BLDC Control method selection

Hello, and thanks for your question!

Trapezoidal control is preferred for high speed for a few reasons:

1. Less computational complexity - FOC requires trigonometric functions to be calculated every loop cycle. Sensorless FOC also requires sensorless position estimation calculations to be done every cycle. High speed requires these functions to be performed at a very, very high rate. A one pole pair motor at 100,000 rpm would be spinning at 1.7kHz electrical frequency (based on the equations from this FAQ), 1.7kHz is already quite high for your typical FOC. A two pole pair motor would be 3.3kHz, is above the speed we see FOC normally being implemented. If you develop a custom FOC algorithm running on a fast enough MCU, you can make it work - however it becomes expensive very quickly.
2. Current sensing is related to computational complexity is. In traditional low-side current sensing, you need to synchronize your ADC sampling with the low-side MOSFET ON. With high output PWM frequency and high duty cycles, there is not very much time to sample the current in each phase. You can get around this using all three current sense (instead of just two) or implement in-line current sense amplifiers, but this adds cost. Single shunt techniques are better for high speed but result in worse performance at low speeds.
3. Higher applied voltage to the motor - FOC does not apply full duty cycle to the motor to achieve a sinusoidal profile. This results in a lower maximum speed than trapezoidal unless you implement overmodulation, which is a trapezoidal-type of extension to FOC.
4. The majority of high-speed motors are wound trapezoidally, meaning driving them with a sinusoidal or FOC control will not result in audible noise or efficiency gains because their back-EMF profile is trapezoidal.
5. Higher switching losses are present in FOC versus trapezoidal control due to three phase switching simultaneously (only one phase is switching in trapezoidal control)
6. Sensorless is much easier to implement in trapezoidal control compared to sine or FOC because you can always open up a Hi-Z window to directly measure the back-EMF. Sine/FOC do not have a Hi-Z window and so they require more complicated estimation (V&I measurement & calculation) of the back-EMF where information about the motor parameters will be required.

Thanks,

Matt

Let me address your questions:

1.Why does the use of FOC control for the back EMF trapezoid not reduce noise and efficiency? In my understanding, the back-EMF is trapezoidal, which makes it easier to detect the zero-crossing point, and it seems to be able to control FOC

• To achieve maximum efficiency & minimum audible noise, you need to drive a waveform similar to the back-EMF profile. While you certainly can drive a trapezoidal motor with a sinusoidal/FOC profile, it will not give optimal results like with a sinusoidal motor.

2.For my customer's motor, it has low inductance and low resistance. If the FOC control is performed, the over-current protection will be triggered when starting. I want to know if the starting current is very large due to low inductance and low resistance? So can the use of trapezoidal wave control solve the problem of large starting current? why?

• For a very low inductance/resistance motor. the motor will have a very large start-up current. To operate this motor properly you need a very high PWM frequency to properly control the motor current - if the PWM frequency is too high the FOC algorithm will have difficulty sensing the current & calculating quick enough. Trap algorithms normally do not have as muc PWM frequency dependence. You also need to very carefully control the open loop spin-up of the motor until you reach closed loop, but this does not have any dependence on trap or FOC control.

3., And the back-EMF coefficient Ke I measured is also very small.If the BEMF of the motor itself is very small, will it make it impossible to use the FOC control mode (due to the inability to change the direction normally)

• Since the Ke is very, very small, it means this motor will be very difficult to control at low speed. For example in a hypothetical 100,000 rpm motor, if we operate at 8,000 rpm that would be 8% of max speed. You need to accurately control an 8% duty cycle PWM to get the motor spinning in trap. For FOC you need your average duty cycle to be 8% but for the applied sine profile you need to further divide down the duty cycle from 8% - this leads to a control duty cycle that is too small.

4.In the FOC control, I know that the BEMF constant and motor phase resistance is very important ,But I don't know how they affect the work of the motor, can you explain it to me in detail?

• In sensorless FOC or sine, we need to calculate/estimate the back-EMF because there is no zero-crossing window open to directly measure it. Trapezoidal control has these zero-crossing windows due to the waveform shape. To calculate/estimate the back-EMF, you need to do the following calculation (see the DRV10987 datasheet for how it is done on this device). Motor resistance as well as the back-EMF constant are needed to calculate the back-EMF if you know the motor current, supply voltage, and motor speed.
• 

Thanks,

Matt

You're welcome!

Please verify the answer to this post for tracking & search!