Why Do Robot Joint Modules Use Dual Encoders?

Dual encoder robot joints are becoming increasingly important in high-performance robotics. But adding a second encoder is not simply a matter of installing another sensor.

The real purpose is to answer two different questions:

How much has the motor moved?

and

How much has the robot joint actually moved?

In a robot joint module consisting of a frameless torque motor, precision reducer, encoder and output flange, the motor-side encoder can accurately measure motor position and speed. However, the actual joint output can still be affected by transmission error, torsional deformation, friction, thermal drift, assembly tolerances and changes in external load.

This is why a dual encoder robot joint can provide a more complete feedback architecture. One encoder monitors the motor side, while the other measures the joint output side. Together, they allow the controller to observe both the motor dynamics and the actual joint position.

What Is a Dual Encoder Robot Joint?

A typical robot joint dual encoder architecture can be simplified as:

Motor → Motor-Side Encoder → Precision Reducer → Output-Side Encoder → Joint Output

The motor-side encoder is installed near the motor rotor or input shaft. It primarily provides high-speed feedback for motor commutation, field-oriented control (FOC), velocity control and motor state estimation.

The output-side encoder, also known as a joint-side encoder, is installed after the reducer and measures the actual angular position of the robot joint.

This architecture is commonly described as:

• Dual encoder

• Dual feedback

• Dual-loop control

• Motor-side and joint-side feedback

The two encoders do not necessarily have to use the same technology. Depending on the application, a robot joint may use incremental encoders, single-turn absolute encoders or multi-turn absolute encoders.

Why Is a Motor-Side Encoder Not Enough?

In a conventional single-encoder robot joint, the encoder is usually installed on the motor side.

The controller knows the motor position and calculates the theoretical output position according to the reduction ratio:

θout ≈ θmotor / N

where N is the reduction ratio.

This approach works well when the transmission system can be considered sufficiently rigid and predictable.

However, a real robot joint is not an ideal rigid transmission system.

The reducer may introduce transmission error, while flexible components can deform under load. Bearings, couplings and output flanges can also introduce small mechanical deviations. Temperature changes may further affect both the mechanical structure and encoder measurements.

As a result:

Calculated Joint Position ≠ Actual Joint Position

This difference is one of the main reasons why output-side encoders are used in high-performance robot joints.

A secondary encoder can provide additional feedback for identifying and compensating transmission errors, especially in robot joints affected by flexibility, backlash, friction and other mechanical disturbances.

The Core Advantage of Dual Encoders: Measuring What Happens After the Reducer

The most important concept behind a dual encoder actuator is simple:

The motor-side encoder measures the input side of the transmission.

The output-side encoder measures the actual joint output.

The controller can then compare the two feedback signals and identify differences between the motor motion and the joint motion.

For example, the motor may rotate to the commanded position while the output flange is still slightly displaced because of torsional deformation, transmission error or load-dependent mechanical behavior.

A single-encoder system may assume:

The motor has reached the commanded position, so the joint has reached the commanded position.

A dual encoder system can identify:

The motor has reached the commanded position, but the joint output still has a position error.

This is one of the fundamental advantages of dual-loop feedback in robot joint modules.

1. Improve Robot Joint Position Control

One of the main advantages of a dual encoder architecture is that the position loop can use feedback from the actual joint output.

In a conventional motor-side feedback system, the position controller primarily observes motor position.

With dual feedback, the control architecture can separate the two functions:

Motor Encoder → Motor Control

Output Encoder → Joint Position Feedback

This allows the controller to regulate not only the motor but also the actual output position after the reducer.

This architecture is particularly useful for:

• High-precision robot joints

• Collaborative robots

• Humanoid robot actuators

• Precision motion systems

• CNC rotary axes

• High-performance servo mechanisms

For applications requiring accurate trajectory tracking, output-side position feedback can help reduce the influence of transmission-related errors on the final joint position.

2. Measure and Compensate Reducer Transmission Error

A common misunderstanding is that a dual encoder can simply “eliminate backlash.”

This is not technically accurate.

A dual encoder robot joint does not mechanically eliminate backlash. Instead, the second encoder allows the control system to measure the difference between motor-side and output-side motion and compensate for part of the transmission error through control algorithms.

Depending on the reducer and mechanical design, relevant error sources may include:

• Transmission error

• Torsional deformation

• Hysteresis

• Friction

• Elastic deformation

• Assembly error

• Thermal drift

By comparing motor-side and joint-side position feedback, the controller can build an error model and apply compensation according to joint position, direction, speed, load and other operating conditions.

Therefore, a more accurate description is:

Dual encoders enable transmission-error measurement and compensation.

Rather than:

Dual encoders eliminate backlash.

This distinction is important when evaluating the actual performance of a harmonic drive robot joint or other precision reducer-based actuator.

3. Improve Actual Joint Position Feedback

High encoder resolution does not automatically mean high robot positioning accuracy.

Three technical concepts should be clearly distinguished:

Resolution

Resolution describes the smallest position increment that an encoder can distinguish.

For example, a 23-bit single-turn encoder provides:

2²³ = 8,388,608 counts per revolution

The theoretical angular increment is approximately:

0.154 arcsec per count

However, this is encoder resolution, not the overall positioning accuracy of the robot joint.

Accuracy

Accuracy describes how closely the measured position corresponds to the actual physical position.

Repeatability

Repeatability describes how consistently the system returns to the same position when the same motion is repeated.

Therefore:

High Encoder Resolution ≠ High System Accuracy

The final performance of a robot joint depends on the complete system, including encoder accuracy, reducer transmission error, bearing runout, mechanical stiffness, thermal effects, calibration and control algorithms.

This is why dual encoder feedback should be evaluated as part of the complete motion-control architecture rather than by encoder bit count alone.

4. Separate Motor Motion From Actual Joint Motion

This is another important advantage of a robot joint dual encoder architecture.

Consider a robot joint using a 100:1 reduction ratio.

The motor may rotate through a relatively large angle while the output joint rotates only a small fraction of that angle.

At the same time, the reducer and mechanical structure may deform slightly as the external load changes.

The controller can therefore calculate a theoretical joint position from the motor encoder, but this does not necessarily represent the actual output position.

The output encoder provides direct feedback from the joint side.

The control system can therefore observe two different states:

Motor Dynamics

and

Joint Dynamics

This distinction becomes increasingly important in high-speed robot motion, dynamic load changes and high-precision trajectory control.

Can Dual Encoders Prevent Position Loss After Power-Off?

The answer is:

They can help, but dual encoders alone do not automatically eliminate homing.

Whether a robot joint can retain its position after power loss depends mainly on the encoder type and the multi-turn position retention mechanism.

Incremental Encoder

An incremental encoder generally does not retain absolute position information after power loss and may require a reference or homing procedure.

Single-Turn Absolute Encoder

A single-turn absolute encoder can identify the angular position within one revolution, but it does not by itself provide unlimited multi-turn position information.

Multi-Turn Absolute Encoder

A multi-turn absolute encoder can provide both single-turn position and accumulated rotational position over multiple revolutions.

Multi-turn position retention can be implemented using different technologies, including battery-backed systems, mechanical counting mechanisms and battery-free position retention technologies.

For example, Oriental Motor's ABZO encoder uses a battery-free mechanical absolute position system that can retain position information after power is removed.

Therefore, the engineering relationship is better described as:

Dual Encoder + Appropriate Multi-Turn Absolute Feedback → Potentially Reduced or Eliminated Conventional Homing

rather than:

Dual Encoder → Automatically Eliminates Homing

Can Dual Encoders Improve Robot Functional Safety?

Yes, but an important distinction must be made:

Two encoders do not automatically mean functional safety redundancy.

If two encoders are simply connected to a conventional controller without appropriate diagnostics, fault detection and safety architecture, the system cannot automatically claim PL d, PL e, SIL 2 or SIL 3.

The value of dual feedback in safety-related applications is that the controller can compare two different position measurements.

For example:

Motor Encoder Position ≠ Joint Encoder Position

may indicate:

• Encoder failure

• Mechanical transmission failure

• Reducer damage

• Shaft or coupling problems

• Unexpected mechanical displacement

• Communication or signal errors

The control system can then detect the abnormal condition and execute an appropriate fault response.

IEC 61800-5-3 addresses safety-related encoders, while IEC 61800-5-2 covers safety-related functions for power drive systems.

Therefore, the technically correct statement is:

Dual encoders can support a functional-safety architecture, but the final PL or SIL capability depends on the complete safety-related system, diagnostics, architecture and validation.

How Do Dual Encoders Relate to Functional Safety Standards?

Functional safety in a robot joint is not determined by a single encoder.

It is based on the complete safety-related control system.

Relevant standards may include:

IEC 61508

Provides the general framework for functional safety and Safety Integrity Levels (SIL).

IEC 61800-5-2

Defines safety-related functions for power drive systems, including functions such as Safe Torque Off (STO) and other drive-related safety functions.

IEC 61800-5-3

Provides requirements for safety-related encoders used in functional-safety applications.

EN ISO 13849-1

Addresses safety-related parts of control systems and defines Performance Levels (PL).

ISO 10218-1:2025

Specifies safety requirements for industrial robot systems.

ISO 10218-2:2025

Addresses safety requirements for industrial robot applications and robot cells.

The key point is that a dual encoder can be part of a safety-related feedback architecture, but the encoder count alone does not determine the final safety level.

System architecture, diagnostic coverage, fault response, common-cause failure protection and validation must all be considered.

Dual Encoder vs. Single Encoder Robot Joints

The choice between a single encoder and a dual encoder depends on the application.

Feature Single Encoder Dual Encoder

Motor position feedback Yes Yes

Direct output position feedback Usually no Yes

Transmission error monitoring Limited Improved

Dual-loop control Limited Yes

Reducer error compensation Limited Improved

Multi-turn absolute position Depends on encoder Depends on encoder

Fault cross-checking Limited Improved

System complexity Lower Higher

Cost Lower Higher

Suitable for high-performance joints Application dependent Strong advantage

A single encoder may be sufficient for cost-sensitive robot joints or systems where the transmission accuracy is already adequate.

A dual encoder architecture becomes more attractive when the application requires high positioning performance, transmission-error compensation, advanced motion control or additional position diagnostics.

Why Are Dual Encoders Not Used in Every Robot Joint?

If dual encoders provide additional feedback, why not use them everywhere?

The answer is system complexity.

Adding an output encoder increases:

• BOM cost

• Mechanical integration complexity

• Wiring requirements

• EMC design requirements

• Calibration requirements

• Controller complexity

• Diagnostic requirements

• Assembly requirements

The output-side encoder can also be sensitive to mechanical installation errors.

Encoder eccentricity, shaft runout, flange deformation and coupling errors can all influence the measured output position.

Therefore, the objective is not to install as many sensors as possible.

The objective is to use the right feedback architecture for the required robot joint performance.

Typical Dual Encoder Robot Joint Architecture

A high-performance robot joint can be understood as three main feedback layers.

Motor-Side Feedback

The motor-side encoder primarily supports:

Commutation

Field-Oriented Control (FOC)

Current and velocity control

Motor position feedback

The feedback needs to be fast and stable enough for dynamic motor control.

Joint-Side Feedback

The joint-side encoder primarily provides:

• Actual joint position

• Output position feedback

• Transmission-error monitoring

• Mechanical state monitoring

Because it is installed after the reducer, it provides information that cannot be obtained directly from the motor-side encoder.

Controller

The controller combines both feedback sources:

• Motor Position → Motor State

• Output Position → Joint State

• Position Difference → Transmission and Mechanical Error Information

This creates a more complete feedback loop for the robot joint.

Dual Encoders Are About More Than Higher Accuracy

The most important reason for using two encoders is not simply to increase encoder resolution.

It is to make the measurement target more complete.

A single encoder primarily measures:

• Motor

• A dual encoder system measures:

• Motor + Joint

This allows the control system to distinguish between motor movement and actual joint movement.

That distinction is particularly valuable when the robot joint includes a high-ratio precision reducer such as a harmonic drive or RV reducer.

When Should You Use a Dual Encoder Robot Joint?

A dual encoder architecture is worth considering when the robot joint requires one or more of the following:

High-precision joint positioning

When the final output position matters more than the calculated motor position.

Transmission-error compensation

When reducer and mechanical transmission errors need to be measured and compensated.

Advanced motion control

When the control system needs accurate feedback from both the motor and joint sides.

Absolute multi-turn positioning

When the robot needs to retain joint position information after power loss.

Position-based diagnostics

When motor-side and joint-side positions need to be continuously compared to detect abnormal mechanical behavior.

Functional-safety architecture

When redundant or cross-checked position feedback is required as part of a complete safety-related control system.

Why Dual Encoder Feedback Matters for Next-Generation Robot Actuators?

As collaborative robots, humanoid robots and high-performance industrial robots continue to demand higher precision, faster response and better diagnostic capabilities, robot joint architectures are becoming more sophisticated.

The traditional structure:

Motor + Reducer + Encoder

is increasingly evolving toward:

Motor + Motor Encoder + Precision Reducer + Joint Encoder + Controller

The key change is not simply the addition of another sensor.

It is the transition from estimating the joint position from motor motion to directly measuring the actual joint state.

For robot joints using harmonic reducers, this distinction can be particularly important. Precision reducers provide high reduction ratios, high torque density and compact mechanical structures, but the final joint performance is still influenced by transmission error, torsional stiffness, thermal effects, mechanical tolerances and external loads.

A dual encoder architecture gives the controller access to both sides of the transmission chain.

This provides a stronger foundation for:

• High-precision position control

• Transmission-error compensation

• Joint-state monitoring

• Dual-loop motion control

• Advanced robot motion control

• Selected functional-safety architectures

In other words, a dual encoder is not simply “one more sensor.”

It creates a more complete input–transmission–output feedback chain inside the robot joint.

For next-generation robot actuators, this shift from estimating joint position to directly measuring joint behavior can be more valuable than simply increasing encoder resolution.