A compound epicyclic train is a planetary gear arrangement where each planet carries two rigidly-coupled gears of different tooth counts, so a single stage can deliver reduction ratios from roughly 20:1 up to 2000:1 in a coaxial package. The stepped planets mesh with two different sun or ring gears, multiplying the ratio without adding a second stage. Engineers use this layout when they need extreme reduction in a short axial space — wind turbine yaw drives, robot joints, and the Wolfrom-type drives inside Maxon and Harmonic Drive servo gearheads all rely on it.
Compound Epicyclic Train Interactive Calculator
Vary the ring and stepped-planet tooth counts to see the Wolfrom-style ring-difference reduction ratio and output motion.
Equation Used
The worked example shows a fixed 100-tooth ring and a 99-tooth output ring. With a one-tooth ring difference, the approximate reduction is 100 / 1 = 100:1. The 24T and 22T planet halves are shown as the stepped-planet detail that creates the coupled differential motion.
- Uses the worked-example Wolfrom ring-difference approximation.
- Fixed and output ring tooth counts are treated as integer teeth.
- Planet teeth are included to show the stepped planet ratio, but detailed assembly phasing is not checked.
How the Compound Epicyclic Train Works
A simple planetary set has one sun, one ring, three or four planets, and a carrier — and tops out around 10:1 per stage before the ring gear gets impractically large. A compound epicyclic train fixes that ceiling by splitting each planet into two gears pressed onto the same shaft. One half meshes with the sun, the other half meshes with a ring (or a second sun). Because the two halves share a shaft, they rotate at the same speed, but their different tooth counts create a velocity differential between the two meshes. That differential is what generates the high ratio. The classic Wolfrom drive uses one sun, stepped planets, and two ring gears — one fixed to the housing, the other coupled to the output. Tooth counts are typically picked so the two rings differ by just 1 or 2 teeth, which is exactly how you get a 100:1 ratio out of a single compact stage.
Tooth count selection is where this mechanism lives or dies. The two ring gears must satisfy a strict assembly condition — the planets have to engage both rings simultaneously without binding. If you pick tooth counts that don't share the right common factors with the planet count, you can't even slide the gearset together. We've seen prototypes where the designer ignored the assembly equation, machined everything to spec, and then couldn't insert the third planet because the teeth were 0.3 mm out of phase. Backlash is the other quiet killer. Each mesh contributes its own backlash, and in a compound stage those errors stack — a Wolfrom drive with 0.05 mm backlash per mesh ends up with 0.2 to 0.3 mm of output lash once you multiply through the ratio. That shows up as positioning hunt on a servo system. Efficiency also drops fast at high ratios because the two meshes work against each other; a 100:1 compound planetary often runs at 60-70% efficiency, not the 95% you'd see from a simple 5:1 stage.
Key Components
- Sun Gear: The central input gear, typically 12-24 teeth, that drives the inner mesh of every stepped planet. Concentricity to the carrier bore must hold within 0.02 mm or you get cyclic torque ripple at planet-pass frequency.
- Stepped Planets: Two gears rigidly joined on a common shaft — usually 3 or 4 planets equally spaced. The two halves typically differ by 4-12 teeth, and the tooth-count pair must satisfy the compound assembly condition (Zr1 + Zs) and (Zr2 + Zp2 × something) divisible by the planet count.
- Fixed Ring Gear: Bolted to the housing and meshes with one half of each stepped planet. Its tooth count differs from the output ring by just 1-3 teeth — that small difference is exactly what creates the high reduction ratio.
- Output Ring Gear: Rotates freely and couples to the output shaft. Carries full output torque, so wall thickness and tooth-root strength dominate sizing — for a 100 Nm output you typically need a ring with at least 6 mm root thickness in 4140 steel.
- Planet Carrier: In a Wolfrom configuration the carrier floats — it isn't an output, it's just a positioning frame for the planet shafts. Its bearings must hold the planets parallel to within 0.05 mm/100 mm or load distribution between planets goes uneven and one planet eats most of the torque.
- Planet Bearings: Needle rollers or full-complement cages, typically rated for the planet shaft speed which can hit 3000 RPM in a high-ratio drive. These are usually the first wear point — when a compound planetary fails, 70% of the time it's a planet bearing seizing.
Who Uses the Compound Epicyclic Train
Compound epicyclic trains turn up wherever you need huge reduction in a coaxial package and can tolerate moderate efficiency. The geometry is more expensive to manufacture than a 2-stage simple planetary, so designers reach for it specifically when axial length, weight, or coaxial input/output is non-negotiable. Robotics, aerospace actuation, and wind turbine pitch drives are the heaviest users.
- Robotics: The Wolfrom-style drives inside Nabtesco RV-series gearheads used in FANUC and ABB robot wrist joints, delivering 81:1 to 192:1 in a single coaxial stage.
- Aerospace: Pratt & Whitney's PW1000G geared turbofan uses a compound planetary fan drive gear system to reduce the LP turbine speed by roughly 3:1 — a simple planetary couldn't fit the radial envelope.
- Wind Energy: Pitch and yaw drives on Vestas V90 turbines use compound epicyclic gearboxes to take the 1500 RPM motor down to under 1 RPM at the blade root.
- Automotive: The Ravigneaux gearset inside Ford's 4R70W and many ZF automatic transmissions is a compound epicyclic — one long planet and one short planet on the carrier deliver four forward ratios from a single gear set.
- Servo Actuators: Maxon GP series and Harmonic Drive CSG-2A units use compound planetary stages internally to hit 100:1 to 500:1 in motor diameters under 50 mm — common in surgical robots like the Intuitive da Vinci.
- Heavy Industry: Slew drives on Liebherr LR 11000 crawler cranes use compound epicyclic reductions to swing a 1000-tonne boom at controlled rates.
The Formula Behind the Compound Epicyclic Train
The Wolfrom-drive ratio formula tells you how the input sun speed relates to the output ring speed when the housing ring is fixed. The dramatic part is how sensitive the ratio is to tooth-count choice. At the low end of the typical range — picking ring tooth counts that differ by 4 or 5 teeth — you get ratios around 20:1 with reasonable 80% efficiency. The sweet spot for most servo applications sits at a 1-2 tooth difference between rings, which yields 80-150:1 at 65-72% efficiency. Push to a 1-tooth difference with carefully chosen sun and planet counts and you can theoretically hit 2000:1, but efficiency collapses below 40% and the gearbox starts behaving more like a brake than a transmission.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| i | Overall reduction ratio from input sun to output ring (housing ring fixed) | dimensionless | dimensionless |
| Zr1 | Tooth count of the fixed (housing) ring gear | teeth | teeth |
| Zr2 | Tooth count of the output ring gear | teeth | teeth |
| Zp1 | Tooth count of the planet half meshing with the fixed ring | teeth | teeth |
| Zp2 | Tooth count of the planet half meshing with the output ring | teeth | teeth |
Worked Example: Compound Epicyclic Train in a surgical robot end-effector drive
Sizing the compound epicyclic reduction inside the roll-axis joint of a laparoscopic surgical robot end-effector similar to the CMR Surgical Versius arm. The motor is a Maxon EC-i 40 brushless running at 6000 RPM, and the joint needs to rotate at roughly 60 RPM with backlash under 3 arcmin. Total axial length available between the motor face and the tool flange is 28 mm, which rules out a 2-stage simple planetary. You decide on a Wolfrom layout with 3 stepped planets.
Given
- Zr1 = 100 teeth (fixed ring)
- Zr2 = 99 teeth (output ring)
- Zp1 = 42 teeth (planet half meshing fixed ring)
- Zp2 = 41 teeth (planet half meshing output ring)
- Nin = 6000 RPM
Solution
Step 1 — compute the nominal ratio with the chosen tooth counts (1-tooth difference between rings, 1-tooth difference between planet halves):
Step 2 — convert that to output speed at the nominal 6000 RPM motor input:
That sits comfortably above the required 60 RPM target, so you've got headroom. The joint will feel responsive — at 83.7 RPM the surgical tool rolls a full revolution every 0.72 seconds, which matches what surgeons expect from the Versius and da Vinci platforms.
Step 3 — at the low end of the practical Wolfrom range, swap to a less aggressive tooth-count pair (Zr1=100, Zr2=96, Zp1=42, Zp2=40):
Wait — that's a higher ratio, not lower. That's the counter-intuitive part of Wolfrom drives: closing the gap between the rings actually INCREASES the ratio. To get a LOWER ratio you spread the rings apart. Try Zr1=100, Zr2=90, Zp2=37 — that yields about 25:1, which is the practical low end before the geometry collapses into a regular simple planetary.
Step 4 — at the high end, push the rings to differ by exactly the planet count (3 teeth here would let you tighten further). With Zr1=99, Zr2=100, Zp1=40, Zp2=41 you can hit ratios beyond 400:1:
The negative sign just flips output rotation direction. Push tooth counts further and ratios climb past 500:1, but efficiency falls below 45% — the joint starts feeling sluggish and back-driving becomes impossible, which is actually useful for a surgical hold-position joint but disastrous for a haptic feedback joint.
Result
The 71. 7:1 nominal ratio gives 83.7 RPM at the joint output from a 6000 RPM motor — fast enough for the 60 RPM specification with 40% speed headroom for accel/decel ramps. Across the Wolfrom range, dropping toward 25:1 turns the joint into a fast, lightly-geared drive that needs a brake to hold position, while pushing past 200:1 gives you a self-locking joint at the cost of dropping efficiency from ~70% to under 45%, meaning the motor draws nearly twice the current for the same output torque. If your measured output speed is significantly off the predicted value, three failure modes dominate: (1) a planet tooth-phasing error during assembly — even one tooth out on one of the three planets causes binding and a step-change in ratio, (2) excessive output ring runout above 0.03 mm TIR which produces a velocity ripple that reads as a wrong average speed on a low-resolution encoder, or (3) planet bearing preload drift that lets the planets walk axially and partially disengage one of the two meshes, dropping effective ratio by 5-15%.
Choosing the Compound Epicyclic Train: Pros and Cons
A compound epicyclic isn't always the right answer. It earns its keep when you need very high coaxial reduction in a short axial space — but it costs more, runs less efficiently, and is harder to manufacture than the alternatives. Here's how it stacks against the two mechanisms most often considered alongside it.
| Property | Compound Epicyclic (Wolfrom) | Two-Stage Simple Planetary | Strain Wave (Harmonic Drive) |
|---|---|---|---|
| Typical reduction ratio range | 20:1 to 2000:1 in one stage | 10:1 to 100:1 in two stages | 30:1 to 320:1 in one stage |
| Efficiency at typical ratio | 45-75% | 85-94% | 65-85% |
| Backlash (typical) | 3-15 arcmin | 5-20 arcmin | <1 arcmin |
| Axial length for 100:1 | 25-35 mm | 60-90 mm | 20-30 mm |
| Torque density | High | Medium | Very high |
| Manufacturing cost (relative) | High — stepped planets need precision grinding | Low — standard parts | Very high — flexspline is exotic |
| Service life at full load | 8,000-15,000 hours | 20,000-40,000 hours | 10,000-20,000 hours |
| Best application fit | High-ratio coaxial drives, robot wrists | General industrial reduction | Precision positioning, robotics |
Frequently Asked Questions About Compound Epicyclic Train
In a Wolfrom drive the two meshes work in opposition — the fixed-ring mesh and the output-ring mesh both transmit nearly the full torque, but the net output speed is the small difference between them. The closer the two rings are in tooth count, the smaller that net difference, and the larger the circulating power inside the gearbox.
At 70:1 maybe 1.5× input power circulates internally and 70% of it makes it to the output. At 200:1 you can have 4-5× input power circulating, with sliding losses scaling proportionally. A two-stage simple planetary doesn't have circulating power — each stage just transmits power forward — so its efficiency stays in the high 80s regardless of ratio.
The compound assembly condition fails. For a Wolfrom drive with Np planets you need (Zr1 × Zp2 − Zr2 × Zp1) to be divisible by Np. Plug your numbers in: (100 × 41) − (99 × 42) = 4100 − 4158 = −58. With 3 planets, −58 is not divisible by 3, so the third planet won't seat without forcing teeth out of phase.
Fix it by adjusting one tooth count — try Zp1=43 which gives (100 × 41) − (99 × 43) = 4100 − 4257 = −157, still not divisible by 3. Zr2=96, Zp2=39 gives 100 × 39 − 96 × 42 = 3900 − 4032 = −132, divisible by 3. Always run the assembly check before cutting metal.
Pick compound epicyclic when peak torque transients are high and shock loading matters. Strain waves have zero backlash and are the default for precision robotics, but the flexspline fatigue-fails under repeated peak loads above 3× rated torque. A Wolfrom drive with proper case-hardened gears shrugs off 5-7× rated peak torque without permanent damage.
Pick strain wave when positioning accuracy under 1 arcmin matters more than peak torque capacity, like in metrology or semiconductor handling. The CMR Versius and Intuitive da Vinci use strain waves for the precision joints and compound planetary for the higher-load proximal joints — that mix is industry standard.
Most likely circulating-power losses you didn't account for. The textbook calculation Tout = Tin × i × η assumes a single efficiency value, but Wolfrom efficiency varies sharply with operating point. At cold start with stiff lubricant, efficiency can be 20 percentage points below the warm-running value — losing 30% torque on a 70% nominal-efficiency drive matches a cold-start scenario almost exactly.
Check oil temperature first. Then check planet-to-ring contact pattern by running blueing compound through the gearbox at low load — if contact concentrates at one end of the tooth face, you have planet-shaft misalignment that's wasting 5-10% in extra sliding friction. A correctly-aligned Wolfrom drive at operating temperature should match the calculated torque within 8-12%.
Usually no, and that's often a feature rather than a bug. Back-drivability requires forward efficiency above roughly 50% — the threshold below which the output torque can't overcome internal friction to spin the input. A Wolfrom at 100:1 typically runs at 60-70% forward efficiency, which corresponds to 20-40% reverse efficiency. Below 50% reverse efficiency the gearbox is self-locking.
This is why surgical robots use compound planetary on weight-bearing joints — power loss to the motor doesn't drop the surgical tool. If your application needs back-drivability for haptic feedback or compliance, drop ratio below 30:1 or switch to a strain wave or cycloidal drive.
Three planets, three pulses per input revolution — you're seeing planet-pass frequency. Some ripple is normal because each planet enters and exits mesh under slightly different load. But a ripple amplitude above 0.5% of mean output speed points to one of two issues.
Either the three planets aren't sharing load equally (typical cause: carrier-bore positional tolerance worse than 0.02 mm, letting one planet take 60% of the load while the other two share 40%), or one stepped planet has a tooth-count phasing error from assembly — easy to verify by marking the gears and running a slow rotation test. A correctly-built 3-planet Wolfrom drive should show under 0.2% velocity ripple at full load on a 17-bit encoder.
References & Further Reading
- Wikipedia contributors. Epicyclic gearing. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
