Key Takeaways
- Both drive the same three-phase brushless motor; the difference is commutation. Six-step switches two phases at a time, FOC drives all three with continuous sinusoidal current.
- Ideal six-step drive carries roughly 13 percent peak-to-peak torque ripple; FOC reduces it to a few percent, which is why FOC platforms sound and fly smoother.
- Sensorless FOC cannot start or hold low speed reliably because the back-EMF observer needs speed to work; six-step has the same limit but is far more forgiving about it.
- For very high-KV racing motors, six-step often wins outright: the electrical frequency outruns what a cost-effective FOC loop can regulate.
FOC vs BLDC is usually presented as new versus old, which is wrong and leads people to buy the wrong ESC. As of 2026 both topologies remain in volume production because they are two different answers to the same physics problem: how to commutate a three-phase brushless motor without wasting torque. FOC wins on smoothness, efficiency and noise; six-step BLDC wins on burst thrust, motor tolerance, cost and the ability to spin a very high-KV motor at all. The right choice is a function of the airframe, not of which one is more modern.
What an ESC actually does
An electronic speed controller sits between the flight controller and the motor and performs three jobs:
- Commutation. Energise the right stator phases in the right sequence so the magnetic field rotates and drags the rotor with it.
- Power modulation. Vary the average voltage across the windings, usually by PWM, to set speed and torque.
- Protection. Limit current, detect desynchronisation, cut off on over-temperature, and brake if asked.
Everything that follows is a consequence of how the first job is done. Both ESC families use the same power stage: six MOSFETs in three half-bridges. The difference is entirely in the gate signals.
BLDC: six-step commutation
Six-step, also called trapezoidal or block commutation, energises exactly two phases at any instant. One phase is pulled high, one is pulled low, and the third floats. The six valid switch states step the stator field around in 60-degree increments, which is why the method is called six-step.
To know when to step, the controller needs rotor position. Two approaches dominate:
- Hall sensors. Three digital outputs give 60-degree resolution directly. Simple, robust, and still common in industrial and hobby motors.
- Sensorless back-EMF sensing. The floating third phase carries the back-EMF of the unenergised winding. Its zero crossing tells the controller where the rotor is. This is what almost every FPV and multirotor ESC does, because it removes three wires and a connector.
The consequences of energising only two phases at a time are structural, not implementation details:
- Torque ripple. Current is a square wave, so the torque produced has a sixth-harmonic ripple. The textbook figure for ideal 120-degree conduction is about 13 percent peak-to-peak, and in practice it is worse because back-EMF is never perfectly trapezoidal.
- Harmonic current. A square-wave current contains the fundamental plus 5th, 7th and higher harmonics. Those harmonics produce no useful torque. They produce copper loss, iron loss and audible noise.
- Cheap compute. Six steps per electrical revolution means a low-end MCU or even a comparator-plus-gate-driver design is enough. That is the whole reason this topology is inexpensive.
Six-step is also extremely tolerant. It does not need motor resistance, inductance, or flux linkage. It does not care much if the motor constants are wrong, because it never uses them.
FOC: field-oriented control
Field-oriented control drives all three phases with continuous sinusoidal current and controls the stator field in a rotating reference frame. The sequence is:
- Measure two or three phase currents.
- Clarke transform: three-phase quantities become two orthogonal stationary components, alpha and beta.
- Park transform: using the rotor electrical angle, alpha and beta become two rotating DC quantities, d and q.
- Run two PI loops, one on d and one on q. Normally the d-axis current command is held at zero, so all current produces torque and none is wasted magnetising the rotor.
- Inverse Park and inverse Clarke back to three-phase voltage commands.
- Space-vector PWM synthesises those commands on the six MOSFETs.
The loop closes in the rotor frame, which is the key point: in that frame a correctly commutated motor looks like a DC motor with a constant torque constant. Current maps linearly to torque, so the outer speed or thrust loop becomes trivial to tune.
That elegance costs three things:
- Rotor angle at all times. FOC needs the electrical angle continuously, not just at six discrete points. With sensors this is easy; sensorless it becomes the hard part of the whole design.
- Current sensing. At least one shunt is required, and two or three are better. Single-shunt sensing needs a specific PWM pattern with a measurable window, which constrains modulation index at high duty.
- Compute. Two transforms, two PI loops and SVPWM at 20 to 40 kHz is roughly one to two orders of magnitude more work than six-step. That is affordable on a modern 32-bit MCU and not affordable on the cheapest ones.
The comparison that matters
| Dimension | Six-step BLDC | FOC |
|---|---|---|
| Phases energised | Two at a time | All three, continuously |
| Current waveform | Square / trapezoidal | Sinusoidal |
| Torque ripple | About 13 percent ideal, worse in practice | A few percent |
| Throttle linearity | Noticeable steps and dead band at low duty | Linear and smooth |
| Efficiency at part load | Lower, harmonic losses | Higher, typically a few percent |
| Acoustic noise | Audible sixth-harmonic whine | Quieter, higher-frequency residual |
| Low-speed torque | Poor, cogging and ripple dominate | Good, smooth from near zero |
| Sensorless startup | Open-loop ramp, generally works | Hard, needs a dedicated strategy |
| Very high electrical frequency | Handles it well | Loop bandwidth becomes the limit |
| Motor parameter knowledge | Not needed | Needed, or the observer misbehaves |
| MCU cost | Low | Higher |
| Tuning effort | Almost none | Real work, per motor |
Where the efficiency gain actually comes from
Vendors like to quote a single efficiency number. It is more honest to say that FOC’s advantage is load-dependent and comes from three specific places.
Copper loss. Loss is current squared times resistance. A square-wave current with the same average torque carries extra harmonic current that produces heat but no torque, so for the same output the winding runs hotter. Sinusoidal current removes most of that penalty.
Iron loss and eddy currents. High-order current harmonics excite the stator laminations at frequencies well above the fundamental, and iron loss rises with frequency. A cleaner waveform means less of this.
Operating point. The gain is largest at part load and low speed, where harmonic current is a big fraction of total current, and smallest at full throttle near the motor’s peak efficiency point. A multirotor that hovers at 40 percent throttle spends most of its life exactly where FOC helps most. A racing quad that lives at 90 percent throttle sees much less benefit.
That is why the same comparison produces wildly different numbers in different reviews: they are measuring different operating points, sometimes on different motors, sometimes with the throttle simply held at full.
The hard part of sensorless FOC
Sensorless FOC needs the rotor angle, and without a sensor it has to estimate it. The dominant methods all share one weakness.
- Back-EMF observers such as the sliding-mode observer estimate the flux linkage and derive angle from it. Back-EMF is proportional to speed, so at low speed the signal disappears into measurement noise.
- Flux linkage integration works at moderate speed and drifts at low speed because of integrator offset.
- High-frequency injection superimposes a carrier and reads the rotor saliency. It works at standstill but needs a salient rotor, adds audible noise, and is sensitive to inductance saturation.
In practice a sensorless FOC ESC starts the motor open-loop, ramping voltage and frequency until the observer has enough signal, then hands over. That handover is where most field failures live: too early and the observer diverges, too late and the motor has already drawn a current spike.
Six-step has exactly the same physical limit, because back-EMF zero crossing also vanishes at low speed. The difference is that six-step’s open-loop ramp is far more forgiving. It does not depend on a model, so a wrong ramp just means a rough start rather than a runaway.
When six-step is genuinely the better answer
This is the part the marketing skips.
Very high-KV motors. A small racing motor can run at electrical frequencies well above what a cost-effective FOC loop can regulate at a sane PWM frequency. When the electrical period shrinks toward the control period, the discrete-time loop loses phase margin and the current loop cannot track. Six-step only needs six commutations per electrical revolution, so it keeps working where FOC starts to smear.
Burst thrust. Six-step responds to a throttle step in one commutation state change. There is no observer to converge and no PI loop to slew. For an acro quad or a car where the throttle is a switch, that directness is a feature.
Motor tolerance. Six-step does not know or care about motor constants, so a cheap ESC drives almost any motor of the right size. FOC tuned for one motor can misbehave on another, which is a real cost when you sell hardware to people who mix and match parts.
Cost and failure modes. No current sensor, no observer, fewer things to get wrong. A six-step ESC that desynchronises usually recovers; an FOC ESC with a bad current-sense offset can drive full current into a stationary winding and destroy the motor.
When FOC earns its keep
Camera and survey platforms. Torque ripple is vibration, and vibration is visible in the footage. The smoothness advantage is not cosmetic here, it is the product.
Heavy lift and precision work. At low disc loading and low throttle, harmonic losses are a larger share of total current, so the efficiency gain is real and translates into flight time.
Noise-sensitive deployments. Inspection, survey and any platform flying near people benefit from removing the sixth-harmonic whine.
Applications that need controlled torque rather than controlled speed. Because current maps linearly to torque in the d-q frame, FOC makes torque control and active braking straightforward in a way six-step does not.
A practical selection checklist
| Airframe / use | Default choice | Reason |
|---|---|---|
| Small acro or racing quad, high KV | Six-step | Burst response and electrical frequency ceiling |
| 5-inch freestyle, mid KV | Either, lean six-step | FOC gain is small at high average throttle |
| Cinema or survey multirotor | FOC | Torque ripple shows up as visible vibration |
| Heavy lift, long endurance | FOC | Part-load efficiency converts directly into flight time |
| Budget build, mixed motors | Six-step | No per-motor tuning, wide motor tolerance |
| Precision position or torque control | FOC | Linear current-to-torque mapping in the d-q frame |
Failure modes worth knowing
FOC
- Wrong motor parameters. Resistance and inductance feed the observer and the current loop. If they are off, the estimated angle lags and efficiency collapses before anything visibly fails.
- Current-sense offset drift. A small DC offset in a shunt amplifier becomes a false torque command. This is the failure mode that cooks windings.
- Observer divergence at handover. Usually a startup ramp that is too aggressive for the prop load.
Six-step
- Desynchronisation. At high electrical frequency under heavy acceleration the commutation can fall behind the rotor. The motor stutters and current spikes.
- Demagnetisation. Excessive current at high temperature can permanently weaken the magnets, after which the motor is simply down on torque forever.
- Zero-crossing misdetection. PWM noise on the floating phase shifts the detected crossing, which retards timing and costs efficiency without any obvious symptom.
Have questions about this article? Feel free to contact us at [email protected] — we’re happy to help!
Frequently Asked Questions
Is FOC always more efficient than BLDC?
No. FOC is usually more efficient, but the margin depends on the operating point. It is largest at part load and low speed, where harmonic current is a big share of total current, and smallest at full throttle near peak efficiency.
Why do racing quads still use six-step ESCs?
Because high-KV motors run at electrical frequencies that a cost-effective FOC loop cannot regulate cleanly, and because six-step responds to a throttle step with one commutation change instead of a converging observer plus a PI slew.
Can a BLDC ESC be converted to FOC with firmware?
Only if the hardware supports it. FOC needs current sensing, a faster MCU and a gate driver able to switch at the required PWM rate. Many six-step ESCs have none of those, so the answer is usually no.
What causes the whine that FOC removes?
Sixth-harmonic torque ripple from square-wave current, plus the switching frequency itself. FOC removes the low-order harmonic content, leaving mostly a higher-frequency residual that is easier to damp and less audible.
Do I need to tune an FOC ESC for each motor?
Generally yes. The current loop and the sensorless observer both depend on motor resistance and inductance, so parameters that suit one motor can cause a lagging angle or a rough start on another.
About Aomway
Aomway is a radio-frequency engineering company that designs and manufactures long-range FPV video transmitters, antennas, goggles and UAV communication links, and it publishes technical notes on propulsion, flight dynamics and embedded control. Its hardware is built for builders who need dependable links at range, and its documentation aims to explain the engineering behind each product.
