Wheel Train

← Back to Engineering Library

A Wheel Train is a series of meshed gears — alternating large wheels and small pinions — that transmits rotation between a power source and an output while changing speed and torque. It exists to solve a fundamental drive problem: a mainspring or weight delivers slow, high-torque rotation, but the output (clock hands, escapement, or driven shaft) needs a different speed. Each wheel-pinion pair multiplies the ratio, so a four-stage train can step a 1 RPH barrel up to 18,000 BPH at the escape wheel. That's how a mechanical watch turns one wind into 40 hours of timekeeping.

Watch the Wheel Train in motion
Video: Sector wheel baling press by Nguyen Duc Thang (thang010146) on YouTube. Used here to complement the diagram below.
Wheel Train Gear Ratio Diagram Animated three-stage gear train showing speed multiplication from input to output. BARREL (Input) 72 teeth 12 leaves Arbor 60 teeth 10 leaves 56 teeth ESCAPE WHEEL 8 leaves 1 RPH 252 RPH 72÷12 = 6 × 60÷10 = 6 × 56÷8 = 7 = Total: 252×
Wheel Train Gear Ratio Diagram.

How the Wheel Train Actually Works

A Wheel Train works by stacking gear pairs in series. Each stage has a large wheel driven by the previous pinion, and a small pinion fixed to the same arbor that drives the next wheel. The ratio of one stage equals the wheel tooth count divided by the pinion tooth count — typically 6:1 to 10:1 per stage in a clock, 6:1 to 8:1 in a wristwatch. Multiply the stage ratios together and you get the total train ratio. A going train in a typical mechanical watch runs the barrel through centre wheel, third wheel, fourth wheel, and escape wheel — four stages, total ratio around 1:3600 from barrel turn to escape-wheel turn, so the escape wheel ticks fast while the barrel unwinds slowly over 40+ hours.

The Wheel Train, also called the Wheel train (horology) when used in clocks and watches, only works cleanly if the tooth geometry, depthing, and arbor alignment are right. Pivot holes must sit within roughly 0.02 mm of true centre distance for a watch — drift further and the teeth either bind (centres too close) or skip and chatter (centres too far). Pinion leaves are usually polished to a mirror finish because a single pinion in a watch sees the entire torque of the train at multiplied speed, and surface roughness above Ra 0.1 µm chews through lubricant in months. If you notice the seconds hand stuttering or the watch losing amplitude over a few days, the most common cause isn't the escapement — it's a worn pivot or a dry pinion deep in the train.

Wheel-and-pinion geometry matters more than people think. Clockwork uses cycloidal tooth profiles, not the involute teeth you see on industrial gearing, because cycloidal teeth roll cleanly with very low pinion leaf counts (6, 7, 8 leaves) where involute teeth would undercut and bind. That's the reason a 200-year-old longcase clock still runs — the cycloidal action tolerates wear and dirt that would freeze a modern involute pair.

Key Components

  • Wheel: The larger of each gear pair, typically 50 to 80 teeth in a watch and 60 to 120 in a clock. The wheel is driven by the previous stage's pinion and carries the arbor that drives the next pinion. Wheel rims are usually 0.10 to 0.30 mm thick brass with crossings (spokes) cut to reduce inertia.
  • Pinion: The small driving gear, usually 6 to 12 leaves cut directly into hardened steel arbor stock. Pinions transmit the multiplied speed and reduced torque of each stage. Leaves must be polished to Ra ≤ 0.1 µm — anything rougher accelerates wear because the pinion sees more tooth engagements per hour than any other part.
  • Arbor: The shaft that carries one wheel and one pinion, running between two pivot holes. Arbor pivot diameters are typically 0.08 to 0.12 mm in a wristwatch — smaller than a human hair — and must run in jewelled bearings to survive. Lateral play above 0.02 mm causes depthing errors and audible knocking.
  • Pivot and bearing (jewel): Each arbor end runs in a synthetic ruby jewel hole. The jewel reduces friction by an order of magnitude versus brass and resists wear from the steel pivot. A typical 17-jewel watch has jewels at every train pivot for exactly this reason.
  • Tooth profile (cycloidal or involute): Horological trains use cycloidal teeth for low-leaf pinion compatibility and dirt tolerance. Industrial wheel trains use involute teeth because they accept centre-distance error without changing the velocity ratio — important when the housing isn't machined to watch tolerances.

Real-World Applications of the Wheel Train

Wheel trains show up wherever a slow, high-torque source needs to drive a faster, lower-torque output — or vice versa. The Wheel train (horology) is the classic case, but the same principle drives industrial gearboxes, music boxes, kitchen timers, and mechanical counters. The dominant industry is still watchmaking, where every mechanical movement on the planet runs a wheel train between barrel and escapement.

  • Watchmaking: The going train in an ETA 2824-2 automatic movement — barrel, centre wheel, third wheel, fourth wheel, escape wheel — running at 28,800 BPH from a barrel that turns roughly once every 7 hours.
  • Clockmaking: The eight-day going train in a Howard Miller longcase clock, stepping a falling weight (one turn of the great wheel per 24 hours) up to roughly 14,400 escapement beats per hour.
  • Music boxes: The wheel train in a Reuge 72-note cylinder music box, gearing the spring barrel down to the cylinder shaft so a 30-second tune plays at the correct tempo.
  • Mechanical counters: The Veeder-Root rotary counter wheel train, where each digit wheel advances 1/10 turn per full revolution of the wheel below it — a 10:1 wheel train per digit.
  • Kitchen and darkroom timers: The wheel train inside a vintage Lux Minute Minder, gearing the mainspring down to a 60-minute dial through three stages.
  • Industrial low-power gearmotors: The reduction train in a Maxon GP series planetary gearhead, where multiple wheel-and-pinion stages step a 12,000 RPM brushed DC motor down to usable output speeds for instrumentation drives.

The Formula Behind the Wheel Train

The train ratio formula tells you how many turns of the input give one turn of the output. At the low end of the typical horological range — say a single wheel-pinion stage at 6:1 — you get gentle reduction with minimal friction loss. At the high end, a four-stage going train multiplies stage ratios to reach 1:3600 or more, but every stage adds friction, so a poorly finished train can lose 30% of barrel torque before it reaches the escapement. The sweet spot for a watch going train sits at 4 stages with stage ratios between 6:1 and 8:1, giving long runtime without starving the escapement.

Rtotal = (Zw1 / Zp1) × (Zw2 / Zp2) × ... × (Zwn / Zpn)

Variables

Symbol Meaning Unit (SI) Unit (Imperial)
Rtotal Total train ratio from input to output dimensionless dimensionless
Zw Tooth count on the driven wheel of a stage teeth teeth
Zp Leaf count on the driving pinion of a stage leaves leaves
n Number of wheel-pinion stages in the train count count

Worked Example: Wheel Train in a vintage marine chronometer restoration

A horological restorer in Glashütte is rebuilding the going train of a 1935 Wempe two-day marine chronometer. The barrel turns once every 12 hours when fully wound, and the restorer needs to confirm the escape wheel runs at the original 14,400 BPH (4 Hz, half-seconds beat). The train has four stages with the following tooth/leaf counts: centre wheel 80 / barrel pinion drives it via barrel teeth 96 (stage 1 = 96/12 effective for hour wheel, but for the going train proper start at centre), centre pinion 12 driving third wheel 75, third pinion 10 driving fourth wheel 70, fourth pinion 8 driving escape wheel 15 with 15 teeth (each tooth gives 2 beats).

Given

  • Zw1 / Zp1 = 75 / 12 teeth/leaves
  • Zw2 / Zp2 = 70 / 10 teeth/leaves
  • Zw3 / Zp3 = 64 / 8 teeth/leaves
  • fcentre = 1 / 3600 rev/s (one turn per hour)
  • Teeth on escape wheel = 15 teeth

Solution

Step 1 — at the nominal centre-wheel speed of 1 turn per hour, compute the stage ratios from centre wheel through to escape wheel:

R1 = 75 / 12 = 6.25
R2 = 70 / 10 = 7.00
R3 = 64 / 8 = 8.00

Step 2 — multiply the stage ratios for the total train ratio from centre wheel to escape wheel:

Rtotal = 6.25 × 7.00 × 8.00 = 350

Step 3 — the centre wheel turns once per 3600 seconds, so the escape wheel revolution rate at nominal is:

fesc = 350 / 3600 = 0.0972 rev/s

Each escape-wheel revolution produces 2 × 15 = 30 beats, so beats per hour at nominal:

BPHnom = 0.0972 × 30 × 3600 = 14,400 BPH

That matches the chronometer's design beat exactly. At the low end of the practical range — if the train is dirty and amplitude drops to where the balance only completes 200° instead of 270° — the escape wheel still has to turn at 14,400 BPH or the chronometer loses time, but the torque margin shrinks. At the high end, a builder who substitutes a 76-tooth third wheel by mistake (one tooth high) shifts R1 to 76/12 = 6.333, pushing total ratio to 354.7 and BPH to 14,587 — the chronometer would gain roughly 19 minutes per day. A single mis-cut wheel destroys timekeeping.

Result

Nominal escape-wheel beat rate is 14,400 BPH, exactly the design value for a 4 Hz marine chronometer. In practice this is the half-second tick the navigator hears against the deck clock — slow enough to count by ear, fast enough to interpolate sextant timing to ±0.25 s. Compared to a wristwatch train at 28,800 BPH the chronometer beats half as fast, which is the sweet spot for long-runtime, low-torque, high-stability marine use. If you measure 14,250 BPH instead of 14,400, check three things in order: (1) third-wheel pinion endshake — anything over 0.04 mm causes intermittent depthing slip and a slow rate, (2) escape-wheel pivot polish, since a scuffed pivot drags the wheel and shows up as low amplitude before it shows as bad rate, and (3) miscount of teeth on a replacement wheel — always count three times before installing, because one wrong tooth is invisible to the eye and catastrophic to the rate.

Wheel Train vs Alternatives

Wheel trains aren't the only way to step speed and torque between two shafts. Belt drives, chain drives, and harmonic drives all do the same job with different tradeoffs. Here's how the Wheel train (horology) and its industrial cousin compare to the main alternatives on the dimensions a designer actually cares about.

Property Wheel Train Belt Drive Harmonic Drive
Typical stage ratio range 3:1 to 10:1 per stage 1:1 to 5:1 per stage 30:1 to 320:1 in a single stage
Backlash 0.5° to 2° per stage (more with wear) Effectively zero (toothed belt) to high (V-belt slip) Under 1 arcmin
Efficiency per stage 96-99% (well-finished horological) 95-98% (toothed belt) 70-85%
Lifespan at rated load 50+ years (jewelled watch train) 5,000-20,000 hours (belt fatigue) 10,000-35,000 hours
Cost (production scale) Low for stamped, high for cut/polished horological Very low High — single-source precision component
Best application fit Clocks, watches, low-power instruments Power transmission with shaft offset Robotics joints, precision indexing
Sensitivity to centre-distance error ±0.02 mm (watch), ±0.05 mm (clock) ±2-5 mm (belt tension absorbs it) Fixed by housing — no centre-distance variable

Frequently Asked Questions About Wheel Train

The textbook 1-2% loss per stage assumes ideal lubrication and zero pivot side-load. In a real watch you also pay for: pivot side-load from gravity in vertical positions (adds 3-5% per stage when crown-down), dried oil that's gone gummy after 5+ years (doubles bearing friction), and any depthing error which makes teeth slide instead of roll. A train calculated at 92% efficiency commonly measures 75-80% in a 10-year-old unserviced movement.

Stage ratios above 10:1 force you into pinion leaf counts of 5 or 6, where cycloidal tooth action gets twitchy and any depthing error becomes a rate error. Adding a stage costs you efficiency (each stage ≈ 2-4% loss) but keeps each ratio in the comfortable 6:1 to 8:1 band. Rule of thumb: for total ratios up to ~50:1 use 2 stages, 50-500:1 use 3 stages, above 500:1 use 4. The four-stage going train in nearly every wristwatch follows exactly this logic.

Tooth count alone doesn't define a wheel. Module (or diametral pitch), tooth profile (cycloidal vs involute), addendum height, and pressure angle must all match the original. A wheel cut to a 0.08 module won't mesh cleanly with a pinion designed for 0.07 module even at the correct centre distance — teeth either tip-strike or fail to engage at the pitch line. Always match all five parameters, and verify by rotating the train slowly with tweezers feeling for stiff spots before reassembly.

Involute teeth need at least 12-14 teeth per gear before undercutting kicks in and weakens the tooth root. Watch pinions have 6, 7, or 8 leaves — way below the involute threshold. Cycloidal profiles allow leaf counts as low as 5 with full-strength teeth, and they tolerate dirt because the contact rolls instead of sliding. The price you pay is sensitivity to centre distance: cycloidal velocity ratio changes if pivot holes drift, where involute is immune. In a watch you machine the plates accurately enough that centre-distance error never becomes an issue.

No. Escape wheel tooth count is set by the train ratio designed into the centre, third, and fourth wheels — change only the escape wheel and you'll either run the new beat rate at the wrong frequency or starve the escapement of torque. To raise beat rate from 18,000 to 21,600 BPH you need to redesign at least two stages and verify the pallet fork geometry still impulses correctly. This is why hi-beat conversions are full re-trains, not drop-in swaps.

For a wristwatch, pivot-to-jewel clearance under 0.005 mm is the spec. At 0.010 mm you'll see amplitude drop 10-15° in vertical positions because the arbor cocks and increases tooth-flank sliding. At 0.020 mm the wheel can lift enough to skip a tooth under shock, which shows up as the watch occasionally jumping minutes ahead. Use a jewel hole gauge — eyeballing it through a loupe is not enough. Worn jewel holes get re-jewelled, not re-bored, because brass plate jewels lose their press fit if you enlarge the seat.

Vertical orientation transfers all arbor weight onto the pivot side-flanks instead of distributing it across the pivot end. A pivot polished to Ra 0.05 µm flat will drag noticeably at Ra 0.15 µm against a side-load. Three causes in order of frequency: (1) pivot polish degraded by old oil residue — clean and repolish, (2) jewel hole worn oval, fine in the flat plane but binding in the vertical, (3) wheel out of flat, so it wobbles and binds at one rotational position. Spin each wheel solo with the others removed to isolate which arbor is the culprit.

References & Further Reading

  • Wikipedia contributors. Wheel train. Wikipedia

Building or designing a mechanism like this?

Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.

← Back to Mechanisms Index
Share This Article
Tags: