is a note that a revision is needed. Raise the change, issue the new revision, and pull the marked copy.

The withdrawal problem

Issuing a new revision is the easy half. Removing the old one is the half that gets missed, and it is where the register earns its keep.

Superseded paper on a shop-floor board is simultaneously the most common audit finding in document control and one of the most common causes of scrap. The mechanism is always the same: a new revision goes out electronically, everyone assumes the printed copies have been dealt with, and nobody actually walked the route. Six months later a part is made to the old hole pattern.

Two fields exist for exactly this. Copies to withdraw records how many hard copies of a revision are physically out there — on boards, laminated in cells, in inspection folders, in a supplier's hands. Withdrawal confirmed is the tick that somebody went and got them. Until that tick is present on a superseded revision that had copies out, the register flags it as an uncontrolled copy risk and lists it at the top of the control-gaps table with the number to collect.

Note what the tool deliberately does not do: it does not stop you saving a superseded revision with copies still out and no confirmation. That state is not an error to be blocked, it is the reality to be surfaced. Blocking it would just teach people to tick the box.

When you do the withdrawal, destroy the copies or mark them clearly and permanently as void. A superseded drawing kept "for reference" in the same folder as the current one will be used eventually.

Author, checker, approver — and why they are three people

The register records three names against a revision, and the separation is the point:

• Author or draughtsman — the person who made the change. They know what they intended. They are the worst possible person to verify it, because they will read what they meant rather than what is on the page.
• Checker — a competent second pair of eyes on the technical content. Do the dimensions close? Is the tolerance achievable? Does the material call-up match the specification? Will it fit the mating part? This is a technical review, not a proofread.
• Approver — the person with the authority to release the document into use, and to accept the consequences of doing so. This is a decision, not a check: is this change right for the business, has the customer been told where the contract requires it, is the supplier ready, does stock of the old part need using up first?

In a small organisation one person may hold two of these roles, and that is workable. Author and approver being the same person on a change of substance is where it stops being workable, because nobody independent has looked. The tool will not let you set a revision to Issued without an approver recorded — and where a row was imported or created without one, it appears in the control-gaps table as an issue released by nobody in particular.

Distribution lists that are actually maintained

The distribution field is a plain description of where controlled copies live: "Cell 2 board, QC office, supplier portal". Keep it in the words the people who have to walk the route would use. A distribution list written in department codes is a list nobody can act on.

Three habits keep it honest. Record the place, not the person — people leave, boards do not. Include external holders, because a supplier working to your old drawing is your problem and not theirs. And when the distribution changes, change it on the current revision rather than starting a note somewhere else; the register is only useful if it is where people look.

Drafts are the exception, and the tool enforces it: a document with status Draft cannot carry a distribution. A draft that has been sent out is not a draft, it is an unapproved issue, which is the worst of both. Either issue it properly or keep it in the drawing office.

Periodic review

Drawings tend to change when the product changes, which is a natural trigger. Procedures and work instructions have no such trigger, so they drift: the job changes, the document does not, and eventually the document describes a way of working that nobody has used for two years. That is worse than having no document, because it trains people to ignore the documents that do matter.

Set a next review due date on issued documents — twelve months is a common interval for procedures and work instructions, twenty-four for stable drawings, but choose intervals you will actually honour. The tool flags issued documents past their date as overdue and shows those coming up inside the window you set in Settings.

A review has two honest outcomes. Either the document still describes reality, in which case record that it was reviewed and set the next date; or it does not, in which case raise a revision. Quietly moving the date forward is the third outcome and it is how registers become fiction.

Linking to change notes

An engineering change note and a document register answer different questions and you need both. The change note says why: what problem was found, what options were considered, what the disposition is for stock and work in progress, who agreed it, what it costs. The register says what is current: this document, this revision, from this date, approved by this person.

Recording the change note reference against the revision joins the two, so that anyone looking at a revision can find the reasoning behind it, and anyone closing a change note can confirm the documents it called for were actually issued. A change note whose documents were never revised is a change that did not happen.

Electronic control and paper control

Electronic control is generally better, and it is not automatic. Serving one file from one place, with read-only access and the revision in the file name, removes the whole class of problems caused by copies: there is nothing to withdraw because there was never more than one. Access is logged, the archive is free, and a change reaches everyone the moment it is saved.

What it does not remove is printing. The moment a screen is inconvenient — oily hands, a welding bay, a supplier without access, a tablet with a flat battery — somebody prints, and that print is an uncontrolled copy from the second the next revision is issued. Paper, by contrast, is honest about its own weakness: you can see it, count it and collect it, which is why the copy count in this register is worth keeping even in a mostly electronic system. A workable middle position is to control electronically, to print only onto paper that carries a visible "uncontrolled if printed" or a printed-on date, and to keep the small number of genuinely necessary standing hard copies in the register with their locations named.

How the flags are worked out

Two notes on the arithmetic. Documents controlled counts document numbers, not rows, so a drawing with five revisions on file counts once — the register would otherwise flatter itself the more history it kept. And the copy counts are only ever as good as somebody's honest count of what is out there; the tool adds up what you enter and cannot know what it does not know.

FAQ

Should I delete superseded rows to tidy the register? No. They are the history, and they are the reason the register can answer questions a list cannot. Filter by status if the table feels crowded.

What is the difference between superseded and obsolete? Superseded means a later revision of the same document replaced it — the document lives on. Obsolete means the document itself has been withdrawn and nothing replaces it: the part is no longer made, the process no longer exists. Keep obsolete drawings for spares and warranty history, and make sure they are clearly marked so nobody manufactures to them by accident.

Do certificates and test reports belong here? They can. They are controlled documents in the sense that people need the right one, though they are rarely revised. Many organisations keep them in a separate certificate register; use whichever puts them where people look.

What counts as a controlled document? Anything where using the wrong version would cause harm, scrap, a failed inspection or a broken commitment. Marketing brochures are not controlled documents. The setting sheet taped inside a machine cabinet is, whether or not anyone has ever called it one.

How long should I keep old revisions? Longer than you expect. Product liability, warranty periods and customer contracts all reach back years, and the revision history is often the only thing that can establish what was built. Check what your industry and your customers require, and keep exports accordingly.

Is a high revision count a bad sign? Not by itself — a live product changes. It is worth investigating when one document churns far more than its neighbours, because that usually points at a requirement that was never settled rather than at the drawing office.

Saving your work

Revisions, 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 and includes every filtered record. Reset asks twice, then erases everything this tool has stored. There is no undo. Because the register is itself a record you may need years from now, export it on a schedule and keep the exports where your other engineering records live.

Accuracy & disclaimer

This tool records what you enter and flags what your entries imply. It cannot tell whether the revision marked as issued is the one on the board, whether the copies you counted are the copies that exist, whether the approver had the authority, or whether the change summary describes what actually changed. A tick in the withdrawal box is a claim by whoever ticked it.

Document control is verified by going and looking. Walk the route on the distribution list, read what is pinned up, and compare it with the current revision in this register — the discrepancies you find that way are the real state of your document control, and no register can find them for you.

Retention periods, approval authority, what must be controlled and what a customer or certification body will expect to see differ by industry, contract and country. This is an internal record-keeping aid, not legal, contractual or certification advice.