Slow Forward and Quick Back Circular Motion: How the Elliptical Gear Quick-Return Mechanism Works

← Back to Engineering Library

Slow forward and quick back circular motion is a gear-driven scheme that produces one full output rotation per input rotation, but with the output crawling through most of its travel and snapping back through the rest. A pair of mating non-circular gears — usually elliptical, mounted on offset centres — does the work by varying the instantaneous gear ratio across the cycle. The point is to give a tool more time on the workpiece during the cutting stroke and waste less time on the idle return. Shapers, slotting machines and indexing drives use this trick to lift throughput by 30-50% versus a constant-speed drive.

Slow Forward and Quick Back Circular Motion Interactive Calculator

Vary the elliptical gear axis ratio and semi-major axis to see the quick-return speed swing and required gear center distance.

Center Distance
--
Slow Ratio
--
Fast Ratio
--
Speed Swing
--

Equation Used

D_c = 2a; q_fast ~= 1 + 1.5(A - 1); q_slow = 1 / q_fast; swing = q_fast / q_slow

The calculator estimates the slow and fast instantaneous velocity ratios for a matching elliptical gear pair. The required shaft spacing is the article relation D_c = 2a. For the stated 2:1 ellipse example, the ratio range is calibrated to about 0.4x slow and 2.5x fast.

FIRGELLI Automations - Interactive Mechanism Calculators.

  • One output revolution occurs for each input revolution, so the average ratio is 1:1.
  • Both elliptical gears have matching pitch curves and are mounted at the required center distance.
  • Speed swing is an article-level teaching estimate calibrated to the stated 2:1 ellipse example.
  • Final tooth geometry, backlash, and pressure angle must be checked with detailed non-circular gear design.
Elliptical Gear Quick-Return Mechanism Static engineering diagram showing two meshing elliptical gears that produce variable angular velocity - slow forward motion and quick return motion from a constant speed input. DRIVER DRIVEN Input Output Long radius Short radius Mesh Center distance = 2a Angular Velocity (Same Time) Constant speed 90° swept Variable speed ~30° swept (SLOW) Long radius meets short radius → SLOW output
Elliptical Gear Quick-Return Mechanism.

Inside the Slow Forward and Quick Back Circular Motion

The whole thing rests on one idea — if you mesh two elliptical gears on their geometric centres, the gear ratio at any instant depends on which part of each ellipse is currently in contact. When the long radius of the driver meshes against the short radius of the driven gear, the output spins fast. Half a turn later the radii swap, and the output crawls. One input revolution still produces one output revolution, so the average ratio is 1:1, but the instantaneous angular velocity ratio swings between roughly 0.4 and 2.5 for a standard 2:1 axis ratio ellipse. That swing is where the slow-forward and quick-back behaviour comes from.

Getting the pitch curves right is where most builds fail. The two ellipses must have identical eccentricity, identical perimeter, and they must be mounted on centres separated by exactly 2a — twice the semi-major axis. If your centre distance is off by even 0.3 mm on a 60 mm gear, the teeth will bind at the long-radius mesh point and skip at the short-radius point. The teeth themselves are not standard involutes either — each tooth profile is computed against the local pitch curve, so tooth thickness and pressure angle vary slightly around each gear. Cut them on a CNC hob or wire EDM, never on a standard generating cutter.

The most common failure modes are pitch-curve mismatch from sloppy CNC tolerancing, backlash that varies with rotation angle, and bearing wear at the high-radius mesh where tooth load peaks. If you hear a periodic clack once per output revolution, that is almost always tooth interference at the long-radius crossover — recheck centre distance and tooth correction factor before blaming the bearings.

Key Components

  • Driver Elliptical Gear: The input gear, mounted on a shaft that passes through one focus of the ellipse — not the geometric centre — to convert uniform input rotation into varying surface velocity at the pitch curve. Eccentricity is typically 0.3 to 0.6 for useful quick-return ratios; below 0.2 the speed variation is too small to bother with.
  • Driven Elliptical Gear: Identical pitch curve to the driver, mounted on a centre offset by 2a (twice the semi-major axis) so the two ellipses always mesh tangentially. Profile shift on individual teeth must match within ±0.02 mm or you get angular position error that walks around the cycle.
  • Centre Distance Frame: Holds the two shafts at the exact computed centre distance — usually 2a within ±0.05 mm for a 50 mm gear. Any thermal growth or frame flex shows up immediately as binding at one mesh point and backlash at the opposite mesh point.
  • Output Shaft and Coupling: Transmits the now-non-uniform rotation to the working element — ram, slotting tool, or indexing dial. Must include enough torsional stiffness that the inertia spike during the quick-return phase doesn't wind up the shaft and smear the velocity profile.
  • Tooth Profile (Modified): Each tooth is generated against the local pitch curve radius, so tooth thickness varies around the gear by typically ±5%. Standard involute cutters cannot make these — you need CNC hobbing with continuous correction or wire EDM from a CAD pitch curve.

Industries That Rely on the Slow Forward and Quick Back Circular Motion

You see this mechanism wherever a tool spends real time on a workpiece during the working stroke and contributes nothing during the return. The aim is always the same — pack more cutting time into each cycle without raising the average shaft speed. Modern servo systems can fake the same velocity profile electronically, but a properly cut elliptical gear pair gives you the same behaviour with no controller, no encoder, and no software, which is why the mechanism still earns its keep on dedicated production machines.

  • Metalworking: Ram drive on a Cincinnati 24-inch shaper, where the ram cuts on the forward stroke and returns idle — elliptical gears give roughly a 3:2 cutting-to-return time ratio.
  • Sheet Metal: Slotting attachment on a Pratt & Whitney slotter, where the slow downstroke clears chip and the quick upstroke recovers tool position before the next index.
  • Packaging: Cross-seal jaw drive on a Bosch SVE 2520 vertical form-fill-seal machine, where the jaws dwell longer on the seal than on the open-return phase.
  • Printing: Reciprocating ink-distribution roller drives on Heidelberg sheet-fed presses, where slower forward travel improves ink film transfer.
  • Textile Machinery: Picking-stick drive on traditional Sulzer projectile looms — gives the projectile a slow accelerating push and a quick return for the next pick.
  • Mechanical Watches: Variable-rate motion in horological complications such as the Breguet sympathique escapement variants, where non-circular gears modulate transmission rate across the cycle.

The Formula Behind the Slow Forward and Quick Back Circular Motion

The defining number for this mechanism is the quick-return ratio Q — the time spent on the slow forward stroke divided by the time spent on the quick return. Q depends only on the eccentricity e of the elliptical gears. At the low end of the practical range, e = 0.2 gives Q ≈ 1.2 — barely worth the cost of cutting non-circular gears. At nominal e = 0.5 you get Q ≈ 2.0, the sweet spot used on most production shapers. Push to e = 0.7 and Q reaches roughly 3.5, but tooth contact stress at the long-radius mesh climbs steeply and gear life drops fast.

Q = (1 + e) / (1 − e)

Variables

Symbol Meaning Unit (SI) Unit (Imperial)
Q Quick-return ratio — forward stroke time divided by return stroke time dimensionless dimensionless
e Eccentricity of the elliptical pitch curve, e = √(1 − b2/a2) dimensionless dimensionless
a Semi-major axis of the ellipse mm in
b Semi-minor axis of the ellipse mm in

Worked Example: Slow Forward and Quick Back Circular Motion in a tobacco-leaf cutting reciprocator

You are sizing the elliptical gear pair on the reciprocating leaf-knife drive of a Hauni Protos 90 cigarette-maker rebuild, where the knife must descend slowly through a 28 mm tobacco mat to make a clean cut and retract quickly to clear the next slug. The input motor runs at 180 RPM steady, and you want the cutting stroke to take about twice as long as the return stroke. Semi-major axis a = 40 mm. You need to pick an eccentricity, predict the resulting cycle time split, and check what happens if you push the design harder.

Given

  • Ninput = 180 RPM
  • a = 40 mm
  • Target Q = 2.0 dimensionless
  • Cycle stroke = 28 mm

Solution

Step 1 — solve the formula for the eccentricity that gives the target nominal quick-return ratio of 2.0:

e = (Q − 1) / (Q + 1) = (2.0 − 1) / (2.0 + 1) = 0.333

That gives a semi-minor axis b = a × √(1 − e2) = 40 × √(1 − 0.111) = 37.7 mm. So the ellipse is only mildly squashed, which is good — tooth correction stays manageable.

Step 2 — compute the cycle time split at nominal 180 RPM. One full output revolution takes 60 / 180 = 0.333 s. With Q = 2.0, the slow forward stroke takes 2/3 of the cycle and the quick return takes 1/3:

tforward,nom = 0.333 × (2/3) = 0.222 s ; treturn,nom = 0.111 s

The knife spends 222 ms in the tobacco and only 111 ms recovering. That is the sweet spot — long enough for the blade to shear cleanly without tearing fibres, short enough on the return to keep the line at 180 cuts per minute.

Step 3 — check the low end of the typical eccentricity range, e = 0.2:

Qlow = (1 + 0.2) / (1 − 0.2) = 1.5

At e = 0.2 the forward stroke takes 200 ms and the return takes 133 ms — barely different from a plain circular drive. Not worth cutting non-circular gears for that.

Step 4 — check the high end, e = 0.6:

Qhigh = (1 + 0.6) / (1 − 0.6) = 4.0

That looks great on paper — 267 ms in the cut and 67 ms on return — but the instantaneous angular acceleration on the return spike hits roughly 4 times the nominal value, and tooth contact stress at the long-radius mesh climbs about 60% above nominal. On a Protos running 24/7, gear life drops from years to months. Stay near e = 0.33 for production work.

Result

At nominal e = 0. 333 and 180 RPM, the elliptical gear pair gives a 222 ms cutting stroke and a 111 ms return — exactly the 2:1 split the cigarette-maker needs for clean leaf shearing. At the low end (e = 0.2, Q = 1.5) you barely beat a constant-speed drive and the extra cost of non-circular gears is wasted; at the high end (e = 0.6, Q = 4.0) you get a beautifully fast return on day one but the gears chew themselves up inside a quarter. If your measured cycle split comes out closer to 1.7:1 instead of the predicted 2.0:1, the usual culprits are: (1) centre distance off by more than 0.1 mm — recheck the frame against 2a, (2) one gear cut to the wrong eccentricity from a CAM file with rounded ellipse parameters, or (3) excessive torsional wind-up in the output coupling smearing the velocity profile across the crossover points.

When to Use a Slow Forward and Quick Back Circular Motion and When Not To

Three real options compete here for slow-forward, quick-return circular output. Elliptical gears give you a clean mechanical solution with no electronics. Whitworth quick-return linkages are cheaper and simpler but only deliver reciprocating linear motion, not continuous rotation. Servo motors with cam profiles fake the same behaviour electronically and let you change the profile in software. The right choice depends on duty cycle, ratio range, and whether you need true continuous rotation.

Property Elliptical Gear Pair Whitworth Quick-Return Linkage Servo with Cam Profile
Output motion type Continuous rotation, non-uniform velocity Reciprocating linear, not continuous rotation Any profile, fully programmable
Achievable Q ratio 1.2 to 4.0 (e = 0.2-0.6) 1.5 to 2.5 (geometry-limited) 1.0 to 10+ (software-limited)
Cost (50 mm scale, single unit) $400-$900 wire-EDM gears $80-$200 cast linkage $1,500-$3,500 servo + drive + controller
Reliability and lifespan 20,000+ hours if e ≤ 0.5 30,000+ hours, simple pin joints Servo life 30,000 hours, electronics 5-10 yr
Maintenance interval Re-grease every 2,000 hours Pin lubrication every 500 hours No mechanical service, firmware/encoder checks
Load capacity at scale High — direct gear mesh Medium — pin joints limit shock load Limited by servo torque rating
Best application fit High-cycle production with fixed ratio Shapers, slotters, simple reciprocators Variable-product packaging, R&D
Complexity to commission Set centre distance once, done Adjust crank length, simple Tune servo, write cam profile, debug

Frequently Asked Questions About Slow Forward and Quick Back Circular Motion

That is the signature of incorrect centre distance, not gear wear. On a circular gear pair, the centre distance error gives you one constant backlash value. On an elliptical pair, the same physical centre distance error produces backlash that swings between tight at the long-radius mesh and loose at the short-radius mesh — because the local gear radius and therefore local pressure angle changes around the rotation.

Diagnostic check: rotate the pair slowly by hand and feel for the tight spot, then mark its angular position. If the tight spot is exactly 180° from the loose spot, your centre distance is off — usually too short by 0.05-0.2 mm. Shim the bearing block out until the tight and loose readings equalise.

Below e ≈ 0.15 you can sometimes get away with standard involute teeth and the mesh will still work, but you give up most of the quick-return effect anyway. For any useful Q ratio above 1.3, the local pitch curve radius changes enough around the gear that a constant involute profile causes interference at the long-radius mesh and excessive clearance at the short-radius mesh.

The right way is to generate each tooth profile against the local instantaneous radius — modern CAM packages like KissSoft and the open-source Gearotic do this directly from the ellipse parameters. Wire EDM the gear from that profile. Trying to file or grind correction into involute teeth never works repeatably.

The decision rests on cycle count and product variety. If you are running one fixed product 24/7 — say a dedicated cigarette-maker, a tube cutter, or a shaper ram — the elliptical gears win on total cost of ownership inside about 18 months. They draw no power for control, never need encoder calibration, and survive contamination that would kill a servo.

If you change products often, run short batches, or need to tune the velocity profile during commissioning, the servo wins. You can change Q from 1.5 to 4.0 by editing one line in the cam table. Trying to change Q on a gear pair means cutting new gears.

That is angular acceleration kicking the driven inertia. At the crossover from slow forward to quick return, the angular acceleration peaks — for e = 0.5 it spikes to roughly 5 times the average value over a few degrees of rotation. If your output shaft and its coupling do not have enough torsional stiffness, the driven inertia winds up the shaft like a spring and then unwinds, putting a 50-200 Hz oscillation on top of the intended profile.

Two fixes: stiffen the coupling (replace a jaw coupling with a disc-pack or rigid coupling), or add a small flywheel between the gear pair and the load to smooth the torque demand. Do not try to solve it by reducing eccentricity — that throws away the quick-return benefit you bought the gears for.

Three causes, in order of likelihood. First, the gear maker rounded the ellipse axis ratio when generating the tool path — if your CAD says a/b = 1.060 and the CAM file imported it as 1.05, your eccentricity drops from 0.333 to 0.31 and Q falls to about 1.9. Always check the actual cut geometry on a CMM before blaming anything else.

Second, drive train compliance. If you measured Q at the load side rather than at the gear output, shaft torsion and coupling backlash compress the apparent quick-return phase. Measure with an encoder mounted directly on the driven gear shaft.

Third, motor speed regulation. A standard induction motor slows down under the peak torque demand at the crossover, which lengthens the apparent return time. A servo or a heavier flywheel solves it.

Tooth contact stress at the long-radius mesh, plus inertial torque demand at the crossover, set the limit. For a 50 mm semi-major axis gear pair in case-hardened steel at e = 0.4, you can run 600-800 RPM input continuously. Above that, surface fatigue at the long-radius contact zone shows up as pitting within a few thousand hours.

Rule of thumb: for any e above 0.3, derate the allowable continuous input speed to about 60% of what you would run an equivalent circular gear pair at the same module and material. If you need higher speeds, cut the eccentricity and accept a smaller Q — or move to a servo cam.

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.

← Back to Mechanisms Index
Share This Article
Tags: