In this guide
How to Design a Simple Gearbox for a Student Robot
Quick answer
A gearbox trades speed for torque: pick the overall ratio from the wheel speed and torque your robot needs, split it into two or three stages of about 5:1 each, choose spur gears of one module throughout, and size the shafts from the output torque. A 3,000 RPM motor with 0.10 N·m becomes about 120 RPM and 2.3 N·m through a 25:1 two-stage box.
| Item | Worked design value |
|---|---|
| Motor | 3,000 RPM, 0.10 N·m |
| Overall ratio | 25:1 (two stages of 5:1) |
| Output | 120 RPM, about 2.3 N·m |
| Gears | Module 1 spur gears, 12T driving 60T per stage |
| Output shaft | 8 mm, on ball bearings |
Why your robot needs a gearbox
A small DC motor spins fast and pushes weakly — typically a few thousand RPM and a fraction of a newton-metre of torque. A robot wheel needs the opposite: roughly a hundred RPM and several newton-metres. No motor gives you both at once, so the gearbox sits between them and trades one for the other. Power is conserved (minus losses), so halving the speed roughly doubles the torque.
The same trade appears everywhere in a student build: a robot arm joint needs high torque at low speed, a conveyor needs steady slow torque, a gripper needs a large force from a small motor. Wherever the motor's natural speed and torque do not match the job, a gearbox is the answer — and a gearbox you design yourself, from gears you can actually buy, is cheaper and more repairable than an imported gearmotor with unknown internals.
This guide designs a real one: a two-stage spur gearbox for a robot drive wheel, worked from the motor datasheet to the shaft diameter, with every number shown so you can repeat the process for your own motor and target.
The five decisions
Every gearbox design is five decisions in order. Skip one and the box fails in a way that looks like bad luck but is actually a missed decision.
| Decision | Question it answers | Covered in |
|---|---|---|
| 1. Overall ratio | How much speed must become torque? | Ratio from speed and torque |
| 2. Number of stages | How is the ratio split? | Stage splitting |
| 3. Gear type | Spur, helical, worm, or planetary? | Gear-type table |
| 4. Module and teeth | What size are the gears? | Module and centre distance |
| 5. Shafts, bearings, housing | What holds it all together? | Shafts and housing |
Decision 1: the overall ratio
Start from the output, not the motor. Two numbers define the job: the wheel speed you want and the torque you need.
Ratio from speed. Divide the motor speed by the target output speed. A motor at 3,000 RPM driving a wheel at 120 RPM needs 3,000 ÷ 120 = 25, so a 25:1 reduction.
Check the torque. Output torque is input torque times the ratio times the efficiency. Spur-gear stages are commonly estimated near 95% efficient each in clean, well-aligned mesh, so two stages give about 0.95 × 0.95 = 0.90. With a 0.10 N·m motor: 0.10 × 25 × 0.90 ≈ 2.3 N·m at the output. If that is less than your robot needs, you need a bigger ratio, a stronger motor, or both — do not discover this after building.
Check the speed is useful, not just possible. A 100 mm wheel at 120 RPM moves the robot at π × 0.1 m × 120 ÷ 60 ≈ 0.63 m/s — a brisk walking pace, sensible for a demo robot. If the answer had been 3 m/s, the ratio would be wrong for a small indoor robot no matter how neat the maths looked. Always convert the output RPM back into a real speed or a real arm-swing time and ask whether it fits the demo.
Work backwards from torque when the load is known. If your arm joint needs 2 N·m and your motor gives 0.10 N·m, the minimum ratio is 2 ÷ (0.10 × 0.90) ≈ 22:1 before efficiency, so a 25:1 box covers it with a small margin. Design from the harder requirement — torque or speed — and verify the other one.
Decision 2: how many stages
A single stage of 25:1 would need a driven gear 25 times the diameter of the pinion — a 12-tooth pinion driving a 300-tooth gear, a disc wider than most robots. That is why large reductions are built from several modest stages: the ratios multiply, so two stages of 5:1 give 25:1 in a package that fits on a robot.
Keep each spur stage between about 3:1 and 7:1. Below 3:1 you are wasting a stage; above 7:1 the driven gear gets unwieldy and the pinion gets too small. For ratios up to about 50:1, two stages are enough. For 50:1 to 300:1, use three. Beyond that, consider a worm stage or a planetary box instead of stacking more spur stages — each extra stage costs about 5% efficiency and adds a shaft, two bearings, and another alignment to get wrong.
Stages multiply, they do not add: 5:1 followed by 5:1 is 25:1, not 10:1. And an idler gear — a gear sitting between two others only to bridge a distance or reverse direction — does not change the ratio at all; it only changes the direction and the spacing. Students lose marks in vivas by counting an idler as a stage, so count reductions, not gears.
Decision 3: which gear type
| Type | Strengths | Weaknesses | Use it when |
|---|---|---|---|
| Spur | Simple, cheap, easy to buy and mount, efficient | Noisy at speed, teeth take load one at a time | Most student gearboxes; the default choice |
| Helical | Quieter, stronger, smoother — teeth engage gradually | Axial thrust needs thrust bearings; costlier | A fast or quiet box, or a heavily loaded one |
| Worm | Huge ratio in one stage (up to about 100:1), compact, often self-locking | Low efficiency (typically 50–70%), wears faster, needs bronze wheel | A slow lift or tilt that must hold position unpowered |
| Planetary | Very compact, high ratio per stage, load shared across planets | Complex to build, harder to buy as loose parts | A tight space with a big ratio, e.g. a compact arm joint |
For a first gearbox, use spur gears. They are the cheapest to buy, the easiest to mount on parallel shafts, and the simplest to reason about — and a well-built spur box is efficient and reliable. Move to helical only if noise or load demands it, to worm only if you need a big ratio in one stage or self-locking, and to planetary only when space forces you to.
One rule covers all types: meshing gears must share the same module (metric) or diametral pitch (imperial). A module 1 gear will not mesh with a module 1.5 gear, however similar they look. Buy all gears for one stage — and ideally the whole box — in the same module.
Decision 4: module, teeth and centre distance
Module is the size unit of a gear: module = pitch diameter ÷ number of teeth. A module 1 gear with 12 teeth has a pitch diameter of 12 mm; with 60 teeth, 60 mm. For student robots, module 0.5 to 1.5 covers almost everything — module 1 is a good default because gears are widely stocked and the teeth are big enough to handle in a workshop.
Centre distance is fixed by the module and the tooth counts: centre distance = module × (teeth₁ + teeth₂) ÷ 2. For a 12-tooth pinion driving a 60-tooth gear at module 1, that is 1 × 72 ÷ 2 = 36 mm between shaft centres. Get this distance right and the gears mesh with the correct backlash; get it wrong by even half a millimetre and the mesh is either tight and noisy or loose and sloppy. Drill the bearing holes from one measured layout, not by eye, and check the distance with a vernier caliper before assembly.
Minimum teeth. A standard 20° involute spur pinion needs at least 17 teeth to avoid undercut — a weakening of the tooth root that happens when the cutter removes too much material from a small pinion. Use 12-tooth pinions only when they are bought as proper profile-shifted gears; otherwise keep pinions at 17 teeth or more. In our worked design the 12-tooth pinions assume bought profile-shifted pinions; with standard cut gears, use 18 teeth driving 90 teeth for the same 5:1 stage.
Face width. The width of the gear teeth should be roughly 8 to 12 times the module — 8 to 12 mm for module 1. Narrower teeth overload and wear; much wider teeth add weight and demand more precise alignment for no benefit.
Backlash. A small gap between meshing teeth is deliberate: without it, the gears bind as they warm up and as manufacturing errors stack. For module 1 student gears, a few hundredths of a millimetre per mesh is typical. You can feel it as a tiny rock of the output shaft with the input held still. Too much backlash makes a robot arm sloppy and a drive jerky on direction changes; too little makes the box stiff and noisy.
Decision 5: shafts, bearings and housing
Gears do not hold themselves up. The shaft, bearing, and housing decisions decide whether the box runs for a semester or eats itself in a week.
Shaft diameter from torque. A shaft fails in torsion, and the diameter follows from the torque it carries. For a solid shaft, diameter³ = 16 × torque ÷ (π × allowable shear stress). Using 40 MPa as a design allowable shear stress for a mild-steel shaft — a conservative student-design value — the output shaft carrying 2.3 N·m needs diameter³ = 16 × 2.3 ÷ (π × 40 × 10⁶) ≈ 2.93 × 10⁻⁷ m³, giving a diameter of about 6.6 mm. Use an 8 mm shaft: the next standard size up, with margin for keyways and shock loads. Input shafts carry far less torque and can be smaller, but keep them at least 6 mm for stiffness and for standard bearings.
Bearings. Support every shaft at both ends with ball bearings — one on each side plate — so the gears stay aligned under load. Bushings (plain bearings) are cheaper and acceptable on slow, lightly loaded shafts, but ball bearings are the right default for a drive gearbox: they handle radial load well, add little friction, and are cheap in standard sizes such as 608 (8 mm bore). Press bearings into the housing plates; never rely on a drilled hole alone to locate a shaft.
Housing. Two side plates, held apart by spacers, with the shafts running between them. Acrylic lets you see the gears mesh — genuinely useful for debugging — but it flexes; aluminium or mild-steel plates stay rigid. The plates must hold the centre distances from Decision 4, so mark and drill them together, clamped as a pair, or the two sides will disagree and the shafts will bind. Add a lid or cover: an open gearbox collects dust, drops screws into the mesh, and is a finger hazard.
Keyways and fixing gears to shafts. Gears must not spin on their shafts. Use a keyway and key, a grub screw onto a flat filed on the shaft, or a press fit — and put a collar or circlip on each side so gears cannot wander along the shaft and out of mesh. A gear that has drifted 2 mm sideways is only half-engaged and will strip its teeth under load.
Lubrication. Grease the mesh lightly before closing the box — a thin smear of lithium grease on the teeth, not a bath of oil. Re-grease when the box is opened for inspection. A dry gearbox is noisy, inefficient, and short-lived; an over-greased one flings grease everywhere and picks up grit.
Worked design: a robot drive gearbox
Putting it together for a 5 kg robot with 100 mm wheels, driven by a 3,000 RPM, 0.10 N·m DC motor:
| Item | Value |
|---|---|
| Target wheel speed | 120 RPM |
| Overall ratio | 25:1 |
| Stages | Two, 5:1 each |
| Stage 1 | 12T pinion → 60T gear, module 1, centre distance 36 mm |
| Stage 2 | 12T pinion → 60T gear, module 1, centre distance 36 mm |
| Output torque | 0.10 × 25 × 0.90 ≈ 2.3 N·m |
| Output speed | 3,000 ÷ 25 = 120 RPM |
| Robot speed | π × 0.1 × 120 ÷ 60 ≈ 0.63 m/s |
| Output shaft | 8 mm, mild steel, on 608 ball bearings |
| Housing | Two 3 mm aluminium plates, 36 mm shaft spacing, greased |
Check the torque against the robot: a 5 kg robot needs roughly 0.5–1 N·m per wheel to accelerate and climb small obstacles, so 2.3 N·m per wheel leaves comfortable margin without being absurdly over-geared. If the number had come out at 0.3 N·m, the robot would stall on carpet; if it had come out at 10 N·m, the robot would crawl. The check takes a minute and it is the difference between a design and a guess.
Build it in this order: cut the plates and drill the bearing holes as a clamped pair, fit the bearings, assemble the shafts with gears and spacers on the bench, check the mesh by hand, then close the box and run it unloaded before fitting it to the robot.
Assembly and testing checklist
- Mesh check by hand. With the box open, turn the input shaft slowly. The gears should turn smoothly with a faint, even resistance — no tight spots, no catching, no grinding. A tight spot at one point of rotation usually means a bent shaft or a plate hole drilled off-centre.
- Backlash feel. Hold the input still and rock the output. A tiny movement is correct; a large clunk means the centre distance is too big or a gear is the wrong module.
- Marking test. Smear a thin coat of marking compound or even a felt-tip pen on a few teeth, run the gears through a few turns, and look at the contact pattern. It should sit centred on the tooth face. A pattern hard against one edge means the shafts are not parallel.
- No-load run. Run the motor unloaded for a minute and listen. A steady hum is fine; a rhythmic knock or a rising whine is a misaligned stage. Feel the housing afterwards — warm is normal, hot means something is binding.
- Loaded run. Fit the wheel, hold the robot back gently, and check it still turns smoothly under load. Measure the output RPM with a tachometer or a phone strobe app and compare with the design 120 RPM — a large gap means slipping, a wrong ratio, or a motor weaker than its datasheet.
- After the first hour. Open the box, look for metal dust or shiny wear strips on the teeth, re-grease lightly, and check every grub screw and collar is still tight. Most gearbox failures announce themselves at this inspection if you do it.
5 common mistakes
1. One giant stage instead of two small ones. A 25:1 single stage needs a 300-tooth gear. Split the ratio, keep each stage under about 7:1, and the box stays compact, efficient, and buyable.
2. Adding stage ratios instead of multiplying them. Two 5:1 stages give 25:1, not 10:1. The same error in reverse — dividing where you should multiply — is the most common arithmetic mistake in gearbox design, and it produces a robot that is five times too slow or five times too weak.
3. Ignoring efficiency. Torque out is torque in times ratio times efficiency, not just times ratio. Two stages at 95% each leave 90%, and a worm stage can leave 50–70%. Forgetting this is how a robot that "should" climb a ramp stalls on it.
4. Undersized shafts and no bearing support. A shaft that is too thin twists and lets the gears separate under load, and a shaft held at one end only bends. Size the output shaft from the torque, support every shaft at both ends, and keep gears from wandering with collars.
5. Wrong centre distance. Drilling bearing holes by eye, or measuring from the wrong edge, gives a mesh that is too tight or too loose. Mark and drill plates as a clamped pair from one layout, verify the centre distance with a caliper, and test the mesh by hand before closing the box.
Frequently asked questions
Can I just buy a gearmotor instead of designing a gearbox?
Yes, and for many projects you should — a bought gearmotor is compact, tested, and cheap. Design your own when the ratio you need is not sold, when you need to repair or modify the drive, or when the design itself is part of the project assessment. Knowing how to design one also tells you whether a bought gearmotor's claimed ratio and torque are believable.
How do I pick the module?
Module 1 is the default for student robots: gears are widely stocked, teeth are big enough to handle, and the sizes work out to sensible centre distances. Go to module 0.5 for very small, light mechanisms and module 1.5 or 2 for heavy, high-torque builds. Whatever you pick, keep one module through the whole box.
What is the difference between a gearbox and a gear train?
A gear train is the set of meshing gears; a gearbox is the gear train plus its shafts, bearings, housing, and lubrication — the complete working unit. Designing the train is the maths; designing the box is the engineering.
Do I need to consider the motor's stall torque?
Yes, for the worst case. Size the gears and shafts for the torque the motor can actually deliver when stalled, not just its rated torque, or a jammed wheel can strip teeth. A small safety margin on the rated torque, and a current limit or fuse on the motor driver, covers most student situations.
Why is my gearbox noisy?
Usually backlash that is too large, centre distance slightly off, shafts not parallel, or dry gears. Check in that order: re-grease first (cheapest fix), then check the mesh by hand for tight spots, then verify the centre distances and shaft parallelism.
Can I 3D-print the gears?
For light loads and demonstration, yes — printed PLA or PETG gears work for low-torque mechanisms and are excellent for proving a layout before buying metal gears. They wear quickly under real drive loads and they flex, so do not trust them for the drive wheels of a heavy robot. Print the housing and buy the gears is the usual compromise.
Limitations
This guide covers spur-gear box design for student robots: ratio, staging, module, shaft sizing, and testing. It does not cover helical-gear thrust calculations, worm-gear efficiency and self-locking analysis, planetary-gear layout, or fatigue-life calculation — those need a machine-design textbook and the gear manufacturer's data. The efficiency figures (about 95% per spur stage), the 40 MPa shaft allowable, and the torque requirement for a 5 kg robot are design assumptions labelled as such; always confirm the final design against your motor's datasheet and a test of the built box. Tooth counts below 17 assume profile-shifted pinions bought as such; with standard cut gears, keep pinions at 17 teeth or more.
Get help with your gearbox
If you are designing a gearbox for a robot drive, arm, or mechanism and want the ratio, gear selection, and shaft sizing checked before you buy parts, message us on WhatsApp at wa.me/918275719193 with your motor specifications, target speed and torque, and deadline. The reply is a quotation stating the scope, the price, and the delivery timeline.