Assembled pin joint with clevis: separate lugs and pin

The most reliable joint you can print is the one you don't print as a single piece. A U-shaped clevis, a tab that fits between its two arms, and a pin that runs through them: three parts that reach the bed separately and assemble by hand. It sounds more laborious than leaving an in-place pivot —print-in-place— with its clearance already formed, and it is. In exchange, the surface the axle turns on isn't built by a nozzle welding overhangs tenth by tenth; you set it yourself when you choose the pin and the hole's fit. That control is the difference between a joint that turns clean and lasts for years, and one that comes out rough, oval, and with play you don't control. Let's see how to dimension it so the kinematics are exactly one degree of freedom and don't end up being zero (seized) or three (with play in every direction).

One single degree of freedom, and why the clevis guarantees it

The clevis —the classic clevis— is two parallel lugs with their holes facing each other and coaxial. The tab on the mating part slides between them, with its own hole aligned on the same axis, and the pin runs through all three. What you have then is a pure revolute joint: the pin defines an axis, and the only freedom left is rotation about it. No translation, no pitch, no lateral sway. That purity is no accident: it comes from two separated supports on the same axis, not from one alone.

Compare it with a single lug and a cantilevered pin. There the axle is held at a single point, and any lateral load makes it tilt: the pin works as a lever with the hole as fulcrum, the plastic yields, and the degree of freedom you wanted clean gets contaminated with play in every direction. The double lug closes that off. It splits the reaction between two supports on either side of the tab, so the load reaches the pin in double shear instead of single, and with a much smaller bending moment: the pin still bends between the lugs, but the span between supports is short and the reaction is shared, so the moment that used to make the axle pitch is nearly balanced by the pair of reactions. More load capacity, less deformation, and a rotation that stays rotation and nothing else.

Why the assembled joint beats print-in-place on the surface that matters

A print-in-place pivot is born with its clearance already formed between axle and hole, with no later assembly. It's convenient, but the contact surface that actually turns was built by the printer layer by layer, spanning an overhang: the underside of any horizontal hole or axle comes out hanging over air, stepped and sagging, and the downward-facing wall is always the worst of the lot. That roughness is precisely what rubs as it turns.

By separating the parts you take that compromised face out of the path. The pin can be a real metal axle —a ground steel rod, a screw, a hardened steel wire, a repurposed drill bit used as an axle—, and then the face that rubs against the hole is polished steel, not stepped plastic. Steel lowers friction and, above all, barely wears, so the friction torque stays stable over time instead of climbing as the plastic gradually files itself away. And since you control the axle's true diameter to hundredths of a millimeter, you can push the fit to fine clearances that a print-in-place pivot can never reach: with an axle whose diameter you know, the only error left to correct is the hole's.

Orient the holes and recover the true diameter

The lug's hole has two possible orientations, and it's worth separating two decision axes that don't move together: hole quality and resistance to delamination.

For hole quality, vertical wins. With its axis perpendicular to the bed, each layer traces a circle in the plane and the contour comes out round and predictable; the cylindrical wall is vertical, so there's barely any layer stepping on it. What it does have is a narrowing from extrusion-width compensation and a small neck from elephant's foot at the mouth. The horizontal hole, with its axis parallel to the bed, puts the overhang right in the turning surface: the upper half prints over air and comes out oval and sagging, wider than it is tall. For roundness, vertical is almost always better.

The second axis is which way the pin's load pulls. The layers should sit so the stress doesn't tend to separate them, because a hole whose layer line coincides with the plane of maximum tension delaminates sooner than a well-oriented one. There the orientation that suits you depends on the load vector, not on hole quality, and that's why the two decisions can pull in different directions. That logic —orienting the layers according to where the motion pushes— is what Layer orientation for motion develops, and here it marks the difference between a clevis that holds and one that delaminates through the hole at the first serious load.

Whatever the orientation, the step that really rescues the fit is recovering the diameter after printing. A printed hole almost never measures what you drew —it's born undersized and, depending on orientation, oval—, so run a reamer, a drill bit of the exact diameter, or the pin itself turning with a little abrasive through it until it's round and to size. It's the only way for the hole to be truly cylindrical and coaxial with the one in the other lug, which is the condition for the pin to enter straight and the joint not to bind. Measure on the part, not on the model, and dimension against that measured number. It's the same discipline as Tolerances for moving parts, applied to a hole you're also going to calibrate by hand.

The clearance decides whether it turns free, brakes, or seizes

The hole and the pin aren't one dimension each: they're a pair, and the only thing that matters is the gap left between them. Reason about it per side —on the radius— and convert to diameter only at the end: the diametral difference is twice the per-side clearance, and confusing the two is the error that ruins the most joints. With the pin at its known true diameter, you put all the clearance in by opening up the hole.

The first row is the free-turn fit on a ground axle: 0.10–0.20 mm per side is enough for the link to swing on its own with no appreciable torque. Don't go above that thinking it'll "turn looser"; at 0.25 mm per side on a Ø3 mm pin you already have 0.50 mm of diametral gap, and the link doesn't turn clean, it rattles —the very play the double lug exists to fight.

The third row, the asymmetric one, solves a real problem: a pin that turns free in both lugs tends to back out on its own and spin like a loose screw. Press-fit it into one of the lugs —or into the tab— with a small interference, on the order of 0.0–0.05 mm on the diameter against the measured hole, and leave the turning clearance only in the other. That way the joint has its clean degree of freedom at a single interface, and the pin is captured with no need for a head, ring, or clip.

If you want real friction —a joint that stays where you leave it—, closing the clearance to zero isn't enough: at 0.05 mm per side there's still a gap and the axle turns free. The brake demands press, interference between pin and hole, and letting the friction act as the stop. But there the material comes in: plastic turning against plastic under press heats up, files itself, and, over time, loosens. It's one of the failure modes in the next section.

The dimensions the radial fit doesn't see: bearing length and axial clearance

The gap between pin and hole isn't the only thing that decides a clean degree of freedom. Two more dimensions, easy to forget because they don't show up in the radial fit, decide as much as it does.

The first is the bearing length: the thickness of each lug, which is the length over which the hole wraps the pin. A lug too thin over a Ø3 mm pin gives a low length-to-diameter ratio, and then the pin pitches inside the hole again despite the double support, because each individual hole holds it over too short a slice. Give each lug a thickness on the order of its hole diameter or more, so the support is genuinely a support and not an edge.

The second is the axial clearance between the tab and the inner faces of the lugs. If the tab fits tight between the two arms, its faces rub against the inner walls and the joint seizes from lateral friction, even if the radial fit is perfect. Leave a few tenths of clearance on each side —on the order of 0.2–0.3 mm— so the tab turns without scraping the lugs. It's the same idea of putting all the gap where it belongs, carried to the pin's axis instead of its radius.

When to choose this joint, and the three failure modes that ruin it

The clevis pin joint is the right choice when the articulation carries load and has to last or be precise: links of a mechanism, articulated arms, connections you want to be able to take apart later by simply pulling the pin. Against a print-in-place pivot, which shines in the disposable and undemanding, this one earns its keep the moment there's force, turns, or years involved. And it has a maintenance advantage worth not overlooking: if the hole wears, you pull the pin, replace the cheap part, and carry on; nothing is welded together.

Knowing the kinematics and the fit, the three failure modes can be anticipated and designed out.

Hole ovalization under load. Plastic flows slowly under sustained load (creep), and the pin–hole contact is a narrow line that concentrates the pressure. Over time that line yields and the round hole turns oval: play appears where there used to be a fit, and the rotation starts to have slop. The defense is to lower the contact pressure —a larger-diameter pin, more wall around the hole, perimeters instead of infill in that zone— or, better, to embed a bushing in the lug, which offers a turning surface that doesn't flow. At this scale, that bushing is usually a plain bronze sleeve or an insert, not a ball bearing: the smallest common one already has a Ø10 mm outer diameter, so the bearing only comes into play above a certain joint diameter.

Failure of a single lug. The point of the double lug is to split the load into double shear; if one of the two is weaker —less wall, a poorly welded layer, the pin not bearing equally on both for lack of coaxiality—, it takes the whole reaction and breaks. Print the two lugs identical and well supported, make sure the pin enters straight in both —hence the insistence on calibrating both holes to the same axis— and give them a generous root where they join the body.

Fast plastic-on-plastic wear. If pin and hole are both plastic and the joint turns a lot, the pair files itself away: the friction torque rises, swarf appears, the clearance grows, and the fine fit you calibrated evaporates in a few hours of use. This is the case where a metal pin or an embedded bushing stop being a luxury and become the solution: a hard surface against a soft one lasts far longer than two soft ones rubbing, because plastic wears very little against polished steel. With one caveat: steel isn't immune, it just wears the plastic little; if abrasive grit gets between the two, the axle's hardness turns against you and files the hole faster. How to house that bushing or metal axle so it sits firm and flush is covered in Embedded hardware: magnets, bearings, and inserts.