Which fit class does this joint need?

Most clearance advice collapses every printed joint into a single question: how much gap do I leave? The answer arrives as a flat millimetre value, usually 0.2 mm, and it works often enough that nobody interrogates it. Then somebody uses it on a bearing seat and the bearing falls out, or on a hinge knuckle and the hinge seizes. The flat rule did not fail because the number was slightly wrong. It failed because a hinge and a press fit are not the same problem, and no single number can serve both.

This page is the selection step. It covers the six fit classes the calculator offers, what each one is actually for, what clearance each needs at real diameters, and the four things that change the answer once you leave the happy path: hoop stress, elastic materials, which member owns the nominal dimension, and whether you are printing one part or both.

The six classes, and what each one is for

Each class is defined by a coefficient C, a multiple of the ISO 286 fundamental tolerance unit i. The functional clearance is

clearance_mm = C × i / 1000, where i = 0.45 × D1/3 + 0.001 × D in microns.

That cube-root term is the whole point. Clearance has to grow with size, but sub-linearly, which is exactly what real engineering fits do and exactly what a flat rule cannot do. A negative C means interference: the shaft is deliberately larger than the bore and has to be forced in.

Worked through the formula, here is what those coefficients mean in millimetres at six common diameters. Negative figures are interference.

Selection table: symptom to class

Read this by finding your joint, not by finding a number you like.

A press fit is a stress problem, not a clearance problem

Press is the one class the calculator flags as needing wall thickness, and that flag is the tell. Every other class is satisfied when the parts assemble with the intended gap. A press fit is only satisfied when the boss survives.

Interference works by elastically expanding the bore. That expansion puts the material around the bore into hoop tension, which is a circumferential stress trying to split the boss along a line running away from the hole. So a press fit has two independent failure modes pulling in opposite directions. Too little interference and the joint does nothing: the insert spins, the pin walks out, the bearing creeps in its seat under load. Enough interference but a thin wall and the boss cracks, either on assembly or quietly over the next few thermal cycles.

The calculator's guidance is to keep the wall around the bore at roughly 0.25 × the bore diameter and add a 0.5 mm lead-in chamfer, so the peg self-aligns instead of shaving the bore on entry. Treat both as a starting point rather than a specification; actual margin depends on load, geometry and material, and it is not something a one-dimensional model can certify.

Elastic materials need one class tighter

TPU 95A is flagged elastic, and the model treats it differently on purpose. For any non-press class it notes that TPU deflects rather than binding, so it tolerates roughly one class tighter than a rigid material for the same felt fit.

The mechanism is simple: a rigid part resists, so the clearance you designed is the clearance you feel. An elastic part yields, so some of the interaction you were relying on disappears into deformation. Build a TPU joint at PLA sliding clearance and it will feel loose, because the material gives way before the surfaces load each other. When you are choosing a class for a flexible member, move one step tighter than the table suggests: running instead of free for a soft pivot, sliding instead of running for a located cover.

Press fits in TPU are a separate matter and generally a bad idea. Elastomers relax under sustained load, so an interference fit loses grip over days rather than holding indefinitely. If a soft part must be retained, use a geometric feature, a groove, a lip, a shoulder, and let the fit locate rather than grip.

Basic hole or basic shaft: whichever side you bought

The datum decides which member owns the nominal dimension and which one absorbs the clearance.

Basic-hole system. The hole stays at nominal; the shaft is reduced by the full clearance. Use this when the bore has to accept a purchased part whose diameter you cannot negotiate: a bearing outer race, a metric rod, a standard bushing.

Basic-shaft system. The shaft stays at nominal; the hole grows by the clearance. Use this when the shaft is the fixed component and you are printing the part that goes around it.

This is not a matter of taste. It is decided by which side you bought. Fitting a printed bracket to an 8 mm ground rod, the rod is unchangeable, so the printed bore has to carry the entire clearance: at a free running fit that is 0.363 mm at 8 mm, all of it on the hole. Pressing a printed pin into a printed plate, neither side is fixed and the clearance can be split evenly between them, which halves the deviation each member carries and is usually the more forgiving arrangement.

One printed part or two changes the answer

Shrinkage is the quiet variable. When both members are printed in the same material, both shrink by the same fraction and the error largely cancels: the hole and the peg move together. The moment one member is purchased, that cancellation disappears and shrinkage becomes a full error term. The calculator raises this explicitly, and calls it the single most common reason a fit that worked print-to-print fails against a bought shaft or bearing.

It also flags when shrinkage exceeds the fit itself, comparing against the requested clearance rather than a flat millimetre threshold, because the same shrinkage can be harmless or fatal depending on the class.

ABS shrinks 0.80%. At 20 mm that is 0.160 mm of contraction. Set against a free running fit of 0.497 mm it is a nuisance you could almost ignore. Set against a press fit of 0.062 mm it is 2.6 times the whole interference, which does not merely loosen the joint, it inverts it: the intended interference becomes a clearance and the part drops out. A single-printed press fit in a high-shrinkage material is the most failure-prone combination on this page, and it fails for a reason that has nothing to do with the class you picked.

Frequently asked

Why does everyone say to leave 0.2 mm?

Because it is right once. A sliding fit at 20 mm works out to 0.199 mm, so the folk rule is a sliding fit at one diameter, quoted as if it were universal. At 5 mm the same class wants 0.124 mm and 0.2 is sloppy; at 50 mm it wants 0.273 mm and 0.2 binds. For a hinge at 20 mm the correct figure is 0.497 mm, two and a half times the rule.

What clearance does a 3D printed hinge need?

A free running fit, because hinges rotate continuously. That is 0.310 mm at 5 mm, 0.392 mm at 10 mm and 0.497 mm at 20 mm. Free rather than running because a printed bore has layer texture and a printed pin is rarely perfectly straight, and both eat into the gap.

What is the difference between a sliding fit and a running fit?

Intent. A sliding fit locates and assembles by hand with no felt play but is not meant to move in service. A running fit is meant to move: 240 against 160, so about 50% more clearance, enough for slow rotation or occasional sliding without binding.

Can I use a press fit in PLA?

Yes, with care. PLA is flagged brittle, so hoop stress goes into crack initiation instead of relaxation. Keep the wall generous, chamfer the lead-in, and above roughly 25 mm expect split bosses. A transition fit plus adhesive is often the better engineering answer.

How much interference should a press fit have?

Start with the class figure, −0.049 mm at 10 mm and −0.062 mm at 20 mm, then verify on a coupon. Interference figures beyond that are load and geometry dependent and this model does not supply them: TBC for your specific joint.