How to select a linear encoder

In short:Build an error budget at the tool or carriage first, then pick the technology the environment allows: magnetic tape for long, dirty axes at tens of µm per metre, capacitive for compact absolute axes with passive heads, exposed optical for µm-level accuracy in clean machines, and enclosed optical where you need that accuracy among chips and coolant. Then check ride height and alignment tolerances, thermal expansion, Abbe offset and whether the axis must know its position at power-up.

The decisions, in order

A linear encoder measures carriage position directly, removing screw lead error, belt stretch, backlash and screw thermal growth from the measurement. On linear motor axes it is essential: there is no other position feedback.

Select one in this order:

1. Error budget at the tool, probe or part: accuracy, repeatability and resolution, and the temperature range they apply over.
2. Absolute or incremental.
3. Technology: magnetic tape, capacitive, exposed optical or enclosed optical.
4. Mechanical fit: length, ride height and alignment tolerances, mounting method, Abbe offset.
5. Thermal behaviour of scale and machine.
6. Speed and interface to the drive.
7. Environment: contamination, temperature, shock, vibration, magnetic fields.

For rotary axes and the general encoder process, see how to select an encoder; this page covers what is specific to linear axes.

Step 1: build the error budget

Separate three figures, because datasheets quote them separately:

• Resolution: the smallest step reported, from 1 nm on high-end optical systems to tens of µm on simple magnetic ones.
• Accuracy: the worst deviation of the reading from true position over the travel, quoted as ±µm/m or ±µm over a stated length. Renishaw’s RTLA tape, for example, is ±5 µm/m at 20 °C, including slope and linearity (Renishaw).
• Repeatability: the spread when returning to the same position, usually much better than accuracy.

Then list everything else that adds error at the working point: thermal growth, Abbe error from carriage tilt, structure deflection, and installation error. The encoder’s accuracy is only one line in the budget.

Rules of thumb: choose resolution four to ten times finer than the accuracy you need, and finer still on linear motor axes, where the velocity loop differentiates position and coarse counts appear as velocity ripple and audible noise. If the machine returns to taught positions, budget mostly for repeatability; if it must hit programmed coordinates, budget for accuracy.

Step 2: absolute or incremental

• Incremental scales output pulses or sine waves and need a homing move to a reference mark at start-up. They are simple, cheap and fast.
• Absolute scales carry a code that gives position immediately at power-up. That matters on linear motor axes, which need the rotor (forcer) position for commutation before moving; on gantries where a homing move could collide; on long axes where homing takes time; and on multi-carriage tracks.

Absolute options exist in each technology: two-track pseudo-random coded magnetic tapes such as Bogen LMSA, capacitive systems such as Kappasense FREEe, and optical systems such as Renishaw RESOLUTE. Our absolute vs incremental guide covers the trade-offs in depth.

Step 3: choose the technology

Sources: Renishaw, Power Transmission Engineering and the Bogen and Kappasense records in our catalogue. For the physics behind these differences, see magnetic vs capacitive vs optical encoders.

Magnetic tape

A flexible steel-backed tape with magnetised poles, read by a Hall or magnetoresistive head. It cuts to length from a reel and sticks to the machine, which makes it the natural choice for long axes: Bogen’s LMSA absolute scale runs to 32.7 m. Accuracy depends on grade and pole pitch; a cover strip protects the tape against mechanical damage. Keep ferrous swarf and strong magnets away from it.

Capacitive

Electrodes on scale and head measure position by capacitance. Kappasense’s FREEe uses a passive 3 g read head with no moving cable, so several heads can share one scale, which simplifies multi-carriage systems. Design World notes capacitive encoders tolerate dust, dirt and oil better than optical and are immune to magnetic interference (Design World).

Exposed optical

A read head over a tape, spar or glass scale with a fine grating, 30 µm pitch on Renishaw’s RESOLUTE scales. It gives the highest resolution (to 1 nm) and speed (Renishaw quotes up to 100 m/s) (Renishaw). Contamination on the scale is the main risk.

Enclosed optical

The scale sits inside a sealed aluminium extrusion and a carriage inside rides on it, coupled to the machine. HEIDENHAIN’s LC 115 and LC 415 use two pairs of sealing lips with compressed air fed between them to keep chips and coolant off the graduation (Power Transmission Engineering). The coupling must allow for misalignment between machine and housing without adding error.

Step 4: ride height, alignment and mounting

Exposed read heads have a nominal gap to the scale and tolerances on gap and angles:

• Renishaw specifies ±150 µm ride height tolerance and ±0.5° yaw, pitch and roll for RESOLUTE on RTLA (Renishaw).
• Bogen’s ALS21 absolute magnetic head reads at 0.1 to 0.6 mm from the tape (Bogen).

The carriage, guide rails and mounting surface must hold these over the whole travel, including deflection under load and thermal distortion. Check the flatness and straightness of the surface the scale sticks to: on a long fabricated frame it is often the limit.

Mounting methods matter for thermal behaviour: an adhesive tape scale fixed at one datum point and free elsewhere expands at its own rate; a tape held in a rail and clamped at one point behaves similarly; a scale clamped at both ends to the substrate follows the substrate. Renishaw offers both approaches across its tape scales.

Abbe error

If the scale is offset from the line where accuracy matters (the tool tip, probe or part) and the carriage pitches or yaws as it moves, the reading and the working point move by different amounts:

Abbe error ≈ offset × tilt angle (radians)

With 1 arcsec = 4.85 × 10⁻⁶ rad, a 50 mm offset and 20 arcsec of carriage pitch give 50 × 20 × 4.85 × 10⁻⁶ ≈ 0.005 mm, or 5 µm. Mount the scale as close to the working line as you can, and use stiff, accurate guides.

Step 5: thermal behaviour

Scale and machine expand with temperature, usually at different rates:

ΔL = α × L × ΔT

Renishaw gives RTLA at 10.1 µm/m/°C (Renishaw), close to steel. Aluminium frames expand at roughly twice that. Over 1.2 m and a 2 °C change, an RTLA-type steel tape grows about 24 µm. Options:

• Accept and budget it when the range is small.
• Fix the scale to the substrate at both ends so it follows the machine, when the part and machine share a material and temperature.
• Use a low-expansion scale (Renishaw’s ZeroMet spar is 0.75 µm/m/°C) for metrology.
• Compensate in the controller with temperature sensors.

Step 6: speed and interface

• Speed: check the maximum at the resolution you will use, as many heads trade resolution for speed. Bogen’s ALS21 datasheet gives 10 m/s.
• Interface: BiSS-C or SSI for absolute; A/B/Z (RS422) for incremental; 1 Vpp sine/cosine where the drive interpolates; proprietary protocols where the drive needs them. Confirm the drive reads the protocol natively and the update rate suits the current loop on linear motor axes.
• Cable: moving cable to the read head, unless the system uses a passive head. Rate the cable for continuous flexing in a carrier.

Worked example: absolute feedback for a 1.2 m linear motor gantry

Requirement: A pick-and-place gantry X axis, 1.2 m stroke, driven by an iron-core linear motor at up to 2 m/s. Positioning accuracy ±50 µm along the axis at 20 ± 2 °C. Light dust and occasional oil mist from nearby machines. The axis must commutate and know its position at power-up without moving, because a homing move risks collision with the Y axis tooling.

Absolute: required.

Budget:

Technology:

• Exposed optical: easily accurate enough, but the dust and oil mist count against it.
• Capacitive: Kappasense FLEXe is listed at ±20 µm/m with strokes to 1.9 m, tolerant of contamination, but ±20 µm/m × 1.2 m = ±24 µm exceeds the scale allowance. FREEe is more accurate (±5 µm) but its standard strokes are 280 and 800 mm.
• Magnetic absolute tape: Bogen LMSA is a two-track absolute tape with accuracy grades from ±3 to ±40 µm/m. A ±10 µm/m grade gives ±12 µm over 1.2 m, inside the allowance, and tolerates dust and oil.

Resolution and speed: the ALS21 head reads LMSA at up to 1 µm resolution, fifty times finer than the accuracy target, and its datasheet speed of 10 m/s is well above 2 m/s. Check the gap of 0.1 to 0.6 mm can be held along the frame.

Choice: Bogen LMSA in a ±10 µm/m grade with an ALS21 head on BiSS-C, with the drive confirmed to commutate from absolute BiSS-C position, the tape fixed at one datum near the centre of travel, and the thermal term reviewed: if the machine sees more than ±2 °C, add temperature compensation in the controller.

Products in our range to consider

• Bogen LMSA: absolute two-track magnetic scale, accuracy ±3 to ±40 µm/m by grade, up to 32.7 m, no homing at power-up.
• Bogen ALS21: absolute linear magnetic read head for LMSA scales, 1 µm resolution, SSI or BiSS-C, IP67.
• Bogen LMSI: incremental magnetic tape with configurable pole pitch and up to nine tracks, on reels to 25 or 50 m.
• Bogen LMSB: incremental magnetic scale bar up to 2,000 mm, accuracy grades ±3, ±10 or ±20 µm/m.
• Bogen LMSN: Nonius-coded absolute magnetic scale for custom OEM absolute systems.
• Bogen IKS23: rugged incremental magnetic head with RS422 or HTL output, 0.5 to 50 µm resolution.
• Bogen xMARS: miniature 16 × 10 × 3 mm absolute head for ring and tape scales.
• Kappasense FREEe: capacitive absolute linear encoder, ±5 µm accuracy, ±1 µm repeatability, passive 3 g read head.
• Kappasense FLEXe: robust capacitive linear encoder for vibrating, dirty or hot environments, ±20 µm/m, strokes to 1.9 m.

What else the build needs

• Drive that reads the encoder protocol and, on linear motors, commutates from it. See servo drives and how to select a servo drive.
• Linear motor or actuator for the axis. See how to select a linear motor and linear actuators.
• Guides and structure straight, flat and stiff enough for the ride height and Abbe budget.
• Cable carrier and flex-rated cable for the moving read head.
• Controller with temperature compensation if needed. See motion controllers.
• Rotary feedback on other axes. See encoders.

Things that catch engineers out

• Resolution mistaken for accuracy.
• Thermal growth ignored on long axes, especially aluminium frames with steel scales.
• Scale mounted far from the tool, creating Abbe error.
• Mounting surface not flat or straight enough for the ride height tolerance.
• Exposed optical in “mostly clean” machines that see coolant or oil mist.
• Ferrous swarf and magnets near magnetic tape.
• Speed limit at full resolution not checked.
• Drive cannot commutate from the absolute protocol chosen.
• Read head cable not rated for continuous flex.

Checklist

Send these to our engineers, or enter them in the spec builder:

• Measuring length and total scale length
• Accuracy, repeatability and resolution required, and the temperature range
• Absolute or incremental; need to commutate a linear motor at power-up
• Maximum speed and acceleration
• Contamination: dust, chips, coolant, oil mist, water
• Ride height and alignment your carriage can hold; mounting surface material
• Abbe offset between scale and working point
• Interface and drive model
• Number of read heads on one scale
• Cable length and flex requirements
• Temperature, shock, vibration and nearby magnetic fields
• Quantities and timing

For the wider family, see our encoders overview.