METHOD OF CONTROLLING A BRUSHLESS DC MOTOR

Abstract

A method of controlling the commutation of a brushless direct current motor includes providing sensors which provide a variable output dependent on rotational angle or the relative position of the stator and rotor of the motor. Output from the sensors is sampled at a time between a past commutation event and the next commutation event to be implemented. An angular position between the rotor and stator is determined at the time. The time of the next commutation event is determined based on the next commutation angle, motor speed, and the determined angular position.

Claims

1-5. (canceled)

6. A method of controlling commutation of a brushless direct current motor, said method comprising: a) providing one or more sensors, said one or more sensors being adapted to provide a variable output dependent on a rotational angle of the brushless direct current motor or a relative position of a stator and a rotor of the brushless direct current motor; b) sampling the variable output from said one or more sensors at a time t.sub.n between a past commutation event C.sub.m1 and a next commutation event C.sub.m to be implemented; c) determining an angular position .sub.n between the rotor and the stator at time t.sub.n at said time since the past commutation event from step b); and d) determining a time T.sub.m of the next commutation event C.sub.m based on said time t.sub.n instantaneous motor speed , and output from step c).

7. A method as claimed in claim 6, wherein in step d), the time T.sub.m is based also on a next commutation angle.

8. A method as claimed in claim 6, wherein said time T.sub.m is determined from the equation:
T.sub.m=t.sub.n+(C.sub.m,angle.sub.n)/ where .sub.n is the angular position at time t.sub.n calculated from step c), is a determined or modelled rotational speed, and C.sub.m,angle is an angle of the next commutation event.

9. A method as claimed in claim 6, wherein a speed of the brushless direct current motor is determined by determining an angular change in rotation of the brushless direct current motor from sensor output at two or more known time instances (tn, tn1).

10. A method as claimed in claim 6, where a speed w of the brushless direct current motor is determined from a model of the brushless direct current motor.

11. A method as claimed in claim 1, wherein: said time T.sub.m is determined from the equation:
T.sub.m=t.sub.n+(C.sub.m,angle.sub.n) where .sub.n is the angular position at time t.sub.n calculated from step c), w is a determined or modelled rotational speed, and C.sub.m,angle is an angle of the next commutation event; and the rotational speed w of the brushless direct current motor is determined by determining an angular change in rotation of the brushless direct current motor from sensor output at two or more known time instances (tn, tn1).

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The present invention is now described by way of example with reference to the accompanying drawings in which:

[0015] FIG. 1 shows common BLDC motor control;

[0016] FIGS. 2a, 2b, 2c and 2d show the prior art methodology for determination of commutation time (event);

[0017] FIG. 3 shows specifically an example of the hardware/functional blocks used to examples of the invention;

[0018] FIG. 4a shows a phasor or angular diagram showing computational timing of a BLDC motor and FIG. 4b shows the timeline therefor.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] The present invention provides a method to control BDC motors using simple linear sensors whilst but achieving optimal commutation timing.

[0020] FIG. 1 shows common BLDC motor control apparatus including a schematic representation of the BLDC motor 1 itself. The rotor of the brushless motor is equipped with a permanent magnet 3, creating an excitation field. The stator here consists of three phase windings: 2a, 2b, and 2c, supplied with current from an inverter 4. The commutation is handled by the inverter and its controlling logic circuitry 5. Several commutation patterns are used in industry, most of the time so called sine commutation and block commutation. In known techniques, the raw position of the rotor is determined by a linear angular position sensor 6, processed by a signal processing block 7 and input to the control logics.

[0021] In block commutation, the commutation pattern changes at every block commutation event, in order to change the direction of the stator induced magnetic field. A high timing precision of the commutation event is required. Many angular sensors provide magnetic field raw values requiring signal post processing in order to determine the correct angular position. This processing takes time, making the determination of the commutation event imprecise. As the A/D conversion is on a typical A/D converter is not continuous there is also an undesirable time discretization leading to further error. In addition, scheduling jitter may heavily affect the time precision of the commutation event at high motor speeds. Impacts are torque ripple and lower efficiency.

[0022] FIGS. 2a, 2b, 2c and 2d show the prior art methodology for determination of commutation time (event). FIGS. 2a and 2b show the output signals 20 from one or more sensors (e.g. two sensors), indicative of the rotational position of the motor. FIG. 2c shows the timeline of the processing of these signals. The signals are sampled at intervals e.g. periodically at time points 21. The signals are then converted to digital signals during time period 22 and processed during time periods 23 to determine the commutation timing. FIG. 2d shows an example of the commutation event timeline. Arrow A shows the optimal commutation timing event. However, this is not calculated (or rather able to be implemented) until the time indicated by the arrow B, due to sampling being late, as well as conversion and processing times. In order to overcome previously described issues, commonly a sensor package integrating the position processing and providing timing signals is used. Sometimes the processing speed of the logics is increased in order to limit the negative effects of jitter. Nevertheless, at high speeds, a reasonable dimensioned microcontroller system will always deal with commutation errors.

[0023] Aspects of the invention provide methodology to optimize timing precision of the commutation event in a very cost efficient way. In other aspects, the methodology may be implemented using microcontroller hardware.

[0024] FIG. 3 shows a block diagram of the hardware/functional blocks which may be used to implement examples of the invention. The entire signal chain from angular position sensor to the actual commutation in the inverter is optimized for the best possible commutation timing, while making use of an inexpensive linear output angular position sensor. FIG. 3 shows specifically an example of the hardware/functional blocks used to examples of the invention. Analog signals 12 provided by a sensor 11 are fed in the A/D converter module 13 of a microcontroller 112. If several analog signals are to be sampled, different options are possible: they can be sampled simultaneously with distinct A/D converters, or converted sequentially, potentially with a fixed delay. Simultaneous sampling simplifies the position signal logicbut this invention can be used with different setups as long as the position signal logic can calculate angle/timestamp pairs to be used for extrapolation.

[0025] The converted values of the analog signals are shown by reference 14a and the timestamp of the sampling event 14b are provided to the position signal processing logics 15a and commutation logic 15b. The output of commutation logic 15 b is a timestamp 17 which is input to program a hardware timer 19 to provide an event 110 at T.sub.m. Based on this event the inverter logic block 111 is provides commutation control signals to the inverter for the next commutation table entry (either in hardware or via interrupt or DMA). The problem of jitter and meeting the high precision timing of the block commutation event is solved by using the hardware timer module. A time stamp is sent to the commutation controller/inverter logic block 111 from block 15a/15b.

[0026] The precision problem of the position and speed estimation due to runtime and processing delays (especially at high motor speeds) is solved by doing processing and calculations in a time-based fashion on the processing sensor data at a (relative) known time in the past. From signal data and past commutation data provided at a previous times (e.g. with a corresponding, known times/time-stamps) the next commutation timing/event is determined.

[0027] The time stamp based calculation allows in addition lowering the CPU load at high speeds, because processing time and commutation angle error are uncorrelated.

EXAMPLE 1

[0028] Methodology of aspects of the invention will now be described with reference to FIGS. 4a and 4b. FIG. 4a shows a phasor or angular diagram (with reference to electrical revolution) of a BLDC showing a series of ideal temporally arranged (clockwise) ideal commutation events C.sub.0 C.sub.1 C.sub.2 C.sub.3 C.sub.4 C.sub.5. The angle a can be regarded as a relative angle between rotor and stator. The angle illustrates the angular interval between commutation events, i.e. commutation times. In the example this occurs at fixed intervals/relative positions of 60. FIG. 4b shows a plot 30 of motor position, in terms of angular displacement against time. In the figure, it is assumed the previous commutation event occurred at Tm1 at corresponding commutation event C.sub.0(m1). The correct next commutation event/time is shown at time T.sub.m and is referred to as C.sub.1(m).

[0029] The sensor outputs from one or more sensors are sampled at a time t.sub.n, to determine the angular position at the corresponding time-point e.g. .sub.n. So in other words, the position signal is used (e.g. by logic circuitry) to computes the measured actual motor position .sub.n (see FIG. 3) at a past time-point t.sub.n. A timestamp (representing t.sub.n may be determined by e.g. A/D conversion logic or A/D conversion trigger logic circuitry, and may originate from e.g. a hardware timer in the A/D converter, in A/D trigger logic circuitry or is latched in an interrupt or DMA.

[0030] The motor (angular) speed is also estimated. This may be based on one or more older timestamp position pairs .sub.n1/t.sub.n1 and/or modeling of the system. So for example the speed of the motor w can be determined from the equation (.sub.n.sub.n1)/(t.sub.nt.sub.n1)

[0031] To determine the next desired commutation event/time Tm for the next commutation (C.sub.m), the method calculates the commutation time T.sub.m based on the past measured position .sub.n, the timestamp t.sub.n of that past position and the instantaneous speed of the motor.

[0032] In the hardware, the timestamp may be is used to program a hardware timer 19 to provide an event 10 at T.sub.m.

[0033] Based on this event the inverter is reconfigured e.g. in block 110 of FIG. 3 for the next commutation table entry (either in hardware or via interrupt or DMA). The problem of jitter and meeting the high precision timing of the block commutation event is solved by using the hardware timer module.

[0034] So in summary, the next commutation event is (T.sub.m) is a function of t.sub.n, .sub.n, and .

[0035] The time of the next commutation event Tm may be

[0036] T.sub.m=t.sub.n+(C.sub.m,anglen)/ where C.sub.m,angle is the next commutation angle.

[0037] Of course, the skilled person would be readily aware of other methods or variation which use these basic parameters to determine the timing of the next commutation even.

[0038] Extrapolation methodology may optionally be enhanced to compensate for acceleration (measured or estimated based on a model):

[00001]$T_m=t_n+\frac{C_m-{}_n}{\left(M\right)+\left(M\right)},$

where A(M) is the acceleration derived from the model M

[0039] External parameters parameters 16 may be input to block 15a/15b in FIG. 3 in order to increase the precision of the calculation model.

[0040] If necessary the timestamps may need to be converted between the A/D converter and event timer domains. For this invention no common hardware timer is needed between the event generation and the A/D converterboth timers may in practice run at different clock and have an offset. For the same commutation event T.sub.m the hardware timer value may be updated several times in order to increase accuracy.