How to Select a Precision Planetary Gearbox
for Servo Motor Applications

A servo motor without a gearbox runs at 1,000–5,000 rpm with low output torque — far from what most industrial axes require. A planetary gearbox converts that high-speed, low-torque input into the low-speed, high-torque output the load needs, while simultaneously resolving the inertia mismatch between the compact motor rotor and the often much heavier load it must accelerate.

When engineers select a precision planetary gearbox for a servo motor correctly, the result is a closed-loop axis with repeatable positioning, efficient energy conversion, and a service life measured in years. When they select incorrectly, three failure modes dominate:

The five-step planetary gearbox selection framework below walks through each parameter in the correct sequence — starting with torque, then ratio, then backlash grade, then inertia, and finally physical interface. Skipping steps or reversing the order is the single most common source of servo axis specification errors in Korean machine design.

Output torque is the first parameter to establish because it determines both the gearbox frame size and the torque rating. Two torque values matter for every axis: the continuous rated torque (T2N) that the gearbox handles throughout a production cycle, and the peak torque (T2B) occurring during acceleration and deceleration. Peak loads can reach two to three times the continuous value, and a gearbox sized only for continuous duty will suffer accelerated gear tooth wear under repeated peak loads.

Worked example: A Korean packaging machine cross-seal jaw requires 85 N·m continuous at the jaw shaft. The servo motor delivers 8.5 N·m at rated speed. Required ratio: 85 / (8.5 × 0.97) ≈ 10:1. Applying a 2.5× peak factor for jaw impact → gearbox must handle 212 N·m peak. The selected gearbox must have T2B ≥ 212 N·m at i=10.

Output Torque Reference by Application Type

Safety factors shown are starting points — always confirm with your full duty cycle analysis.

• Highest efficiency: ≥97%
• Shortest axial housing depth
• Highest allowable input speed
• Best inertia ratio for fast dynamics

• Wider ratio range for slower axes
• Efficiency ≥94%
• Longer housing — check axial space
• Reflected inertia drops sharply (i² benefit)

• Ratios up to 10,000:1 in one unit
• Efficiency ≥90–92% (3–4 stage)
• Heavy industrial and energy applications
• Larger frame sizes (AH/AHK/AFHK series)

Inertia matching is the most frequently skipped step in servo gearbox selection — and the most frequently blamed when a newly commissioned axis behaves unpredictably. The inertia ratio problem is straightforward: a servo motor rotor typically has a rotor inertia of 50–500 g·cm², while the load it must accelerate may have an inertia of thousands of g·cm². Without a gearbox, the motor is trying to swing a mass 50–100× its own rotational equivalent — leading to overshoot, oscillation, and ultimately a control loop that no gain setting can stabilise.

Worked example: Robot elbow axis with J_load = 800 g·cm², servo motor J_rotor = 120 g·cm²:

This is why increasing the ratio from 5:1 to 10:1 — even when either could achieve the speed — often produces dramatically better servo response: the i² denominator effect reduces reflected inertia by 4× for every doubling of ratio.

Frame Size (Body Diameter)

Frame size sets the physical scale: output shaft diameter, radial load capacity, and mounting dimensions. Once the output torque is confirmed, the minimum frame size follows from the torque rating table for the selected series. Always cross-check that the chosen frame’s radial load capacity (F_rad) exceeds the actual radial force applied at the shaft end — this is particularly critical for belt drives, gear meshes, and chain sprockets mounted directly on the output shaft.

Flange Geometry

The output flange type determines how the gearbox mounts to the machine structure. Square flanges (EP-AB, EP-AF, EP-ABR) are the most common for direct machine bed mounting. Round flanges (EP-AD, EP-ADS) suit bore-mounted rotary tables and spindle heads. Large flanges (EP-AE, EP-AER) provide higher overturning moment resistance for conveyor head drives — and are the only series in the Korea Ever-Power range with an IP67 option.

Temperature Range

Standard Korea Ever-Power planetary series operate from −10 °C to +90 °C. This covers Korean outdoor industrial winter conditions. The sole exception is the EP-KF/KH hypoid gear series, whose gear oil specification limits the lower bound to 0 °C minimum. Do not specify KF/KH for outdoor Korean winter installations, cold-room applications, or any environment where temperatures may drop below 0 °C.

What Torsional Stiffness Actually Means for Servo Performance

Torsional stiffness (C_T) is the torque required to produce one arcminute of angular deflection between the gearbox input and output shafts under load — expressed in N·m/arcmin. A gearbox with high torsional stiffness transmits a motor torque command to the load with minimal spring-back. A gearbox with low torsional stiffness behaves as a torsional spring in the drive train: the motor encoder accurately reports input position, but the load is at a different angle because the gearbox body is elastically deflecting.

This elastic compliance between motor and load defines the anti-resonance frequency — the frequency at which the motor and load begin to oscillate in opposition. The governing formula is:

Worked example: A robot elbow axis with C_T = 80 N·m/arcmin (converted: 274,960 N·m/rad), J_motor = 80 g·cm² = 8×10⁻⁵ kg·m², J_load reflected = 12 g·cm² = 1.2×10⁻⁵ kg·m²:

With f_res ≈ 258 Hz, this axis can support a servo bandwidth up to ~86 Hz (258 ÷ 3) — sufficient for high-performance robot joint control. If C_T were halved to 40 N·m/arcmin, f_res drops to 182 Hz and the usable bandwidth ceiling falls to 60 Hz, which may be marginal for high-speed pick-and-place cycles.

Backlash vs Torsional Stiffness — Two Independent Problems

These two specifications are sometimes confused because both relate to angular error at the output shaft — but they arise from completely different mechanisms and affect servo performance in different ways.

This distinction explains why the enlarged output shaft of the EP-AF और EP-AFR high-rigidity series contributes to servo performance beyond just radial load capacity: a shaft of larger diameter has a polar moment of area proportional to diameter⁴, which directly increases the shaft’s own torsional stiffness contribution. At the same frame size, the enlarged shaft of EP-AF vs a standard shaft at EP-AB can raise the shaft torsional contribution by 50–100% depending on the diameter difference.

Ignoring peak torque during acceleration and jaw-close impacts. A gearbox rated for 100 N·m continuous exposed to 250 N·m peak loads will reach its emergency stop torque rating and suffer premature gear tooth fatigue.

Over-engineering every axis with P0 ≤1 arcmin adds 20–40% unit cost without functional benefit on axes where P1 or P2 is technically sufficient. Apply P0 only where the positioning specification genuinely requires it.

A gearbox that meets the torque and backlash specification but creates a 50:1 inertia ratio at the motor will produce an unstable servo axis that no amount of PID tuning can fix. Calculate J_reflected before finalising ratio selection.

Selecting frame size by torque alone without verifying the output shaft radial load rating. Belt drives, open gear meshes, and chain sprockets mounted on the shaft end impose radial forces that can exceed standard shaft ratings — requiring the high-rigidity enlarged shaft of EP-AF or EP-AFR.

The P0/P1/P2 specification for EP-ABR, EP-ADR, and EP-AFR is measured at the right-angle output shaft with the bevel stage contribution already included. The stated P0 ≤1 arcmin is the total, not the planetary stage alone — there is no additional bevel penalty.

The EP-KF/KH hypoid series uses gear oil with a 0 °C minimum operating temperature. Operating below 0 °C risks inadequate lubrication and accelerated gear wear. For outdoor Korean winter applications or cold-room drives, specify any planetary series with the standard −10 °C lower limit.

क्यू
What is the practical price difference between P0, P1, and P2 backlash grades?

Exact pricing depends on series and frame size, but as a general guideline, P1 adds approximately 15–25% over P2 at the same frame and ratio, while P0 adds approximately 30–50% over P2. For a machine with 12 servo axes where only 4 genuinely require P0, specifying P1 or P2 on the remaining 8 axes can reduce gearbox BOM cost by 15–25% without any functional compromise. Korea Ever-Power supplies grade certification documentation with every unit, confirming the measured backlash value at the time of manufacture.

क्यू
Can a stepper motor be used with a precision planetary gearbox?

Yes, physically — the motor adapter plate system accommodates stepper motor flanges. However, the precision (P0/P1/P2) backlash specification of the gearbox will be underutilised with an open-loop stepper, since the stepper itself has no encoder feedback to compensate for positional uncertainty at the axis level. For stepper motor drives where backlash below 6–8 arcmin is not a functional requirement, the Korea Ever-Power Economic Line provides the cost-appropriate match. Reserve precision series for closed-loop servo applications where the encoder feedback can actually exploit the tight backlash specification.

क्यू
How do I quickly estimate the gear ratio I need?

Start with: i = Motor rated speed ÷ Required output speed. Then check the inertia ratio at that value using J_reflected = J_load / i². If the inertia ratio exceeds 10:1, try the next higher standard ratio (e.g. 25 instead of 20, or 50 instead of 40) and recalculate. The standard two-stage ratios available in most Korea Ever-Power precision series are: 12, 15, 16, 20, 25, 28, 30, 32, 35, 40, 45, 50, 60, 70, 80, 90, 100. For the EP-AD/ADS and EP-ADS series, additional non-standard ratios of 16, 21, 31, 61, and 91 are available — useful when a standard ratio does not perfectly match the required output speed.

क्यू
Is a two-stage gearbox’s backlash simply twice the single-stage value?

No — the accumulation is less severe than doubling. The correct approximation is: Total backlash ≈ Output-stage backlash + (Input-stage backlash ÷ Output-stage ratio). For example, with a two-stage gearbox where each stage has 1.0 arcmin backlash and the output stage ratio is 5: Total = 1.0 + (1.0/5) = 1.2 arcmin. The output stage dominates, and the input-stage contribution is divided by the output-stage ratio. This is why Korea Ever-Power specifies P0 two-stage at ≤3 arcmin rather than ≤2 arcmin — the bevel stage in right-angle units contributes in the same way, and the specification already accounts for this at the final output shaft measurement. For multi-axis servo systems requiring CV drive shafts to connect gearbox outputs to offset load positions, precision CV joint drive shafts allow torque transmission through angular offsets without adding backlash to the system.