A rotary encoder, also called a shaft encoder, is an electromechanical device that converts the angular position or motion of a shaft or axle to analog or digital output signals [S1]. The three labels point to the same hardware family but imply different accuracy and mounting assumptions, so a process engineer reading a motor datasheet should map the term to the job before locking the part number.
"Rotary encoder" is the term Wikipedia and most US suppliers use for the general class [S1][S3]. "Shaft encoder" is a synonym for the same device, still common on US and European datasheets including US Digital and Encoder Products lines such as the Model 15S, 755A, and 802S incremental shaft encoders [S3]. "Angle encoder" is reserved by metrology-grade suppliers like Renishaw for high-accuracy partial-arc or full-ring optical systems where absolute angular position in degrees, arc-seconds, or nanometres of arc is the deliverable, not just pulses per revolution [S2]. The choice between them is driven by resolution, environmental sealing, and whether the application needs absolute or incremental feedback, which maps directly to the rotary encoder decision tree used in motion-control work.
Terminology mapping: when each word applies
Wikipedia treats "rotary encoder" and "shaft encoder" as direct synonyms in the lead sentence, with the article organised around absolute vs incremental and the four sensing technologies: mechanical, optical, on-axis magnetic, and off-axis magnetic [S1]. Renishaw separates the class by the type of motion measured (linear, partial arc, full rotation) rather than by mounting, and uses "angle encoder" for the partial-arc and full-ring optical product line where graduation accuracy, not CPR, is the headline specification [S2]. US Digital mirrors that split, calling incremental counting devices "rotary encoders" and reserving angle-encoder language for absolute, high-resolution indexing and metrology [S5]. The practical rule: "shaft encoder" and "rotary encoder" are interchangeable for sourcing motors and feedback devices; "angle encoder" usually means an absolute optical ring or partial-arc unit from a metrology vendor.
Datasheets reinforce this split. Encoder Products lists "Incremental Shaft Encoders" and "Absolute Encoders" as separate product lines with shared model families such as the Model 15S, 25SF, 755A, and 802S incremental shaft encoders, and the Model 925, 960, A36R, and A58SE absolute encoders, with bus protocols including EtherCAT, EtherNet/IP, PROFINET, SSI, CANopen, SAE J1939, IO-Link, and BiSS [S3]. DigiKey's reference guide (2024-09-02) repeats the same hierarchy: incremental vs absolute on the operating-principle axis, magnetic, mechanical, optical, and capacitive on the sensing-technology axis [S4]. The two axes are independent: a given device can be described by both, for example an "absolute optical rotary encoder" or an "incremental magnetic shaft encoder" [S1][S4].
Operating-principle split: absolute vs incremental
An absolute encoder maintains position information when power is removed and reports the actual shaft position at any time, while an incremental encoder reports motion and must be homed or indexed on power-up to know absolute position [S1][S3][S5]. US Digital frames the absolute encoder as one "whose output corresponds to a specific angle of shaft position", with the absolute code carried in multiple code rings of binary weighting for parallel units, or transmitted as a serial data word for SSI, BiSS, and fieldbus variants [S5]. The incremental unit produces quadrature pulses; the controller counts edges and infers position and speed from the count and the count rate [S1][S4].
On-shaft magnetic on-axis encoders using a 2-pole neodymium magnet commonly run from a few degrees accuracy down to under 1 degree, with resolutions as coarse as 1 degree or as fine as 0.09 degree at 4000 counts per revolution (CPR) [S1]. Optical encoders can resolve movement to 1 nm on linear scales through digital interpolation and are the dominant choice where accuracy and resolution trump environmental tolerance [S2]. Magnetic encoders trade resolution for robustness, tolerating higher humidity, dust, vibration, and condensation where an optical disc would fog, which is why they are common on outdoor solar trackers, packaging lines, and food-grade washdown axes [S5].
Sensing technology comparison for rotary encoders
Optical, magnetic, mechanical (conductive), and capacitive are the four sensing families in current use, and the decision usually comes down to resolution, environment, and cost [S1][S4][S5]. Optical units use a light source and photodiode array reading slits on a glass, metal, or plastic disc, achieve the highest resolution and accuracy, but are sensitive to dust and oil contamination and typically need an enclosure on machine tools [S1][S2]. On-axis magnetic units mount a 2-pole neodymium magnet to the motor shaft, reach sub-1-degree accuracy only with care, and are widely used where an optical disc would fail [S1]. Off-axis magnetic units use rubber-bonded ferrite magnets on a metal hub and offer design flexibility and low cost for custom pole counts [S1].
Mechanical (conductive) encoders use etched copper tracks on a PCB read by contact brushes, are the lowest-cost option, but wear out in service and are mostly used in human-interface devices like digital multimeters, not motion control [S1]. Capacitive encoders round out the field and are listed as a separate technology in DigiKey's reference guide alongside magnetic, mechanical, and optical [S4]. For higher-end feedback loops, the linear encoder family uses the same optical or magnetic sensing principles on a graduated strip rather than a disc, and the two are often specified together when an axis has both linear travel and rotary stages.
Selection criteria: resolution, environment, protocol, mounting
Four criteria drive most selections. First, resolution and accuracy: optical absolute ring encoders for metrology reach 1 nm linear equivalent or sub-arc-second angular resolution, while on-axis magnetic encoders typically bottom out at 0.09 degree at 4000 CPR [S1][S2]. Second, environment: magnetic units win where dust, humidity, or condensation would fog an optical disc, and enclosed optical units win on machine tools where ingress protection from cutting fluid and swarf is required [S2][S5]. Third, feedback protocol: incremental quadrature for cost-sensitive loops, SSI, BiSS, EtherCAT, EtherNet/IP, PROFINET, CANopen, IO-Link, and SAE J1939 for absolute multi-turn or networked axes [S3][S4]. Fourth, mounting: hollow-shaft, thru-bore, NEMA motor-mount, and modular encoder kits for large motors cover most motor frames, with stainless-steel housings (Model 802S, 858S, 865T) for washdown and corrosive areas [S3].
Absolute units that use parallel output with multiple code rings are increasingly replaced by serial or fieldbus versions where cable count and noise immunity matter [S5]. Renishaw's optical encoder scales are typically made of stainless steel or glass for stability, and the readhead is a separate non-contact unit with rideheight gap, which is why the same scale can be specified as linear, partial arc, or full ring by changing the readhead path [S2]. For motor feedback on a servo drive loop, the encoder choice cascades into current rating and feedback type; the servo drive spec checklist covers the downstream selection once the encoder resolution and protocol are fixed.
Use cases and where each label is appropriate
Industrial controls, robotics, photographic lenses, computer input devices, controlled-stress rheometers, and rotating radar platforms all use rotary encoders as the generic class [S1]. The "shaft encoder" label is most common on datasheets for general-purpose motor feedback: incremental models like the US Digital Encoder Products 15S, 25SF, 711, 715, 716, 725, 755A, 758, 802S, 858S, 25SP, and 58TP, plus programmable and thru-bore/motor-mount variants for NEMA frame motors [S3]. Machine tools, CMMs, semiconductor lithography, and large telescope pointing systems use "angle encoder" in the metrology sense, where the deliverable is an absolute angular position traceable to a calibration standard, and the scale is read by a remote readhead on a partial arc or full ring [S2][S5].
Human-interface applications, including digital multimeters, audio volume knobs, and instrument panel selectors, are dominated by mechanical (conductive) encoders because resolution is low and contact life is acceptable at hand speed [S1]. Harsh-environment solar trackers, packaging lines, and food and beverage washdown use magnetic encoders because humidity, dust, and cleaning fluids would fog or contaminate an optical disc [S5]. Automotive and off-highway use industrial-grade absolute encoders with CANopen, SAE J1939, or PROFINET, where the protocol and the connector seal are as important as the resolution [S3].
Limitations, failure modes, and what to watch for
Each technology has known failure modes. Optical encoders are very sensitive to dust and oil on the disc, which is why machine-tool variants are enclosed and rated for cutting-fluid ingress [S1][S2]. On-axis magnetic encoders can show output jitter when internal interpolation is poorly implemented, and this is normally fixed with internal sample averaging in the encoder electronics [S1]. Magnetic encoders in general trade resolution for robustness and do not reach optical-class accuracy because of inherent non-linearities in the magnetic field [S5]. Mechanical encoders are susceptible to mechanical wear at the brush- track interface and are inappropriate for continuous high-speed rotation [S1].
Incremental encoders lose absolute position on power-cycle and need a homing routine or an index pulse to re-establish a reference [S1][S3][S5]. Absolute encoders solve the homing problem but historically used parallel multi-track wiring that is bulky and noise-sensitive, and modern fieldbus variants shift to SSI, BiSS, EtherCAT, EtherNet/IP, or PROFINET to keep cable count down and noise immunity up [S3][S5]. When the encoder is part of a high-speed vision loop, the industrial camera frame rate for closure inspection article covers the matching trigger and shutter side, which is often the next bottleneck after encoder bandwidth.
Track the following two signals over the next 12 months: which protocol names (EtherCAT, PROFINET, IO-Link, single-pair Ethernet) show up on new absolute encoder datasheets from the major encoder vendors, and whether on-axis magnetic interpolation ICs break the 16-bit (65536 CPR) barrier in motor-mount packages, since both directly change the rotary/shaft/angle encoder cost/performance trade-off for new builds.
The underlying component specifications are covered under angle grinder.