Demountable tab and slot: boxes and panels that assemble by hand

A screwless box isn't magic: it's a tab that drops into a slot, plus a little something to keep it from sliding back out on its own. Tab-and-slot is the flat joint par excellence, the one that assembles a chassis, a jig, or a paneled box by hand with no tools, and that — unlike a permanent snap — you actually want to be able to reopen. But that virtue is also its trap: it doesn't clamp on its own. Design it as though it did, and you end up with a stack of panels that fit perfectly on the bench and rattle the moment you pick them up. Everything hinges on how you constrain the degrees of freedom, on a notch almost nobody adds, and on the two tenths of a millimeter that separate "goes in by hand" from "won't go in."

Which degrees of freedom it constrains — and which it doesn't

Start with the kinematics, because that's where the misunderstanding starts. A tab seated in the slot of the neighboring panel constrains, above all, the panel spacing: it stops the panels from separating in the direction perpendicular to the panel plane and holds them in the plane of the joint. What a single tab doesn't constrain is sliding along the slot — which is exactly what a slot, by definition, lets run — nor the relative rotation of the panel: the tab can come straight back out the way it went in. That's why a real joint is never one tab: it's several, distributed along the edge, so that together they constrain the rotation and leave the panel located. Two spaced tabs are already enough, kinematically, to constrain the in-plane rotation and locate the edge; add more for robustness, not because the kinematics demands it — and be careful: overconstraining with real FDM tolerances is exactly what produces an impossible assembly.

And here's the difference that decides everything else. A dovetail, when you push it in, self-tightens: the wedge geometry turns any attempt to pull apart into a force that closes the joint harder. A straight tab does none of that. It goes in and rests there, but nothing in its shape keeps it from backing out; the only force holding it is the friction of the fit, and the friction of a sliding fit is, by definition, low. A tab-and-slot with nothing else is a locator, not a latch. It locates the panel in the plane and lets it slide out along the insertion axis. If that's all you need, it's perfect for aligning; if you expect it to hold a box upright, you're missing half the joint.

Add the retention a dovetail provides for free

Because the shape doesn't self-tighten, you have to add the retention by hand, and there are two clear solutions. The first is a retention notch on the tab itself: a small recess or boss that, once the tab passes through the slot, catches against the far face of the panel and stops it from backing out. It works like a small snap: the tab flexes just enough for the notch to clear the panel thickness, then straightens on the other side, hooked on the edge of the slot. Press the notch and pull it out and it's demountable, but it no longer comes loose on its own. If you go this way, everything you know about a cantilever tab applies: the bending lives at the root, it wants a fillet radius and a sensible travel. And a warning: for a joint that's made and unmade many times, the snap-notch fatigues and loses force with use — in PLA it loses retention force or breaks at the root after enough assemblies; for many cycles, the pin described in the next paragraph lasts longer.

The second solution, sturdier for structure, is a transverse pin or wedge: a through-hole in the part of the tab that pokes out the other side of the slot, into which you drive a perpendicular pin — or a tapered key — that locks the joint. The tab can no longer back out because the pin won't let it pass the slot again. It's the version of "screwless" with no catch: the torque is carried by a separate part, trivial to print, and to disassemble you just pull the pin. It doesn't fatigue, so it survives the cycles the notch can't. The wedge has an extra advantage: being tapered, it draws its own joint tight against the stop as you drive it in, eliminating the local play of that connection. That's not the same as recovering the play accumulated in series along a long panel — but it does close the rattle of each fit, one by one. Without a notch or a pin you don't have a stable joint; you have an assembly that holds as long as nobody touches it.

Orient the tab: the root works in bending

The tab is a short cantilever, and in service its weak point is the root, the start where it grows out of the panel: when you push a panel, the load enters through the face of the tab and bends it at its base, putting the outer fiber of the root in tension. In an FDM part, that root is the vulnerable spot. If you print the panel so the layer plane runs transverse at the root — the stacked layers perpendicular to the load — the bending tension falls across the layer lines, and the tab doesn't break across the plastic — it delaminates, peeling cleanly along the layer line where it grows. Orient the panel so the beads run lengthwise along the tab, so the fiber follows the material and the bending doesn't pry apart the weld between layers. It's the same logic that governs any printed flexing arm, and it's worth keeping in mind from the moment you decide how to lay the panel on the bed.

The good news is that the geometry cooperates. A through slot — an opening that crosses the panel from side to side — is about the easiest thing to print: vertical walls, no overhang, repeatable tolerance. Whenever you can, make the slot a through slot. A blind slot, one that doesn't break out the other face, forces you to close the bottom with an overhanging roof, and that roof comes out sagging and rough right on the face where the tab is supposed to stop and locate. If the function doesn't require a closed bottom, don't put one in: a through slot gives you a better fit, a better finish, and on top of that lets the tab poke out so you can drive the pin in.

Add a lead-in chamfer: a countersink at the mouth of the slot, or a chamfered tip on the tab, is the most effective way to make it "go in by hand" without touching the fit. With six tight fits that all have to engage at once, the chamfer is what makes simultaneous assembly viable without chipping edges: it guides each tab into place in the first few millimeters, before the nominal clearance starts to matter.

The tab-to-slot clearance: the deciding parameter

As with any printed fit, the number that decides whether the thing works isn't the nominal dimension but the real clearance between tab and slot, and here you want it sliding: on the order of 0.2 mm per side in PLA for a hand-assembled joint that gets made and unmade many times. Work it per side, as always: each wall lays its bead into the gap independently; the slot comes out narrower than drawn and the tab thicker, and the two shifts consume the clearance before you change a thing. Why the hole closes up and the part grows wider is covered in detail in Tolerances for moving parts; here it's enough to calculate on the measured part, not on the nominal.

The material isn't neutral. PETG has more elephant's foot, more oozing, and worse repeatability on vertical walls than PLA; allow 0.05 to 0.1 mm more clearance per side — or tune the flow — otherwise you risk parts that won't go in, especially in a joint worn by repeated assembly and disassembly. The range is bounded on both sides, and each bound is a failure mode. Too tight and it won't go in — or goes in forced, chipping the edge — which in a box of several tabs multiplies: with six tight fits, assembling it by hand becomes impossible. Too loose and the box wobbles: each joint has its own play, and the play of one flat joint doesn't cancel against the next — it accumulates. That's why the clearance of a tab-and-slot is more delicate than that of a loose pivot: you don't set it part by part, but by thinking about how many joints in series the structure will have and how much error you'll tolerate once they all add up.

When to choose it and what breaks it

Tab-and-slot is the joint for anything that has to come apart: boxes and chassis that travel disassembled, jigs that get reused, panels that open for maintenance. It's the screwless alternative to a bolted joint when you want quick assembly by hand and reversibility, and where a dovetail would be too rigid or too hard to insert. If the joint is permanent, another edge geometry gives you more; if it's demountable, this is the natural one. The comparison with the joint that does self-tighten — and why you sometimes prefer it — is in Printed joinery: dovetails and joints.

There are three failure modes, and it's worth naming them so you can design them out. The first is loosening from accumulated play: many loose joints in series add up their play and the whole structure wobbles, even though each joint on its own looks acceptable — it's the one that ruins big boxes most often, and you attack it with retention per joint, not with a tighter fit. The second is breakage at the base of the tab, almost always from delamination when the print orientation left the layer plane transverse to the root bending. And the third is the impossible assembly on large parts: shrinkage on cooling moves the dimensions, errors accumulate along a long panel, and a tolerance that was perfect on a 50 mm coupon leaves the tab unable to seat on a 300 mm edge. On large parts, give a little more clearance and let the retention — not the fit — do the work of keeping it from coming loose.