A star-wheel escapement is a horological gating mechanism that uses a multi-pointed star-shaped wheel — typically 5, 6, or 8 points — to release one increment of stored mainspring or weight energy per oscillator beat. It runs at the same beat rate as the controller, often 14,400 to 28,800 vph in watch sizes and 3,600 to 7,200 vph in clock sizes. It exists to deliver a clean, repeatable impulse to a pendulum or balance with minimal recoil. You see star-wheel layouts in jumping-hour modules, repeater work, and certain Breguet-style retrograde complications.
Star-wheel Escapement Interactive Calculator
Vary the number of star points and released beats to see the escapement tooth pitch, total angular advance, and wheel revolutions.
Equation Used
The star-wheel escapement advances by one tooth pitch each beat. With N evenly spaced points, the pitch angle is 360 / N, so a 6-point star advances 60 degrees per beat.
FIRGELLI Automations - Interactive Mechanism Calculators.
- One star tooth is released per oscillator beat.
- Star points are uniformly spaced around 360 degrees.
- No missed beats or double-jumps are included.
Inside the Star-wheel Escapement
The star-wheel escapement works by parking a spring-loaded jumper or impulse pallet against the V-shaped notch between two adjacent points of the star. Energy from the going train wants to rotate the star, but the jumper holds it locked until the controller — pendulum or balance — arrives at the impulse position and lifts the jumper clear. The star then snaps forward by exactly one tooth pitch, the jumper drops into the next notch, and the cycle repeats. That snap is what gives the mechanism its characteristic tick and what delivers the energy pulse back to the oscillator.
Geometry is everything here. The flank angle of each star point typically sits between 30° and 45° measured from the radial centreline. Go shallower than 30° and the jumper struggles to lock — you get backlash and the star drifts mid-beat. Go steeper than 45° and the unlocking force climbs sharply, robbing the oscillator of amplitude. If you notice the watch losing 30 to 90 seconds a day with healthy mainspring torque, the star flank angle or jumper spring rate is almost always the culprit. The other classic failure is jumper-spring fatigue: once the spring loses 10 to 15% of its rated force, the star can double-jump under shock, and the complication advances two steps instead of one.
Lubrication matters less than you would expect — the star tip and jumper roller are usually run dry or with a microdot of 9415 thixotropic grease. Too much oil migrates onto the impulse face and changes the friction coefficient, which shows up as positional rate variation between dial-up and crown-down.
Key Components
- Star Wheel: The toothed wheel itself, with 5 to 8 symmetrical points machined to a flank angle of 30° to 45°. Tooth pitch must be uniform within ±3 arc-minutes; any more and you see beat-error rates climb into the 10 ms range.
- Jumper Spring: A leaf or beam spring carrying a polished roller or V-tip that rides between star points. Spring rate is typically 0.05 to 0.15 N/mm — enough to hold the star against drive-train torque, light enough that the oscillator can lift it without losing more than 15° of amplitude.
- Impulse Pallet: Mounted on the balance staff or pendulum crutch, this is the part that contacts the star tip during the unlocking arc. Surface finish must be Ra ≤ 0.1 µm — anything rougher and you double the impulse-phase friction.
- Drive Pinion: Couples the going train to the star wheel. Backlash between pinion and star arbor must stay below 0.02 mm; sloppy coupling here causes the star to chatter under impulse and shows up as audible double-ticking.
- Banking Pin or Limit Stop: Restricts the star's overshoot when the jumper releases. Set the gap to 0.05 to 0.10 mm beyond one tooth pitch; tighter than that and thermal expansion can lock the star, looser and you risk skip-jumps under shock.
Real-World Applications of the Star-wheel Escapement
Star-wheel escapements show up wherever a designer needs a clean indexed advance synchronised to an oscillator, but the load is too low or the geometry too constrained for a conventional Swiss lever. The mechanism is commonly cited in jumping-hour and retrograde-display work, in some minute-repeater all-or-nothing pieces, and in a handful of experimental wristwatch escapements where designers want to reduce sliding friction.
- Wristwatch complications: Jumping-hour modules in pieces like the F.P. Journe Vagabondage I, where a star wheel indexes the hour disc once every 60 minutes.
- Pocket watches: All-or-nothing safety mechanisms in Patek Philippe minute repeaters use a star-wheel detent to prevent partial chime activation.
- Precision regulators: Some experimental Daniels-school regulators have used star-wheel detents in remontoire stages to deliver constant force to the escapement proper.
- Tourbillon variants: Andreas Strehler's Sauterelle a Lune Perpetuelle uses star-wheel indexing for its lunar display, advancing one step per sidereal day.
- Retrograde displays: Maurice Lacroix Masterpiece Calendrier Retrograde uses a star wheel in the date-jump train to snap the retrograde hand back at month-end.
- Clockwork automata: Jaquet-Droz singing-bird boxes use small 6-point star wheels to gate cam advances in the bellows-and-whistle train.
The Formula Behind the Star-wheel Escapement
The core question for any star-wheel design is: how much torque does the jumper spring resist, and how much amplitude does the oscillator lose during unlocking? At the low end of the typical range (5-point star, light spring) you get easy unlocking but poor positional stability — the star drifts under shock. At the high end (8-point star, stiff spring) lock is rock-solid but you bleed 25%+ of balance amplitude per beat. The sweet spot for most wristwatch-scale work sits around 6 points and a spring force that costs you 10 to 15° of amplitude.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Munlock | Torque required at the star arbor to lift the jumper out of the notch | N·mm | oz·in |
| Fspring | Jumper spring preload force at the contact point | N | lbf |
| rstar | Effective radius from star centre to jumper contact point | mm | in |
| α | Star-point flank angle measured from radial centreline | degrees | degrees |
| φ | Friction angle, typically arctan(0.10 to 0.15) for hardened steel on steel | degrees | degrees |
Worked Example: Star-wheel Escapement in a 6-point jumping-hour module
A boutique movement designer in Le Locle is prototyping a 6-point star-wheel jumping-hour module for a 32 mm calibre. The star is 4.2 mm in radius, the jumper spring delivers 0.18 N at the contact point, and the flank angle is set at 38°. They need to know the unlocking torque so they can verify the going train can deliver it without dropping balance amplitude below 240°.
Given
- Fspring = 0.18 N
- rstar = 4.2 mm
- α = 38 degrees
- φ = 8.5 (μ ≈ 0.15) degrees
Solution
Step 1 — combine the flank and friction angles for the nominal 38° design:
Step 2 — compute the nominal unlocking torque:
That sits comfortably inside the torque budget of a typical 28,800 vph calibre, which can deliver 1.0 to 1.4 N·mm at the escape pinion before amplitude collapses. Step 3 — at the low end of the typical flank-angle range, drop α to 30° and recompute:
Easier unlock, but at 30° the star starts to back-drive under wrist shock — you will see the hour disc creep forward by half a step on a hard tap, which is unacceptable on a jumping-hour. Step 4 — at the high end, push α to 45°:
Now you are eating nearly the entire torque budget just to unlock the star. Amplitude will drop 30°+ at the moment of jump, and on a 240° starting amplitude you would dip into the 200° range — exactly where rate stability falls apart. The 38° nominal design is the sweet spot: secure lock, manageable unlock cost.
Result
Nominal unlocking torque is 0. 797 N·mm at α = 38°, which leaves about 0.3 N·mm of headroom in a typical 28,800 vph going train — enough to keep balance amplitude above 240° during the jump. The low-end 30° design drops to 0.601 N·mm but invites back-drive under shock; the high-end 45° design climbs to 1.021 N·mm and crushes amplitude into the 200° danger zone. If your prototype measures torque 25%+ above the predicted 0.797 N·mm, check three things: (1) jumper-spring preload that has crept past 0.22 N from over-bending during assembly, (2) galled or rough star points where surface finish has degraded above Ra 0.2 µm, or (3) a star arbor pivot running dry without 9010 oil, which can double the effective friction angle φ.
Choosing the Star-wheel Escapement: Pros and Cons
Star-wheel escapements compete with detent, lever, and co-axial escapements depending on the application. The choice comes down to torque budget, isochronism, shock resistance, and how much sliding friction you can tolerate.
| Property | Star-wheel Escapement | Swiss Lever Escapement | Detent Escapement |
|---|---|---|---|
| Typical beat rate | 3,600–28,800 vph | 18,000–36,000 vph | 14,400–18,000 vph |
| Daily rate accuracy (good example) | ±10 to ±30 s/day | ±2 to ±10 s/day | ±0.5 to ±3 s/day |
| Shock resistance | Moderate — risk of double-jump | Excellent (with safety pin) | Poor — set risk under shock |
| Sliding friction in impulse | Low — mostly rolling | High — pallet stones slide | Very low — single impulse arc |
| Manufacturing complexity | Moderate | High but well-tooled | Very high — hand-fit |
| Best application fit | Jumping/retrograde complications | General wristwatch and clock | Marine chronometers, precision pocket watches |
| Service interval | 5–7 years | 5–7 years | 3–5 years |
Frequently Asked Questions About Star-wheel Escapement
This is almost always a jumper-spring force problem, but in the opposite direction from what most people first check. If the spring is too soft — below about 0.10 N at the contact point on a 4 mm star — the star has enough kinetic energy from the drive-train torque to overshoot the next notch entirely and drop into the one after. The fix is to add 20-30% preload to the jumper spring or fit a banking pin that physically limits star rotation to one tooth pitch plus 0.05 mm.
The other cause is a banking-pin gap that has opened up over time from repeated impacts. Measure the gap with a feeler gauge; anything over 0.15 mm on a wristwatch-scale module needs the pin reset.
The point count needs to match the display divisions, but it also drives the unlocking torque budget. A 5-point star has a 72° pitch — large angular advances per beat, easier to see and audibly tick, but the impulse spike is large enough that you cannot easily run it off a balance wheel. Use 5-point for cam-driven retrograde displays where a separate spring resets the hand.
An 8-point star at 45° pitch gives smoother, smaller advances and works directly off oscillator impulse, but each beat is a smaller fraction of the full retrograde sweep, which can look mechanically lazy on a 60-second retrograde. 6-point is the most common compromise — used in the Maurice Lacroix Masterpiece line and several Chronoswiss retrograde calibres.
Positional rate variation in a star-wheel system usually comes from oil migration onto the impulse face. In dial-up, gravity pulls oil away from the star tip; in crown-down, it pools at the contact point and changes the friction coefficient by 30 to 50%. That shifts the effective friction angle φ in the unlocking equation and changes how much amplitude the oscillator loses per beat.
The diagnostic is simple: clean the star tip and jumper roller with hexane, run the watch dry for a week, and re-check positional rates. If variation drops below ±5 ms, you have your answer — switch to a microdot of 9415 grease applied with a 0.3 mm oiler, no more.
You can, but it rarely helps and usually hurts. The lever escapement in a calibre like the ETA 2824-2 is tuned around a specific impulse angle (42°) and amplitude window (270-310°). Drop a star-wheel detent into the impulse train and you change the load on the balance during the unlocking arc, which detunes the hairspring and shifts the rate by 30 to 90 seconds a day in the wrong direction.
Star-wheel makes sense when designed in from scratch with matched balance inertia and hairspring stiffness. As a retrofit, it is almost always a downgrade unless you are also re-cutting the balance and reselecting the hairspring.
Ra 0.1 µm or better on the contact flanks. The friction angle φ in the unlocking torque equation assumes μ ≈ 0.15 for hardened-steel-on-hardened-steel with a polished finish. At Ra 0.4 µm — typical of a fine-ground but unpolished surface — μ climbs to 0.25 and your unlocking torque jumps by 40%.
Practically, this means the star has to be black-polished after hardening. Diamond paste at 1 µm followed by 0.25 µm on a tin lap is the workshop-standard sequence. Skip this and the calculated torque budget is meaningless.
If you have removed the jumper and the star still feels notchy, the problem is in the arbor pivots or the drive pinion mesh, not the escapement geometry. Check the upper pivot first — a bent arbor will create a periodic tight spot every revolution. A 0.01 mm runout at the pivot tip translates to noticeable binding when the star is loaded.
The second suspect is the drive-pinion depthing. If the pinion is too deep into the star arbor wheel, you get tooth-by-tooth resistance that mimics jumper notching. Loosen the pinion bearing eccentric and back it off until the wheel spins freely under finger pressure with no sticky points.
References & Further Reading
- Wikipedia contributors. Escapement. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
