Rotary Encoder

A rotary encoder is an electro-mechanical feedback device that converts the angular position or motion of a rotating shaft into an electrical signal a controller can read. It is one of the most widely deployed sensors in motion control, closing the loop on servo drives, robot joints, machine tool spindles, conveyor indexing, valve actuators, and crane positioning. Two families dominate: incremental encoders that emit a pulse train representing relative movement, and absolute encoders that report a unique digital code for every shaft angle.

This page is a procurement and design reference. It explains how the two families differ, which sensing technology suits which environment, how to translate PPR, CPR, and resolution bits into real loop performance, and how output stages and serial protocols decide wiring and controller compatibility. Every figure traces to a manufacturer datasheet or a published standard.

Industrial rotary incremental encoder with a thru-bore housing, a shaft inserted through the hollow bore, an anti-rotation tether bracket, and an M12 threaded electrical connector

Photo: Lambtron, CC BY-SA 4.0, via Wikimedia Commons

This guide is written for industrial purchasing engineers and machine designers. It spans 6 chapters, from what a rotary encoder is, through incremental and absolute classification, sensing technologies, resolution and signal interfaces, key spec-sheet parameters, to a structured selection sequence, with 7 selection FAQs and manufacturer comparisons. Parameters reference public standards including IEC 61800-5-2 for safe motor feedback, IEC 61508 and ISO 13849-1 for functional safety, IEC 60529 for ingress protection, and the published EnDat, BiSS-C, SSI, and HIPERFACE protocol definitions.

Chapter 1 / 06

What is a Rotary Encoder

A rotary encoder is a position feedback transducer that translates the rotation of a shaft into a quantified electrical output: either a stream of pulses that a counter accumulates, or a digital word that names the exact angle. Together with current and velocity feedback, encoder position feedback is what lets a servo or stepper system hold a commanded angle, follow a velocity profile, and coordinate multiple axes. Without trustworthy feedback, a closed-loop machine degrades into open-loop guesswork.

Structurally, an encoder has three functional layers. First, a measuring standard rotates with the shaft: a glass or metal disc patterned with optical slits, a magnetized ring, an inductive coil pattern, or a shaped capacitive electrode. Second, a sensing head reads that standard, generating raw electrical signals as the pattern sweeps past it. Third, signal-conditioning electronics interpolate, digitize, and format the output into the pulses or serial telegram the controller expects. In an absolute encoder this third stage also holds the position code and, in multi-turn variants, the accumulated revolution count.

The mechanical build splits into solid-shaft and hollow-shaft styles. A solid-shaft encoder couples to the driven shaft through a flexible coupling that absorbs misalignment and protects the bearings; common housing diameters are 58 mm and 60 mm with a synchro, clamping, or servo flange. A hollow-shaft or hollow-bore encoder slides directly onto the motor or machine shaft and is held against rotation by a torque arm, saving axial length and eliminating the coupling. Bearingless modular kits, where a sensing head reads a separate magnetic or optical ring, push the principle further for large-bore and direct-drive axes.

The historical arc runs from coarse to fine. Early position feedback used commutator-style contact discs and resolvers. The optical incremental encoder, reading light through a slitted disc with a photodetector pair, became the workhorse of numerical-control machine tools through the second half of the twentieth century. Absolute optical encoders followed, using Gray-coded discs so that only one track bit changes between adjacent positions, which removes the ambiguity that plagues straight binary discs. From the mid-2000s, self-powered multi-turn counting using the Wiegand effect let absolute encoders track revolutions through a power cut without a battery, and magnetic and inductive sensing matured into rugged alternatives to glass discs.

The application scale is broad. A simple hand-wheel detent encoder may resolve a few dozen positions per turn for an operator panel, while a precision optical encoder on a wafer stage resolves tens of millions of steps and holds accuracy to a few arc-seconds. There is no universal encoder. The engineering task is to map the axis requirement, its accuracy, speed, environment, safety rating, and controller interface, onto a specific measuring principle, mechanical mount, and protocol.

Chapter 2 / 06

Incremental vs Absolute Types

The first and most consequential branch in encoder selection is incremental versus absolute. The choice changes the wiring, the controller logic, the behavior after a power failure, and the cost. Choosing wrong forces either an unnecessary homing routine on every restart or an oversized budget on an axis that never needed absolute feedback. The table below summarizes the core differences.

PropertyIncrementalAbsolute (single-turn)Absolute (multi-turn)
Position at power-onUnknown, needs homingKnown within one turnKnown across many turns
Typical outputA/B/Z quadrature pulsesSerial code (SSI/BiSS/EnDat)Serial code with turn count
Resolution metricPPR (e.g. 1024 to 65536)Bits (e.g. 13 to 25)ST bits + turn bits
Survives power lossNo (count lost)Yes (re-reads code)Yes (battery/gear/Wiegand)
Relative costLowMediumHigh
Best fitSpeed loops, simple positioningSingle-axis positioning, safetyGeared linear axes, robots

Incremental encoders emit two square-wave channels, conventionally labelled A and B, that are 90 degrees out of phase, hence the term quadrature. The phase relationship tells the controller the direction of rotation: whether A leads B or B leads A. By counting edges, the controller derives speed and relative displacement. A third optional channel, the index or Z pulse, fires once per revolution at a fixed mechanical angle, giving a reference for homing and for detecting lost counts. Because an incremental encoder reports change rather than absolute position, the controller must establish a datum after every power-up, typically by driving the axis to the index or to a limit switch.

Absolute encoders assign every shaft angle a unique digital code, classically using a Gray-coded disc so that adjacent positions differ by a single bit, which eliminates transition glitches. The instant power is applied, the encoder reads its current code and reports true position with no motion required. This is decisive for vertical axes that could drop during a homing move, for multi-axis robots where homing all joints is impractical, and for any safety function that must know position immediately.

Single-turn absolute encoders report a unique position within one revolution and then repeat. They suit rotary tables, valve actuators, and any axis whose travel stays within 360 degrees of the encoder shaft. Multi-turn absolute encoders add a revolution counter, so they distinguish turn 5 from turn 5000. This is required wherever the load moves through many encoder turns, for example a ball-screw linear stage driven through a gearbox, or a crane hoist. The revolution count must persist while the machine is off, which leads directly into the three multi-turn counting methods covered in Chapter 5: backup battery, mechanical gear train, and self-powered Wiegand energy harvesting.

Chapter 3 / 06

Sensing Technologies

Underneath the incremental or absolute output, four physical sensing principles compete: optical, magnetic, inductive, and capacitive. Each trades accuracy against robustness, cost, and environmental tolerance, and no single principle wins every axis. The table below compares the four on the metrics that drive selection.

TechnologyAccuracy / resolutionVibration & contaminationRelative costTypical applications
OpticalHighest (to arc-seconds)Sensitive to dust, oil, shockHighMachine tools, precision servo
InductiveNear-optical accuracyExcellent, resists dirt & vibrationMedium-highRobot joints, harsh motion control
MagneticModerate (approx 0.1 deg)Excellent, tolerates 30 g+LowMobile machinery, rugged absolute
CapacitiveModerateGood, low power drawLow-mediumBattery and OEM designs

Optical sensing passes a light source, typically an LED, through a disc patterned with fine transparent and opaque lines, and a photodetector array reads the resulting light-dark pattern. Optical encoders deliver the highest resolution and accuracy of any rotary feedback and remain the reference for machine tools and demanding servo axes; precision optical units hold accuracy to a few arc-seconds. The trade-off is environmental fragility: dust, oil mist, condensation, and mechanical shock can degrade or block the optical path, so optical encoders favor sealed housings and clean or controlled environments.

Magnetic sensing reads a magnetized ring or a two-pole magnet with Hall-effect or magnetoresistive sensors that track the rotating field. Magnetic encoders are compact, inexpensive, and tolerate vibration well above 30 g along with heavy contamination, which makes them the default for mobile equipment, mill drives, and rugged absolute duty. Their accuracy is more modest, with magnetic absolute units such as the Baumer MAGRES EAM580 rated to about plus-or-minus 0.15 degree and the Pepperl+Fuchs ENA58IL specified under 0.1 degree. They are also vulnerable to strong external fields, for example from a nearby magnetic motor brake.

Inductive sensing energizes a printed coil pattern and measures how a passing conductive or patterned rotor changes the coupling, an evolution of resolver physics into a flat printed-circuit form. Inductive encoders are the middle ground: they resist vibration, dirt, and magnetic interference like magnetic units, yet reach accuracy approaching optical performance, which is why they have grown rapidly in robotics and harsh motion control. Heidenhain, which combined inductive expertise after acquiring AMO in 2015, offers inductive families such as the ECI/EQI series for exactly these duties.

Capacitive sensing rotates an asymmetric disc between electrodes; the changing overlap varies the capacitance, which is decoded into angle. Capacitive encoders inherit much of the environmental robustness of magnetic units, draw little power, and avoid the glass disc of optical designs, which makes them attractive in compact, battery, and OEM applications. They are a newer entry and generally target moderate-accuracy, cost-sensitive positioning rather than the highest-precision servo loops.

Chapter 4 / 06

Resolution, Output Signals, and Protocols

Resolution is the most quoted, and most misread, encoder number. For incremental encoders it is stated as PPR, pulses per revolution, on each channel. A quadrature decoder that triggers on both rising and falling edges of channels A and B extracts four counts from each pulse, so CPR, counts per revolution, equals four times PPR. A 2500 PPR encoder yields 10000 counts per turn after this x4 decoding. Maximum output frequency caps usable speed: at a given PPR the controller must read all edges before the next arrives, so a high-PPR encoder spun fast can exceed both the encoder electronics and the counter input bandwidth.

For absolute encoders, resolution is expressed in bits. A 13-bit single-turn encoder divides one revolution into 8192 steps, about 0.044 degree; a 17-bit device reaches 131072 steps; and high-end optical units such as the Heidenhain ROC 425 deliver 25-bit single-turn resolution, roughly 33.5 million steps per turn. Multi-turn resolution adds turn bits: 12 turn bits count 4096 revolutions, as on the Heidenhain ROQ 437, while higher-end and energy-harvesting modules extend the revolution range further. Crucially, resolution is not accuracy. A magnetic encoder may report 16-bit resolution yet hold accuracy only to plus-or-minus 0.1 degree, so the extra bits describe granularity, not truth.

Incremental output stages come in three electrical forms, and matching them to the controller input is a wiring decision, not a preference. The table below summarizes them.

Output stageSignal levelWiringNoise immunity / rangeTypical use
RS-422 line driver (TTL)Fixed 5 V differentialDifferential A/A̅, B/B̅Highest, longest cable runsServo drives, long cables, noisy plants
HTL push-pull10 to 30 V, supply-referencedSingle-ended (or with complements)Good, robust in industrial noisePLC inputs, heavy machinery
Open collectorPull-up dependentSingle-ended, needs pull-upLowest, short rangeSimple OEM, short connections

RS-422 line driver outputs, often called TTL outputs, drive each channel as a differential pair at a fixed 5 V level independent of supply voltage. Differential transmission rejects common-mode noise and supports the longest cable runs, which is why servo-drive feedback almost always uses line drivers. HTL push-pull, also called high-threshold logic or totem-pole, swings between roughly 0 V and the 10 to 30 V supply, giving large signal margins that survive electrically noisy plant wiring; it is the PLC-friendly choice. Open collector outputs need an external pull-up and suit only short, simple connections.

Absolute encoders communicate their digital position over a serial protocol rather than raw pulses. The dominant choices are summarized here. SSI (Synchronous Serial Interface) is a clocked, unidirectional protocol over an RS-422 physical layer, with classic word lengths of 13 bits for single-turn and 25 bits for multi-turn devices. BiSS-C is an open, bidirectional successor to SSI: it reuses the RS-422 physical layer, adds CRC-secured data, supports point-to-point or multi-slave topologies, and clocks up to 10 MHz on short runs, with the rate falling on longer cables as with any RS-422 link. EnDat 2.2, developed by Heidenhain, is a bidirectional digital interface carrying position, parameters, and diagnostics, used on the ROC/ROQ and ECN/EQN families. HIPERFACE and the two-wire all-digital HIPERFACE DSL, originated by SICK, integrate position and power for motor feedback. For network installations, the same data is wrapped into fieldbus and industrial Ethernet telegrams: PROFIBUS, CANopen, PROFINET, EtherCAT, and EtherNet/IP.

Chapter 5 / 06

Key Specification Parameters

Reading an encoder datasheet means separating the headline number from the parameters that actually govern machine behavior. Beyond type and resolution, eight parameters drive most selection decisions: accuracy, repeatability, maximum mechanical speed, maximum output frequency, ingress protection, operating temperature, multi-turn retention method, and functional safety rating. Each is explained below.

Accuracy is how close the reported angle is to the true angle, stated in arc-seconds or degrees, and is independent of resolution. A precision optical encoder reaches single-digit arc-seconds, while a rugged magnetic absolute encoder may hold only plus-or-minus 0.1 to 0.15 degree. Repeatability is the spread of readings when the same true position is approached repeatedly; an axis can be highly repeatable yet not perfectly accurate, which is acceptable for relative moves but not for absolute referencing. Always read accuracy and repeatability as separate specifications, never inferred from the bit count.

Maximum mechanical speed is the rpm limit set by the bearings and disc, often 6000 to 12000 rpm for bearing-style 58 mm encoders, with bearingless modular kits going far higher. Maximum output frequency, typically a few hundred kHz on incremental units, limits how fast pulses can be transmitted: at high PPR the electrical frequency rises quickly with speed, so a high-PPR encoder cannot necessarily run at full mechanical speed. The two limits must both be checked against the worst-case axis speed.

Ingress protection per IEC 60529 governs survival in dust and water. Indoor cabinet duty accepts IP54 to IP65; washdown, outdoor, and contaminated duty needs IP66, IP67, or IP69K. Manufacturers frequently rate the housing higher than the shaft seal, for example the Heidenhain ROC/ROQ at IP67 on the housing and a lower rating at the shaft inlet, so read both numbers. Operating temperature ranges commonly span roughly -40 to +85 or +100 degrees C; magnetic and inductive units generally tolerate wider thermal and shock extremes than glass-disc optical units.

Multi-turn retention method applies only to multi-turn absolutes and decides reliability and maintenance:

  • Battery-backed: a low-power circuit counts turns while the machine is off; simple and high-resolution, but the battery is a wear item requiring scheduled replacement.
  • Mechanical gear train: a geared chain of code wheels stores the turn count with no power at all; robust and maintenance-free, at the cost of size and a finite gear-defined turn range.
  • Self-powered Wiegand: energy harvested from a Wiegand wire pulse as the shaft turns powers the counter and writes the count to non-volatile memory, giving battery-free multi-turn tracking introduced in the mid-2000s and now common in servo feedback.

Functional safety rating matters whenever the drive performs safety functions. An encoder feeding a Safe Torque Off, Safely Limited Speed, or Safe Stop function must be part of a chain certified to IEC 61800-5-2, supporting SIL2 or SIL3 per IEC 61508 and Category 3 or 4, PL d or PL e per ISO 13849-1. Safety-rated encoders use protocols with embedded CRC and diagnostics, such as EnDat 2.2 Safety, BiSS Safety, or HIPERFACE DSL Safety, and vendors such as Sensata BEI Sensors offer SIL3-rated incremental encoders for these duties.

Chapter 6 / 06

Selection Decision Factors

To turn the preceding chapters into a specific part number, work through the sequence below in order. Most selection errors come not from a single wrong parameter but from deciding a later step, such as protocol, before an earlier one, such as type, is settled. These eight steps double as a reusable RFQ template.

  1. Incremental or absolute: decide first. If the axis cannot tolerate a homing move after power-up, if it is a safety axis, or if it is a multi-axis robot, choose absolute. If a quick homing cycle is acceptable and cost matters, incremental is sufficient. For axes spanning many encoder turns, choose multi-turn absolute.
  2. Resolution and accuracy target: set the required accuracy in arc-seconds or degrees first, then the resolution in PPR or bits. Remember CPR equals four times PPR after quadrature decoding, and that bits describe granularity, not accuracy.
  3. Sensing technology and environment: map contamination, vibration, and accuracy onto optical (precision, clean), inductive (precision, harsh), magnetic (rugged, moderate accuracy), or capacitive (compact, low power) per Chapter 3.
  4. Mechanical mount: solid shaft with coupling, hollow or hollow-through shaft on the machine shaft with a torque arm, or bearingless modular ring for large-bore and direct-drive axes. Confirm flange type (synchro, clamping, servo) and shaft or bore diameter.
  5. Output and protocol: for incremental, choose RS-422 line driver, HTL push-pull, or open collector to match the controller input. For absolute, choose SSI, BiSS-C, EnDat 2.2, HIPERFACE/DSL, or a fieldbus (PROFIBUS, CANopen, PROFINET, EtherCAT, EtherNet/IP) per the drive and network.
  6. Environmental ratings: ingress protection (IP54 to IP69K) per IEC 60529, operating temperature, and rated vibration and shock. Verify both housing and shaft-seal IP ratings, not just the headline number.
  7. Functional safety: if the axis performs a safety function, specify an encoder certified within an IEC 61800-5-2 chain to the required SIL (per IEC 61508) and PL/Category (per ISO 13849-1), using a CRC-secured safety protocol.
  8. Total cost of ownership: add purchase price, wiring and connector cost, homing time saved by absolute feedback, battery maintenance for battery multi-turn units, and downtime risk. An absolute encoder that removes a fragile homing routine often repays its premium across the machine life.

One frequently overlooked dimension is manufacturer serviceability and ecosystem: availability of the chosen protocol on your drive, electronic datasheet and parameter files, connector and cable compatibility, local spare-part stock, and field-replacement lead time. Heidenhain, SICK, Baumer, Pepperl+Fuchs, Renishaw, Kuebler, Hengstler, Sensata BEI Sensors, FAULHABER, and Broadcom maintain broad catalogs and support networks; aligning the encoder with a drive that natively reads its protocol, EnDat with a Heidenhain-compatible drive or HIPERFACE DSL with a SICK-aware servo, avoids integration surprises years into production.

FAQ

What is the difference between an incremental and an absolute rotary encoder?

An incremental encoder outputs a stream of pulses as the shaft turns. It reports relative motion, not position, so the controller must count pulses from a known reference and the count is lost on power failure unless a homing routine re-finds the index. An absolute encoder assigns a unique digital code to every shaft angle, usually via a Gray-coded disc, so it reports true position the instant power is applied with no homing move. Incremental encoders are cheaper and dominate speed and simple positioning loops. Absolute encoders are chosen for safety, multi-axis robots, and any machine that cannot tolerate a homing cycle after a power cut.

How is rotary encoder resolution defined, and how do PPR, CPR, and quadrature relate?

For incremental encoders, resolution is stated in PPR (pulses per revolution) on each channel. A quadrature decoder that triggers on both the rising and falling edges of channels A and B produces four counts per pulse, so CPR (counts per revolution) equals 4 times PPR. A 2500 PPR encoder therefore yields 10000 counts per turn after x4 decoding. For absolute encoders, resolution is stated in bits: a 13-bit single-turn encoder divides one revolution into 8192 steps (about 0.044 degree), and a 25-bit device reaches roughly 33.5 million steps. Do not confuse resolution with accuracy: a high count does not guarantee the position is correct to that fineness.

Which sensing technology should I choose: optical, magnetic, inductive, or capacitive?

Optical encoders deliver the highest resolution and accuracy and remain the reference for machine tools and precision servo feedback, but the glass disc is sensitive to dust, oil mist, and shock. Magnetic encoders are compact, low cost, and tolerate vibration above 30 g and heavy contamination, at the price of lower accuracy and sensitivity to nearby magnetic brakes. Inductive encoders are the middle ground: they resist vibration and dirt like magnetic units but reach accuracy close to optical, which suits robot joints and harsh motion control. Capacitive encoders share the environmental robustness of magnetic units and add low power draw, fitting battery and OEM designs. Match the technology to the contamination, vibration, and accuracy budget of the axis.

What output signals and serial protocols do rotary encoders use?

Incremental encoders use three electrical output stages: RS-422 line driver (TTL-compatible, fixed 5 V differential, best noise immunity and longest cable runs), HTL push-pull (single-ended, supply-referenced 10 to 30 V, robust in electrically noisy plants), and open collector (simple, short range). Absolute encoders use a parallel code or, far more commonly, a serial protocol: SSI (clocked, typically 13-bit single-turn or 25-bit multi-turn), BiSS-C (open, bidirectional, CRC-secured, up to 10 MHz over RS-422), EnDat 2.2 from Heidenhain, and HIPERFACE plus the two-wire HIPERFACE DSL from SICK. Fieldbus and industrial Ethernet variants (PROFIBUS, CANopen, PROFINET, EtherCAT, EtherNet/IP) embed the same position data into a network telegram.

What is the difference between a single-turn and a multi-turn absolute encoder?

A single-turn absolute encoder reports a unique position within one 360-degree revolution and resets to zero at each full turn, so it cannot tell turn 1 from turn 1000. A multi-turn encoder adds a revolution counter, reporting both the angle within a turn and how many turns have accumulated, for example 12 bits of turns counting 4096 revolutions on top of 13 to 25 bits of single-turn resolution. The turn count is preserved while powered down by one of three methods: a backup battery, a mechanical gear train, or a self-powered Wiegand energy-harvesting pulse that needs no battery. Multi-turn encoders are essential on linear axes driven through a gearbox, where position spans many motor revolutions.

How do I read encoder accuracy versus resolution on a datasheet?

Resolution is the smallest step the encoder can report; accuracy is how close the reported angle is to the true angle. They are independent numbers. A magnetic absolute encoder can offer 16-bit resolution yet hold accuracy only to about plus-or-minus 0.1 to 0.15 degree, while a precision optical encoder reaches a few arc-seconds of accuracy. Always check accuracy (often stated in arc-seconds or degrees), repeatability, and the integral non-linearity separately. Resolution sets the granularity of the feedback loop; accuracy sets how truthfully the machine knows its position. For closed-loop velocity smoothness, also look at the number of sine periods or line count, not just the digital bit count, because interpolation noise affects control quality.

What protection rating and functional safety level does an encoder need?

Match ingress protection to the installation. Indoor cabinet or clean machine duty accepts IP54 to IP65; washdown, outdoor, or contaminated environments need IP66, IP67, or IP69K, and many encoders rate the housing higher than the shaft inlet, for example IP67 housing with IP64 at the seal. For drives that perform safety functions such as Safe Torque Off or Safely Limited Speed, the encoder must be part of a feedback chain certified to IEC 61800-5-2, supporting SIL2 or SIL3 per IEC 61508 and Category 3 or 4, PL d or PL e per ISO 13849-1. Safety encoders use protocols with built-in CRC such as EnDat 2.2 Safety, BiSS Safety, or HIPERFACE DSL Safety. Specifying a safety rating you do not need adds cost; omitting one a safety machine requires fails the assessment.

Ask SpecForge AI