METHOD FOR OPTIMIZING A MAGNETIC ANGULAR POSITION SENSOR
Patent Information
- Application Number
- FR2023001949
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2023-03-02
- Publication Date
- 2025-03-21
- Estimated Expiration
- 2043-03-02
Smart Images
Figure 00000013_0000 
Figure 00000013_0001 
Figure 00000013_0002
Abstract
Description
Title of the invention: METHOD FOR OPTIMIZING A MAGNETIC ANGULAR POSITION SENSOR
[0001] The invention relates to a method for optimizing a magnetic angular position sensor or magnetic encoder.
[0002] Subsequently, for greater readability, the expressions “angular position sensor” and “magnetic encoder” will preferably be replaced by the term “encoder” which is the term commonly used by those skilled in the art.
[0003] There are various types of encoders for determining an angular position of an object: optical encoders, magnetic encoders, encoders of other types such as capacitive or inductive encoders. A magnetic angular position sensor or magnetic encoder is a device for determining, without contact, the angular position of an object by monitoring the variation of a magnetic field according to the position of the object. In the case of a magnetic encoder, a permanent magnet, at least bipolar, is fixed to a rotating shaft and generates a magnetic field. This magnetic field varies according to the movement of the magnet, therefore according to the rotation of the latter. The magnetic field is detected by a sensor configured as a disk which makes it possible to obtain two sinusoidal curves phase-shifted by 90° and corresponding for one to a sine and for the other to a cosine. By calculating the arc tangent, the measurement of an angle is obtained.It is noted that four positions are determined, by measurement with the encoder, phase-shifted by 90° at each complete rotation of the magnet over 360°, the positions making it possible to generate two sinusoidal curves phase-shifted with respect to each other. The use of electronic processing of the measurements, therefore of the values read by the encoder, of the different positions makes it possible, overall, to refine by a factor of 100 the value obtained from the angular position. Thus, as the period of each curve is given by the relationship 360° / number of pairs of magnetic poles, measurement accuracy is improved when using multipolar magnets, for example with a number of poles, south and north, generally but not exclusively between 16 and 64. Typically, a measurement accuracy, therefore detection of a variation in the angular position, of the order of 1.2 arc seconds can be obtained with a magnetic encoder.
[0004] Compared to an optical encoder, a magnetic encoder is not sensitive to pollutants such as dust. It is less fragile than an optical encoder due to the absence of a glass part. A magnetic encoder consumes little energy because there is no light source and it has a long service life due to the low number of mechanical wear parts. Magnetic encoders therefore have a cost price lower than optical encoders. Moreover, if in the case of an optical encoder we note an average form factor, that is to say a significant shift between the values of the sine and cosine obtained from the measurements and the values of the theoretical sine and cosine called perfect, this is not the case with a magnetic encoder, bipolar or multipolar, which has an excellent form factor, the measurements allowing to obtain almost perfect values of the sine and cosine.
[0005] Nevertheless, the use of magnetic encoders remains limited because they have, overall, a lower measurement performance than optical encoders. This is, among other things, due to the impossibility of obtaining a polar pitch as fine as the engraving of an optical disc and, for certain magnetoresis type encoders, due to magnetic hysteresis, that is to say the fact that the curves observed during the increase and during the decrease of the magnetic field are not perfectly superimposable.
[0006] The expression "measurement performance" refers to a set of parameters that may affect a measurement in one way or another. These include, among others, the accuracy, therefore the veracity of the measurements provided, the resolution, therefore the smallest variation in angular position that can be detected, the precision of the encoder illustrated by the degree of repeatability of the measurements and the linearity which is assessed by the difference between the measurement and the actual angular position.
[0007] The electronic part of a magnetic encoder or ASIC (Application-Specified Integrated Circuit) is an integrated circuit which groups together on a single electronic chip all the functions necessary for the operation of the encoder. However, since the ASIC is integrated into the silicon constituting the electronic chip and placed at the level of the magnets, it turns out that a so-called polar pitch of the magnets does not correspond with a so-called silicon pitch, because the ASIC is fixed and does not move, unlike the magnets which, by definition, are mobile in a rotation plane parallel to the main plane of the electronic chip. It should be noted that the silicon pitch has a fixed value and is the same for all ASICs. As a reminder, the silicon pitch is the minimum distance separating two sensors integrated on the same electronic chip.In other words, by nature, any electronic chip with an ASIC induces a constant and known shift between the curves corresponding to the sine and cosine measurements. To remedy this, one can either vary the diameter of the circular track on which the magnets are located, or increase or decrease the number of poles. Thus, the pole pitch is adjusted to twice the silicon pitch, which represents the optimal ratio. Such a modification, although effective and simple, is nevertheless limited because it only applies within a certain limit of sizes of magnetic encoders. Indeed, the use and location of the encoder imply constraints in terms of dimensions of the magnetic encoder. Therefore, if the use of magnetic encoders is possible. In many fields, such as land, rail, maritime transport, wind power, space and aeronautics, the performance of magnetic encoders limits their use to applications where high precision and / or reliability are not required.
[0008] In order to improve the performance and / or use of magnetic encoders, several solutions have been considered. Thus, by way of examples, FR-A-2 893 410 describes a magnetic angular position sensor in which the magnetosensitive elements are located at the same point and provide sinusoidal signals phase-shifted by 90°. FR-A-2 952 430 discloses a bidirectional magnetic encoder with field rotation. FR-A-2 965 347 relates to a magnetic encoder comprising a relative movement probe with respect to the magnet in order to provide a value of the position calculated as a function of the arc tangent with a correction coefficient which makes it possible to minimize measurement variations. EP-A-3 667 346 describes an encoder with several magnetic elements oriented so as to minimize the deviation of the magnetic field in a given range.EP-A-1 014 039 describes a magnetic encoder with an arrangement of the constituent elements of the encoder allowing to have a simple structure and with a high resolution. US-B-6 326 781 discloses a magnetic encoder adapted to determine the angular position of a rotating shaft over 360°. Most of the solutions envisaged modify the structure of the encoder and / or only intervene on some of the factors affecting the performance of the magnetic encoder. Thus there is a need for optimization of magnetic encoders, in order to improve their performance to develop their uses in many fields of application.
[0009] The object of the invention is to propose a method for optimizing the performance of a magnetic encoder, without structural modifications and without affecting the manufacturing costs of a magnetic encoder.
[0010] To this end, the invention relates to a method for optimizing a magnetic encoder comprising a multipolar magnet consisting of two tracks each formed from a plurality of pairs of bipolar magnets, the south and north poles being alternated between neighboring magnets located on the same track, characterized in that it comprises at least the following steps:
[0011] - a) carry out an initial survey of the values of the following characteristics of the two sinusoidal curves: origin along the ordinate axis or offset, one for a sine and one for a cosine, amplitude and relative phase, representative of the performance of the magnetic encoder, this by taking the characteristics one by one,
[0012] - b) determining a correction of at least one characteristic to be carried out in order to obtain optimal performance of the magnetic encoder defined by a phase shift between the sinusoidal curves equal to a quarter of a period,
[0013] - c) carry out for each sinusoidal curve at least one correction of a characteristic, by choosing the type of correction to be carried out among the offset adjustment, the amplitude correction and the phase correction,
[0014] -d) carry out a second survey of the values of the following characteristics of the two sinusoidal curves: origin along the ordinate axis or offset, one for a sine and one for a cosine, amplitude and relative phase, representative of the performance of the magnetic encoder, this by taking the characteristics one by one,
[0015] - e) compare the characteristics obtained in step d) with the characteristics determined in step b),
[0016] - f) repeat steps c) to e) for each of the uncorrected characteristics in choosing the type of correction to be performed from offset adjustment, amplitude correction and phase correction until the encoder performance as defined in step b) is obtained.
[0017] Thanks to the invention, without structural modification of the encoder, simple and rapid corrections can be made allowing optimization of the encoder, its performance being improved.
[0018] According to advantageous but not obligatory aspects, such a method may comprise one or more of the following steps:
[0019] The offset correction is performed first.
[0020] The encoder error corrections are carried out by Fast Fourier Transform (FFT).
[0021] Encoder error corrections are performed by iteration.
[0022] In the case of a magnetic encoder having two concentric tracks of bipolar magnets, the correction of the internal track is only carried out if the initial errors are greater than a quarter of the period of the sine and cosine signals of the external track.
[0023] The correction of the external track is carried out first.
[0024] The invention will be better understood and other advantages thereof will appear more clearly on reading the description which follows, given solely by way of non-limiting example and with reference to the appended drawings in which:
[0025] [Fig.l] is a simplified perspective view of a magnetic angular position sensor according to one embodiment of the invention and
[0026] [Fig.2] is an illustration, on another scale, of two sinusoidal curves phase shifted by 90°, one for a sine and one for a cosine, obtained when measuring angular positions by the magnetic encoder of [Fig.l], illustrating one type of error, namely the shift of the origin on the ordinates or offset and
[0027] [Fig.3] is an illustration similar to [Fig.2], illustrating other types of errors, the offset of the origin on the ordinates having been corrected.
[0028] [Fig.l] is an illustration of an embodiment of the invention. Here, a magnetic angular position sensor 1, hereinafter referred to as a magnetic encoder, is schematically illustrated. It is configured as a disk 2 free to rotate in a plane, here horizontal, around a vertical central axis illustrated by the reference AA. On one face, the disk 2 comprises a multipolar magnet 3. The magnet 3 is formed by two concentric tracks 4 and 5. Each track 4, 5 is composed of a regular alternation of bipolar magnets with their alternating south S and north N poles. For the record, the polar pitch Pp is defined as the distance between two consecutive N or S poles of two neighboring magnets located on the same track 4 or 5. The alternation of the N and S poles concerns the bipolar magnets located on the same track 4 or 5, this for each track 4, 5.The number of pole pairs being different between tracks 4 and 5, there is an angular zone in which there is an alternation of the S and N poles between tracks 4 and 5 and an angular zone in which poles of the same nature, south S or north N, located on tracks 4 and 5 are opposite each other. Thus, each track 4, 5 defines a multipolar magnet formed of a plurality denoted n of pairs Pa of magnets located on two tracks 4, 5 having alternating south S, north N magnetic poles. The period of a pair Pa, during a complete rotation of the encoder around the axis AA, is 360° / n, with a value of n, number of pairs of magnets, different between tracks 4 and 5, because the latter have a different number of magnets. Such a period corresponds to twice the polar pitch.
[0029] Each pair Pa provides, due to its constitution with two magnets with alternating poles, a magnetic field whose spatial evolution is sinusoidal. The sensors located in the ASIC provide at least two signals per track 4, 5. During a complete revolution each pair Pa produces an oscillation which makes it possible to generate, for each signal, from all the pairs, a sinusoidal curve with n periods. Each track 4, 5 thus provides at least two sinusoidal curves. These sinusoidal curves are offset, as shown in [Fig. 2], it being understood that one curve corresponds to a sine, denoted Si, and the other to a cosine, denoted Co. A cell 6 comprising the electronic elements for measuring the magnetic field, therefore the ASIC, is positioned above the magnet 3, in a position where it detects the south poles S and north poles N of the magnets constituting the two tracks 4 and 5 when the latter pass under the cell 6 during the rotation of the encoder 1.By electronic signal processing, the fineness of the signal obtained can be divided by a factor of 100, it being understood that factors corresponding to powers of 2 are often preferred. For the record, the fineness of an encoder is also called resolution. The resolution of an encoder is the smallest variation in the signal which produces a variation perceptible, and therefore measurable and measured, by the encoder 1. For example, a number n of pairs Pa of magnets is generally between 16 and 64, which corresponds approximately to what . is technically feasible due to the common dimensions of encoders. For example, with a number of pairs Pa of 32, the period T is: T= (360 / 32) / 100 following electronic processing, instead of: T=360 / 32. In other words, the higher the number n of pairs Pa, the greater the fineness of encoder 1, and therefore de facto the lower the variation in the detectable angular position will be, after electronic processing.
[0030] The electronic part, therefore the ASIC, is integrated in the electronic cell 6, more precisely in the silicon constituting the electronic cell, in order to obtain maximum protection of the electronics in a minimum of space. It is known that such a construction of the ASIC induces a specific shift between the sinusoidal curves Si and Co. This shift corresponds to the ratio between the polar pitch Pp and the silicon pitch as defined previously. It is this ratio which determines the phase shift between the curves Si and Co. The silicon pitch is known and invariable for all ASICs. It is therefore appropriate to adjust the polar pitch Pp between the two magnet tracks 4, 5 according to the value of the silicon pitch, in order to obtain the desired correct phase shift. The correct phase shift is equal to a quarter period noted ji / 2 - a sine or a cosine having a period of 2 ir - when the polar pitch Pp is twice the silicon pitch.In fact, the polar pitch Pp represents a half-period and the silicon pitch represents a quarter of a period. For this, one of the known solutions is to vary the diameter of the multipolar magnet by varying the diameter of the tracks 4, 5. Such a solution quickly finds its limits due to the size constraints that the encoder must respect.
[0031] In order to be able to overcome the constraints of encoder size and to best adjust the polar pitch Pp relative to the silicon pitch, several solutions are envisaged by the applicant. The various solutions are described in an order which does not reflect the relative importance of one solution compared to the others nor its relative degree of effectiveness compared to the effectiveness of the other solutions. The order is that which was envisaged by the applicant to validate the solutions.
[0032] First of all, it was considered to carry out a phase correction between the Si curves of the sine and Co of the cosine to overcome the differences between the Pp polar pitch and the silicon pitch. Indeed, if, when reading the measured value of the phase shift between the two sinusoidal curves, the phase shift does not correspond to a shift of exactly 90°, a phase correction is carried out between the Si and Co curves to overcome the differences between the Pp polar pitch and the silicon pitch. The shift will be brought back to 90° electronically, in order to have the correct arc tangent and therefore the correct angle.
[0033] Another solution envisaged by the applicant is the use of a Hall effect sensor. In a Hall effect sensor, the measurement of the variation of the magnetic field is associated with electronics which allows excellent linearity of the signal before transmission. Thus, the smaller the distance between the magnet and the sensor, the stronger the Hall effect will be because the distance between the magnet and the sensor influences the amplitude of the received signal. This allows a correction of the amplitude Am, Bm of the signal, it being understood that the Si and Co curves must have the same amplitude Am as illustrated in [Fig.2]. The correction is carried out by electronic signal processing or gain which results in a variation of the amplitude Am.
[0034] Another solution for optimizing a magnetic encoder envisaged by the applicant concerns the offset Of. This term designates, in a manner known per se, the offset of the origin of the curves Si and Co on the ordinate axis, this when they have the same gain and therefore the same amplitude Am. In this case the offset of the curves Si and Co is oriented towards the top or bottom of the curves, as is apparent from [Fig. 2]. The invention modifies this offset Of, so that the offset of the origin between the curves Si and Co is zero, and therefore the curves Si and Co have the same origin on the ordinate axis.
[0035] The method which is the subject of the invention combines three types of corrections: adjustment of the offset Of, gain correction and phase correction, in order to optimize the use of a magnetic encoder. In practice, the first step of the method which is the subject of the invention is to carry out an inventory, therefore a collection, of the following characteristics: origin along the ordinate axis or offset Of, amplitude and relative phase of the sinusoidal curves Si and Co of the magnetic encoder, this by taking the characteristics one by one. These characteristics are representative of the performance of the encoder. We then determine the values of these characteristics to be obtained to optimize the performance of the encoder, therefore to achieve a correct phase shift of the curves Si and Co equal to a quarter of a period. The determination of these values is done for each characteristic taken one by one.In other words, we define, according to the needs, a specification which must satisfy the characteristics mentioned above for a given magnetic encoder.
[0036] The next step consists of carrying out a correction, by choosing the type of correction from among the three types of corrections relating to each of the defined characteristics, namely from among the adjustment of the offset Of, the correction of the gain or amplitude and the phase correction. Once the first correction has been carried out for each curve Si and Co, the other two characteristics of the encoder are rechecked, for each curve Si and Co and, if another correction is necessary because the characteristics determined previously have not yet been achieved, this next correction is carried out by choosing it from among the corrections not yet carried out, this for the characteristics concerned and for each curve Si, Co. These steps are thus repeated as many times as necessary, until the determined values of the determined characteristics are obtained, namely offset, amplitude and relative phase of the curves Si and Co therefore until the desired optimization of the encoder, therefore until the correct phase shift equal to a quarter of a period.
[0037] It should be noted that, when the corrections are carried out by a Fast Fourier Transform method called FFT (Fast Fourier Transform), the order of correction of the characteristics is of little importance. On the other hand, when the corrections are carried out by iteration, the applicant has noted that the order is important and it is the characteristic concerning the origin or offset Of that should be corrected first.
[0038] The corrections are independent of each other and they must, in principle, all be carried out to optimize the encoder. Advantageously but not necessarily, the correction of the offset Of is carried out first, depending on the correction method chosen. It is known that the phase correction shifts the offset Of but not the other way around. Indeed, during the phase correction the signals Si and Co are a linear combination of the initial signals Si and Co. This has the consequence of modifying their respective offsets which follow the same linear combination of the initial offsets, it is therefore preferable to start the corrections with the offset. To carry out this correction of the encoder error, the Fast Fourier Transform method or FFT is used, the acronym being preferentially used. It is possible to use the FFT correction because each error at the offset Of has a unique signature and the combinations of errors also.Thus, by FFT, we determine a constant, in the vector sense of the term. The second coordinate of this vector is a complex constant. It is the real part of this constant which will be added or subtracted from the curve Co and the imaginary part of this constant which will be added or subtracted from the offset Of of the curve Si to cancel the offsets of the curves.
[0039] An FFT performed from 2n error measurements equally distributed over a revolution is a vector comprising 2n complex coordinates. The first coordinate is purely real and corresponds to the average of the 2n measurements. As such, it is of no interest for signal correction. A second coordinate is the complex constant correcting the offsets. A third coordinate is a complex whose real part is the phase correction factor and the imaginary part the differential amplitude. We also obtain a constant that will serve as a multiplier for amplitude correction. This constant is different from the constant mentioned previously because it is real and not vector. It is, for example, the factor by which signal Si must be multiplied so that its amplitude is identical to that of signal Co.The advantage of using the so-called FFT method is that the correction is directly deduced from the result of the FFT, whereas with the iteration method, one must first have measured the amplitude of the two curves, Si or Co, in order to be able to carry out the correction.
[0040] As can be seen from [Fig.3], the amplitude Am of the curve Co is different from the amplitude Bm of the curve Si, namely in the example illustrated in [Fig.3], the amplitude Bm is greater than the amplitude Am. Alternatively, it is the opposite. There is a phase shift noted Or between the two curves, a phase shift which is different from the expected theoretical phase shift. During the phase correction, a single multiplication factor is obtained by which the curve Si is multiplied before adding it to the curve Co to obtain a corrected curve Co. To obtain a corrected curve Si, the approach is identical: the curve Co is multiplied by the same single factor before adding it to the curve Si. To have an angle just after the phase correction, a part of the value of the sine Si is "reinjected" into the cosine Co and vice versa in order to shift the phase without modifying the relative amplitude of the curves Si and Co.The reinjected value is modulated by a phase correction factor calculated from the tangent of half the phase shift error of the cosine with respect to the sine, this error being the difference between the observed phase shift and that expected.
[0041] The advantage of a correction by FFT rather than by iteration is that the constants and correction factors can be obtained in one go. Once the corrections have been made, they are validated by a control which consists of a measurement of the encoder error which must remain lower than the defined maximum value, it being understood that it is preferable to have an optimal amplitude of the signals to make the corrections.
[0042] The correction method applies, in the case of a magnetic encoder 1 having two tracks of pairs of magnets, first on the external track 4 called the master track which makes it possible to generate the weak points of the encoder. By this expression, we designate the bits which are beyond the rank corresponding to the number of poles, knowing that the weight increases by power of two. For example for an encoder with 20 bits of resolution and 32 pairs of poles (n=6), the low-weight bits go from rank 7 to rank 20. The second track 5, internal, generates in combination with the master track the strong points of the encoder, therefore according to the terminology mentioned above, the bits which are below the rank corresponding to the number of poles, that is, with the example above, the bits going from rank 1 to rank 6 are the high-weight bits.If the error measured on the internal track 5 is sufficiently small, typically less than half a period in absolute value with an optimized algorithm and without signal drift, then the error is acceptable and no correction is necessary on the second track 5 because such an error does not affect the accuracy of the encoder 1.
[0043] Making the corrections in the following order: offset, phase, amplitude implies that a final iteration is necessary. If the corrections are made in the following order: offset, amplitude and phase, the corrections are made on the external track 4 first, therefore on the master track. If this correction is sufficient, there is no need to make the corrections on the internal track 5, as long as the error on track 5 does not exceed not a half-period as mentioned above, an optimized algorithm is able to reconstruct the so-called high-weight bits, for example those of ranks 1 to 6 for 64 points, as long as the error on track 5 before correction is acceptable, as indicated above.
Claims
1. Claims Method for optimizing a magnetic encoder (1) comprising a multipolar magnet (3) consisting of two concentric tracks (4, 5) each formed from a plurality of pairs of bipolar magnets, the south (S) and north (N) poles being alternated between neighboring magnets located on the same track (4, 5) and forming a pair of magnets, each track (4, 5) providing at least two offset sinusoidal curves, one curve (Si) corresponding to a sine and the other (Co) to a cosine, characterized in that it comprises at least the following steps: - a) taking a first reading of the values of the following characteristics of two sinusoidal curves (Si, Co) representative of a signal emitted by each track (4, 5) from all the pairs of bipolar magnets during a complete revolution of the encoder (1): origin along the ordinate axis or offset (Of), one for a sine and one for a cosine, amplitude (Am, Bm) and relative phase of sinusoidal curves (Si, Co),this by taking the characteristics one by one, - b) determining a correction to be made of at least one value of one of the following characteristics: origin along the ordinate axis or offset (Of), one for a sine and one for a cosine, amplitude (Am, Bm) and relative phases of the sinusoidal curves (Si, Co), said values being representative of an optimal performance defined by a phase shift between the sinusoidal curves (Si, Co) equal to a quarter of a period, this by taking the characteristics one by one, - c) carrying out, for the external track first, a correction of a characteristic, said correction being chosen from three types of corrections relating to each of the defined characteristics, namely the adjustment of the offset (Of), the correction of the amplitude (Am, Bm) and the phase correction to be carried out to obtain an optimal performance of the magnetic encoder (1) defined by a phase shift between the sinusoidal curves (Si,Co) equal to a quarter of a period, -d) take a second reading of the values of the following characteristics of the two sinusoidal curves (Si, CO) representative of a signal emitted on each track (4, 5) from all the pairs of bipolar magnets during a complete revolution of the encoder (1): origin along the ordinate axis or offset (Of), one for a sine and one for a cosine, amplitude (Am, Bm) and relative phase of the curves,
2.
3.
4.
5. sinusoidal (Si, Co), representative of the performance of the magnetic encoder (1), this by taking the characteristics one by one, - e) compare the values of the characteristics obtained in step d) with the correction of the value of one of the characteristics determined in step b), - f) check the two other uncorrected characteristics for each curve (Si, Co) to check if the values defined in step b) are reached and - g) repeat steps c) to f) for each of the uncorrected characteristics by choosing the type of correction to be carried out from the type of corrections not yet carried out until the performance of the encoder (1) as defined in step b) is obtained. Method according to claim 1, characterized in that the correction of the offset (Of) is carried out first. Method according to claim 1, characterized in that the error corrections of the encoder (1) in steps b) to f) are carried out by Fast Fourier Transform (FFT). Method according to claim 1, characterized in that the error corrections of the encoder (1) in steps b) to f) are carried out by iteration. Method according to claim 1, characterized in that in the case of a magnetic encoder (1) having two concentric tracks (4, 5) of bipolar magnets, the correction of the internal track (5) is carried out only if the correction of at least one characteristic to be carried out to obtain optimal performance of the magnetic encoder (1) defined by a phase shift between the sinusoidal curves (Si, Co) is greater than a quarter of the period of the sine and cosine signals of the external track (4).