From caliper to printed part

Everything you print has to meet something real

The whole reason for this course is parts that fit the world: a bracket that fits the shelf, a case that fits the board, an adapter that fits the hose you already own. Which makes measuring an actual physical object the first step of most projects, and it deserves more care than it usually gets.

• Buy digital calipers. A cheap pair is transformative and costs less than a spool of filament. A ruler cannot resolve the 0.2 mm that decides whether a lid closes.
• Measure three times, in three places. Injection moulded parts have draft, extrusions bow, and the thing you are copying may not be the shape you assume. Take the measurement that matters for the fit — usually the largest.
• Write down what you measured, not just the number. “19.8 mm across the flats, not the corners” is a measurement. “19.8” is a number you will misinterpret next week.
• Put measured values into variables. #boardWidth = 56.2 mm at the top of the tree means the day you remeasure and find it was 56.5, one edit fixes the model.

Checking fit before you spend three hours of filament

When two of your own parts have to fit each other, model them in the same Part Studio. They then share one coordinate system, one set of variables, and you can see interference on screen rather than discovering it on the bed.

Onshape gives you two cheap checks before committing. A section view slices the display so you can look inside a closed assembly — the fastest way to spot a lid lip that bottoms out. And Measure reports the true distance between any two faces, which is how you confirm that the gap you designed is the gap that exists.

Assemblies are the next step up: separate parts joined by mates that let you move them the way the real thing moves, and check that a hinge clears its housing through the whole swing. For a two-part enclosure you can skip assemblies entirely. For anything with a moving joint, they will save you a print.

Getting it out cleanly

Export is where a well-built model can still be spoiled. Three settings matter.

• Format. STL is universal and stores triangles only. 3MF stores units, colour and metadata, is far harder to get wrong, and every current slicer reads it — prefer it when you have the choice. STEP is the format to send someone who needs to edit the geometry; never send an STL to a machinist.
• Resolution. The default coarse setting facets curved surfaces visibly. Choose fine, or set a chord tolerance around 0.01 mm — small holes are where the faceting shows first and hurts fit most.
• Units. STL carries no units. If a part arrives in the slicer at 1/25th of its intended size, the file was written in inches and read as millimetres. 3MF makes this impossible.

The loop that actually makes you good

Model, export, slice, print, measure the result, change the variable, repeat. The skill is not getting it right first time — nobody does. The skill is making each lap short: parametric model, named variables, measured clearances, one number to change.

Two habits shorten the loop dramatically. Print a fit fragment rather than the whole part: if only the lid lip is in question, slice off a 20 mm section of it and print that in four minutes instead of printing the enclosure in four hours. And create a version in Onshape before each print, so the physical object on your desk maps to a state you can return to.

Where to go after this

You now have the loop that everything else attaches to. The obvious next moves, roughly in order of usefulness to a maker: threaded inserts and fastener bosses, snap fits and living hinges, assemblies with mates, and sheet-metal or surfacing if your projects pull you that way. The resources page lists where to learn each, and which communities are worth asking in — because the last third of this skill is not knowledge or practice, it is other people telling you what they already broke.