DEV Community: Peter Hong

How to Select the Right Robot Joint Actuator: A Practical Engineering Guide

Peter Hong — Mon, 08 Jun 2026 06:01:21 +0000

How to Select the Right Robot Joint Actuator: A Practical Engineering Guide

TL;DR: A step-by-step framework for engineers selecting robotic joint actuators — covering torque requirements, transmission type, communication protocol, and control interface. Includes downloadable spec comparison template.

The Problem: Too Many Options, Too Little Guidance

If you're building a robot arm, cobot, or humanoid and searching for actuators, you've probably noticed:

Some suppliers list peak torque, others list continuous torque
"Backdrivability" means different things to different manufacturers
Communication protocols range from CANopen to EtherCAT to RS485, with no universal standard
Price varies by 10x for seemingly similar specs

This guide gives you the decision framework I wish someone had given me when I started.

Step 1: Define Your Torque Requirements

The single most common mistake is using peak torque for sizing. Here's the correct approach:

Continuous Torque (T_cont)

This is what drives sustained operation. For a robot joint:

T_cont = m × g × L / (GR × η)

Where:

m = link mass (kg)
g = 9.81
L = distance from joint to center of mass (m)
GR = gear ratio
η = transmission efficiency (~0.6–0.8 for harmonic drives, ~0.85–0.95 for planetary)

Peak Torque (T_peak)

This matters for acceleration and emergency stops. As a rule of thumb, design for:

T_peak = 1.5× to 2× T_cont

Why This Matters

If you size on peak torque, you'll end up with an actuator that's 2× heavier and 3× more expensive than needed — and the extra mass propagates upstream, requiring larger actuators in the base joints.

Step 2: Choose Your Transmission Type

Harmonic Drive

Pros	Cons
Zero backlash (<20 arcsec)	Lower efficiency (~60-80%)
High reduction in single stage (50:1–160:1)	Cannot be backdriven without torque sensor
Compact, coaxial design	Higher cost
Good for precision positioning	Shock loads can damage flexspline

Best for: 6-DOF arms, cobots, surgical robots, semiconductor equipment

Planetary Gear

Pros	Cons
Higher efficiency (~85-95%)	Backlash (3-15 arcmin typical)
High shock load capacity	Larger diameter for same ratio
Lower cost per Nm	Higher noise at high speed
Can be backdriven with proper motor sizing	Needs multi-stage for high reduction

Best for: Mobile robots, AGVs, parallel robots, high-cycle pick-and-place

QDD (Quasi-Direct Drive)

Pros	Cons
Very high backdrivability	Low torque density
Excellent for force control	Needs extra brake for safety
Low friction, high efficiency	Complex thermal management

Best for: Humanoid robots, exoskeletons, haptic devices

Step 3: Select Your Communication Protocol

This is often the most underestimated decision.

CANopen (CiA 402)

Maturity: Extremely well-established
Speed: Up to 1 Mbps
Real-time: Good for up to 10 axes
Tooling: Widespread — LinuxCNC, ROS2 CANopen packages
Complexity: Simple wiring, differential pair

EtherCAT

Maturity: Growing rapidly in robotics
Speed: 100 Mbps
Real-time: Sub-μs DC synchronization
Tooling: SOEM, IgH, ROS2 EtherCAT drivers
Complexity: Requires master PC with compatible NIC

RS485 / Modbus

Maturity: Declining for new robotic applications
Speed: Up to 10 Mbps
Synchronization: None
Best for: Simple positioning, not coordinated multi-axis

Recommendation: If your robot has more than 4 joints and requires coordinated motion, choose EtherCAT. For 1-4 axes with simpler profiles, CANopen is sufficient and easier to set up.

Step 4: Evaluate Control Modes

Make sure the actuator supports the control modes your application needs:

Mode	When to Use
Cyclic Synchronous Position (CSP)	Trajectory following, point-to-point
Cyclic Synchronous Velocity (CSV)	Conveyor tracking, mobile base
Cyclic Synchronous Torque (CST)	Force control, compliant motion
Profile Position	Simple indexing with predefined ramps
Homing Mode	Power-on reference

For ROS2 ros2_control integration, ensure CSP and CST modes are available and well-documented.

Step 5: Spec Template

Here's a 10-point spec checklist to send to any actuator supplier:

Continuous torque (Nm) at rated speed
Peak torque (Nm), duration limit
Backlash (arcmin or arcsec)
Reduction ratio
Communication protocols supported
Control modes (CiA 402 support)
Operating voltage and current (continuous / peak)
IP rating
Temperature range
CAD model availability

If a supplier can't provide clear answers for all 10, that's a red flag.

About the Author

I design and build robotic joint actuators for industrial and collaborative robots. Our ZHR series covers 1–91 Nm continuous torque with CANopen, EtherCAT, and RS485 options. Reach out through the robotics community — I'm happy to discuss your project requirements.

Have questions about selecting an actuator for your specific robot design? Drop them in the comments below.

BLDC Motor Torque Density Optimization for Robot Joints: A Practical Design Guide

Peter Hong — Mon, 01 Jun 2026 06:37:14 +0000

BLDC Motor Torque Density Optimization for Robot Joints: A Practical Design Guide

Introduction

In robotic joint design, torque density (torque per unit mass or volume) is arguably the most critical performance metric. It directly determines whether your robot arm can lift its own weight, how fast it can move, and how compact the overall system can be.

For most industrial and collaborative robot joints, the target range is 1–5 Nm/kg depending on the application. Achieving this requires careful optimization across magnetic, electrical, and thermal domains.

This guide focuses on practical, implementation-ready techniques — not abstract theory.

What Is Torque Density?

Torque density is defined as:

Torque Density = Output Torque / Motor Mass (or Volume)

Units: Nm/kg or Nm/L

A higher torque density means:

Smaller, lighter robot arms
Higher payload-to-weight ratios
Lower material costs
Better dynamic response

The Three Pillars of Torque Density Optimization

1. Magnetic Circuit Design

The torque constant (Kt) is the foundation:

Kt = 2 × N × B × L × R

Where N = turns per phase, B = air-gap flux density, L = stack length, R = rotor radius.

Practical recommendations:

Parameter	Recommendation	Rationale
Magnet grade	N52SH or higher (Br ≥ 1.42 T)	Higher flux density = higher torque
Pole count	8–14 poles for 12–18 slots	Good balance of torque and cogging
Air gap	0.3–0.8 mm (w.r.t. diameter)	Smaller gap = higher B, but tighter tolerances
Magnet shape	Segmented arc magnets	Reduces cogging torque by 30–50%

2. Winding Design

The winding configuration directly impacts both torque and thermal performance.

Key decision: concentrated vs. distributed winding

Aspect	Concentrated	Distributed
Torque density	Higher (shorter end turns)	Lower (~10% longer end turns)
Cogging torque	Higher (needs skewing)	Lower
Manufacturing	Simpler (automatic winding)	More complex
Copper utilization	Better	Worse

Slot fill factor is the practical bottleneck. Aim for ≥ 50% fill factor:

Use rectangular (flat) wire instead of round wire → +15–20% copper volume
Needle winding for automated production
Insulation grade: Class H (180 °C) minimum for robotic applications

# Slot fill factor estimation
def fill_factor(wire_diameter, turns, slot_area):
    wire_area = 3.14159 * (wire_diameter/2)**2 * turns
    return wire_area / slot_area  # Target: ≥ 0.50

3. Thermal Management: The Real Limiter

This is the most overlooked factor. Thermal constraints, not magnetic saturation, typically limit continuous torque.

Heat sources in a BLDC motor:

Copper loss (I²R): Dominant at low speed / high torque
Iron loss (hysteresis + eddy current): Dominant at high speed
Mechanical loss (friction + windage): Minor

Thermal model:

T_motor = T_ambient + (I² × R_phase) × R_th
I_continuous = sqrt((T_max − T_ambient) / (R_th × R_phase))

Cooling strategies for robot joints:

Housing conduction — Aluminum housing with thermal interface material to the structural frame
Internal air circulation — Through ventilation holes in the rotor
Integrated liquid cooling — For high-power joints (> 500 W), micro-channel cooling in stator housing
PT100 / NTC monitoring — Thermally couple the sensor to the end-turn copper (hottest point)

💡 Rule of thumb: Every 10 °C rise halves the motor's insulation lifetime. Keep continuous operation below 120 °C for long-term reliability.

Practical Design Example

Target specification: 2 Nm/kg, 60 mm diameter, 40 mm length

Parameter	Value	Notes
Poles	10	12-slot stator
Magnet	N52SH	Br = 1.45 T
Winding	Concentrated	0.2 mm flat wire
Slot fill	55%	0.2 mm flat wire
Air gap	0.5 mm	Single-sided
Max. continuous torque	0.8 Nm	Thermal limit
Peak torque	1.6 Nm	For 5 seconds
Torque density	2.1 Nm/kg	Target achieved
Thermal resistance	1.2 °C/W	Housing-to-ambient

Integration with FOC Control

Proper field-oriented control (FOC) maximizes the usable torque:

I_d = 0 control → MTPA (Maximum Torque Per Ampere) in field-weakening region

In ROS2, this can be integrated as a motor controller node:

motor_controller:
  type: bldc_foc
  current_loop_freq: 20000  # 20 kHz
  current_loop_kp: 0.5
  current_loop_ki: 50.0
  torque_limit: 2.0  # Nm
  temperature_limit: 85  # °C
  flux_weakening: true

Conclusion

Optimizing BLDC motor torque density for robot joints is a multi-domain engineering challenge. The key takeaways:

Magnetic design sets the theoretical ceiling — choose the right magnet grade, pole count, and air gap
Winding design determines how much of that ceiling you reach — flat wire and high fill factor are essential
Thermal management is the real bottleneck — invest in cooling before chasing higher magnet grades
FOC control extracts the full potential — proper MTPA tuning adds 5–10% usable torque

For more technical details on robotic joint actuators, visit our engineering resources.

This article is based on hands-on development experience with custom BLDC actuator designs for collaborative robot joints. All torque density values are from validated prototypes.

Harmonic Drive vs Planetary Reducer: A Practical Guide to Transmission Selection for Robot Joints

Peter Hong — Sat, 30 May 2026 02:49:43 +0000

Introduction

Selecting the right transmission for a robot joint is one of the most consequential decisions in actuator design. The gearbox determines the joint's precision, stiffness, torque capacity, and cost — and the choice between harmonic drive and planetary reducers is rarely straightforward.

This guide provides a structured comparison based on application requirements, not marketing claims.

1. Fundamental Differences

Harmonic Drive (Strain Wave Gear)

|| Parameter | Harmonic Drive |
||-----------|----------------|
|| Reduction ratio | 30:1 to 160:1 (single stage) |
|| Backlash | <1 arcmin (typically <20 arcsec) |
|| Efficiency | 60-85% (varies with ratio) |
|| Stiffness | Moderate (flexspline elasticity) |
|| Torque density | 50-100+ Nm/kg |
|| Cost per unit | $$-$$$$ |
|| Service life | 7,000-20,000 hrs |

Planetary Gearbox

|| Parameter | Planetary (2-stage) | Planetary (3-stage) |
||-----------|---------------------|---------------------|
|| Reduction ratio | 10:1 to 100:1 | 100:1 to 1000:1 |
|| Backlash | 3-10 arcmin (std) / <1 arcmin (precision) | 5-15 arcmin |
|| Efficiency | 85-95% | 80-90% |
|| Stiffness | High | Moderate |
|| Torque density | 40-80 Nm/kg | 30-60 Nm/kg |
|| Cost per unit | $-$$ | $$ |
|| Service life | 10,000-30,000+ hrs | 8,000-20,000 hrs |

2. When to Choose Harmonic Drive

Harmonic drives excel in applications where position accuracy and smooth motion are critical:

Collaborative robots (cobots): The near-zero backlash (typically <20 arcsec) enables precise force control and backdrivability. Most cobots on the market use harmonic drives for these reasons.

Humanoid robots: High torque density in a compact package makes harmonic drives ideal for space-constrained joints in legs and arms. Many humanoid platforms target 70-90 Nm/kg with harmonic drives.

Precision positioning: Semiconductor manufacturing, optical alignment, and medical robotics require the repeatability that only zero-backlash transmissions can provide.

Considerations:

Harmonic drives have lower torsional stiffness due to the flexspline — this must be modeled in servo control loops
They generate more heat at high speeds than planetary gearboxes
Not recommended for high-impact loads or shock applications

3. When to Choose Planetary Reducers

Planetary gearboxes are the workhorses of industrial robotics:

Industrial robots: For welding, painting, and material handling where high stiffness and repeatability (±0.05mm) are sufficient, planetary reducers offer better value.

High-speed applications: The higher efficiency (85-95%) and lower moment of inertia make planetary gearboxes superior for fast pick-and-place operations.

High-load, low-speed: The higher stiffness of planetary gearboxes (typically 2-3x that of equivalent harmonic drives) is advantageous for heavy payload applications.

Cost-sensitive projects: When unit cost dominates the BOM, a precision planetary gearbox can achieve acceptable performance at 30-50% of the cost of a harmonic drive.

Considerations:

Standard planetary gearboxes have 3-10 arcmin backlash — unacceptable for precision force control
Precision planetary gearboxes (<1 arcmin) approach harmonic drive costs, reducing the cost advantage

4. Hybrid Approaches

Some actuator designs use both transmission types:

Two-stage joints: Harmonic primary + planetary secondary for high reduction with low backlash
Dual-drive arms: Harmonic drives on critical joints (wrist, shoulder), planetary on base joints
Cable-driven + planetary: For lightweight, high-backdrivability designs

5. ROS2 Integration Notes

When integrating either transmission type into ROS2:

ros2_control hardware interfaces must model the gearbox stiffness and friction. For harmonic drives, include the nonlinear stiffness curve; for planetary, include the backlash dead zone.
Joint state estimation benefits from transmission models — especially for harmonic drives where elastic deformation under load can be significant.
Safety constraints: The lower inertia of planetary-geared joints means lower kinetic energy in collisions — this can be an important safety consideration.

6. Practical Decision Framework

|| Criteria | Choose Harmonic | Choose Planetary |
||----------|----------------|-----------------|
|| Backlash requirement | <1 arcmin | ✅ | ❌ |
|| Peak torque requirement | High | Moderate | ✅ |
|| Speed requirement | High-speed | ❌ | ✅ |
|| Cost sensitivity | Low | ✅ | ❌ |
|| Shock load resistance | Moderate | ❌ | ✅ |
|| Compact size priority | Critical | ✅ | Moderate |
|| Service life >15,000h | Desirable | ❌ | ✅ |
|| Force control precision | High | ✅ | ❌ |

Conclusion

There is no universal "best" transmission for robot joints. The choice between harmonic drive and planetary reducers depends on specific application priorities:

Harmonic drives for precision, compactness, and zero-backlash applications (cobots, humanoids, medical)
Planetary reducers for high stiffness, high speed, high efficiency, and cost-sensitive applications (industrial robots, automation)

For applications that need both precision and robustness, consider hybrid approaches or precision planetary gearboxes with advanced compensation control.

For more detailed specifications and application engineering support, visit Zhinno Robotics.

This article was originally written for robotics engineers and system integrators evaluating joint actuator technologies.

Torque Density Optimization in Robot Joint Actuators: A Practical Analysis

Peter Hong — Sun, 24 May 2026 08:00:32 +0000

Introduction

Torque density — the ratio of output torque to actuator mass — is arguably the single most important metric in robot joint design. For collaborative robots, mobile manipulators, and humanoid platforms, every gram of joint mass directly impacts payload capacity, dynamic performance, and energy efficiency.

This article provides a quantitative analysis of torque density optimization strategies, comparing harmonic drive and planetary gearbox solutions across different joint size classes.

1. Defining Torque Density

For robot joint actuators, torque density is typically expressed as:

TD = τ_rated / m_total (Nm/kg)

Where:

τ_rated = rated continuous output torque at the joint
m_total = total mass of the actuator module (motor + gearbox + encoder + housing)

A typical collaborative robot joint achieves 50-80 Nm/kg, while high-performance designs can reach 100+ Nm/kg.

2. Harmonic Drive vs Planetary: Torque Density Comparison

Harmonic Drive (Wave Gear)

Joint Size	Torque Range	Mass	Torque Density
Small (<20mm bore)	5-15 Nm	0.3-0.5 kg	30-50 Nm/kg
Medium (20-35mm)	20-80 Nm	0.6-1.2 kg	50-66 Nm/kg
Large (35-60mm)	100-300 Nm	1.5-3.0 kg	60-100+ Nm/kg

Key advantage: Single-stage reduction ratios of 30:1 to 160:1 with near-zero backlash (<20 arcsec)

Planetary Gearbox

Joint Size	Torque Range	Mass	Torque Density
Small	3-10 Nm	0.2-0.4 kg	25-40 Nm/kg
Medium	15-50 Nm	0.4-0.8 kg	37-62 Nm/kg
Large (2-stage)	50-200 Nm	0.8-2.0 kg	62-100 Nm/kg

Key advantage: Higher stiffness, lower cost, but 3-10 arcmin backlash in single-stage configurations

3. Thermal Management and Continuous Rating

Torque density is fundamentally limited by thermal performance. Key factors:

Copper fill factor: Higher fill (45-55%) improves torque density by 15-25%
Magnetic circuit optimization: Halbach arrays can increase torque density by 10-15%
Housing thermal conductivity: Aluminum (200 W/mK) vs steel (50 W/mK) — a 4x difference
Active cooling: Liquid-cooled joints can sustain 2-3x rated torque continuously

4. Motor-Gearbox Matching

The optimal torque density requires matching the motor's torque-speed curve to the gearbox ratio.

For a given joint requirement:

Select gearbox ratio N = ω_motor / ω_joint
Motor torque requirement: τ_motor = τ_joint / (N × η)
System mass: m_total = m_motor(N) + m_gearbox(N)

The optimal N minimizes m_total while meeting torque and speed requirements. For harmonic drives, N=50-100 is typical; for planetary, N=10-30 per stage.

5. ROS2 Integration Considerations

For ROS2-based systems, torque density optimization affects:

ros2_control hardware interface: The joint torque limits must match actual thermal capacity, not just peak ratings
State estimation: Higher torque density joints may exhibit more elastic deformation under load — compensation models improve control accuracy
Safety: Lower inertia (from lighter joints) reduces kinetic energy in collisions

6. Practical Recommendations

Application	Recommended Actuator Type	Target Torque Density
Collaborative arm (7kg payload)	Harmonic drive	55-70 Nm/kg
Industrial SCARA	Planetary (2-stage)	60-80 Nm/kg
Humanoid leg	Harmonic drive	70-90 Nm/kg
Mobile manipulator base	Planetary + belt	40-60 Nm/kg

Conclusion

Torque density optimization requires a systems-level approach: gearbox selection, motor design, thermal management, and control strategy must be considered together. Harmonic drives generally offer superior torque density for precision applications, while planetary gearboxes provide better value where cost and stiffness are prioritized.

For more detailed specifications and application engineering support, visit Zhinno Robotics.

This article was originally written for robotics engineers and designers evaluating joint actuator technologies.

Harmonic Drive vs Planetary Reducer: A Practical Selection Guide for Robot Joints

Peter Hong — Sat, 23 May 2026 12:04:41 +0000

When designing a robotic joint, the choice between harmonic drive and planetary gear reducers is one of the most important decisions an engineer makes. Each type brings different trade-offs in precision, torque density, stiffness, and cost.

Harmonic drives achieve near-zero backlash (<20 arcsec) while planetary gears have <1 deg uncompensated backlash. Integrated harmonic modules achieve up to 36 Nm/kg, compared to 2.6-15 Nm/kg for planetary systems.

Harmonic drive reducers excel in cobot joints, humanoid robot arms, and precision positioning. Planetary gearboxes offer higher torque per dollar and better shock load tolerance.

For a detailed comparison with full specification data, see the complete engineering guide: https://robotics.zhinno.com/blog/harmonic-reducer-vs-planetary-reducer.html