Home machining

Home machining is cutting metal to a dimension. A lathe rotates the workpiece against a fixed tool and produces anything round; a mill rotates the tool against a fixed workpiece and produces everything else. Between them they make parts that cannot be bought, repair parts that are no longer made, and produce the fittings every other workshop craft turns out to need.

The craft divides by what it serves rather than by technique. Model engineering — locomotives, stationary engines, clocks — is the traditional heart of it in Britain and the reason the society network exists. Repair and restoration is the largest practical use: a bush, a shaft, a thread, a part for a machine whose manufacturer disappeared in 1974. One-off making covers everything else, from tooling for the workshop itself to parts for other hobbies entirely.

What beginners consistently underestimate is that the machine is the cheap part of the problem and workholding is the craft. Getting metal to sit exactly where you want it, rigidly, so that it can be cut without moving, is most of the skill and nearly all of the frustration. A cut that goes wrong is almost always a setup that was not rigid enough.

The actual skill

Workholding, not cutting

Best first move

A model engineering society

Budget for tooling

As much again as the machine

What this page will not give you

Feeds and speedsDeliberately. See below.

1. 1

   Step 1: Join a model engineering society before buying anything

   Most have equipped workshops, training sessions and members who will teach you on their machines. Thirty to eighty pounds a year against several thousand, and you will learn what you actually want before you buy it.

   You will know it worked when: Real experience, and a much better purchase later.

2. 2

   Step 2: Learn to hold work before learning to cut it

   Three-jaw, four-jaw, between centres, faceplate, and how each one affects concentricity and rigidity. This is the part that determines whether anything else works.

   You will know it worked when: Setups that do not move.

3. 3

   Step 3: Buy measuring equipment early

   A caliper, a micrometer and a dial indicator. Machining is the act of measuring repeatedly and cutting occasionally, and without instruments you are guessing at the only thing that matters.

   You will know it worked when: The ability to know rather than hope.

4. 4

   Step 4: Make tooling for the machine

   A mandrel, a boring bar holder, a set of soft jaws, a tailstock die holder. Traditional first projects because they are useful, forgiving, and teach the machine while improving it.

   You will know it worked when: A better workshop and a trained operator at once.

5. 5

   Step 5: Then buy a lathe, second-hand, with tooling

   An older British or a decent import machine with a chuck set and a toolpost included is worth far more than a bare new machine at the same price. Tooling is where the money goes.

   You will know it worked when: A working setup rather than a machine and a shopping list.

• Rigidity, above every other specification — Mass, bearing quality and how well the machine is bolted down determine what cuts are possible far more than headline capacity does. A small rigid lathe beats a large flexible one at everything.
• A three-jaw and a four-jaw chuck — The three-jaw is fast and approximately concentric; the four-jaw is slow and exactly whatever you set it to. You need both and beginners frequently buy only the first.
• Measuring instruments you trust — A digital caliper for setting out, a micrometer for anything that must fit, and a dial indicator for setup. Cheap calipers are fine; a cheap micrometer is not.
• A quick-change toolpost — The upgrade that changes the experience most. Repeatable tool height and thirty-second changes turn a discouraging setup task into nothing.
• Cutting tools, in one system — Ground high-speed steel or indexable carbide inserts. Both work; mixing systems randomly is how a beginner ends up with a drawer of tools and no consistent results.
• A digital readout, eventually — Removes backlash arithmetic from every operation and is a genuine accessibility aid as well as a convenience. Worth more than most people expect.
• What can wait indefinitely — A large lathe, a full collet set, exotic tooling and a second machine. The most common beginner error is buying capacity instead of tooling.

Machining splits cleanly into two kinds of knowledge and only one of them belongs on a page. Executing a cut is judged live, by sound, by the colour and shape of the swarf, and by what the machine is telling you through your hands — and the parameters that govern it depend on the material, the tool, the rigidity of your specific machine and how the work is held, which is why this site publishes none of them. Planning the job is the opposite: it happens before the machine is switched on, it is written down by every competent machinist anyway, and getting it wrong is the reason most amateur parts end up scrap.

The organising idea is datums. A datum is the surface or axis every subsequent measurement is taken from, and the single most common cause of a part that will not assemble is measuring each feature from the previous one, so that four small errors accumulate into one large one. Choose a face and an axis at the start, machine them first, and refer everything to them. Where a part is turned round and re-held, the question that decides whether it works is how you re-establish that datum.

The second idea is order of operations, and it follows from the first. Rough before you finish, so the stresses released by heavy cutting are removed before the surfaces that matter are made. Do the operations that need the most rigidity while the workpiece is largest. Drill before you bore. Leave the feature that is hardest to hold until the part still has something convenient to hold it by — and if you cannot answer "what am I gripping for the last operation" before you start, you do not yet have a plan.

The third is workholding, which the safety section already treats as a safety system and which is equally a precision one. A part held in a way that lets it move produces a surface that is wrong in a way no amount of care at the handles can correct, and a part held in a way that distorts it will spring back into a different shape when released. Both are decided before cutting.

Measurement, which is the discipline the craft actually runs on:

• Measure the thing, not the handwheel — Dials and readouts describe what the machine was asked to do. Only a measurement describes what happened, and the difference between the two is the backlash, deflection and tool wear that make machining interesting.
• Sneak up on a dimension — Take a light cut, stop, measure, and take the rest. Material removed cannot be replaced, and the entire cost of a cautious approach is a few minutes.
• Know what your instrument can resolve — A measuring tool has a resolution and an accuracy and they are different numbers. Working to a tolerance finer than your instrument can resolve is a way of generating confident nonsense.
• Let things reach temperature — Metal expands measurably at the tolerances this craft works to, and a part measured hot from cutting will be a different size cold. This surprises everybody once.
• Deburr before you measure — A burr is a raised edge and calipers will happily include it. A great many mystifying half-millimetre errors are a burr.

• The chuck key never stays in the chuck — A key left in and the machine started throws it with real force. The universal habit is that the key is never let go of while it is in the chuck.
• Workholding is a safety system, not a convenience — A part that comes loose in a chuck or a vice becomes a projectile. Most dramatic amateur incidents are workholding failures rather than tool failures.
• Stop the machine to measure — Every time, without exception. Measuring a rotating part is the second most common route to an entanglement injury after hand-held abrasive.
• Swarf is not touched — Long turnings are sharp along their whole length and hot when fresh. A brush and a hook, never fingers, and never while anything is turning.
• Isolate before changing anything — Tooling, chucks, belts. Switched off is not isolated, and a machine that starts while your hand is in it does not care why it started.

• Model engineering — Locomotives, stationary and traction engines, built to drawings over years. The traditional core of the craft in Britain and the reason the society and exhibition network exists. Published here as its own subject.
• Repair machining — Bushes, shafts, threads and replacement parts for machines nobody supports any more. The most practically useful branch and the one that justifies the machine to everybody else in the house.
• Toolmaking — Making tooling for the workshop itself. Traditional first projects, immediately useful, and unusually forgiving of a beginner’s tolerances.
• Clockmaking and horological work — Machining at very small scale with its own tooling, thread standards and lathes. A distinct discipline that shares only the principles. Published here separately.
• Firearms and engine work — Both use the same machines and both bring separate regulatory or safety regimes that sit entirely outside this page.
• CNC conversion — Adding stepper motors and control to a manual machine. A large and rewarding project, and a different craft afterwards — the CNC routing subject covers the workflow side of that world.

The culture of this craft is the model engineering society, and it is one of the most generous institutions in British hobby life. Societies own equipped workshops, run tracks, hold exhibitions and contain a very high density of people who will spend an afternoon showing a beginner how to grind a tool. The demographic skews old and the societies know it, which means a new member of any age is met with something close to relief.

The second cultural feature is documentation. This is a craft with a century of magazines, drawings and books behind it, and the practice of publishing a design so that other people can build it is deeply embedded. That makes it unusually learnable from print compared with crafts that live on video, and it is why so much of the useful material is older than the internet.

The tension worth naming is between the traditional and the modern approaches — hand-ground high-speed steel and a dial against carbide inserts and a digital readout. Both work, both have advocates who overstate the case, and a beginner is best served by ignoring the argument entirely and using whatever the person teaching them uses.

1. 1

   Step 1: Turning something round to a dimension

   Facing, turning to diameter, measuring, and hitting a size on purpose. Sounds trivial and is the whole foundation.

   You will know it worked when: Control over one axis.

2. 2

   Step 2: Holding work several ways

   Four-jaw setup, between centres, faceplate work. The point at which the machine stops limiting what you can attempt.

   You will know it worked when: Setups rather than a single method.

3. 3

   Step 3: Threading, boring and parting

   The three operations that intimidate beginners and that between them cover most real work.

   You will know it worked when: Parts that fit other parts.

4. 4

   Step 4: Working from a drawing to a tolerance

   Order of operations planned, datums chosen, fits achieved deliberately rather than by luck.

   You will know it worked when: Engineering rather than turning.

5. 5

   Step 5: A long project, or a converted machine

   A locomotive over several years, a full restoration, or CNC conversion and a different craft again. All three are common destinations and none of them is an end point.

   You will know it worked when: A workshop that can make what you need.