Linear guide variants: V-groove, telescoping tubes, slides, and crossed rollers

The rail with its sliding carriage is the default linear guide, but it isn't the only one and not always the best. The moment you ask for long travel, tolerance to manufacturing error, low friction, or stiffness in several directions at once, the plain rail falls short and a family of alternatives appears, each one solving a different problem at the cost of complicating another. It pays to know them all before you size anything, because the decision of which type of guide to use weighs more than any clearance you tune afterward: a badly chosen slide isn't fixed with tolerances, it's fixed by replacing it. This article is the menu, with the physics that explains why each variant does what it does.

V-groove: trading rubbing for rolling

The V-groove guide replaces sliding with rolling, and that single change decides almost everything. A wheel with a V cut into its rim rolls on a rail that is also V-shaped, so the two angled faces of the wheel bear against the two faces of the rail. The geometry does two things, but neither for free. First, it self-aligns, provided there is load or preload keeping the wheel seated in the channel: once the two faces of the V are bearing, any lateral displacement meets a face that pushes it back toward center, because the wedge contact turns that displacement into a component that corrects it. Without a force holding it in the channel, a single wheel does not self-center: it lifts out. That's why the self-alignment lives in the mounting—wheels on both sides, opposing Vs, or a preload—not in the profile by itself.

The tolerance to manufacturing error comes out of that same condition, and it's worth not confusing where it comes from. The V does not absorb, by some magic of geometry, the fact that the rail isn't straight or that the spacing between wheels isn't exact: with both wheels mounted rigidly, an incorrect spacing either binds or leaves play, plain and simple. What absorbs the misalignment is the degree of freedom you give the mounting—one wheel on an eccentric or flexible mount that moves toward or away from the rail until it bears properly on both faces. The wedge spreads that adjustment across the angled faces instead of fighting it; but what tolerates the error is the mount that gives, not the profile.

The second thing the V does, and the most important one for plastic, is to swap sliding friction for rolling friction. A surface sliding against another drags the whole time; a rolling wheel only has to overcome the internal resistance of its bearing, which is far lower. Here is the condition that makes or breaks a printed V-groove guide: the wheel has to actually roll. A printed V-wheel turning on a printed axle, plastic against plastic, is not clean rolling: it's a bad plain bearing, one that binds and wears. The V variant only pays off if you mount wheels with a real bearing—a pair of ball bearings pressed into each wheel—and then the printed body is just the support that positions them. That logic of when plastic should give way to a metal part is covered by Embedded hardware: magnets, bearings, and inserts; in a V-groove guide it isn't optional, it's what separates it from a rail that rubs.

Telescoping tubes: long travel, short at rest

A telescope solves a problem the rail doesn't solve well: providing long travel while taking up little space at rest. One tube slides inside another and, retracted, measures little more than the longest section; extended, it adds the lengths together. The kinematics is almost trivial—axial sliding—so all the difficulty falls on the fit between tubes and on the geometry of the overlap. Almost trivial, because a round tube inside another has one extra degree of freedom: the inner one rotates inside the outer. If the orientation of the tip matters, you need an anti-rotation guide—a polygonal profile or a key—; with a smooth circular section, the tip rotates freely and nothing stops it.

The gap between the inner and outer tube sets both the play and the smoothness at once, and they are opposing goals. Tight, it doesn't nod, but it rubs and can bind; loose, it slides smoothly, but the inner tube wobbles inside the outer. The right clearance is that of a clean sliding fit, the same logic as Tolerances for moving parts. Remember that a printed tube tends to come out with its inside a little narrower than drawn—the bead of the inner perimeter eats into the diameter—while the outside depends on your flow and shrinkage calibration, so the real gap rarely matches what's on screen: confirm it with a fit test and open it on purpose if you need to.

But the parameter that really governs a telescope isn't the clearance, it's the overlap at maximum extension: how much inner tube is still inside the outer when you've pulled it all the way out. A lever effect governs here, and it's worth seeing it as two arms, not as a clearance that grows large. The inner tube nods inside the outer by a maximum angle that the clearance allows: roughly the clearance divided by the overlap length. That angle translates into displacement at the tip when you multiply it by the overhang, the part that has come outside. On extending, both things happen at once: the overlap shrinks—so the allowed angle grows—and the overhang increases—so each degree weighs more at the tip. The error at the tip is on the order of the clearance times the ratio of overhang to overlap, and that's why a telescope that seemed firm retracted turns floppy and shaky fully extended: it isn't that the clearance amplifies, it's pure lever geometry with both arms working against you. The practical rule isn't stated against "the tube length," which defines nothing stable, but against the overhang: keep the overlap generous relative to the length you pull out as overhang—and quite a bit more if it will carry side load—and add stops that prevent going past that point, because the failure gives no warning: the tube simply drops out.

Drawer slides and crossed rollers: full access and the greatest stiffness

When you want full extension—pulling the entire carriage clear of the rail, like a drawer that opens all the way—the telescope doesn't reach and you enter slide territory. A commercial drawer slide chains two or three rail stages with ball cages between them, so each stage contributes travel and the sum lets the carriage come all the way out. Reproducing that in print, with balls, is hardly realistic; the printable version is usually rail-in-rail, one profile sliding inside another on friction, with an end-of-travel stop that keeps it from coming out. You give up the smoothness of the balls and accept that this is plastic sliding against plastic—with its rubbing and its wear—in exchange for simplicity and being able to open all the way. For something that opens and closes now and then, it's enough; for heavy cycling, the wear will eventually take its toll.

At the other extreme of demand are crossed rollers. The principle is elegant: two rows of cylindrical rollers alternated, each turned 90° from the previous one and bearing on crossed V-tracks, so each roller takes the load on one of the two diagonals at ±45° from the plane. Between them, the assembly resists in all four directions precisely because no track is aligned with the pure vertical or pure lateral. Since each roller is a line of contact, not a point, they spread the load over far more area than balls do and the guide comes out stiff and precise in every direction at once, without the play a plain rail has on its secondary axes. It is the guide with the greatest stiffness and precision under multidirectional load. And it is, of the whole family, the one that makes the least sense to print in solid plastic: the stiffness and precision that justify it come from ground metal rollers running on hard tracks. What you can print is the body that houses and positions real rollers; expecting the roller itself to be PLA is wasting precisely what sets this variant apart.

What FDM imposes on all of them

Beneath the choice there's a layer of process physics, and the first rule rules over the rest: orient the part so the sliding axis runs along the bed, not vertical. The reason isn't the overhang—a prismatic profile extruded in Z hangs nothing, its side faces are vertical and print layer on layer without trouble. The reason is the direction of the layer stepping. Print the rail standing up and every layer change leaves its step crossing the working surface, the one that will roll or slide; you turn the track into a file. There's a second reason, anisotropy: standing up, the joint between layers ends up perpendicular to the stress and the part is weaker right where you load it most. Laid flat along the bed, the direction of movement runs parallel to the beads and the contact surface comes out continuous instead of serrated.

That said, the "smooth printed surface" criterion only bites where the contact surface is printed plastic: the smooth rail-carriage, the rail-in-rail, the telescope. In the variants with an embedded bearing or metal roller, the rolling track is no longer plastic—it's the rail or the roller—so there the printed smoothness of the track stops being the criterion; what matters is that the housing positions the rolling element well. And from there comes the second common rule: knowing when plastic is not the material. In the variants with rollers or balls—V-groove with bearing, crossed rollers, ball slides—the rolling element must be genuine metal; the printed body is the housing that positions it, not the contact surface. Asking a PLA track to bear the concentrated load of a roller is condemning it to creep and premature wear. The boundary between what you print and what you embed is drawn by Embedded hardware: magnets, bearings, and inserts.

The failure modes they share

Five failures run through the whole family, and recognizing them before printing saves iterations. The first is nodding from insufficient overlap, native to the telescope but present in any short guide: when the bearing arm falls short against the overhang, an innocent clearance in the contact translates into a large angular play at the tip. You fight it with generous overlap and stops, not with tighter clearances—tightening to kill the nodding is what leads to the second failure.

The second is binding from lack of parallelism. A linear guide demands that its two bearing ways be parallel; if they aren't—because the part warped as it cooled, or because two separately mounted rails didn't end up aligned—the carriage moves a few tenths and jams, because you're asking it to travel along two straight lines that aren't. The defense is one of design and of printing: integrate the two ways into a single part whenever you can, so their parallelism is guaranteed by the model and not by the assembly, and watch for warping in long, thin parts.

The third is accumulated play. Each interface contributes its clearance, and in a guide with several stages or several wheels those clearances add up and the carriage's final position turns imprecise and soft. This is where preload—the eccentric of the V, a stage with a light interference—earns its place: removing the play at the source is more reliable than chasing ever finer tolerances at each interface.

The fourth is stick-slip, native to the variants that slide plastic against plastic—rail-carriage, rail-in-rail, telescope. PLA has a static friction that is high and markedly greater than its dynamic friction, so the carriage sticks, breaks the start abruptly, and jumps; moving a printed slide by hand feels jerky, not continuous. It's the first thing that gives away a plastic slide, and it's mitigated without touching the clearance: a dry PTFE lubricant, swapping the PLA for PETG on one of the two faces, or a lower-friction sacrificial surface at the contact.

The fifth is wear of the plastic at the contact surfaces under load, the slow failure, and it's worth separating two mechanisms that get confused. One is wear from cycling: wherever two plastic surfaces slide under load, the material polishes, flows, and ends up creating the very play you took such care to avoid. The other is creep under sustained static load: a carriage holding weight while parked for days slowly deforms the clearance even without a single cycle, because loaded PLA flows slowly even at room temperature. They are two distinct mechanisms with two distinct defenses, but they point to the same conclusion: rolling beats sliding and metal beats plastic at the contact point not because of the friction on day one, but because of the dimension a month later. If a guide is going to work hard and loaded, that's the question that decides its service life, and the answer usually lies in getting the contact out of the plastic—bearings, rollers, a sacrificial surface—rather than in tuning a clearance that wear is going to eat anyway.

With the variant chosen and its orientation settled—the layer criterion is developed by Layer orientation for motion—the next step is to size the specific contact you've picked: the gap between tubes, the carriage clearance, the seat of each bearing. And there the criterion for how much gap to leave is always the same: Tolerances for moving parts.