Magnetic Sensor

What is a Magnetic Sensor

A magnetic sensor is a transducer that responds to a magnetic field and produces a proportional or threshold electrical output. The field can come from a permanent magnet attached to a moving part, from current flowing in a nearby conductor, or from the Earth itself. Unlike a mechanical contact, an optical beam, or a capacitive plate, a magnetic sensor needs no physical link to the object it observes: the field passes through plastic, aluminum, stainless steel, dust, oil, and water. This single property explains why magnetic sensing dominates contactless position, speed, and proximity measurement in vehicles, motors, hydraulics, and sealed industrial equipment.

Functionally, a magnetic sensor pairs a field-sensitive element with signal-conditioning electronics. The element may be a Hall plate, a magnetoresistive thin-film bridge, or a wound fluxgate core. The conditioning stage amplifies the raw signal, compensates for temperature, cancels offset, and formats an output the control system can read: a switching logic level, a ratiometric analog voltage, a pulse-width-modulated duty cycle, or a digital word over I2C, SPI, SENT, or PWM. When the element measures field direction rather than magnitude, the device becomes an angle sensor; when it measures magnitude near a busbar, it becomes a current sensor.

The physics has a long lineage. Edwin Hall discovered the Hall effect in 1879, demonstrating that a magnetic field perpendicular to a current-carrying conductor produces a transverse voltage. Practical Hall integrated circuits arrived once silicon processing matured in the 1960s and 1970s. Anisotropic magnetoresistance (AMR) was understood from the nineteenth century but became a commercial sensor technology in the 1980s. The pivotal modern breakthrough was giant magnetoresistance (GMR), discovered independently by Albert Fert and Peter Grunberg in 1988, a result that won the 2007 Nobel Prize in Physics and first transformed hard-disk read heads. Tunnel magnetoresistance (TMR), exploiting electron tunneling through an ultra-thin insulating barrier, followed and now delivers the highest sensitivity of any room-temperature solid-state magnetic sensor.

In terms of scale, magnetic sensors cover an enormous dynamic range of field strength. Industrial Hall switches operate in the range of tens of millitesla produced by small permanent magnets. Magnetoresistive bridges resolve fields below one millitesla. Fluxgate magnetometers measure the Earth's field of roughly 25 to 65 microtesla with nanotesla resolution, and superconducting quantum interference devices (SQUIDs) reach into the femtotesla regime for biomagnetic research. No single device covers this entire span; the engineering task is to match the field a given application produces to a sensor whose range, sensitivity, and noise floor fit that field.

Four engineering attributes ultimately decide whether a magnetic sensor succeeds in service: sensitivity and resolution at the working field, offset and temperature drift, the saturation or linear-range ceiling, and the output and interface the control system expects. A sensor that is too sensitive saturates; one that is not sensitive enough buries the signal in noise. The remainder of this guide builds the vocabulary to balance these attributes against real datasheet numbers.

Sensing Technology Families

From the physical principle of the sensing element, magnetic sensors fall into five families: Hall effect, AMR, GMR, TMR, and fluxgate. The first four are solid-state silicon or thin-film devices, while the fluxgate is a coil-and-core instrument for weak fields. Each family occupies a distinct band of sensitivity, field range, cost, and power. There is no universal element. The table below compares the key engineering metrics of the five families.

Hall effect sensors apply a bias current across a thin semiconductor plate; a perpendicular magnetic field deflects the carriers and produces a transverse Hall voltage proportional to field magnitude. Hall elements have no inherent saturation up to very high fields, tolerate the strong fields of permanent magnets directly, and integrate cheaply with CMOS conditioning. Their drawback is comparatively low sensitivity and higher offset and noise than magnetoresistive elements, which modern ICs counter with chopper stabilization. Hall remains the default for digital switches, isolated current sensing, and rugged position detection.

AMR (anisotropic magnetoresistance) changes the resistance of a ferromagnetic permalloy strip depending on the angle between the current and the magnetization. AMR offers higher sensitivity than Hall and excellent low-field performance, making it a classic choice for electronic compasses and 180-degree angle measurement. Its limitations are a restricted unique-angle range and a thin film that saturates at low field, so it cannot tolerate strong magnets.

GMR (giant magnetoresistance) stacks alternating ferromagnetic and non-magnetic conductive layers. When the two magnetic layers align in parallel, resistance is low; when anti-parallel, resistance is high. The effect, discovered by Fert and Grunberg in 1988, gives a larger signal than AMR and high sensitivity in weak fields. GMR spin valves serve gear-tooth speed wheels, rotational sensing, and historically the hard-disk read heads that drove storage density. GMR has a comparatively narrow linear field span and can show hysteresis.

TMR (tunnel magnetoresistance) separates two magnetic layers with an ultra-thin insulating barrier through which electrons quantum-tunnel; an external field changes the tunneling probability and thus resistance. TMR yields the largest resistance change, the highest sensitivity and signal-to-noise ratio, and the lowest power consumption of the solid-state families, with response times in the nanosecond range. It is the leading choice for high-precision angle sensors, low-power position sensing, and compact current sensing, and increasingly displaces resolvers in electric-motor angle measurement.

Fluxgate magnetometers drive a high-permeability core into periodic saturation with an AC excitation coil; an external field unbalances the saturation symmetry, inducing even-harmonic components in a sense coil that are proportional to the external field. Fluxgates deliver vector accuracy and nanotesla resolution for measuring the Earth's field, geophysical surveying, spacecraft attitude, and navigation, but they are larger, slower, and more expensive than silicon sensors and are unsuited to strong-field industrial switching.

Hall Switch and Output Types

Within the Hall family, the most common industrial product is the digital switch IC, defined entirely by its magnetic operate point (BOP) and magnetic release point (BRP). Choosing the wrong switching class is the most frequent beginner mistake, because a magnet oriented the wrong way may never trip the device. Four digital classes exist, plus the linear and ratiometric analog outputs. The table below summarizes them.

A unipolar switch turns on when a sufficiently strong field of a single polarity (conventionally a south or positive face) reaches its operate point, and turns off when the field is removed below the release point. It is the simplest and cheapest digital Hall device, used for lid detection, end-of-travel sensing, and basic presence detection where the magnet approaches from one known direction. Because BOP and BRP share a polarity, a unipolar device ignores a magnet of the opposite orientation entirely.

A bipolar switch has a positive operate point and a negative release point, so it requires fields of both polarities to cycle. This makes it ideal for alternating ring magnets and rotating multipole wheels where north and south poles pass the sensor in sequence. An omnipolar switch trips on either a strong north or a strong south field, which simplifies factory assembly because the actuating magnet can be mounted with either pole facing the device. A latch resembles a bipolar switch but holds its last state until the opposite polarity arrives, giving tightly controlled symmetric thresholds. Latches are the standard choice for brushless DC motor commutation and for speed sensing against a multipole rotor, where predictable, hysteresis-controlled switching is essential.

For continuous measurement, a linear Hall sensor produces an output voltage proportional to field within a specified linear window; beyond that window sensitivity falls and the response bends. Most linear devices are ratiometric, meaning the output scales with the supply voltage and rests near half the supply (the quiescent or null output) when no field is present, swinging up for one polarity and down for the other. Ratiometric behavior lets a downstream analog-to-digital converter share the same reference and cancel supply variation. As a concrete reference point, the Honeywell SS495A linear Hall IC specifies a magnetic range of roughly -600 to +600 Gauss (-60 to +60 mT) with a response time near 3 microseconds, illustrating the linear span a typical industrial-grade analog Hall sensor covers.

Beyond the classic single-axis devices, modern parts integrate two or three orthogonal sensing axes to report field vectors or rotational angle directly. Infineon's TLV493D-A1B6, for example, is a low-power 3D Hall sensor that measures Bx, By, and Bz up to roughly plus-or-minus 130 mT with 12-bit resolution per axis over an I2C interface, while drawing on the order of microamperes in low-power mode. Melexis Triaxis devices such as the MLX90393 provide 16-bit X, Y, and Z output over I2C or SPI at a few milliamperes of supply current. These multi-axis parts power joysticks, contactless rotary knobs, and three-dimensional position tracking that single-axis switches cannot serve.

Applications, Magnets, and Standards

The magnetic sensor families above map onto four dominant application classes: position and angle, speed and rotation, isolated current sensing, and field or proximity detection. Understanding the application class first, then the field source, then the qualification regime, prevents the common error of choosing a sensor element before the magnetic circuit is defined.

Position and angle sensing is the largest industrial use. A magnet fixed to a shaft or slider presents a rotating or translating field to a stationary sensor, which reports angle or displacement contactlessly. Angle sensors based on TMR or AMR measure field direction, not magnitude, so they are largely immune to magnet aging and air-gap variation. TDK's TAS-series TMR angle sensors, for instance, hold an angle error of around plus-or-minus 0.6 degrees across a magnetic field range of 20 to 80 mT and a temperature range of -40 to +150 degrees C, which is why TMR angle sensors increasingly replace resolvers in electric-vehicle traction motors. Infineon's TLE5012 is a magnetoresistive angle sensor used with an external rotating magnet for similar duty.

Speed and rotation sensing uses a toothed ferrous wheel or a multipole ring magnet passing the sensor; each tooth or pole edge produces a field reversal that the sensor converts to a pulse train. GMR and Hall gear-tooth sensors handle crankshaft, camshaft, transmission, and wheel-speed measurement in vehicles, and conveyor and motor feedback in industry. Isolated current sensing exploits the field a conductor creates around itself: a Hall or magnetoresistive element near the busbar, with or without a flux-concentrating core, reads current while staying galvanically isolated from hundreds of volts, which is indispensable in motor drives, EV inverters, and solar converters.

The performance of every one of these applications depends on the magnet as much as the sensor. The table below summarizes the common permanent-magnet materials and the temperature behavior an engineer must account for, because magnet remanence drift directly becomes measurement error in any magnitude-based sensor. Published magnet datasheets cite a reversible remanence coefficient near -0.1 percent per degree C for sintered NdFeB, far less for SmCo and AlNiCo.

On the standards side, the governing regime depends on the market. For automotive electronics, the dominant qualification is AEC-Q100, which stress-tests the sensor integrated circuit across temperature grades, with Grade 0 covering -40 to +150 degrees C. Safety-critical functions such as electric power steering, throttle position, and brake-by-wire additionally require ISO 26262 functional safety, where angle sensors can be qualified to ASIL D, the highest automotive integrity level, sometimes within a single chip carrying redundant channels. Industrial switching sensors reference the IEC 60947-5 family for control-circuit and switching devices, while electromagnetic compatibility follows the IEC 61000 series. Material compliance (RoHS, REACH) applies broadly, and weak-field magnetometers depend on calibration traceability to a national metrology institute rather than any single product standard.

Key Specification Parameters

Reading a magnetic sensor datasheet is a core skill for purchasing engineers. A single device may list 20 or more parameters, but only a handful drive the selection decision: sensitivity, full-scale range and saturation, offset and quiescent output, temperature drift, linearity and hysteresis, bandwidth and response time, and the output or interface format. Each is explained below.

Sensitivity is the output change per unit field, expressed in millivolts per millitesla, millivolts per gauss, or for digital devices as the operate and release thresholds. TMR offers the highest sensitivity, followed by GMR, then AMR, then Hall; vendor figures for TMR bridges reach the millivolt-per-volt-per-fraction-of-a-gauss level. Higher sensitivity reduces the magnet strength and air gap a design needs, but it also means the device saturates at lower field, so sensitivity and range trade off directly.

Full-scale range and saturation define the largest field the sensor measures linearly. Hall elements extend from about a millitesla to well above one tesla and do not saturate within normal magnet fields, which is why they suit current sensing and strong-magnet position work. Magnetoresistive thin films saturate at low field, often a few to a few tens of millitesla, which is acceptable for angle sensors that only need a saturating in-plane field but disqualifies them near strong magnets. Always confirm that the worst-case field at the closest air gap stays within the linear span.

Offset, quiescent output, and drift determine zero-field accuracy. A linear ratiometric sensor outputs roughly half its supply voltage at zero field; deviation from that null is offset. Both the sensitivity and the offset drift with temperature, which is why quality ICs add chopper stabilization and on-chip temperature compensation. Critically, the permanent magnet drifts too, with NdFeB losing roughly 0.08 to 0.12 percent of remanence per degree C, so a magnitude-based sensor accumulates magnet drift and sensor drift together. Direction-based angle sensors avoid the magnet-amplitude term, which is the central reason TMR angle sensors hold tight accuracy over wide temperature.

Linearity and hysteresis describe how faithfully the output tracks the field. Linearity is the maximum deviation from a straight line across the range, while hysteresis is the output difference between rising and falling field at the same point. GMR and TMR films can exhibit hysteresis from domain effects, which vendors mitigate with bias structures; Hall devices are essentially hysteresis-free. Bandwidth and response time matter for speed sensing and current measurement: the Honeywell SS495A linear Hall, for example, lists a response time near 3 microseconds, and TMR elements respond in the nanosecond range, enabling high-frequency current and rotation measurement.

Output and interface is the link to the control system. Common formats include:

• Digital switch (logic level): open-drain or push-pull on/off output for unipolar, bipolar, omnipolar, and latch devices, the default for proximity and presence.
• Ratiometric analog voltage: output proportional to field and to supply, resting near half supply at null, read directly by a PLC or microcontroller analog input.
• PWM: a duty cycle proportional to field, robust over long cables and easy for microcontrollers to decode without an analog-to-digital converter.
• SENT (SAE J2716): a one-way digital protocol common in automotive position and pressure sensing, carrying data plus diagnostics.
• I2C and SPI: bidirectional digital buses on 3D and programmable parts (for example the Melexis MLX90393 at 16-bit and the Infineon TLV493D at 12-bit per axis), giving full configuration and multi-axis readout.

Two further parameters deserve attention on industrial and automotive parts: the supply voltage and current (low-power 3D Hall parts can idle in the microampere range, while bridge sensors with conditioning draw a few milliamperes), and the operating temperature grade, where the compensated range, not the storage range, is what guarantees accuracy. An AEC-Q100 Grade 0 part is rated to +150 degrees C, but a commercial part may compensate only to +85 or +125 degrees C.

Selection Decision Factors

To turn the preceding chapters into a specific part number, follow the decision sequence below. Most selection mistakes come not from a single wrong choice but from deciding the sensor element before the magnetic circuit and the system interface are defined. These eight steps double as a fixed RFQ template.

1. Define the measurand: presence (switch), linear position, rotary angle, speed or rotation, isolated current, or weak-field magnetometry. The measurand alone narrows the family, for example angle favors TMR or AMR, current favors Hall or TMR, and switching favors a digital Hall.
2. Quantify the field at the sensor: compute or measure the worst-case minimum and maximum flux density at the sensor face across the full mechanical travel and air-gap tolerance. Field from a small magnet falls roughly with the cube of distance, so air gap dominates this number.
3. Match range and sensitivity: confirm the working field sits inside the linear window for analog parts, or clears BOP with margin and drops below BRP for switches. Reject any part that saturates at the worst-case high field or buries the signal in noise at the worst-case low field.
4. Choose the output and protocol: digital switch, ratiometric analog, PWM, SENT, or I2C and SPI. Match the controller input and cable length; long runs favor PWM, SENT, or current-style outputs over raw analog voltage.
5. Specify temperature and the magnet: set the compensated temperature range (not storage), then choose a magnet material whose remanence drift the system can tolerate, or move to a direction-based angle sensor to remove magnet-amplitude drift entirely.
6. Pin down qualification and safety: AEC-Q100 grade for automotive, ISO 26262 ASIL level for safety-critical functions, IEC 60947-5 and IEC 61000 for industrial switching and EMC, plus RoHS and REACH material compliance.
7. Set mechanical and environmental limits: package and footprint, mounting and air-gap repeatability, ingress protection of any housed assembly, vibration, and exposure to stray fields from nearby motors or busbars that may need shielding.
8. Evaluate total cost of ownership: unit price plus the cost of the magnet and magnetic circuit, calibration, and field-failure risk. A direction-based angle sensor may cost more per unit but eliminate recurring recalibration that a drift-prone magnitude sensor demands.

One last dimension is commonly overlooked: supply continuity and serviceability. Magnetic sensor ICs are subject to long product lifecycles and occasional end-of-life notices, so for production designs verify the part's lifecycle status, second-source availability, and the vendor's automotive or industrial longevity commitment. Established suppliers such as Allegro MicroSystems, Honeywell, Texas Instruments, Infineon, Melexis, TDK, TDK-Micronas, Sensitec, NVE, and MultiDimension Technology maintain broad portfolios and qualification documentation, which reduces the risk of a sensor obsoletion forcing a redesign five to ten years into a product's life.