LEGO System: the 8 mm pitch and the press-fit clutch

You print a LEGO-compatible brick. It comes off the bed with clean studs and straight walls. You press it onto a genuine brick and one of two things happens: it won't go on, or it goes on under a hammer and splits a wall along a layer line. What you never get is the crisp click, the firm grip that holds without play — the thing that makes a LEGO a LEGO. The problem isn't your model or your nominal dimension: the grip in this system lives inside a band a few hundredths of a millimetre wide, so narrow that your printer's own process variation swamps it. This article explains why the LEGO standard is easy to draw and one of the hardest things to reproduce in FDM.

The 8 mm pitch and the stud grid

The whole system comes down to a single figure: 8 mm between stud centres. That's the grid. A 2×4 brick measures a touch under 16 by 32 mm in plan, because the outline is set back to leave play between neighbouring pieces; the studs, on the other hand, land exactly on the 8 mm grid. Layered onto that grid are the two dimensions that define the male half of the joint: the stud is 4.8 mm in diameter and stands 1.8 mm proud of the top face.

Look at the arithmetic of the gap. Between two neighbouring studs there is 8 − 4.8 = 3.2 mm of clear space, and that channel isn't empty by accident: it's where the walls and tubes of the piece above come down. In a brick two or more studs wide, what sits at the centre of each square of four studs is the inner tube, roughly 6.5 mm in outside diameter, and its centre lands on the diagonal of the grid, √(4²+4²) = 5.66 mm from each stud. Add the two radii that face each other — 2.4 mm of stud and 3.25 mm of tube, 5.65 mm — and you'll see the surfaces touch almost exactly: the inner clutch rides on that contact, an overlap of barely a few hundredths of a millimetre. At the edge studs, where there's no tube facing them, it's the inner wall that pinches the outer flank of the stud. Change the stud diameter by a tenth and you're already off that sum: you clamp where you shouldn't, or you stop rubbing where you ought to be gripping.

Heights: the brick, the plate and the exact third

The vertical dimension has its own clean logic. A brick is 9.6 mm tall — not counting the stud — and a plate is 3.2 mm, exactly one third of a brick. That's no factory coincidence: three stacked plates equal exactly one brick, which is why you can swap a brick for three plates anywhere in a model without introducing a height error. The 1.8 mm stud that sticks up doesn't add to the stacking height, because it sinks into the cavity of the piece above.

The clutch: a press-fit measured in hundredths

The grip — the clutch — isn't in the stud diameter on its own, but in the interference between the stud and the surfaces that pinch it as it seats. When you lower one brick onto another, each stud of the lower brick enters the cavity of the upper one and ends up clamped between the central tube and the inner walls of that cavity. The tube pushes outward, the wall pushes inward, and the stud is bitten between the two. That bite is the whole physics of the system: a press-fit, of the family described in Choosing the fit: clearance, transition, interference, but refined to a degree that metalwork almost never demands.

The real interference is on the order of a hundredth of a millimetre per side. Enough for friction to hold the weight of a tower of pieces, and for you to feel the snap on assembly and the tug on disassembly, but fine enough that the joint repeats thousands of times without wearing out or seizing. A few hundredths too many and the pieces won't come off; a few too few and they wobble: the useful margin is narrower than your printer's own spread. LEGO stays inside that band because it injects ABS into steel moulds held to microns. You have neither that material nor that mould.

Why the nominal dimensions come out too tight in FDM

Here the standard rebels against the printer, and it does so in the direction you already know from Real printed clearances: the FDM bias fattens the stud and shrinks the cavity, both at once and both against you. The stud is a male cylinder printed vertically, so the bead width and the perimeter overlap leave it fatter than its nominal 4.8 mm. The mating cavity is a female hole, so those same perimeters pinch it inward, and its mouth — the first layer resting on the bed — comes out narrower still, squashed by elephant's foot. Add the two sides together and, at nominal, the clutch is no longer a hundredth of interference: it's two or three tenths. That isn't a fine press-fit; it's an impossible one.

The result is always one of two failures with names of their own. Either the stud won't go in — a dead stop, the piece left half-seated — or it goes in by force and splits the wall along a layer line, because an FDM part under pressure doesn't yield the way injected ABS does: it opens between layers along its weakest plane. That's why PLA, rigid and brittle, is the worst material for a clutch: it splits without warning. ABS, ASA or PETG, with a little give, come closer in feel and stand up better to repeated assembly, though PETG creeps under load and can, over time, loosen a joint that went on tight.

You can't print the nominal dimension and expect a clutch. You have to shave the stud or open the cavity by a few tenths. You can do it in the model and calibrate it with a test coupon, but the fastest lever is the slicer's horizontal compensation — the XY compensation of contours and holes — which shrinks the outer contours and enlarges the holes independently, correcting the stud and the cavity each in its own direction without remeshing the CAD. It's the reasoning of Real printed clearances, only here you're chasing not a clearance but a tiny, controlled interference. A stud printed at 4.6–4.7 mm in diameter, or a cavity opened up by 0.1–0.15 mm, is the realistic starting point; the exact value comes from your machine.

Orientation: studs up

Orientation on the bed decides most of the joint's quality. Always print with the studs up, the way the piece is used. That way the studs come out as vertical cylinders of constant diameter — with the slight ripple of layer lines on the flank, not a staircase — and the mating cavity prints as a vertical hole, which is the clean case: narrow, but cylindrical and not collapsed.

The base pays the price. The first layer is squashed against the bed and fattens the low walls of the cavity right where the stud has to enter, so the mouth of the joint comes out tighter than its interior. That's elephant's foot, acting on the very surface you care about most, which is why it pays to switch on first-layer compensation or to chamfer the mouth of the cavity slightly so the stud finds its way in. Printing the piece on its side to spare yourself the texture on the stud flank is a cure worse than the disease: a cavity printed sideways comes out oval and unsupported inside, and its clutch is nothing like that of a vertical hole. The physics behind that squashing and that tight mouth is covered in Holes, pivots and first-layer squish.

Plates, tiles and adapters

With the pitch and the heights fixed, the rest of the catalogue is combinatorics. A plate is a brick one third as tall; a tile is a plate with no studs on its top face — no clutch on top, so it only grips underneath. Anything that respects the 8 mm grid and the 9.6 and 3.2 mm heights fits with everything else, which is the whole point of the system and the reason it's worth printing compatible pieces rather than inventing your own grid.

If you're going to design an adapter — a piece that's LEGO on one face and threaded hardware, a profile or any other system on the other — the LEGO side is always the one that sets the tolerances, because it's the one with the margin measured in hundredths. Calibrate that side first with a test coupon and leave the other for last. And if what you want is the other half of LEGO, the one with pins and cross axles, that has its own friction-fit physics and its own tenths: it's covered in LEGO Technic: holes, pins and cross axles.