A pinion is the smaller of two meshing gears, or the smaller gear in a rack-and-pinion pair, that drives a larger gear or a toothed rack to transmit torque and motion. The term traces to early French clockmakers — Antide Janvier documented pinion-and-wheel pairings in the 1780s. The pinion typically takes the input shaft from a motor and steps speed down or torque up by its tooth-count ratio. Modern uses range from automotive steering racks to CNC gantry drives moving 30 m/min with sub-0.05 mm positioning error.
How the Pinion Actually Works
A pinion meshes with a larger gear or a flat toothed rack. Drive the pinion shaft and its teeth push against the mating teeth — that contact force, multiplied by the pitch radius, is the torque transferred. Because the pinion has fewer teeth, one rotation of the pinion produces only a fraction of a rotation on the larger gear, which is why pinions sit on the high-speed, low-torque side of almost every gear pair. In a rack-and-pinion, each rotation of the pinion advances the rack by π × D<sub>pitch</sub>, so the pitch diameter directly sets your linear feed per turn.
The geometry is unforgiving. Module (or diametral pitch in imperial) must match exactly between pinion and mate — a Module 2 pinion will not run cleanly against a Module 2.5 rack, even though they look almost identical. Centre distance has to land within roughly ±0.05 mm of the theoretical (R<sub>pinion</sub> + R<sub>gear</sub>) value, otherwise you get either binding or excessive backlash. Tooth profile is involute on virtually every modern pinion because the involute keeps the contact point sliding along a straight line of action regardless of small centre-distance errors.
What goes wrong? Undercutting on pinions with fewer than 17 teeth at 20° pressure angle — the cutter takes a bite out of the tooth root and you lose bending strength. Pitting on the flank from contact stress above the material's allowable Hertzian limit. And on rack drives specifically, drive pinion misalignment of more than about 0.1° tilts the load to one tooth edge and you get rapid wear on a single-flank stripe. If you notice chatter or a clicking that follows pinion rotation rate, check tooth-to-tooth runout first — anything above ISO grade 8 (about 25 µm for a Module 2) and you'll feel it on the output.
Key Components
- Pinion body and bore: The hub carries the input torque from a motor shaft, usually through a keyway, taper, or shrink fit. Bore tolerance is typically H7 to match a k6 or m6 shaft for an interference-free slip fit with key, and concentricity between bore and pitch circle must hold within 20 µm or you get cyclic backlash.
- Pinion teeth: Cut to an involute profile at a standard pressure angle of 20° (or 14.5° on older equipment, 25° on heavy-duty drives). Tooth count below 17 needs profile shift to avoid undercutting. Surface hardness on a hardened steel pinion runs 58-62 HRC for contact-fatigue life.
- Mating gear or rack: Must share identical module and pressure angle. On rack-and-pinion CNC gantries the rack is usually Module 2 or Module 3, hardened and ground, supplied in 1 m or 2 m segments that bolt end-to-end with a pin-aligned joint. Joint accuracy below 10 µm is what separates a quality install from a noisy one.
- Pinion shaft and bearings: Two bearings minimum, straddling the pinion, to keep tooth alignment under load. Radial play above 30 µm shows up as audible whine because the pinion walks across the rack face under torque reversal.
- Lubrication interface: Open gears use NLGI 1 or 2 grease with EP additives; enclosed pinions run in oil bath or splash. Without proper film thickness (calculated by elastohydrodynamic theory) you transition to boundary lubrication and pitting starts inside 100 hours.
Who Uses the Pinion
Pinions show up anywhere you need to convert rotary motor output into either a slower-but-stronger rotation or a clean linear motion. The application set spans hand tools to heavy industry — what changes is module size, material, and accuracy class. On a CNC plasma cutter you want Module 1.5 ground steel for ±0.03 mm position accuracy. On a quarry conveyor drive you want Module 20 cast steel and you don't care about microns. The trade-offs around tooth count, pressure angle, and helix angle all flow from what the application demands.
- Automotive: Rack-and-pinion steering on the Toyota Corolla and virtually every passenger car since the 1970s — a small pinion on the steering column drives a horizontal rack to swing the tie rods.
- CNC machine tools: Gantry drive on Haas GR-712 router and similar large-format CNCs use a Module 2 helical pinion against a hardened rack for X-axis travel up to 3.6 m at 30 m/min.
- Rail and transit: Rack railways like the Pilatus Railway in Switzerland use a horizontal pinion engaging a vertical Locher rack to climb 48% grades.
- Construction equipment: Mast-climbing work platforms from Alimak Hek run a pinion gearbox up a fixed toothed mast, lifting 1500 kg payloads at 12 m/min.
- Industrial automation: Pinion-driven linear axes on FANUC palletisers and ABB IRB-series robot 7th-axis tracks, where rack lengths up to 20 m make ball-screws impractical.
- Heavy machinery: Drive pinion in a final-drive differential on a Caterpillar 793F mining truck — takes 2,500 hp from the engine and steps down to wheel torque through a hypoid pinion.
The Formula Behind the Pinion
The most useful pinion calculation for a designer is the linear feed per pinion revolution in a rack-and-pinion drive — it tells you motor RPM for any required feed rate. At the low end of typical CNC operating range (around 600 RPM input) a Module 3 pinion gives gentle, accurate motion good for finish cuts. At the nominal mid-range (~1500 RPM) you hit the productive sweet spot. Push toward 3000 RPM and you start fighting pinion-tooth dynamic loads, mesh noise, and the sliding velocity at the tooth flank that lubrication can no longer support. The formula assumes zero-backlash mesh and ideal rolling — real systems lose 1-3% to elastic deflection.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| vlinear | Linear travel speed of the rack | m/s | in/s |
| m | Module of the pinion (pitch diameter ÷ tooth count) | mm | in (use diametral pitch reciprocal) |
| z | Number of teeth on the pinion | count | count |
| N | Pinion rotational speed | RPM | RPM |
Worked Example: Pinion in a CNC fibre-laser gantry drive
A sheet-metal fabricator in Hamilton is sizing the X-axis pinion drive for a 4 m × 2 m fibre-laser cutter. They've selected a Module 3, 20-tooth helical pinion with a 19° helix angle, driven by a servo through a 10:1 planetary reducer. They need to know the cutting feed rate at three operating points: a slow profiling pass, the nominal cutting speed, and the rapid-traverse limit.
Given
- m = 3 mm
- z = 20 teeth
- Npinion,nom = 500 RPM
- Reducer ratio = 10:1 —
Solution
Step 1 — calculate the pitch diameter of the pinion. This sets travel per revolution:
Step 2 — at nominal pinion speed of 500 RPM (servo at 5000 RPM through the 10:1 reducer), compute linear feed:
That's a productive cutting feed for 3 mm mild steel on a 6 kW fibre laser — fast enough for throughput, slow enough that kerf quality stays clean.
Step 3 — at the low end of typical operating range, 200 RPM pinion (slow profiling pass on 10 mm plate):
At 38 m/min the head moves at a measured walking pace — you can watch the cut form. This is where you do edge-quality-critical work.
Step 4 — at the high end, rapid traverse at 1000 RPM pinion:
188 m/min is achievable on rapids, but pushing the pinion past about 1200 RPM continuously starts to drive tooth-mesh frequency into the gantry's structural resonance band on a typical welded steel frame, and you'll hear it as a singing whine. Most shops cap continuous operation around 800-900 RPM pinion speed.
Result
The nominal cutting feed is 1. 57 m/s (94 m/min) at the rack. The low-end profiling speed of 38 m/min gives you the time you need to hold edge quality on thick stock, while the 188 m/min rapid traverse at the high end is fine for non-cutting moves but will excite gantry resonance if held continuously. The sweet spot sits around 80-110 m/min for production cutting on 3-6 mm mild steel. If your measured feed comes out 5-8% below predicted, the most likely causes are: (1) servo following error climbing under acceleration because the inertia mismatch between pinion-gantry and motor wasn't tuned, (2) rack-segment joints opening more than 10 µm so the encoder reads more rotation than the gantry actually travels, or (3) helical pinion thrust load deflecting the support bearings axially — a common miss when the bearing pair isn't preloaded against the helix-induced thrust.
When to Use a Pinion and When Not To
A pinion drive isn't always the right answer. For long travels above 2 m it usually wins on cost and stiffness — but for short, high-precision moves, a ball screw is hard to beat. Belt drives sit in between. Compare on the dimensions that actually matter for a motion-system designer.
| Property | Rack-and-pinion | Ball screw | Toothed belt drive |
|---|---|---|---|
| Practical travel length | Up to 20+ m (joinable rack) | Limited to ~6 m by whip and critical speed | Up to 10 m, limited by belt stretch |
| Maximum linear speed | 3-5 m/s | 1-2 m/s typical | 5-10 m/s |
| Positioning accuracy | ±0.05 mm with ground rack | ±0.005 mm with C3 ball screw | ±0.1 mm typical |
| Backlash (typical) | 20-50 µm single pinion, <5 µm dual-pinion preloaded | <5 µm preloaded | 0 µm if belt tensioned |
| Stiffness | High and constant over length | High but degrades with length cubed | Low — belt acts as a spring |
| Cost per metre of travel | Low at long length | High and rises sharply with length | Lowest |
| Maintenance interval | Re-grease every 500-1000 hours | Re-grease ball nut every 100 km travel | Belt inspection every 2000 hours |
Frequently Asked Questions About Pinion
A single-pinion drive has irreducible backlash equal to the manufactured tooth-to-tooth clearance, typically 30-50 µm for a Module 2-3 system in DIN quality 7. The fix isn't tighter mesh — that just causes binding when temperature shifts the centre distance. The fix is mechanical: either a spring-preloaded split pinion, or a dual-pinion drive where two pinions on parallel shafts are torque-biased against each other so one always pushes against the leading flank and one against the trailing flank.
Dual-pinion preload kills backlash to under 5 µm but costs roughly 2× the drive hardware and demands a second servo or a mechanical biasing gearbox like the Wittenstein alpha TP+.
You're undercut. Below 17 teeth at a 20° pressure angle, the standard hob cuts into the tooth root and removes material from the bending-stress region. The tooth fillet ends up thinner than the involute profile assumes, so root bending stress can be 30-50% higher than your calculation predicts.
Two fixes: switch to a 25° pressure angle pinion (allows tooth counts down to about 12 without undercut), or specify positive profile shift (x-factor of +0.3 to +0.5) which moves the cutter outward and restores the root thickness. Most CAD gear generators handle this if you tell them the pinion needs profile correction.
Helical wins on noise and load capacity above about 3 m/s pitch-line velocity. The angled tooth engagement gives you contact ratio above 2.0, so at any moment more than two teeth share the load — that smooths torque ripple and drops mesh noise by 6-10 dB compared to a spur of the same size.
The cost is axial thrust. A 19° helix on a 60 mm pitch pinion under 200 Nm produces about 1.2 kN of axial force on the bearings. If you don't preload an angular-contact pair against that thrust, the pinion walks under torque reversal and you lose your positioning accuracy. For low-speed, light-duty axes, a spur is simpler and cheaper.
You're seeing rack-joint and thermal expansion error that the rotary encoder on the motor cannot detect. A motor-side encoder counts pinion rotation, not actual gantry position — so any rack-segment joint gap, pinion runout, or thermal growth in the rack between cold start and warm running shows up as positioning error invisible to the servo.
The fix is a linear glass scale or magnetic tape encoder reading directly off the moving axis. Heidenhain LC scales or Renishaw RGH magnetic tape give you sub-5 µm absolute position regardless of what the pinion is doing. On long axes, also check that the rack is bolted to a thermally stable beam — an aluminium gantry with steel rack will fight you at every shift change.
Compute the tooth bending stress with the Lewis equation at your peak acceleration torque, not your steady-state torque. For a 200 kg gantry accelerating at 5 m/s² you need roughly 1000 N of drive force, which translates through a 60 mm pitch pinion to about 30 Nm. Module 2 with a 20 mm face width is right at the edge of allowable bending stress for a hardened steel rack. Module 3 with the same face width has roughly 1.7× the bending strength.
Rule of thumb: Module 2 for payloads under 100 kg and accelerations under 3 m/s². Module 3 for the 100-500 kg range. Module 4 and up for heavy gantries or vertical lifts where gravity loads on top of acceleration. Don't oversize blindly — bigger module means more inertia at the pinion, which eats your acceleration headroom.
Misalignment leaves a single-flank wear stripe — wear concentrated on one edge of the tooth face across all teeth, with the opposite edge looking nearly new. That's a clear shaft-parallelism problem, usually traceable to the gearbox mounting or rack mounting surface being out of square by more than 0.05 mm/m.
General wear or overload produces uniform face-width contact with pitting starting near the pitch line and spreading. If you see scuffing (matte, torn-looking flanks) instead of polished pitting, that's lubrication failure — film thickness collapsed and you went into metal-to-metal contact. Each pattern points to a different fix: re-shim the gearbox for the stripe, drop the load or upgrade the lubricant for the scuffing.
References & Further Reading
- Wikipedia contributors. Pinion. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
