Six-step commutation method for brushless motors

The timing of commutation depends only on the position of the rotor. A relatively simple way is to use a photoelectric encoder disk as a position sensor, which is widely used in industry.

Secondly, Hall effect devices are used as position sensors, which can give high or low level outputs according to the different magnetic field direction distributions at different positions of the rotor. Generally, three Hall sensors are installed at different positions of the motor to measure the position of the rotor. This is the so-called "sensored brushless motor". It is worth mentioning that the electric speed controllers in car models and ship models mostly use the "sensored" method, because their motors need to be started, stopped, and reversed frequently, and the weight of the entire power system is not very particular, so it is more appropriate to use a sensored brushless motor electric speed controller. The non-sensor method omits the position sensor and uses the back electromotive force of the third phase that is not energized at a certain moment to estimate the position of the rotor. This makes the whole system lighter and simpler in structure. Its disadvantage is that it is more troublesome to start, and the controllability is poor when starting, and it becomes controllable only after reaching a certain speed.

Basic principle of sensorless commutation

How does the inductive commutation estimate the rotor position based on the back electromotive force of the third phase? As shown in Figure 3-1, during the AB phase power-on period, the C side of coil CC' cuts the magnetic force lines of the N pole in Figure (a) and generates a positive induced electromotive force, while in Figure (b) it cuts the magnetic force lines of the S pole and generates a reverse induced electromotive force. The situation of the C' side is similar. This shows that during the AB phase power-on period, if you measure the voltage on coil CC', you will find that there is a change process from positive to negative. Similarly, the other power-on conditions can also be analyzed in this way.

Figure 3-1

How does the inductive commutation estimate the rotor position based on the back electromotive force of the third phase? As shown in Figure 3-1, during the AB phase power-on period, the C side of coil CC' cuts the magnetic force lines of the N pole in Figure (a) and generates a positive induced electromotive force, while in Figure (b) it cuts the magnetic force lines of the S pole and generates a reverse induced electromotive force. The situation of the C' side is similar. This shows that during the AB phase power-on period, if you measure the voltage on coil CC', you will find that there is a change process from positive to negative. Similarly, the other power-on conditions can also be analyzed in this way.

Figure 3-2

During the AB phase energization period, the induced electromotive force of CC' will completely change direction, which is the so-called "zero crossing point". At t0 in Figure 3-2, it is the situation when the AB phase energization just begins, and the equivalent circuit diagram of the induced electromotive force generated by CC' is shown in Figure 3-3 (a). At t1, it is the situation when the AB phase energization is about to end, and the equivalent circuit diagram of the induced electromotive force generated by CC' is shown in Figure 3-3 (b).

Figure 3-3

It should be noted here that during the AB phase power-on period, not only the coil CC' generates an induced electromotive force, but AA' and BB' also cut the magnetic lines of force and generate an induced electromotive force. The direction of the electromotive force is opposite to the direction of the external power supply, so it is called "backward induced electromotive force" (BEMF). The back electromotive force generated on the coil windings AA' and BB' is very large, and the sum of the two is almost slightly less than the power supply voltage (assuming it is 12V). The equivalent resistance of the coil winding itself is very small (about 0.1 ohms). If the back electromotive force is not large, the terminal voltage will be loaded on the equivalent resistance of the coil winding, which will generate a huge current and the coil will burn out. For the sake of convenience, let's assume that at the rated speed, AA' and BB' each generate a back electromotive force of 5.7V, then they are connected in series to generate a back electromotive force of 11.4V, then the voltage applied to the equivalent resistance is (12-11.4)=0.6V, and the current passing through the winding AB is 0.6/(2*0.1)=3A. Similarly, since the structure of each winding is the same and the speed of cutting magnetic lines of force is the same, the coil CC' should also generate an induced electromotive force of about 5.7V. Since the midpoint potential value is always 6V, the induced electromotive force generated by the CC' coil can only be superimposed on the basis of the midpoint 6V potential as the reference point. It is still assumed that an induced electromotive force of 5.7V will be generated on CC' at the rated speed. Then at time t0, if the voltage at point C is measured, its value should be 6+5.7=11.7V; at time t1, the voltage value at point C should be 6-5.7=0.3V. In other words, during the AB phase power-on period, as long as the voltage of the C phase lead of the motor is monitored, once it is detected to be lower than the midpoint voltage, it means that the rotor has rotated 30° to the position between t0 and t1, and it only needs to wait another 30° to switch phases. If the MCU of the ESC is fast enough, continuous AD sampling can be used to measure the C phase voltage, but it will waste CPU resources because most of the AD values ​​collected are useless. We only care about when it is lower than the midpoint voltage. An analog comparator can be used to monitor the zero-crossing signal. Once the C phase output voltage is lower than the midpoint voltage, the comparator can immediately sense it and give a falling edge at the output. Similarly, when the motor is in the AC phase, the B phase output voltage is monitored; when the motor is in the BC phase, the A phase output voltage is monitored. Continuing forward, when the motor starts to enter the BA phase, the C phase output voltage will be in a lower state at first. When the zero-crossing event occurs, the C phase output voltage will exceed the midpoint voltage. At this time, the comparator will sense and output a rising edge. The following CA and CB phase power-on conditions are similar and will not be repeated.

Delayed commutation

According to the previous article, we know that at the zero crossing point of phase C, the rotor has only rotated 30°, and it needs to rotate another 30° before the phase can be changed. How do we know how long it takes to rotate the remaining 30°? A relatively simple approach is to approximately assume that the rotor speed is basically constant in this small range of 0°~60°: the first half of the time from the start of power-on of phase AB to the detection of the zero crossing of phase C is basically equal to the second half of the time. So just use the timer to count the time interval T1 in the first half, wait for the zero crossing event to occur, and then wait for the same amount of time before changing phases. Of course, a more direct method is to directly change phases instead of delaying 30° after detecting the zero crossing. This method is certainly feasible, but it will lose a little efficiency, and the torque will also be reduced under the same conditions.

Demagnetization phenomenon

In the actual commutation process, for example, when the AB phase switches to the AC phase, the current of the B phase suddenly decreases (it will not disappear suddenly, and will continue to flow for a period of time until its own energy is exhausted). The self-inductance of the coil will become an electromotive force generator during the continuous flow period, and the direction is opposite to the original, and it is superimposed on the midpoint. As shown in Figure 3-4, the potential of the B terminal is higher than the midpoint potential. Note that the induced electromotive force of the B coil in Figure (a) is generated by the conductor cutting the magnetic lines of force, while the continuous flow electromotive force in Figure (b) is generated by the inductance of the B coil itself (its magnitude is higher than the magnitude of the induced electromotive force cutting the magnetic lines of force).

When the energy of coil B is exhausted, the cutting of magnetic lines of force becomes the main factor that dominates the induced electromotive force of phase B again, so the potential of terminal B will be lower than the midpoint potential at this time, which is the so-called "demagnetization phenomenon". Figure 3-5 shows the demagnetization phenomenon that occurs during phase switching.

Figure 3-4

When the energy of coil B is exhausted, the cutting of magnetic lines of force becomes the main factor that dominates the induced electromotive force of phase B again, so the potential of terminal B will be lower than the midpoint potential at this time, which is the so-called "demagnetization phenomenon". Figure 3-5 shows the demagnetization phenomenon that occurs during phase switching.

Figure 3-5

Due to the existence of demagnetization events, the B terminal will generate two overshoot zero-crossing events during the AC phase power-on period (that is, the analog comparator will capture two zero-crossing signals). The first one is invalid, and the second one is a valid zero-crossing signal. The following strategy can be used to filter out the interference of demagnetization phenomenon in software. After the comparator captures the zero-crossing signal, check whether the current time from the last phase change is greater than the delayed phase change time. If it is greater, it is a valid zero-crossing signal, and if it is less, it is a zero-crossing caused by demagnetization.