JP5374422B2 - Magnetic field detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a magnetic field detector for accurately calculating the rotary motion and angular velocity of a magnetic vector while reducing the effect of noise by using a triaxial sensor for detecting magnetism such as geomagnetism. <P>SOLUTION: A magnetic vector is found as coordinate point data on a spherical coordinate G based on detection outputs from three magnetic sensors for magnetic vector detection. A regression plane P with the least residual error is found while taking into account computed variance and mean of a plurality of groups of coordinate point data D1, D2, D3, D4, ..., to assume an intersecting line of the regression plane P with the spherical coordinate G to be a movement locus circle Cr. An opening angle from the center of the locus circle Cr to two groups of coordinate point data is differentiated in terms of time, allowing angular velocity to be acquired. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a magnetic field detection device that detects a magnetic vector with a magnetic sensor directed in each of three orthogonal directions, and more particularly to a magnetic field detection device that can reduce the influence of noise and know the rotational motion of a magnetic vector. .

  A magnetic field detection device using a triaxial magnetic sensor that detects magnetic field strengths in three directions orthogonal to each other is used as a geomagnetic detection device that detects geomagnetism.

  The magnetic gyro described in Patent Document 1 has a three-axis magnetic sensor that detects geomagnetism arranged on three-axis orthogonal coordinates. When this magnetic gyroscope is rotated in a three-dimensional space, a difference vector between two different time points is obtained using output data of three axes, and the difference vector becomes smaller than a predetermined threshold value. It is determined whether or not one of the three axes is rotating as a center.

  The magnetic gyro described in Patent Document 1 can detect the rotation state when rotating around any of the three axes determined by the orientation of the magnetic sensor. When it is rotated about the rotation axis, the rotation axis cannot be recognized, and it cannot be specified in which rotation plane it is rotating. That is, the angular velocity when rotating around an arbitrary axis in the three-dimensional space cannot be detected with only one magnetic gyro described in Patent Document 1.

  Patent Document 2 discloses an attitude sensor mounted on an airplane or the like. This attitude sensor has a geomagnetic detection device, and is provided with a load weight and a force detection device that detects gravity acting on the load weight. When the attitude sensor is tilted with an airplane, etc., the inclination with respect to the direction of gravity is detected by the detection output of the force detection device, and the azimuth output obtained by the geomagnetism detection device is used as information on the inclination posture obtained by the force detection device. To correct.

  The posture sensor described in Patent Document 2 includes not only a geomagnetic detection device but also a load weight and a force detection device that detects gravity acting on the load weight. It is difficult to install in small equipment for the purpose.

  The three-axis attitude detection device described in Patent Document 3 detects the attitude of a target object, and is equipped with both a magnetic sensor capable of detecting in three directions and a gyro sensor capable of detecting in three directions. Has been. For this reason, it is not suitable for mounting on a portable small device or the like, and both the magnetic sensor and the gyro sensor are mounted.

JP 2008-224642 A JP-A-2-238336 JP 11-248456 A

  The present invention solves the above-described conventional problems, and can estimate a moving locus circle of a magnetic vector from a detection output of a magnetic sensor directed in three orthogonal directions, and reduce the influence of noise to move the moving locus circle. It is an object of the present invention to provide a magnetic field detection apparatus capable of accurately estimating the magnetic field.

  It is another object of the present invention to provide a magnetic field detection apparatus that can accurately estimate a movement locus circle with a small number of data.

The present invention includes a magnetic detection unit in which the X direction, the Y direction, and the Z direction orthogonal to each other are determined as reference directions, and a calculation unit.
An X-axis sensor in which the absolute value of the detection output becomes a maximum value when the X direction is directed to the magnetic direction and the absolute value of the detection output when the Y direction is directed to the magnetic direction. Is equipped with a Y-axis sensor in which the maximum value is detected, and a Z-axis sensor in which the absolute value of the detection output is a maximum value when the Z direction is directed to the magnetic direction,
In the calculation unit, magnetic vectors obtained from a plurality of the detection outputs are converted into coordinate point data on spherical coordinates, a regression plane is obtained from the plurality of coordinate point data, and an intersection between the regression plane and the spherical coordinates is obtained. The line is a moving locus circle.

  In the present invention, the plane equation in the three-dimensional coordinates of XYZ is Y = A · X + B · Z + C, and the constants A and B are the variance Sx of x, the variance Sy of y, the variance Sz of z, and The covariance Sxy of x and y, the covariance Sxz of x and z, the covariance Syz of y and z, the constant C is obtained from the average xa of x, the average ya of y, and the average za of z, and the above equation From this, the regression plane is obtained.

  In the present invention, a regression plane is calculated in consideration of the dispersion tendency and the average of the dispersion of a plurality of coordinate point data on the three-dimensional coordinates, and the intersection line between the regression plane and the spherical coordinates is used as a movement locus circle. Since the distribution tendency of a plurality of coordinate point data is statistically taken into consideration, the obtained regression plane and moving locus circle are close to the actual tendency of the rotating locus circle. That is, the rotation locus circle can be estimated while reducing the influence of noise. Further, even if the number of data is small, a regression plane with a small error can be obtained, and the movement locus circle can be estimated from the small number of data.

  In the present invention, the angular velocity is obtained from the opening angle of a plurality of coordinate point data from the center of the movement locus circle and the sample time of the coordinate data.

  In this case, a plurality of coordinate point data is projected onto the movement locus circle, and the angular velocity is obtained from the opening angle of the projection point from the center of the movement locus circle and the sampling time of the coordinate data.

  By projecting the coordinate point data onto the moving trajectory circle as described above, the opening angle of the coordinate point data can be obtained as being less influenced by noise and less error.

  According to the present invention, the relative rotational motion of the magnetic vector can be estimated by using the detection output of the magnetic sensor directed in the orthogonal three-axis directions. In addition, since the regression plane is obtained by taking into account the variance and average of a plurality of coordinate data, the influence of noise is small, and even with a small number of data, the movement locus circle can be obtained with high accuracy.

The circuit block diagram of the magnetic field detection apparatus of embodiment of this invention, Illustration of data buffer, Explanatory drawing of the X-axis sensor, the Y-axis sensor, and the Z-axis sensor provided in the magnetic detection unit, Explanatory diagram of 3D coordinates showing the detection principle of geomagnetic vector, A perspective view showing the regression plane and spherical coordinates; Explanatory drawing explaining the calculation method of angular velocity,

  A magnetic field detection apparatus 1 according to an embodiment of the present invention shown in FIG. 1 has a magnetic detection unit 2 that detects geomagnetism. As shown in FIG. 3, in the magnetic detection unit 2, the X axis, the Y axis, and the Z axis, which are reference axes orthogonal to each other, are determined as fixed axes. The magnetic field detection device 1 is mounted on a portable device or the like, and the magnetic detection unit 2 can freely move in the space while maintaining the orthogonal relationship between the X axis, the Y axis, and the Z axis.

  As shown in FIG. 3, the X-axis sensor 3 is fixed along the X-axis, the Y-axis sensor 4 is fixed along the Y-axis, and the Z-axis sensor is fixed along the Z-axis. Has been. The X-axis sensor 3, the Y-axis sensor 4, and the Z-axis sensor 5 are all composed of GMR elements. The GMR element is composed of a pinned magnetic layer and a free magnetic layer made of a soft magnetic material such as a Ni—Co alloy or a Ni—Fe alloy, and a nonmagnetic material such as copper sandwiched between the pinned magnetic layer and the free magnetic layer. And a conductive layer. An antiferromagnetic layer is laminated under the pinned magnetic layer, and the magnetization of the pinned magnetic layer is pinned by antiferromagnetic coupling between the antiferromagnetic layer and the pinned magnetic layer.

  The X-axis sensor 3 detects a component of the geomagnetism facing in the X direction, and the magnetization direction of the pinned magnetic layer is fixed in the PX direction along the X axis. The direction of magnetization of the free magnetic layer responds to the direction of geomagnetism. When the magnetization direction of the free magnetic layer is parallel to the PX direction, the resistance value of the X-axis sensor 3 is minimized, and when the magnetization direction of the free magnetic layer is opposite to the PX direction, the resistance value of the X-axis sensor 3 is maximized. become. Further, when the magnetization direction of the free magnetic layer is orthogonal to the PX direction, the resistance value is an average value of the maximum value and the minimum value.

  In the magnetic field data detection unit 6 shown in FIG. 1, the X-axis sensor 3 and the fixed resistance are connected in series, and a voltage is applied to the series circuit of the X-axis sensor 3 and the fixed resistance. The potential between the resistor and the resistor is taken out as an X axis detection output. When a magnetic field directed in the X direction is not applied to the X-axis sensor 3, or when a magnetic field orthogonal to PX is applied, the X-axis detection output becomes a midpoint potential.

  When the entire magnetic detection unit 2 is tilted and the fixed direction PX of the magnetization of the fixed magnetic layer of the X-axis sensor 3 is set in the same direction as the geomagnetic vector V, the magnetic field component applied to the X-axis sensor 3 becomes a maximum value. The X-axis detection output at this time has a maximum value on the plus side with respect to the midpoint potential. Conversely, when the fixed direction PX of the magnetization of the fixed magnetic layer of the X-axis sensor 3 is directed opposite to the geomagnetic vector V, the reverse magnetic field component applied to the X-axis sensor 3 has a maximum value. At this time, the detected output of the X axis has a maximum value on the minus side with respect to the midpoint potential.

  Each of the Y-axis sensor 4 and the Z-axis sensor 5 is also connected to a fixed resistor in series, and a voltage is applied to the Y-axis sensor 4 or a series circuit of the Z-axis sensor 5 and the fixed resistor. Is taken out as a Y-axis or Z-axis detection output.

  When the fixed direction PY of the magnetization of the fixed magnetic layer of the Y-axis sensor 4 is set in the same direction as the geomagnetic vector V, the detected output of the Y-axis becomes a maximum value on the plus side with respect to the midpoint potential. When the fixed direction PY of the magnetization of the fixed magnetic layer of the Y-axis sensor 4 is directed opposite to the geomagnetic vector V, the detection output of the Y-axis becomes a negative maximum value with respect to the midpoint potential. Similarly, if the fixed direction PZ of the magnetization of the fixed magnetic layer of the Z-axis sensor 5 is set in the same direction as the geomagnetic vector V, the detection output of the Z-axis becomes a maximum value on the plus side with respect to the midpoint potential. When the fixed direction PZ of the magnetization of the fixed magnetic layer of the Z-axis sensor 5 is directed opposite to the geomagnetic vector V, the detection output of the Z-axis becomes a negative maximum value with respect to the midpoint potential.

  If the magnitude of the geomagnetic vector V is constant, the detection outputs from the X-axis sensor 3, the Y-axis sensor 4, and the Z-axis sensor 5 are all of the absolute value of the plus side maximum value and the minus side maximum value. The absolute value is the same.

  As the X-axis sensor 3, a positive detection output and a negative detection output are obtained depending on the direction of the geomagnetic vector, and the absolute value is the same between the maximum value of the positive detection output and the maximum value of the negative detection output. If it becomes, it can also be comprised with magnetic sensors other than a GMR element. For example, a Hall element or MR element that can detect only the positive magnetic field intensity along the X axis and a Hall element or MR element that can detect only the negative magnetic field intensity may be combined and used as the X axis sensor 3. Good. The same applies to the Y-axis sensor 4 and the Z-axis sensor 5.

  As shown in FIG. 1, the detection outputs of the X axis, the Y axis, and the Z axis detected by the magnetic field data detection unit 6 are given to the calculation unit 10. The arithmetic unit 10 includes an A / D converter, a CPU, a clock circuit, and the like. According to the measurement time of the clock circuit of the calculation unit 10, the detection outputs of the X axis, the Y axis, and the Z axis detected by the magnetic field data detection unit 6 are intermittently sampled in a short cycle and read out to the calculation unit 10. . Each detection output is converted into a digital value by the A / D conversion unit provided in the calculation unit.

  A memory 7 is connected to the CPU constituting the arithmetic unit 10. In the memory 7, software for arithmetic processing is programmed and stored. The arithmetic processing of the arithmetic unit 10 is executed by the software.

  The X axis detection output, the Y axis detection output, and the Z axis detection output converted into digital data are subjected to calculation processing by the calculation unit 10 and are coordinate points on the three-dimensional coordinates of XYZ shown in FIG. Data D (xb, yb, zb) is converted and stored in a data buffer (buffer memory) 11 provided in the arithmetic unit 10. The coordinate point data D sampled and calculated in a short cycle in synchronization with the clock circuit is given to the storage unit 11a of the data buffer 11 shown in FIG. Each time the coordinate point data D is given to the storage unit 11a, the coordinate point data D is sequentially sent from the storage unit 11a to 11m, and the oldest coordinate point data D stored in the final storage unit 11m is discarded. While the magnetic field detection device 1 is operating, the latest data is continuously read from the magnetic field data detection unit 6 at regular intervals, and the coordinate point data D after calculation is sequentially stored in the data buffer 11.

  As shown in FIG. 4, XYZ three-dimensional coordinates are set in the calculation unit 10 on the software. The directions of the XYZ axes set in the calculation unit 10 coincide with the directions of the XYZ axes fixed and set in the magnetic field detection device 1 shown in FIG.

  As shown in FIG. 4, when the magnetic detection unit 2 is placed anywhere on the earth, the detection output xb is obtained from the X-axis sensor 3 of the magnetic detection unit 2, and the detection output yb is obtained from the Y-axis sensor 4. The detection output zb is obtained from the Z-axis sensor 5. In the calculation unit 10, coordinate point data D (xb, yb, zb) is calculated from the three-dimensional detection output. The coordinate point data D (xb, yb, zb) are obtained one after another for each sampling period, and are sequentially stored in the data buffer 11.

  The magnetic field detector 1 is calibrated immediately after the power is turned on or at the start of use. The calibration is instructed by the display on the display of the portable device in which the magnetic field detection device 1 is mounted, and is executed by the user according to the instruction.

  Calibration is performed by rotating a portable device or the like equipped with the magnetic field detection device 1 several times in an arbitrary direction. The arithmetic unit 10 samples some of the coordinate point data D obtained one after another in the calibration. By obtaining at least three coordinate point data D, it is possible to specify a circle that matches the rotation locus of the coordinate point data D at that time. A plurality of circles are obtained, a center line that passes through the center of each circle and is perpendicular to the plane including the circle is obtained, and the intersection of these center lines is calculated. In the calculation unit 10, the intersection obtained as a result of the calibration is corrected so as to become the reference origin O of the three-dimensional coordinates.

  When the reference origin O of the three-dimensional coordinates is corrected by calibration, the coordinate point data D (xb, yb, zb) calculated from the subsequent detection output is the reference on the three-dimensional coordinates as shown in FIG. It appears as a point on the spherical coordinate G with the origin O as the center. The radius R of the spherical coordinate G is proportional to the absolute value of the maximum value of the detection output obtained from the X-axis sensor 3, the Y-axis sensor 4, and the Z-axis sensor 5. The radius R of the spherical coordinate G differs depending on the measurement location at that time, and the radius R of the spherical coordinate G also changes depending on the magnitude of the detected absolute value of the geomagnetic vector V.

  FIG. 4 shows a state where the Z axis set in the magnetic field detection device 1 is directed in the direction of gravity. A line connecting the coordinate point data D (xb, yb, zb) obtained from the detection output and the reference origin O of the three-dimensional coordinates is the geomagnetic vector V. When the magnetic field detector 1 is rotated about the Z axis (centered about the axis facing the direction of gravity), the coordinate point data D (xb, yb, zb) sampled one after another is centered on the Z axis and X -Move along a horizontal latitude line Ha parallel to the Y axis. If the calculation unit 10 recognizes that the direction of the reference axis Z is directed in the direction of gravity, the opening of a plurality of coordinate point data D centering on the intersection of the horizontal latitude line Ha and the Z axis is opened. The angular velocity can be obtained from the angle and the sampling time.

  However, when the mobile device or the like actually mounted with the magnetic field detection device 1 is rotated in an arbitrary direction in the three-dimensional space, the direction of the axis of the rotation trajectory is determined only from the coordinate point data D, It is impossible to recognize where the coordinates of the center of the rotation locus are. As a result, the angular velocity cannot be calculated.

  Therefore, in the calculation unit 10 shown in FIG. 2, the following software is executed so that when the magnetic field detection device 1 is rotated around the axis in an arbitrary direction, the coordinate point data is along which movement locus circle. It is possible to estimate whether it is moving by calculation.

  As shown in FIG. 5, when the magnetic field detection device 1 rotates around an arbitrary axis other than the XYZ axes, coordinate point data D1, D2,..., Dn-1, Dn appears. The coordinate point data D1 is the latest data, and becomes the old data dating back in the order of D2, D3, D4,... Dn−1, Dn. The latest coordinate point data D1 is stored in the storage unit 11a of the data buffer 11 shown in FIG. 2, and the coordinate point data D2, D3, D4,... Are stored in the storage units 11b, 11c,. The

  The calculation unit 10 uses a plurality of coordinate point data D1, D2, D3,..., Dn-1, Dn stored in the data buffer 11, and considers the distribution tendency and average of these coordinate point data. Then, a regression plane P that minimizes an error from each coordinate point data is obtained. A circle that is an intersection line of the regression plane P and the spherical coordinate G is estimated as a movement locus circle Cr.

  When the equation of the regression plane P in the three-dimensional coordinates of XYZ shown in FIG. 5 is Y = A · X + B · Z + C, the constants A and B are the variance Sx of x, the variance Sy of y, and the variance of z Sz, x and y covariance Sxy, x and z covariance Sxz, y and z covariance Syz, and constant C is obtained from x average xa, y average ya, and z average za It is done.

  The coordinates of the coordinate point data D1, D2, D3,..., Dn-1, Dn used for the calculation are (x1, y1, z1), (x2, y2, z2), (x3, y3, z3), respectively. ,... (Xn-1, yn-1, zn-1), (xn, yn, zn), and the coordinates of the spherical coordinate G are (0, 0, 0). The coordinate point data D1, D2, D3,..., Dn−1, Dn used for the calculation may use all data stored in a certain amount in the buffer memory 11 shown in FIG. Any one of them may be selected and used.

The average xa of x, the average ya of y, and the average za of z are obtained from the following formula 1.

The variance Sx of x, the variance Sy of y, and the variance Sz of z are obtained from the following formula 2.

  The covariance Sxy of x and y, the covariance Sxz of x and z, and the covariance Syz of y and z are obtained from the following Equation 3.

  Constants A and B of the equation Y = A · X + B · Z + C on the regression plane P are obtained from the following equation (4).

The constant C of the equation Y = A · X + B · Z + C of the regression plane P is obtained from the following equation (5).

Next, from the equation Y = A · X + B · Z + C of the regression plane P and the equation X ² + Y ² + C ² = R ² (R is a radius) of the spherical coordinate G, the intersection line of the regression plane P and the spherical coordinate G And the circle of this intersection is estimated as the movement locus circle Cr.

  When a portable device equipped with the magnetic field detection device 1 rotates around a certain axis, the coordinate point data D1, D2, D3,..., Dn−1, Dn obtained during this time are correlated. Yes, the residual between the actual movement locus circle and each coordinate point data substantially follows a normal distribution, and the average value of the residual is very close to zero. Therefore, the regression plane P calculated from the variance, covariance and average of a plurality of coordinate point data is close to the tendency of the rotation locus of the geomagnetic vector V, and the influence of noise superimposed on each coordinate point data is affected. It will be less. Therefore, the movement trajectory circle Cr is also less affected by noise and is highly accurate in accordance with the actual movement tendency.

  As shown in FIG. 6, when the movement locus circle Cr is obtained, its center M can be obtained. The angular velocity is obtained by selecting any two coordinate point data Da and Db from the plurality of coordinate point data and differentiating the opening angle from the center M with the sampling time of the two coordinate point data Da and Db. be able to.

  At this time, the opening angle from the center M of the two coordinate point data Da and Db may be obtained by using the two coordinate point data Da and Db as they are. However, as shown in FIG. 6, the projection points Dav and Dbv projected perpendicularly on the regression plane P from the coordinate point data Da and Db are obtained, and the opening angle θ of the projection points Dav and Dbv is differentiated by the sampling time to obtain the angular velocity. Is preferably obtained.

The coordinates (xm, ym, zm) of the center M of the movement locus circle Cr are obtained by the following equation (6).

  From the coordinates, a center point obtained by projecting the center M onto the XY plane coordinates, the XZ plane coordinates, and the YZ plane coordinates is obtained. Similarly, projection points obtained by projecting the two coordinate point data Da and Db onto the XY plane coordinates, the XZ plane coordinates, and the YZ plane coordinates are obtained. From the projected center point and the projection points of the two coordinate point data, the angular velocity around the X axis, the angular velocity around the Y axis, and the angular velocity around the Z axis can be respectively obtained.

  As described above, when the projection points Dav and Dbv projected perpendicularly to the regression plane P from the coordinate point data Da and Db are obtained, and the angular velocity is obtained by differentiating the opening angle θ of the projection points Dav and Dbv with the sampling time, Since the moving angular velocity in a circle having the same tendency as the actual moving trajectory circle can be obtained, an angular velocity with less influence of noise can be calculated.

  Further, even if the number of samples of the coordinate point data D is relatively small, it is possible to estimate the movement locus circle in accordance with the actual movement from the tendency of dispersion, so that the calculation load is small and the magnetic field detection device 1 is slightly rotated. The angular velocity at that time can be calculated quickly.

  The magnetic field detection apparatus of the present invention can calculate the angular velocity by specifying the rotational motion using the coordinate point data from the triaxial magnetic sensor. Therefore, it can be used for a portable game device or an input device of a game device. It can also be used as a detection unit that detects changes in posture of the robot's arms and joints.

  Furthermore, the magnetic field detection device of the present invention can be used as a device for detecting the motion of a magnetic vector of an external magnetic field other than geomagnetism. For example, it is possible to fix the magnetic detection device and detect what direction and how the external magnetic vector moves.

DESCRIPTION OF SYMBOLS 1 Magnetic field detection apparatus 2 Magnetic detection part 3 X-axis sensor 4 Y-axis sensor 5 Z-axis sensor 6 Magnetic field data detection part 7 Memory 10 Calculation part 11 Data buffer D1, D2, D3, D4, ..., Dn Coordinate point data G Spherical coordinate P Regression plane Cr Movement locus circle

Claims (4)

A magnetic detection unit in which the X direction, the Y direction, and the Z direction orthogonal to each other are determined as reference directions, and a calculation unit;
An X-axis sensor in which the absolute value of the detection output becomes a maximum value when the X direction is directed to the magnetic direction and the absolute value of the detection output when the Y direction is directed to the magnetic direction. Is equipped with a Y-axis sensor in which the maximum value is detected, and a Z-axis sensor in which the absolute value of the detection output is a maximum value when the Z direction is directed to the magnetic direction,
In the calculation unit, magnetic vectors obtained from a plurality of the detection outputs are converted into coordinate point data on spherical coordinates, a regression plane is obtained from the plurality of coordinate point data, and an intersection between the regression plane and the spherical coordinates is obtained. A magnetic field detection device characterized in that a line is a moving locus circle.   The equation of the plane in the three-dimensional coordinates of XYZ is Y = A · X + B · Z + C, and the constants A and B are the variance Sx of x, the variance Sy of y, the variance Sz of z, and the variance of x and y. The covariance Sxy, the covariance Sxz of x and z, the covariance Syz of y and z, the constant C is obtained from the average xa of x, the average ya of y, the average za of z, and the regression plane is calculated from the above equation. The magnetic field detection apparatus according to claim 1 to be obtained.   The magnetic field detection device according to claim 1, wherein an angular velocity is obtained from an opening angle of a plurality of coordinate point data from the center of the movement locus circle and a sampling time of the coordinate data.   The magnetic field detection apparatus according to claim 3, wherein a plurality of coordinate point data is projected onto the movement locus circle, and an angular velocity is obtained from an opening angle of the projection point from the center of the movement locus circle and a sampling time of the coordinate data.

Images