Variable Rotary Motion is a power transmission scheme where a uniformly rotating input shaft drives an output shaft whose angular velocity changes within each revolution. The motion principle relies on a non-constant transmission ratio — typically achieved with elliptical or non-circular gears, eccentric linkages, or cam-modulated couplings — so the output speeds up and slows down on a repeating cycle. Engineers use it to time a tool against a moving workpiece, dwell at a specific angle, or shape torque delivery. You see it in flying-shear cutoffs, rotary printing presses, and wire-forming machines where output position must match a moving target.
Variable Rotary Motion Interactive Calculator
Vary the input speed and instantaneous pitch radii to see the elliptical gear velocity ratio, output speed, and timing response.
Equation Used
The article diagram gives the instantaneous velocity relationship for an elliptical gear pair: output speed equals constant input speed multiplied by the current driver-to-follower pitch radius ratio. Larger r1 with smaller r2 produces a faster output phase; the opposite produces a slower phase.
- Instantaneous pitch radii are measured at the mesh point.
- Gear teeth maintain conjugate action with no slip.
- Output speed magnitude is shown; rotation direction is opposite for external gears.
- Radii represent one operating angle of the non-circular gear cycle.
Operating Principle of the Variable Rotary Motion
The core idea is simple: keep the input rotating at a steady RPM but make the output's angular velocity vary as a function of input angle θ. With a pair of elliptical (non-circular) gears mounted on parallel shafts, the radius of contact at the pitch point changes through the rotation, and so does the instantaneous velocity ratio. When the long radius of the driver mates against the short radius of the follower, the output runs faster than the input. Half a revolution later, the geometry inverts and the output runs slower. The whole cycle repeats every revolution, smoothly, with no clutching, no indexing pawl, and no hard stops.
Why build it this way instead of just using a servo? Because the motion is purely mechanical and timed by geometry. There's no encoder loop to tune, no drive lag, no jitter from a controller missing a step. On a high-speed rotary printing press running 12 m/s web speed, you cannot afford a servo to glitch — the elliptical gear pair gives you deterministic non-uniform rotation locked to the line shaft. The same logic applies to cam-modulated rotation in packaging machines and to eccentric four-bar linkages used as Geneva mechanism alternatives.
Tolerance matters more than people expect. If the gear pair's centre distance is off by even 0.05 mm on a 60 mm pitch ellipse, the conjugate action breaks down and you get backlash that varies through the cycle — meaning the output shaft chatters at the slow phase and lags at the fast phase. Common failure modes are bearing wear letting the centres drift, tooth pitting at the high-velocity-ratio zone (where Hertzian contact stress peaks), and shaft windup if the output inertia is too high for the peak angular acceleration. If you notice torque ripple where there shouldn't be any, suspect the centre distance first.
Key Components
- Driver (input) non-circular gear: Mounted on the constant-speed input shaft, this gear carries a non-circular pitch curve — most often elliptical with eccentricity 0.2 to 0.5. The pitch curve must be mathematically conjugate to the follower; a 0.01 mm profile error at the long-radius zone produces measurable velocity ripple.
- Follower (output) non-circular gear: Identical or matched-pair pitch curve mounted on the output shaft. The two pitch curves must roll without slipping, so centre distance is fixed and equal to the sum of conjugate radii at every angle. On a 60 mm major-axis elliptical pair, centre distance must hold ±0.02 mm or conjugate action degrades.
- Bearings and shaft supports: Bearings carry cyclic radial loads that vary with the instantaneous velocity ratio. Peak load occurs at the maximum-radius contact point, often 2 to 3× the mean. Use angular contact bearings if axial slop would shift the gears off centre.
- Timing reference (key or pin): Both gears must be phased correctly on assembly. A keyway or dowel pin sets the angular zero so the slow phase aligns with the work event — for example, the dwell phase synchronised with a print impression on a rotary press.
- Output shaft load (driven member): Whatever the output drives — a cutter drum, an indexing carousel, a print cylinder — sees a non-constant angular acceleration. Its polar moment of inertia must be low enough that peak torque demand stays inside the gearset's rating. Oversize the inertia and you crack a tooth at the fast-phase reversal.
Industries That Rely on the Variable Rotary Motion
Variable Rotary Motion shows up wherever an output needs to track a moving target, dwell at a specific angle, or deliver shaped torque without resorting to electronic servos. The mechanism trades flexibility for determinism — once cut, the gear or cam profile is locked, but it never drifts, never needs tuning, and never misses a cycle. Industries that run 24/7 production at high line speeds rely on it for exactly this reason.
- Rotary printing: Heidelberg Speedmaster and Koenig & Bauer Rapida sheet-fed presses use cam-modulated gripper drums where rotation accelerates during sheet hand-off and slows during the impression dwell.
- Packaging machinery: Bosch Pack Technology cartoners use elliptical gear pairs to match flying-shear cutter rotation to incoming web speed without a servo correction loop.
- Wire and spring forming: Wafios spring coilers drive the wire feed cam through a non-circular gear set so feed velocity matches the coil pitch as it changes mid-spring.
- Textile machinery: Picanol Sumo air-jet looms run reed-beat-up motion through a four-bar variable-rotation drive to dwell at the cloth fell during pick insertion.
- Rotary indexing: Mikron NRG-50 rotary transfer machines use cam-driven rotation as a Geneva mechanism alternative, indexing 12 stations at 60 cycles per minute with smoother acceleration than a Geneva drive can deliver.
- Agricultural machinery: John Deere round-baler pickup reels use eccentric drive geometry so tine velocity slows at ground contact and speeds up during retraction.
The Formula Behind the Variable Rotary Motion
The instantaneous velocity ratio for an elliptical gear pair changes with input angle θ. At low eccentricity (e ≈ 0.1) the output barely varies — you get maybe 20% speed swing peak-to-peak, useful only for gentle modulation. At nominal eccentricity (e ≈ 0.3) you get a 2:1 speed ratio between fast and slow phase, which suits most flying-shear and indexing jobs. Push e past 0.5 and you exceed 4:1 speed swing, but tooth contact stress at the long-radius zone climbs sharply and the gears need premium case-hardened steel. The sweet spot for industrial pairs sits between e = 0.25 and e = 0.4.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| ωout(θ) | Instantaneous output angular velocity at input angle θ | rad/s | rev/min |
| ωin | Constant input angular velocity | rad/s | rev/min |
| e | Eccentricity of the elliptical pitch curve (0 = circular, →1 = extreme) | dimensionless | dimensionless |
| θ | Input shaft angle measured from the long-axis alignment | rad | deg |
Worked Example: Variable Rotary Motion in a flying-shear cutoff on a corrugated board line
You are specifying an elliptical gear pair to drive a flying-shear cutter on a BHS Corrugated single-facer line in Weiherhammer Germany. The web travels at 200 m/min and the cutter must briefly match web speed during the cut, then return to a faster idle rotation between cuts. Input shaft runs at a constant 180 RPM from the line's main drive. You're choosing eccentricity e = 0.30 and need to verify the speed swing across the cycle.
Given
- ωin = 180 RPM
- e = 0.30 dimensionless
- θ range = 0 to 360 deg
Solution
Step 1 — convert input speed to a working unit and compute the slow-phase output (θ = 180°, cos θ = −1):
Step 2 — compute the fast-phase output at θ = 0° (cos θ = +1):
Step 3 — compute the speed-swing ratio across the cycle:
At the low-eccentricity end of the typical range (e = 0.15) the same input would give roughly 153 to 211 RPM — a 1.4:1 swing — barely enough to track a flying-shear web speed change. At nominal e = 0.30 you get the 96.9 to 334.3 RPM swing computed above, which lines up cleanly with the 200 m/min web requirement during the cut phase. Push to e = 0.50 and the swing climbs to roughly 60 to 540 RPM (9:1) — the cutter would slam into the cut so hard you'd see tooth pitting on the driver gear inside 500 hours, and the line shaft would need a flywheel to absorb the torque ripple.
Result
The output swings from 96. 9 RPM (slow phase, during the cut) up to 334.3 RPM (fast phase, between cuts) — a 3.45:1 ratio across each input revolution. In practical terms this means the cutter dwells near web speed for about 60° of input rotation, long enough for a clean square cut on 8 mm corrugated, then accelerates clear of the web before the next sheet arrives. At the low end (e = 0.15) the swing collapses to 1.4:1 and the cutter cannot reach web speed at the slow phase — you'll see torn edges. At the high end (e = 0.50) the cycle is overdriven and tooth contact stress passes 1,400 MPa at the long-radius mesh. If your measured swing comes in below predicted, suspect (1) centre distance opened up by 0.05 mm or more from worn shaft bearings, letting backlash steal angular position at the fast phase, (2) gear-pair phasing off by 5° at assembly, which shifts where the slow phase lands relative to the cut event, or (3) line shaft flexing under cut-load reaction torque, dragging the input below its nominal 180 RPM during the impact.
Choosing the Variable Rotary Motion: Pros and Cons
Variable Rotary Motion competes against three other ways of getting non-uniform output: a Geneva drive for hard intermittent motion, a cam-driven oscillator for precise dwell, and a servo motor with electronic cam profiling. Each one targets a different cost, speed, and precision envelope. The right choice depends on your cycle rate, your dwell-accuracy budget, and how often you'll change the motion profile.
| Property | Variable Rotary Motion (elliptical gears) | Geneva Drive | Servo with electronic cam |
|---|---|---|---|
| Maximum cycle rate | Up to 600 RPM continuous | Limited to ~300 RPM before dwell shock | Up to 3000 RPM with sized motor |
| Output smoothness | Continuous sinusoidal velocity, no shock | Hard start/stop, high jerk at index | Smooth, profile-dependent |
| Position accuracy | ±0.1° at output (geometry-locked) | ±0.5° (slot/pin clearance) | ±0.01° (encoder-resolution-limited) |
| Profile flexibility | Fixed at gear-cut time | Fixed (4, 6, 8 station only) | Fully reprogrammable |
| Capital cost (typical) | $800–$3,000 per pair | $200–$1,500 | $2,500–$8,000 with drive |
| Maintenance interval | 10,000+ hours, oil-bath lubrication | 5,000 hours, pin/slot wear | Bearing service ~20,000 hours |
| Best application fit | High-speed continuous lines, fixed cycle | Low-speed indexing, simple stations | Variable-product lines, short runs |
Frequently Asked Questions About Variable Rotary Motion
A flywheel on the output shaft helps with torque ripple but does not help with tooth contact stress at the long-radius mesh — and that's what limits the catalogue RPM. Hertzian contact stress scales with the square root of load divided by radius of curvature, and at the long-radius point of an elliptical gear the curvature is sharper than the equivalent circular pair, so stress is already 30–50% higher there at rated load.
If you push past the rated RPM you'll see pitting initiate at the long-radius zone within a few hundred hours. Flywheel mass on the output also raises peak angular acceleration torque at the fast-phase entry, which makes the problem worse, not better. If you need more speed, spec a case-hardened pair (60 HRC) or move to a larger module.
Most likely the slow-phase dwell isn't landing where you think it is. The slow phase of an elliptical pair is centred on θ = 180° from the long-axis alignment of the driver, and if the gears were keyed onto the shafts even 3–5° off from the design phase, the cut event lands on the accelerating flank of the velocity curve instead of the dwell. The cutter is then accelerating during the cut, so the cut-to-cut length drifts.
Check the timing mark alignment between the driver and the line-shaft index. A dial indicator on the cutter at top-dead-centre against an encoder on the input shaft will tell you within 0.1° whether the phasing matches the drawing.
Choose the elliptical pair when the motion profile is fixed for the life of the machine, the cycle rate is above ~300 RPM, and you cannot tolerate any drive-loop variation between cycles. Print impression cylinders, web cutoffs, and high-speed cartoners all fit that profile.
Choose the servo when you change product format more than once a week, your line speed varies by more than 20%, or you need to sync against a vision system. The servo costs 2–3× more in capital and adds tuning complexity, but it pays back fast on flexible lines.
Above e = 0.5 you start running into involute interference at the short-radius zone — the tooth pressure angle effectively swings outside the 14.5°–25° workable range as the pitch radius shrinks, and the teeth either undercut or skip contact. Quality gear shops will quote e = 0.4 as a comfortable maximum and e = 0.5 as the practical ceiling using corrected tooth profiles.
Beyond that you're looking at form-cut custom teeth or a different mechanism entirely — a cam-modulated coupling or a four-bar linkage with variable-length crank handles those extreme ratios more cleanly.
Two-per-rev torque spikes from a single elliptical pair almost always trace to an out-of-round driver or follower — the pitch curve has been ground or hobbed with a profile error that adds a second-harmonic component on top of the intended elliptical shape. A good metrology check is to mount the gear on a CMM and trace the pitch curve against the design ellipse; deviations above 0.02 mm at any point will show up as a second-harmonic torque ripple on the output.
The other possibility is shaft misalignment — if the input and output shafts are not parallel within 0.05 mm/m, the contact line walks across the tooth face twice per revolution, producing a similar two-per-rev signature.
Size for peak instantaneous torque, not average. The peak torque demand at the input occurs at the fast-phase entry, where you're accelerating the output inertia from slow to fast. Compute peak torque as Jout × αpeak, where αpeak is the maximum angular acceleration of the output — at e = 0.3 this runs roughly 4× the mean ωin2.
A common mistake is sizing the motor for the average load and then watching it stall on cold starts when oil viscosity adds drag at the worst phase. Add a 1.5× service factor on top of the peak calculation and pick the next motor frame size up.
References & Further Reading
- Wikipedia contributors. Non-circular gear. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
