1. A method for operating an absolute measurement position detection system having a sensor arrangement (100) and a single-track magnetic code object (105) with a non-repeating code region, wherein the sensor arrangement (100) is formed by a plurality of magnetic field sensors (110) arranged substantially linearly, characterized in that the relative position of the sensor arrangement (100) with respect to the respective code object (105) is determined by searching for a partial pattern (710, 720), which partial pattern (710, 720) is most similar to a partial pattern (705, 715) detected by a current sensor based on reference data comprising magnetic curve variations or magnetic patterns of magnetic field vector components (500, 505) detected by the sensors for the entire code object (105) according to positions on the code object (105).

2. Method according to claim 1, characterized in that between the respective partial patterns (705-720) to be compared a position is searched for where a deviation between the respective partial patterns of the sensor arrangement (100) on the code object (105) forms a minimum.

3. The method according to claim 2, characterized in that the deviation is formed as a squared deviation between the two partial modes (705-720).

4. A method according to claim 2, characterized in that said deviation is formed as the sum of the absolute values of the differences between the corresponding values of said two partial modes.

5. The method of claim 2, wherein the deviation is determined by means of an artificial neural network as the square or absolute value of the distance between the detected positions of the two partial patterns estimated by the artificial neural network.

6. The method according to one of the claims 2 to 5, characterized in that finding the minimum of the deviation between the partial patterns (705-720) to be compared for the entire code object (105) occurs based on a (dissimilarity) similarity curve (725).

7. The method according to one of the preceding claims, characterized in that during operation of the position detection system, partial patterns used as reference data for the code object (105) are automatically learned, which partial patterns together describe the course of a magnetic field along the single-track code object.

8. Method according to one of the preceding claims, characterized in that self-diagnosis of the position detection system is performed on the basis of the determined quality value of the similarity between the two partial patterns to be compared.

9. The method according to one of the preceding claims, characterized in that during operation of the position detection system, the learned reference data is then relearned in case of any change to the code object (105).

10. The method according to one of the preceding claims, characterized in that the learned reference data comprises values of magnetic field vector components (500, 505) detected by at least two sensors and corresponding phase information (600).

11. Method according to one of the preceding claims, characterized in that the reference data is stored in a reference table or on a reference map.

12. The method according to one of the preceding claims, characterized in that, taking into account the phase change (600), a corresponding bit pattern is generated from at least two field vector components (500, 505) detected by the sensor and phase-shifted with respect to each other.

13. A method as claimed in claim 12, characterized in that the at least two phase-shifted field vector components (605, 610) detected by the sensor are matched as well as possible in phase dependence by means of a linear integral transformation in order to generate corresponding bit patterns therefrom.

14. An absolute measuring position detection system having a sensor arrangement (100) and a single track magnetic code object (105) with a non-repeating code region, wherein the sensor arrangement (100) is formed by a plurality of magnetic field sensors (110) arranged substantially linearly, characterized in that the position detection system is operable according to the method of any of the preceding claims.

Background

A method for determining the relative position or movement of a physically linearly encoded bit sequence and a corresponding sensor head with an arrangement of sensor elements for detecting a corresponding physical parameter are known from DE 19518664C 1. Each bit of the bit sequence of the sensor head comprises two sensor elements.

Further, the present applicant has developed and marketed a magnetic tape position detection system having an incremental position sensor for detecting the precise position of a target object (e.g., a magnetic code strip) in which two sensor elements that detect a magnetic field are arranged on a line parallel to the magnetic code strip. The distance between these sensor elements corresponds to a quarter of the pole spacing of the code bar. The SIN/COS position dependence of the components of the respective magnetic field vectors is resolved or interpolated by means of an interpolator.

On the other hand, the present applicant has also developed and sold a magnetic tape position detecting system having a position sensor that absolutely measures the position of a target object (e.g., a magnetic code bar), wherein the code bar is divided into two parallel tracks. One track allows incremental measurements to be made with relatively high resolution position determination within one magnetic cycle of the code strip. On the other hand, the other track carries an absolute positioning code encoded as a non-repeating regular 12-bit or 14-bit sequence of maximum length.

Disclosure of Invention

The invention is based on the following recognition: the absolute measuring magnetic position sensor referred to herein and based on one or more magnetic code bars is designed for use only in a single magnetic mode, e.g., only in a particular extremely wide mode. Therefore, the corresponding position detection system is not easily adaptable to different magnetic code objects of uniform bit length. It is therefore an object of the present invention, firstly, to specify a method for operating a sensor arrangement referred to herein, which method enables such a sensor arrangement or a corresponding position detection system to be operated even in the case of magnetic code objects having in each case a different magnetic pattern.

The aforementioned parallel tracks of code objects with e.g. absolute code bars lead to magnetic perturbations of the adjacent magnetic fields, which leads to a significant reduction of the measurement resolution and to a rather limited possible read-out distance between the code bar strip and the magnetic sensor element. It is therefore also an object of the present invention that such a method also enables the operation of a position detection system with a sensor arrangement of the type referred to here with a large reading distance.

The aforementioned inherently known absolute sensors are also very sensitive to misalignment of the sensor elements with respect to the code object or code bar. This applies in particular to position detection systems having a sensor arrangement with sensor elements which detect magnetically in three spatial directions. According to the three-axis coordinate system shown in fig. 1, this misalignment may be:

a change in the vertical distance Δ z between the sensor arrangement and the respective code object in the z-direction shown in fig. 1, as seen from the x-direction shown in fig. 1;

the pitch α between the sensor arrangement and the respective code object, when viewed from the y-direction shown in fig. 1_Pitching(ii) a change;

the roll angle α between the sensor arrangement and the respective code object, when viewed from the x-direction as shown in fig. 1_{Side tipping}A change in (b);

the yaw angle α between the sensor arrangement and the respective code object, when viewed from the z-direction shown in fig. 1_YawA change in (b);

a lateral displacement ay between the plane of symmetry of the sensor arrangement and the plane of symmetry of the respective code object, when viewed from the x-direction shown in fig. 1;

a vertical inclination of the sensor arrangement with respect to the direction of movement of the respective code object, when viewed from the y-direction shown in fig. 1, and thereby resulting in a continuous change Δ z of the distance between the sensor arrangement and the respective code object in the x-direction;

horizontal tilting of the sensor arrangement relative to the direction of movement of the respective code object, when viewed from the z-direction shown in fig. 1, and thus results in a continuous change Δ y of the sensor arrangement relative to the respective code object in the y-direction.

It is therefore also an object of the present invention that such a method also enables the operation of a position detection system with a sensor arrangement as referred to herein, which is as fault-tolerant as possible with respect to the aforementioned misalignments.

Errors occurring in one or more of the code bars (e.g., due to mechanical wear or due to changes in magnetization, such as undesired magnetic reversal of a single magnetic pole) result in reduced measurement resolution, or even make the entire position detection system unusable. It is therefore also an object of the present invention that such a method for operating a position detection system with a sensor arrangement as referred to herein can also be adapted to such changes of the respective magnetically encoded target or measurement object.

In sensor applications with circularly curved code bars, it must always be ensured that the circumference of the respective code bar is an integer multiple of the respective bit width. It is therefore also an object of the present invention to specify a method for operating a position detection system with a sensor arrangement as referred to herein, which is also compatible with any diameter of such a code bar.

According to a first aspect, in particular to solve the aforementioned objects, the invention proposes a method for operating an absolute measuring linear position detection system as referred to herein, wherein the position detection system has a plurality of magnetic field sensors arranged substantially linearly, and a single and single track magnetic code object with non-repeating code regions, e.g. a magnetic code strip or a magnetic code strip of uniform bit length with a bit pattern encoded on the code strip.

It should be noted here that the presence of substantially uniformly magnetized regions of substantially equal length is a prerequisite for the operation based on bit identification ("bitmap"). Although for a bit-sequence magnetic code object it is very reliable to identify the actual position of the sensor head involved here based on the minimum dissimilarity between the actually measured pattern and the bit pattern previously stored in the reference map based on field vector components, phase, etc., the code object need not be formed so regularly. In contrast, the code object should be a magnetic object only, in which the magnetic pattern of the essentially non-magnetic short-circuit does not repeat, as the sensor head itself.

In the proposed method, it is provided in particular that the relative position of the sensor arrangement with respect to the respective code object is determined by searching for a partial pattern of the detected magnetic field signal that most closely resembles the currently complete measurement curve of the predetermined pattern. As a measure of similarity, an inherently known correlation technique can be applied, for example, by means of an inherently known "least squares" method to search for the least square deviation between the two signals to be compared.

Other possible methods for determining a similarity or distance measure are other measures, such as the sum of the absolute values of the differences, rather than the sum of the squares of the differences. Even with these methods, it is advantageous if the two patterns to be compared are normalized before performing the dissimilarity test to minimize the effect of the distance in the z-direction between the sensor and the code object.

Another possibility to determine the distance measure is to apply an artificial neural network that has been previously trained with respect to distance estimates between the possible detection positions of the two patterns to be compared.

According to another aspect, the partial mode is automatically learned by the position detection system, for example, during installation of the automatic execution system. In doing so, the system creates or learns a map of a given code object, as well as, in effect, a spatial profile of the partial pattern that collectively describes the magnetic field along a single track code object (e.g., a code strip).

According to another aspect, in normal operation of the position detection system, the relative position of the sensor arrangement with respect to the code object is determined by searching for a local or partial pattern in the map generated by the learning method that is most similar to the currently detected pattern.

According to another aspect, the position detection system may also perform self-diagnostics based on the respective quality of similarity or match.

According to another aspect, the learned graph may be relearned during operation of the system in the event of any changes occurring, for example, due to damage to code objects. In this way, the misalignment between the sensor arrangement and the code object can be reliably detected.

According to another aspect, the learned map preferably includes values of magnetic field vector components and corresponding phase angles. In particular, the map may be learned, expanded or updated during normal operation, i.e. during movement of the sensor arrangement. The new partial pattern learned in this way is learned based on the corresponding partial pattern of the already-known map after the partial pattern is successfully located.

With respect to the phase information, it can further be provided that, taking into account the phase change, a corresponding bit pattern is generated from at least two magnetic field vector components which are detected by the sensor and are phase-shifted with respect to one another. At least two field vector components detected by the sensors and phase-shifted relative to one another can be matched as well as possible in phase dependence by means of a linear integral transformation in order to generate a corresponding bit pattern.

It should be noted for this purpose that the main objective of the proposed method or position detection system is to use as far as possible sensor elements with single-axis sensing detection, or in the case of sensor elements with multi-axis sensing detection, to read out only one vector coordinate per sensor element, or to calculate or simulate the pattern detected in this way with respect to the respective other vector components by means of a hilbert transform. Therefore, the operation speed of the position detection system can be significantly improved.

According to yet another aspect, the position detection system may already be operated in a normal mode during learning, wherein the learning process, which is preferably continuously performed, may maintain or even improve the measurement resolution or measurement characteristics of the system throughout its lifetime. Furthermore, by means of a learning process, operation under changing measurement conditions is also possible.

It should be noted that in the present context, a so-called "magnetic code" denotes a sequence of magnetic bits, wherein longitudinally arranged regions having substantially equal bit lengths correspond to the bits of the bit sequence, and the magnetization direction of each region is determined by the value of the corresponding bit. Such regions of substantially equal bit length are referred to herein as "magnetic bits". The regions themselves are substantially magnetically uniformly polarized, wherein the polarization direction is perpendicular to the surface of the code strip, and wherein the polarizations of the regions have substantially equal strength, but opposite directions for binary "0" and "1" values, respectively.

In the present context, a "bitmap" is a sequence of bits stored in a memory of the sensor device to represent the binary value of the corresponding sequence of bits corresponding to said magnetic bits of the code strip.

The position detection system referred to herein identifies the values of the magnetic bits over their length, i.e. in a manner corresponding to the sequence of detected bits, and determines the position of each 0- >1 and 1- >0 bit shift along the longitudinal axis of the code strip relative to the coordinate system of the sensor arrangement. The position of the bit shift may occur with an accuracy better than the 1/4 bit length of the magnetic bits.

The position detection system referred to herein locates the sequence of bits detected at the respective code object in the mentioned bitmap and calculates therefrom the approximate absolute position of the sensor arrangement along the code object as the mathematical product of the magnetic bit width and the sequence number of the respective first bit of the sequence observed within the bitmap.

The position detection system referred to herein further calculates an absolute position from the bit sequence detected at the code object as the sum of the starting position and the coarse absolute position of the first bit of the observed bit sequence in the coordinate system of the sensor arrangement.

The proposed method for operating a sensor arrangement or a corresponding position detection system referred to herein has in particular the following technical effects or advantages resulting therefrom:

by means of the method, the position detection system can be easily adapted to the respective (magnetic) environment and can be calibrated without an external reference system;

thanks to said method, the position detection system has a high degree of fault tolerance, since it can restore its normal operation in case of various damages, misalignments and failures;

the method enables a very robust operation of the sensor arrangement or the position detection system;

the sensor arrangement or the position detection system can operate very reliably by the method even in the case of extreme misalignments, for example with respect to pitch, roll and/or yaw, or spatial displacements in three spatial directions of the sensor elements or spatial displacements of the sensor arrangement with respect to the respective code object;

by means of the method, the sensor arrangement or the position detection system can be used with any code object (e.g. with absolute or incremental code bars or code strips);

the position detection system can also operate reliably by means of the proposed learning method in the case of a manual creation of the code strip and in the case of a damage of the code strip, for example due to magnetic reversal;

the method may use a sensor arrangement or a position detection system without being limited by the diameter of the circular code object.

Drawings

FIG. 1 illustrates, in a schematic isometric view, a position detection system as referred to herein;

fig. 2a, 2b schematically show two possible designs of the magnetic code strip or magnetic code strip referred to herein;

fig. 3 shows typical changes of the magnetic field vector when moving along the code strip shown in fig. 2a and 2b by means of the sensor arrangement of the position detection system referred to herein, on the basis of which a phase estimation is performed;

fig. 4a, 4b show typical measurement curves obtained with the sensor arrangement of the position detection system referred to herein, with the phase transitions shown in fig. 3 and the corresponding position data (4a) and bit patterns (4b) detected on these measurement curves;

fig. 5a to 5c show exemplary embodiments of a learning method according to the invention, which is based on typical magnetic sensor data measured on the code object (in this case in the form of a code strip) referred to herein, depending on the position of the sensor arrangement along the code strip;

6 a-6 e illustrate the creation of a first map resulting from a magnetic field component detected by a sensor and a second map having a phase angle derived from the detected sensor data;

FIGS. 7 a-7 d illustrate an exemplary embodiment of a process for finding partial patterns stored on referenced graphs based on similarity comparisons in accordance with the present invention;

fig. 8a to 8g show exemplary measurement curves illustrating the noise immunity or robustness of the method according to the invention to external magnetic fields present in the vicinity of the code strip.

Detailed Description

The sensor arrangement (or sensor head) 100 shown in fig. 1 forms together with the magnetic target object shown here (in the present case a stretched magnetic code strip 105) a linear absolute measurement position detection system.

The magnetic code strip 105 has a plurality of magnetic poles with an upward pole direction 107 or a downward pole direction 108. The linear arrangement of these different poles in the x-direction represents the encoding of the magnetic code strip 105.

In the present exemplary embodiment, the sensor arrangement or sensor head 100 has a plurality of eighteen (18) magnetic field sensor elements 110, which magnetic field sensor elements 110 are irregularly spaced in the x-direction, as indicated by arrows 125. The sensor head 100 also includes a measurement unit and digital signal processing unit (DSP unit) 115, and a digital communication interface 120.

In addition to this, the typical spatial arrangement of the axes of the coordinate system 130 of the sensor arrangement 100 with respect to the magnetic code strip 105 provided in the present exemplary embodiment is marked.

The measurement/DSP unit 115 arranged on the sensor arrangement detects and processes raw signals from the magnetic field sensor element 110 and communicates with external devices (not shown here) via the digital communication interface 120, i.e. for transmitting sensor data, parameter data and diagnostic data. The magnetic field sensor element 110 is designed to be magnetically sensitive, in particular in two axes, in order to be able to perform a phase evaluation of the measurement signal, as mentioned and described in more detail below.

The magnetic field sensor element 110 has in particular the following technical properties or characteristics:

they are designed to be substantially equal;

they are arranged along the magnetic code object in the direction of movement of the sensor element;

-they are arranged along a straight line or along a curved trajectory according to the spatial configuration of the magnetic code objects or the movement trajectory of the respective target object to be detected;

they are arranged with a substantially constant distance between their respective sensor elements, or with a different or varying distance between the respective sensor elements;

each having at least two sensitive axes for detecting a magnetic field generated by a magnetic target object. Thus, the sensitive axis spans a plane substantially coinciding with both the arrangement of magnetic field sensor elements and a line connecting the arrangement of magnetic field sensor elements and the center of the respective magnetic target object. The center is the center line of the magnetic code strip or the center of the discrete magnet, depending on the type of target object.

However, the sensor arrangement proposed herein is also applicable to sensor elements that perform magnetic detection on only a single axis. Furthermore, the sensor arrangement may (optionally) still have a third axis of sensitivity oriented substantially perpendicular to the first two axes.

In particular, the signal processing unit 115 has the following technical properties or characteristics:

it has programmable components (e.g. a microcontroller, FPGA or similar, or a combination of such components), and operating memory (e.g. RAM as fast as possible), and rewritable non-volatile memory, e.g. FLASH, FRAM or similar;

it periodically reads out the signal from the magnetic field sensor element;

it converts the signal detected by the sensor into a series of conventional sensor signals in a self-adjusting manner (as it were, by means of background correction and by means of gain compensation, for example, to eliminate small sensing differences between the magnetic field sensor elements of the sensor arrangement based on the spatial rotation of the rectified signal relative to the coordinate system of the sensor arrangement);

it determines the relative position of the sensor arrangement or sensor head with respect to the magnetic target object based on the detected sensor signals;

it provides diagnostic information and tools for installation, maintenance and normal operation of the position detection system;

it is capable of two-way communication with external devices via a digital interface.

Fig. 2a and 2b schematically illustrate two exemplary magnetic code strips (or bars) of the position detection system referred to herein.

The absolute code band shown in fig. 2a has only a single track 200, the single track 200 having an absolute code with a uniform bit length. The absolute code consists of a linear arrangement of poles with an upward pole direction (see fig. 1)205 and with a downward pole direction 210. Thus, the illustrated encoding includes both a single pole 207 surrounded by poles of different polarity, and multiple poles, i.e., multiple connected poles 205, 210 of the same polarity.

In contrast, the code strip shown in fig. 2b, which is also suitable for the sensor arrangement according to the invention or a corresponding position detection system, has a corresponding code target object with both a uniform, increasing bit length code 215 and a relatively short code segment 220 or in each case a bit code with a uniform bit length.

According to fig. 2a and 2b, only a single absolute code track 200, 220 is required in each case according to the invention. This has various advantages over magnetic code bars. First, the result is lower manufacturing costs and overall operating costs compared to the prior art. Furthermore, the result is simpler installation and greater assembly tolerance tolerances.

Fig. 3 illustrates the magnetic field vectors generated by the sensor arrangement according to the invention when moving along the code strip illustrated in fig. 2a and 2b in the illustrated x-direction 300. The dots 305 included in the figure indicate the corresponding scanning positions. The lines 310, which are also shown, are aligned in each case in the direction of the field vector occurring during magnetic induction, which corresponds to the possible phase values 320 indicated below the figure. The length of the line has been normalized to the maximum value of the absolute field value present.

The staircase line 315 drawn in the lower portion of the figure corresponds to the magnetic code produced by the scan. The angle of the magnetic induction vector is measured relative to the x-axis. Hereinafter, this angle is referred to as the phase angle or phase of the magnetic induction vector.

At any distance of the sensor element from the respective code strip (i.e. in the vertical z-direction as shown in fig. 1), wherein this distance should not be larger than the longitudinal dimension of the magnetic code bits, the magnetic induction vector rotates in the opposite direction (i.e. clockwise in this illustration) as the sensor element moves from left to right. Above the code bit boundaries with alternating (bit) polarity, the field is substantially horizontal. However, since the magnetic field distribution of individual magnetic code bits (see the bits included in the top row) becomes wider, it is difficult or impossible to detect or measure some bits at a larger distance.

Fig. 4a illustrates a typical measurement curve of the phase transitions at the code bits detected in each case. Here, the curve 400 corresponds to the phase variation of the magnetic induction vector along a (horizontal) position on the code strip, and the staircase line 405 corresponds to the corresponding value of the detected magnetic code bit.

If the sensor arrangement or the sensor head is moved from left to right over a respective target object (e.g. a magnetic code strip, a dipole magnet, etc.), the magnetic induction vector rotates in the opposite direction, i.e. clockwise in the present case. The phase variation 400 now has performance characteristics corresponding to the structure of the magnetic code strip. The plateau-shaped phase change region 410 corresponds to a longer, magnetically uniform segment. However, at magnetic transition 415 between poles or pole regions of different polarity, the corresponding code bits are inverted. It is therefore possible in the present case to determine the respective magnetic bit sequence by means of "inverse analysis".

In fig. 4b, a code strip bit pattern 420 is depicted in the upper part of the illustrated diagram, where 425 and 430 represent the respective x-and z-components of the corresponding regular (and equalized) signal detected by the sensor. The detected binary pattern 435 is depicted in the middle of the figure. Here, point 440 corresponds to the detected bit and point 445 corresponds to the actual reference position of the sensor. In the lower part of the figure, a detected bit pattern 450 and a monotonic phase 455 generated from the regular signal detected by the sensor are depicted.

Fig. 5a to 5c show exemplary embodiments of the learning method according to the invention, which use typical magnetic sensor data measured on a code strip and which depend on the position of the sensor arrangement along the code strip.

According to said method, operating during the position detection system (i.e. during the movement of the sensor arrangement along the code object), the magnetic field vector component (field component B in the exemplary embodiments described below) detected by the sensor is based on_xAnd B_z) And generating a reference table (referred to herein as a "reference map") having magnetic curve variations or magnetic patterns based on the phase angles of the vector components. Thus, new (local) magnetic partial patterns are learned by finding already existing partial patterns corresponding to or related to these partial patterns on an existing map by means of a similarity check and by replacing, refining or correcting these corresponding partial patterns with the corresponding currently detected partial patterns.

It should be noted here that the identification of the actual position based on dissimilarity does not depend on the number of sensitive axes of the sensor elements. This applies to both single-axis and three-axis sensitive sensor elements.

In the present exemplary embodiment, the mentioned learning process comprises the following four process steps based on the already existing reference map:

1. the existing reference map of the code objects involved in each case is read out (note: a separate reference map is created for each code object),

2. the partial pattern is found in the reference map by means of correlation calculations, wherein this partial pattern resembles the currently measured partial pattern or the corresponding magnetic curve variation as much as possible,

3. learning new or corresponding changed patterns or corresponding complete curve changes by mathematical interpolation, an

4. The changed pattern or curve change is idealized on the basis of a given physical model of the magnetic system, which is referred to here, and the already existing reference map is updated accordingly.

The advantage of this procedure is that the position detection system can operate in normal mode even during the learning process, and as a result of the learning process, the system can retain or even improve its measurement performance even in the case of drastic changes in external conditions (for example, due to parasitic magnetic fields described below).

In fig. 5a to 5c, the field vector component B detected by the sensor on the code strip_x(Black dots) 500 and B_z(white point) 505 is in units of [ mm ] along the code strip in the sensor arrangement]Is depicted in each case in arbitrary units above the actual position. The lines 507, 509 also drawn correspond to simple connecting lines between the measurement points, drawn differently for each of the two field components (B)_x: thick line, B_z: dashed line).

It should be noted that in the present example, the number of points 500, 505 shown in fig. 5a to 5c is 11 (eleven), corresponding to the sensor arrangement 100 shown in fig. 1 with only eleven (11) sensor elements 110, wherein, in the present exemplary embodiment, the distance between the individual sensor elements is approximately 1.38 times the bit length.

Fig. 5a therefore shows the currently detected spatially limited (local) partial pattern (point distribution) or the corresponding curve variation for the two field vector components. These patterns or corresponding curve variations are stored or saved on a reference map.

In the measuring situation shown in fig. 5b, the sensor arrangement has been moved approximately 40mm to the upper right of the code strip. As a result, further partial patterns or corresponding curve variations are shown for the two field vector components. In the present exemplary embodiment, this further reference data is also stored on the reference map.

In the measuring case shown in fig. 5c, the sensor arrangement above the code strip is shifted to the left by approximately 130mm starting from the last position according to fig. 5 b. As a result, still further partial patterns or corresponding curve variations 510 for the two field vector components are shown. This additional data is also stored on the reference map in addition to the already stored data 515.

As a result, the two field vector components B for the current code band as a whole, depicted in fig. 5c by the solid line 520 and the dashed line 525_xAnd B_zA change in (c).

FIGS. 6a to 6e show the creation of a first diagram with two field vector components B starting from the (local) partial mode of the two magnetic field components 500, 505 shown in FIG. 5a_xAnd B_zA complete variation along the entire code strip. Furthermore, from this data it is shown that a second diagram with corresponding phase angles is created.

From the point values 500, 505 (fig. 6a), the phase offset or corresponding phase angle of the measurement point between the two measured field vector components is calculated in units of "radians" (fig. 6b), depending on the position of the sensor arrangement formed by the respective difference.

By moving or traversing the sensor arrangement along the entire code strip in the manner shown in fig. 5a to 5c, a complete first reference map of the two magnetic field vector components is generated (fig. 6 c). Correspondingly, a second reference map 615 (fig. 6d) of the phase paths or phase offsets between the two field vector components along the entire code band is generated. In this depiction, the local phase route 600 according to fig. 6b is depicted or highlighted by window 620.

According to two field vector components B as shown in FIG. 6c_xAnd B_zThe two phase offset curves 605, 610, the corresponding bit pattern 625 is calculated or formed taking into account the phase variation 615. The (binary) bit sequence 630 that appears in this example is also listed above the bit pattern 625. This is because the two field components almost represent harmonic conjugates and therefore the phase-dependent matching or overlapping is preferably done as good as possible by means of an integral linear transformation (e.g. the overall known "hilbert transformation") so that they can be evaluated in connection with determining the corresponding bit pattern.

It should be noted here that a joint evaluation can also be carried out if each sensor element has only one sensitivity direction. Thus, in case the sensor element is sensitive to only one field vector component (e.g. Bx), the course of the respective other field component (i.e. in this case Bz) can be at least approximately calculated or simulated from the course of the field vector component Bx by means of the hilbert transform.

Fig. 7a to 7d show a method for finding a partial pattern stored on a graph by means of a similarity check. Here, the position detection system determines the position of the sensor arrangement relative to the code strip by looking for a local or partial pattern stored on the reference map that is most similar to the partial pattern detected by the current sensor. In the present exemplary embodiment, the similarity is calculated using a known method of minimum deviation or squared error. However, other known correlation methods, such as the method mentioned at the outset, are also conceivable.

Fig. 7a shows a code strip 700 which is assumed here and has only one track, with the irregular sequence of the poles 108, 108 shown in fig. 1.

Fig. 7b again shows only the local detection sensor signals for the currently detected sensor data 705 (black dots) and for the sensor data 710 (white dots) stored on the reference map and very similar to the current sensor data 705, both values being expressed in arbitrary units and actually depending on the relative position in mm between the sensor arrangement and the code strip (see fig. 1). The two measurement curves, again generated by simple connecting lines between the measurement points, show a clearly visible high match.

In contrast, fig. 7c shows sensor data 715 (black dots) currently locally detected at a different location on the reference map and corresponding data 720 (white dots) stored at this location on the reference map, both values again being expressed in arbitrary units. However, a considerable deviation between the two curves can be seen in comparison with fig. 7 b.

Fig. 7b shows a dissimilarity curve 725 calculated from the matching results of fig. 7b and 7c, which may be used to determine the location on the reference map with the highest degree of matching. Thus, the local values according to fig. 7b result in relatively low dissimilarity values, and the local values according to fig. 7c result in relatively high dissimilarity values. In this example, the two partial modes shown in FIG. 7b result in a very low dissimilarity value 730 due to the relatively high degree of matching.

Thus, it is assumed that the current code strip position is the relevant position on the reference map, wherein the partial pattern shown in fig. 7b is the most likely current position of the sensor arrangement relative to the code strip.

Finally, fig. 8a to 8g show typical measurement curves to illustrate the robustness of the method with respect to external magnetic dipole fields or corresponding parasitic interference fields that are only locally present near the code strip in the strip direction of the code strip.

Fig. 8a and 8b show characteristic curves 800, 805 of the sensor arrangement in accordance with the invention, in which the position determined using the described method (y-axis) is plotted in the corresponding units [ mm ] against the actual position (x-axis). The two figures also show the curve changes 810, 815 (each expressed in arbitrary units) of the relative position deviations that occur between the determined position data and the actual position data.

Fig. 8a shows the situation without the previously mentioned disturbing field, while fig. 8b shows the situation with the presence of a disturbing field. As can be seen by comparing the two characteristic curves 800, 805, the influence of the interference field on the characteristic curve 800 of the sensor arrangement is relatively small or negligible.

Fig. 8c to 8e show two field vector components B which have already been depicted in fig. 6c_xAnd B_zAnd in fact there is no so-called parasitic interference field in fig. 8c, whereas such interference field is present in fig. 8 e. In this case, the curve change 825 is relearned in the described manner during the contact of the interference field. It should be noted that the difference 830 between the two curves 820, 825 substantially corresponds to the field path 835 of the interference field shown in fig. 8d, which is also detected by means of the sensor arrangement.

In fig. 8f and 8g, the dissimilarity curves 840, 845 already shown in fig. 7d are depicted, and in fact no disturbing field is present in fig. 8f, whereas in fig. 8g a disturbing field is present, wherein the reference map (with corresponding curve variations) created without being disturbed by parasitic disturbing fields is used as a basis for creating the dissimilarity curve 845. In the first dissimilarity curve 840, a minimum 850 (and therefore a maximum for similarity) occurs where the dissimilarity value is zero at a-25 mm position of the code strip, while in the second dissimilarity curve a minimum 855, 860 occurs at two positions of-53 mm and +55mm of the code strip, but where the dissimilarity value is 0.12. Therefore, even in the presence of an interference field, a curve change of the non-similarity curve 845 occurs, which is fully evaluable and yields an absolute position value. In the case of careful observation, location 855 would even be selected to be relatively close to location 850 because the minimum at location 855 is slightly higher than the minimum at location 860.