FOC vs. Six-Step BLDC Control: How to Choose

Field-oriented control can deliver smooth torque, low acoustic noise and precise current regulation. Six-step commutation can deliver a simpler, economical drive with less processing and sensing overhead. Neither method is automatically the right answer for every brushless motor system.

The useful question is not “Which algorithm is best?” It is “Which control architecture meets the motor, load, motion, acoustic, safety, cost and production requirements with acceptable engineering risk?”

This guide compares field-oriented control (FOC) and six-step commutation at the system level. It also explains why the motor label alone does not settle the decision, what each method requires from the electronics and firmware, and what should be verified on the bench before the architecture is frozen.

Start with the motor and the application

BLDC means brushless direct-current motor. In practice, BLDC and permanent-magnet synchronous motor (PMSM) terminology is not always used consistently. Vendors often associate “BLDC” with a machine designed for trapezoidal back EMF and six-step commutation, and “PMSM” with a machine designed for sinusoidal excitation. Both are permanent-magnet synchronous machines, and real motors do not always fit a clean naming boundary.

The drive should be selected from measured or supplied electrical and mechanical behavior rather than from the nameplate label. Useful inputs include:

• phase resistance and inductance, including d-axis and q-axis inductance when saliency matters;
• back-EMF shape and motor constant;
• pole-pair count and rotor inertia;
• torque-speed operating points and transient load;
• minimum and maximum speed, including startup and zero-speed requirements;
• acceptable torque ripple, vibration and audible noise;
• DC-bus range, regeneration behavior and source impedance;
• thermal limits, cooling and duty cycle;
• available rotor feedback and current-sensing architecture;
• controller cost, processing budget, diagnostic coverage and production-test constraints.

A smooth, position-controlled actuator and a fixed-speed fan can use the same broad class of permanent-magnet motor while requiring very different control systems.

How six-step commutation works

Six-step commutation divides one electrical revolution into six 60-degree sectors. In the common 120-degree conduction scheme, two motor phases conduct while the third is left undriven. The active phase pair changes at each sector boundary to advance the stator field. Microchip’s six-step application note illustrates the phase sequence and explains how the undriven phase can support back-EMF zero-crossing detection in a sensorless implementation.

Rotor sector information can come from Hall sensors, an encoder-derived state, or a sensorless estimator. Speed or current can still be closed-loop quantities. Six-step therefore does not mean open-loop control; it describes the commutation pattern.

The architecture is attractive when the application values straightforward implementation, modest processor requirements and a relatively simple feedback path. ST’s six-step overview notes that scalar six-step drives can be implemented on cost-effective microcontrollers and can regulate torque when current-mode control is used.

The tradeoff is that the commanded stator field moves in discrete steps. The effect on torque ripple, vibration, acoustic noise and efficiency depends on the motor’s back-EMF shape, winding, load, speed and commutation timing. A well-matched motor and six-step controller can be a sound engineering choice. A poor match can make the discontinuities visible in motion or sound.

How field-oriented control works

FOC regulates the stator-current vector relative to the rotor’s magnetic field. The controller samples phase currents and converts them from the three-phase stationary system into a two-axis representation. A Clarke transform produces stationary alpha-beta components; a Park transform rotates them into the rotor-aligned d-q frame.

In that rotating frame, the current components behave like DC quantities in steady state. The q-axis component is normally associated with torque production. The d-axis component is associated with flux and is often commanded near zero for a surface-mounted permanent-magnet motor below base speed, although saliency, field weakening and other objectives can change that command. Current regulators generate d-q voltage commands, an inverse transform returns them to the stationary frame, and a modulation stage produces inverter PWM commands.

The NXP PMSM FOC application note describes the separation of torque and field-producing current components and the different time scales of the current and speed loops. A TI sensored-FOC design guide shows the current measurements, Clarke and Park transforms, d-q current regulators, inverse transform and PWM generation in a complete position-control loop.

FOC needs an accurate rotor electrical angle. That angle can be measured with an encoder or resolver, derived from sufficiently precise Hall information in a suitable implementation, or estimated from motor voltages and currents. It also needs current samples with adequate accuracy and timing, deterministic execution of the fast loop, and a motor and inverter model good enough for the selected control and estimation methods.

These requirements make FOC more demanding to design and validate. They also give the engineer direct control over torque-producing and flux-producing current, which is valuable when smooth torque, dynamic response, position control, acoustic performance or a broad operating envelope matters.

FOC and six-step comparison

The comparison below is directional. Actual results depend on the motor, inverter, sensing, tuning, PWM strategy, operating point and mechanical system.

When six-step is the better fit

Six-step deserves serious consideration when its behavior already satisfies the product requirements. Common decision signals include:

• the product needs speed control but does not need servo-grade torque or position behavior;
• commutation ripple and acoustic content are acceptable at the load;
• the motor’s back-EMF and winding are a good match for block commutation;
• Hall sectors or back-EMF detection provide adequate rotor-state information over the required operating range;
• controller cost, processor load, development time or validation scope is tightly constrained;
• startup occurs under a predictable load and can be verified across bus voltage and temperature;
• the system benefits from a smaller firmware state space and easier production diagnostics.

Six-step is not automatically inefficient, crude or unsuitable for professional products. It remains widely used because it can solve the intended motion problem with fewer implementation layers. The correct evaluation is measured system performance against requirements, not algorithm prestige.

When FOC earns the added complexity

FOC becomes compelling when the application places a high value on one or more of these outcomes:

• smooth torque at low speed;
• low vibration or audible noise;
• controlled torque through load transients;
• precise velocity or position loops built around a fast current loop;
• deliberate use of motor saliency, field weakening or other advanced operating strategies;
• high utilization of the motor and inverter across a broad torque-speed envelope;
• coordinated current limiting, bus-voltage handling and regeneration behavior;
• diagnostics that benefit from explicit d-q current commands and feedback.

Those benefits are conditional. An FOC implementation with noisy current measurements, an inaccurate electrical angle, missed real-time deadlines, poor motor parameters or unstable loop tuning can perform worse than a well-executed six-step drive. The choice of FOC creates a requirement for stronger measurement, timing and validation discipline.

Sensorless control is a separate architecture decision

FOC does not mean sensorless, and six-step does not mean sensored. Either commutation family can use physical rotor feedback or an estimator, but the information needed and the low-speed behavior differ.

Back-EMF-based estimators have limited information at standstill and low speed because back EMF scales with speed. Microchip’s sensorless FOC startup documentation describes a common sequence: rotor alignment, an open-loop ramp, stabilization and transition to a back-EMF observer after the signal becomes usable. The required current and ramp depend on the electrical and mechanical system.

Other sensorless methods can extend the controllable region, but they add assumptions, tuning and test cases. Initial-position detection, high-frequency injection, saliency tracking and model-based observers each have operating constraints. If the product must start against an unknown load, produce controlled holding torque, reverse smoothly through zero speed or know absolute position, a physical position sensor may be the lower-risk system choice even when a sensorless algorithm is technically possible.

For sensorless six-step, the floating phase provides a convenient back-EMF observation window, but startup still needs an initial commutation strategy and enough motion to create a useful signal. PWM state, blanking time, bus ripple, switching noise and load acceleration all influence detection quality.

Hardware consequences of the control choice

Both methods use a three-phase inverter, gate drive, DC-link network, sensing, protection and a real-time controller. The required quality and topology of those blocks can differ.

Current measurement and ADC timing

FOC normally reconstructs or measures phase currents at defined points in the PWM cycle. One-, two- and three-shunt approaches trade component count, common-mode behavior, measurable windows, reconstruction logic and low-duty-cycle performance. Amplifier settling, offset, gain error, ADC acquisition time and PWM synchronization become part of the control-loop design.

Six-step voltage-mode control may use only bus-current monitoring for protection, while closed-loop torque or current control needs a measurement path suitable for its regulation objective. Simpler commutation does not remove the need for fast hardware overcurrent protection.

Processor and real-time budget

The relevant question is not whether a microcontroller can execute a transform in isolation. The worst-case schedule includes ADC triggering, current reconstruction, angle acquisition or estimation, current regulators, modulation, protection checks, communications and application tasks. Execution jitter and missed deadlines must be measured under representative load.

FOC may also require more calibration and commissioning data. Resistance, inductance, flux linkage, pole pairs, sensor alignment and current offsets can influence loop or observer behavior. The exact set depends on the algorithm.

Power stage, thermal design and EMC

Commutation strategy changes switching patterns and current waveforms, which can change switching loss, ripple current, conducted emissions, acoustic excitation and heat distribution. It does not replace the core work of selecting MOSFETs, designing gate loops and current-return paths, controlling dead time, protecting the DC bus, handling regeneration, and moving heat out of the enclosure.

Efficiency should be measured at the operating points that matter. Controller efficiency, motor copper and iron losses, switching frequency, modulation, torque ripple and mechanical load all contribute to the system result. A universal efficiency claim based only on “FOC” or “six-step” is not defensible.

A practical selection process

1. Define operating points and transients

Map continuous and peak torque against speed, bus-voltage range, acceleration, inertia, duty cycle, braking and regeneration. Include startup, stall, reversal and externally driven conditions.

2. Rank behavior that the customer can observe

Set measurable limits for speed error, position error, torque ripple, acoustic noise, vibration, settling time and thermal rise. Separate mandatory limits from preferences.

3. Characterize the motor and mechanical load

Confirm phase parameters, back-EMF shape, pole pairs, rotor inertia, friction, transmission effects and load variation. Supplier data is useful, but the production motor, wiring and load should be measured when risk warrants it.

4. Choose feedback and startup behavior

Decide what the controller must know at standstill, during startup and throughout the speed range. Evaluate sensor cost and reliability against the estimator’s assumptions and failure modes.

5. Budget hardware and software resources

Select current sensing, bus and phase-voltage sensing, rotor feedback, PWM/ADC resources, processor margin, memory, communications and diagnostic coverage together. Avoid choosing the algorithm before confirming that the signal chain can support it.

6. Prototype the highest-risk conditions

Run the motor, inverter and load through low bus voltage, peak load, temperature extremes, startup, reversal, rapid transients and fault recovery. Compare candidate algorithms using the same acceptance criteria.

What to verify on the bench

A credible comparison uses repeatable tests, calibrated measurements and defined pass/fail limits. At minimum, consider:

• startup success and peak current at minimum bus voltage and maximum expected load;
• operation through the minimum speed, zero-speed and reversal requirements;
• torque ripple, speed ripple, vibration and acoustic spectra at representative points;
• current-loop and speed-loop response without saturation-driven instability;
• phase-current waveform quality and current-reconstruction validity;
• current-sensor offset and drift across temperature;
• MOSFET, gate-driver, shunt, capacitor, connector and motor temperatures over duty cycle;
• DC-bus overvoltage behavior during braking or an overhauling load;
• short circuit, stall, loss of feedback, desynchronization and sensor-plausibility response;
• conducted and radiated emissions in the intended wiring and enclosure configuration;
• repeatability across motor, sensor, component and assembly tolerances.

The result may be that six-step passes every product requirement. It may show that FOC materially improves the constraints that drive customer value. It may also reveal that the decisive change is better feedback, a different motor, a stiffer power source or a redesigned mechanical load rather than a different commutation algorithm.

Frequently asked questions

Is FOC always more efficient than six-step control?

No. FOC can improve current utilization and reduce torque ripple for many motor and load combinations, but system efficiency depends on the motor, operating point, sensing, tuning, switching frequency, modulation, inverter loss and mechanical load. Compare measured input power and mechanical output over the real duty cycle.

Can FOC drive a motor sold as a BLDC motor?

Often, yes, but suitability depends on the motor’s back-EMF shape, inductance, saliency, required operating range and the quality of rotor-angle and current information. The “BLDC” label alone is not enough to guarantee a good result.

Does FOC require an encoder?

No. FOC can use an encoder, resolver or another position sensor, and it can use a sensorless rotor-angle estimator. Sensorless operation requires explicit treatment of startup, low-speed observability, parameter variation and estimator faults.

Is six-step control open loop?

Not necessarily. Six-step describes how the inverter commutates the phases. Rotor sectors can come from Hall sensors or sensorless back-EMF detection, and speed or current can be regulated in closed loop.

Should an existing six-step product be converted to FOC?

Only when the expected improvement addresses a defined product requirement and justifies changes to sensing, processing, firmware, calibration and verification. A measured baseline is necessary before the conversion can be evaluated.

Choose from requirements, then prove it

FOC is a strong architecture for smooth, precise and dynamically controlled permanent-magnet motor systems. Six-step remains a practical architecture when it meets the motion requirements with lower implementation burden. The correct choice emerges from the motor, load, feedback, operating envelope, product behavior and validation plan.

If you are defining a new drive or trying to correct startup, torque ripple, acoustic, thermal or control-stability problems in an existing one, begin with the system requirements and measured waveforms. Review Outer Reef’s custom motor-control design approach or schedule an engineering discussion.

Technical references

• Microchip AN1160: Sensorless BLDC Control with Back-EMF Filtering
• STMicroelectronics: 3-Phase Six-Step Control
• NXP AN4656: PMSM Field-Oriented Control of Industrial Drives
• Texas Instruments TIDUEW0: Sensored FOC Current and Position Loops
• Microchip: Field-Oriented Control
• Microchip: Sensorless FOC Startup Procedure