Differential Signaling and Transmission Line Termination in Servo Encoder Cables

The signal interface between a servo drive and its encoder is a frequent source of installation problems that are misdiagnosed as encoder failures or drive firmware issues. 

The root cause is almost always a signaling or termination problem: single-ended signals running over distances appropriate only for differential pairs, or high-speed digital signals running on unterminated cables that behave as transmission lines. 

Understanding when a cable must be modeled as a transmission line, and how to terminate it,  eliminates a class of intermittent position errors and communication faults that cannot be resolved by tuning or configuration.

Single-Ended vs. Differential Signaling

Single-Ended Signaling

A single-ended signal uses one conductor for the signal and a shared ground return. All signals in the cable share the same ground wire. Problems:

• Return currents from multiple signals flow through the shared ground conductor.
• High currents raise the ground potential at the receiver relative to the transmitter, reducing noise margin.
• Ground connections separated over long distances create ground loops, especially when signal grounds are connected to earth ground at both ends.
• Any noise induced on the cable appears directly in the received signal.

Single-ended signaling is not suitable for connecting an encoder to a servo drive. It is only appropriate for on-board signal routing where a common ground plane eliminates the return current and ground loop problems.

Differential Signaling

Differential signaling transmits two complementary signals (equal in magnitude, opposite in polarity) to a differential receiver that computes the difference. Each signal has its own dedicated return conductor.

Noise immunity mechanism: Common-mode noise (interference that appears identically on both conductors) is subtracted at the receiver and has no effect on the output. 

This is the operating principle of twisted pair cables: the equal and opposite currents in the two conductors produce equal and opposite electromagnetic fields that cancel each other, reducing both emission and susceptibility.

The RS-422 standard (ANSI TIA/EIA-422) implements differential signaling for long-distance data transmission:

• Driver outputs (A+ and A-) must not exceed ±6 V with respect to ground.
• Differential voltage between A+ and A-: > ±2 V and not exceeding ±10 V.
• Receiver input sensitivity: 200 mV minimum differential voltage for valid state changes.
• The RS-485 standard is similar but supports multiple drivers; RS-422 supports only one.

All standard digital encoder outputs are RS-422 compatible. Differential signaling over twisted pairs is required for all encoder-to-drive connections.

When a Cable Must Be Modeled as a Transmission Line

A cable behaves as a simple lumped resistance-capacitance circuit at low frequencies, and as a distributed transmission line at high frequencies, where wavefront reflections become significant.

Rule of Thumb: Analog Signals

For analog signals, model the cable as a transmission line if:

Cable length > (wavelength of highest frequency) / 4

Example calculation: A sin/cos encoder with 256 periods/revolution rotating at 100 rev/sec produces sine waves at 25,600 Hz ≈ 25 kHz.

Wavelength = (speed of light × velocity factor) / frequency.

Secure Your Components Stock Now with Torquety

Reliable automation components for high-performance applications.

Check Availability Chat on WhatsApp

For typical twisted pair cable: velocity factor ≈ 0.66.

Wavelength = (3 × 10⁸ × 0.66) / 25,000 ≈ 7,920 m.

One-quarter wavelength ≈ 1,980 m.

Conclusion: For analog encoder signals at typical motor speeds, the cable does not need to be modeled as a transmission line at any practical installation length.

Rule of Thumb: Digital Signals

For digital signals, model the cable as a transmission line if:

Propagation delay through cable > 10% of signal risetime.

Example: At 10 Mbps, a typical RS-422 driver risetime is 10 ns. Propagation delay of typical cable: approximately 1 ns/ft.

10% of risetime = 1 ns → maximum cable length without transmission line behavior = 1 ft (0.3 m).

Conclusion: At 10 Mbps digital encoder output frequencies, any cable longer than ~0.2 m must be modeled as a transmission line. This includes virtually all real-world encoder cable runs.

Transmission Line Termination Methods

When the cable must be modeled as a transmission line, the goal of termination is to eliminate reflections by matching the termination impedance to the cable’s characteristic impedance.

For a typical twisted pair cable: characteristic impedance Z ≈ 100–120 Ω.

When a propagating signal encounters an impedance mismatch at the cable end, a reflection is generated. This reflection returns to the source, encounters another mismatch, generates additional reflections, producing the characteristic ringing pattern visible on signal edges. Ringing causes false state changes and degrades noise margin.

Method 1: No Termination

Appropriate when:

• Data rate < 200 kbps, or
• Signal risetime > 4 × cable propagation delay

The receiver input resistance (~4 kΩ for RS-422) generates a reflection, but at low data rates the reflection damps before the next transition. This is the simplest approach, requiring no additional components.

Tradeoff: Minimizes power consumption; unsuitable for high data rates or long cables.

Method 2: Series (Source) Termination

Series resistors at the driver output, chosen so that (driver output impedance + series resistor) = characteristic impedance Z.

• A reflection still occurs at the receiver end.
• But it encounters proper termination when it returns to the driver, eliminating secondary reflections.
• Power dissipation is lower than parallel termination.
• Data rate must remain low (single reflection settles before next transition).

Method 3: Parallel Termination

A single resistor across the differential inputs at the receiver, with value equal to Z (±20%):

• Most widely used termination method
• Effectively makes the cable appear resistive, no impedance discontinuity at the receiver
• Supports the highest data rates and longest cable runs.
• Increases driver current demand, the termination resistor draws continuous current.

For RS-422 differential pairs, the resistor is placed across the + and – inputs of the differential receiver. For bidirectional bus signals (CAN), termination is required at both ends.

Method 4: AC Termination

A capacitor in series with the termination resistor:

• During a state change: capacitor acts as a short → parallel termination behavior
• During steady state: capacitor charges and blocks DC → unterminated behavior

Tradeoff: Reduces DC power consumption vs. parallel termination, but introduces an RC time constant that limits maximum data rate.

Capacitor value selection: C ≤ (round-trip cable delay) / Z

For a 100 ft cable with Z = 100 Ω and propagation delay 1.6 ns/ft: C ≤ (100 ft × 2 × 1.6 ns/ft) / 100 Ω = 3,200 pF

Maximum switching rate with this C: C × Z ≤ 10% of unit interval → switching rate ≤ 312.5 kHz at 3,200 pF and 100 Ω.

Comparison Summary

Pull-Up Resistors and 3.3 V Encoder Outputs

Some encoder digital outputs operate at 3.3 V (driven directly from FPGA I/O). If the receiving circuit has pull-up resistors to 5 V, pulling a 3.3 V output to 5 V exceeds the output drive capability of the FPGA. In this case:

• Do not use pull-up termination for 3.3 V encoder outputs.
• Use no-termination or parallel termination within the 3.3 V output voltage range.
• Verify the output voltage specification in the encoder datasheet before selecting termination.

Before you leave, you might find it interesting to dive deeper into: 

• Multi-Turn Absolute Encoders: Architecture and Application in Multi-Revolution Actuation Systems, 
• discover the details of Slip Ring Assemblies for Offshore Oil and Gas: ATEX Certification and Hazardous Area Requirements, 
• or check out our analysis of Optical Encoder Performance in Harsh Environments: Contamination Failure Modes and Tolerant Designs.