A hypocyclic gear train is an epicyclic arrangement where a smaller external gear (the planet) rolls inside a larger internal ring gear, tracing a hypocycloidal path. The configuration was formalised in the early 1930s by Lorenz Braren, whose patents on cycloidal speed reducers became the foundation of the modern Sumitomo Cyclo Drive. It converts a single high-speed input into a single low-speed, high-torque output through eccentric motion rather than meshing tooth-on-tooth in the conventional sense, achieving reduction ratios from 10:1 up to 200:1 in one stage with minimal backlash. You see it in robot wrists, aerospace actuators, and machine-tool indexers where compactness and precision both matter.
Hypocyclic Gear Train Interactive Calculator
Vary the fixed ring and planet tooth counts to see the reduction ratio, tooth difference, and slow output motion.
Equation Used
The tooth difference is d = Zring - Zplanet. The ideal speed reduction is i = Zplanet / d, so a one-tooth difference makes the reduction equal to the planet lobe count.
- Single-stage hypocyclic or cycloidal reducer.
- Fixed internal ring gear with external planet disc.
- Valid calculation requires Z_ring greater than Z_planet.
- Ideal kinematics only; losses, backlash, and tooth strength are not included.
How the Hypocyclic Gear Train Works
The mechanism runs on a simple geometric trick. Put a circle inside a bigger circle, fix the big one, and roll the small one along the inside surface — any point on the small circle traces a hypocycloid. In a hypocyclic gear train, the small circle is a planet gear (or a cycloidal disc with lobes), the big circle is the internal ring gear, and the input shaft drives an eccentric cam that forces the planet to wobble around inside the ring. The output is taken either from the planet's own rotation through a set of pin-and-roller couplings, or from a carrier that picks up the planet's slow precession.
The reason engineers reach for this layout is the reduction ratio per stage. A standard spur planetary gives you maybe 3:1 to 10:1 in one stage. A hypocyclic or cycloidal drive gets you 30:1, 87:1, even 119:1 in the same axial length, because the ratio is set by the difference between the planet tooth count and the ring tooth count — typically a difference of just one or two teeth. With zring = 60 and zplanet = 59, you get a 59:1 reduction in a package barely longer than the bearing stack.
If the eccentricity tolerance drifts — say the cam is ground to 1.52 mm of throw instead of the design 1.50 mm — the planet teeth contact too aggressively on one flank and the drive develops a clicking noise under load, then pits the ring gear within a few hundred hours. Common failure modes are eccentric-bearing wear (which shows up as rising backlash measured at the output), pin-and-roller fretting on the output coupling (audible as a metallic chirp), and tooth-tip scuffing when the lubricant film breaks down at low speed and high torque. Backlash-free reducers built to this principle hold under 1 arc-minute when new, but climb fast if the eccentric bearing is undersized for the load.
Key Components
- Internal Ring Gear: The fixed outer member with internal teeth, typically zring = 40 to 200. Tooth profile is usually a modified cycloid or pin-roller geometry rather than involute, with profile tolerance held to ISO grade 5 or better for low-backlash service.
- Planet (Cycloidal Disc): The rolling element with one or two fewer teeth than the ring, often two discs phased 180° apart to balance the eccentric load. Disc thickness runs 8 to 25 mm in industrial sizes, ground flat within 5 µm to keep the pin-bore alignment true.
- Eccentric Cam / Input Shaft: Drives the planet's wobble. Eccentricity is set by (Dring − Dplanet) / 2, typically 0.5 to 3 mm. Cam grind tolerance must be within ±5 µm — drift past that and you get tooth-flank pounding.
- Output Pin-and-Roller Coupling: Picks up the planet's rotation through holes in the disc that are larger than the pins by exactly 2 × eccentricity. The pins ride against hardened rollers to convert wobble into pure rotation. Roller-to-bore clearance under 10 µm gives you sub-arc-minute backlash.
- Support Bearings: High-precision angular contact or crossed-roller bearings carry the eccentric loads. The eccentric bearing alone sees the highest combined radial and shock load in the assembly — undersizing it is the single most common cause of premature drive failure.
Industries That Rely on the Hypocyclic Gear Train
Hypocyclic and cycloidal drives show up wherever you need a lot of reduction in a short axial space, with low backlash and the ability to take shock loads without damage. The cycloidal tooth contact spreads load across many teeth simultaneously — typically 60 to 70 percent of teeth in mesh at any moment — so the drives shrug off impact loads that would chip a conventional spur planetary. That tolerance for shock is why you find them in robot joints, indexers, and any application where the output occasionally has to absorb a hit.
- Industrial Robotics: Joint reducers in FANUC LR Mate and ABB IRB 1200 6-axis robots, where Sumitomo and Nabtesco cycloidal units provide 81:1 to 121:1 reduction with under 1 arc-minute backlash.
- Aerospace: Flap actuators and trim drives on Boeing 737 secondary flight control systems, where the high reduction ratio in a compact housing saves weight versus a multi-stage spur reducer.
- Machine Tools: Rotary indexing tables on Mazak Integrex multi-tasking lathes, where the backlash-free reducer holds part position within 30 arc-seconds during heavy interrupted cuts.
- Wind Turbine Pitch & Yaw: Yaw drives on Siemens Gamesa SG 5.0-145 turbines using cycloidal gearboxes to handle the shock loads from gusting wind on the nacelle.
- Medical Imaging: Gantry rotation drives inside Siemens SOMATOM CT scanners, where smooth low-vibration output keeps image artefacts below detection threshold.
- Semiconductor Equipment: Wafer-handler robot wrists in Applied Materials Endura platforms, where sub-arc-minute repeatability is needed across millions of cycles.
The Formula Behind the Hypocyclic Gear Train
The core relationship is the reduction ratio of a hypocyclic stage with a fixed ring gear, output taken from the planet's rotation. The ratio depends only on the tooth counts. At the low end of the practical range — say a 10:1 ratio with zring = 11 and zplanet = 10 — the eccentricity becomes large relative to the gear diameter and the drive runs rough with high vibration at the input. At the high end, beyond 200:1, the tooth-count difference is still 1 but the ring gets large and manufacturing the matched cycloidal profile to grade-5 tolerance gets expensive fast. The sweet spot for industrial drives sits at 30:1 to 120:1, where the eccentricity is comfortable, the lobe count gives smooth output, and the cost-to-ratio curve is favourable.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| i | Reduction ratio (input revs per output rev), with the ring gear fixed | dimensionless | dimensionless |
| zplanet | Number of teeth (or lobes) on the cycloidal planet disc | teeth | teeth |
| zring | Number of teeth (or pin-rollers) on the fixed internal ring | teeth | teeth |
| e | Eccentricity of the input cam, equal to (Dring − Dplanet) / 2 at the pitch line | mm | in |
Worked Example: Hypocyclic Gear Train in a Universal Robots UR5e wrist joint retrofit
You are sizing a replacement cycloidal reducer for the wrist-3 joint of a Universal Robots UR5e collaborative robot, swapping in a custom hypocyclic stage to lower backlash for a precision dispensing application. The motor input runs at 3000 RPM nominal, and the joint must rotate at roughly 30 RPM at the output. You have a planet disc with zplanet = 39 lobes and a ring with zring = 40 pin-rollers. Eccentricity is set at e = 1.5 mm.
Given
- zplanet = 39 lobes
- zring = 40 pin-rollers
- Ninput = 3000 RPM
- e = 1.5 mm
Solution
Step 1 — compute the nominal reduction ratio from the tooth-count difference:
Step 2 — convert input speed to output speed at nominal 3000 RPM:
That is above the 30 RPM target, so in practice the controller throttles the servo to roughly 1170 RPM to land at 30 RPM at the joint. The drive sweet spot sits comfortably here — the input is well below the 4500 RPM upper limit typical of this class of cycloidal reducer, and the eccentric bearing runs cool.
Step 3 — check behaviour at the low end of the operating range. If the dispensing program calls for slow positioning at 300 RPM input:
At 7.7 RPM output the joint creeps — about one revolution every 8 seconds. This is where boundary lubrication matters: below roughly 5 RPM output, the cycloidal tooth contact transitions out of full EHL film and you start seeing micro-pitting on the lobe flanks within a few thousand hours. Stay above 5 RPM output for continuous duty.
Step 4 — at the high end, suppose the controller commands full motor speed 4500 RPM for a fast retract:
The joint sweeps quickly but the eccentric bearing inner race now sees about 4500 RPM of orbiting load — at this point bearing temperature, not gear capacity, is the limit. Most industrial cycloidal drives in this size derate continuous torque above about 3000 RPM input for exactly that reason.
Result
The nominal reduction is 39:1, giving 76. 9 RPM at the joint output for 3000 RPM motor input. That feels about right for a UR5e wrist on a dispensing task — fast enough to move between dispense points without lag, slow enough that the servo loop holds position cleanly. Across the range, the drive runs smoothly from 7.7 RPM at the low end up to 115 RPM at the high end, with the comfortable continuous zone sitting between 30 and 90 RPM output. If you measure backlash above 2 arc-minutes after installation when the spec is under 1, the most likely causes are: (1) eccentric-bearing preload set too loose during assembly, letting the planet wobble axially, (2) pin-roller bore clearance machined oversize — anything past 12 µm and you lose the backlash spec, or (3) a planet disc warped during heat treat, which shows up as a once-per-revolution position error at the output that you can pick up on an encoder trace.
Hypocyclic Gear Train vs Alternatives
Hypocyclic and cycloidal drives are not always the right answer. They cost more than spur planetaries, they are harder to manufacture to tight tolerances, and the input vibration from the eccentric mass is real. Here is how they stack up against the two alternatives an engineer most often considers — a multi-stage spur planetary gearbox and a strain-wave (harmonic) drive.
| Property | Hypocyclic / Cycloidal Drive | Multi-Stage Spur Planetary | Strain-Wave (Harmonic) Drive |
|---|---|---|---|
| Single-stage reduction ratio | 10:1 to 200:1 | 3:1 to 10:1 per stage | 30:1 to 320:1 |
| Backlash (new, typical) | under 1 arc-minute | 5 to 15 arc-minutes | under 30 arc-seconds |
| Peak torque capacity vs nominal | 3× to 5× nominal (shock-tolerant) | 1.5× to 2× nominal | 2× to 3× nominal |
| Maximum input speed | 3000 to 4500 RPM continuous | 5000+ RPM continuous | 3500 RPM continuous |
| Service life under rated load | 20,000+ hours | 10,000 to 15,000 hours | 10,000 hours (flexspline-limited) |
| Relative cost (similar ratio) | medium-high | low-medium | high |
| Typical application fit | robot joints, indexers, shock-load drives | general-purpose machinery, conveyors | precision robotics, semiconductor |
| Manufacturing complexity | High — cycloidal profile grinding | Low — standard involute tooling | Very High — flexspline forming |
Frequently Asked Questions About Hypocyclic Gear Train
The eccentric mass on the input shaft is unbalanced by design — the planet disc orbits off-axis, and that creates a once-per-input-rev radial force. Two-disc designs phase the discs 180° apart to cancel this, but single-disc drives run with residual imbalance.
Under load, the tooth-mesh forces dominate the reaction at the input bearing and the imbalance signature gets buried in the noise. Unloaded, the bearing has nothing else to react against, so you feel the imbalance directly. If the vibration is excessive even on a two-disc drive, check that the discs are actually phased correctly — a 180° phase error on assembly turns a balanced drive into a single-disc drive's worth of vibration.
Technically yes, but you will hate the result. Cycloidal geometry has efficiency around 90 to 94 percent in the reduction direction. Run it backwards as an increaser and efficiency drops fast — often below 70 percent — because the small tooth-count difference means the drive is close to self-locking at high ratios.
Above about 30:1 ratio, back-driving torque has to overcome significant friction at every pin-roller contact simultaneously, and you may find the output jerks rather than spins smoothly. If you need speed increase, pick a different topology — a planetary running in reverse handles back-drive far better.
1-tooth difference gives you maximum reduction per stage and the smoothest tooth-engagement profile, but the eccentricity has to grow in proportion to keep the gears engaged, which raises input vibration and bearing load. You see this in high-precision robot joints where ratio matters most.
2-tooth difference cuts the eccentricity roughly in half for the same gear size, dropping vibration and bearing wear, but you get half the reduction. It is the choice when you need a moderate ratio in the 15:1 to 40:1 range and want long bearing life — many industrial indexers use this approach. Rule of thumb: if you need above 50:1 single-stage, go 1-tooth-difference; below 40:1, 2-tooth-difference gives better bearing life.
Probably not — you are seeing the difference between the nominal naming convention and the actual gear math. Many cycloidal manufacturers label drives by the ring tooth count rather than the true reduction ratio. A drive marked 40:1 with zring = 40 and zplanet = 39 actually delivers i = 39 / (40 − 39) = 39:1 with the ring fixed.
If you are doing closed-loop position control and counting encoder pulses, this 2.5 percent error compounds fast. Measure the actual ratio by clamping the input, rotating the output one full turn by hand, and counting input revs — trust that number, not the label.
The eccentric bearing is the culprit. In a cycloidal drive, that single bearing carries the full radial load from the planet's wobble at the input shaft speed — so at 3000 RPM input you have a heavily loaded bearing turning at 3000 RPM. A spur planetary spreads the load across three or more planet bearings, each turning slower.
Heat output scales with bearing load × speed, so the cycloidal can run 15 to 25 °C hotter than an equivalent planetary at the same output torque. This is normal up to about 80 °C housing temperature; above that, check that the lubricant viscosity is correct for the temperature class and that the eccentric-bearing preload was not set too tight at assembly.
For a hobby arm, almost certainly overkill — and worse, hard to source affordably. A genuine cycloidal reducer in the size you would want for a desktop arm runs $300 to $800 even from budget Chinese suppliers, and the cheap clones often skip the precision pin-roller grinding that makes the geometry work, so backlash is no better than a spur planetary anyway.
For hobby use, a quality planetary gearmotor or a 3D-printed cycloidal drive (yes, these work surprisingly well at low torque) is the better call. Reserve the real hypocyclic drive for builds where backlash under 5 arc-minutes is a hard requirement and the budget supports it.
References & Further Reading
- Wikipedia contributors. Cycloidal drive. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
