Understanding the methods of trapezoidal control of BLDC motors

onsemi, which supplies a wide range of MOSFETs and power transistors, explains the factors to take into account when designing sensored or sensorless motor-control systems.

Trapezoidal control, also called six-step or 120-block commutation, is one of the simplest methods for controlling a brushless dc (BLDC) motor. The basic concept of trapezoidal control is to apply square-wave currents to the motor phases that align with the BLDC motor’s trapezoidal back EMF profile, resulting in optimal torque generation. 

The six-step name reflects the inverter MOSFETs’ assumption of just six on/off state combinations, resulting in six possible stator field orientations within the rotor magnetic field’s plane of rotation. The six possible inverter states must follow a specific sequence depending on the desired direction of rotation of the motor, so that the stator and rotor field orientation arrangement produces maximal torque. 

Rotor position feedback, which determines the proper commutation timing, is generally achieved in one of two ways: 

  • With Hall sensors mounted on the motor, a sensored configuration
  • Or by sensing the back EMF of the motor phases while rotating, a sensorless configuration

Fig. 1: Timing diagram of Hall sensor commutation 

Sensored trapezoidal control does not require any voltage or current feedback signals for operation. Instead, it uses position feedback from Hall sensors to determine the correct sequence for energizing the phases of the motor. Hall sensors are mounted on the motor and sense rotor position through the Hall effect caused by the rotation of the rotor’s permanent magnet. Sensored trapezoidal control is generally easier to implement and allows for proper commutation even at start-up, because the rotor position information is present even at zero speed. A typical Hall sensor/phase-voltage commutation timing diagram is shown in Figure 1.

Sensorless trapezoidal control uses the back EMF generated by the motor’s rotation to determine the correct motor commutation sequence. For trapezoidal control, only two motor phases are energized at a time. Because no current flows in the non-energized phase, the back EMF of that phase can be directly sensed during that time. 

For BLDC motors, the back EMF profile is trapezoidal: during the non-energized time frame, the phase back EMF is either linearly increasing or decreasing. Most back EMF position feedback techniques for trapezoidal control rely on a technique involving back EMF zero-crossing detection (ZCD). The back EMF is monitored to determine the moment when it crosses a reference point, either the neutral motor voltage or half the dc bus voltage. 

One important drawback of sensorless control is that, because the magnitude of back EMF is proportional to rotary speed, the motor must rotate at a minimum speed to produce a strong enough back EMF signal for proper rotor position sensing. This means that a start-up mechanism is required to kick-start the motor until it achieves sufficient rotary speed.

While sensored trapezoidal control is easier to implement, it entails a higher bill-of-materials due to the requirement for Hall sensors. It also requires more wiring from the motor, which might not be possible in some environments. 

Sensorless control is more complex, must be tuned for specific loads or operating conditions, and might have difficulty starting up under heavy loads. Sensorless control is well suited to applications where the load profile is well known, however, and/or the load profile increases with speed, such as a fan.

Figures 2 and 3 illustrate examples of sensored and sensorless trapezoidal motor-control systems.

Fig. 2: Block diagram of a sensored trapezoidal motor-control circuit

Figure 2 shows the required signals for sensored trapezoidal control. Hall sensors must be powered externally and are generally open-drain outputs, though they can be push-pull. When placing Hall sensors at 60 or 120 intervals on the motor, determining the correct commutation sequence requires the Hall signal sequences to be decoded differently. It is important to refer to the motor manufacturer’s Hall sensor commutation timing diagrams to determine the correct Hall sensor-phase commutation sequence. If no timing diagram is available, the correct commutation sequence can be determined empirically. Hall sensor signals can also be noisy and might benefit from hardware or software filtering.

Fig. 3: Block diagram of a sensorless trapezoidal motor-control circuit

Figure 3 shows the signals required for sensorless trapezoidal control. Back EMF ZCD can be performed either in firmware or hardware. For software ZCD, back EMF detection requires an ADC with a minimum of four inputs, 3x phase voltage, Vin/2, or neutral reference, with all voltages divided to adjust to the controller ADC’s full-scale range. 

Hardware ZCD implements comparators to compare the phase voltages to the reference points and provide the zero-crossing signals directly to the controller via GPIO pins. If using hardware ZCD, the recommendation is to filter the divided phase voltages and implement comparator hysteresis to prevent glitches in the ZCD signals due to non-ideal and noisy back EMF signals. Note, however, that this might limit the maximum rotary speed available due to the delay caused by filtering. 

Digital filtering can also be implemented in software ZCD: this is advantageous because motor speed can adjust filtering operation, for instance to implement heavy filtering at lower speeds and reduced filtering at higher speeds. 

While using a comparison to half the dc bus voltage is possible, a comparison to the motor neutral (virtual or actual) is preferred as it is more tolerant of phase imbalance. If the motor neutral is not accessible from the motor, a simple Wye-connected resistor network connected to the motor phases can recreate virtual neutral.


Additional Design Considerations

Over-current protection (OCP) can be implemented in hardware, software, or both. Typically, hardware-based OCP will provide faster response, but software-based OCP has more flexibility. Another limitation of software-based OCP is that the full-scale current measurement range of the ADC limits the maximum trigger point. 

A combined hardware/software implementation might implement a latching OCP to mitigate catastrophic hard faults quickly, and use software-based OCP for dynamic events such as cycle-by-cycle phase current limiting.

Over-voltage protection (OVP): in certain applications, such as when regenerative braking might cause excessive voltages on the dc bus, it might be necessary to implement OVP in hardware by diode-clamping or a crowbar circuit. 

Software-based OVP can also be implemented by monitoring the dc bus, and protecting the motor from potentially damaging voltages above the rated voltage of the motor by disabling the inverter output.

Over-temperature protection (OTP): it is generally recommended to monitor the inverter MOSFET and board temperature when using any motor-control method, especially when the system is subject to varying ambient temperatures, or in the event of failure of the cooling system failure. 

For example, PWM duty cycle limits can be dynamically reduced as temperature increases, and thermal monitoring can also help to determine component degradation over time. 

MOSFET selection: selecting the best MOSFET or switch is important for any motor-control system and should be specifically tailored to system requirements. The output inverter stage influences the system’s overall efficiency, and incorrect MOSFET selection could result in substantial performance degradation and even catastrophic system failure. 

It is beyond the scope of this design note to discuss all specific factors related to proper switch selection. onsemi offers a wide variety of highly efficient MOSFETs and transistors for motor-control application, in various voltage ranges.