Paper UDC 621.313.3-573: 621.316.71:681.532.8:621.382 Brushless Motor without a Shaft-Mounted Position Sensor By Tsunehiro Endo Fumio Tajima Member Member Kenichi Iizuka Member Summary Hideo Uzuhashi Non-member A brushless motor with a microcomputer-based speed control system, but without a shaft-mounted position sensor, is described. The motor terminal voltages are used to obtain rotor position information. Speed control in the system is carried out through chopper control of the inverter itself, much as in chopper-fed DC motors. Various processes, including the speed control process, are controlled by developing the appropriate software. The limits of position sensing under a motor load are also described. 1. Introduction Brushless motors are superior to inverter-fed induction motors in regard to their noise and efficiency and, therefore, they have recently begun to be used in variable speed drive systems. However, brushless motors need such rotor position sensors as shaft encorders, Hall effect transducers, etc., in order to obtain necessary rotor position information. A position sensor is usually mounted on the motor shaft, but this tends to reduce a system's overall ruggedness. If motor terminal voltages are used to supply rotor position information, the shaft-mounted position sensor can be eliminated(1). As is well known, brushless motors are generally similar to DC motors in regard to their speed control properties. Accordingly, the PWM inverter that is usually used in the speed control system of induction motors can also be used in the speed control system of brushless motors. This paper presents a brushless motor control system, which includes a voltage source inverter, in which motor terminal voltages are used to obtain position sensor signals and speed control is performed by an inverter chopper. The following items are described in Tsunehiro Endo, Fumio Tajinaa, Kenichi Iizuka, & Hideo Uzuhashi are with Hitachi, Ltd. Manuscript received Aug. 12, 1983, revised Sept. 8, 1984. this paper. (1) The hardware and software structures of the brushiess motor control system. (2) A method of controlling the inverter output voltage. (3) A method of detecting rotor position using motor terminal voltages and a position sensing limit for the motor load. (4) Experimental results. 2. Description of the Control System 2.1 Hardware structure A brushless motor control system without a shaftmounted position sensor is shown in Figs. I and 2. The brushless motor consists of a synchronous motor with permanent magnet excitation, an inverter with a voltage source and 120 K conduction, a rectifier, a position sensor circuit in which position sensor signals are obtained from term inat'voltages, a control circuit, and a drive circuit. The inverter itself has an output voltage control function for adjusting motor speed. The control circuit consists of a microcomputer and a custom LSI 1/0 processing unit which processes input/output data for the inverter and operates as a peripheral circuit of the microcomputer"', as shown in Fig. 2. The development of appropriate software allows various processes, including speed control, to be
Fig. 2. Block diagram of control circuit. carried out with the above mentioned 2.2 Structure of software hardware. Fig. 3 shows a block diagram of the software for the brushless motor control system. In the figure, IRQ and SWI are entrances for different programs started respectively by interrupt request signals from the I/O processing unit. to the microcomputer and by a software interrupt instruction. The software consists of an operating system, which manages different programs, and various tasks required to operate the brushless motor. The operating system is composed of a restart process, IRQ controller, task dispatcher, etc. Four processes are controlled by the IRQ controller, examples of which are a process which interrupts motor operations or suppresses an instantaneous motor current while motor operations continue thus protecting the inverter from excessive motor current and a process which performs a periodic operation to obtain motor speed from position sensor signals. The processes that protect the inverter from destruction by an over voltage or over current are performed by over-voltage and over-load tasks. In the speed control task, the inverter output voltage is determined by the motor speed obtained from the operating system and a speed command from a command reading task. One of the remarkable features of this brushless motor control system is the motor start task which allows the inverter to be operated as a separatelycontrolled inverter until the motor reaches a speed at which position sensor signals become available, as shown in Fig. 4. During the motor start task operation, the inverter drive signals are generated in the microcomputer although there are also position sensor Fig. 4. Operation pattern at motor start. signals. The frequency of these drive signals is accelerated by degrees. In other words, the time during 60 electrical degrees is calculated for a desired inverter output frequency, and the inverter current path modes are changed whenever the calculated time, which is shortened by degrees, elapses. When the position sensor signals become available, the microcomputer changes the inverter operation to selfcontrolled if the inverter current path mode of the position sensor signals coincides with this mode or the next one of the drive signals generated in the microcomputer. The drive signals are decided according to the position sensor signals during the inverter
self-controlled operation. 3. Speed Control Method 3.1 General considerations The characteristics of brushless motors are similar to those of DC motors. Accordingly, the speed of a brushless motor can be controlled by adjusting motor terminal voltages. In the present control system, terminal voltages are adjusted by operating the inverter in the same manner as a PWM inverter in induction motors. The equivalent circuit for the brushless motor during one. of six current path modes (60 electrical degrees) is shown in Fig. 5. Two transistors are placed between the motor terminals and the DC power supply. Motor terminal voltage can be adjusted by operating either transistor as a chopper with the other one in the on-state, as in chopper-fed DC motors. In order to prevent interruptions in motor current during periods of 120 electrical degrees, the transistor located opposite the chopper action transistor (either transistor shown by the broken line in Fig. 5) is switched on and off during two current path modes. This complementary switching methed facilitates speed control over a wide range of motor loads. 3.2 Actual control method Variovs chopper control methods are possible, depending on which of the six transistors in the inverter is the chopper. In the present control system a transistor acts as a chopper only during the latter half of the transistor's 120 electrical degrees conduction period. Fig. 6 shows signals and waveforms obtained for the terminal voltage and the motor current, where the terminal voltage is equal to the transistor collectoremitter voltage on the minus arm side. Drive signals for the inverter (Fig. 6 (c) ) are generated by the I/O LSI according to position sensor signals and a chopper signal. Speed control is accomplished by regulating a chopper signal duty factor. The duty factor is determined by writing duty factor data for one of the many registers in the I/O LSI in software. The motor current waveform, for the case in which chopper signal off-time is longer than the interval represented by the motor winding's time constant and the motor voltage, and drive signal and terminal voltage waveforms are shown in Fig. 7. In the above case, when chopper signal off-time is relatively long, (a) drive signals for A+ and A-, (b) motor current for phase A, (c) teminal voltage for phase A. Fig. 7. Waveforms for the case in which a chopper signal's off-time is relatively long. motor current flows alternately in positive and negative directions even during 120 electrical degrees periods, because of the effect of the complementary switching method described above. 4. Position Detection 4.1 Relation between phase voltage and commutation In order to obtain an average motor torque that is as high as possible, the commutation leading angle must be nearly zero. That is, commutation should occur at the instant when two of three motor phase voltages are equal. On the one hand, such a commutation event is equal to zero crossings of the integrated motor phase voltages. For example, in the case of the phases of the B+ and B- transistors in the inverter at turn-on, va is equal to vb and vc integrated is equal to zero, where va, vb and vc are the voltage of the A,B and C phases, respectively. If the motor phase voltages are then obtained from the terminal voltages and integrated, it becomes possible to detect rotor position. 4.2 Phase voltage detection An additional center for phase voltage detection is made available by connecting three resistances on
Fig. 8. Circuits and waveforms illustrating method of detecting phase voltage. (a) without chopping, (b) with chopping, (c) chopper signal. Fig. 9. Terminal voltage waveforms. lines led from the motor terminals in a start-shaped configuration as shown in Fig, 8 ( a) 3. If the system is symmetrical, the two centers of resistance and the motor windings will have the same potential. The voltages corresponding to the motor phase voltages are obtained by comparing the terminal voltages with the voltage at the center of the star-shaped configuration of resistors. These corresponding voltages contain phase voltages and motor winding impedance drops. The phase relationships between the terminal voltage V and the center voltage Vm are illustrated by the waveforms shown in Fig. 8 ( b), in which no chopping occurrs. The waveform imposed on terminal voltages by a voltage source inverter are nearly trapezoidal in shape and contain no visible evidence of motor voltages. The only information required for detecting rotor position is the zerocrossing information contained in the phase voltages. No amplitude information is necessary. As explained below, the necessary information can be obtained without interference from the above mentioned motor winding impedance drops. The equivalent circuit for the current flow period during which the transistors C+ and B- are in the on-state (mode (I) in Fig. 8(b)) is shown in Fig. 8 (c). The following equations can be obtained from the figure. voltage Va and the center voltage Vm is equal to the phase voltage va. That is, the phase voltage can be obtained uninfluenced by the motor winding impedance drops. Assuming that the three phase voltages va, t b and vc are balanced and that the winding impedances are equal to each other, the terminal voltage is equal to one half the DC voltage Ed in the phase in which the terminal voltage V is equal to the center voltage Vm. This also applies to mode (II) in Fig. 8(b), and to modes (I) and (H) with chopping when there is no motor current flowing in those modes. Therefore, position sensor signals are obtained by phase-shifting the signal generated in a comparison of the terminal and center voltages in modes (I) and (II), by 90 electrical degrees, as shown in Fig. 8(b). Fig. 9 shows the terminal voltage waveforms in detail, including the impedance drops. 4.3 Position sensor circuit In order to obtain position sensor signals, it is necessary to phase-shift the phase voltages obtained from the teminal voltages by 90 electrical degrees as indicated above. However, the terminal voltages contain many transistor commutation and chopper caused transients. Consequently, it is better first to phase-shift and filter the terminal voltage in order to attenuate these transients. A block diagram of the proposed position sensor circuit is shown in Fig. 10. The position sensor circuit consists of simple high-and low-pass filters, resistors connected in a star-shaped configuration, and comparators. Fig. 11 shows the waveforms characteristic of the voltages present at selected points in the position sensor circuit when there is no chopper control. The high-pass filters attenuate the DC components in the terminal voltages and the low-pass filters attenuate the transients and shift the terminal voltage back by 90 electrical degrees. The center voltage Vnf that
Fig. 14. Overlap angle and motor current at instant of commutation versus commutation leading angle. angle depends on the motor winding time constant, the motor current at the instant of commutation and the commutation leading angle(1). Fig. 14 shows the relation between the overlap angle p and the motor current at the instant of commutation Iao and the commutation leading angle S for a given motor load and speed(4). As shown in the figure, for a commutation leadig angle smaller than about 25 electrical degrees, the overlap angle p and the motor current at the instant of commutation Iuo decrease as the size of the commutation leading angle increases. Consequently, even if the fundamental components of the terminal voltages are shifted forward by the commutation transients, the overlap angle tends to decrease and the commutation leading angle does not continue to increase indefinitely. But for a commutation leading angle larger than about 25 electrical degrees, the overlap angle p and the motor current I,oo increase as the size of the commutation leading angle increases. As a result, a positive feedback loop is produced and the size of the commutation leading angle continues to increase until, finally, the motor stops. In order to extend the upper limit of the motor load, either comparators with a hysteresis function or six low-pass filters can be used. 5. Experimental Results The experimental setup used a brushless motor with a rated output of 1.2 kw. An 8 bit microcomputer based on the 6802 and the I/O LSI are responsible for system control functions. The filtered terminal voltage, motor current and motor speed immediately after motor start with a motor load of 130% are shown in Fig. 15. In this example, when the motor speed has reached 250 rpm, (a) position sensor signals, (b) filtered terminal voltage, (c) motor speed, (d) motor current. Fig. 15. Motor start characteristics. Fig. 16. Speed versus torque. inverter operation is changed from separately-to selfcontrolled. The maximum value of the motor current is limited during separately-controlled operation. Fig. 16 shows the speed/torque characteristics for various duty factors. The broken lines represent the characteristics in the case without complementary chopping, where the motor speed increases sharply as torque decreases. Fig. 17 shows the waveforms characteristic of the voltages present at selected points in the position sensor circuit and the motor current waveform in the case where the duty factor equals 0.5. The voltage waveform for the differential between the filtered terminal and center voltages is shown in Fig. 17(c).
(a) terminal voltage: 50 V/div, (b) filtered terminal voltage: 0.2 V/div, (c) differential between filtered terminal and center voltage : 0.2V/div, (d) motor current: 5A/div, horizontal: 2ms/div. Fig. 17. Wave forms. results shown in Fig. 14. 6. Conclusion (a) when additional coils are connected, (b) when no additional coils are connected. Fig. 18. Relation between commutation leading angle and limit of detecting position. Fig. 18 shows results for examinations of the relation between the commutation leading angle and the position sensing limit. In the figure, variations in the commutation leading angle and the overlap angle are shown plotted against motor torque. The (a) curves plot the case for additional coils intentionally connected in series to the motor windings for the purpose of understanding exactly the position sensing limit phenomenon. As is evident them the motor stops when the commutation leading angle reached a size of about 25 electrical degrees. This confirms the calculated A brushless motor control system in which terminal voltages are used to detect rotor position has been developed. All the system control functions were assumed by a microcomputer and I/O LSI. A chopper control method was adopted in the, inverter to control motor speed. A complementary switching method facilitated speed control over a wide range of motor loads. The position sensor circuit consisted of only filters and comparators and was therefore, very simple. In the proposed position sensor circuit, the commutation leading angle increased in size as the motor load increases. It was found that motor speed could not be controlled when the commutaton leading angle was larger than about 25 electrical degrees. References (1) H. Le-Huy, A..Jakubowicz & R. Perret: "A self-controlled synchronous motor drive using terminal voltage system," IEEE Trans. Industr. Applic., IA-18, 46 (1982) (2) S. Morinage :"Microprocessor control system with I/O processing unit LSI for motor drive pwm inverter," in Conf. Rec. 1981 IEEE Ind. Applic. Soc, Annu. Meeting, p. 1197 (3) P. Ferrais, A. Vagati, & F. Villata: "P. M brushless motor drives:- a self-commutation system without rotor-position sensors," Proc, of the Ninth Annual Symp. on Incremental Motion Control Systems and Devices, p. 305 (1980) (4) S. Miyairi & Y. Tsunehiro: "The Analysis of a Commutatorless Motor as a DC Motor and its Characteristics" J. IEE of Japan (in Japanese), 85, 1585 (1965)