Why One Planetary Gearbox Series Cannot Serve All Six Robot Joints
The six axes of a standard industrial robot differ not just in torque requirement — they differ fundamentally in what physical property of the gearbox matters most. J1 and J2 are dominated by inertia and torsional stiffness requirements that standard precision planetary gearboxes cannot adequately address at their torque class. J3 is a torque-and-efficiency balance problem. J4 and J5 are primarily a packaging problem where axial depth determines whether the robot wrist stays within its target envelope. J6 is a speed-and-mass minimisation problem.
Applying the same gearbox series across all six joints — a common shortcut in early-stage robot design — results in some joints being overspecified (heavy, expensive, high inertia) and others being underspecified (insufficient stiffness or axial load capacity). The correct approach is to treat each joint as an independent selection problem, resolved in sequence from J1 outward.
| Joint | Primary Design Driver | Typical Torque Range | Typical Ratio | IP Requirement | Recommended EP Series |
|---|---|---|---|---|---|
| J1 — Waist | Torsional stiffness Inertia always >5:1 |
800–3,000+ N·m | 20:1 – 40:1 | IP65 preferred | EP-ZDS-142/190 |
| J2 — Large Arm | Torque + Stiffness Peak gravity torque |
600–2,000+ N·m | 16:1 – 25:1 | IP65 preferred | EP-ZDS-115/142 |
| J3 — Small Arm | Torque + efficiency | 250–800 N·m | 10:1 – 20:1 | IP54 | EP-ZDS-115 or EP-ZDE-160 |
| J4 — Wrist Roll | Axial depth (compact) | 20–80 N·m | 8:1 – 16:1 | IP54 | EP-ZDWE-80 or EP-ZDE-80 |
| J5 — Wrist Bend | Axial depth (compact) | 15–60 N·m | 8:1 – 16:1 | IP54 | EP-ZDWE-60/80 |
| J6 — Tool Rotation | Mass minimisation | 5–20 N·m | 3:1 – 8:1 | IP54 | EP-ZDE-60 |
J1 and J2 — Why Torsional Stiffness Matters More Than Backlash
J1 (waist rotation) and J2 (large arm) are the most demanding joints in any 6-axis robot. At J1, the entire robot body plus maximum payload rotates about the base. At J2, the combined weight of the forearm, wrist, and payload acts at maximum moment arm when the arm is fully extended horizontally. Both joints have one defining characteristic: their load inertia structurally exceeds the servo motor rotor inertia by 10–35× even at gear ratios of 20:1.
For a 100 kg payload robot, the effective load inertia at J1 is approximately 540 kg·m² — the entire robot body and payload rotating about the base. A large servo motor for this class has rotor inertia J_motor ≈ 0.15 kg·m². At 20:1 gear ratio: J_reflected = 540/20² = 1.35 kg·m², giving an inertia ratio of 1.35/0.15 = 9:1 — well above the “safe” 3:1 target. At J2 with 20:1 ratio, the ratio improves to approximately 2:1, making 20:1 the preferred ratio for J2.
The Engineering Solution: Torsional Stiffness Raises the Resonant Frequency
When inertia ratio exceeds 3:1, the standard approach — increasing servo Kv gain — excites the drivetrain’s mechanical resonant frequency. For J1 and J2, this resonant frequency must be pushed above the servo control bandwidth (typically 50–100 Hz for robot joint controllers) to prevent oscillation. The resonant frequency of the load-gearbox system is:
This calculation explains why robot OEMs historically used strain wave gearboxes (zero-backlash, extremely high stiffness) for J1 and J2, and why the EP-ZDS high-stiffness series — with torsional stiffness up to 130 N·m/arcmin and 28,000 N axial capacity — is the relevant EP series for these joints rather than the standard EP-ZDE. The backlash specification (<8 arcmin for EP-ZDS) is secondary to the Ct value at this axis.
- Torque: calculate full body + payload inertia × peak angular acceleration, SF = 2.0–2.5
- Stiffness: Ct ≥ 44 N·m/arcmin (EP-ZDS-142 or -190)
- Axial: typically low at J1 (waist is horizontal) — EP-ZDE-160 may suffice if no vertical offset
- IP65 for welding and automotive body-shop environments
- Ratio: 20:1–25:1 to bring inertia ratio below 10:1
- Torque: gravity torque at full horizontal extension + acceleration torque, SF = 2.0
- Use 20:1 ratio to reach inertia ratio ≈ 2:1 (see calculation above)
- Stiffness: Ct ≥ 20 N·m/arcmin — EP-ZDS-115 at 20:1 delivers Ct = 22 N·m/arcmin
- Axial: significant — arm weight creates axial load on J2 output shaft; verify against limit
- IP65 for harsh environments; IP54 acceptable for clean room or general automation
J3 — Small Arm: The Torque-Efficiency Balance Point
J3 drives the forearm, wrist, and payload — typically 50–80 kg in a 100 kg payload robot. At maximum extension, this creates a gravity torque of 350–500 N·m. Combined with acceleration torque and a service factor of 1.75 for moderate shock, the required output torque is typically 600–900 N·m. This positions J3 at the boundary between the EP-ZDE-160 (rated to 800 N·m) and the EP-ZDS-115 (rated to 260 N·m at 20:1, or 780 N·m at a two-stage ratio through EP-ZDS-142).
At J3, the inertia ratio at 16:1 is approximately 1.7:1 — ideal territory for stable servo tuning without needing exceptional torsional stiffness. This makes J3 the first joint where efficiency (and therefore heat management) becomes a relevant differentiator. A 96% single-stage efficiency at EP-ZDE-160 produces significantly less heat in the arm housing than a two-stage unit at 94% efficiency during continuous-duty pick-and-place cycles.
| Configuration | Max Torque | Eficiência | Ct (N·m/arcmin) | Weight (2-stage) | Best for J3 |
|---|---|---|---|---|---|
| EP-ZDE-160, 16:1 | 800 N·m | 94% | 38 | 22 kg | ✅ T ≤ 700 N·m |
| EP-ZDS-142, 16:1 | 910 N·m | 94% | 44 | 18.5 kg | ✅ High-torque J3 |
| EP-ZDS-115, 20:1 | 260 N·m | 94% | 22 | 11.6 kg | ⚠ Only if T ≤ 250 N·m |
J3 decision rule: If the combined torque requirement (gravity + acceleration × SF) exceeds 700 N·m, specify EP-ZDS-142 at 16:1. If it falls below 700 N·m and IP65 is not required, EP-ZDE-160 at 16:1 is the more cost-effective choice with equivalent efficiency. The EP-ZDS-142 delivers higher torsional stiffness (44 vs 38 N·m/arcmin) and IP65 as additional engineering margin for J3 applications where the arm housing faces environmental exposure.
J4 and J5 — Wrist Joints: Where Axial Depth Defines the Design
Robot wrist joints J4 (roll) and J5 (bend) have comparatively modest torque requirements — typically 20–80 N·m depending on wrist mass and tool payload. The design challenge at J4/J5 is not torque — it is physical space. The wrist must fit within the robot arm envelope, and every millimetre of gearbox axial depth directly adds to the wrist outer diameter or length. In collaborative robot designs targeting a 100 mm wrist diameter, the difference between an inline EP-ZDE-80 and a right-angle input EP-ZDWE-80 at J4 is the difference between a feasible and an infeasible wrist cross-section.
The right-angle input EP-ZDWE series has wider backlash than the inline EP-ZDE at the same frame size (<25–30 arcmin vs <8 arcmin), as explained in the backlash guide. For J4/J5 in servo-controlled robots, this is not a concern — the servo position loop compensates for the backlash completely in closed-loop position mode. The backlash becomes relevant only in open-loop stepper systems, which are not used for precision robot joints.
- Wrist outer diameter target ≤ 130 mm
- Motor cannot be coaxially stacked with the gearbox output
- Collaborative robot wrist where cable routing requires the motor to exit laterally
- Servo-controlled axis (closed-loop position feedback)
- Wrist envelope allows coaxial motor + gearbox stacking
- Positioning accuracy requirements require <8 arcmin backlash for partial open-loop holding
- Industrial robot (not cobot) where wrist size is less constrained
- Force-control mode where gearbox stiffness is critical
J6 — Tool Rotation: Mass Is the Primary Specification Criterion
J6 rotates the end-effector or tool. It has the lowest torque requirement of any joint (typically 5–20 N·m), the highest continuous speed (often 360–720 rpm output), and the tightest mass budget — because every gram added at J6 adds to the load torque at J5, J4, J3, J2, and J1 in a compounding chain. The correct approach is to specify the smallest EP-ZDE frame that meets the torque requirement, choose a single-stage unit for maximum efficiency, and minimise mass absolutely.
| EP-ZDE Frame | Torque @ 3:1 | Torque @ 5:1 | Weight (1-stage) | Max Input Speed | J6 Suitability |
|---|---|---|---|---|---|
| EP-ZDE-60 | 12 N·m | 16 N·m | 0.9 kg | 4,500 rpm | ✅ Best for most J6 |
| EP-ZDE-80 | 40 N·m | 50 N·m | 2.1 kg | 4,500 rpm | ⚠ Heavy payload tools only |
| EP-ZDE-40 | 4.5 N·m | 6 N·m | 0.4 kg | 4,500 rpm | Lightest; for tool changers <5 N·m |
J6 rule of thumb: Select EP-ZDE-60 at 3:1 or 5:1 for standard 100 kg payload robot J6. The inertia ratio at J6 is excellent (≈1.1:1 at 3:1 ratio), efficiency is 96% (single stage), and 0.9 kg gearbox weight adds negligible load to upstream joints. Reserve EP-ZDE-80 for heavy-tool applications where tool mass exceeds 15 kg and tool rotation torque peaks above 30 N·m.
Complete Axis-by-Axis Selection Matrix — 100 kg Payload 6-Axis Robot
The following matrix consolidates the complete specification recommendation for a 100 kg payload, 1.5 m reach, 6-axis industrial robot. All torque values include a service factor of 2.0 for J1/J2, 1.75 for J3, and 1.5 for J4–J6. Adjust frame size proportionally for lighter-payload robots by scaling torque requirements.
| Joint | T_required (N·m) | Razão | Inertia Ratio | Min Ct (N·m/arcmin) | IP | Recommended Unit | Rated Torque (N·m) |
|---|---|---|---|---|---|---|---|
| J1 Waist | 800–2,000+ | 20:1–25:1 | ≈9:1 (structural) | ≥44 | IP65 | EP-ZDS-142, 20:1 | 910 |
| J2 Large Arm | 600–1,500+ | 20:1 | ≈2:1 ✅ | ≥20 | IP65 | EP-ZDS-115, 20:1 | 260 |
| J3 Small Arm | 400–900 | 16:1 | ≈1.7:1 ✅ | ≥30 | IP54 | EP-ZDS-142, 16:1 | 910 |
| J4 Wrist Roll | 20–80 | 8:1 – 16:1 | ≈1.6:1 ✅ | ≥4 | IP54 | EP-ZDWE-80, 8:1 | 45 |
| J5 Wrist Bend | 15–60 | 8:1 – 16:1 | ≈1.6:1 ✅ | ≥4 | IP54 | EP-ZDWE-60, 10:1 | 12 |
| J6 Tool | 5–20 | 3:1 – 5:1 | ≈1.1:1 ✅ | ≥1 | IP54 | EP-ZDE-60, 3:1 | 12 |
100 kg payload, 1.5 m reach, 6-axis industrial robot reference design. Torques include SF 2.0 (J1/J2), 1.75 (J3), 1.5 (J4–J6). Scale proportionally for different payload classes. Confirm with Korea Ever-Power application engineering for final specification.
Collaborative Robot (Cobot) Joint Selection — Where the Specification Differs
Collaborative robots (cobots) operate alongside human workers without protective fencing, which imposes design constraints that differ significantly from conventional industrial robots. The payload class is typically lower (3–25 kg versus 50–200 kg for industrial robots), the arm speed is deliberately limited, but the wrist diameter and overall form factor targets are more demanding — cobots must be visually compact and ergonomic.
Korean cobot OEMs in Suwon, Seongnam, and Ansan typically target wrist diameters of 60–100 mm for their product lines. At these dimensions, the right-angle input EP-ZDWE series at J4 and J5 is not merely preferred — it is often the only viable solution within the target wrist envelope. The EP-ZDWE-60 at 1-stage (L1 = 150 mm, total height L12 = 93 mm) allows the motor to route inside the arm body while keeping the wrist cross-section within 100 mm.
- Lower payload → smaller frames: 10 kg cobot J1 uses EP-ZDS-115 instead of EP-ZDS-190; J6 uses EP-ZDE-40 at 0.4 kg
- Force-torque sensing at J6: if backdrivability is required for force control, verify that gearbox efficiency is sufficient for reliable back-calculation of joint torque from motor current
- Noise: cobots operate near human workers — EP-ZDE/ZDS noise levels (55–70 dB(A)) are within acceptable range; avoid 3-stage units which trend toward the upper end
- IP54 is generally sufficient for typical cobot deployments unless the cobot is in a food-processing or washdown zone — in which case IP65 (EP-ZDS) applies
Three Specification Mistakes Robot OEMs Commonly Make
Applying EP-ZDE across all joints means J1/J2 are under-stiffness (Ct too low, resonance risk) and J6 is overweight. Using EP-ZDS across all joints adds 12–30 kg of unnecessary mass to the distal joints, compounding upstream torque requirements and reducing dynamic performance. The correct BOM has at least three different EP series across the six joints.
Engineers sometimes specify <3 arcmin backlash at J1/J2 believing this improves precision. At these joints, the dominant position error under load is torsional elastic deflection (θ = T/Ct), not backlash. At 1,000 N·m on EP-ZDE-160 (Ct=38), elastic deflection is 26 arcmin — far larger than any backlash specification. Tightening backlash from 8 to 3 arcmin saves 5 arcmin while ignoring 26 arcmin of load-dependent error. Specifying EP-ZDS with Ct=130 reduces the same elastic deflection to 7.7 arcmin — a 3.4× improvement for the same or lower cost.
Korean automotive body-shop robots operate in welding spatter, cooling mist, and periodic line-washdown environments. IP54 sealing resists splash but not sustained exposure or pressure washing. J1/J2 gearboxes — the largest and most expensive in the robot — are typically at the base, closest to floor-level splash and washdown water. An IP54 unit in this environment has an effective service life of 3,000–5,000 hours before lubricant contamination. Specifying IP65 (EP-ZDS) at J1/J2 from the outset costs less than one unscheduled replacement and line stoppage.
Provide your robot payload class, arm reach, cycle time, and operating environment. Korea Ever-Power’s application engineering team will return a complete joint-by-joint EP series specification with torque margins, inertia ratios, and torsional stiffness analysis — in Korean and English — at no charge for qualified OEM projects.
View specifications →
View specifications →
View specifications →
Editor: Cxm
