Servo System Repeatability Positioning Accuracy Testing: Standards, Methods, and Technical Implement

Servo system repeatability positioning accuracy testing evaluates how consistently a servo-driven motion axis returns to the same commanded position over multiple approach cycles, and it is a decisive acceptance criterion for CNC machine tools, industrial robots, and precision motion platforms. Performed in accordance with international standards such as ISO 230-2, the test quantifies positioning scatter statistically, typically as four times the standard deviation measured with high-resolution instruments like laser interferometers. This article provides a comprehensive technical reference covering test principles, sample types, measurement items, procedures, data evaluation methods, and typical industrial applications.

Technical Overview

A servo system achieves precise motion control through the closed-loop interaction of a servo motor, a drive amplifier, a feedback device such as an optical or magnetic encoder, and a motion controller. Repeatability positioning accuracy, commonly referred to simply as repeatability, describes the closeness of agreement among successive positioning movements toward the same target position when the approach is made under the same conditions, including the same direction of travel, the same feed rate, and the same load state. It is one of the most important dynamic performance indicators of a servo axis because it directly determines process consistency in applications such as machining, assembly, and pick-and-place operations.

Repeatability must be clearly distinguished from absolute positioning accuracy. Accuracy describes how close the achieved position is to the true commanded position, whereas repeatability describes the dispersion of results around the mean achieved position. A servo axis can be highly repeatable while still being offset from the commanded target, for example due to lead-screw pitch error or thermal expansion, and this offset can often be compensated in the controller. Scatter caused by backlash, friction variation, vibration, encoder quantization, and servo loop instability is far more difficult to compensate and is therefore the primary focus of repeatability testing.

Statistically, repeatability is evaluated by commanding the axis to approach a series of target positions for a defined number of cycles, recording the positional deviation at every stop, and processing the data according to ISO 230-2 (Test code for machine tools, Part 2: Determination of accuracy and repeatability of positioning of numerically controlled axes) or its equivalent national standard GB/T 17421.2. The unidirectional repeatability at a target position is defined as four times the standard deviation of the measured deviations, which corresponds to approximately ±2 standard deviations and covers about 95 percent of results for a normally distributed population. Some industries, notably robotics under ISO 9283, specify repeatability as ±3 standard deviations, so the reported value must always state the evaluation basis.

The main technical factors influencing repeatability include mechanical backlash and lost motion in the drivetrain, friction characteristics of guideways, servo loop gains and tuning quality, encoder resolution and signal quality, rigidity of the mechanical transmission, thermal drift during operation, payload variation, and ambient vibration. A well-designed test isolates the servo system contribution from environmental effects through controlled laboratory conditions and appropriate warm-up procedures.

Testing Samples

Rotary Servo Systems

Rotary servo systems consisting of an AC or DC servo motor, drive, and rotary encoder or resolver are tested for angular positioning repeatability. Typical samples include servo motors with integrated 17-bit to 23-bit absolute encoders, rotary indexing tables, and torque-motor-driven direct-drive rotary stages. Angular repeatability is commonly required at the arc-second level for high-end rotary axes used in machine tool fourth and fifth axes and in optical scanning systems.

Ball Screw Driven Linear Servo Axes

The most common test sample in machine tool and automation contexts is a linear axis driven by a servo motor through a ball screw and linear guideway, with feedback from either a rotary encoder on the motor or a linear scale on the moving table. Samples range from compact actuators with 100 mm stroke to large machining center axes exceeding several meters of travel. Repeatability requirements typically fall between 0.001 mm and 0.01 mm depending on the machine class.

Linear Motor Direct Drive Systems

Iron-core and coreless linear motor axes eliminate mechanical transmission elements and therefore exhibit minimal backlash, making them the preferred architecture for semiconductor equipment, lithography stages, and high-speed inspection machines. These samples use sub-micron linear encoders and are tested to repeatability levels down to tens of nanometers, which places stringent demands on measurement instrumentation and environmental stability.

Multi-Axis Servo Tables and Electric Actuators

XY positioning stages, goniometers, rod-type electric cylinders, and servo presses are also submitted for repeatability verification. For electric actuators, positioning is often measured indirectly through an external reference scale rather than the internal feedback device, so that the true mechanical output position, including compliance and lost motion of the rod and coupling, is evaluated rather than the commanded motor angle.

Servo Joints of Industrial Robots

For articulated robots, repeatability testing is performed on the complete kinematic chain in accordance with ISO 9283, where the end-effector position is measured with a laser tracker or coordinate measuring machine while individual joints execute commanded cycles. This verifies the combined performance of all servo axes, reducers, and the control algorithm under realistic payload and speed conditions.

Testing Items

The repeatability test campaign comprises the following evaluation items, each derived from the raw positioning deviation data collected at every target position:

• Unidirectional repeatability: the dispersion of positioning deviations when every approach to a target is made from the same direction, calculated as four times the standard deviation of the deviations for that direction.
• Bidirectional repeatability: the dispersion when approaches alternate between the two opposite directions, which includes the influence of reversal behavior and is always equal to or worse than the unidirectional value.
• Mean positional deviation: the arithmetic mean of the positioning deviations at each target, evaluated separately for each approach direction, indicating systematic offset.
• Reversal error, also called lost motion or backlash: the difference between the mean positional deviations approached from opposite directions at each target.
• Standard deviation of positioning deviation: the per-target scatter value that feeds directly into the repeatability calculation.
• Axis repeatability: the maximum repeatability value found among all target positions along the tested travel, which is the headline result reported for the axis.
• Optional systematic accuracy items: the range of mean positional deviations and the positioning accuracy of the axis, which are frequently measured simultaneously with repeatability since they share the same dataset.

Testing Methods

Test Conditions and Preparation

Before measurement, the servo system must be fully assembled, leveled, and functionally commissioned, with servo parameters set to their production values because loop gains, feed-forward settings, and positioning filter time constants directly affect the scatter. A warm-up phase, typically consisting of several continuous travel cycles at the test feed rate, is mandatory to bring the drive, screw, and guideways to a thermally stable state. The test environment should be maintained as close as possible to the reference temperature of 20 degrees Celsius, with ambient temperature variation during measurement preferably kept within ±0.5 degrees Celsius and drafts, direct sunlight, and external vibration excluded. The payload condition, either no load or a specified test load, must be documented together with the dwell or settling time allowed at each target before the reading is taken.

Measurement Instrumentation

The reference instrument for linear repeatability testing is the helium-neon or stabilized laser interferometer, which provides traceable length measurement with resolution down to 1 nanometer and measurement uncertainty on the order of fractions of a micrometer over machine-tool travels. A complete system comprises the laser head, beam splitter and retroreflector optics, an environmental compensation unit with air temperature, pressure, and humidity sensors that continuously corrects the laser wavelength, and material temperature sensors attached to the scale, ball screw, and machine structure to correct thermal expansion of the workpiece coordinate system. For rotary axes, an autocollimator with a precision polygon mirror, a calibrated circular grating, or a laser angular interferometer is used. Alternatives for coarser axes include high-accuracy magnetic scales, laser Doppler displacement meters, and for robots, laser trackers or ballbar systems. The instrument uncertainty should normally be less than one quarter, and preferably one tenth, of the repeatability tolerance being verified.

Test Procedure

The measurement program follows the cycle defined in ISO 230-2. At least five target positions are selected and distributed over the useful travel, including positions near both ends of the stroke and, where the axis has known critical zones such as ball-screw anchor points, additional intermediate targets. Starting from a park position, the axis moves through the sequence of targets in one direction for the unidirectional test, or approaches each target alternately from each direction for the bidirectional test. At every stop the axis dwells for the defined settling time, and the interferometer records the actual position. The deviation is computed as the difference between the actual position and the target position. The cycle is repeated at least five times so that each target accumulates at least five measurements per direction, producing a statistically meaningful sample.

Data Processing and Evaluation

For each target position, the deviations are averaged to obtain the directional means and the directional standard deviations. The unidirectional repeatability at that target is four times the standard deviation, and the bidirectional repeatability additionally accounts for the reversal value between the two directions. The reversal error at each target is the difference of the two directional means. The final reported axis repeatability is the worst, meaning the largest, value among all targets, and the report must state the number of cycles, the approach feed rate, the dwell time, the payload, the environmental conditions, and the measurement uncertainty. When results are compared against a specification, the measurement uncertainty should be taken into account, either through a simple guard-band rule or a full decision-rule approach agreed between the parties.

Common Corrective Actions Following Failure

When measured repeatability exceeds the specification, the diagnosis typically proceeds from mechanical to control causes. Frequent findings include excessive backlash in the ball screw nut or support bearings requiring preload adjustment, worn or undertensioned couplings, guideway friction irregularities, insufficient servo loop stiffness corrected through gain tuning and notch filtering, encoder signal contamination from electrical noise, and thermal drift caused by inadequate warm-up or poor screw pre-stretch design. Re-testing after each corrective action verifies effectiveness and is documented as part of the commissioning record.

Applications

Repeatability positioning accuracy testing serves a wide range of industrial and research purposes:

• Acceptance testing of new CNC machine tools and machining centers, where axis repeatability is a contractual guarantee item verified before handover to the end user.
• Periodic verification and condition monitoring of installed machines, allowing degradation of drivetrain preloads and guideways to be tracked over time and maintenance to be scheduled predictively.
• Commissioning and optimization of servo drives and motion controllers, where before-and-after repeatability measurements quantify the benefit of autotuning, feed-forward calibration, and backlash compensation tables.
• Type testing and quality control by servo motor, drive, and linear motor manufacturers, providing datasheet values and production release criteria for feedback devices and complete axes.
• Robot calibration and certification under ISO 9283 for industrial robots deployed in welding, dispensing, and assembly, where pose repeatability defines process capability.
• Semiconductor and electronics manufacturing equipment qualification, where wafer stages, bonders, and inspection systems demand sub-micron verified repeatability as a prerequisite for high-yield production.
• Research and benchmarking in precision engineering laboratories, supporting the development of new drive architectures, encoders, and control algorithms.

FAQ

What is the difference between repeatability and accuracy in a servo system?

Accuracy describes how close the achieved position is to the commanded target, including systematic offsets such as screw pitch error, while repeatability describes only the scatter of results when returning to the same target. A system can be very repeatable but inaccurate, and the systematic component can usually be compensated by the controller, whereas poor repeatability indicates stochastic mechanical or control problems that compensation cannot fully remove.

Which standards govern repeatability positioning accuracy testing?

The primary standard for machine tool axes is ISO 230-2, which is adopted nationally as GB/T 17421.2, JIS B 6192, and ASME B5.54 references similar methods. Industrial robots are evaluated under ISO 9283, and servo drive and motor performance is additionally covered by IEC 61800 series documents for adjustable-speed electrical power drive systems.

Why is repeatability defined as four times the standard deviation?

Assuming the positioning deviations follow a normal distribution, an interval of ±2 standard deviations around the mean contains approximately 95.4 percent of results, so the full width is four standard deviations. This convention provides a statistically robust, manufacturer-neutral figure; users requiring higher confidence, such as ±3 standard deviations covering 99.7 percent, must request the corresponding evaluation basis explicitly.

How many target positions and measurement cycles are required?

ISO 230-2 specifies a minimum of five target positions distributed over the travel and a minimum of five measurement cycles per direction. More targets and cycles increase statistical confidence and are recommended for long-travel axes, for axes with suspected local defects, and for dispute resolution between supplier and customer.

What environmental conditions are necessary for a valid test?

The test should be performed at an ambient temperature near 20 degrees Celsius, with temperature drift during measurement limited, typically to within 0.5 degrees Celsius, and with material temperatures of the screw, scale, and structure recorded for compensation. Air pressure and humidity are measured continuously to correct the laser wavelength, and the machine must be thermally stabilized through a documented warm-up routine.

What repeatability levels can modern servo systems achieve?

Typical CNC machining centers achieve bidirectional axis repeatability between 0.003 mm and 0.01 mm, precision grinding machines reach 0.001 mm to 0.003 mm, direct-drive linear motor stages with sub-micron encoders achieve better than 0.001 mm, and state-of-the-art semiconductor stages demonstrate repeatability in the tens of nanometers. Industrial robots typically specify pose repeatability between 0.02 mm and 0.1 mm depending on payload class.

Can backlash compensation replace a good mechanical repeatability result?

Backlash compensation can correct the mean reversal offset but not the variation of lost motion, which changes with position, load, and wear. Consequently, compensation improves measured accuracy while genuine repeatability still depends on the mechanical quality of the drivetrain, preload settings, and servo tuning, which is why mechanical correction remains the primary path to compliance.