Diametrically Magnetized NdFeB Rings for Rotary Encoders: Magnetic Angle Error (MAE), Runout Tolerances, and Sensor Magnet Sourcing

Diametrically magnetized NdFeB rings and cylinders provide a rotating magnetic reference for non-contact rotary encoders, brushless DC motor commutation, and robotic joint position sensing. Their magnetization vector lies across the diameter, perpendicular to the rotational axis. A compatible sensor measures field components as the magnet rotates; the magnet, mounting arrangement, and sensor together determine usable accuracy.

The sourcing decision: Select the sensor IC and sensing geometry before assigning magnet tolerances. Specify the field at the sensor's active sensing plane, measure magnetic angle error (MAE) over a full revolution, and qualify the magnet on the intended shaft. An OD tolerance, a surface-Gauss reading, or a high energy-product grade cannot independently establish encoder accuracy.

Reviewed: November 6, 2026. Scope: Global OEM sourcing of custom sintered NdFeB diametrical rings and cylinders for on-axis and off-axis rotary position sensing. This guide covers drawing requirements, error budgeting, magnetic inspection, and supplier qualification. Numerical examples are analytical illustrations, not measured supplier capability, customer results, or guaranteed acceptance limits. Application-specific simulation, sensor qualification, and functional safety assessment remain separate tasks.

A 14-bit encoder has a nominal angular code increment of about 0.022°, while a 16-bit encoder has about 0.0055°, calculated as 360° / 2ⁿ. These increments describe digital resolution. Absolute accuracy also depends on field distortion, sensor errors, assembly position, temperature, noise, and calibration. Adding output bits does not remove a repeatable magnetic error.

When periodic angle distortion or torque ripple appears, inspect both the mechanical assembly and the signal chain. Candidate causes include sensor offset, gain mismatch, magnet eccentricity, tilt, nearby steel, and material variation. A repeatable error spectrum helps organize tests, but does not uniquely identify the defective component.

For a practical starting point, submit your sensor magnet envelope and target air gap with the IC part number, shaft material, and required accuracy. Those inputs make a tolerance discussion more useful than requesting a generic “precision sensor magnet.”

1. Electromagnetic Geometry: Diametrical Vector and Field Distribution

In an ideal uniformly diametrically magnetized cylinder or ring, the bulk magnetization vector lies in the transverse plane. Do not confuse this with axial magnetization through the two flat faces, radial magnetization around a ring wall, or a multipole pattern. The magnetization drawing should show both the direction and its relationship to a mechanical locating feature.

Axial magnetization                  Diametrical magnetization
Vector parallel to shaft axis        Vector across the diameter
Poles on opposing flat faces         Two opposing pole regions
Typical holding applications         Compatible rotary sensing geometries

The Rotating Vector Field

For an ideal on-axis arrangement, matched orthogonal field components can be represented as Bx = B0 cos(θ + φ0) and By = B0 sin(θ + φ0). Here, B0 is the in-plane field amplitude at the active sensor location and φ0 is a fixed mounting offset. Select the permitted field range from the specific IC datasheet, including its noise floor, saturation limit, and temperature conditions.

For full-turn sensing, calculate the angle with the quadrant-aware expression θsensor = atan2(By, Bx), then apply the required zero offset, direction convention, and angular wrapping. The simpler expression atan(By / Bx) is ambiguous by 180° and has a division problem at zero Bx. Some ICs perform the calculation internally; others require host processing.

An ideal plot of By against Bx forms a centered circle. A displaced center suggests an additive offset; unequal amplitudes create an ellipse. These are signal descriptions, not proof that a particular magnet manufacturing process caused the error. The TI multi-axis Hall sensing application report discusses sensing arrangements, calibration, and practical design examples.

On-Axis vs. Off-Axis (Hollow-Shaft) Geometries

1. On-axis, end-of-shaft: A cylinder or ring rotates opposite a sensor near the rotation axis. Two transverse components can support a full-turn angle calculation when the IC and field orientation are compatible. State the distance to the active die, including package and PCB tolerances, rather than only the visible package clearance.
2. Off-axis, hollow-shaft: A sensor sits beside the ring OD or near its face to preserve a central passage. The measured components may have unequal amplitudes and a nonlinear relationship with mechanical angle. Confirm the permitted geometry with the IC vendor, then simulate and calibrate it. A two-pole diametrical ring and a multipole ring require different decoding assumptions; a simple full-turn atan2 model is not interchangeable between them.

On-axis placement is often the easier architecture to characterize when shaft-end space is available. Hollow-shaft designs can be useful where cables or structures must pass through the bore, but their sensitivity must be established for the actual dimensions. Neither architecture carries a universal allowable runout or air-gap specification.

2. Deconstructing Magnetic Angle Error (MAE) and Harmonic Signatures

Define the signed angle error as e(θ) = wrap180(θsensor − θreference), where wrapping returns the difference within −180° to +180°. State whether a constant zero offset has been removed and whether the result is before or after calibration. Report maximum absolute error, peak-to-peak error, repeatability, and test conditions separately. An IC's integral nonlinearity specification may use different correction and reporting conventions from a supplier's “MAE” test.

A full-turn sweep against an independently characterized reference encoder separates repeatable spatial error from random noise. Reference resolution alone is insufficient: document angular accuracy, fixture runout, sampling, rotation direction, speed, and uncertainty. Record the raw component signals when available so that a corrected angle trace does not conceal a field-range problem.

Observed signal feature       Useful checks; these are not unique diagnoses
Once-per-revolution error     Additive field offsets, mounting, surrounding steel
Twice-per-revolution error    Relative channel gain, quadrature, tilt, geometry
Higher spatial harmonics     Field shape, magnet defects, nearby parts, electronics
Speed-dependent error        Sampling delay, filtering, channel timing, vibration

Eccentricity and Radial Runout: Define the Geometry First

An often-used geometric illustration is δθ ≈ e / R radians, or (e / R) × 180 / π degrees. It describes a small tangential displacement e of a point referenced to a nonzero radius R. It is not a general magnetic-field error model, and it cannot predict an on-axis sensor's MAE by assigning an arbitrary sensing radius. Magnetic sensitivity requires the field geometry or measured response.

Worked numerical example: For an illustrative off-axis geometric construction with R = 2.50 mm and tangential displacement e = 0.10 mm, the small-angle estimate is 0.10 / 2.50 × 57.296 = 2.29°. This corresponds to approximately 26 counts at 12-bit resolution. If that mechanical angle error were present in a motor with six pole pairs, its electrical angle equivalent would be 6 × 2.29° = 13.7°. These arithmetic conversions do not establish the MAE or torque loss of a real magnet assembly.

Also distinguish displacement from total indicated runout (TIR). For an ideal circular surface whose indicated variation comes solely from eccentricity, TIR = 2e. Form error, datum choice, and fixture error can invalidate that simple conversion. A 0.10 mm eccentricity and a 0.10 mm TIR are therefore different drawing conditions. Establish a runout limit by sweeping the proposed tolerance in simulation and confirming it on representative assemblies.

Channel Amplitude Mismatch and Ellipticity

A reproducible signal-only model is Bx = cos θ, By = k sin θ, with positive gain ratio k, zero offsets, and exact quadrature. Its maximum absolute angle error is asin(|1 − k| / (1 + k)). The small-mismatch approximation replaces asin(x) with x in radians.

For k = 0.92, the predicted maximum error is asin(0.08 / 1.92) = 2.388°. This example shows why an apparently modest channel amplitude imbalance matters. It does not imply that any particular tilt angle creates an 8% imbalance. Use the measured signals or the complete magnetic model to connect assembly tilt to gain and phase changes.

Higher-Order Harmonics and Calibration Limits

Higher-order error can arise from magnetization variation, geometry, surrounding steel, sensor nonlinearity, and other signal-processing effects. A third- or fourth-order component alone cannot identify sintering density variation. Compare a bare-magnet scan with the mounted assembly and, where possible, exchange the sensor and fixture to isolate causes.

Calibration can compensate repeatable higher-order error when the chosen IC or host lookup table supports sufficient correction points and bandwidth. Its practical limits include memory, end-of-line cycle time, added latency, and stability across temperature and load. It cannot recover a saturated channel or restore the signal margin of an insufficient field. Evaluate calibrated residual error under conditions beyond the calibration sweep.

3. Visualizing Misalignments and the Lissajous Error Trajectory

The schematic below illustrates alignment and signal changes for discussion with a magnet supplier. The curves are conceptual and are not simulated or measured predictions. Use the adjacent signal models to reproduce the analytical examples; use a qualified fixture to establish actual acceptance values.

4. Mechanical & Material Interfaces: Shaft Permeability, Air Gap, and Runout

A free-space magnet scan does not reproduce every feature of an installed actuator. Include the shaft, adhesive layer, sleeve, nearby fasteners, and relevant motor or brake states when qualifying the sensing assembly.

The Shaft Permeability Trap: Ferromagnetic Shunting

A carbon or alloy steel shaft changes the magnetic boundary conditions. Depending on geometry, it can change both the field amplitude and its angular shape at the sensor. Do not assume a fixed 40% to 60% field loss: the result depends on dimensions, the steel's magnetic properties, sensor location, and the operating magnetic state.

Qualification comparison
A. Bare magnet or approved nonmagnetic fixture
B. Magnet on the intended steel shaft
C. Proposed sleeve or alternative mounting design
Compare field range and full-turn angle error at the same sensing location.

A brass, aluminum, or suitably low-permeability stainless sleeve may reduce coupling by adding separation. It does not form a perfect magnetic shield, and no universal sleeve-thickness-to-ring-wall ratio guarantees isolation. Model candidate thicknesses, then compare the assembled field and MAE. Confirm the actual permeability of stainless parts after forming or machining; a material label alone does not guarantee negligible magnetic response.

Air-Gap Sensitivity and Axial Play

Field amplitude changes with distance, but the near field of a finite ring is not generally exponential. Far from a compact magnet, a dipole approximation gives field magnitude proportional to inverse distance cubed; it should not be used as a universal near-gap design equation. For on-axis transverse sensing, the useful components are normally Bx and By, rather than an assumed axial Bz signal.

Map the permitted axial and radial position envelope, including bearing end play, adhesive thickness, thermal expansion, and sensor package offsets. At every corner, confirm that the relevant component amplitudes remain inside the IC's valid range. Amplitude changes can cancel in an ideal angle ratio, but offsets, noise, unequal channel response, and dynamic filtering prevent assuming perfect rejection in the real system.

Thermal Selection and Stray Fields

Request the supplier's temperature-dependent demagnetization curves for the proposed grade and geometry. Reversible loss of field with temperature differs from irreversible demagnetization, and a grade's advertised maximum temperature is not an assembly guarantee. The operating point depends on magnet dimensions, surrounding materials, temperature, and opposing fields.

Evaluate the magnet's actual temperature rather than only ambient temperature. Include brake coils and motor fields in the exposure assessment. Compare field amplitude and angle error at temperature and again after cooling to distinguish reversible behavior from permanent change. Choose higher coercivity where the operating point requires it; choosing the highest available remanence alone can reduce useful margin or overdrive the sensor.

5. Error Budget and Sourcing Specifications

Separate requirements from demonstrated capability. The following allocation is an illustrative planning budget, not a claim that these errors are independent or that a magnet supplier can guarantee them. Suppose the complete encoder requires maximum absolute error no greater than 0.80° before application-specific linearization. Reserving the contributions below yields a conservative arithmetic sum of 0.80°; confirm the assembled result because interactions can invalidate the allocation.

Table 1: Illustrative Encoder Error Allocation and Verification

Do not use root-sum-square combination merely to obtain a smaller total: correlated systematic errors require a justified model. Resolve overlap between allocations during design review. A magnet's magnetic-axis offset can be a removable constant rather than nonlinearity; specify whether an index mark must align with absolute mechanical zero.

Table 2: Material and Process Choices to Discuss in the RFQ

These are candidate families for comparison. Actual remanence, coercivity, temperature limits, and dimensional capability must come from the quoted material datasheet and validated process.

For a sourcing discussion, send the proposed grade and tolerance drawing for review, identifying which values are mandatory and which remain open to design adjustment. This helps suppliers separate material cost, grinding, magnetization fixtures, and inspection cost in their quotations.

6. Quality Assurance & Factory Verification Protocols

1. Magnetic Moment and Vector Alignment

IEC 60404-14:2002 describes open-circuit magnetic dipole moment measurement using withdrawal or rotation methods. A calibrated multi-axis coil fixture can use component measurements to assess a magnet's net moment direction. The standard does not assign an encoder-grade angular tolerance or guarantee the local field shape at the sensing die.

For moment components mx, my, and mz expressed relative to a defined mechanical datum, deviation from intended positive x magnetization is α = acos(mx / sqrt(mx² + my² + mz²)). This is a nonnegative magnitude; specify a limit such as α ≤ αmax, not a signed plus/minus bound. A rotationally symmetric ring needs a defined index feature if its in-plane magnetic direction must have an absolute mechanical reference.

2. Full-Turn Field and Angle Scanning

Agree on a production test that measures the required behavior rather than only a convenient magnetic proxy:

1. Record the fixture datum, shaft material, sensor or probe identity, and distance to its active location.
2. Sweep a complete mechanical revolution against a calibrated angular reference, recording both rotation directions where hysteresis or fixture motion could matter.
3. Capture the necessary field components and compute angle using the agreed geometry and correction state.
4. Report field extrema, maximum absolute MAE, peak-to-peak MAE, and repeatability. Preserve the raw trace for failure analysis.
5. If waveform THD is specified, define the component, normalization, harmonic bandwidth, and sampling. Scalar THD does not uniquely determine vector angle error.
6. State the acceptance rule and measurement uncertainty, then establish lot sampling or 100% inspection according to risk, demonstrated process capability, and test cost.

Coil moment measurements and local Hall scans answer different questions. ASTM A977/A977M-07(2020) addresses magnetic property measurement using hysteresigraphs; it does not certify the angle accuracy of a finished encoder ring. Require an appropriate material report and a separate application-relevant angle test.

3. Chamfer and Coating Control

Control edge damage and dimensional changes after coating. Specify finished OD, ID, thickness, permitted chips, coating buildup on mounting surfaces, and the edge treatment in the released drawing. Select a chamfer that fits the actual wall thickness; a generic 0.20 mm chamfer can remove a significant fraction of a miniature ring's section.

Compare coating options against corrosion exposure, adhesive compatibility, abrasion, and final dimensions. Electroless nickel or Parylene may be candidates, but neither should be assigned a universal thickness-uniformity guarantee without process evidence. Qualify geometry and field after the final manufacturing operations. Ask the supplier to identify available inspection equipment and subcontracted steps rather than assuming every quoted process is performed in-house.

7. Engineering & Procurement Checklist for Diametric Sensor Magnet RFQs

Use the checklist to turn the design intent into comparable quotations. Mark unresolved values explicitly so that a supplier does not silently substitute a standard commercial requirement.

• Magnetization pattern: Show diametrical two-pole orientation and any mechanical index. Keep multipole alternatives on separate controlled drawings.
• Sensor and geometry: Identify the IC, active sensing plane, nominal gap, off-axis position if applicable, and complete position tolerance envelope.
• Mechanical datums: Specify finished mounting dimensions, assembled runout, face orientation, adhesive thickness, and whether TIR or eccentric displacement is being limited.
• Field envelope: State minimum and maximum component amplitudes at defined positions and temperatures, tied to the IC's valid range.
• Angle acceptance: Define zero-offset removal, pre/post-calibration state, maximum absolute error, peak-to-peak error, reference uncertainty, and guard band.
• Material and environment: Identify the proposed grade, magnet temperature, opposing field conditions, coating exposure, and required aging or recovery tests.
• Shaft and sleeve: Document actual shaft material, nearby steel, sleeve geometry if used, and which mounted assembly the acceptance test represents.
• Inspection deliverables: Agree on sampling, full-turn traces, material reports, traceability, and available metrology. Treat 100% scanning or PPAP as requirements to quote and confirm.
• Commercial conditions: State prototype quantity, annual volume, tooling ownership, change-control rules, lead-time assumptions, and the criteria for moving from samples to production.

8. Frequently Asked Questions (FAQ)

Q1: Can digital calibration eliminate the need for tight magnet tolerances?

Calibration can reduce stable, repeatable error, including higher-order components when the IC or host algorithm supports them. Its coverage and added latency are device-specific. It cannot correct saturation, insufficient signal, or changing assembly geometry with a single static correction map. Balance magnet cost against calibration time, then validate residual error over temperature, speed, and load.

Q2: When should an engineer choose an off-axis ring over an on-axis cylinder?

Use the mechanical architecture and the selected IC's supported geometry to decide. An on-axis arrangement often simplifies field interpretation. An off-axis ring preserves a hollow passage, but requires characterization of component imbalance and mechanical-angle mapping. Do not impose a universal 10 µm runout limit: determine the needed tolerance from the actual error budget and assembly tests.

Q3: How does edge chamfering affect sensor-ring accuracy?

Removing material changes the field geometry, and asymmetric edge treatment can alter the waveform. The magnitude depends on ring dimensions and sensing location. Specify a controlled, manufacturable edge treatment and permitted damage, then compare final coated parts at the intended gap. A chamfer dimension by itself cannot establish an angle-error limit.

Q4: Can a diametric ring be mounted directly on a steel motor shaft?

Possibly, if the complete magnetic design supports it. Steel can substantially change field amplitude and shape, so test the intended shaft assembly. A nonmagnetic spacer is one possible design variable, not an automatic requirement or a complete shield. Approve the mounting design using the measured field envelope and angle error under operating conditions.

Q5: What changes when choosing bonded instead of sintered NdFeB?

Sintered material generally offers greater magnetic density, while bonded processes can offer shape and magnetization flexibility. Actual dimensional capability, thermal behavior, corrosion protection, and coercivity depend on the selected material and process. Compare finished-part field data, binder or grade documentation, and lifecycle cost rather than assuming either process is universally better for encoders.

9. Sourcing Support & Next Steps for Precision Sensor Programs

Review our custom NdFeB magnet product families to compare the available shapes and customization options. For related purchasing workflows, browse the NdFeB engineering and sourcing guides. For an encoder inquiry, contact our engineering and procurement team for a drawing review with:

1. A 2D/3D drawing identifying the envelope, mounting datums, and magnetization direction.
2. The sensor IC model, nominal position, permitted field range, and full positional tolerance stack.
3. Shaft and nearby-part materials, the magnet's temperature envelope, and relevant motor or brake field states.
4. Required accuracy before and after calibration, reference method, and inspection deliverables.
5. Sample quantities, annual demand, delivery expectations, and qualification milestones.

The review can identify open requirements and the feasibility work needed before a quotation or sample plan is finalized. Obtain written confirmation of tolerances, metrology coverage, inspection cost, and any process-capability commitment. Supplier approval should follow agreed sample evidence; this article does not establish guaranteed MAE, Cpk, or functional safety performance.

Sources & References

The sources support specific methods, not universal runout limits, grade suitability, sleeve dimensions, or supplier performance. Publisher records identify the standards; full procedures may require access to the standards themselves.

• Texas Instruments — Angle Measurement With Multi-Axis Hall-Effect Sensors, SBAA463A: Multi-axis Hall sensing arrangements, calibration, and practical on-axis/off-axis design discussion. Sensor-specific operating limits still require the selected IC's datasheet.
• IEC — IEC 60404-14:2002: Open-circuit magnetic dipole moment measurement by withdrawal or rotation. Used here to distinguish net moment measurement from local sensor-plane field and angle testing.
• ASTM International — ASTM A977/A977M-07(2020): Hysteresigraph measurement of high-coercivity permanent magnet materials. Material characterization is separate from finished-assembly encoder accuracy acceptance.