Blog · 2026-07-05 · Gearing
Ratio selection isn't taste — it's a five-step calculation with a check at the end. Follow the process once and you'll have a repeatable method for every drivetrain you ever build.
Builders usually pick gear ratios one of two wrong ways: copying whatever a similar-looking robot used, or buying the gearmotor that happened to be in stock and living with the result. Both work often enough to be dangerous. The right way takes ten minutes, needs four inputs you already know, and ends with a robot whose top speed and pushing force are choices, not accidents. Here is that process, followed by two complete examples.
Everything downstream flows from the wheel speed you actually need, so resist the instinct to specify a thrilling number. Indoor navigation robots live happily at 0.3–0.8 m/s; a brisk hobby rover at 1–2 m/s; anything above 3 m/s is genuinely hard to control indoors and chews battery outdoors. Remember the trade you learned in Gear Ratios Explained: every unit of unused top speed was purchased with torque and acceleration you could have kept. Gearing for a speed you'll never safely command is the most common ratio mistake.
Wheel RPM = (speed in m/s × 60) ÷ (π × wheel diameter in m)
For 1 m/s on 80 mm wheels: (1 × 60) ÷ (π × 0.08) ≈ 239 RPM. Note the lever hidden here: wheel diameter is itself a gearing decision. Doubling wheel size halves the RPM requirement but also halves pushing force per unit torque — bigger wheels are, mechanically, a gear-up. Choose diameter for terrain (bigger clears obstacles) and let the gearbox handle the rest.
Datasheets quote no-load RPM; a healthy loaded motor runs at roughly 75–85% of it. Use 80% as your planning figure — a 6,000 RPM motor plans as 4,800 RPM. Skipping this correction is why so many finished robots come out mysteriously 20% slower than designed.
Required ratio = loaded motor RPM ÷ target wheel RPM
Continuing the example: 4,800 ÷ 239 ≈ 20:1. You won't find exactly 20:1 on a shelf, and you don't need to — round toward more reduction (a stock 25:1 or 30:1 here). The robot arrives slightly slower and slightly stronger than spec, which is the failure direction nobody notices. Rounding the other way delivers a robot that can't reach its own top speed on carpet, which everybody notices.
A ratio that hits your speed is only half-validated. Now verify the geared output torque covers the requirement. First get the requirement itself from the Motor Sizing Calculator (weight, incline, surface, acceleration → kg·cm per motor). Then compute what your candidate delivers:
Output torque = motor rated torque × ratio × gearbox efficiency
Use honest efficiency: ~92% for one or two spur stages, ~85% for planetary boxes and belts, as low as 50% for worm drives. If output torque falls short of requirement, your options in order of preference: more reduction (if you have speed margin to spend), a torquier motor, or smaller wheels. If it exceeds requirement massively, you likely have room to gear for more speed — or to buy a smaller, cheaper, lighter motor, which is the secret bonus of doing this math.
Spec: 2.5 kg robot, two motors, 65 mm wheels, 0.6 m/s on smooth floor, gentle 5° ramps. Wheel RPM: (0.6 × 60) ÷ (π × 0.065) ≈ 176 RPM. Motor: a common 6 V "TT" class motor, no-load ~9,000 RPM → 7,200 loaded. Ratio: 7,200 ÷ 176 ≈ 41:1 → buy 48:1 stock (the classic TT gearbox — no coincidence; this math is why it's the classic). Torque check: requirement from the sizing calculator ≈ 0.55 kg·cm per motor; the 48:1 TT delivers roughly 0.8 kg·cm rated. Pass, with margin. Verdict: the cheapest gearmotor in robotics is genuinely correct for this build — now you know why.
Spec: 9 kg garden robot, four motors, 125 mm wheels, 1.2 m/s, 15° slopes on grass. Wheel RPM: (1.2 × 60) ÷ (π × 0.125) ≈ 183 RPM. Motor: 12 V, 37 mm class, no-load 10,000 RPM internal → the shopping question becomes "which 37 mm gearmotor variant lands near 183 RPM?" — the 1:56 variant (no-load ~178 RPM) is the fit. Torque check: the sizing calculator (9 kg, 15°, grass Crr ≈ 0.055, 1.5 SF, four motors) demands ≈ 6.5 kg·cm per motor; the 1:56 variant is rated ~10 kg·cm. Pass. Traction check: 4 driven wheels carrying 9 kg on grass gives ample grip headroom. One design note: grass plus 15° is exactly the regime where stall events happen — confirm the pack can feed four motors' stall current using the Battery Runtime Calculator's C-rating line.
For drivetrains, buy it: gearmotors with integrated boxes are cheaper, tighter and more reliable than anything hand-assembled from loose gears. Build custom reduction when you need what stock boxes don't offer — an offset drive (belt across the chassis), a specific in-between ratio (add a 2:1 belt stage after a stock gearmotor; stage ratios multiply), silence (belts), or holding torque (worm). The Gear Ratio Calculator suggests tooth pairings and stage splits when your ratio exceeds single-stage territory.
Yes — pivot turns scrub wheels sideways and can demand 2–4× straight-line torque on grippy surfaces. Four-wheel skid-steer builds should run the torque check with a 2.0 safety factor.
Your motor is slower than the wheel needs, so you'd be gearing up and dividing torque. Almost always the right fix is a faster motor or a smaller target speed, not an overdrive.
With integrated gearmotors, you swap motors — one argument for belt final drives, where a pulley change re-gears the robot in minutes. Competition teams design this in deliberately.
That's the whole method: speed → wheel RPM → loaded motor RPM → ratio → torque and traction check. Run your own numbers in the Gear Ratio Calculator, with the Motor Sizing Calculator supplying the torque requirement — the pair together is a complete drivetrain design session.
After enough builds, patterns emerge. Small indoor robots (1–3 kg, 60–70 mm wheels, 0.5–1 m/s) on 6 V TT-class motors land at 48:1–120:1. Mid-size platforms (3–8 kg, 80–100 mm wheels, ~1 m/s) on 12 V 37 mm gearmotors land at 30:1–75:1. Fast rovers (1.5–3 m/s) push down toward 10:1–20:1 and pay for it in ramp performance. Arm joints and lifts, where speed barely matters and holding force does, run 100:1 and beyond — worm territory. None of this replaces the calculation; it's a sanity band. If your math says 400:1 for a driving robot or 5:1 for a lift, re-check the inputs before trusting the output.
Quoting no-load speed in step 3 anyway. It's the most reverted correction in drivetrain design because listings make the big number so visible. Write "×0.8" on the whiteboard. Checking torque against stall instead of rated. A candidate that passes on stall torque fails on a warm afternoon; the distinction is covered fully in Motor Torque Explained. Forgetting the drivetrain adds mass. Bigger motors and gearboxes chosen in step 5 make the robot heavier than the estimate used in the sizing calculator — for marginal designs, run one more iteration with updated mass. Convergence takes two passes at most, and a design that only closes without its own motors' weight was never closed.
Sometimes step 5 leaves you with two stock gearmotors that both satisfy speed and torque — say a 30:1 and a 50:1 variant. Break the tie by asking where the design is most likely to be wrong. If your mass estimate is soft (payloads tend to grow) or the terrain is rougher than assumed, take the higher reduction: the cost is top speed you specified honestly and can afford to lose. If the robot's job genuinely depends on pace — racing, coverage rate for a cleaning robot — take the lower reduction and compensate with the conservative safety factor already baked into your torque requirement. And if the two candidates differ in price, remember the slower, torquier variant usually runs cooler and draws less peak current, which quietly relaxes your driver and battery specs too. Ties in gearing are rarely ties once the whole power system is in view.