Calibration of a current sensor

Abstract

Method for calibrating a sensor unit, - wherein the sensor unit comprises: - a sensor housing, - a first magnetic sensor, - a second magnetic sensor and - a third magnetic sensor, - wherein the first magnetic sensor, the second magnetic sensor and the third magnetic sensor are arranged on a semiconductor chip, - wherein the sensor unit is arranged on a device during calibration, the device comprising a first conductor and a second conductor, - wherein a first magnetic field is generated by the first current-carrying conductor and a second magnetic field is generated by the second current-carrying conductor by a magnetic field generating device, wherein the first current-carrying conductor and the second current-carrying conductor have a different distance from the sensor unit and / or a different orientation to the sensor unit, - wherein the first magnetic sensor and the third magnetic sensor are each configured to detect magnetic field components in a first direction, - wherein the second magnetic sensor is configured to detect magnetic field components in a second direction, - where the first direction is not parallel to the second direction, - wherein the calibration of the sensor unit is carried out based on responses of the magnetic sensors to the magnetic field by - a first current is sent through the first conductor and the first signals from the three sensors are measured, - a second current is sent through the second conductor and second signals from the three sensors are measured, - whereby the magnetic sensitivities of the three sensors are determined based on the first signals and the second signals.

Description

[0001] The invention relates to a device and a method for calibrating a current sensor.

[0002] DE 10 2012 221 803 A1 relates to a current sensor component for detecting a current flowing in a primary conductor of a substrate.

[0003] There are current sensors that measure an electric current through an electrical conductor. The current generates a magnetic field, and the current sensor measures this magnetic field and uses this to determine the current level.

[0004] The exact position of the current sensor (or the sensor elements of the current sensor) to the electrical conductor is critical: Since the magnetic field of the current conductor is highly inhomogeneous, even a small change in the position of the current sensor can strongly influence or change the magnetic field on the sensor elements.

[0005] Current sensors are known in which the electrical conductor is an integral part of the sensor housing. This ensures that the conductor and current sensor are positioned at a fixed spatial distance from each other. The manufacturer of the current sensor is responsible for calibration, i.e., the positioning of the sensor element in relation to the electrical conductor.

[0006] Current sensors are also known in which the electrical conductor is not part of the sensor housing. Such current sensors have a sensor housing that is attached to the electrical conductor, for example, by a customer. However, due to tolerances in the attachment, it often happens that the sensor element does not detect the exact desired magnetic field, but rather a deviating one. This leads to an error in the current measurement, which can, for example, be in a range of 5% to 15%. This is disadvantageous when higher current measurement accuracy, in particular an error of no more than 1%, is required.

[0007] DE 10 2012 110 406 A1 discloses a proposal for current measurement in which the current sensor comprises two types of sensor elements (e.g., Hall plates and vertical Hall sensors). The first type of sensor detects a magnetic field component perpendicular to a conductor surface, while the second type of sensor detects a magnetic field component parallel to the same conductor surface. If the relative magnetic sensitivity of both sensor types is known, the ratio of the two measured values ​​can be used to determine the position of the sensor elements with high accuracy, thus achieving precise calibration of the current sensor.

[0008] One problem here, however, is that two different types of sensors are required for which, for example for technological reasons, the relative magnetic sensitivities to each other cannot be known with sufficient accuracy. For example, different types of sensors can lead to different process variations or individual sensor variations, which require additional adjustment measures (so-called "trimming procedures"). Without such adjustment measures, the ratio of the magnetic sensitivities on the chip is only known to within 1±4%. This results in an uncertainty of 8% (tolerance band), for example, which arises due to variations in layer thicknesses, different doping concentrations, different outdiffusions and / or different influences of mechanical stresses on both sensor elements.

[0009] Therefore, it is a task to create an efficient approach to calibrating the sensor elements.

[0010] This object is achieved according to the features of the independent claims. Preferred embodiments can be found in particular in the dependent claims.

[0011] To achieve the object, a method is provided for calibrating a sensor unit according to the features of claim 1.

[0012] In particular, the sensor unit can have more than two magnetic sensors. It is also possible to use two types of magnetic sensors, with at least one magnetic sensor per type.

[0013] The magnetic field generating device can be any type of device, element, or arrangement of multiple elements capable of generating at least one magnetic field acting on the sensor unit. Based on the at least one magnetic field, the magnetic sensors provide a reaction (referred to herein as a "response"), e.g., in the form of a measurement signal. It should be noted that the at least one magnetic field can be one or more magnetic fields or changes in at least one magnetic field. In this case, the response can also be a signal that corresponds to a temporal progression.

[0014] The calibration of the sensor unit may be performed by any processing unit configured to calibrate the sensor unit based on responses of the first magnetic sensor and the second magnetic sensor to the magnetic field.

[0015] This processing unit can be embodied, in particular, as a processor unit and / or an at least partially hard-wired or logical circuit arrangement, which is configured, for example, to carry out the method as described herein. Said processing unit can be or include any type of processor or computer or computer with the correspondingly necessary peripherals (memory, input / output interfaces, input / output devices, etc.).

[0016] It is a further development that the magnetic field generating device is designed separately from the sensor housing.

[0017] It is a further development that the magnetic field generating device is arranged in the sensor housing.

[0018] It is a further development that the magnetic field generating device generates at least two magnetic fields of different strengths, which act on at least one of the magnetic sensors.

[0019] Optionally, a rotating magnetic field of constant strength can be used with an angle sensor. It is also possible to move (e.g., rotate) a magnet at two different distances to test whether the angle sensor functions correctly between a minimum and a maximum magnetic field. The magnet can be moved at two different distances by first rotating the magnet, for example, 2 mm and then 4 mm above the magnetic sensor. In this example, the magnet is moved to two different positions along the rotation axis.

[0020] It is a further development that the first magnetic sensor and the second magnetic sensor have a predetermined distance from each other.

[0021] For example, an advantageous distance is in a range between 0.5mm and 3mm.

[0022] For example, a current sensor uses two magnetic sensors that are a certain distance apart in order to separate the magnetic field of the conductor from interfering background fields. In particular, the conductor can be arranged between the two magnetic sensors. In this case, the two magnetic sensors detect a magnetic field of (essentially) equal strength in different directions. If the signals from the two magnetic sensors are subtracted from one another, the contributions from the current conductor are added together and, at the same time, homogeneous interfering fields are canceled out. The distance between the magnetic sensors is preferably chosen to be large enough to allow the entire current to fit between them, i.e., so that the magnetic field can be detected as completely as possible. On the other hand, however, the distance should not be unnecessarily large for reasons of space efficiency.

[0023] If the current is small (e.g., 100mA), the conductor can be thin (e.g., 0.5mm thick), and the distance between the magnetic sensors can also be relatively small (e.g., 0.7mm). However, if the current is large (e.g., 100A), the conductor cross-section is preferably approximately 10mm. 2 , because the current density in the conductor should not exceed 10A / mm 2 up to 15A / mm 2 to prevent the conductor from overheating. Such a conductor can have a diameter of 11mm, and the magnetic sensors can be spaced accordingly. However, since a semiconductor chip is usually not that large for cost reasons, for currents of more than 5A, for example, the existing chip size of 1.5mm to 3mm is used for the sensor spacing.

[0024] It is a further development that the sensor unit is a current sensor.

[0025] It is a further development that the magnetic field generating device generates a magnetic field that is essentially proportional to an electric current.

[0026] It is a further development that the magnetic field generating device generates at least two magnetic fields which act on at least one of the magnetic sensors.

[0027] It is a further development that at least one of the two magnetic fields acts inhomogeneously on the at least one magnetic sensor.

[0028] In particular, the inhomogeneous magnetic field at the magnetic sensor has an inhomogeneity of more than 1% per 0.5 mm.

[0029] It should be noted that the magnetic fields of the conductors are usually inhomogeneous.

[0030] In the example of a cylindrical conductor, the magnetic field is proportional to the reciprocal radial distance (1 / radial distance), where the radial distance is the distance between the cylinder axis (the core of the conductor) and the reference point. For example, if the radial distance is 10 mm and the positioning tolerance is 0.1 mm, the magnetic field changes by 0.1 / 10 = 0.01 = 1%. This is an estimate of a real inhomogeneity. If two conductors with a rectangular cross-section are present, one will be slightly closer, the other slightly farther apart, so the latter will have a slightly better homogeneity than 1% / 0.1 mm.

[0031] It is a further development that the first direction is essentially perpendicular to the second direction.

[0032] It is a further development that the at least one magnetic field generated by the magnetic field generating device comprises an inhomogeneous magnetic field on at least one sensor element.

[0033] It is a further development that the first magnetic sensor and the second magnetic sensor are sensors of different types.

[0034] For example, the following types of magnetic sensors can be used: Hall plates, vertical Hall sensors, horizontal Hall sensors, magneto-resistive (MR) sensors, e.g. anisotropic MR (AMR) sensors, giant MR (GMR) sensors, tunneling MR (TMR) sensors, colloidal MR (CMR) sensors, MAG-FETs.

[0035] It is a further development that the two magnetic sensors are Hall sensors that detect magnetic field components acting in different directions.

[0036] It is a further development that the at least one magnetic field comprises at least two magnetic fields whose spatial distribution is such that the ratio of the responses of the first magnetic sensor to both magnetic fields is not equal to the ratio of the responses of the second magnetic sensor to both magnetic fields.

[0037] It is a further development that a first magnetic field is generated by a first current-carrying conductor and a second magnetic field is generated by a second current-carrying conductor, wherein the first current-carrying conductor and the second current-carrying conductor have a different distance from the sensor unit and / or a different orientation to the sensor unit.

[0038] It is a further development that the magnetic field generating device comprises at least one current-carrying conductor.

[0039] It is a further education that - the sensor unit further comprises: - a semiconductor chip on or in which the first magnetic sensor and the second magnetic sensor are arranged, - a frame on which the semiconductor chip is arranged, - the sensor housing is a plastic housing.

[0040] In particular, the sensor housing is at least partially a casing for the components, i.e. the sensors, the semiconductor chip, the frame.

[0041] It is a further development that the semiconductor chip is glued to the frame.

[0042] It is a further development that the sensor housing is arranged on a base plate.

[0043] In particular, the sensor housing can be arranged in a recess of the base plate. Furthermore, it is an option for the base plate to have at least one conductor. The current through this at least one conductor can be detected by the sensor unit described here (after successful calibration).

[0044] The above explanations regarding the devices apply accordingly to the method. The respective device can be implemented in a single component or distributed across multiple components.

[0045] It is a further development that at least one magnetic sensor has a sensitivity substantially in a direction perpendicular to the semiconductor chip.

[0046] It is a further development that the at least one magnetic field comprises at least two magnetic fields whose spatial distribution is such that the ratio of the responses of the first magnetic sensor to both magnetic fields is not equal to the ratio of the responses of the second magnetic sensor to both magnetic fields.

[0047] It is a further development that a first magnetic field is generated by a first current-carrying conductor and a second magnetic field is generated by a second current-carrying conductor, wherein the first current-carrying conductor and the second current-carrying conductor have a different distance from the sensor unit and / or a different orientation to the sensor unit.

[0048] It is a further development that the calibration of the sensor unit is determined based on the responses of the first magnetic sensor and the second magnetic sensor as magnetic sensitivities of the magnetic sensors.

[0049] The magnetic sensitivities can be stored in coded form in a memory that is optionally assigned to the sensor elements. This could be, for example, an EEPROM on the same chip.

[0050] After calibration, the test specimen can be used in its position to measure the electrical current through the conductor, taking into account the determined sensitivities of the sensor elements.

[0051] The properties, features, and advantages described above, as well as the manner in which they are achieved, are further explained in connection with the following schematic description of exemplary embodiments, which are explained in more detail in conjunction with the drawings. For clarity, identical or equivalent elements may be provided with the same reference numerals.

[0052] They show: Fig. 1 an exemplary arrangement for calibrating a test object; Fig. 2 based on the representation of Fig. 1 an exemplary sketch to illustrate the positioning tolerances ϑ, δ x , δ y .

[0053] An exemplary solution proposes using at least two selectively activatable magnetic field sources for calibration instead of a single magnetic field source. Preferably, the two magnetic field sources have a known location relative to each other. In particular, the relative positioning of the two magnetic field sources is thus known.

[0054] A calibration can be performed, for example, as follows: (i) A test specimen is positioned close to a magnet array. A relatively accurate position of the test specimen is advantageous, allowing for compensation of minor inaccuracies. In particular, it is advantageous that the position of the test specimen remains (essentially) unchanged during calibration. The device under test is, in particular, an arrangement comprising a semiconductor chip (also referred to as a "chip") arranged on a leadframe. Several sensor elements can be arranged on the chip. The arrangement of chip, leadframe, and sensor elements can be encapsulated in plastic (i.e., surrounded by a plastic casing). As already explained, the chip with the sensor elements can have a certain positioning tolerance relative to the frame, which should be taken into account during calibration.

[0055] The sensor elements can also be referred to as sensors. (ii) A magnetic field is generated by a first part of the magnetic field source and a response of the sensor elements is measured. The response is, in particular, a measurement response, i.e., at least one signal measured as a result of the applied magnetic field. Different magnetic fields, in particular, result in different signals, i.e., "responses" in this sense. (iii) A magnetic field is generated by a second part of the magnetic field source, and the response of the sensor elements is measured. Preferably, there is a predetermined geometric relationship between the first and second parts of the magnetic field source. Also, there is preferably a predetermined relationship between the strength of the first magnetic field source and the strength of the second magnetic field source. For example, the two magnetic field sources can be in a specific spatial relationship to one another. The two magnetic field sources can also provide magnetic fields whose strengths are in a specific ratio to one another. (iv) Magnetic sensitivities of the sensor elements are determined from the measured data. (v) The magnetic sensitivities can be stored, for example, in coded form in a memory that is optionally assigned to the sensor elements. This can be, for example, an EEPROM on the same chip, or a package insert on which these values ​​are recorded.

[0056] After calibration, the test specimen can be used in its position to measure the electrical current through the conductor, taking into account the determined sensitivities of the sensor elements. Example implementation:

[0057] Fig. 1 shows an exemplary arrangement for calibrating a test object 104. The test object 104 is embodied, for example, as a sensor element in a plastic casing. The test object comprises a chip 106, on which, for example, three sensor elements 101, 102, and 103 are arranged. The chip is arranged, e.g., glued, on a leadframe 105. The test object 104 is placed in a recess of a base plate 109. Embedded in the base plate 109 beneath the recess and thus beneath the test object are a first conductor 110 and, beneath this, a second conductor 111.

[0058] An example is Fig. 1 an xy-plane (coordinate system) is entered. The middle sensor element 102 has a distance y0 from the origin of the coordinate system. The distance between the sensor element 101 and the sensor element 102 is s / 2 and the distance between the sensor element 101 and the sensor element 103 is s. The sensor elements 101 to 103 are located in Fig. 1 on a straight line parallel to the x-axis.

[0059] Conductors 110 and 111 are arranged one above the other, for example, as strip-shaped, parallel conductors. The test object 104, for example, has the same sensor elements as those described in DE 10 2012 110 406 A1. The test object 104 is arranged above the two conductors 110, 111. The conductors 110, 111 are manufactured with high precision, for example, so that their geometry and position (e.g., relative to one another) are known.

[0060] In the Fig. In the example shown in Figure 1, the test specimen 104 lies above the conductor 110 in the recess of the base plate 109.

[0061] The sensor elements 101 to 103 are positioned relatively precisely to both ribbon-shaped conductors (for example, to within a few 1 / 10ths of a millimeter and with a maximum tilt of 5 degrees).

[0062] In addition, there are tolerances that cannot be neglected: For example, the test piece 104 can comprise the chip 106, which has been glued to the frame 105 (leadframe), with the chip 106 and frame 105 being encased in the plastic-like potting compound. The thickness of the chip 106 can, for example, have a tolerance of 5 µm or even up to 20 µm, and an adhesive joint can additionally have a tolerance of 50 µm. Furthermore, the adhesive joint can be wedge-shaped, so that the surface of the chip 106 is not exactly parallel to the surface of the lead 110. The frame 105 can have a thickness tolerance of 10 µm and can possibly be provided with burrs that contribute to the frame 105 being tilted relative to the horizontal (here: the x-axis). Laterally, the chip 106 may be shifted by up to 200µm due to positioning tolerances (e.g.as a result of a so-called pick and place positioning of a machine used for the positioning process). Furthermore, the chip 106 may float during assembly of the chip on the frame due to the adhesive not having fully cured during placement, which causes an additional tolerance in the actual position of the chip 106. The potting compound may be inaccurate in its shape due to a potting process, a subsequent cure-shrink process (i.e., continued crosslinking of the chemical bond structure), swelling due to moisture absorption from the ambient air, etc., and may contribute to further tolerances.

[0063] Thus, - an unknown tilt (ϑ) of the chip 106 against the surface of the conductor 110 and - one unknown offset (δ x , δ y ) in both directions normal and parallel to the surface of the conductor 110.

[0064] The chip 106 comprises, for example, the three sensor elements 101, 102, 103 at three positions which lie on a line and each have a distance s / 2 from the adjacent sensor element.

[0065] The relative positions of the sensor elements 101 to 103 are relatively precise, as they can be manufactured using microelectronics manufacturing techniques. For example, tolerances of less than 1µm can be ensured.

[0066] The sensor elements 101 and 103 detect vertical magnetic fields B y1 , B y3 with the magnetic sensitivities S y1 , S y3 . The sensor element 102 detects horizontal fields B x2 with the magnetic sensitivity S x2 . The sensitivity S x2 for example, significantly (e.g. by 10%) from the sensitivities S y1 or S y3 . The sensitivities S y1 and Sy3 be different, whereby their difference (e.g. by 2%) is, for example, smaller than the difference to the sensitivity S x2 The difference between the sensitivities of the sensor elements 101 and 103 can be referred to as the mismatch.

[0067] This results in a system with a total of six unknowns: ϑ,δx,δy,Sy1,Sx2,Sy3.

[0068] A calibration determines these six unknowns or at least the three sensitivities S y1 , S x2 and S y3 .

[0069] For example, a first current is sent through the first conductor and the signals S1', S2', S3' of the three sensor elements 101, 102, 103 are measured.

[0070] A second current is then sent through the second conductor and the signals S1", S2", S3" of the three sensor elements 101, 102, 103 are measured.

[0071] Thus, there are six measured values ​​with which the six unknowns can be determined.

[0072] Preferably, the ratio of the two currents, i.e., the first current to the second current (or vice versa), is known. Optionally, the ratio of the two currents is equal to one.

[0073] The problem presented here can be defined as follows:

[0074] The signals S1, S2, S3 of the three sensors 101, 102, 103 depend on the magnetic fields at the three locations of the sensors as follows: S1=Sy(1+MM / 2){By1cosϑ+Bx1sinϑ} S2=Sx{−By2sinϑ+Bx2cosϑ}, S3=Sy(1−MM / 2){By3cosϑ+Bx3sinϑ}, where (x1,y1) the location of the sensor element 101 (x2,y2) the location of the sensor element 102, (x3,y3) the location of the sensor element 103 describe.

[0075] If the field components (B x1 ,B y1) measured at the location (x1, y1) of the sensor element 101, the following applies to this location: (x1,y1)=(δx−0.5×scosϑ,y0+δy−0.5×ssinϑ).

[0076] If the field components (B x2 ,B y2 ) measured at the location (x2, y2) of the sensor element 102, the following applies to this location: (x2,y2)=(δx,y0+δy).

[0077] If the field components (B x3 ,B y3 ) measured at the location (x3, y3) of the sensor element 103, the following applies to this location: (x3,y3)=(δx+0.5×scosϑ,y0+δy+0.5×ssinϑ).

[0078] These include: s is the distance between the sensor element 101 and the sensor element 103; s / 2 is the distance between the sensor element 101 and the sensor element 102 or the distance between the sensor element 102 and the sensor element 103; y0 a so-called nominal y-position of the chip surface; S xthe magnetic sensitivity of the vertical Hall sensor to the magnetic fields in the x-direction (B x ); S y (1+MM / 2) the magnetic sensitivity of the sensor element 101 (horizontal Hall sensor element) to magnetic fields in the y-direction (B y ); S y (1-MM / 2) the magnetic sensitivity of the sensor element 103 (horizontal Hall sensor element) to magnetic fields in the y-direction (B y ); MM a mismatch between the two sensor elements 101 and 103.

[0079] In this example, the sensor element 101 and the sensor element 103 are a sensor element of the first type (e.g. a horizontal Hall sensor) and the sensor element 102 is a sensor element of the second type (e.g. a vertical Hall sensor).

[0080] The calibration device explained here can be used advantageously, especially when the test specimen is part of a strip.

[0081] For example, an arrangement of 15 columns and 5 rows can be used, in which the sensor packages (chips with sensor elements) are arranged on a strip. This strip first consists of a structured copper sheet, the leadframe. The chips are glued to this, and then the connections between bond pads on the chips and leads or pins on the leadframe are created using bond wires. The entire strip is then placed in a mold tool, where the chips and bond wires are overmolded or potted with plastic compound. The connections (with the possible exception of the ground pins) are punched out, but the sensor packages remain attached to the frame by bars (known as dam bars).

[0082] The entire frame is electrically tested by a test device (a so-called "in-strip tester"). A robotic arm, for example, places the strip on a base plate, and fine needles on a contact card contact the terminal pins of one or more components. Predefined electrical parameters (e.g., current consumption at various voltages) are then measured.

[0083] In this case, the two current conductors are advantageously accommodated in the base plate of the test device. The robot arm can then place the strip on the base plate with considerable tolerance. The base plate can also have a profile that defines the rough positioning of the components on the so-called chuck or plunger head (also referred to as the base plate). The base plate is in particular a base plate that is laterally movable. The base plate can be moved to a loading position, where, for example, a gripper places the pre-tempered leadframe strip onto the plate. The base plate then moves to a contact position where the tests for the devices to be tested are carried out. The base plate is then moved to an unloading position, where a gripper lifts the tested leadframe strip and places it in a removal cassette.

[0084] As shown above, finer positioning tolerances are calculated using the 2 currents and 6 measured variables of the 3 sensor elements and can thus be reduced or eliminated in the subsequent application of the respective current sensor.

[0085] The conductors can be arranged in rows or columns beneath the strip, preferably in such a way that their number is minimized. In particular, switches or relays can be provided that selectively send the current via the first or second conductor (alternatively, the current can always be sent via both conductors, but the direction of current flow in one of the two can be selectively switched). Optionally, several devices under test (DUTs) can be subjected to a magnetic field simultaneously or essentially simultaneously if they are located over the same conductors. For this purpose, the conductor can be shaped, for example, in a serpentine or spiral form, so that more than one row or column of the matrix-arranged DUTs can be subjected to the magnetic field essentially together. In one example, all or a selection of the DUTs can be calibrated simultaneously.

[0086] Fig. 2 shows based on the representation of Fig. 1 an exemplary sketch to illustrate the positioning tolerances ϑ, δ x , δ y . In Fig. 2 again shows the test piece 104, which comprises the chip 106 with sensor elements 101 to 103 and the frame 105 in a plastic sheath. The test piece 104 is positioned in a recess of the base plate 109, which again surrounds the two conductors 110 and 111. Fig. 2 shows the xy coordinate system.

[0087] The conductors 110 and 111 have - as in Fig. 1 - has a surface that is parallel to the x-axis. In contrast to Fig. 1, the frame 105 is shown slightly inclined (tilted) relative to the x-axis and the chip 106 is shown more inclined relative to the x-axis. This results in a y-offset δ for the chip 106 and the sensor elements 101 to 103 positioned on the chip 106. y , an x-offset δ xand a tilt angle ϑ.

[0088] For example, instead of measuring a sensor signal for a certain current or magnetic field, a signal change resulting from a change in current or magnetic field can be measured. This allows static background or interference fields (DC components) to be eliminated. Optionally, the current direction can be inverted for this purpose. Another option is to temporally integrate and / or average the measurement over an integer multiple of the period of the dominant AC disturbance (e.g., the mains frequency).

[0089] If the positioning tolerances ϑ, δ x , δ y are sufficiently small (e.g. ϑ less than 1 degree, δ x and δ y each smaller than 50µm), they can be set to zero. In this case, the current through one conductor is sufficient to obtain 3 measured values ​​from the 3 sensors and from these the 3 unknowns Sx , MM, S y to calculate.

[0090] If the deviation MM of the y-sensors is sufficiently small (e.g., MM less than 0.4%), it can be set to zero. Then, measuring the signal combination S1-S3 or S1+S3 per current is sufficient, instead of measuring both signals S1 and S3 (i.e., one measurement can be saved).

Claims

[1] Method for calibrating a sensor unit, - wherein the sensor unit comprises: - a sensor housing, - a first magnetic sensor, - a second magnetic sensor and - a third magnetic sensor, - wherein the first magnetic sensor, the second magnetic sensor and the third magnetic sensor are arranged on a semiconductor chip, - wherein the sensor unit is arranged on a device during calibration, the device comprising a first conductor and a second conductor, - wherein a first magnetic field is generated by the first current-carrying conductor and a second magnetic field is generated by the second current-carrying conductor by a magnetic field generating device, wherein the first current-carrying conductor and the second current-carrying conductor have a different distance from the sensor unit and / or a different orientation to the sensor unit, - wherein the first magnetic sensor and the third magnetic sensor are each configured to detect magnetic field components in a first direction, - wherein the second magnetic sensor is configured to detect magnetic field components in a second direction, - where the first direction is not parallel to the second direction, - wherein the calibration of the sensor unit is carried out based on responses of the magnetic sensors to the magnetic field by - a first current is sent through the first conductor and the first signals from the three sensors are measured, - a second current is sent through the second conductor and second signals from the three sensors are measured, - whereby the magnetic sensitivities of the three sensors are determined based on the first signals and the second signals. [2] The method of claim 1, wherein the at least one magnetic field comprises at least two magnetic fields whose spatial distribution is such that the ratio of the responses of the first magnetic sensor to both magnetic fields is unequal to the ratio of the responses of the second magnetic sensor to both magnetic fields. [3] A method according to claim 2, wherein the calibration of the sensor unit is determined from the responses of the magnetic sensors as magnetic sensitivities of the magnetic sensors. [4] A method according to any one of the preceding claims, wherein the ratio of the first current to the second current is known. [5] Method according to one of the preceding claims, in which the magnetic field generating device is designed separately from the sensor housing. [6] Method according to one of claims 1 to 4, wherein the magnetic field generating device is arranged in the sensor housing. [7] Method according to one of the preceding claims, in which the magnetic field generating device generates at least two magnetic fields of different strengths which act on at least one of the magnetic sensors. [8] Method according to one of the preceding claims, wherein the first magnetic sensor and the second magnetic sensor are spaced apart by a predetermined distance. [9] Method according to one of the preceding claims, wherein the sensor unit is a current sensor. [10] A method according to any one of the preceding claims, wherein the magnetic field generating means generates a magnetic field which is substantially proportional to an electric current. [11] Method according to one of the preceding claims, wherein the magnetic field generating device generates at least two magnetic fields which act on at least one of the magnetic sensors. [12] Method according to one of the preceding claims, in which at least one of the two magnetic fields acts inhomogeneously on the at least one magnetic sensor. [13] A method according to any one of the preceding claims, wherein the first direction is substantially perpendicular to the second direction. [14] Method according to one of the preceding claims, wherein the at least one magnetic field generated by the magnetic field generating device comprises an inhomogeneous magnetic field on at least one sensor element. [15] A method according to any one of the preceding claims, wherein the first magnetic sensor and the second magnetic sensor are sensors of different types. [16] The method of claim 15, wherein the first magnetic sensor and the second magnetic sensor are Hall sensors that detect magnetic field components acting in different directions. [17] Method according to one of the preceding claims, wherein the first magnetic sensor and the third magnetic sensor are sensors of the same type. [18] Method according to one of the preceding claims, in which the at least one magnetic field comprises at least two magnetic fields whose spatial distribution is such that the ratio of the responses of the first magnetic sensor to both magnetic fields is not equal to the ratio of the responses of the second magnetic sensor to both magnetic fields. [19] Method according to one of the preceding claims, - wherein the sensor unit further comprises: - a frame on which the semiconductor chip is arranged, - the sensor housing is a plastic housing. [20] Method according to one of the preceding claims, in which the semiconductor chip is glued to the frame. [21] Method according to one of the preceding claims, wherein the sensor housing is arranged on a base plate. [22] Method according to one of the preceding claims, wherein at least one magnetic sensor has a sensitivity substantially in a direction perpendicular to the semiconductor chip.

Images