Can You 3D Print Gears?

Yes. The harder question is what the gear drives, because that decides whether printing it is clever or a waste of your money.

A gear is not really a shape — it is a pair of shapes that have to agree with each other. Get that agreement right and FDM hands you a working drive for a few grams of plastic. Get it wrong and you get a gear-shaped part that binds, whines and chews itself flat. The geometry is in the gears and mechanisms category of the print library.

Where printed gears genuinely work

• Low-torque, low-speed drives. Printer and plotter parts, model mechanisms, hand-cranked movements, display pieces.
• Jigs and indexing — anything that turns occasionally rather than continuously and hard.
• Proving a ratio before cutting metal. Print the train, turn it by hand, confirm the arithmetic and the packaging, then commit to the machined version. One of the best uses of the process there is.
• Obsolete gears nothing else sells. When the machine is out of production and the alternative is scrapping it, a plastic gear that runs gently is a real answer — the ground industrial machine spares covers.

The common thread: intermittent duty, modest load, and a failure that costs an afternoon rather than a finger.

Module and teeth

Module (m) is the size unit of a metric gear: pitch diameter divided by number of teeth. It is not a preference — both gears in a mesh must share it, or the teeth do not correspond. Two gears at module 2 mesh. A module 2 and a module 1.5 do not, whatever their tooth counts.

Once you have the module, the centre distance falls out of it:

a = m(z₁ + z₂) / 2

where z₁ and z₂ are the tooth counts. That line turns "I want a 3:1 reduction" into a hole position on a plate.

The FDM constraint is resolution. Under the ISO 54 standard proportions, tooth thickness at the pitch circle is πm/2 and whole tooth depth is 2.25 m. At module 1 that is a tooth about 1.57 mm thick and 2.25 mm deep — roughly four extrusion widths from a 0.4 mm nozzle, which works. Drop to module 0.5 and the tooth is 0.79 mm thick: two perimeters back to back with nothing between them. Module 1 is a sensible floor; 1.5 to 2 is comfortable. Below that, the geometry is smaller than the tool making it.

Why the tooth profile matters

Gear teeth are not arbitrary bumps. An involute profile satisfies the law of gearing: the common normal at the contact point always passes through the pitch point, so the ratio stays constant through the whole mesh. Contact rolls at the pitch point and slides increasingly either side of it — but the ratio never fluctuates.

Invent your own tooth shape and you lose that. The action becomes non-conjugate, the output surges and stalls within every tooth, teeth interfere at the roots, and load concentrates where it was never meant to. Use a generator that produces a real involute; do not sketch teeth by eye.

The honest limits

• Sustained torque and creep. Plastic under constant load slowly deforms and does not recover, even below its breaking stress. A tooth held hard for months quietly loses its shape.
• Heat. Meshing teeth generate heat by sliding friction, and plastics conduct heat poorly — so it stays at the flank instead of leaving through the gear. As local temperature climbs towards the material's glass transition, the flank softens, wear accelerates and it runs away with itself. This, not tooth breakage, is what usually kills a continuously-running plastic gear.
• Layer lines are the wear surface. A printed flank is a stack of ridges, not a ground finish.
• Anisotropy. Print a spur gear flat, axis vertical, and the tooth load runs in the plane where FDM is both strongest and most accurate. The trouble starts when geometry forces the axis horizontal — a worm, a bevel, a gear on a long shaft. Then every tooth is a stack of layers loaded across the weld, the weak axis. See how strong printed parts are.
• Backlash is not optional. At nominal, a pair binds — holes shrink and posts grow. Design the clearance in deliberately, as in the tolerances and fit guide.

We cannot promise a printed gear will match cut steel, because it will not. Where there is real power, real duty and real consequences, it is a poor substitute.

Material

PLA is fine for a display piece or proving a ratio — it softens in modest heat and is brittle, so not for something that runs. PETG is the sensible default for a gear that actually turns: tougher, more forgiving, better under creep. Nylon is the real engineering choice — naturally slippery, tough, wear-resistant. Same logic as threads in printed parts: the material decides whether the feature survives being used.

Print-in-place mechanisms and clearance

Hinges, drag chains, universal joints and springs can come off the bed already assembled, because the printer lays plastic across a gap it never bonds. The trick is the gap: around 0.3 to 0.4 mm between moving surfaces. Too tight and it fuses solid; too loose and it rattles. Test it on a small piece before committing to a large part.

Herringbone: where printing genuinely wins

A herringbone gear is two helical gears of opposite hand on one body. Each helix generates axial thrust and the two cancel, so it self-centres, needs no thrust bearing, and — with more than one tooth pair always in contact — runs smoother and quieter than a spur gear.

Cutting one is awkward and expensive: it needs a specialist gear shaper or 5-axis milling, which is why herringbones are reserved for special applications in metal. To a 3D printer, a herringbone costs exactly the same as a spur gear — the difficulty that makes it expensive to machine does not exist when you are stacking layers.

That is the real case for printing gears. Not that it beats steel, but that printing is indifferent to a complexity machining charges dearly for.

When to have it machined instead

Plainly, so you do not waste a print:

• Continuous running under load. Have it machined.
• Anything transmitting real power — a gearbox, a drive train, a winch. Have it machined.
• Safety-critical or lifting. Metal, machined, no discussion.
• Fine pitch below about module 1. FDM cannot resolve the tooth. Machined, or a bought stock gear.
• Precision indexing or a rated ratio. Printed plastic moves with temperature and humidity. Machined.

If your gear is on that list we will say so rather than take the order.

The practical route

Tell us the module, the tooth counts, the centre distance and — most usefully — what the gear drives and how often. A one-off to prove a mechanism before you spend on metal is prototype printing, and exactly the right call. Browse the print library, and see how pricing works.