Blog · 2026-07-05 · Gearing
Gearing is the trick that lets a motor the size of your thumb move a robot the size of a dog. Here's the complete picture: the speed–torque trade, the gear types, the losses, and the math.
Electric motors are naturally fast and weak: a small brushed motor happily spins at 6,000–20,000 RPM while producing torque measured in fractions of a kilogram-centimetre. Robots need the opposite — wheels turning at a few hundred RPM with real pushing force, arm joints moving slower still with much more. Gearing is the currency exchange between those two worlds, and understanding its exchange rate is one of the highest-leverage pieces of knowledge in robotics.
A gear ratio of N:1 means the input shaft turns N times for each output turn. Speed divides by N; torque multiplies by N (minus friction). The deep reason is conservation of energy — power in ≈ power out, and since power = torque × speed, squeezing one number inflates the other:
Output speed = Input speed ÷ N
Output torque = Input torque × N × efficiency
Concrete numbers: a motor spinning 6,000 RPM at 0.5 kg·cm behind a 30:1 gearbox at 90% efficiency delivers 200 RPM at 0.5 × 30 × 0.9 = 13.5 kg·cm. Twenty-seven times the torque, from the same motor, for the price of speed you didn't need anyway. That single transformation is why nearly every robot motor is a gearmotor.
Two corollaries worth internalizing. First, gearing never adds power — a 5 W motor is a 5 W motor behind any gearbox, which is why the Motor Sizing Calculator reports power as well as torque: if the power isn't there, no ratio will save you. Second, reflected inertia: the load the motor "feels" shrinks with the square of the ratio, which is why geared motors accelerate loads so much more crisply than direct drive.
For meshing gears, the ratio is simply the tooth count quotient: a 12-tooth pinion driving a 60-tooth gear gives 60 ÷ 12 = 5:1. (Diameters work too, since teeth are proportional to diameter at the same module.) Practical single-stage spur reductions top out around 5–6:1 — beyond that the pinion becomes fragile and the big gear becomes enormous — so larger ratios are built by stacking stages, whose ratios multiply: two 5:1 stages give 25:1, three give 125:1. Idler gears placed between two gears change rotation direction but not ratio — a fact that resolves a lot of beginner confusion about gear trains that "should" be different.
Total ratio = stage₁ × stage₂ × stage₃ … (efficiencies multiply too)
That last parenthesis matters: efficiencies compound. Three 95% spur stages deliver 0.95³ ≈ 86% overall. It's why very high-ratio gearmotors feel "stiffer" and waste more input power than their single-stage cousins.
Spur gears — straight-cut, cheap, efficient (92–95% per stage), and the noisiest of the family. What's inside most low-cost hobby gearboxes. Planetary gearboxes — a sun gear, orbiting planets and a ring gear sharing the load across multiple teeth: compact, strong, coaxial (output inline with motor), typically 80–90% efficient overall, and the standard for quality gearmotors. Worm gears — a screw driving a wheel, achieving huge ratios (10:1–100:1) in one stage with a property no other type offers: most worm sets can't be back-driven, so an arm or lift holds position with power off. The price is brutal efficiency, often 30–70%. Belts and pulleys (GT2 timing belts) — quiet, tolerant of imprecise shaft spacing, great for 1:1–5:1 transmission across a chassis at ~85–90%. Chain and sprocket — belts' heavy-duty sibling for high-torque outdoor builds.
| Type | Per-stage ratio | Efficiency | Signature strength |
|---|---|---|---|
| Spur | up to ~5:1 | 92–95%/stage | Cheap and simple |
| Planetary | 3–10:1/stage | 80–90% overall | Compact strength, coaxial |
| Worm | 10–100:1 | 30–70% | Self-locking under load |
| Belt (GT2) | 1–5:1 | 85–90% | Quiet, forgiving spacing |
| Chain | 1–6:1 | 90–95% | High torque, robust |
Backlash is the small free play between meshing teeth — the "clunk" when a gear train reverses. Drive wheels barely care. Robot arms and anything doing closed-loop positioning care a lot: backlash appears as a dead zone the controller can't see through, and it's a classic source of the limit-cycle oscillation you can reproduce in our PID Tuning Visualizer's badly-tuned presets. Remedies, in ascending cost: belts (near-zero backlash), quality planetary boxes, split anti-backlash gears, and harmonic drives at the exotic end.
"12 V, 1:75, no-load 120 RPM, rated torque 10 kg·cm" now decodes fluently: the internal motor spins ~9,000 RPM; 75:1 of reduction trades that down to 120 RPM while multiplying the motor's ~0.18 kg·cm into 10 kg·cm at the shaft (the gap from the ideal 13.5 reveals the gearbox's ~75% efficiency — you can audit a manufacturer this way). The same motor is sold as 1:30 (300 RPM, ~4 kg·cm) and 1:150 (60 RPM, ~20 kg·cm) variants: one motor, one gearbox family, a whole catalog.
Output-shaft limits: very high-ratio micro gearmotors can generate more torque than their own plastic gears or shaft can survive — a stalled N20 with a 1000:1 box strips itself. Check the listing's "maximum momentary torque" and keep stall events below it. Support your shafts: gears mesh correctly only at their designed centre distance; a wheel cantilevered far off an unsupported gearbox shaft levers the mesh apart and chews teeth. A single outboard bearing transforms drivetrain lifespan.
No — past the ratio your speed target implies, you're donating top speed and efficiency for torque you can't use (traction becomes the limit: wheels slip before the motor stalls). Match the ratio to the requirement; that's the entire job of the Gear Ratio Calculator.
Robots gear down (reduce speed) in almost every case. Gearing up — multiplying speed at the cost of torque — appears only in special cases like flywheel shooters.
Plastic (POM/nylon) is quiet, cheap and fine at low torque; metal survives shock loads and stall events. The hybrid pattern in good gearboxes — metal final stages, plastic early stages — puts strength exactly where torque is highest.
With the theory in hand, the practical question remains: which ratio for your robot? That's a short, worked process — covered next in How to Choose the Right Gear Ratio, or jump straight to the Gear Ratio Calculator with your motor's numbers.
Ratio math assumes the gears physically mesh, and that requires matching tooth geometry. Metric gears are specified by module (tooth size in mm of pitch diameter per tooth): two gears mesh only if their modules match, and their centre distance is (teeth₁ + teeth₂) × module ÷ 2. Hobby robotics lives around module 0.5 (small, precise, fragile — micro gearmotors), module 1 (the sweet spot for 3D-printed and small machined trains) and module 1.5–2 for torque-carrying stages. Imperial gears use diametral pitch instead — 32 DP, 20 DP — and the systems do not intermix. When 3D-printing gears, module 1 or larger prints reliably on hobby FDM machines; module 0.5 mostly produces decorative sadness. Belt equivalents are simpler: GT2 means 2 mm tooth pitch, and ratio is just pulley tooth count quotient — one reason belts are so beginner-friendly.
Manufacturer efficiency claims deserve auditing, and the test is pleasantly simple. Method one, electrical-to-mechanical: drive the gearmotor lifting a known weight on a pulley at steady speed; mechanical power out = weight × g × lift speed, electrical power in = voltage × current from your bench meter or inline watt-meter, and the quotient is total efficiency (motor × gearbox combined). Method two, comparative: measure the current the motor draws spinning free with and without the gearbox attached — the difference is the gearbox's friction tax made visible. Do this once on a cheap gearmotor and a quality planetary and you'll never again treat the efficiency column in the table above as pedantry: the spread between 60% and 90% is the difference between a 40-minute and a 60-minute battery on the same pack — which loops straight back into the Battery Runtime Calculator's inputs.
Used or long-stored gearmotors deserve a quick audit before earning a place in a build. Spin the output shaft by hand (unpowered): smooth, even resistance is healthy; a rhythmic catch once per revolution is a damaged tooth. Wiggle the output shaft radially: a little play is normal, a lot means worn bushings and imminent mesh problems under load. Run it free at rated voltage and listen — a steady whir is fine, a periodic click is a tooth, a grinding note is debris or dry gears (a drop of light machine grease, never household oil, on open trains). Finally compare free-run current against a known-good sibling: elevated no-load current is friction announcing itself before it becomes failure. Sixty seconds, no tools beyond a power supply, and it has saved more competition weekends than any exotic technique in this article.