Motion Fundamentals

1.1Why Motion Control Matters

Optical systems rarely function in isolation. A lens must be positioned at the correct focal distance. A fiber end must be aligned to a waveguide with sub-micrometer precision. A sample must be scanned beneath a microscope objective at constant velocity. In every case, the quality of the optical result depends directly on the quality of the mechanical positioning that supports it.

Precision motion control is the engineering discipline concerned with constraining, actuating, and measuring mechanical displacement with quantified accuracy. In photonics, the demands on motion systems are unusually severe: positioning tolerances often fall below one micrometer, angular errors of a few arc-seconds can measurably degrade system performance, and thermal drift of a few tenths of a degree can shift an alignment beyond acceptable limits [1, 2].

This topic establishes the foundational vocabulary, physics, and specification framework for precision positioning. It covers the parameters that define how well a stage performs — travel, resolution, accuracy, repeatability — and, critically, explains why those parameters are not consistently defined across the industry. Subsequent topics in this category build directly on this foundation: Manual Stages covers hand-driven positioning; Motorized Positioning covers automated systems and their control; and additional topics address optic mounts, optomechanical hardware, and multi-axis systems.

1.2Scope

The stage types addressed here span the range of photonics positioning hardware:

Translation stages provide straight-line motion along one axis (X, Y, or Z). They range from simple manual dovetail slides to air-bearing stages with nanometer-class trajectory quality.

Rotation stages provide angular motion about a single axis, typically the vertical axis (θZ). Precision variants achieve sub-arc-second resolution and are used for polarimetry, diffractometry, and wafer inspection.

Goniometers (goniometric cradles) provide angular motion about a horizontal axis, tilting a payload about a fixed point in space above the stage surface. Stacking two goniometers at 90° creates a tip/tilt system with a common center of rotation — an arrangement widely used in crystallography and beam alignment.

Hexapods are parallel-kinematic platforms that provide simultaneous motion in all six degrees of freedom (three translational, three rotational) using six actuators working in parallel. They offer higher stiffness and better multi-axis path accuracy than stacked serial-kinematic stages, and are widely used in fiber alignment, telescope mirror positioning, and semiconductor inspection [2, 4].

1.3The Specification Problem

A researcher comparing stages from different suppliers will encounter a persistent difficulty: the same physical quantity is described with different terms, measured under different conditions, and reported in different units. One supplier's "resolution" is another's "minimum incremental motion." One datasheet reports unidirectional repeatability; another reports bidirectional — without always labeling the distinction. Accuracy values may assume factory calibration that requires a specific controller, or may reflect uncalibrated performance that is five to ten times worse.

This inconsistency is not accidental. The precision motion industry lacks a single universally adopted specification standard. ASME B5.57, ISO 230-2, and JIS B 6190 all define positioning performance metrics, but suppliers choose which standard to follow — or whether to follow any formal standard at all. Test conditions (temperature, load, measurement height above the stage surface, number of measurement points) vary and are not always disclosed.

The consequence for the engineer is that datasheet comparison is not straightforward. A stage with a "±0.2 µm accuracy" specification and a stage with "±5 µm accuracy" may not differ by 25× in real-world performance — the gap may be partly or largely explained by differences in how the measurement was made. This topic devotes Section 9 to a systematic treatment of this problem, including a cross-industry terminology reference table.

2.1Six Degrees of Freedom

A rigid body in three-dimensional space possesses six independent degrees of freedom (DOF): three translational and three rotational. In motion control, these are conventionally labeled [1, 2]:

A translation stage is designed to provide controlled motion along one translational DOF while constraining the remaining five. The quality of a stage is defined not only by how well it controls the intended DOF, but by how effectively it suppresses parasitic motion in the five constrained DOFs. A stage with excellent on-axis accuracy but poor yaw control may produce larger total positioning errors at the point of interest than a less accurate stage with tighter angular error control. See Manual Stages for specific bearing types, actuator mechanisms, and stage selection.

2.2Coordinate Conventions

The right-hand coordinate system is standard in precision engineering. The X axis points along the direction of travel, Y is transverse (horizontal), and Z is vertical. Rotations follow the right-hand rule: curl the fingers of the right hand around the axis in the positive rotation direction, and the thumb points along the positive axis direction [1].

In multi-axis assemblies, the distinction between stage-local and global (world) coordinate systems becomes important. Each individual stage defines its own local coordinate frame, with its travel axis aligned to local X. When stages are stacked — for example, an X stage carrying a Y stage carrying a rotation stage — the global coordinate frame remains fixed to the base, and the relationship between commanded local motion and resulting global displacement depends on the orientations and offsets of all stages in the stack. Hexapods avoid this complexity by operating in a single user-defined coordinate frame, with the controller performing the inverse kinematics to translate Cartesian commands into individual actuator motions [2, 4].

2.3Serial vs. Parallel Kinematics

Stacking individual single-axis stages to build a multi-DOF system is called serial kinematics. The bottom stage carries the full weight of all stages above it plus the payload. Each successive stage adds its own errors to the stack, and cable management becomes progressively more difficult as upper stages move relative to the base.

In a parallel-kinematic system — the hexapod being the primary example — all actuators connect in parallel between a fixed base and a moving platform. The moved mass is lower (just the platform and payload, not the accumulated mass of intermediate stages), stiffness is higher for a given footprint, and path accuracy in multi-axis motion is generally superior. The tradeoff is a more complex kinematic relationship between actuator displacements and platform position, which requires a controller capable of real-time coordinate transformation [2, 4]. For detailed coverage of hexapod architecture, inverse kinematics, virtual pivot points, and workspace, see the Motorized Positioning guide, Section 7.

3.1Travel Range

Travel range is the total distance a stage can move along its intended axis, expressed in millimeters or inches. It is the most straightforward specification on any datasheet — but even here, details matter. Some suppliers specify mechanical travel (the physical limit of the bearing system), while others specify usable travel (the range over which performance specifications are guaranteed, which may be shorter due to end-of-travel effects on bearing preload or encoder quality) [1].

For stages with limit switches, the switch activation point defines the usable travel, and rubber bumper hard stops provide a few millimeters of overtravel for emergency braking.

3.2Resolution vs. Minimum Incremental Motion

This is the single most important distinction in motion control specifications, and it is widely misunderstood.

Resolution, as commonly published in datasheets, reflects the smallest position increment that the feedback system can detect — typically the encoder count size after interpolation. A stage with a linear encoder interpolated to 1 nm count size will be listed with "1 nm resolution." This number describes the sensor, not the stage. It does not mean the stage can execute a 1 nm step.

Minimum incremental motion (MIM) is the smallest actual motion the stage can reliably produce. It is limited by friction (particularly stiction in the bearing system), drive train compliance, controller servo bandwidth, and noise. The MIM of the same stage with 1 nm encoder resolution may be 20 nm, 50 nm, or even 1 µm — one to three orders of magnitude larger than the encoder resolution [1, 2].

Resolution as commonly published reflects encoder count size, not achievable motion, and can overstate real-world positioning capability by an order of magnitude or more. Engineers designing systems around positioning precision must use MIM, not resolution, as the operative specification. Unfortunately, MIM is harder to measure and less flattering to publish, so not all suppliers disclose it. When a datasheet lists only "resolution" without a separate MIM specification, treat the number with skepticism and request MIM data from the supplier [1].

Some suppliers have adopted more descriptive terminology: "minimum repeatable incremental movement" (Thorlabs), "practical resolution" (Aerotech), or define MIM as the smallest closed-loop step for which the standard deviation is less than 25% of the mean step size (SmarAct). These definitions all point to the same physical quantity — the smallest motion the stage can actually execute with useful repeatability.

Where: $d$ = theoretical step size (mm), $p$ = screw pitch (mm/rev), $N$ = motor steps per revolution, $M$ = microstepping ratio (dimensionless). This equation gives the theoretical encoder-limited step size, not the MIM. Real MIM is always larger due to friction, stiction, and compliance in the drive train.

3.3Accuracy

Accuracy is the difference between the commanded position and the actual position achieved, measured across the full travel range. It is typically expressed as a ± value in micrometers (e.g., ±2 µm) and represents the worst-case deviation observed during a measurement sweep [1, 6, 7].

The accuracy of a stage depends on the combined contributions of the feedback sensor (encoder linearity, interpolation errors), drive mechanism (screw pitch errors, periodic error), and bearing system (straightness, thermal effects). Stages with rotary encoders mounted on the motor shaft measure screw rotation, not carriage position — any error in the screw (pitch variation, thermal expansion, backlash) is invisible to the encoder. Stages with linear encoders mounted on the carriage measure position directly, and can correct for screw errors in closed-loop operation [1, 2].

A critical detail often buried in datasheet footnotes: accuracy may be specified as calibrated or uncalibrated. Calibration involves measuring the actual position error at multiple points across the travel using a laser interferometer, storing the error map, and applying corrections in the controller. Calibrated accuracy can be 5–10× better than uncalibrated. For example, a stage may specify ±1 µm accuracy calibrated, but ±8 µm uncalibrated. Calibration is only valid with the specific controller used during the calibration procedure — a stage calibrated with one manufacturer's controller will not achieve its calibrated accuracy specification when driven by a third-party controller [6, 7].

Where: $r$ = encoder resolution (length), $p_e$ = encoder grating pitch (length), $I$ = electronic interpolation factor (dimensionless), 4 = quadrature multiplication factor.

3.4Repeatability

Repeatability is the range of positions attained when a stage is commanded to the same target position multiple times. It quantifies the stage's ability to return to a location, independent of whether that location is the correct one (that is accuracy's job) [1, 6].

Unidirectional repeatability measures scatter when the target is always approached from the same direction. This eliminates the effects of backlash and hysteresis, and produces the most favorable number.

Bidirectional repeatability measures scatter when the target is approached from both directions. It includes backlash and hysteresis effects, and is the more demanding — and more practically relevant — specification. A stage used in a scanning application that reverses direction at each end of the scan cares about bidirectional repeatability. A stage used only for one-directional indexing may be adequately characterized by unidirectional repeatability [6, 7].

The distinction matters enormously: a stage with ±0.5 µm unidirectional repeatability may exhibit ±3 µm bidirectional repeatability. Datasheets that list "repeatability" without specifying unidirectional or bidirectional should be treated as unidirectional until confirmed otherwise — suppliers naturally publish the more favorable number.

For a detailed analysis of why encoder resolution overstates system MIM in motorized stages — and how different encoder and drive combinations affect the ratio — see the Motorized Positioning guide, Section 4.5.

3.5Backlash, Hysteresis, and Lost Motion

Backlash is mechanical looseness in the drive train that causes lost motion when the direction of travel reverses. In a lead screw system, backlash arises from clearance between the screw threads and the nut. Preloaded ball screws and anti-backlash split nuts reduce backlash to near zero at the cost of increased friction and reduced efficiency [1, 3].

Hysteresis is non-repeatability on reversal of input that persists even in zero-backlash systems. It arises from elastic deformation in the drive train, bearing compliance, and — in piezoelectric systems — from the inherent hysteresis of the piezoelectric material itself. Hysteresis is load-dependent and can increase significantly under heavy or cantilevered payloads [1, 2].

Lost motion is a term used by some suppliers to describe the total dead zone on direction reversal, combining backlash and hysteresis. It is measured as the deviation between the stop position after a forward move and the stop position after a backward move to the same target.

These three effects are the primary contributors to the gap between unidirectional and bidirectional repeatability. In high-precision applications, they are managed through closed-loop control with direct-measuring linear encoders, which detect and correct for lost motion regardless of its source.

4.1Parasitic Motion

A perfect translation stage would move in a perfectly straight line with zero angular deviation. Real stages deviate from this ideal in all five constrained degrees of freedom as the carriage traverses the travel range. These parasitic motions — often called geometric errors or error motions — determine the trajectory quality of the stage and can dominate the total positioning error at the point of interest [1, 2].

The six geometric error components for a single-axis translation stage are:

Some suppliers report straightness and flatness as separate specifications, each in micrometers over the full travel range. Others combine them into a single "straight-line accuracy" or "runout" specification defined as the radius of the smallest cylinder containing the path of a point on the carriage. These are not directly comparable: a stage with ±0.5 µm straightness and ±0.5 µm flatness does not necessarily have 0.5 µm runout — the runout depends on the geometric combination of the two components [1].

Angular errors (pitch, yaw, roll) are specified in microradians (µrad) or arc-seconds. Their impact on positioning accuracy at the point of interest depends on the distance between the measurement point and the stage surface — this is the Abbe error, treated in Section 4.2.

4.2Abbe Error

Abbe error is the positioning error introduced when the point of interest is offset from the axis of motion. If a stage has an angular error $\theta$ (pitch or yaw) and the measurement point is a distance $d$ from the travel axis, the resulting lateral displacement at the measurement point is [1, 2, 4]:

Where: $\varepsilon_{\text{Abbe}}$ = lateral positioning error (µm), $d$ = offset distance from the travel axis (mm), $\theta$ = angular error in radians (pitch or yaw). The small-angle approximation (tan θ ≈ θ) is valid to within 0.01% for angles below 50 mrad (approximately 2.9°), which covers virtually all precision stage angular errors [1].

Abbe error is significant because optical components are rarely measured at the stage surface. A sample on a rotation stage may be 50 mm above the bearing plane. A fiber tip in an alignment fixture may be 30 mm above the stage carriage. At these offsets, even small angular errors produce micrometer-scale positioning errors that can exceed the stage's on-axis accuracy specification.

For a detailed discussion of how rotary vs. linear encoder placement affects Abbe error in motorized systems, see the Motorized Positioning guide, Section 4.6.

4.3Cosine Error

Cosine error arises when the axis of measurement is not perfectly parallel to the axis of stage travel. If the measurement axis is tilted by an angle $\alpha$ relative to the true travel direction, the measured displacement is shorter than the actual displacement [1, 2]:

Where: $\varepsilon_{\cos}$ = positioning error (same units as L), $L$ = travel distance, $\alpha$ = misalignment angle between travel axis and measurement axis.

4.4Thermal Errors

Thermal expansion is frequently the largest single error source in precision positioning systems, particularly in laboratory environments without active temperature control. A 100 mm aluminum stage that experiences a 1 °C temperature change expands by 2.3 µm — comparable to or larger than the on-axis accuracy of many mid-range stages [1, 3].

Bi-material thermal effects are particularly insidious. A stage with an aluminum body and steel bearing rails expands asymmetrically when the temperature changes, producing bowing or warping that manifests as pitch, yaw, or flatness errors that are not present at the calibration temperature. Several suppliers use FEM-optimized extruded aluminum body designs specifically to minimize this bi-material bending effect [1].

Test standards specify the temperature at which measurements are taken — ISO 230-2 uses 20 ±0.5 °C, PI specifies 22 ±3 °C, and Aerotech requires ±0.25 °C environmental stability for long-term specification validity. Accuracy claims are only valid at the stated test temperature.

5.1Angular Travel and Resolution

Rotation stages provide angular displacement about a single axis, most commonly the vertical axis (θZ). Travel may be continuous (360° or unlimited) for stages with slip rings or wireless data transfer, or limited to a defined angular range for stages with cable wraps or hard stops.

Angular resolution follows the same resolution-vs.-MIM distinction as translation stages. A rotation stage with an encoder resolution of 0.001° (3.6 arc-seconds) may have an angular MIM of 0.005° (18 arc-seconds) due to friction in the worm gear or bearing system. The same caution applies: encoder resolution is not achievable angular step size [1].

For worm-gear-driven rotation stages, the theoretical angular step size is:

Where: $\theta_{\text{step}}$ = angular step size (degrees or radians), $\theta_{\text{motor}}$ = motor step angle (degrees; typically 1.8° for a 200-step motor), $M$ = microstepping ratio, $G$ = worm gear ratio (e.g., 360:1 for a single-start worm with 360-tooth wheel).

5.2Wobble

Wobble is the tilt of the rotation axis as the stage rotates, producing a conical motion of the rotation axis rather than pure rotation about a fixed line. It is measured as the peak-to-peak angular deviation of the surface normal of the rotating platform over a full revolution, expressed in arc-seconds or microradians [1, 2].

Wobble creates a lateral displacement at any point above the stage surface. At a distance R from the rotation center, wobble of $\theta_w$ produces a lateral displacement:

Where: $\delta$ = lateral displacement (same units as R), $R$ = distance from the rotation center to the point of interest, $\theta_w$ = wobble (half-angle, in radians); 1 arc-second = 4.848 × 10⁻⁶ rad.

5.3Eccentricity and Concentricity

Eccentricity (also called concentricity or radial runout, depending on supplier) is the lateral displacement of the actual rotation center from the ideal rotation center during rotation. It produces a circular orbit of the rotation center with each revolution, displacing any centered payload by the eccentricity amount [1, 2].

Eccentricity is measured in micrometers and is a property of the bearing quality and assembly precision. Unlike wobble, it does not scale with distance from the stage surface — the displacement is the same at any height above the stage. High-precision rotation stages achieve eccentricity below 1 µm; general-purpose stages may have 5–25 µm.

5.4Rotary-Specific Accuracy and Repeatability

Angular accuracy and repeatability follow the same definitions as their linear counterparts, expressed in arc-seconds, arc-minutes, or degrees rather than micrometers. The same unidirectional vs. bidirectional distinction applies, and the same caution about which is being reported.

For worm-gear-driven stages, backlash in the gear mesh contributes significantly to the gap between unidirectional and bidirectional repeatability. Direct-drive rotation stages (torque motors) eliminate gear backlash entirely, achieving bidirectional repeatability limited only by encoder resolution and bearing runout.

6.1Goniometers

A goniometer (goniometric cradle) rotates a payload about a horizontal axis located at a fixed point in space above the stage mounting surface. This fixed point — the center of rotation — is the defining geometric parameter of a goniometer [1, 2].

Center of rotation height is the distance from the stage base to the center of rotation, specified in millimeters. When two goniometers of different sizes are stacked at 90° to create a tip/tilt (Euler cradle) system, they are designed so that both axes of rotation pass through the same point. This ensures that the payload tilts about a single point rather than describing a complex arc. The center of rotation height is therefore not just a dimensional parameter — it is a compatibility constraint for multi-axis assembly.

Center of rotation wander is the maximum deviation of the actual rotation center from its ideal position as the goniometer traverses its angular range. It is specified in micrometers and determines how well the stage maintains a true rotation (as opposed to rotation combined with parasitic translation).

Goniometers typically have limited angular range (±5° to ±45°) compared to full rotation stages. They are driven by precision worm gears or ball screws acting on a lever arm, and the angular resolution depends on the screw pitch, gear ratio, and motor step size.

6.2Hexapods and Parallel Kinematics

Hexapods are parallel-kinematic positioning systems based on the Stewart platform principle: six length-adjustable actuators connect a fixed base plate to a moving platform through precision joints (typically cardanic joints with ball bearings for zero backlash). By coordinating the extension of all six actuators, the platform can be positioned in all six degrees of freedom simultaneously [2, 4].

Travel ranges are interdependent. The maximum travel in any single axis (e.g., ±50 mm in X) is only achievable when all other axes are at their zero position. If the platform is simultaneously displaced in Y and rotated in θZ, the available X travel is reduced. Datasheets report single-axis maximum travel, and users must verify that their required multi-axis motion envelope is achievable — hexapod suppliers provide simulation software for this purpose [4].

Programmable pivot point. Unlike a rotation stage with a fixed axis, a hexapod's center of rotation can be defined in software and changed on the fly. This allows the user to rotate a payload about any arbitrary point — the center of a lens, the tip of a fiber, or the surface of a wafer — without physically repositioning the stage. The controller performs the inverse kinematics to translate the desired Cartesian motion about the chosen pivot into the required actuator displacements [4].

Actuator resolution vs. platform resolution. Because the relationship between actuator motion and platform motion is nonlinear and depends on the current platform position, the positioning resolution at the platform level varies across the workspace. Suppliers specify both the individual actuator resolution and the resulting platform MIM in each axis.

The full treatment of hexapod selection, multi-axis assembly design, and parallel vs. serial kinematic tradeoffs is covered in the Hybridized Positioning guide.

7.1Drive Mechanisms

The drive mechanism converts motor torque or electrical energy into controlled linear or angular displacement. The choice of drive determines the achievable resolution, speed, force, backlash, and self-locking behavior of the stage [1, 2, 3].

Manual actuators — micrometers, fine-adjustment screws, and differential micrometers — are the simplest and lowest-cost positioning devices. A standard micrometer with 0.5 mm pitch provides 10 µm per graduation, with Vernier scales extending display resolution to 1 µm. Differential micrometers achieve 0.1 µm display resolution by using two opposing threads of slightly different pitch. Manual actuators are inherently self-locking and free of electrical noise, making them suitable for set-and-hold alignment tasks where automation is not required [1, 3].

Lead screws convert rotary motor motion into linear travel through a threaded shaft and nut. The thread is a sliding contact, producing relatively high friction (efficiency 20–40%) but also inherent self-locking — the stage holds position with no power applied. Lead screws are compact, inexpensive, and available in fine pitches (down to 0.25 mm/rev) for high mechanical reduction. Their drawbacks are limited speed, wear over time, and stiction at motion onset that limits MIM [1, 3].

Ball screws replace the sliding contact with recirculating ball bearings between the screw and nut, reducing friction dramatically (efficiency 85–95%). This enables higher speeds, longer life, and smoother motion. Preloaded ball screws eliminate backlash by applying an axial preload to the ball nut, removing clearance at the cost of slightly increased friction. The tradeoff is that ball screws are generally not self-locking — the stage will backdrive under gravity or external force unless the motor or a brake holds position [1, 3].

Where: $F$ = axial force (N), $\eta$ = screw efficiency (dimensionless; ~0.90 for ball screw), $T$ = motor torque (N·m), $p$ = screw pitch (m/rev).

Where: $\eta$ = mechanical efficiency, $\lambda$ = lead angle, $\varphi$ = friction angle (arctan of coefficient of friction). Self-locking condition: A screw is self-locking when the lead angle λ is less than the friction angle φ. Lead screws with Acme or trapezoidal threads (λ typically 2–5°, φ typically 8–15°) are self-locking. Ball screws (φ < 1° due to rolling contact) are almost never self-locking [1, 3].

Worm gears are the standard for rotation stages and goniometers. A single-start worm meshing with a 360-tooth wheel provides a 360:1 reduction ratio, converting motor rotation to very fine angular increments. Worm gears are inherently non-backdrivable (self-locking) due to the low lead angle and sliding contact, which is advantageous for holding angular position without power. The disadvantages are friction, wear, and backlash in the gear mesh that limits bidirectional repeatability [1, 3].

Direct-drive motors couple the motor directly to the load with no intermediate transmission — no screw, no gearbox, no belt. Eliminating mechanical coupling removes backlash, compliance, and wear, enabling the highest achievable accuracy, speed, and acceleration [1, 2].

For translation stages, the drive consists of a stationary magnet track and a moving coil assembly (or vice versa) that generates force along the travel axis. Two variants exist:

Ironless (moving coil): Zero cogging force, extremely smooth motion, lower force output. Preferred for scanning, metrology, and applications requiring constant-velocity motion.

Iron-core: Higher force per unit size due to the iron concentrating the magnetic flux, but with periodic cogging forces that must be compensated in the controller. Preferred for high-force, high-acceleration applications.

For rotation stages, direct-drive torque motors integrate the rotor directly into the stage platform, providing backlash-free rotation with sub-arc-second repeatability.

The fundamental tradeoff: direct-drive systems are not self-locking. The motor must be actively powered to hold position, and the stage will move freely if power is lost. For vertical or gravity-loaded applications, a counterbalance or brake mechanism is required [1, 2].

Piezo-based drives use the expansion and contraction of piezoelectric ceramic elements to generate motion. Multiple drive principles exist, each with distinct performance characteristics:

All piezo drive types are self-locking (the stage holds position with no power due to friction coupling), generate no magnetic fields (important near electron microscopes and sensitive detectors), and are vacuum-compatible. Their limitations are lower speed and force compared to electromagnetic drives, and — for stick-slip types — audible noise during rapid moves [2].

7.2Bearing Systems

The bearing system constrains the stage to its intended degree of freedom while supporting the payload. The choice of bearing determines the stiffness, friction, motion quality, load capacity, lifetime, and cost of the stage [1, 2].

8.1Encoder Types

Position feedback is the mechanism by which the controller knows where the stage is. The encoder type fundamentally determines what errors the controller can detect and correct [1, 2].

Rotary encoders mount on the motor shaft or the screw end and measure the angular position of the drive train. They are compact and inexpensive. However, they measure motor position, not carriage position — any error between the motor and the carriage (screw pitch errors, thermal expansion of the screw, backlash, compliance) is invisible to the controller. Rotary encoders are adequate for applications where the accuracy requirements are relaxed relative to the screw quality.

Linear encoders mount on the stage body and read a scale attached to the carriage (or vice versa), measuring carriage position directly. Errors in the screw, backlash, and thermal expansion of the drive train are all visible to the encoder and can be corrected in closed loop. Linear encoders are essential for high-accuracy applications and are the only way to achieve sub-micrometer positioning accuracy in screw-driven stages.

The practical impact is significant. A stage with a rotary encoder may specify ±5 µm accuracy. The same mechanical stage with a linear encoder and closed-loop control may achieve ±0.5 µm — a 10× improvement from the encoder alone.

8.2The Encoder-Accuracy Chain

Adding a better encoder does not always improve the stage accuracy. The encoder measures what it is designed to measure — and the relevance of that measurement depends on where the encoder is in the kinematic chain [1].

Consider three configurations:

1. Open-loop stepper motor (no encoder): Position is inferred from the number of commanded steps. Any missed step, stall, or mechanical error is undetected. Accuracy is limited to the screw pitch accuracy and the motor step repeatability.

2. Rotary encoder on motor shaft: The controller can detect and correct motor positioning errors, ensuring the screw turns to the correct angle. But screw pitch errors, thermal expansion, and backlash are uncorrected. Accuracy improves over open-loop but remains limited by screw quality.

3. Linear encoder on carriage: The controller measures the actual carriage position and corrects for all errors between the motor and the carriage. This is the highest-accuracy configuration for screw-driven stages.

For direct-drive stages, there is no screw — the encoder is inherently measuring the load position, and the encoder-accuracy chain is as short as possible.

8.3Limit Switches and Homing

Limit switches define the boundaries of safe travel and prevent the carriage from overrunning the mechanical stops. They are typically optical (non-contact, no wear) or mechanical (contact, simpler but subject to wear).

A home or origin switch provides a repeatable reference position for initializing the coordinate system on power-up. Absolute encoders eliminate the need for a homing routine because they know the position immediately. Incremental encoders require the stage to move to the home switch, establish a reference, and count from there — a process that takes time and may be unacceptable in applications where the stage must not move from its current position on power-up [1].

9.1Why Specifications Are Inconsistent

The precision motion industry serves a wide range of markets — semiconductor fabrication, photonics research, biomedical instrumentation, aerospace — each with different performance priorities and purchasing cultures. No single specification standard has achieved universal adoption. The result is that suppliers choose terminology, test methods, and reporting conventions that best represent their products' strengths [6, 7, 8].

Three formal standards exist for positioning performance:

ASME B5.57 (American): Defines accuracy, repeatability, and reversal error for CNC axes. Uses laser interferometer measurement with specific statistical treatments.

ISO 230-2 (International): Similar in scope to ASME B5.57, with harmonized definitions but slightly different statistical methods for computing accuracy and repeatability from measurement data.

JIS B 6190 (Japanese): The Japanese test code for machine tools, broadly compatible with ISO 230 but with specific measurement protocols.

Some suppliers explicitly state which standard they follow. Many do not. And even among those that cite a standard, the test conditions (temperature, load, measurement height, number of targets, number of approaches per target) may differ from the standard's recommendations or may not be fully disclosed.

9.2What "Guaranteed" vs. "Typical" Means

Datasheets contain two classes of specification values:

Guaranteed specifications are values the supplier commits to for every unit shipped. They are measured and recorded during production testing. If a stage does not meet its guaranteed specification, it is rejected or reworked.

Typical specifications represent the statistical average — most units will meet or exceed this value, but any individual unit may not. "Typical" is by definition a looser commitment. One major supplier notes that their typical values are "two times better than the guaranteed specifications" — meaning the guaranteed specs are the conservative floor, and most stages actually perform better. Another supplier explicitly states that "the designation 'typ.' indicates a statistical average for a property; it does not indicate a guaranteed value for every product supplied."

For system design, use guaranteed values. Typical values are useful for estimating expected performance but should not be used as design constraints.

9.3Test Conditions That Change the Numbers

Several test conditions significantly affect measured specifications:

Temperature: All materials expand with temperature. A stage measured at 20.0 ±0.2 °C will show different accuracy than the same stage measured at 22 ±3 °C — not because the stage changed, but because thermal expansion introduces positioning errors that vary with the temperature stability of the test environment [6, 7].

Load: Specifications are typically measured unloaded. Adding a payload deflects the bearings, changes the preload state, and can alter friction characteristics. A stage with 0.1 µm accuracy unloaded may degrade to 0.5 µm under a 10 kg cantilevered payload.

Measurement height: Abbe error means that the point at which position is measured matters. Some suppliers measure at the stage surface; others measure at a defined height (e.g., 25 mm above the tabletop). The same stage will produce a worse accuracy number when measured at a greater height.

Single-axis vs. multi-axis: Stacking stages in an XY or XYZ configuration degrades the performance of each individual axis due to the accumulated weight, compliance, and Abbe offsets. Multi-axis specifications require factory characterization of the specific stack, and cannot be reliably inferred from single-axis data.

9.4Cross-Industry Specification Terminology Reference

The following table maps common specification concepts to the terms used across industry leaders in precision motion. The terminology is drawn from published datasheets and technical notes of major photonics-grade motion suppliers†.

10.1Step-by-Step Stage Selection

Selecting a positioning stage is a requirements-driven process. The following workflow progresses from application definition to final specification, with each step narrowing the field of candidates [1, 2].

Step 1 — Define the application. What is being positioned? What is the optical or mechanical function? What are the consequences of positioning error? A fiber alignment system with sub-micrometer coupling loss sensitivity has fundamentally different requirements than a sample translation stage for a teaching lab.

Step 2 — Determine degrees of freedom. How many axes of motion are needed? Which are translational and which are rotational? Can the required DOFs be achieved with a single multi-axis stage (hexapod), or is a stacked serial-kinematic assembly more appropriate?

Step 3 — Specify travel range. What is the required travel in each axis? Include margin for alignment and homing. For hexapods, verify that the required multi-axis motion envelope is within the workspace — not just the single-axis travel limits.

Step 4 — Specify MIM (not resolution). What is the smallest positioning step the application requires? Specify this in terms of MIM, not encoder resolution. If the supplier does not publish MIM, request it.

Step 5 — Specify accuracy. What is the maximum acceptable deviation between commanded and actual position? Determine whether calibrated or uncalibrated accuracy is needed, and whether the application requires a specific controller to achieve the calibrated value.

Step 6 — Specify repeatability. What is the acceptable position scatter on repeated moves to the same target? Determine whether unidirectional or bidirectional repeatability is relevant — if the application involves direction reversals (scanning, step-and-repeat), bidirectional repeatability is required.

Step 7 — Evaluate geometric errors and Abbe offset. Calculate the Abbe error contribution at the actual point of interest (not the stage surface). If the point of interest is elevated, pitch and yaw specifications may dominate the total error budget. Consider whether straightness and flatness are critical (scanning applications) or secondary (point-to-point indexing).

Step 8 — Select drive mechanism. Match the drive type to the application requirements using Section 7's framework: lead screw for low-cost manual; ball screw for general motorized; direct-drive for high speed, high accuracy, and zero backlash; piezo for compact, vacuum, and nanopositioning.

Step 9 — Select bearing type. Match the bearing to the load, stiffness, travel, speed, and motion quality requirements using Section 7's bearing comparison table. Key question: is the application stiffness-limited (crossed-roller or air bearing) or travel-limited (recirculating ball)?

Step 10 — Select feedback. Match the encoder to the accuracy requirement: open-loop stepper for coarse indexing; rotary encoder for moderate accuracy; linear encoder for high accuracy. Absolute encoders for applications that cannot tolerate homing routines.

Step 11 — Consider environment. Vacuum compatibility, cleanroom class, magnetic sensitivity, temperature range, and vibration environment all constrain the available options.

Step 12 — Verify specifications under comparable conditions. Before final selection, ensure that competing suppliers' specifications are compared on equal footing: same type (calibrated/uncalibrated, unidirectional/bidirectional), same test standard if possible, and with disclosed test conditions.