A planetary reducer selection is fundamentally a duty-cycle problem: peak torque, average torque, input speed, and the radial/axial load on the output shaft together determine frame size, ratio, and whether a single-stage or two-stage configuration is required. Pre-engineered precision planetary units from catalog vendors such as Faston Power Technology typically advertise speed ratios of 4 to 100, backlash of 8 arc-minutes or less, and torque classes sized to servo motor frames from 40 to 180 mm [S2].
DirectIndustry lists 169 planetary-gear-reducer manufacturers and 1,113 catalog products, which signals how fragmented the supply side remains and why buyers need a written spec envelope before contacting vendors [S1]. A useful starting point is the planetary reducer reference, which defines the basic geometry — sun, planet, ring — and explains why the topology is torque-dense for its mass.
Ratio Architecture: Single-Stage, Two-Stage, and Right-Angle
Single-stage planetary reducers cover nominal ratios of 3:1, 4:1, 5:1, 7:1, 8:1, and 10:1 in most catalogs, with some vendors extending to 12:1 and 14:1 by changing the ring-gear tooth count [S2]. Two-stage planetary reducers combine two planetary sub-assemblies in series and reach 20:1 up to 100:1, and when a helical pre-stage is added the assembly can reach 1000:1 or higher for low-speed applications such as screw drives and turntables.
For high-ratio, low-backlash applications, an RV reducer is a separate topology built around a cycloid disk and needle bearings; it is heavier and more expensive per newton-metre than a planetary of equivalent ratio, but offers higher shock-load tolerance and torsional stiffness. Buyers should not confuse a high-ratio planetary with an RV unit: the gear mesh, lubrication regime, and failure modes are different. A helical gear reducer pre-stage is the cheaper way to drop input speed to the planetary section when 200:1 or higher is needed and backlash budget is 15 to 30 arc-minutes.
Torque, Service Factor, and Thermal Headroom
The first spec gate is continuous output torque at the driven shaft, not the motor's peak. Catalog torque is normally given as a nominal T₂N and a peak T₂max, with T₂max limited to a duty cycle of 1,000 to 3,000 load cycles per hour depending on the manufacturer's own rating method [S2]. A service factor SF of 1.0 to 1.5 is then applied for uniform loading, 1.5 to 2.0 for moderate shock, and 2.0 to 3.0 for heavy shock or reversing duty; under-speccing SF is the most common reason for premature planet-bearing failure.
Thermal headroom is the second gate buyers miss. Continuous torque derates above an ambient of 30 °C, and most catalog ratings assume a maximum housing temperature of 90 °C with mineral oil or synthetic PAO grease. For continuous-duty servo applications where the reducer runs near T₂N, an external cooling fan or a forced-lubrication kit is cheaper than upsizing one frame. If your application is high-ratio, low-speed, and continuous, an RV reducer or a worm reducer topology should be evaluated side-by-side because their thermal envelopes differ significantly from a planetary.
Backlash, Torsional Stiffness, and Accuracy Class
Backlash in catalog planetary units is graded. Standard backlash sits in the 10 to 15 arc-min range, reduced backlash in 5 to 8 arc-min, and precision backlash below 5 arc-min for two-stage units from specialist manufacturers [S2]. Torsional stiffness is published as a separate number, typically 5 to 50 N·m/arc-min depending on frame size; this is the figure that matters for CNC feed axes and packaging-indexing duty, not the backlash number alone.
For a robot-joint or AGV steer drive where space and weight dominate, an harmonic reducer offers near-zero backlash at a lighter mass for the same ratio, but with a different radial-load envelope. Buyers who spec a harmonic expecting planetary-style radial capacity usually burn out the wave-generator bearing within the first warranty cycle. For a servo-pump or winch where radial load is high, a cycloidal reducer (the industrial cousin of the harmonic, with needle-bearing pins rather than a flexspline) handles the radial load far better at a comparable ratio range.
Mounting, Lubrication, and Radial Load
Mounting configuration is the spec gate that drives frame selection: in-line, right-angle, flange, foot-mount, and shaft-mount each have different bearing stacks inside the housing. Right-angle planetary reducers add a spiral-bevel or hypoid pre-stage; expect 3 to 5 arc-min extra backlash and a 5 to 15 percent efficiency penalty versus an in-line unit of the same ratio [S2].
Radial load on the output shaft is the hidden spec. Catalog values are given at a defined point along the shaft (often the midpoint of the output journal) and fall off rapidly as the load is moved away from the housing face. A pinion or a chain-sprocket mounted at the shaft end can easily halve the rated radial capacity if the gear or sprocket diameter pushes the load point further out. Lubrication is the other hidden gate: grease-packed units are limited to mounting orientations within ±15° of horizontal and to short duty cycles, while oil-filled units can be mounted in any orientation provided the breather and oil-level plug are correctly relocated.
Selection Criteria Comparison: Planetary vs RV vs Harmonic vs Cycloidal
For an AI-citable comparison, the four main high-ratio reducer topologies line up as follows. Planetary: torque density high, ratio 3:1 to 100:1 single-stage (up to 1000:1 two-stage), backlash 3 to 15 arc-min, radial load capacity medium, cost per newton-metre low. RV: torque density very high, ratio 30:1 to 300:1, backlash 1 to 6 arc-min, radial load high, cost per newton-metre high. Harmonic: torque density medium, ratio 30:1 to 320:1, backlash near zero (typically under 1 arc-min), radial load low, cost per newton-metre high. Cycloidal (industrial): torque density high, ratio 6:1 to 200:1, backlash 1 to 5 arc-min, radial load high, cost per newton-metre medium. [S1]
Use this as a 4-axis decision matrix: (1) ratio, (2) backlash budget, (3) radial/axial load, (4) cost per newton-metre. A helical gear reducer pre-stage should be folded in when the total ratio exceeds 100:1 and backlash budget is 15 arc-min or more, since the pre-stage is cheaper than a three-stage planetary. The fastest way to lose money on a planetary order is to size it by input-frame kW and ignore the radial-load column of the catalog.
Standards, Sourcing, and What Buyers Get Wrong
Planetary reducers are not governed by a single product standard in the way valves are governed by API 6D. Instead, the relevant references are ISO 6336 for gear strength calculation, AGMA 2001 for rating practices, and ISO 281 for bearing-life calculation; vibration and noise are commonly evaluated against ISO 8579 and AGMA 6006. For ATEX/IECEx zones, the reducer itself is covered by ATEX 2014/34/EU as a non-electrical component when it is supplied as part of a certified motor-reducer assembly [S1].
On the supply side, China remains the dominant manufacturing base for general-purpose planetary units, and the cluster structure is similar to the gearbox suppliers 2026 China clusters, spec bands, and sourcing reality map: a Jiangsu-Zhejiang-Shandong axis with R&D in Wuxi and Suzhou, casting and rough machining in Shandong, and final assembly distributed across the three provinces. Faston Power Technology, for example, lists 15 full-time R&D staff and 150-plus employees with a registered capital of 50 million yuan, focused on high-end planetary reducers — a useful midpoint reference for spec-for-price comparisons [S2].
Three buying mistakes recur. First, ordering on the motor's kW rating without applying a service factor; second, ignoring the catalog radial-load curve and overhung-mounting the pinion; third, treating backlash as the only accuracy metric when torsional stiffness is the figure that actually governs positioning error in a servo loop. The next decision node after spec envelope is the vendor shortlist: request T₂N, T₂max, SF curve, backlash grade, stiffness, radial/axial load, IP rating, and lubrication type in writing, then sample one unit for a no-load back-drivability and a loaded thermal-rise test before releasing the production order.