Hall Effect Sensor in BLDC Motors: A Guide to Principles, Control, and Application Selection

Hall Effect Sensor in BLDC Motors: Principles, Placement and Signal Integrity

What the Sensor Actually Detects

A Hall effect sensor in a BLDC motor has one job, and it is narrower than most descriptions suggest. It does not measure position, it does not measure speed, and it does not close a servo loop. It reports which of six states the rotor is in, so the drive knows which pair of phases to energise during the next 60 electrical degrees.

Everything that goes wrong with sensored commutation comes from that narrowness being ignored. A sensor placed two mechanical degrees off, a latch with the wrong threshold, a pull-up resistor picked for the BOM rather than the cable, or a magnet ring that has weakened with temperature will all keep the sequence legal while quietly moving the edges. The motor still turns; it turns with more current, more ripple and more heat than its datasheet implies.

The Hall effect is a transverse voltage that appears when a current-carrying conductor sits in a magnetic field perpendicular to the current. Charge carriers are pushed to one side by the Lorentz force, and the resulting charge separation produces a voltage across the conductor that is proportional to the flux density and to the bias current, and inversely proportional to the carrier density and thickness of the sensing element.

The sensor in a motor is not that element on its own. It is an integrated circuit: a thin semiconductor Hall plate, a bias current source, a threshold comparator with hysteresis, a temperature-compensating amplifier and an output transistor, all in one package of a few square millimetres. The magnetic field never appears at the output. What appears is a decision, taken against an internal threshold, and the threshold is the only part of the measurement the drive ever sees.

On a datasheet, therefore, the parameters that matter are the thresholds, not the sensitivity curves. Two devices with identical magnetic sensitivity can switch at completely different flux levels, and the drive cannot tell the difference until the motor is hot, the air gap has grown and one of them is missing poles.

Latching, Bipolar or Linear: Which Device Belongs in a Motor

Four families of Hall device are sold into motor work, and only one of them is the default for six-step commutation. The distinction is not about accuracy; it is about what the output does when the field returns to zero, which happens twice per electrical cycle at every pole boundary.

The reason a latch is mandatory is the pole boundary. Between a north and a south pole the field passes through zero, and a unipolar or omnipolar switch would release there, producing an extra edge in the middle of a sector and a false commutation command. A latch holds through that region and switches only when the opposite pole has actually arrived, which is exactly the behaviour a state machine needs.

BOP, BRP and the hysteresis that keeps the output clean

A latching device is specified by two thresholds: the operate point BOP, at which the output changes state as flux rises in one direction, and the release point BRP, at which it changes back as flux rises in the other. The gap between them is the hysteresis, and it is what stops the output chattering when the field sits near a threshold with noise on it. Devices with a low operate point are described as more sensitive and tolerate a larger magnet-to-sensor gap, at the price of being easier to trigger with stray fields ^[10].

Why Three Sensors, and Why 120 Electrical Degrees

Rotor position in a motor is counted in electrical degrees, not mechanical ones. One electrical cycle is the distance between a north pole and the next north pole, so a rotor with more poles packs more electrical cycles into one revolution. The conversion is a single multiplication by the pole-pair count: θe = p · θm, where p is the number of pole pairs.

Three sensors spaced 120 electrical degrees apart produce a three-phase square-wave set. Each of the six combinations then corresponds to exactly one 60-degree sector of the electrical cycle, so six-step commutation needs no interpolation: the state itself is the sector. In mechanical terms the spacing shrinks as pole count rises, and this is where most custom motor designs get into trouble.

Mechanical spacing shrinks as the pole count rises

An eight-pole rotor needs its sensors 30 mechanical degrees apart. That happens to equal the slot pitch of a 12-slot stator, so the sensors can sit in the slot openings with a simple PCB and no adjustment. Put the same PCB on a nine-slot stator, whose slot pitch is 40 degrees, and there is no slot at 30 degrees: the sensors have to move to their own carrier, or the motor is built with a sensor housing that allows angular trimming. A one-degree mechanical error on an eight-pole rotor becomes four degrees of electrical timing error. High-pole motors are the least forgiving of the same mechanical tolerance.

Six States, One Truth Table

The whole interface between the motor and the drive is the table below. Each row is a valid Hall combination, the phase pair that should be driven for that sector, and the phase that must be left floating so that it can carry the back EMF the drive may later use for sensorless handover or for diagnostics ^[10].

The sector numbering is a convention, not a property of the motor: whether sector 1 sits at the north-pole reference or 60 degrees away depends on the winding phase order and on which way the machine is expected to turn. The two invalid states carry more weight. A shorted harness, a dead sensor or a missing magnet ring all land on 000 or 111, and a drive that trusts three pins on every sample will happily commutate on all-low and destroy a bridge.

The sequence also explains the shape of the current waveform. Each phase conducts for 120 electrical degrees in the positive direction, floats for 60, conducts for 120 in the negative direction and floats again, hence the name 120-degree conduction; each phase carries current for two thirds of the cycle. The motor constant that turns that current into torque is a property of the winding and magnet, not of the sensors; the sensors only decide when the current is applied ^[5].

Reading the Signal: Pull-Up, Filter, Shield and Route

Between the sensor package and the controller there is a wire, and on most machines that wire is the weakest part of the commutation system. The output of a latching Hall device is usually open-drain, which means it can only pull low; the high level is created by a pull-up resistor, and the value of that resistor sets both the edge speed and the noise immunity of the whole channel. How the drive turns those six states into a bridge pattern is covered in our guide to the BLDC motor controller.

Pull-up value is an EMC decision

The pull-up trade-off is worth stating with numbers. With a 10 kΩ pull-up and 200 pF of cable and input capacitance, the rising edge has a time constant of 2 µs. At a low electrical frequency that is invisible; at 2 kHz electrical, where a state lasts 83 µs, it is still acceptable. What is not acceptable is the other half of the same choice: 10 kΩ also makes the node a high-impedance target for capacitive coupling from the phase cables running beside it.

Dropping to 2.2 kΩ costs a milliamp or two of supply current and buys a much stiffer node. 2.2 to 4.7 kΩ at 5 V is a better default than the 10 kΩ in most reference designs, particularly on a motor with a harness longer than half a metre.

Harness: three rules that cover most failures

Route the Hall bundle away from the phase bundle, not alongside it. Bond the shield at the controller end and, on a long run, at both ends. Give the sensor its own supply and ground pins on the connector instead of sharing the logic supply of a power stage. On vehicles and AGVs the harness is long enough that a differential or buffered arrangement is worth considering, and the EMC requirements of the complete drive are assessed as a power drive system, not as a motor ^[2].

Timing Error and What It Costs in Torque

Commutation timing is the angle between the stator current vector and the rotor flux. In an ideal six-step drive each sector is centred so that this angle stays near 90 electrical degrees, where torque per amp is highest. Move every edge by the same angle and the whole waveform shifts: the motor still runs, but at a lower torque per amp, which the drive compensates for by drawing more current for the same load. Converting that current into a shaft torque follows the motor constant arithmetic set out in our note on BLDC motor torque calculation.

The instantaneous factor is the cosine of the error. A 15-degree shift costs about three per cent of torque, which nobody notices. A 30-degree shift costs more than thirteen per cent, which shows up as a hot motor on a machine that used to run cool. Beyond 60 degrees the sign of the torque in part of the cycle reverses and the drive is effectively braking in one sector and driving in the next.

Uniform shift versus non-uniform placement

Uniform misalignment is the easy case because it can be compensated by advancing the commutation angle in the drive. The harder case is non-uniform placement, where the three sensors are not exactly 120 degrees apart from each other. That produces unbalanced phase currents and low-frequency torque harmonics, not a simple loss of torque, and no single advance angle fixes it.

The measurement and compensation methods for that case are documented in the literature, including filtering of the Hall signals and online adjustment of the advance angle ^{[6] [8]}. Some of that harmonic content is audible as a buzz that tracks load and not speed, so the complaint often arrives as a noise problem. Our guide to reducing electric motor noise covers the other mechanisms in a machine.

Hall Edges Are Not Position Feedback

The most expensive misunderstanding in this subject is treating the Hall signals as an encoder. They are not: they are a six-step state machine, and the resolution that comes with it is coarse by design. Each state spans 60 electrical degrees, which becomes 60 divided by the pole-pair count in mechanical degrees.

On a four-pole motor one state is 30 mechanical degrees. On an eight-pole motor it is 15. A positioning axis that needs a tenth of a degree cannot be closed on that signal, and no amount of edge interpolation changes the fact that the sensor was chosen to answer a yes-or-no question. What the edges can do is provide a speed estimate: each electrical cycle contains six edges, so an eight-pole motor turning at 1,000 rpm produces 400 edges per second, and a capture timer with microsecond resolution measures the period between them well enough for speed control ^[5].

The practical consequence is architectural. A machine that needs both commutation and positioning normally carries both devices: Hall sensors for the drive, an encoder or a linear Hall pair for the controller. Designs that try to run positioning from three latches end up with a slow loop, a coarse following error and a tuning exercise that never converges. The trade-offs between the two topologies are set out in our comparisons for lawn mower drives and for AGV traction.

Faults That Look Like Motor Failures

Most sensored motors returned as “weak” have a winding that is fine and a sensor chain that is not. Three physical layers are involved and the repair differs for each, so the symptoms in the table overlap.

The thermal failure is not a motor fault

The thermal case is the one that gets misdiagnosed as a motor problem. A latching sensor has a maximum junction temperature that is often lower than the winding insulation class allows, and it sits on a board that is bolted to a stator carrying the winding heat. A sensor that worked on the bench and fails after twenty minutes is usually sitting in a cavity that is hotter than the designer assumed, and the enclosure rating of the motor says nothing about the temperature inside it ^[1].

On a traction machine the same defect presents as a general overheat, not a sensor fault, so it is worth reading our note on AGV motor overheating before condemning a motor.

Selecting the Device: Eight Parameters

The eight parameters below decide whether a latch works in a given motor. They are listed in the order in which they cause redesigns, not in the order in which they appear on a datasheet.

Frequency bandwidth is the parameter most often skipped, and it can be calculated before a part is chosen. Electrical frequency is fe = n · p / 60, and each electrical cycle contains six states, so the state rate is six times that. An eight-pole motor at 6,000 rpm runs at 400 Hz electrical and 2,400 states per second, with each state lasting about 420 µs. A device specified for a few kilohertz handles that comfortably; the same device on a 12-pole motor at 20,000 rpm faces 2 kHz electrical and 12,000 states per second, with 83 µs per state, which is where a slow sampled output begins to distort the timing it is supposed to report.

Incoming Inspection: Six States, in Order

A Hall sensor chain is easy to verify and almost never verified. The check takes a minute per motor and catches the faults that survive a functional test on a bench supply, because a motor can spin with a misaligned sensor set and still pass a no-load run.

The width-balance check is the one that earns its keep. Six roughly equal intervals mean the sensors are 120 electrical degrees apart and the magnet ring is uniform; one short interval means a placement error, and a gradual shortening across the whole set means the ring or the air gap is varying with angle. Neither condition produces an alarm on a drive that only validates states, and both consume torque margin for the life of the machine. The ten per cent figure in the table is a workshop rule rather than a standard limit, and it should be tightened on a high-pole motor, where the same mechanical tolerance costs proportionally more electrical angle.

When Hall Feedback Is Mandatory

Sensorless control has improved to the point where the choice is no longer “sensored for performance, sensorless for cost”. It is a question about the duty, and five duty conditions decide it in practice.

Where the duty falls in the first four rows, the motor is specified with sensors and the work moves to getting them right: placement, latch thresholds, the harness and the acceptance test described above. Where it falls in the last row, the sensorless route is cheaper and the comparison is a different article; the machine-level trade-offs for a traction application are covered in our note on AGV speed and rpm selection, and the thermal consequences of getting commutation timing wrong surface as overheating, not as a sensor fault, so they are usually chased in the wrong place ^[3].

Why Greensky Power for Sensored BLDC Motors

We build sensored and sensorless BLDC motors from 20 mm to 130 mm frames, and for the sensored types the sensor carrier, the magnet ring and the winding are designed together rather than assembled from three catalogues. That matters because the 120-degree spacing is a relationship between the sensor PCB and the stator slots: get it wrong and no amount of drive tuning recovers the torque.

Every sensored motor we ship is checked in Hall-monitor mode, one revolution by hand, with the six state widths recorded in the test file. That record is what makes a warranty conversation short: either the motor left with balanced states or it did not. The application areas we build for are indexed under electric motor applications, and our Hall sensor BLDC motor sourcing guide sets out the specifications we would want to see from any supplier, which is the same list we hold ourselves to.

Custom work covers sensor phasing to match an existing drive, 60-degree or 120-degree output sets, integrated harnesses with shielded cable, and connectors rated for the enclosure the motor sits in. If the requirement is a motor that has to produce torque from zero speed or hold a load without back-driving, start with the duty and the harness, not the frame size. The wider range is on our brushless DC motor page, the fundamentals of how the commutation loop closes are in BLDC motor basics, and the programme that runs a design from sample to series is described under custom electric motor manufacturing.

References

1. International Electrotechnical Commission. (2026). IEC 60034-1:2026 — Rotating electrical machines, Part 1: Rating and performance. IEC. https://webstore.iec.ch/en/publication/89961
2. International Electrotechnical Commission. (2022). IEC 61800-3:2022 — Adjustable speed electrical power drive systems, Part 3: EMC requirements and specific test methods for PDS and machine tools. IEC. https://webstore.iec.ch/en/publication/65056
3. International Electrotechnical Commission. (2020). IEC 60034-5:2020 — Rotating electrical machines, Part 5: Degrees of protection provided by the integral design of rotating electrical machines (IP code). IEC. https://webstore.iec.ch/en/publication/64224
4. International Electrotechnical Commission. (2013). IEC 60529:1989+AMD2:2013 — Degrees of protection provided by enclosures (IP code). IEC. https://webstore.iec.ch/en/publication/2452
5. Krishnan, R. (2010). Permanent magnet synchronous and brushless DC motor drives. CRC Press. https://doi.org/10.1201/9781420014235
6. Samoylenko, N., Han, Q., & Jatskevich, J. (2008). Dynamic performance of brushless DC motors with unbalanced Hall sensors. IEEE Transactions on Energy Conversion, 23(3), 752–763. https://doi.org/10.1109/TEC.2008.921555
7. Alaeinovin, P., Chiniforoosh, S., & Jatskevich, J. (2008). Evaluating misalignment of Hall sensors in brushless DC motors. 2008 IEEE Canada Electric Power Conference. https://doi.org/10.1109/EPC.2008.4763350
8. Park, J. S., & Lee, K.-D. (2017). Online advanced angle adjustment method for sinusoidal BLDC motors with misaligned Hall sensors. IEEE Transactions on Power Electronics, 32(11), 8247–8253. https://doi.org/10.1109/TPEL.2017.2694042
9. Yao, X., Jiang, X., & Zhang, Y. (2017). Compensation method for commutation torque ripple reduction of BLDC motor with misaligned hall sensors. IECON 2017 — 43rd Annual Conference of the IEEE Industrial Electronics Society. https://doi.org/10.1109/IECON.2017.8216315
10. Andrews, C., Soltero, M., & Mesganaw, M. (2022). SLVAEG3B — Brushless DC motor commutation using Hall-effect sensors (Application brief, rev. B). Texas Instruments. https://www.ti.com/lit/an/slvaeg3/slvaeg3.pdf

FAQ