Linear Encoder

What is a Linear Encoder

A linear encoder is a sensor, transducer, or readhead paired with a scale that encodes linear position. The scale carries a precise periodic structure (the graduation), and the readhead scans it without contact, converting the encoded geometry into an analogue sine and cosine pair or a digital position word. It is the linear counterpart of the rotary encoder, and together they form the position feedback layer of nearly every closed-loop motion system, sitting alongside the velocity and current loops as the outermost control loop.

The decisive engineering property of a linear encoder is that it is a direct measuring system. The scale is fixed to the machine bed and the readhead rides the carriage, so the measurement captures the real position of the moving part. Any backlash in the ballscrew, any pitch error in the screw, any thermal growth of the drive train, and any compliance in the coupling all lie inside the measured loop and are therefore corrected by the servo. A rotary encoder mounted on the motor shaft, by contrast, sees only the motor angle and must infer linear travel through the transmission; it cannot detect those mechanical errors. This is why machine tools and stages that require positioning accuracy better than roughly 10 micrometers close the loop on a linear encoder rather than on a motor encoder alone.

A linear encoder system has three functional layers. First, the scale: a glass, glass-ceramic, or steel substrate carrying the graduation, in lengths from a few tens of millimeters to over 4 meters in one piece, and beyond 20 meters in tape form. Second, the readhead, which contains the light source or magnetic sensor, the scanning optics or coil array, the photodetectors or magnetoresistive elements, and the interpolation electronics that subdivide each signal period. Third, the interface, which delivers the position to the control system as analogue 1 Vpp sine and cosine, digital TTL quadrature, or a serial absolute protocol such as EnDat, BiSS, Fanuc, Mitsubishi, or DRIVE-CLiQ.

The historical roots run through optical metrology. Moire-fringe and diffraction-grating scales matured through the 1950s and 1960s, and photolithographic graduation processes let manufacturers such as HEIDENHAIN apply periodic structures with periods of tens of micrometers onto glass and steel carriers. The arrival of dedicated interpolation ASICs, then digital signal processing, pushed effective resolution from micrometers into the nanometer range while the physical graduation period stayed in the tens of micrometers. Magnetic and inductive scales followed, trading ultimate resolution for tolerance of oil, coolant, and contamination on the shop floor.

In application scale, linear encoders span an enormous dynamic range. At the high-resolution end, exposed optical and interferometric scales deliver 1 nanometer resolution for semiconductor lithography and wafer inspection stages. At the rugged end, magnetic tape encoders with a 1 to 5 millimeter pole pitch survive on plasma cutters and woodworking routers where an optical scale would clog within a shift. No single encoder spans that range; selection is the act of mapping a specific axis, with its accuracy target, travel length, speed, and contamination level, onto the right physical principle and construction.

Encoder Classifications

Before choosing a technology, an engineer must fix two orthogonal classifications: how the scale encodes position (absolute versus incremental), and how the system is built (exposed versus sealed). Getting either wrong forces a redesign later, because they dictate the wiring, the homing strategy, the contamination tolerance, and the achievable accuracy. The table below summarizes the absolute versus incremental decision.

Incremental encoders carry a single uniform grating plus one or more reference marks. The readhead counts signal periods relative to the last reference mark, so the controller knows displacement, not absolute position, until it drives the axis over a reference mark to establish datum. This reference run, or homing move, happens after every power cycle. Incremental scales support the simplest and cheapest output formats (analogue 1 Vpp sine and cosine, or digital TTL quadrature with index pulse) and remain the standard for digital readouts, manual machine retrofits, and cost-sensitive OEM axes. Distance-coded reference marks shorten the homing move to a few millimeters because the spacing between marks encodes the absolute position.

Absolute encoders carry a coded track, either a pseudo-random code (PRC) or a serial absolute pattern, alongside or merged with the incremental grating. The readhead resolves the full position the instant it powers on, with no reference run, communicating over a serial protocol such as EnDat 2.2, BiSS, Fanuc, Mitsubishi, or DRIVE-CLiQ. Absolute is mandatory where a homing move would risk a collision, where the machine must restart in place after a power failure, and where functional-safety axes require a position that survives power loss. Modern CNC and direct-drive linear-motor axes increasingly default to absolute to eliminate the homing sequence entirely.

The second classification is mechanical construction. Exposed (open) encoders have no housing: the readhead flies over a bare scale with a controlled air gap and no mechanical contact. With no sealing lips to drag and no housing to constrain the scale, exposed systems reach the highest accuracy, the highest traverse speed, and the lowest hysteresis, which makes them the choice for metrology, semiconductor, and clean direct-drive stages. Sealed (enclosed) encoders house the scale and readhead inside an aluminium extrusion fitted with elastic sealing lips, often with a compressed-air purge connection. The housing keeps coolant, swarf, and dust off the graduation, protecting to roughly IP53 with purge or up to IP64, at the cost of some friction, speed, and ultimate accuracy. Sealed encoders are the default on machine tools and any contaminated environment.

These two axes combine into four practical categories: incremental exposed (high-speed OEM stages), absolute exposed (semiconductor and metrology positioning), incremental sealed (machine-tool retrofit and DRO), and absolute sealed (modern CNC axis feedback). A clear answer on both axes, before any brand or model is discussed, prevents the most expensive selection mistakes.

Sensing Technologies

Four physical principles dominate linear position encoding: optical, magnetic, inductive, and capacitive. Each occupies a different point on the accuracy-versus-ruggedness map, and there is no universal winner. The table below compares the key engineering metrics; the prose that follows explains each principle and where it belongs.

Optical encoders dominate the high-resolution and high-accuracy market. A light source illuminates a graduation of fine lines (the signal period typically ranges from 40 micrometers down to a few micrometers), and the readhead detects the resulting moire fringes, diffraction orders, or imaged pattern with a photodetector array. The analogue sine and cosine signals are interpolated electronically: a 20 micrometer signal period interpolated several thousand times yields nanometer-class resolution. HEIDENHAIN exposed LIC 4000 and LIC 4100 absolute scales reach 1 nanometer resolution over measuring lengths up to 27 meters with the serial EnDat 2.2 interface, while Renishaw RESOLUTE absolute optical encoders combine 1 nanometer resolution with high speed on a 30 micrometer pitch. The weakness of optics is contamination: oil film, dust, or condensation on the graduation degrades the signal, which is why optical scales for the shop floor are usually sealed and air-purged.

Magnetic encoders use a scale magnetized into alternating north and south poles, read by Hall-effect or magnetoresistive sensors. The pole pitch is comparatively coarse (commonly 1 to 5 millimeters), so native accuracy is lower, but modern interpolation pushes resolution to about 1 micrometer or finer, and a premium magnetic tape bonded to a matched steel substrate holds roughly plus-or-minus 5 to 15 micrometers per meter. The decisive advantage is robustness: magnetic scales shrug off coolant, swarf, oil mist, and vibration, and a non-magnetic film of contamination does not blind them. They are the standard on heavy machinery, plasma and waterjet cutters, and woodworking routers, where an optical scale would foul quickly. Magnetic scales tolerate a larger and less critical read gap than optics, easing installation on long or imperfect axes.

Inductive encoders sense the position of a passive scale by the inductive coupling between a transmitter coil and receiver coils in the readhead, with the scale modulating the field. Mitutoyo ABSOLUTE scales, such as the AT715 and AT500 series, use this electromagnetic-induction principle: the AT715 is rated to IP67 with accuracy of plus-or-minus 5, 7, and 10 micrometers across the 100 to 500, 600 to 1800, and 2000 to 3000 millimeter length bands, and the AT500 reaches 0.05 micrometer resolution at a maximum response speed of 150 meters per minute. Inductive technology is immune to oil and coolant films and survives the harsh machine-tool environment, which is why inductive sealed scales dominate CNC retrofit and DRO installations.

Capacitive encoders measure the changing capacitance between a patterned scale and the readhead electrodes as they move. They are inexpensive and compact, which is why they appear in digital calipers and low-cost position devices, with resolution around 1 micrometer. Capacitive scales tolerate liquids reasonably well but are more sensitive to humidity, conductive contamination, and dielectric variation than the other principles, so they are rarely used for precision machine feedback. For completeness, eddy-current scales (varying the permeability or conductivity along the scale and detecting the inductance change) occupy a niche for rugged, non-contact sensing in demanding thermal and contamination conditions.

Scales, Materials and Standards

The scale substrate and its graduation set the ceiling on system accuracy and, crucially, govern how the encoder behaves with temperature. The two variables that matter most are the graduation accuracy itself and the coefficient of thermal expansion (CTE) of the carrier material, because thermal growth of the scale routinely exceeds the rated graduation accuracy on any axis longer than a few hundred millimeters.

Three substrate families cover most industrial encoders. Glass and glass-ceramic scales carry photolithographic graduations with the finest periods and the best graduation accuracy; HEIDENHAIN DIADUR glass scales sit near a CTE of 8 x 10^-6 per kelvin. Steel scales and steel tapes (for example HEIDENHAIN AURODUR graduation on steel) trade a slightly coarser limit for ruggedness and very long single-piece lengths, with a CTE near 10 x 10^-6 per kelvin that conveniently matches steel and cast-iron machine beds, so scale and machine grow together. Low-expansion substrates, such as Renishaw ZeroMet spar or glass-ceramics like Zerodur, reach a CTE near zero for metrology that must hold accuracy across a temperature swing without active compensation.

Thermal behavior is the most under-appreciated selection factor. Because a steel scale grows about 10 micrometers per meter per degree Celsius, a 1 degree rise on a 1 meter axis can introduce 10 micrometers of error, dwarfing a plus-or-minus 3 micrometer graduation grade. Two strategies resolve this: CTE matching, where the scale expansion coefficient is chosen to match the workpiece or machine bed so they grow in step and the relative error cancels, and thermal stabilization, where a near-zero-expansion scale is combined with controller compensation or a temperature-controlled enclosure. All accuracy grades are certified at the ISO 1 reference temperature of 20 degrees Celsius; a calibration chart is only valid at that reference unless compensation is applied.

The standards landscape for linear encoders spans the component datasheet and the machine that hosts it. The table below lists the references engineers cite most often.

Note that ISO 230-2 and VDI/DGQ 3441 describe the accuracy of the whole machine axis, not the encoder in isolation. The encoder graduation accuracy is necessary but not sufficient: mounting strain, Abbe offset, guideway straightness, and thermal state all add to the axis budget. A datasheet figure of plus-or-minus 3 micrometers describes the scale on the calibration bench at 20 degrees Celsius, not the installed axis on the shop floor.

Key Specification Parameters

A linear encoder datasheet lists many numbers, but only a handful drive the selection decision: accuracy grade, resolution, signal period and interpolation, sub-divisional error, repeatability and hysteresis, maximum traverse speed, measuring length, and the interface. Each is explained below, with the traps that catch first-time buyers.

Accuracy grade is the maximum deviation of the reported position from the true position over the measuring length, certified at 20 degrees Celsius. It is quoted either as an absolute figure per length band (for example HEIDENHAIN LIC scales in plus-or-minus 3 micrometer and plus-or-minus 5 micrometer grades, or Mitutoyo AT715 at plus-or-minus 5, 7, and 10 micrometers across rising length bands) or as a per-meter figure (for example Renishaw RTLA tape at plus-or-minus 5 micrometers per meter). The accuracy grade is the single most important spec and must be backed by a calibration certificate traceable to a laser interferometer.

Resolution is the smallest position step the system outputs. It is set by the signal period divided by the interpolation factor: a 20 micrometer signal period interpolated 4096 times gives roughly 5 nanometer resolution. The fatal beginner error is to equate resolution with accuracy. A magnetic encoder can advertise 1 nanometer resolution and still hold only plus-or-minus 10 micrometers per meter accuracy. Resolution is how finely the display can change; accuracy is how truthful that change is.

Signal period and interpolation describe the analogue layer beneath the resolution. Optical scales use signal periods of 4 to 40 micrometers; the interpolation electronics subdivide each period into thousands of counts. A finer signal period and a higher interpolation factor both increase resolution, but interpolation also amplifies any imperfection in the sine and cosine signals, which leads directly to the next parameter.

Sub-divisional error (SDE), also called interpolation error, is the position error that repeats within each single signal period because the sine and cosine signals are never perfectly sinusoidal, equal in amplitude, or 90 degrees apart. SDE is what limits the usable accuracy and the smoothness of low-speed motion and velocity control; it is independent of the graduation accuracy and is a key differentiator between premium and budget readheads. On a high-grade optical encoder SDE can be tens of nanometers; on a coarse magnetic system it can reach a micrometer or more.

Repeatability and hysteresis describe consistency. Repeatability is the spread of readings when the axis returns to the same point from the same direction; it is usually a small fraction of the accuracy grade and is what matters for relative moves and back-to-the-same-place operations. Hysteresis (reversal error) is the difference in reading at one point approached from opposite directions, and is minimized in exposed encoders because there is no sealing-lip friction to deflect the readhead.

Maximum traverse speed must exceed the axis rapid rate, and it interacts with resolution: a serial absolute interface has a finite clock, so very high resolution at very high speed can saturate the data rate. Sealed encoders are speed-limited by the sealing lips (often a few meters per second), while exposed encoders reach higher speeds. Mitutoyo quotes the AT715 at 50 meters per minute and the higher-performance AT500 at 150 meters per minute response speed, illustrating how speed and resolution trade against each other.

Measuring length and interface complete the picture. Measuring lengths run from tens of millimeters to over 4 meters in one rigid scale and beyond 20 meters in tape form. The interface (analogue 1 Vpp, digital TTL quadrature, or a serial absolute protocol such as EnDat 2.2, BiSS C, Fanuc, Mitsubishi, or DRIVE-CLiQ) must match the target drive or counter exactly, and the ingress rating (IP53 with purge through IP67) must match the environment.

Selection Decision Factors

To turn the preceding five chapters into a specific model, follow the ordered decision sequence below. Most selection failures come not from one wrong number but from deciding a downstream parameter before settling an upstream one. These eight steps make a reusable RFQ template.

1. Accuracy target and error budget: Fix the required axis positioning accuracy first, then work backward. Reserve budget for mounting strain, Abbe offset, guideway straightness, and thermal state, and only then specify the encoder graduation grade, which is typically the smallest term in the budget.
2. Absolute versus incremental: Choose absolute if power-on position, no-homing safety, or collision risk demand it; choose incremental for cost-driven retrofits and DRO. This decision dictates the interface and the wiring.
3. Sealed versus exposed: Assess the dirtiest condition the axis will ever see. Coolant, swarf, and dust point to sealed (or magnetic/inductive); a clean room or metrology lab allows exposed optical for maximum accuracy and speed.
4. Sensing technology: Map accuracy and contamination jointly. Optical for the highest accuracy in clean or sealed conditions, magnetic for rugged heavy machinery, inductive for oil-immune machine-tool feedback, capacitive only for low-cost calipers.
5. Measuring length and scale form: Confirm the travel plus overtravel fits a rigid scale, or move to tape for long axes. Verify the mounting interface, the scale cross-section, and the read gap tolerance against the available space.
6. Thermal and CTE strategy: Match the scale CTE to the machine bed or workpiece so they grow together, or specify a low-expansion scale with controller compensation. Remember all grades hold only at 20 degrees Celsius.
7. Interface and protocol: Match the encoder output exactly to the target drive or counter: 1 Vpp, TTL, or serial absolute (EnDat 2.2, BiSS C, Fanuc, Mitsubishi, DRIVE-CLiQ). Confirm functional-safety capability (for example a two-channel safe-position output) if the axis is a safety axis.
8. Ingress protection and installation: Set the IP rating against the environment (IP53 with purge, IP64, IP67), and verify mounting flatness and parallelism tolerances. A strained or non-parallel scale carries its distortion straight into every reading.

One dimension first-time buyers consistently overlook is installation discipline and serviceability. Abbe error makes the mounting position a first-order accuracy factor: an angular carriage error multiplied by the scale offset becomes linear error the encoder cannot detect, so the scale should sit as close as possible to the line of action and within the specified parallelism. On the service side, check readhead-to-scale interchangeability, the availability of installation diagnostics or setup LEDs, the length of the readhead cable and its flex rating in a cable chain, and the lead time on spare scales for long axes. Established suppliers, HEIDENHAIN, Renishaw, Mitutoyo, RLS, Fagor Automation, Magnescale, and MicroE among them, maintain calibration and spare-part support that determines repair response years after purchase, which is exactly when an encoder failure stops a production line.