Compound gear train: stages in series

When you need to move something slowly and with a lot of torque from a fast, low-torque motor, the reduction you need won't fit in a single pair of gears. A 50

reduction at a sensible module would demand an enormous output gear, barely printable, its diameter overrunning the bed. The answer isn't to make the gear bigger: it's to chain stages. A compound train achieves huge ratios with sensibly sized gears, and the trick lies in how those stages multiply and in what accumulates as they chain.

3D

Stages multiply, they don't add

A compound train is a succession of gear pairs in series, with one defining detail: each intermediate shaft carries two gears locked together with different tooth counts. The large gear on a shaft receives motion from the previous stage; the small gear sharing that same shaft delivers it to the next. And here is the key to the whole family: the overall ratio is not the sum of each stage's ratios, it's their product.

If you follow the rotation, the reason is straightforward. If the first stage reduces 5

, the intermediate shaft turns five times slower than the input. The second stage, mounted on that already-slow shaft, reduces another 5:1: five times five, 25 at the output. Each stage operates on what the previous one delivers, not in parallel; that's why their factors chain together by multiplication. Two modest 6 stages give 36; three give 216. That's why a reduction that would be unworkable as a single pair becomes manageable when split across stages: each individual mesh keeps a small ratio, comfortably sized teeth, and gears that fit on the bed.

Locked gears versus idlers

Don't confuse a compound train with a simple train full of intermediate gears, because they serve opposite purposes. In a simple train, a gear meshing between two others — an idler, running free on its own shaft — passes the rotation along and reverses its direction, but doesn't change the magnitude of the overall ratio: mathematically, its tooth count enters the ratio and cancels out. Adding or removing an idler changes the output's direction of rotation, or bridges a center distance you otherwise couldn't reach. That's all it does: it doesn't reduce the ratio.

What does reduce the ratio is exactly what defines the compound train: the two gears of different size that share a shaft and turn as a single piece. That rigid coupling is the indispensable condition. If the large gear that receives and the small one that delivers didn't turn together — if one slipped relative to the other on the shaft — the next stage wouldn't receive the reduced rotation from the previous one, and the product would fall apart: you'd have two loose mechanisms instead of a train. That's why those two gears have to be locked together in rotation, whether fused into one piece or keyed onto a shaft that prevents any relative slip. The whole multiplication of ratios depends on that rigid union.

How it splits into printable parts

In FDM each stage wants to be printed flat, with the gear axes perpendicular to the bed. It's the orientation that produces the most faithful tooth profile — the flanks come out defined by each layer's contour, with no overhang stair-stepping — and the one that loads the tooth correctly: it works in bending within its plane, along the beads, not pulling on the weak bond between layers. Why this orientation wins is explained in full in Layer orientation for motion.

The crux of the design is how you build the locked pair on each intermediate shaft. There are two paths. The first is to print the two gears as a single piece of two diameters, a large gear and a small one, coaxial and fused into one body. It's the most robust option: there's no joint to fail, no play between the two, and the locking is perfect by construction. The second is to print separate gears and join them with an anti-rotation shaft — a hex or square profile, or a keyway — that prevents relative slip; you gain modularity and the ability to swap one gear without reprinting the pair, at the cost of introducing a joint you must size so it transmits torque without slipping.

Watch out for one subtlety of the fused piece: printed flat, the diameter step between the large gear and the small one leaves the top face of the larger gear overhanging where it extends beyond the smaller one. It's not a showstopper, but it's worth putting the smaller diameter on top or introducing a transition cone between the two so the step doesn't print as a messy overhang.

Beyond the choice of part, two dimensions are unforgiving. The module has to be consistent within each mesh: the two gears that mesh share a module or their teeth simply won't fit; between different stages you can change it. And the center distance of each pair has to be precise: it's the sum of the pitch radii of the two meshing gears, and an error there pushes the teeth too deep, where they bind, or too far apart, where they lose contact and the play grows. In a train, each intermediate shaft appears in two center distances at once (the one tying it to the previous stage and the one tying it to the next), so a poorly placed shaft ruins two meshes at once.

Backlash accumulates stage by stage

Backlash — that small gap between the flank of a tooth and the flank of the tooth space opposite it — is necessary: without it the teeth bind the moment thermal expansion, a print-tolerance shift, or a speck of plastic tightens the mesh. But it has a cost that comes to the fore in a compound train: it accumulates.

The kinematics make it clear. When you reverse the direction of rotation, each stage has to travel through its own backlash with nothing engaged before its teeth start pushing again on the other face. And since the stages are in series, that lost motion passes along and adds up across the chain. Referred to the output — which is what matters for positioning — each stage's backlash is reflected, amplified or attenuated according to the ratios that lie behind it: the first stage, multiplied before it reaches the end, is heavily attenuated as seen from the output, while the final stages, the low-speed ones, contribute nearly all of their play. The result is that the total backlash at the output grows with the number of stages. A four-stage train, however well each stage is calibrated, will have on reversal a perceptible play that a single stage never would. To move a load in one direction only, that doesn't matter; to position, for something that starts, stops, and returns, it's the difference between a precise mechanism and a loose one.

That's where the practical decisions come from. Calibrate the mesh of each stage — flank clearance is designed, not inherited, just like any other printed fit; it's covered in Tolerances for moving parts, where you'll also find the reasonable order of magnitude for your printer. If position accuracy matters, minimize the number of stages: spread the reduction across fewer pairs with somewhat larger ratios, rather than across many small pairs, because every stage you add contributes its share of play. And if you still need both the high ratio and the precision, turn to an anti-backlash arrangement — typically a gear split into two halves preloaded by a spring, each bearing against one flank of the tooth space — which eliminates that stage's backlash at the cost of friction and complexity.

Where it's used and how it fails

The compound train is the answer when you need a large reduction with sensibly sized gears. The textbook case is the motor reducer: dropping the speed sharply and raising the torque without a giant output gear, spreading the ratio across stages that fit on the bed. The same cascade of stages can also multiply speed instead of reducing it — it's what a mechanical clock's train does, carrying the slow rotation of the barrel up to the fast escape wheel — but the reason to chain stages is always the same: to achieve a large ratio without a disproportionate gear.

The three failures to watch for show up across the whole train. The accumulated backlash you've already seen: it produces play at the output and is fought by calibrating and simplifying. Tooth breakage appears where the torque is highest, and in a reducer that's the output stage: it turns slowly and carries a lot. The way to reinforce it in FDM is to increase the face width rather than the module, because width adds strength by working with the bond between beads, not against it, without enlarging the gear — raising the module thickens the tooth, yes, but it also inflates the diameter and forces you to redo that stage's center distance, exactly the size the train was meant to avoid; reinforce the output, not the input — the input runs fast but lightly loaded. And center misalignment binds: if an intermediate shaft doesn't sit where the sum of the pitch radii dictates, its mesh works under pressure, heats up, wears, and ends up locking a whole stage and stalling everything downstream of it. A train is worth no more than its worst-placed shaft.

Before chaining stages, validate each mesh on its own: the tooth clearance, the print orientation, and the center distance of a single pair are the foundation everything else is built on, and you'll find them in Tolerances for moving parts.