In every magnetic position, speed, or proximity sensor, the magnet is half the system — and the half that gets specified last, tolerated loosely, and blamed first. This guide covers the sensing configurations, how to size the field at the working gap, and how to specify a sensor magnet that performs in production, not just on the bench.
A magnetic sensor system is a transducer pair: a magnet that encodes mechanical position into a field pattern, and a sensing element that reads it. Datasheets, app notes, and design attention overwhelmingly go to the silicon half — while accuracy, temperature drift, and unit-to-unit variation in the finished product are usually dominated by the magnet half: its field level at the gap, its pole geometry and placement tolerance, and its thermal behavior.
Treating the magnet as an engineered component — specified by field at the working point, verified by measurement, toleranced where it matters — is what separates a sensor design that calibrates once from one that fights yield forever.
02
Sensor technologies at a glance
Technology
Measures
Typical working field
Magnet implications
Hall switch / latch
Field threshold crossing (on/off)
Operate points ~1–30 mT typical
Field at gap must clear operate/release with margin over temperature and tolerance
Linear Hall
Field magnitude (analog/ratiometric)
Linear ranges ~±10–100+ mT by device
Field vs. position slope is the signal — magnet geometry sets linearity
2D/3D Hall angle sensors
Field direction (angle)
Often ~20–70 mT at the IC
Diametric magnets; field magnitude stays in window, angle carries the data — tolerant to strength drift
AMR
Field direction (180° ambiguity)
Saturated-mode: tens of mT
Strong-enough field so direction dominates; pairs with diametric/multipole
TMR / GMR
Direction or field, high sensitivity
µT–tens of mT by design
High sensitivity → stray-field management matters as much as the magnet
Reed switch
Field threshold (mechanical contacts)
Specified in ampere-turns (AT)
Orientation-sensitive actuation zones; see section 07
Variable reluctance / inductive
Flux rate of change
—
Often uses a toothed steel wheel + stationary magnet; magnet stability sets baseline
Working-field figures are order-of-magnitude orientations; the sensor datasheet governs. The magnet's job is to place the field at the IC inside that window across all tolerances and temperatures.
03
The five sensing configurations
Configuration
Motion sensed
Typical magnet
Notes
Head-on (proximity)
Approach along the pole axis
Axial disc or block
Field rises steeply near contact — good switching, poor linear range
Slide-by
Lateral pass at fixed gap
Axial disc/block, or 2-magnet pair
Bipolar signal with pair; position of zero-crossing is stable vs. strength drift
End-of-shaft rotary
Absolute angle, 0–360°
Diametric disc on shaft end, sensor on axis
The dominant modern angle-sensing architecture — section 05
Off-axis / through-shaft rotary
Angle or speed around a shaft
True radial ring or OD multipole ring
When the shaft end isn't available — section 06
Linear position
Travel along a stroke
Axial magnet (short strokes) or multipole strip/array
Pole pitch sets resolution for incremental linear encoding
Full geometry definitions for each magnetization option — axial, diametric, true radial, multipole — with diagrams, live in the magnetization directions guide; this page focuses on making them work with a sensor.
04
Designing the field at the gap
The design variable is B at the sensor location — not the grade, not the surface field, not the pull force. Work it in this order:
Establish the real working distance — magnet face to sensing element, including package standoff (the die sits ~0.3–1 mm inside the IC package), housing walls, and the tolerance stack on both sides. The stack usually matters more than the nominal.
Set the field window: from the sensor datasheet — minimum field at maximum gap and temperature; maximum field at minimum gap (linear devices clip, angle sensors have upper windows too).
Size geometry before grade: field at distance is driven by magnet dimensions far more than grade steps (a grade step buys ~3–4% field — the grades chart math). Diameter-to-thickness ratio shapes how fast the field falls with gap: larger faces throw field further.
Then pick the grade and temperature class for the thermal environment (next sections and the temperature guide).
Prefer designs that tolerate strength variation: angle-based and zero-crossing architectures ride out the ±3–5% Br spread between lots; absolute-threshold designs inherit it as error. If your architecture measures field magnitude directly, budget for lot spread explicitly.
The most useful spec line
"Flux density B = X mT ±Y% measured on-axis at Z mm from the marked face, at 25 °C" — a field-at-working-point requirement is measurable, enforceable, and communicates the actual design intent better than any combination of grade and dimensions alone.
05
Rotary: diametric 2-pole design
A diametrically magnetized disc on the shaft end, with a 2D/3D Hall or magnetoresistive angle sensor on the rotation axis, delivers absolute 0–360° sensing with a handful of parts — throttle and pedal position, valve actuators, steering, knobs, BLDC commutation.
Magnet size: typically 3–8 mm diameter discs; the field magnitude at the IC needs to sit in the device's window across the gap tolerance — usually easy to satisfy, which is why this architecture is so forgiving.
What actually drives accuracy: mechanical placement — eccentricity of the magnet on the rotation axis, tilt, and axial gap variation bend the field direction and appear directly as angle error. Specify magnet OD-to-fixture concentricity and the mounting method (press into a brass/aluminum holder, adhesive per the bonding guide) as carefully as the magnet.
Clocking: if the electrical zero must align to a mechanical feature, the pole-axis-to-feature angle and tolerance go on the drawing (e.g., "pole axis to flat: 0° ±2°") — and the supplier verifies it in a rotary fixture, per the testing guide.
Stray fields: for automotive stray-field immunity requirements (ISO 11452-8 class environments), differential/gradiometric sensor variants pair with the same magnets — the magnet spec doesn't change, the IC does.
06
Rotary: multipole & true radial rings
When the shaft end is occupied — through-shafts, hollow shafts, large-diameter joints — the field moves to a ring around the shaft, read by a sensor at the OD:
Incremental encoding — OD multipole rings: alternating poles around the circumference; each pole pair is one signal period, so pole pitch sets resolution (a 32-pole ring gives 16 periods/rev; interpolation multiplies from there). Accuracy lives in the pole-transition placement: specify pole count, peak field at the sensor gap, pole-to-pole balance, and maximum transition angle error — the parameters an automated pole scan verifies.
Speed/direction sensing: the same rings at coarser pitch replace toothed-wheel VR targets in ABS and motor feedback — quadrature from two offset sensors gives direction.
Uniform-field applications — true radial rings: torque sensors, absolute off-axis schemes, and couplings need a continuous, joint-free radial field. A one-piece true radially oriented ring delivers it; a glued arc-segment "radial ring" puts a flux dip at every joint that appears directly as periodic signal error. Specify one-piece construction explicitly — the comparison table and drawing language are in the directions guide, and manufacturing capability is exactly what the True Radial line exists for.
Material choice for rings: sintered NdFeB for maximum signal; bonded NdFeB when fine pole pitch, complex patterns, or over-molding onto a hub matter more than raw field (sintered vs. bonded).
07
Reed switch actuation
Reed switches actuate on flux along the reed axis, with sensitivity specified in ampere-turns — so the magnet's orientation relative to the reeds matters as much as its strength. A magnet approaching parallel to the reed axis gives one clean operate zone; perpendicular approaches create multiple operate/release lobes that surprise designers in testing.
Design to operate AND release: the release point (hysteresis) defines the off state — verify the magnet's far position drops below release across tolerances, or the switch never opens.
Typical pairings: small axial NdFeB discs/blocks for compact security and appliance switches; ferrite or alnico where cost or legacy sensitivity curves apply. Ferrite in cold outdoor duty carries the low-temperature caveat from the material comparison.
Prototyping tip: reed + magnet interaction maps poorly to hand calculation — a quick bench map of operate/release positions with the real switch beats simulation for these devices.
08
Temperature drift & stability
NdFeB drifts ~−0.12%/°C in Br — a threshold-based sensor calibrated at 25 °C sees its switch point move with temperature; a 100 °C swing is a ~12% field change. Angle- and ratio-based architectures cancel most of it; magnitude-based ones must budget or compensate for it (many sensor ICs offer programmable TC compensation matched to NdFeB — use it).
Precision instruments reach for SmCo (−0.035%/°C) or accept the size penalty for 3–4× less drift — the classic aerospace and metering choice from the material comparison.
Irreversible loss is a calibration killer: one over-temperature event permanently shifts every threshold. Spec the temperature class against the real profile with a load-line check — small sensor magnets are often thin discs, exactly the geometry the temperature guide flags as vulnerable below their rated class.
Thermal stabilization before calibration: for calibrated devices, pre-bake magnetized parts above worst-case service temperature so the one-time knock-down happens before, not after, calibration — specify it on the drawing.
09
Common design mistakes
Mistake
Consequence
Fix
Specifying "strongest grade" instead of field-at-gap
Saturated linear sensors, clipped signals
Design B at the IC into the datasheet window; grade follows
Ignoring the die position inside the IC package
Systematic gap error, threshold surprises
Use sensitive-point location from the sensor datasheet
Diametric magnet without clocking spec
Random electrical zero per unit
Pole-axis-to-feature angle + tolerance on the drawing
Segmented ring where the signal needs true radial
Periodic error at segment frequency
"One-piece true radially oriented" drawing note
Calibrating at 25 °C, deploying at 85 °C, no TC plan
Threshold drift, field returns
Architecture that cancels drift, or programmed compensation
Accepting magnets on grade + dimensions only
Lot-to-lot signal spread
Field-at-point or helmholtz moment acceptance limit (testing guide)
Nickel-plated magnet loose in a precision press-fit holder
Position shift over life, cracked magnets
Bond + capture per the bonding guide; no interference fits on bare magnets
10
Specifying a sensor magnet
The sensor-magnet additions to the standard RFQ checklist:
Field at the working point: B (mT/G) at a defined distance from a defined face, with tolerance and temperature of measurement.
Configuration & pattern: axial / diametric / true radial one-piece / multipole with pole count and location — plus clocking angle and tolerance where the application needs it.
For multipole rings: peak field at the sensor gap, pole-to-pole uniformity, and maximum transition angle error.
Thermal: operating range, temperature class per load-line check, and thermal stabilization requirement for calibrated devices.
Acceptance test: the measurement method that defines a good part — fixtured field-at-point, helmholtz moment, or pole scan — with limits anchored to first-article hardware.
Assembly interface: holder material (non-magnetic), mounting method, and concentricity/perpendicularity to the functional axis.
Sensor magnets are a Radial Magnets specialty — from stock diametric discs to custom true radial and multipole encoder rings with pole-scan certification. Send the sensor part number, the gap, and the motion, and our engineers will spec the magnet with you.