Skip to main content

Multi-Axis Test Positioners: Error Sources and Testing Under Full Load

A multi-axis test positioner holds a test article at a commanded angle while the article's weight loads the bearing, winds up the gear train, and adds inertia to the servo, and an empty-table accuracy figure does not state what happens under that load. UTEC Industrial designs, engineers, machines, fabricates, and installs custom material handling systems for aerospace and heavy industry from its Spokane Valley, WA facility, integrating Allen-Bradley PLC and motion control with in-house CNC machining, heat treating, and stress relief. No neutral published source was found that measures the load-dependent accuracy of a multi-axis test positioner, and this article instead builds the error budget from component sources, with every step that combines them labelled as engineering reasoning. Accuracy under load is decided along the whole chain, design → engineering → parts machining → fabrication → assembly → weld fatigue → stress relief → drives → controls → tuning → monitoring, and confirmed only by a loaded test.

What does "accuracy under full load" mean for a multi-axis test positioner?​

Accuracy, repeatability, and reversal error are three different quantities. ISO 230-2:2014, a machine-tool test code, specifies methods for testing and evaluating the accuracy and repeatability of positioning of numerically controlled machine tool axes by direct measurement of individual axes on the machine. Its abstract says the methods apply equally to linear and rotary axes and involve repeated measurements at each position.

At the encoder level, HEIDENHAIN defines system accuracy as the upper limit of the position deviations within a given position. For its angle encoders with integral bearing, that value is determined through five forward and five reverse measurements during final inspection, and the brochure states that the reversal error is not included; the calibration chart documents mechanical hysteresis separately. As engineering reasoning, an encoder's catalog accuracy is therefore not the bidirectional accuracy of the axis it is mounted on.

Neither source states a payload condition. No neutral published source was found that gives the accuracy or deflection of a multi-axis test positioner carrying its full payload; that is the finding of the research for this article, not a statement made by any source. In engineering terms, each of the three quantities needs a stated load, and a useful definition of accuracy under full load is the bidirectional positioning error measured at the test article's reference point, with the payload mass and CG envelope named (ISO 230-2:2014; HEIDENHAIN brochure ID 591109-26, 2025, Measuring accuracy p. 18 and Calibration chart).

Why doesn't a no-load acceptance test predict accuracy with the test article mounted?​

The abstracts of the machine-tool accuracy codes state no payload condition. ISO 230-1:2012 specifies methods for testing the accuracy of machine tools operating either under no-load or under quasi-static conditions, by means of geometric and machining tests, and its abstract adds that the methods can also be applied to other types of industrial machines. It is not applicable to the operational testing of the machine tool, such as vibrations and stick-slip motion of components, or to checking speeds and feeds.

ISO 230-7:2015 covers axis-of-rotation error motion and speed-induced axis shifts of machine-tool axes of rotation, which it says include spindle units, rotary heads, and rotary and swivelling tables. It sends thermally induced axis shifts to ISO 230-3, and it lists angular positioning accuracy among the properties of spindles it does not cover, referring to ISO 230-1 and ISO 230-2. Neither abstract states a condition with a payload applied.

As engineering reasoning, a full payload adds three things an empty-table geometric test leaves out: a tilting moment on the slewing bearing, a torque that winds up the gear train and changes sign when the CG crosses the axis, and a larger inertia for the servo. The machine-tool codes stay useful as methods a buyer and supplier can agree to borrow, but they are not requirements on positioners. The eccentric-load derating article covers how the same payload sets the bearing's capacity; this article covers what it does to accuracy (ISO 230-1:2012; ISO 230-7:2015).

How does slewing-bearing and mounting stiffness let a loaded table move?​

Bearing makers describe the slewing bearing itself as a low-stiffness part. Kaydon's catalog states that slewing ring bearings, due to the nature of their design, have low structural stiffness, making them highly susceptible to any distortions caused by the surrounding structures, and that such distortions adversely affect performance and life. It says the need for increased stiffness and higher precision surfaces is more critical with increasing loads, increased frequency of operation, decreasing diameters, decreasing bearing cross sections, decreasing internal bearing clearance, and decreasing torque limits. SKF states that slewing bearings have limited rigidity due to their relatively small cross-sectional height compared with their diameter, and that the support structure should therefore be designed for maximum axial and radial rigidity.

Kaydon's Figures 3-1 to 3-3 show the maximum permissible deflections that typical four-point and eight-point ball bearings can withstand while maintaining correct function; the deflection must be gradual, similar to a sinusoidal wave pattern, and not occur in a span less than 90° and not more than once in 180°. For ball bearing designs, the allowable dish is dv ≈ 0.003 × Dw × P, where Dw is the rolling-element diameter and P the radial distance of the mounting structure face, and roller designs get 2/3 of the equivalent four-point ball bearing values. Kaydon adds that these values may need to be reduced for application requirements such as lower rotational resistance or higher precision.

These are limits for correct bearing function, not a statement of how far any positioner deflects. As engineering reasoning, a test positioner is a higher-precision application in Kaydon's sense, and a tilt of the bearing plane under the payload's moment moves the test article by that angle times its distance from the bearing, a motion that an angle encoder on the rotation axis does not measure (Kaydon Catalog 390, 2017, Section 3 Part 1 §1, §1.1 and §1.1.1, p. 25; SKF 2019, p. 24).

How do gear backlash and drive-train compliance become positioning error?​

Kaydon states that backlash is required for most geared slewing ring bearing applications, to accommodate manufacturing tolerances of the gears, mounting structures, lubrication, thermal expansion, and deflection of the components under dynamic loading. SKF says practical experience has shown that zero backlash can produce structural overloads that significantly reduce gear life, and its Table 1 gives recommended minimum backlash values for satisfactory meshing under heavy load, measured at the gear's blue marking: 0.25 to 0.375 mm for modules over 3.15 up to 6.3 mm, rising to 1 to 1.5 mm for modules over 20 up to 25 mm.

Backlash shows up as a direction-dependent error. Renishaw's white paper states that any backlash in the rotary drive systems will introduce differences in reported angular position that vary with direction of rotation, and that these can have the most significant impact on system repeatability. HEIDENHAIN's feed-axis paper, written about ball-screw linear axes, says most controls are capable of compensating pitch error and reversal error, but that the reversal error is often unstable over long periods of time and must be regularly recalibrated. It also states that forces deforming the feed drive shift the actual slide position relative to the measured position: a 500 kg slide at a moderate acceleration of 4 m/s² produces deformations of 10 µm to 20 µm that the rotary encoder and ball-screw system cannot recognize.

In UTEC's own arithmetic, with an assumed geometry and the value treated as a circumferential gap at the pitch circle: 0.25 mm of backlash on a 1,000 mm pitch-diameter ring gear is 0.25 ÷ 500 = 0.0005 rad of lost motion, about 103 arc seconds, before any compliance under load is added. That is a geometric conversion of a catalog minimum, not a measured positioner accuracy (Kaydon Catalog 390, 2017, §1.2.2, p. 36; SKF 2019, pp. 28–29, Table 1; Renishaw 2019, p. 3; HEIDENHAIN Technical Information ID 394843-01, 2018, p. 3).

Where should the position encoder sit on a geared test axis?​

HEIDENHAIN compared the two choices on one machine: the C rotary table axis of a high-end 5-axis vertical machining center, driven by a servomotor and worm gear. Feedback from the motor encoder through the gear ratio, which it calls semi-closed loop control, does not account for errors arising in the mechanical transmission elements. With an angle encoder mounted directly on the axis of the rotary table, closed loop control, nearly all of the mechanical transmission errors are taken into account by the position control loop.

Measured per ISO 230-2 at twelve positions 30° apart, five times in each direction, with the axis compensation tables switched off, the closed-loop table stayed within ±1.3″, while semi-closed loop gave ±5″ in each direction and a reversal error of 31″. Twelve-point compensation brought the results to ±0.35″ closed loop and ±1.4″ with a 1.0″ reversal error semi-closed loop, but a repeat semi-closed-loop measurement at 60 points showed ±4.5″ and a reversal error of up to 4.0″. The paper concludes that this compensation is suitable only when the rotary table approaches previously known positions.

The paper says its comparison primarily applies to the rotary tables on machine tools, and its load was cyclic motion and heating, not payload mass; every number belongs to that one test. As engineering reasoning, a test positioner that has to stop at arbitrary commanded angles is the case the paper says table compensation does not cover, and that points to position feedback from an encoder on the loaded axis (HEIDENHAIN Technical Information, 2020, pp. 1–6 and 8).

What does a load-side encoder correct, and what does the encoder add to the error budget?​

HEIDENHAIN's feed-axis paper describes the closed-loop arrangement as one whose position control loop includes the entire mechanical feed-drive system, with transfer errors from the mechanics detected by the linear encoder on the slide and corrected by the controller electronics. It also sets the limit, for cutting force on a machine tool: that force is distributed not only in the feed drive system but over the entire structure of the machine, the deformation of the feed drive system therefore normally has only a small share in the total deformation of the machine, and a linear encoder can recognize and correct only this small portion of the total deformation.

For a test positioner, the following is engineering reasoning. Within the response of the position loop, an angle encoder on the rotation axis can correct backlash, gear-mesh error, and wind-up between the motor and the table. It cannot correct tilt of the bearing plane, bending of the table or of the upper axis's support, or compliance in the fixture, because those move the article without turning the scale, and on a two-axis positioner lower-axis deflection moves the whole upper axis. Those errors are kept small by stiffness or measured at the article itself.

The encoder brings its own terms. HEIDENHAIN lists the eccentricity of the graduation relative to the bearing, the bearing errors, and the method of coupling to the measured shaft among the factors that mainly determine the accuracy of angular measurement, and states that for assessment of the attainable overall accuracy, all of the individual factors must be taken into account. Renishaw gives the bearing-wander term: angular error in arc seconds = bearing wander (µm) × 412.5 ÷ D, with D the scale diameter in mm, and says the radial runout of a spindle on rolling bearings is affected by the design and adjustment of the bearing system, but is typically more than ±1 µm. With an assumed 5 µm of wander on a 500 mm scale, the formula gives 5 × 412.5 ÷ 500 ≈ 4.1″, an illustration of the formula and not a positioner figure. Renishaw also states that error mapping has no impact on the effects of bearing wander, shaft torsion, or other time-dependent error sources (HEIDENHAIN Technical Information ID 394843-01, 2018, pp. 2–3; HEIDENHAIN brochure ID 591109-26, 2025, p. 18; Renishaw 2019, pp. 2–3 and 8).

How do servo bandwidth and structural flexibility limit accuracy under load?​

The system-level sources on this point concern NASA Deep Space Network antennas, heavy two-axis antenna mounts that, as engineering reasoning, are analogous to a large test positioner but are not test positioners. Gawronski's 2006 JPL report evaluates antenna control performance by settling time, bandwidth, steady-state error in rate offsets, and rms servo error while tracking in wind gusts. It defines settling time as the time to reach the commanded value within ±3 percent of the final value, measured from small position offsets, typically 20 mdeg, and bandwidth as the frequency at which the magnitude of the closed-loop transfer function crosses 0.7, or −3 dB. The report says it was observed that the shorter the settling time or the wider the bandwidth, the smaller the servo error in wind gusts.

The report states that for larger, flexible antennas, those with resonance frequencies within the antenna bandwidth, PI controllers become unstable when the proportional gain, and to a lesser extent the integral gain, is too large; the excessive gains excite antenna vibrations. An LQG controller allows a significant increase in the gains without excitation of vibrations. At the component level, Renishaw says it is preferable for the encoder to be as close to the drive as practically possible, which helps to minimise potential shaft resonances that influence servo performance, particularly as servo bandwidths increase.

As engineering reasoning, adding payload mass and inertia to a positioner without adding stiffness lowers its structural resonances, and a gain that was stable on an empty table can excite the loaded structure. Rockwell Automation's Kinetix 5700 manual states that actual bandwidth values depend on the application and can require adjustment once motor and load are connected. The headstock, trunnion, and turntable article covers loaded tuning of positioner servo axes, and hydraulic test axes follow the valve-to-load-resonance guidance in the closed-loop hydraulic control article (Gawronski 2006, Abstract, §§I–II and IV; Renishaw 2019, p. 1; Rockwell Automation 2198-UM002E-EN-P, p. 211).

How does a counter-torque drive hold a geared axis against reversing loads?​

The same DSN antennas show backlash at the gearboxes and at the elevation bull gears, due to a small gap between the gear teeth, and Gawronski and co-authors' 1999 JPL report states that, left uncorrected, backlash deteriorates the antenna pointing precision. The antennas eliminate it with two drives running a specific torque difference, the torque bias or counter-torque: when backlash clearance opens at one drive, the other is still coupled and controls the antenna. For a gearbox of stiffness k and backlash gap b, the report says the bias should be greater than 2kb, large enough to lead the antenna through the gap for the maximal allowable torque load, but small enough that it will not cause excessive local stress, friction, or wear.

The report states that high and steady loads do not need a torque bias, since the backlash is observed for low and reversing axis loads only, and the existing bias profile is largest for low loads and phases out to zero for higher loads. Time-varying loads such as wind gusts can produce high torques that become very low within a short period, causing a backlash gap when the torque-bias dynamics are too slow.

As engineering reasoning, the test-positioner version of a reversing load is gravity. On a horizontal or tilting axis, the payload's gravity torque changes sign each time its CG passes through the vertical plane containing the axis, and a near-balanced test article keeps the drive close to the low-load region where the report says backlash appears; on a vertical-axis turntable, gravity puts almost no torque about the axis at all. Full load at a large, steady CG offset is then not the worst case for lost motion; the crossover is. A two-motor bias, a load-side encoder, or both address it, and the choice belongs in the design (Gawronski et al. 1999, Abstract, §§I–II).

Which test codes can a loaded acceptance test borrow?​

No test code for positioners was found, and the machine-tool codes are the nearest standardized methods. ISO 230-2:2014 says its methods can be used for acceptance tests, among other uses, with uncertainties estimated as described in ISO/TR 230-9:2005, Annex C. ISO 230-3:2020 defines four tests, one of them a test for thermal distortion caused by rotary motion of components. ASME B5.54-2005, for CNC machining centers, provides a series of tests that should be used to perform acceptance testing, or runoff, of new and reconditioned machines, and states that the set of acceptance tests and the specification limits for machine conformance shall be the subject of contractual agreement between the Supplier and the User.

In HEIDENHAIN's 2020 test per ISO 230-3, with the table heated by five cyclic movements at 3,000 °/min between 0° and 180°, the closed-loop axis stayed within 0.5″, while the semi-closed-loop axis changed by up to 8″, with a time constant of about 2 minutes, and its reversal error rose to 3″. The paper equates 8″ to a deviation of 20 µm on a radius of 0.5 m.

As engineering reasoning, a loaded acceptance plan for a test positioner can borrow these methods and add what their abstracts do not state:

  • the payload, or a mass simulator matching its weight, CG offset, and inertia
  • positions approached from both directions, including across the CG crossover
  • a warm-up at the duty the test facility will actually run
  • the measurement taken at the test article's reference point, not only at the table

Those additions are not in the standards' abstracts, and the acceptance limits are a matter for the purchase agreement, as ASME B5.54 says of machining centers (ISO 230-2:2014; ISO 230-3:2020; ASME B5.54-2005; HEIDENHAIN Technical Information, 2020, p. 7).

What do spacecraft test standards ask of a positioner used as test equipment?​

Spacecraft test standards place requirements on the facilities and fixtures of a test, and, as engineering reasoning, a positioner that holds flight hardware during a test falls among them. NASA Goddard's GEVS, which describes itself as providing guidelines for environmental verification programs for GSFC payloads, subsystems and components, with any guidelines intended as institutional requirements captured in GSFC-STD-1000, states that the facilities and fixtures used in conducting tests must be capable of producing and maintaining the test conditions prescribed with the test specimen installed and operating or not operating, as required.

ECSS-E-ST-10-03C Rev.1, which addresses the requirements for performing verification by testing of space segment elements and space segment equipment on ground prior to launch, states that test facilities, tools, and instrumentation shall not prevent fulfilling the test objectives. It also states that test input tolerance bands shall be agreed by the customer and specified in the test specification (TSPE), and that they shall account for the uncertainty budget and confidence level of the measurement instruments and test equipment used to control and monitor the test parameters.

As engineering reasoning, a positioner's error under the test article's load then belongs in that uncertainty budget, characterized before the test rather than inferred from an empty-table figure. The test-facility handling article covers these chambers and antenna-range practice under IEEE Std 149-2021, the IEEE Recommended Practice for Antenna Measurements (GSFC-STD-7000B, §1.12; ECSS-E-ST-10-03C Rev.1, 4.4.1 f and 4.4.2 a–b; IEEE Std 149-2021).

What sensing, drives, and PLC logic hold a test positioner's accuracy under load?​

The controls decide which errors the loop can see. A control package for a loaded test axis can include:

  • Two encoders per axis. As engineering reasoning from the HEIDENHAIN comparison, a motor encoder serves the drive and a load-side angle encoder on the table axis serves as the position reference. Allen-Bradley Kinetix 5700 ERS3 single- and dual-axis inverters support DSL and Hiperface encoder feedback, and the manual gives position-loop, velocity-loop, and torque-loop axis configurations with a current regulator loop; how a load-side encoder is wired into the loop is a configuration to confirm in the drive documentation.
  • A lost-motion monitor. As engineering reasoning, the difference between the motor-encoder and load-encoder angles is the backlash plus wind-up of the drive train, and trending it at each reversal tracks the kind of reversal error HEIDENHAIN says, for ball-screw axes, is often unstable over long periods and must be regularly recalibrated.
  • Counter-torque on two motors. As engineering reasoning, where the axis runs through a CG crossover, a two-drive torque bias of the kind JPL describes for DSN antennas keeps one gear train loaded.
  • Payload recipes. As engineering practice, the PLC stores each test article's mass, CG envelope, and allowed angles, and a load cell or holding-torque check confirms them before motion; the sizing article works that torque check.
  • Tuning with the load on. Kinetix bandwidth values can require adjustment once motor and load are connected; as engineering practice, tuning is repeated with the heaviest and the most flexible payload.
  • Safety functions. The Kinetix 5700 drives feature safe torque-off, but disabling the power transistor output does not provide physical isolation of the electrical output that is required for some applications; the headstock, trunnion, and turntable article covers positioner safety logic and standards.

UTEC Industrial, a Rockwell Automation Recognized System Integrator, builds Allen-Bradley ControlLogix and CompactLogix control and VFD and servo drives into the positioners it fabricates (Rockwell Automation 2198-UM002E-EN-P, p. 16 Table 3, App. D, p. 211 and p. 250; HEIDENHAIN Technical Information ID 394843-01, 2018, p. 3; Gawronski et al. 1999, §II).

Where along the chain from design to monitoring is accuracy under load decided?​

As engineering reasoning, several of the error sources above are fixed before the controls are switched on. Design and engineering choose the bearing-mounting stiffness, the encoder location, and any counter-torque drive.

Machining sets the bearing seat. Kaydon states that bearing mounting surfaces must be machined flat after all welding and stress relief treatment on the structures is complete, that if subsequent welding is necessary, it must be done to avoid distorting the previously machined mounting surface, and that out-of-flatness must be gradual and not occur in a span less than 90° and not more than once in 180°. Fabrication, weld fatigue, and stress relief come before that machining, as covered in stress relief for machine bases and frames.

Assembly sets the backlash. SKF says gear backlash should be checked after positioning the pinion, at the blue marking on the gear, and corrected by adjusting the centres of the gearwheels if it is outside the guideline values. Kaydon says pinions on adjustable centers should be set for proper backlash at the point of minimum backlash on the gear, identified with yellow paint, and checked at both ends to confirm the prescribed alignment.

As engineering reasoning, drives, controls, tuning, and monitoring then work with what the structure delivers; they cannot remove bearing tilt or structural bending that the encoder does not measure. UTEC Industrial stress-relieves welded frames in-house before final machining, machines to tolerances as tight as ±0.001 in, and inspects with NDT and CMM (Kaydon Catalog 390, 2017, §1.1.2.1, p. 30, and §2.3.4, p. 40; SKF 2019, pp. 28–29).

What should a test-positioner specification state about accuracy under load?​

As engineering reasoning drawn from the sources above, a verifiable specification covers:

  • Load basis. The payload mass, CG envelope, and inertia at which each accuracy figure applies, and whether the figure holds across the CG crossover.
  • Quantities. Positioning accuracy, repeatability, and reversal error stated separately, since HEIDENHAIN states the reversal error is not included in its encoder calibration and ISO 230-2 names both the accuracy and the repeatability of positioning in its scope.
  • Measurement point. The encoder, the table, or the test article's reference point.
  • Feedback architecture. Motor encoder only, or a load-side angle encoder on each axis, and any counter-torque drive.
  • Test method. The borrowed procedure, such as ISO 230-2 positions and runs in both directions, an ISO 230-3 style thermal run, and the payload or mass simulator used.
  • Limits and uncertainty. Acceptance limits written into the contract, as ASME B5.54 states for machining centers, and the uncertainty budget that ECSS asks test tolerances to account for.

UTEC Industrial has built custom positioning and rotating fixtures for precision assembly and test for RTX, and performs factory acceptance testing and on-site commissioning, where a loaded accuracy test can be run and witnessed before shipment (ISO 230-2:2014; ASME B5.54-2005; ECSS-E-ST-10-03C Rev.1, 4.4.2 b; HEIDENHAIN brochure ID 591109-26, 2025, Calibration chart).

Related Articles

References​

  • ISO 230-2:2014: Test code for machine tools — Part 2: Determination of accuracy and repeatability of positioning of numerically controlled axes. International Organization for Standardization, 2014.
  • HEIDENHAIN brochure ID 591109-26: Angle Encoders With Integral Bearing. DR. JOHANNES HEIDENHAIN GmbH, 2025.
  • ISO 230-1:2012: Test code for machine tools — Part 1: Geometric accuracy of machines operating under no-load or quasi-static conditions. International Organization for Standardization, 2012.
  • ISO 230-7:2015: Test code for machine tools — Part 7: Geometric accuracy of axes of rotation. International Organization for Standardization, 2015.
  • Kaydon Catalog 390: Slewing Ring Bearings. Kaydon Bearings, 2017.
  • SKF PUB BU/P2 06115/3 EN: Slewing Bearings. SKF Group, 2019.
  • Renishaw plc. White paper: The accuracy of rotary encoders. Renishaw plc, 2019.
  • HEIDENHAIN Technical Information ID 394843-01: Accuracy of Feed Axes. DR. JOHANNES HEIDENHAIN GmbH, 2018.
  • HEIDENHAIN Technical Information: Positioning accuracy of rotary axes: a key factor in 5-axis machining. DR. JOHANNES HEIDENHAIN GmbH, 2020.
  • Gawronski W (2006). "Servo Performance Parameters of the Deep Space Network Antennas." IPN Progress Report, 42-167. NASA Jet Propulsion Laboratory, November 15, 2006.
  • Rockwell Automation 2198-UM002E-EN-P (2018): Kinetix 5700 Servo Drives User Manual. Rockwell Automation, 2018.
  • Gawronski W, Brandt JJ, Ahlstrom HG Jr, Maneri E (1999). "Modifications of the Torque-Bias Profile for Improved Tracking of Beam-Waveguide Antennas." TMO Progress Report, 42-139. NASA Jet Propulsion Laboratory, November 15, 1999.
  • ISO 230-3:2020: Test code for machine tools — Part 3: Determination of thermal effects. International Organization for Standardization, 2020.
  • ASME B5.54-2005 (R2020): Methods for Performance Evaluation of Computer Numerically Controlled Machining Centers. ASME, 2005.
  • GSFC-STD-7000B: General Environmental Verification Standard (GEVS) for GSFC Flight Programs and Projects. NASA Goddard Space Flight Center, 2021.
  • ECSS-E-ST-10-03C Rev.1: Space engineering — Testing. ECSS Secretariat, ESA-ESTEC, 2022.
  • IEEE Std 149-2021: IEEE Recommended Practice for Antenna Measurements. Institute of Electrical and Electronics Engineers, 2021.

Ready to Discuss a Material Handling System?​

UTEC Industrial designs, engineers, machines, fabricates, and installs custom material handling systems for heavy industry, from the stress-relieved structure and drives to the Allen-Bradley PLC controls, tuning, and monitoring that run them, at its Spokane Valley, WA facility. Send UTEC the application, loads, and duty cycle to start a system review.

Request a Quote →

Questions? Call (509) 922-1832 or email sales@utec.co

Last updated on