System for determining at least one rotation parameter of a rotating member

Abstract

The invention relates to a system comprising a coder that has an alternation of North and South magnetic poles separated by transitions extending along an helix of pitch p and of angle α. The invention having N.sub.pp pairs of North and South poles and a polar width L.sub.p measured along a normal (N) to the transitions which are: N.sub.pp=πa/I and L.sub.p=p.cos α. The invention also includes a sensor able to detect the period magnetic field emitted by the coder by means of at least two sensitive magnetic elements which are disposed at a radial reading distance from the magnetic track. The sensitive elements are disposed in relation to one another to deliver signals in quadrature.

Claims

1. A system for determining at least one rotation parameter of a rotating member, the system comprising: a coder intended to be associated rotationally with the rotating member in such a way as to move jointly with it, the coder comprising a body having a cylindrical periphery with a radius (a) around an axis of revolution (X), the periphery having an alternation of North and South magnetic poles of width 1 which are separated by transitions, each one of the transitions extending along an helix of pitch p and of angle □ to form a multipolar magnetic track which is able to emit a periodic magnetic field representative of the rotation of the coder, the track having N.sub.pp pairs of North and South poles and a polar width L.sub.p measured along a normal (N) to the transitions which are: N.sub.pp=πa/I and L.sub.p=p.cos α; a sensor able to detect the period magnetic field emitted by the coder by means of at least two sensitive magnetic elements which are disposed at a radial reading distance from the magnetic track, the at least two sensitive magnetic elements being disposed in relation to one another to deliver signals in quadrature.

2. The system according to claim 1, wherein the sensor comprises at least two sensitive elements which are spaced by a distance d measured along the normal (N) to the transitions which is equal to L.sub.p/2 modulo L.sub.p.

3. The system according to claim 2, wherein the at least two sensitive magnetic elements are aligned circumferentially.

4. The system according to claim 2, wherein the at least two sensitive magnetic elements are aligned along the normal (N) to the transitions.

5. The system according to claim 2, wherein the at least two sensitive magnetic elements are aligned along the axis of revolution (X).

6. The system according to claim 1, wherein the sensor comprises at least two groups of two sensitive elements, with the barycentre of the sensitive elements of one group being separated from the barycentre of the sensitive elements of the other group by a distance d measured along the normal (N) to the transitions which is equal to L.sub.p/2 modulo L.sub.p.

7. The system according to claim 6, wherein the sensor further comprises a device for subtracting signals (V.sub.1, V.sub.2) delivered by the two sensitive elements of each one of the groups.

8. The system according to claim 6, wherein the sensitive elements of one group are spaced by a distance e measured along the normal (N) to the transitions which: e=L.sub.p modulo 2L.sub.p.

9. The system according to claim 6, wherein the sensitive elements of a one group are spaced by a distance e measured along the normal (N) to the transitions which: 0.55L.sub.p<e<0.82L.sub.p, modulo 2L.sub.p; or 1.18L.sub.p<e<1.45L.sub.p, modulo 2L.sub.p.

10. The system according to claim 9, wherein the distance e is equal to 2/3L.sub.p or 4/3L.sub.p modulo 2L.sub.p.

11. The system according to claim 6, wherein the sensitive elements of one group are aligned.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Other particularities and advantages of the invention shall appear in the following description, made in reference to the accompanying figures, wherein:

(2) FIG. 1a and FIG. 1b diagrammatically show a coder of a system for determining according to the invention, respectively in perspective (FIG. 1a) and as a side view (FIG. 1b);

(3) FIG. 2 is a flat representation of the cylindrical periphery of the coder of FIGS. 1a and 1b;

(4) FIG. 3a and FIG. 3b diagrammatically show respectively an alternative embodiment of the arrangement at a radial reading distance of the sensitive elements with respect to a coder according to the invention;

(5) FIG. 4a and FIG. 4b diagrammatically show respectively an alternative embodiment of the arrangement at a radial reading distance of the sensitive elements by group with respect to a coder according to the invention;

(6) FIG. 5a; FIG. 5b and FIG. 5c diagrammatically show respectively an alternative embodiment of the arrangement at a radial reading distance of the sensitive elements by group with respect to a coder according to the invention;

(7) FIG. 6 is a curve showing the filtering of the 3.sup.rd order harmonic according to the distance between the sensitive elements of a group.

DETAILED DESCRIPTION

(8) In relation with these figures, a system for determining at least one rotation parameter of a rotating member with respect to a fixed structure is described. In particular, the parameter of the rotating member can be selected from its position, its speed, its direction of rotation, its acceleration or its direction of movement axially.

(9) In one embodiment, the system can be used in relation with the controlling of a brushless direct current electric motor, making it possible to know the absolute angular position on a pair of motor poles of the rotor with respect to the stator.

(10) The system for determining comprises a coder 1 intended to be integral with the rotating member in such a way as to move jointly with it, the coder comprising a body having a cylindrical periphery with a radius around an axis of revolution X on which is formed a magnetic track 2 which is able to emit a periodic magnetic field representative of the rotation of the coder. The magnetic field emitted can be sinusoidal or pseudo-sinusoidal, i.e. having at least one portion which can be correctly approximated by a sinusoid.

(11) The track 2 has an alternation of North 2n and South 2s magnetic poles of width I which are separated by transitions 3, each one of the transitions extending along a helix of pitch p and of angle α.

(12) Thus, the magnetic track has N.sub.pp pairs of North and South poles and a polar width L.sub.p measured along a normal N to the transitions 3 which are: N.sub.pp=πa/I and L.sub.p=p.cos α. The magnetic track 2 delivers a pseudo-sinusoidal magnetic signal of which the spatial period along the normal N is equal to λ=2.L.sub.p.

(13) The magnetic field generated by the coder 1 on a pair of magnetic poles 2n, 2s is the combination of a perfect fundamental sinusoidal component that is sought to be measured in order to determine the parameter, and of several odd-order harmonics (3, 5, etc.).

(14) If it is assumed that the coder 1 rotates at a constant speed of rotation ω, the magnetic field can be written in the following way:
H(t)=H.sub.1.sin ωt+H.sub.3.sin 3ωt+H.sub.5.sin 5ωt+ . . .

(15) The amplitude H.sub.3 of the 3.sup.rd order harmonic can typically represent 5% of the amplitude H.sub.1 of the fundamental. According to the position of the sensor and the reading distance, this proportion of the amplitude H.sub.3 of the 3.sup.rd order harmonic can be much higher.

(16) The helical geometry of the magnetic track 2 makes it possible the number N.sub.pp of pairs of poles 2n, 2s as well as the polar width L.sub.p to be chosen independently of the radius a of the magnetic track 2. In relation with FIGS. 1a and 1b, the coder 1 comprises four pairs of poles 2n, 2s, which is particularly suitable for the controlling of an electric motor with four pairs of poles, the system providing the absolute position on a pair of motor poles, i.e. 90° mechanical.

(17) According to an embodiment, the coder 1 is formed from a magnet on the cylindrical periphery of which the multipolar magnetic track 2 is carried out. The magnet can be formed from an annular matrix, for example made from a base of a plastic or elastomer material, in which magnetic particles are dispersed, including particles of ferrite or of rare earths NdFeB.

(18) The system for determining comprises a sensor that is intended to be integral with the fixed structure, the sensor being able to detect the periodic magnetic field emitted by the coder 1. To do this, the sensor comprises at least two sensitive magnetic elements 4, 5 which are disposed at a radial reading distance from the magnetic track 2 in order to each deliver a signal representing the rotation of the coder 1, the sensitive elements being disposed in relation to one another to deliver signals in quadrature.

(19) Each one of the sensitive elements 4, 5 can be chosen from magnetically sensitive probes. For example, Hall, tunnel magneto resistance (TMR), anisotropic magneto resistance (AMR) or giant magneto resistance (GMR) probes can measure each one of the two components of the magnetic field (normal and tangential to the coder 1).

(20) As described in document WO-2004/083881, each element 4, 5 forms a tunnel junction by comprising a stack of a reference magnetic layer, of an insulating separation layer and of a magnetic layer that is sensitive to the field to be detected, the resistance of the stack being according to the relative orientation of the magnetisation of the magnetic layers.

(21) Advantageously, each sensitive element 4, 5 can comprise at least one pattern with a magneto resistance material base, with a tunnel effect, of which the resistance varies according to the magnetic field, a sensitive element 4, 5 being able to comprise a single motif or a group of motifs connected in series or in parallel.

(22) Alternatively, the normal component only of the magnetic field delivered by the coder 1 can be measured, for example using Hall elements. Using the normal field alone is favourable because it is more sinusoidal than the tangential field.

(23) In order to be able to determine the rotation parameter of the rotating member, the signals delivered by the sensitive elements 4, 5 must preferably be in quadrature, i.e. geometrically offset by 90° divided by N.sub.pp. By using such signals in quadrature, in the sensor or in an associated calculator, it is known to determine the angular position of the coder 1, for example through a direct calculation of an arctangent function, using a Look-Up Table (LUT) or a method of the CORDIC type.

(24) To do this, in relation with FIGS. 3a and 3b, the sensor comprises at least two sensitive elements 4, 5 which are spaced by a distance d measured along the normal N to the transitions 3 which is equal to L.sub.p/2 modulo L.sub.p. In other terms, when a sensitive element 4 is positioned facing a transition 3, the other sensitive element 5 is positioned on a helix parallel to the transition and spaced from it by a distance d measured along the normal N.

(25) Thus, a good compromise is obtained between sinusoidality and amplitude of the detected signal. This optimum positioning can be obtained with a polar width L.sub.p comprised between 2 and 6 mm, and then even with a number N.sub.pp of pairs of poles 2n, 2s of the coder 1 which is less than 6.

(26) In relation with an application of the system in controlling an electric motor, the good sinusoidality of the signal delivered to the control calculator allows for:

(27) better performance, at start-up, for example the time for reaching the speed or position setting;

(28) a more “gentle” operation, without torque shifts in steady state;

(29) less energy consumption;

(30) a lower operating temperature;

(31) a more substantial maximum torque.

(32) FIG. 3a shows two sensitive elements 4, 5 which are aligned along the normal N to the transitions 3, the elements being aligned circumferentially in FIG. 3b, in the median position of the periphery of the coder 1 in order to be separated as much as possible from the edges of the coder. Alternatively, the two sensitive elements 4, 5 can be aligned along the axis of revolution X.

(33) In relation with FIGS. 4a to 5c, the sensor comprises at least two groups of two sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2, with the barycentre of the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of one group being separated from the barycentre of the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of the other group by a distance d measured along the normal N to the transitions 3 which is equal to L.sub.p/2 modulo L.sub.p.

(34) Thus, in relation with such a duplication of the sensitive elements 4, 5 of the embodiment of FIGS. 3a and 3b, the two groups allow for a filtering of the noise coming from the outside (for example from the motor or neighbouring interconnections).

(35) Indeed, by providing that the sensor further comprise a device for subtracting signals (V.sub.1, V.sub.2) delivered by the two sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of each one of the groups, if the magnetic field includes an identical noise component on the different sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2, the latter will be subtracted from the output signal.

(36) By positioning the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of a group at the magnetic phases respectively φ.sub.1 and φ.sub.2, i.e. by spacing them by a distance e measured along the normal N to the transitions 3 which is such that

(37) ${\phi }_1-{\phi }_2=\frac{e}{2L_p}*360,$
the signals V.sub.1, V.sub.2 delivered by each one of the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 can be written:
V.sub.1(t)=G.H.sub.1.sin(ωt+φ.sub.1)+G.H.sub.3.sin(3ωt+3φ.sub.1)+G.H.sub.5.sin(5ωt+5φ.sub.1)+ . . .
V.sub.2(t)=G.H.sub.1.sin(ωt+φ.sub.2)+G.H.sub.3.sin(3ωt+3φ.sub.2)+G.H.sub.5.sin(5ωt+5φ.sub.2)+ . . .
G being the supposedly identical gain of the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2, ω being the speed of rotation, H.sub.1 being the amplitude of the fundamental for i=1 and of the i-th order harmonics for i=3, 5, etc.

(38) A subtractor circuit calculates the difference which is then written:

(39) $V_1\left(t\right)-V_2\left(t\right)=G.Math.H_1.Math.\left[\sin \left(\mathrm{\omega t}+{\phi }_1\right)-\sin \left(\mathrm{\omega t}+{\phi }_2\right)\right]+G.Math.H_3.Math.\left[\sin \left(3\mathrm{\omega t}+3{\phi }_1\right)-\sin \left(3\omega t+3{\phi }_2\right)\right]+G.Math.H_5.Math.\left[\sin \left(5\mathrm{\omega t}+5{\phi }_1\right)-\sin \left(5\omega t+5{\phi }_2\right)\right]+\mathrm{.Math.}=2.Math.G.Math.H_1.Math.\sin \left(\frac{{\phi }_1+{\phi }_2}{2}\right).Math.\cos \left(\mathrm{\omega t}+\frac{{\phi }_1+{\phi }_2}{2}\right)+2.Math.G.Math.H_3.Math.\sin \left(3.Math.\frac{{\phi }_1-{\phi }_2}{2}\right).Math.\cos \left(3\mathrm{\omega t}+3.Math.\frac{{\phi }_1+{\phi }_2}{2}\right)+2.Math.G.Math.H_5.Math.\sin \left(5.Math.\frac{{\phi }_1-{\phi }_2}{2}\right).Math.\cos \left(5\mathrm{\omega t}+5.Math.\frac{{\phi }_1+{\phi }_2}{2}\right)+\mathrm{.Math.}$

(40) In relation with FIGS. 4a and 4b, e=L.sub.p modulo 2L.sub.p, i.e. the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of a group are offset 180° modulo 360°, this difference is written:

(41) $V_1\left(t\right)-V_2\left(t\right)=2.Math.G.Math.H_1\cos \left(\omega t+\frac{{\phi }_1+{\phi }_2}{2}\right)-2.Math.G.Math.H_3.Math.\cos \left(3\omega t+3.Math.\frac{{\phi }_1+{\phi }_2}{2}\right)+2.Math.G.Math.H_5.Math.\cos \left(5\omega t+5.Math.\frac{{\phi }_1+{\phi }_2}{2}\right)+\mathrm{.Math.}$

(42) It can be seen that the 3.sup.rd and 5.sup.th order harmonics are retained and have the same gain 2 as the fundamental after the subtraction operation.

(43) In order to obtain a precise determination of the rotation parameter, it is sought to measure the filtered signal of at least the 3.sup.rd order harmonic. However, any fixed compensation of the error generated by the harmonics is difficult to carry out, in that it depends on the measurement conditions (gap, position of the sensor). Moreover, a calibration is also difficult to consider for large volume and low cost application.

(44) FIG. 6 shows the filtering of the 3.sup.rd order harmonic according to the value of the offset φ.sub.1-φ.sub.2.

(45) In relation with FIGS. 5a to 5c in which the distance e is substantially equal to 2/3L.sub.p or 4/3L.sub.p modulo 2L.sub.p, the difference is written:

(46) $V_1\left(t\right)-V_2\left(t\right)=\sqrt{3}.Math.G.Math.H_1\cos \left(\omega t+\frac{{\phi }_1+{\phi }_2}{2}\right)+0-\sqrt{3}.Math.G.Math.H_5.Math.\cos \left(5\mathrm{\omega t}+5.Math.\frac{{\phi }_1+{\phi }_2}{2}\right)+\mathrm{.Math.}$

(47) In this case, the 3.sup.rd order harmonic is cancelled, the fundament and the 5.sup.th order harmonic have a gain of 1.73 after the subtraction operation. A 3.sup.rd order harmonic spatial filter was then carried out, while still retaining 86.5% of the fundamental.

(48) Generally and in relation with FIG. 6, considering that the filter of the 3.sup.rd order harmonic plays its role if it removes at least 3 dB from its value without filtering in relation to the amplitude of the fundamental, it is therefore required that:

(49) $.Math.\frac{2.Math.G.Math.H_3.Math.\sin \left(3\frac{{\phi }_1-{\phi }_2}{2}\right)}{2.Math.G.Math.H_1.Math.\sin \left(\frac{{\phi }_1-{\phi }_2}{2}\right)}.Math.\le \frac{\sqrt{2}}{2}.Math..Math.\frac{H_3}{H_1}.Math.\iff .Math.\frac{\sin \left(3\frac{{\phi }_1-{\phi }_2}{2}\right)}{\sin \left(\frac{{\phi }_1-{\phi }_2}{2}\right)}.Math.\le \frac{\sqrt{2}}{2}\iff {\phi }_1-{\phi }_2\in \left[99^{\circ };148^{\circ }\right]\mathrm{modulo}{360}^{\circ }$$\mathrm{or}{\phi }_1-{\phi }_2\in \left[212^{\circ };261^{\circ }\right]\mathrm{modulo}{360}^{\circ }$

(50) Expressed in distance, in order to obtain a filtering of the 3.sup.rd order harmonic, it is therefore required that the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of a group are spaced by a distance e measured along the normal N to the transitions 3 which is such that:

(51) 0.55L.sub.p<e<0.82L.sub.p, modulo 2L.sub.p; or

(52) 1.18L.sub.p<e<1.45L.sub.p, modulo 2L.sub.p.

(53) The distance e between the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of a group can vary within the ranges mentioned hereinabove in order to optimise the couple filtering—gain. Moreover, according to the space available, the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of the groups can be aligned along the normal N to the transitions 3 (FIGS. 4a and 5b) or circumferentially (FIG. 5c). Alternatively, the sensitive elements 4.sub.1, 4.sub.2; 5.sub.1, 5.sub.2 of a group can be offset circumferentially with respect to those of the other group (FIGS. 4b and 5a).

(54) The suppression, or at least the attenuation, of the 3.sup.rd order harmonic in the processed signals to determine the rotation parameter is beneficial relative to the precision of the determination, but also for the processing algorithms of the signal that carry out:

(55) deletion of the offset of the signals;

(56) balancing of the amplitudes of the signals;

(57) phase correction between the signals.