How to measure a replacement part

A photograph shows what a part looks like. It says almost nothing about whether a replacement will fit. This is the measurement routine we use before a part becomes a brief.

Most failed replacement parts do not fail because the model was badly drawn. They fail because one dimension was wrong, or because nobody wrote down where the measurement started. This guide is the short version of the routine: measure the envelope, then the interface, then the clearances, then the details, and write down the datum for every number.

Why a photo is not a measurement

A phone photo has no scale. Even with a ruler in the frame, perspective bends straight lines and the ruler is rarely in the same plane as the feature you care about. Photos are excellent for shape, texture and context: which way the knob turns, where the crack runs, what the surrounding panel looks like. They are poor for numbers.

Use the photo to remember the object. Use callipers to describe it.

Rule of thumb: if a dimension will decide whether the part fits, it has to come from a measuring tool, not from a picture or a product listing.

What to put on the table

  • Digital callipers, 150 mm, resolution 0.01 mm. A €20 pair is enough; a ruler is not.
  • The original part, including the broken pieces. Tape the fragments together if you can.
  • The mating part: the shaft, panel, housing or screw the replacement has to meet. This is often more useful than the broken part.
  • A thread pitch gauge if there are threads, or a known screw that fits.
  • Something to write on, or the brief form open next to you.
  • Good light and a plain background. A ruler in the photo still helps whoever reads your notes later.

The four passes

Work in this order. Each pass answers a different question, and mixing them is how measurements get lost.

1. Envelope: how big is it overall?

Width, depth and height of the part as it sits in use. This is the number most people start with, and the one that matters least for fit, but it decides whether the part collides with anything around it and how much material you need.

2. Interface: what does it attach to?

This is where fit lives. Measure the mating feature, not the worn one:

  • Hole diameter, measured at the top, middle and bottom of the bore, and in two directions 90° apart. Out-of-round holes are common.
  • Shaft diameter, plus any flat, keyway or spline. A "D" shaft needs both the diameter and the distance across the flat.
  • Thread: nominal size, pitch and length of engagement. Measure pitch over ten threads and divide by ten if you have no gauge.
  • Boss or panel thickness at the mounting point, and the depth available behind it.
  • Fastener spacing: centre-to-centre, not edge-to-edge. Measure twice: screw patterns are rarely perfectly symmetric.

3. Clearances: what has to move?

A cover needs millimetres. A rotating knob needs tenths. Write down what the part has to do before you decide a number:

  • Sliding or rotating fit: allow roughly 0.2-0.4 mm total clearance on FDM parts; printed holes usually come out slightly undersized, so a hole drilled or reamed to size afterwards is normal practice.
  • Snap fits: measure the undercut depth and the deflection length, and note the material you intend to use.
  • Gaps to neighbouring surfaces: measure with the part in place, not on the bench.

4. Details: the small things that decide "close enough"

  • Fillets and radii, especially at the base of a shaft or inside a corner.
  • Chamfers, and which edges are actually sharp.
  • Surface texture: is the original smooth, knurled, ribbed?
  • Orientation marks: an arrow, a flat, a notch. These are functional, not decorative.

A worked example: the drawer knob

A knob has come off a wooden drawer. The wooden drawer is worth keeping; the knob is not. Here is what a usable set of measurements looks like:

WhatValueHow it was measured
Knob envelope32 × 32 × 22 mmCallipers across the widest point, height from the panel face
Mounting hole in the drawer faceØ 4.2 mmCallipers inside the bore, two directions
ScrewM4 × 25 mm, pan headThread measured over 10 pitches; length under the head
Counterbore in the knobØ 8.4 mm × 6 mm deepCallipers, depth rod
Clearance to the drawer edge6 mmMeasured with the knob held in place
FunctionPull, rotates freely, no load beyond openingNoted from use

Notice what is not there: the colour, the exact curve of the original, the brand. Those matter to the owner, not to the fit. Notice also that the screw diameter and the hole diameter are both recorded, because a printed knob that grips the screw is a different design from one that lets the screw pass through into the drawer.

Record it once, in one place

Measurements scattered across a notes app, three photos and a WhatsApp thread are how projects stall. Put them in one place, in millimetres, with the datum written next to each number. A brief that says "height 22 mm" is ambiguous; "height 22 mm from the panel face to the top of the knob" is not.

Our project brief builder keeps the part type, the description, the three overall dimensions and your photo reference in one place, then gives you a file you can hand to a print shop or keep for later.

Seven ways measurements go wrong

  1. Measuring the broken part instead of the mating part. If the original snapped off at the shaft, the stub is not the interface.
  2. Measuring a worn part. A knob that rattled for ten years is smaller than it was. Measure the shaft, then choose your own clearance.
  3. Measuring the gap instead of the part. The hole in the panel is not the diameter of the shaft that goes through it.
  4. No datum. "Height 22 mm" means nothing without "from where".
  5. Callipers on a curved surface. The jaws bridge the high points. Measure flats, or use the depth rod.
  6. Reading the wrong unit. Some callipers switch between mm and inches silently. Check the display before every session.
  7. Trusting a listing. Product pages describe the nominal part, not the one on your bench. Verify the two interface dimensions yourself.

When measuring is the wrong move

If the part carries structural load, seals gas or water, insulates mains voltage, touches food, or is part of a safety system, a measured print is usually the wrong answer. Not because it cannot be modelled, but because the failure mode is not worth the saving. We wrote a separate guide on when to print a replacement part and when to buy the spare.

The same applies to a part you cannot measure: a worn cam, a spring, a precision gear. In that case, measure the thing it drives instead, and design from the function rather than the fragment.

Questions

Do I need digital callipers to measure a replacement part?

For anything that clips, screws, rotates or seals, yes. A ruler gets you within a millimetre, and a millimetre is the difference between a press fit and a loose one. Callipers also let you measure bores and depths, which a ruler cannot.

How many measurements does a part need before you can model it?

Three overall dimensions are enough to describe the size, but a fitting part also needs the interface dimensions (hole or shaft, flats, threads), the clearances, and the thickness at the mounting point. Most small parts settle at six to twelve numbers.

What tolerance should I use for a 3D printed replacement part?

For a sliding or rotating fit, roughly 0.2-0.4 mm total clearance is a common starting point on FDM printers, and printed holes usually come out slightly small. For a static cover, you can be much more generous. Always check the first print before committing to a batch.

Written for the Refabrication preview. Sizing, material and safety notes here are general guidance, not engineering or compliance advice. Check any part against your own equipment, loads and local requirements before you print or install it.