Startup and recovery behaviour
Define the machine sequence at power-up and after power loss: whether homing is safe and acceptable, or whether the axis must report a valid position immediately, including multi-turn state.
TECHNOLOGY / ENCODER ARCHITECTURE SELECTION
Incremental vs Absolute Optical Encoder Selection: Output Protocols, Power-Loss Behaviour and Application Fit
Choosing between incremental and absolute optical feedback decides startup behaviour, wiring, controller interface and recovery after power loss. This guide compares output protocols, battery and counter backup, application fit and failure modes, and gives buyers a decision matrix for the axis.
Published
DIRECT ANSWER
What should a buyer prioritize?
An incremental optical encoder reports position change as quadrature signal cycles and needs a reference or homing routine to establish a machine datum; an absolute optical encoder returns a unique position code over a digital interface such as BiSS-C or EnDat, so the axis knows where it is the moment power returns. The selection question is not which architecture is better but which startup, recovery and interface behaviour the machine can accept.
Choose incremental when the axis can home safely, the controller already accepts analog quadrature or square-wave signals and wiring simplicity matters. Choose absolute when homing is slow or risky, position must survive power interruptions, or multi-turn and diagnostic data add value. Require the interface protocol, position validity behaviour and any battery or counter backup terms in writing before the order is placed.
ENGINEERING CONTEXT
The decision starts with the real constraint.
Encoder architecture is often decided by habit: the team that always used quadrature keeps quoting incremental, and teams burnt by long homing cycles default to absolute without checking what the position validity actually covers. Both habits cost money in the wrong place. Incremental feedback hides its risks in the machine sequence, where a poorly designed homing routine collides with fixtures or turns every power dip into a re-reference cycle. Absolute feedback hides its risks in the small print: an absolute singleturn encoder still loses multi-turn count without backup, a battery-backed counter introduces a maintenance item nobody owns, and a protocol that the interface card half-supports surfaces as timing faults during commissioning. Datasheets deepen the confusion by quoting resolution and protocol side by side while leaving startup validity, alarm behaviour and backup policy unstated. The fix is to make the machine sequence the starting input: how the axis starts, what happens at power loss, what the controller accepts, and what evidence the supplier must provide for all three.
Discuss your requirement ↗CRITICAL REQUIREMENTS
Define these before model selection.
Interface and protocol match
Confirm the controller or drive supports the exact output: analog quadrature levels and cable limits for incremental, or the specific BiSS-C, EnDat or similar frame, clock rate and diagnostics for absolute.
Backup and validity policy
For absolute systems, state in writing what position survives power loss, whether a battery or internal counter is used, its expected life and how expiry is signalled.
Failure-mode plan
Cover signal loss, contamination, reference-mark miss and battery expiry: how each is detected by the encoder or controller and what the machine must do when it happens.
BUYER MATRIX
Compare evidence, not feature labels.
Give every shortlisted supplier the same process definition and request configuration-specific proof.
| Decision factor | Incremental optical encoder | Absolute optical encoder |
|---|---|---|
| Position at power-up | Relative cycles only; datum set by homing or reference mark | Unique position code available immediately after power-up |
| Output protocol | Analog quadrature (A/B/Z, 1 Vpp or square wave) up to high frequencies | Digital serial frames: BiSS-C, EnDat, or vendor protocols with CRC and diagnostics |
| Power-loss behaviour | Position lost; machine re-references on restart | Position retained within the stated range; multi-turn count needs backup where applicable |
| Backup requirements | None | Battery or non-volatile counter for multi-turn; battery life and expiry alarm must be specified |
| Wiring and electronics | Simple differential pairs; cable length set by signal frequency and edge rate | Clock/data lines with protocol timing limits; cable length set by clock rate |
| Controller dependency | High-speed counter inputs, homing logic in the controller | Interface card or drive with protocol support and configuration depth |
| Typical failure modes | Lost counts under noise, missed reference, signal degradation with cable length | Battery expiry, protocol timing faults, configuration mismatch between encoder and master |
| Typical fit | Axes that can home safely, high-rate velocity feedback, simple retrofit wiring | Gantries and vertical axes, long recovery cycles, machines where homing is risky or slow |
The decision is the machine sequence, not the datasheet
Both architectures measure position against an optical scale with comparable grating quality, so resolution and accuracy rarely separate them. What separates them is time behaviour: an incremental encoder answers the question how far did I move, an absolute encoder answers where am I. Every downstream difference, from wiring to homing to power-loss recovery, follows from that distinction. The selection therefore starts with the machine sequence, written down before catalogues are opened: how the axis starts, whether homing can run safely with fixtures and payloads present, what a power interruption costs in process terms, and what the controller or drive genuinely supports at the interface.
Treat the sequence as a purchase requirement, not an implementation detail. An axis that re-references for forty seconds after every intervention may be acceptable on a standalone router and unacceptable inside a production cell with a synced cycle. A vertical axis that drops under gravity when power fails may make an incremental solution a safety question rather than a convenience question. When the sequence is explicit, the architecture choice usually becomes obvious, and the remaining work is verifying that the exact model supports it.
Output protocols and what the controller really accepts
Incremental encoders deliver analog quadrature, commonly 1 Vpp sine-cosine signals, or squared digital outputs, and the usable cable length and maximum speed both depend on the signal form and the receiving electronics. The controller needs counter inputs rated for the edge or interpolation frequency at maximum velocity, and the homing logic that consumes the reference mark. These are mature, inexpensive and well understood, which is why incremental feedback remains the default for velocity loops and for retrofit work where wiring simplicity is decisive.
Absolute encoders communicate over synchronous serial protocols such as BiSS-C or EnDat, transmitting a position frame with CRC protection, status bits and often diagnostic data. The master must support the exact protocol version, clock rate, frame format and register map; an interface card that nominally lists the protocol can still fail in commissioning over clock limits or configuration depth. Cable length is bounded by the clock rate rather than by analog signal degradation. Require the protocol implementation sheet for the exact encoder model and confirm it against the master in writing, because this is the most common integration gap between a good encoder and a working axis.
- Incremental: 1 Vpp or square-wave quadrature, frequency limit sets speed and cable length
- Absolute: BiSS-C or EnDat frames with CRC, status and diagnostic data
- Master support must match protocol version, clock rate and frame format exactly
- Request the implementation sheet for the specific model, not the protocol name
Power loss, battery backup and what absolute actually guarantees
Absolute means the position code is unique over the stated range, and the range must be read carefully. A singleturn absolute encoder always knows where it is within one revolution or within the scale length, but a rotary axis that turned twelve times while unpowered has no record of that unless a multi-turn mechanism exists. Mechanical multi-turn designs are self-powered; electronic solutions count turns in a counter that needs power, supplied by a battery or preserved in non-volatile storage. Each route introduces a maintenance or expiry item with an owner: batteries age, alarms must be routed somewhere meaningful, and expiry handling should be defined in the machine logic, not discovered at a service visit.
The honest specification is a validity statement: what position the encoder reports at the moment power returns, over what range, with what confidence, and how any expired or invalid state is signalled on the interface. For linear scales the same questions apply to the absolute code length versus machine travel. Buyers who obtain this statement in writing can decide deliberately between accepting a homing routine, specifying a battery-backed multi-turn counter, or restricting machine motion so the singleturn range is never exceeded while unpowered.
Failure modes differ, and detection belongs in the specification
Incremental failure modes concentrate in signal integrity and the reference event: counts lost to noise or marginal signal quality appear as a slow position shift that no alarm reports, and a missed reference mark produces a datum offset on every re-reference. Detection therefore lives mostly in the controller, through signal-level monitoring where available, plausibility checks against limits, and a homing routine designed so a missed or dirty reference is noticed rather than trusted.
Absolute encoders shift the failure surface toward configuration and backup: protocol timing faults from marginal clocking, mismatches between encoder and master configuration, and battery expiry on multi-turn counters. In exchange, the protocol's CRC and status bits give continuous, built-in supervision that quadrature never offered. Neither architecture is more reliable in general; they fail differently, and the specification should require the detection path for each relevant fault: how signal loss, invalid position, CRC errors and battery warning appear electrically, and what the machine does in each case. An encoder whose faults are visible is worth more than one whose faults are silent, whatever the protocol.
A decision matrix for buyers
Score the axis against five questions and let the pattern choose the architecture. First, can the axis home safely and how often will it have to? Second, what does an unplanned power interruption cost, and must position survive it? Third, does the axis move, or can it move, outside a singleturn or scale-length range while unpowered? Fourth, what does the controller or drive actually support at the interface, with evidence? Fifth, who owns battery or backup maintenance if an absolute multi-turn solution is chosen? Incremental wins when homing is safe, cheap and rare, when velocity feedback quality dominates, and when wiring simplicity matters. Absolute wins when startup position has process value, when homing is slow or hazardous, and when the master's protocol support is proven.
Run the same questions for both architectures on any borderline axis, and require the supplier's written answers on startup validity, backup policy and fault signalling as part of the quotation. The comparison that survives those five questions on paper is the one that survives commissioning on the machine.
- Homing: safe, acceptable cycle time, and how often it runs
- Power loss: process cost and required position survival
- Range: motion outside singleturn coverage while unpowered
- Interface: proven master support with implementation evidence
- Ownership: named responsibility for any battery or backup maintenance
CONCLUSION
Qualify the process, then the platform.
Incremental and absolute optical encoders differ less in measuring quality than in time behaviour: how position is established at startup, how it survives power loss and how faults surface. Write the machine sequence first, verify the exact output protocol and position-validity terms against it, and put backup and fault-signalling requirements in the quotation. An architecture chosen this way matches the machine instead of the catalogue, and the difference shows up as fewer surprises at commissioning rather than as a number on a datasheet.
SOLUTION ROUTES
Continue from requirement to system.
EVIDENCE CHECKLIST
Ask for evidence that closes the decision.
- Output specification for the exact model: signal type, levels, frequency limit or protocol frame, clock rate and cable length limits
- Startup and position-validity statement: when the reported position is trustworthy after power-up and after power interruption
- Multi-turn and backup policy: singleturn or multi-turn range, battery type and life, counter backup behaviour and expiry indication
- Alarm and diagnostic list with the electrical behaviour of each fault on the interface
- Homing method, reference-mark behaviour and the supplier's commissioning requirements for the chosen architecture
FAQ
Questions engineers ask before selection.
Does an absolute encoder always eliminate homing?
It provides a valid position at power-up within its stated range, so the conventional homing move can often be removed. Whether it should be depends on the machine's safety and validation strategy, and multi-turn axes still depend on the backup mechanism having retained the turn count.
What happens to an absolute encoder after a power loss?
Within the absolute range, the position is available again immediately on power-up. Outside that range, for example turns accumulated on a rotary axis beyond singleturn coverage, position is only retained if a mechanical multi-turn mechanism or a backed-up electronic counter is specified.
Do absolute encoders need battery maintenance?
Electronic multi-turn designs that use a battery do, and the battery type, expected life and expiry indication should be part of the purchase specification with a named owner. Designs with non-volatile counter storage or mechanical multi-turn mechanisms avoid the battery item.
Is incremental feedback still a good choice for new machines?
Yes, when the axis can home safely, velocity-loop signal quality matters and the controller already accepts quadrature at the required frequency. It remains the simpler and more economical route where startup position has no process value.
What evidence should I request before ordering?
The output or protocol implementation sheet for the exact model, the position-validity statement after power-up and power loss, the multi-turn and backup policy with battery life where applicable, and the alarm and diagnostic behaviour on the interface.
AUTHORITATIVE SOURCES
Research used for this buyer guide.
TECHNICAL REVIEW
Turn the requirement into a selection brief.
Send the application, critical parameters and any drawing or process information available.
Discuss your application ↗