2020 aluminium extrusion and V-slot: the channel and its T-nuts

You build a printer, a CNC, or a workshop shelf on 2020 aluminium extrusion, and sooner or later you need to fasten something to that frame: a spool holder, an end cap, a cable guide, a bracket to stiffen a corner. You look at the channel, see that it measures 6 mm, and model a 6 mm tongue to fit — and it won't go in. Or it goes in only under a hammer and cracks the part. The extrusion looks like the simplest thing in the world — an aluminium bar with four channels — but the printed fit has a catch, and that catch is always the same FDM bias that adds material to everything you pull off the bed.

This article is about that fit: what the 2020 channel really measures, how the T-nut works, and how much you have to slim a printed tongue for it to slide in without forcing.

What 2020 is and why it's everywhere

2020 is an extruded aluminium profile with a 20 × 20 mm cross-section and a channel running the length of each of its four faces. It's the backbone of open-frame 3D printers, home CNCs, laser cutters, and half the hobby workshops, because it solves the most mundane structural problem — joining bars at right angles and being able to move the joints — with no welding and no drilling.

The channel is what makes it useful. It's a T-shaped slot: a narrow mouth at the surface that opens into a wider inner chamber. Through the mouth you feed the head of a bolt or a special nut; the chamber traps it and stops it pulling back out through the mouth. That's the T-slot of industrial extrusion.

The V-slot variant adds a 45° V-chamfer to the edges of that same channel's mouth. That bevel isn't decorative: it's a track for POM wheels with a V-groove, so the extrusion itself becomes a linear guide. The gantry of a light CNC, or the moving carriage of a cutter, rides on the V-slot with no rail beyond the profile itself.

The 6 mm channel and the T-nut

The most widespread family — the one on almost every printer and CNC kit with a 20 mm frame — uses a 6 mm slot mouth and M5 bolts. The T-nut is a metal part just narrow enough to slip through the 6 mm mouth — you turn it once it's through, or slide it in from the end of the extrusion. Once inside the chamber, its shoulders sit behind the lips of the mouth, and as you tighten the M5 the nut bites into the aluminium. It's a thread you can place anywhere along the extrusion without machining anything.

A printed part uses that same channel in two ways. The first is location: you give it a tongue or a lip that enters the mouth and holds it aligned and square while the bolt does the clamping. The second is retention: a clip that hooks behind the lips and holds by snap fit, with no bolt — handy for caps and cable pass-throughs.

From catalogue dimension to FDM clearance

This is where the catalogue number can no longer be used as-is. A tongue you want to fit into the 6 mm mouth is a positive feature, and positives in FDM grow. The bulk of the growth comes from the extrusion width: every wall lays its bead outward along the whole tongue, so the nominal dimension comes out oversized top to bottom. On top of that, at the base only, there's the squish of the first few layers — elephant's foot — which widens the foot of the tongue but not its height. A tongue modelled at 6.0 mm prints above 6 and won't seat.

The rule of thumb is reliable: holes shrink, and tongues, posts, and printed nuts grow. It almost always works in the direction that tightens the fit. The rule breaks down when the slicer applies hole compensation or horizontal expansion, or with horizontal holes that come out oval from bridging the roof, and there the direction of the bias depends on your profile, not on the geometry. But as a starting point it won't mislead you.

That's why you model the tongue narrower than the channel, subtracting two things: the clearance you want and the growth the machine is going to add. For a lip that only aligns and slides along the extrusion, you want a sliding clearance — around 0.15–0.20 mm per side — over the 6 mm mouth. Add the positive's growth, which usually runs between 0.10 and 0.20 mm per side, and you're taking roughly 0.3–0.4 mm off each face: that means modelling the tongue at around 5.3–5.4 mm nominal width, then confirming it with a test, because how much of the growth lands where depends on your printer. Always reason per side, not by the total dimension: there are two walls, each lays its bead into the gap, and the bias is counted on each surface separately. The concrete per-side number, for your machine and your material, is in Real printed clearances; don't guess it here.

The same thing happens, with the sign flipped, when your part prints a groove that receives the extrusion instead of a tongue that enters it: that inner slot shrinks, so you have to open it up over the aluminium's nominal dimension for the profile to slide in.

Brackets under load: orientation, material, and interlayer adhesion

A printed bracket joining two profiles at an angle doesn't work in clean tension: it works in bending. The load pulls on one arm and levers right at the corner, which is where an FDM part is weakest, because the interlayer bond is the weak plane. If you print the bracket on edge, with the layers running horizontally across the vertical arm, the part doesn't bend like metal: it delaminates, splitting cleanly along a layer line, almost always on the first firm tightening or the first knock.

The defence is to orient the part so the load runs with the layers, not across them. On a bracket loaded in its plane, that usually means printing it flat — the L lying on the bed — so the perimeters of each layer trace the corner continuously instead of the layers cutting across it. Back that up with generous fillet radii at the inner vertex — a sharp corner concentrates stress and gives the crack somewhere to start — and with plenty of perimeters, which on a part in bending do more than infill.

Then there's the material, which on a structural part matters as much as the orientation. PLA is stiff and easy to print, but it has two failings that show up late, and by surprise, in a frame. One is cold creep: under the bolt's sustained preload, PLA yields little by little and the joint loosens over the weeks, even if it never actually delaminates. The other is thermal softening: a PLA bracket near a heated bed, or inside an enclosure, loses stiffness as it warms. For any joint that has to hold a permanent load or work hot, reach for PETG, ABS/ASA, or a fibre-filled nylon; leave PLA for light brackets with no sustained torque.

Caps, inserts, and guides: orienting the mating face up

For parts whose whole job is to fit the channel — end caps, inserts that cover the slot, nut retainers — the mating face governs. The features that define that fit are the ones that have to come out to size, and in FDM the face that points up on the bed comes out cleaner and truer than the one below, which carries the squish of the first layer. Orient the part with the face that enters the channel pointing up, and let the coarser tolerances fall on the visible face or the bearing face, where they don't matter.

With retaining clips there's a tension to resolve. The hook that grabs behind the lips is an undercut: depending on how you orient it, it hangs as an overhang and calls for bridging or a little support so it doesn't droop. And the clip arm is a flexible cantilever that snaps at the first click if the layers cross its root; you want the lines running along the arm, not across it. Reconcile that orientation with the "mating face up" one before you start the print, because the two don't always call for the same thing, and sometimes you have to compromise on one.

The possibilities with this profile are extensive: brackets and corner reinforcements, holders for spools, fans, or boards, end caps that close the end face and hide the central bore, guides and cable pass-throughs that snap into an empty channel, and spacers that occupy a channel to keep something from moving. A printed lug that reproduces the mouth, the lips, and the chamber anchors your part in the channel without remodelling the slot profile by hand.

An honest warning about the V-slot: you can print a guide or a wheel that rides on the V-chamfer, but an FDM surface is rough and won't run as smoothly or last as long as a machined POM wheel. For a light, low-use guide it'll do; for a motion axis you intend to rely on, a manufactured wheel is the better choice.

Once you've settled on the tongue and its clearance, the next step is to pin down the exact number for your printer in Real printed clearances, and if you're going to bolt for real against the frame, go over the patterns and threads in Metric fasteners: M3/M4/M5 patterns, bosses and inserts. And if you're mounting electronics on the same frame, DIN rail: the 35 mm rail and its clip covers the other big fastening standard you'll meet right next to the extrusion.