Engineering Guide 13 min read

EtherCAT vs. CANopen for Robot Joints: Which Protocol Fits Your Application—

ZHR Engineering Team
February 23, 2026

Quick answer: Choose EtherCAT for multi-axis robots requiring >500Hz torque control loops (humanoid locomotion, parallel robots, cobots). Choose CANopen for ≤ axes, lower bandwidth needs, or budget-constrained systems. Use CAN FD when you need CANopen simplicity with 5—× higher bandwidth.

1. Protocol Overview at a Glance

Metric EtherCAT CANopen CAN FD RS485 (async)
Physical layer 100BASE-TX Ethernet CAN bus (2-wire) CAN FD (2-wire) RS-485 (2-wire)
Max update rate 8,000 Hz (125 µs) 500 Hz (2 ms) 2,000 Hz (500 µs) 200—00 Hz
Jitter (typical) <1 µs ~100 µs ~50 µs 1— ms
Max nodes 65,535 127 ~64 ~32
Topology Line / ring / star Bus (passive) Bus (passive) Bus (passive)
Master complexity High (requires RT-Linux or FPGA) Low Medium Very Low
Cable cost per meter ~$0.5 (CAT5e) ~$0.15 ~$0.15 ~$0.10

2. EtherCAT: When Every Microsecond Counts

EtherCAT (Ethernet for Control Automation Technology) is the de facto standard for high-performance multi-axis motion control. It achieves near-zero jitter by processing Ethernet frames on-the-fly at each slave node —no message queuing, no round-trip delays.

✓EtherCAT strengths

  • ✓Deterministic timing (<1 µs jitter across all nodes)
  • ✓Single cable for power + data (with proper topology)
  • ✓Up to 65,535 slave nodes on one master
  • ✓IEC 61158 standard —broad ecosystem
  • ✓Supports PDO (cyclic) + SDO (acyclic) in parallel

✓EtherCAT limitations

  • ✓Master requires real-time OS (RT-Linux, Xenomai, FPGA)
  • ✓Higher implementation cost per node
  • ✓Star topology requires managed switches
  • ✓Overkill for <4 axes with slow motion tasks

ZHR-P with EtherCAT: ZHR-P planetary joint modules support EtherCAT with 2,000 Hz PDO cycle time. Achievable jitter: <5 µs. Ideal for humanoid locomotion control requiring simultaneous torque commands to 12+ joints.

3. CANopen: Simplicity for Smaller Systems

CANopen (based on CAN 2.0B physical layer) is the most widely deployed fieldbus protocol in robotics, medical devices, and industrial automation. Its 2-wire bus, passive topology, and self-identifying nodes reduce integration effort dramatically for smaller robot systems.

✓CANopen strengths

  • ✓Simple 2-wire bus, no switch required
  • ✓Virtually every microcontroller has CAN peripheral
  • ✓DS402 CiA profile standard for servo drives
  • ✓Low node cost (<$2 transceiver IC)
  • ✓Proven in millions of deployments

✓CANopen limitations

  • ✓Max bandwidth 1 Mbit/s ?limits to ~500 Hz at 6— axes
  • ✓Bus contention at high node count degrades cycle time
  • ✓Single master, no redundancy
  • ✓8-byte payload limit per frame (classic CAN)

4. CAN FD: The Middle Ground

CAN FD (Flexible Data Rate, ISO 11898-1:2015) extends classic CAN with two key improvements: up to 8 Mbit/s data phase and 64-byte payload per frame. This makes it 5—× faster than CANopen on the same 2-wire bus infrastructure, without requiring Ethernet.

ZHR-P series supports CAN FD at up to 5 Mbit/s, enabling ~2,000 Hz cycle times for 4— axis systems —bridging the gap between simple CANopen deployments and full EtherCAT implementations.

5. Protocol Selection Decision Tree

≤ axes + <500 Hz control loop sufficient—

?CANopen (ZHR-H RS485/CAN). Lowest integration cost and complexity.

4— axes, 500—000 Hz needed, Ethernet not available—

?CAN FD (ZHR-P CAN FD mode). Same wiring as CAN, 5× bandwidth.

≤ axes, >1000 Hz torque loop, humanoid/exo/parallel robot—

?EtherCAT (ZHR-P EtherCAT). The only choice for deterministic multi-axis force control at this scale.

6. ZHR Protocol Compatibility

Product EtherCAT CANopen CAN FD RS485 MIT FOC
ZHR-P Series
ZHR-H Series
Xiaomi CyberGear ✓(CAN)
2026 Update — Practical Selection Guide

Real-World Decision Framework: Which Protocol Should You Choose?

The choice between EtherCAT and CANopen is rarely a simple technical comparison. In practice, the decision depends on three factors: existing system architecture, required control loop frequency, and team expertise. Below is a practical decision tree used by ZHR engineers when helping customers integrate joint modules.

Choose EtherCAT IF

  • Your control loop requires <1 kHz cycle time
  • You have 6+ axes on a single network
  • You need cycle jitter <1 μs
  • Your team has experience with Industrial Ethernet
  • You are building a new system from scratch
  • Budget for master hardware (TwinCAT, etc.) is available

Choose CANopen IF

  • Your control loop is 1-5 kHz (CAN at 1 Mbit/s is sufficient)
  • You have 1-4 axes on the network
  • You are upgrading an existing CAN/CANopen system
  • Your team has CAN experience (common in automotive)
  • Cost is the primary constraint — CAN transceivers are $2-5 vs $20-50
  • Wiring distance is <40 meters

Consider Hybrid / CAN FD IF

  • You want CAN's simplicity but need higher bandwidth
  • Your system is in transition from CANopen to EtherCAT
  • CAN FD at 5 Mbit/s bridges the gap (30-40% of EtherCAT performance)
  • ZHR-P series supports CAN FD natively
  • Medium-scale humanoid (12-20 joints) can run on CAN FD with <2 ms cycle

Latency Budget Example: Humanoid Robot with 26 Joints

Scenario CANopen (1 Mbit/s) CAN FD (5 Mbit/s) EtherCAT (100 Mbit/s)
Cycle time (26 joints) 5-8 ms 1.5-2.5 ms 0.2-0.5 ms
Control loop stability Marginal for walking Adequate for walking Excellent for running/jumping
System cost premium Baseline +15-25% +50-100%

Recommendation: For first-generation humanoids with <20 joints, CAN FD offers the best cost-performance balance. For production humanoids or systems requiring dynamic running/jumping, EtherCAT is the only viable choice. ZHR actuators support both protocols on the same hardware — you can start with CANopen and migrate to EtherCAT without mechanical changes.