JP2014219364A - Rotation angle detection apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a rotation angle detection apparatus capable of detecting a rotation angle with high accuracy.SOLUTION: An angle interval between a first magnetic sensor 71 and a second magnetic sensor 72, and an angle interval between the second magnetic sensor 72 and a third magnetic sensor 73 are 60° in electric angle, respectively. In a case where the two magnetic sensors including the second magnetic sensor 72 among the first, second and third magnetic sensors 71, 72 and 73 are normal, when both of the two magnetic sensors including the second magnetic sensor 72 detect the same one magnetic pole consecutively over prescribed plural sampling periods, a rotation angle is calculated based on output signals obtained in the prescribed plural sampling periods of the two magnetic sensors. In a case where only the second magnetic sensor 72 is defective, when both of the first and third magnetic sensors 71, 73 detect the same one magnetic pole consecutively over the prescribed plural sampling periods, a rotation angle is calculated based on output signals obtained in the prescribed plural sampling periods of the two magnetic sensors.

Description

  The present invention relates to a rotation angle detection device that detects a rotation angle of a rotating body.

  As a rotation angle detection device that detects the rotation angle of a rotating body, a rotation angle detection device that detects a rotation angle of a rotor of a brushless motor using a detection rotor that rotates in accordance with the rotation of the brushless motor is known. Specifically, as shown in FIG. 17, the detection rotor 201 (hereinafter referred to as “rotor 201”) has a cylindrical shape having a plurality of magnetic pole pairs corresponding to the magnetic pole pairs provided in the rotor of the brushless motor. A magnet 202 is provided. Two magnetic sensors 221 and 222 are disposed around the rotor 201 at a predetermined angular interval with the rotation center axis of the rotor 201 as the center. Each magnetic sensor 221 and 222 outputs a sine wave signal having a predetermined phase difference. Based on these two sine wave signals, the rotation angle of the rotor 201 (rotation angle of the rotor of the brushless motor) is detected.

  In this example, the magnet 202 has five pairs of magnetic poles. That is, the magnet 102 has ten magnetic poles arranged at equiangular intervals. The magnetic poles are arranged at an angular interval of 36 ° (180 ° in electrical angle) around the rotation center axis of the rotor 201. The two magnetic sensors 221 and 222 are arranged at an angular interval of 18 ° (90 ° in electrical angle) with the rotation center axis of the rotor 201 as the center.

A direction indicated by an arrow in FIG. 17 is a positive rotation direction of the detection rotor 201. When the rotor 201 is rotated in the forward direction, the rotation angle of the rotor 201 is increased. When the rotor 201 is rotated in the reverse direction, the rotation angle of the rotor 201 is decreased. As shown in FIG. 18, from each magnetic sensor 221, 222, a sine wave signal having a period during which the rotor 201 rotates an angle corresponding to one magnetic pole pair (72 ° (360 ° in electrical angle)) is one cycle. S ₁ and S ₂ are output.

The angle range for one rotation of the rotor 201 is divided into five sections corresponding to the five magnetic pole pairs, and the rotation angle of the rotor 201 expressed as the start position of each section is 0 ° and the end position is 360 ° The electric angle θ of 201 is assumed.
Here, an output signal of S ₁ = A ₁ · sin θ is output from the first magnetic sensor 221, and an output signal of S ₂ = A ₂ · cos θ is output from the second magnetic sensor 222. . A ₁ and A ₂ are amplitudes. Assuming that the amplitudes A ₁ and A _{2 of the} two output signals S ₁ and S ₂ are equal to each other, the electrical angle θ of the rotor 201 can be obtained based on the following equation using both the output signals S ₁ and S _2. it can.

θ = tan ⁻¹ (sinθ / cosθ)
= Tan ⁻¹ (S ₁ / S ₂ )
In this way, the brushless motor is controlled using the obtained electrical angle θ.

JP 2008-26297 A

In the conventional rotation angle detection device as described above, the rotation angle θ is calculated on the assumption that the amplitudes A ₁ and A ₂ of the output signals S ₁ and S ₂ of the magnetic sensors 221 and 222 are equal. The amplitudes A ₁ and A ₂ of the output signals S ₁ and S ₂ change according to variations in temperature characteristics of the magnetic sensors 221 and 222 and changes in temperature. For this reason, an error occurs in detection of the rotation angle of the rotor due to variations in temperature characteristics and temperature changes of both magnetic sensors 221 and 222.

  An object of the present invention is to provide a rotation angle detection device capable of detecting a rotation angle with high accuracy.

  The invention according to claim 1 is a rotation angle calculation device (77A) for calculating the rotation angle of the rotating body (8), which rotates in accordance with the rotation of the rotating body and has a plurality of magnetic poles. A magnet (61) and three magnetic sensors (71, 72, 73) that are arranged side by side in the circumferential direction of the multipolar magnet and each output a sine wave signal having a phase difference according to the rotation of the multipolar magnet. Of these three magnetic sensors, an angular interval between the central magnetic sensor and the other magnetic sensor, and an angular interval between the central magnetic sensor and the other magnetic sensor, Of three magnetic sensors whose electrical angle is less than 180 degrees, failure determination means (77A, S1) for determining whether or not a failure has occurred in each of the magnetic sensors, and among the three magnetic sensors At least two magnets including a central magnetic sensor In the case where the sensor is normal, when two normal magnetic sensors including the central magnetic sensor satisfy the first condition that both of them detect the same magnetic pole continuously for a predetermined sampling period, those sensors First calculation means (77A, S84, S85, S88, S89) for calculating the rotation angle of the rotating body based on the output signals of the predetermined plural samplings of the two magnetic sensors, and among the three magnetic sensors In the case where only the center magnetic sensor of the sensor fails, the second condition that both normal two magnetic sensors except the center magnetic sensor detect the same magnetic pole continuously for a predetermined number of sampling periods is satisfied. The rotation angle of the rotating body is calculated based on the output signals for the predetermined multiple samplings of the two magnetic sensors. And a second calculation means (77A, S97, S98), a rotation angle calculation unit. In addition, although the alphanumeric character in parentheses represents a corresponding component in an embodiment described later, of course, the scope of the present invention is not limited to the embodiment. The same applies hereinafter.

  In the present invention, an angular interval between the central magnetic sensor and the other one of the three magnetic sensors, and an angular interval between the central magnetic sensor and the other magnetic sensor. The sum is less than 180 degrees in electrical angle. For this reason, even if any one of the three magnetic sensors fails, the two normal magnetic sensors both detect the same magnetic pole continuously for a predetermined number of sampling periods. A condition can occur.

  When two normal magnetic sensors satisfy the condition (at least one of the first condition and the second condition) that the same magnetic pole is detected continuously for a predetermined plurality of sampling periods Can calculate the rotation angle with high accuracy by calculating the rotation angle of the rotating body based on the output signals of a predetermined number of samplings of two normal magnetic sensors. As a result, even if any one of the three magnetic sensors fails, the accuracy is high when at least one of the first condition and the second condition is satisfied. The rotation angle can be calculated.

  According to a second aspect of the present invention, when the first calculation means satisfies the first condition, an output signal for the predetermined plural samplings of two normal magnetic sensors including the central magnetic sensor always or when the first condition is satisfied. When the information satisfies a certain requirement, the information on the magnetic pole width of the magnetic pole detected by the two magnetic sensors and / or the information on the amplitude of the output signal of the two magnetic sensors is calculated and applied to the magnetic pole. Means (77A, S87, S91) for storing in association with each other, and the second calculation means always or normally two magnetic sensors excluding the central magnetic sensor when the second condition is satisfied Information on the magnetic pole widths of the magnetic poles detected by the two magnetic sensors when the output signals for the predetermined plurality of samplings satisfy certain requirements And / or means (77A, S100) for calculating and storing information relating to the amplitudes of the output signals of the two magnetic sensors in association with the magnetic poles, the central magnetic sensor of the three magnetic sensors. When the first condition is not satisfied when at least two magnetic sensors including the normal magnetic sensor are included, information on the magnetic pole width and / or information on the amplitude are stored in association with each other. First, the rotation angle of the rotating body is calculated using output signals for one sampling of two magnetic sensors including one detected magnetic sensor and the information stored by the first calculation means. Only three arithmetic means (77A, S93, S95, S96) and the central magnetic sensor among the three magnetic sensors are out of order. Information on the magnetic pole width in relation to at least one of the magnetic poles detected by the two normal magnetic sensors excluding the central magnetic sensor and not satisfying the second condition. And / or the information on the amplitude is stored, the output signal for one sampling of two normal magnetic sensors and the information stored by the second calculation means are used to obtain the rotating body. The rotation angle calculation device according to claim 1, further comprising fourth calculation means (77A, S102, S104) for calculating the rotation angle.

  According to this configuration, when at least two magnetic sensors including the central magnetic sensor among the three magnetic sensors are normal, even when the first condition is not satisfied, a highly accurate rotation angle is obtained. It becomes possible to calculate. In addition, when only the central magnetic sensor of the three magnetic sensors is out of order, the normal two magnetic sensors excluding the central magnetic sensor detect even when the second condition is not satisfied. In the case where information on the magnetic pole width and / or information on the amplitude is stored in association with at least one of the magnetic poles, the rotation angle can be calculated with high accuracy.

  According to a third aspect of the present invention, when the three magnetic sensors are a first magnetic sensor, a second magnetic sensor, and a third magnetic sensor, and the second magnetic sensor is the central magnetic sensor, the second magnetic sensor The angular interval between a sensor and the first magnetic sensor is 60 degrees in electrical angle, and the angular interval between the second magnetic sensor and the third magnetic sensor is 60 degrees in electrical angle. The rotation angle calculation device according to 1 or 2.

FIG. 1 is a schematic diagram showing a schematic configuration of an electric power steering device to which a rotation angle detection device according to an embodiment of the present invention is applied. FIG. 2 is a schematic diagram showing an electrical configuration of the motor control ECU. FIG. 3 is a schematic diagram schematically showing the configuration of the electric motor. FIG. 4 is a graph showing a setting example of the q-axis current command value I _q ^* with respect to the detected steering torque Th. FIG. 5 is a schematic diagram schematically showing the configuration of the torque sensor. FIG. 6 is a schematic diagram showing the configuration of the first magnet and the arrangement of the three magnetic sensors. FIG. 7 is a schematic diagram showing output waveforms of the first magnetic sensor, the second magnetic sensor, and the third magnetic sensor. FIG. 8A, FIG. 8B, and FIG. 8C are schematic diagrams for explaining examples in which the fourth calculation mode, the fifth calculation mode, and the sixth calculation mode are applied, respectively. FIG. 9 is an explanatory diagram for explaining the fourth calculation mode. FIG. 10 is a flowchart showing the operation of the first rotation angle calculation unit. FIG. 11 is a flowchart showing the procedure of the failure determination process in step S1 of FIG. FIG. 12A is a flowchart showing a part of the procedure of the rotation angle calculation process based on the forced rotation in step S3 of FIG. FIG. 12B is a flowchart showing a part of the procedure of the rotation angle calculation process based on the forced rotation in step S3 of FIG. FIG. 12C is a flowchart showing a part of the procedure of the rotation angle calculation process based on the forced rotation in step S3 of FIG. FIG. 12D is a flowchart showing a part of the procedure of the rotation angle calculation process based on the forced rotation in step S3 of FIG. FIG. 13 is a schematic diagram showing a part of the contents of the memory in the torque calculation ECU. FIG. 14 is a flowchart showing a detailed procedure of the relative pole number setting process. FIG. 15 is a schematic diagram for explaining the relative pole number setting process. FIG. 16A is a flowchart showing a part of the normal rotation angle calculation process in step S5 of FIG. FIG. 16B is a flowchart showing a part of the normal rotation angle calculation process in step S5 of FIG. FIG. 16C is a flowchart illustrating a part of the normal rotation angle calculation process in step S5 of FIG. FIG. 17 is a schematic diagram for explaining a rotation angle detection method by a conventional rotation angle detection device. FIG. 18 is a schematic diagram showing output signal waveforms of the first magnetic sensor and the second magnetic sensor.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
FIG. 1 is a schematic diagram illustrating a schematic configuration of an electric power steering apparatus including a torque sensor to which a rotation angle detection apparatus according to a first embodiment of the present invention is applied.
The electric power steering apparatus 1 includes a steering wheel 2 as a steering member for steering the vehicle, a steering mechanism 4 that steers the steered wheels 3 in conjunction with the rotation of the steering wheel 2, and steering by the driver. And a steering assist mechanism 5 for assisting. The steering wheel 2 and the steering mechanism 4 are mechanically coupled via a steering shaft 6 and an intermediate shaft 7.

  The steering shaft 6 includes an input shaft 8 connected to the steering wheel 2 and an output shaft 9 connected to the intermediate shaft 7. The input shaft 8 and the output shaft 9 are connected via a torsion bar 10 so as to be relatively rotatable on the same axis. That is, when the steering wheel 2 is rotated, the input shaft 8 and the output shaft 9 rotate in the same direction while rotating relative to each other.

  Around the steering shaft 6 is provided a torque sensor (torque detection device) 11 to which the rotation angle detection device according to one embodiment of the present invention is applied. The torque sensor 11 detects the steering torque applied to the steering wheel 2 based on the relative rotational displacement amount of the input shaft 8 and the output shaft 9. The steering torque detected by the torque sensor 11 is input to a motor control ECU (Electronic Control Unit) 12.

  The steered mechanism 4 includes a rack and pinion mechanism including a pinion shaft 13 and a rack shaft 14 as a steered shaft. The steered wheel 3 is connected to each end of the rack shaft 14 via a tie rod 15 and a knuckle arm (not shown). The pinion shaft 13 is connected to the intermediate shaft 7. The pinion shaft 13 rotates in conjunction with the steering of the steering wheel 2. A pinion 16 is connected to the tip of the pinion shaft 13.

  The rack shaft 14 extends linearly along the left-right direction of the automobile (a direction orthogonal to the straight-ahead direction). A rack 17 that meshes with the pinion 16 is formed at an intermediate portion in the axial direction of the rack shaft 14. By the pinion 16 and the rack 17, the rotation of the pinion shaft 13 is converted into the axial movement of the rack shaft 14. The steered wheels 3 can be steered by moving the rack shaft 14 in the axial direction.

When the steering wheel 2 is steered (rotated), this rotation is transmitted to the pinion shaft 13 via the steering shaft 6 and the intermediate shaft 7. The rotation of the pinion shaft 13 is converted into an axial movement of the rack shaft 14 by the pinion 16 and the rack 17. Thereby, the steered wheel 3 is steered.
The steering assist mechanism 5 includes an electric motor 18 for generating a steering assist force, and a speed reduction mechanism 19 for transmitting the output torque of the electric motor 18 to the steering mechanism 4. In this embodiment, the electric motor 18 is a three-phase brushless motor. The speed reduction mechanism 19 includes a worm gear mechanism that includes a worm shaft 20 and a worm wheel 21 that meshes with the worm shaft 20. The speed reduction mechanism 19 is accommodated in a gear housing 22 as a transmission mechanism housing.

The worm shaft 20 is rotationally driven by the electric motor 18. The worm wheel 21 is coupled to the steering shaft 6 so as to be rotatable in the same direction. The worm wheel 21 is rotationally driven by the worm shaft 20.
When the worm shaft 20 is rotationally driven by the electric motor 18, the worm wheel 21 is rotationally driven and the steering shaft 6 rotates. The rotation of the steering shaft 6 is transmitted to the pinion shaft 13 via the intermediate shaft 7. The rotation of the pinion shaft 13 is converted into the axial movement of the rack shaft 14. Thereby, the steered wheel 3 is steered. That is, the wheel 3 is steered by rotating the worm shaft 20 by the electric motor 18.

The rotation angle of the rotor of the electric motor 18 (rotor rotation angle) is detected by a rotation angle sensor 25 such as a resolver. The output signal of the rotation angle sensor 25 is input to the motor control ECU 12. The electric motor 18 is controlled by a motor control ECU 12 as a motor control device.
FIG. 2 is a schematic diagram showing an electrical configuration of the motor control ECU 12.

  The motor control ECU 12 drives the electric motor 18 according to the steering torque Th detected by the torque sensor 11, thereby realizing appropriate steering assistance according to the steering situation. The motor control ECU 12 includes a microcomputer 40, a drive circuit (inverter circuit) 31 that is controlled by the microcomputer 40 and supplies electric power to the electric motor 18, and a current detection unit 32 that detects a motor current flowing through the electric motor 18. It has.

  The electric motor 18 is, for example, a three-phase brushless motor, and includes a rotor 100 as a field and U-phase, V-phase, and W-phase stator windings 101, 102, and 103, as schematically shown in FIG. And a stator 105. The electric motor 18 may be of an inner rotor type having a stator opposed to the outside of the rotor, or may be of an outer rotor type having a stator opposed to the inside of a cylindrical rotor.

Three-phase fixed coordinates (UVW coordinate system) are defined in which the U, V, and W axes are taken in the direction of the stator windings 101, 102, and 103 of each phase. Further, a two-phase rotational coordinate system (dq coordinate system) in which the d axis (magnetic pole axis) is taken in the magnetic pole direction of the rotor 100 and the q axis (torque axis) is taken in a direction perpendicular to the d axis in the rotation plane of the rotor 100. The actual rotating coordinate system) is defined. The dq coordinate system is a rotating coordinate system that rotates with the rotor 100. In the dq coordinate system, since only the q-axis current contributes to the torque generation of the rotor 100, the d-axis current may be set to zero and the q-axis current may be controlled according to the desired torque. The rotation angle (electrical angle) θ- _S of the rotor 100 is the rotation angle of the d axis with respect to the U axis. The dq coordinate system is an actual rotating coordinate system according to the rotor angle θ- _S . By using this rotor angle θ- _S , coordinate conversion between the UVW coordinate system and the dq coordinate system can be performed.

  The microcomputer 40 includes a CPU and a memory (ROM, RAM, non-volatile memory, etc.), and functions as a plurality of function processing units by executing a predetermined program. The plurality of function processing units include a current command value setting unit 41, a current deviation calculation unit 42, a PI (proportional integration) control unit 43, a dq / UVW conversion unit 44, and a PWM (Pulse Width Modulation) control unit. 45, a UVW / dq conversion unit 46, and a rotation angle calculation unit 47.

The rotation angle calculation unit 47 calculates the rotation angle of the rotor of the electric motor 18 (electrical angle; hereinafter referred to as “rotor angle θ _S ”) based on the output signal of the rotation angle sensor 25.
The current command value setting unit 41 sets a current value to be passed through the coordinate axes of the dq coordinate system as a current command value. Specifically, the current command value setting unit 41 refers to a d-axis current command value I _d ^* and a q-axis current command value I _q ^* (hereinafter, collectively referred to as “two-phase current command value I _dq ^* ”). ) Is set. More specifically, the current command value setting unit 41 sets the q-axis current command value I _q ^* to a significant value and sets the d-axis current command value I _d ^* to zero. More specifically, the current command value setting unit 41 sets the q-axis current command value I _q ^* based on the steering torque (detected steering torque) Th detected by the torque sensor 11.

A setting example of the q-axis current command value I _q ^* for the detected steering torque Th is shown in FIG. For the detected steering torque Th, for example, the torque for steering in the right direction is a positive value, and the torque for steering in the left direction is a negative value. The q-axis current command value I _q ^* is a positive value when an operation assisting force for rightward steering is to be generated from the electric motor 18, and the operation assisting force for leftward steering from the electric motor 18 is When it should be generated, it is a negative value. The q-axis current command value I _q ^* is positive for a positive value of the detected steering torque Th and negative for a negative value of the detected steering torque Th. When the detected steering torque Th is zero, the q-axis current command value I _q ^* is zero. The q-axis current command value I _q ^* is set so that the absolute value of the q-axis current command value I _q ^* increases as the absolute value of the detected steering torque Th increases.

The two-phase current command value I _dq ^* set by the current command value setting unit 41 is given to the current deviation calculation unit 42.
The current detection unit 32 detects the U-phase current I _U , the V-phase current I _V, and the W-phase current I _W (hereinafter collectively referred to as “three-phase detection current I _UVW ”) of the electric motor 18. To do. The three-phase detection current I _UVW detected by the current detection unit 32 is given to the UVW / dq conversion unit 46.

The UVW / dq converter 46 converts the three-phase detection current I _UVW (U-phase current I _U , V-phase current I _V and W-phase current I _W ) in the UVW coordinate system detected by the current detector 32 into the dq coordinate system. _Are transformed into two-phase detection currents I _d and I _q (hereinafter collectively referred to as “two-phase detection current I _dq ”). For this coordinate conversion, the rotor angle θ _S calculated by the rotation angle calculation unit 47 is used.

The current deviation calculation unit 42 calculates a deviation between the two-phase current command value I _dq ^* set by the current command value setting unit 41 and the two-phase detection current I _dq given from the UVW / dq conversion unit 46. More specifically, the current deviation calculation unit 42 calculates the deviation of the d-axis detection current I _d with respect to the d-axis current command value I _d ^* and the deviation of the q-axis detection current I _q with respect to the q-axis current command value I _q ^* . To do. These deviations are given to the PI control unit 43.

The PI control unit 43 performs a PI calculation on the current deviation calculated by the current deviation calculation unit 42 to thereby provide a two-phase voltage command value V _dq ^* (d-axis voltage command value V _d ^* and a d-axis voltage command value to be applied to the electric motor 18. q-axis voltage command value V _q ^* ) is generated. The two-phase voltage command value V _dq ^* is given to the dq / UVW converter 44.
The dq / UVW conversion unit 44 performs coordinate conversion of the two-phase voltage command value V _dq ^* into the three-phase voltage command value V _UVW ^* . For this coordinate conversion, the rotor angle θ _S calculated by the rotation angle calculation unit 47 is used. The three-phase voltage command value V _UVW ^* includes a U-phase voltage command value V _U ^* , a V-phase voltage command value V _V ^*, and a W-phase voltage command value V _W ^* . This three-phase voltage command value V _UVW ^* is given to the PWM control unit 45.

The PWM control unit 45 includes a U-phase PWM control signal, a V-phase PWM control signal having a duty corresponding to the U-phase voltage command value V _U ^* , the V-phase voltage command value V _V ^*, and the W-phase voltage command value V _W ^* , respectively. A W-phase PWM control signal is generated and supplied to the drive circuit 31.
The drive circuit 31 includes a three-phase inverter circuit corresponding to the U phase, the V phase, and the W phase. The power elements constituting the inverter circuit are controlled by a PWM control signal supplied from the PWM control unit 45, whereby a voltage corresponding to the three-phase voltage command value V _UVW ^* is set to the stator winding 101 of each phase of the electric motor 18. , 102, 103.

The current deviation calculation unit 42 and the PI control unit 43 constitute a current feedback control unit. By the action of the current feedback control means, the motor current flowing through the electric motor 18 is controlled so as to approach the two-phase current command value I _dq ^* set by the current command value setting unit 41.
FIG. 5 is a schematic diagram schematically showing the configuration of the torque sensor 11.

An annular first magnet (multipolar magnet) 61 is connected to the input shaft 8 so as to be integrally rotatable. Under the first magnet 61, three magnetic sensors 71, 72, and 73 that respectively output sinusoidal signals having a phase difference according to the rotation of the first magnet 61 are arranged.
An annular second magnet (multipolar magnet) 62 is connected to the output shaft 9 so as to be integrally rotatable. Three magnetic sensors 74, 75, and 76 that respectively output sinusoidal signals having a phase difference according to the rotation of the second magnet 62 are disposed above the second magnet 62.

Output signals S _{1 to} S ₆ of the magnetic sensors 71 to 76 are input to a torque calculation ECU 77 for calculating a steering torque applied to the input shaft 8. The power source of the torque calculation ECU 77 is turned on when the ignition key is turned on. When the ignition key is turned off, an ignition off command indicating that is input to the torque calculation ECU 77. In addition, as a magnetic sensor, what was provided with the element which has the characteristic which an electrical characteristic changes with the effect | actions of a magnetic field, such as a Hall element and a magnetoresistive element (MR element), for example can be used. In this embodiment, a Hall element is used as the magnetic sensor.

The magnets 61 and 62, the magnetic sensors 71 to 76, and the torque calculation ECU 77 constitute a torque sensor 11.
The torque calculation ECU 77 includes a microcomputer. The microcomputer includes a CPU and a memory (ROM, RAM, nonvolatile memory, etc.), and functions as a plurality of function processing units by executing a predetermined program. The plurality of function processing units include a first rotation angle calculation unit 77A, a second rotation angle calculation unit 77B, and a torque calculation unit 77C.

The first rotation angle calculator 77A calculates the rotation angle (electrical angle θ _A ) of the input shaft 8 based on the output signals S ₁ , S ₂ , S ₃ of the _three magnetic sensors 71, 72, 73. The second rotation angle calculator 77B calculates the rotation angle (electrical angle θ _B ) of the output shaft 9 based on the output signals S ₄ , S ₅ , S ₆ of the three magnetic sensors 74, 75, 76.
The torque calculator 77C is based on the rotation angle θ _A of the input shaft 8 detected by the first rotation angle calculator 77A and the rotation angle θ _B of the output shaft 9 detected by the second rotation angle calculator 77B. Thus, the steering torque Th applied to the input shaft 8 is calculated. Specifically, the steering torque Th is calculated based on the following equation (1), where K is the spring constant of the torsion bar 10 and N is the number of magnetic pole pairs provided in each of the magnets 61 and 62.

Th = {(θ _A −θ _B ) / N} × K (1)
First magnet 61, the magnetic sensor 71, 72, 73 and the first rotation angle computation section 77A, the first rotation angle detecting device for detecting a rotation angle theta _A of the input shaft 8 is constituted. The second magnet 62, the magnetic sensor 74, 75, 76 and the second rotation angle computation section 77B, and a second rotation angle detecting device is configured to detect the rotation angle theta _B of the output shaft 9 Yes. The operation of the first rotation angle detection device (first rotation angle calculation unit 77A) and the operation of the second rotation angle detection device (second rotation angle calculation unit 77B) are the same. Only the operation of the rotation angle detection device (first rotation angle calculation unit 77A) will be described.

FIG. 6 is a schematic diagram showing the configuration of the first magnet 61 and the arrangement of the three magnetic sensors 71, 72, 73.
The first magnet 61 has four pairs of magnetic poles (M1, M2), (M3, M4), (M5, M6), (M7, M8) arranged at equal angular intervals in the circumferential direction. . That is, the first magnet 61 has eight magnetic poles M1 to M8 arranged at equal angular intervals. The magnetic poles M <b> 1 to M <b> 8 are arranged at an angular interval (angular width) of approximately 45 ° (electrical angle is approximately 180 °) with the central axis of the input shaft 8 as the center. The magnitude of the magnetic force of each of the magnetic poles M1 to M8 is substantially constant.

  The three magnetic sensors 71, 72, 73 are arranged to face the lower annular end surface of the first magnet 61. Hereinafter, the magnetic sensor 71 may be referred to as a first magnetic sensor 71, the magnetic sensor 72 may be referred to as a second magnetic sensor 72, and the magnetic sensor 73 may be referred to as a third magnetic sensor 73. The first magnetic sensor 71 and the second magnetic sensor 72 are arranged at an angular interval of 60 ° in electrical angle (15 ° in mechanical angle) around the central axis of the input shaft 8. Further, the second magnetic sensor 72 and the third magnetic sensor 73 are arranged at an electrical angle of 60 ° in electrical angle with the central axis of the input shaft 8 as the center. Accordingly, the first magnetic sensor 71 and the third magnetic sensor 73 are arranged at an angular interval of 120 ° in electrical angle (30 ° in mechanical angle) around the central axis of the input shaft 8.

That is, the sum of the angular interval between the second magnetic sensor 72 and the first magnetic sensor 71 and the angular interval between the second magnetic sensor 72 and the third magnetic sensor 73 is 120 ° in electrical angle, It is set smaller than the angular width of one magnetic pole (180 ° in electrical angle).
Note that the angular interval between the first magnetic sensor 71 and the second magnetic sensor 72 is based on the second magnetic sensor 72 as a reference between the clockwise angular interval and the counterclockwise angular interval in FIG. Although it may be considered that there are two types, in this specification, the smaller one of the two types is used. In other words, the angular interval between the first magnetic sensor 71 and the second magnetic sensor 72 is the one of the two types that is smaller than the mechanical angle of 180 °. The same applies to the angular interval between the second magnetic sensor 72 and the third magnetic sensor 73.

A direction indicated by an arrow in FIG. 6 is a positive rotation direction of the input shaft 8. When the input shaft 8 is rotated in the forward direction, the rotation angle of the input shaft 8 is increased. When the input shaft 8 is rotated in the reverse direction, the rotation angle of the input shaft 2 is decreased. As shown in FIG. 7, sinusoidal signals S ₁ , S ₂ , S ₃ are output from the magnetic sensors 71, 72, 73 as the input shaft 8 rotates. Note that the rotation angle [deg] on the horizontal axis in FIG. 7 represents a mechanical angle.

In the following, the output signals _{S 1} is referred to as a first output signals _{S 1} or the first sensor value _{S 1,} the output signal _{S 2} of the second magnetic sensor 72 the second output signal _{S 2} or the second of the first magnetic sensor 71 called sensor values _{S 2,} there is a case the output signal _{S 3} of the third magnetic sensor 73 that the third output signal _{S 3} or the third sensor value _{S 3.}
In the following, for convenience of explanation, the rotation angle of the input shaft 8 is represented by θ, not θ _A. Assuming that each output signal S ₁ , S ₂ , S ₃ is a sine wave signal and the rotation angle of the input shaft 8 is θ (electrical angle), the output signal S ₁ of the first magnetic sensor 71 is S ₁ = is expressed as a ₁ · sin [theta, output signal _{S 2} of the second magnetic sensor 72 _is expressed as _{S 2 = a 2 · sin (} θ + 60), the output signal _{S 3} of the third magnetic sensor _{73, S} 3 = a ₃ · sin (θ + 120). A ₁ , A ₂ , and A ₃ represent amplitudes, respectively. First output signals _{S 1} and the phase difference between the second output signal _{S 2} is 60 degrees. Phase difference between the second output signals _{S 2} and the third output signal _{S 3} is also 60 degrees. Accordingly, the phase difference between the first output signals _{S 1} and the third output signal _{S 3} is 120 degrees.

The basic concept of the calculation method of the rotation angle θ by the first rotation angle calculation unit 77A will be described. The rotation angle calculation modes by the first rotation angle calculation unit 77A include a first calculation mode to a seventh calculation mode. Hereinafter, each calculation mode will be described.
[1] First Calculation Mode The first calculation mode is applied when the first and second magnetic sensors 71 and 72 detect the same magnetic pole together for three sampling periods (three calculation periods) continuously. This is a calculation mode that may be In the first calculation mode, the rotation angle θ is calculated based on output signals for three samplings in the first and second magnetic sensors 71 and 72.

Phase difference between the first output signals S ₁ and the second output signal S ₂ (electrical angle) to be represented by C. In addition, the number of the current sampling cycle (the number of the current calculation cycle) is represented by [n], the number of the previous sampling cycle is represented by [n-1], and the number of the previous sampling cycle is represented by [n-2]. To do. A correction value for correcting the rotation angle calculation error based on the variation in the angular width (pitch width) of each of the magnetic poles M1 to M8 is referred to as an angular width error correction value and is represented by E.

Phase difference C, the sampling period number [n], [n-1 ], the use of [n-2] and angular width error correction value E, this first output signals S ₁ and the current sampled the last and second last , the second output signal S ₂ which is sampled last and second last, the following equations (2a), can be expressed by (2b), (2c), (2d), (2e), (2f).
S ₁ [n] = A ₁ [n] sin (E ₁ [n] θ [n]) (2a)
S ₁ [n-1] = A ₁ [n-1] sin (E ₁ [n-1] θ [n-1]) (2b)
S ₁ [n-2] = A ₁ [n-2] sin (E ₁ [n-2] θ [n-2]) (2c)
S ₂ [n] = A ₂ [n] sin (E ₂ [n] θ [n] + C) (2d)
_{S 2 [n-1] =} A 2 [n-1] sin (E 2 [n-1] θ [n-1] + C) ... (2e)
S ₂ [n-2] = A ₂ [n-2] sin (E ₂ [n-2] θ [n-2] + C) (2f)
In the equations (2a) to (2f), E ₁ [x] is an angle width error correction value corresponding to the magnetic pole detected by the first magnetic sensor 71 in the x-th calculation cycle. E ₂ [x] is an angle width error correction value corresponding to the magnetic pole detected by the second magnetic sensor 72 in the x-th calculation cycle.

When the angular width of a certain magnetic pole is w (electrical angle), the angular width error θ _err (electrical angle) of the magnetic pole is defined by the following equation (3).
θ _err = w−180 (3)
The angular width error correction value E for the magnetic pole is defined by the following equation (4).
E = 180 / w
= 180 / (θ _err +180) (4)
The angle width error correction value E of each magnetic pole is information regarding the magnetic pole width of each magnetic pole. The information on the magnetic pole width of each magnetic pole may be the angular width w of each magnetic pole or the angular width error θ _err of each magnetic pole.

Assuming that C is known, the number of unknowns included in the six expressions represented by the expressions (2a) to (2f) is 15. In other words, since the number of unknowns is larger than the number of equations, it is not possible to solve simultaneous equations consisting of six equations.
In this embodiment, by setting the sampling interval (sampling period) to be short, it is considered that there is no change in amplitude due to a temperature change between three samplings. That is, it is assumed that the amplitudes A ₁ [n], A ₁ [n−1], and A ₁ [n−2] of the output signal of the first magnetic sensor 71 during three samplings are equal to each other, and these are defined as A ₁ . . Similarly, the amplitudes A ₂ [n], A ₂ [n−1], and A ₂ [n−2] of the output signal of the second magnetic sensor 72 during three samplings are regarded as being equal to each other, and these are defined as A ₂ . .

When both magnetic sensors 71 and 72 detect the same magnetic pole during three samplings, the angle width error correction value E included in the output signals of both magnetic sensors 71 and 72 for three samplings. ₁ [n], E ₁ [n-1], E ₁ [n-2], E ₂ [n], E ₂ [n-1], E ₂ [n-2] have the same value. Is represented by E. As a result, the equations (2a) to (2f) are represented by the following equations (5a) to (5f), respectively.

S ₁ [n] = A ₁ sin (Eθ [n]) (5a)
S ₁ [n-1] = A ₁ sin (Eθ [n-1]) (5b)
S ₁ [n-2] = A ₁ sin (Eθ [n-2]) (5c)
S ₂ [n] = A ₂ sin (Eθ [n] + C) (5d)
S ₂ [n-1] = A ₂ sin (Eθ [n-1] + C) (5e)
S ₂ [n-2] = A ₂ sin (Eθ [n-2] + C) (5f)
The number of unknowns (A ₁ , A ₂ , E, θ [n], θ [n−1], θ [n−2]) included in these six formulas is 6. That is, since the number of unknowns is less than or equal to the number of equations, simultaneous equations composed of six equations can be solved. Therefore, the rotation angle θ [n] of the input shaft 8 can be calculated by solving the simultaneous equations composed of the six equations (5a) to (5f).

Hereinafter, the case where the phase difference C between the two magnetic sensors is 60 degrees will be specifically described. When the phase difference C is 60 degrees, the six formulas (5a) to (5f) can be expressed by the following formulas (6a) to (6f), respectively.
S ₁ [n] = A ₁ sin (Eθ [n]) (6a)
S ₁ [n-1] = A ₁ sin (Eθ [n-1]) (6b)
S ₁ [n-2] = A ₁ sin (Eθ [n-2]) (6c)
S ₂ [n] = A ₂ sin (Eθ [n] +60) (6d)
S ₂ [n-1] = A ₂ sin (Eθ [n-1] +60) (6e)
S ₂ [n-2] = A ₂ sin (Eθ [n-2] +60) (6f)
Considering Eθ [n] as one unknown, solve the simultaneous equations consisting of four equations (6a), (6b), (6d), and (6e) among the above six equations (6a) to (6f). Thus, Eθ [n] is expressed by the following equation (7) (hereinafter referred to as “Eθ basic operation equation (7)”).

  Further, by solving the simultaneous equations composed of the above six equations (6a) to (6f), the angular width error correction value E is expressed by the following equation (8) (hereinafter referred to as “E operation equation (8)”). Represented.

Therefore, θ [n] is obtained by dividing Eθ [n] calculated by the Eθ basic calculation formula (7) by the angle width error correction value E calculated by the E calculation formula (8). Can do. That is, θ [n] can be obtained by the following equation (9).
θ [n] = Eθ [n] / E (9)
However, when any denominator of the fraction included in the E calculation formula (8) is zero, the angle width error correction value E cannot be calculated based on the E calculation formula (8). Therefore, in this embodiment, when any denominator of the fraction included in E calculation formula (8) becomes zero, the angle width error correction value E calculated last time is used as the current angle width error correction. The value E is used.

  Note that the case where any denominator of the fraction included in E equation (8) is zero is one of the three conditions represented by the following equations (10), (11), and (12): This is a case where at least one of the following conditions is satisfied.

Further, when any denominator of the fraction included in the Eθ basic calculation formula (7) is zero, Eθ [n] cannot be calculated based on the Eθ basic calculation formula (7). Therefore, in this embodiment, when any denominator of the fraction included in the Eθ basic arithmetic expression (7) becomes zero, Eθ [n] is calculated by an arithmetic expression different from the Eθ basic arithmetic expression (7). Is calculated. Furthermore, in this embodiment, Eθ [n] can be calculated based on the Eθ basic calculation formula (7). However, when Eθ [n] can be calculated using a simpler calculation formula, the Eθ basic calculation formula is used. Eθ [n] is calculated by an arithmetic expression different from (7). In this embodiment, the case where Eθ [n] can be calculated more simply than the Eθ basic calculation formula (7) is the case where S ₁ [n] = 0 or S ₂ [n] = 0.

  In this embodiment, 10 types of calculation formulas including Eθ basic calculation formula (7) are prepared as calculation formulas for calculating Eθ [n]. Table 1 shows ten types of arithmetic expressions and conditions to which the arithmetic expressions are applied. When calculating Eθ [n], it is determined whether or not the condition is satisfied in order from the top of Table 1. If it is determined that the condition is satisfied, the subsequent condition determination is not performed. Eθ [n] is calculated by an arithmetic expression corresponding to the condition.

The first arithmetic expression from the top of Table 1 is the Eθ basic arithmetic expression (7). In Eθ basic arithmetic expression (7), neither S ₁ [n] nor S ₂ [n] is zero, and any denominator of a fraction contained in Eθ basic arithmetic expression (7) is not zero. Applicable when conditions are met. The condition that any denominator of the fraction included in the Eθ basic arithmetic expression (7) is not zero is that S ₁ [n−1] ≠ 0 and S ₂ [n−1] ≠ 0, and p ₁ − It is satisfied when p ₂ ≠ 0 and p ₁ ² + p ₁ p ₂ + p ₂ ² ≠ 0. Note that S ₁ [n-1] is the denominator of p ₁ and S ₂ [n-1] is the denominator of p ₂ .

However, p ₁ ² + p ₁ p ₂ + p ₂ ² = 0 holds only when p ₁ = p ₂ = 0, but the first magnetic sensor 71 and the second magnetic sensor 72 have a phase of 60. The sensor values S ₁ and S _{2 of} the two magnetic sensors 71 and 72 do not become zero at the same time because of the degree of deviation. Therefore, p ₁ ² + p ₁ p ₂ + p ₂ ² = 0 does not hold. Therefore, the condition that any denominator of the fraction included in the Eθ basic arithmetic expression (7) is not zero is S ₂ [n−1] ≠ 0, S ₂ [n−1] ≠ 0, and p It is satisfied when ₁₋ p ₂ ≠ 0.

The second arithmetic expression from the top of Table 1 is an arithmetic expression applied when p ₁ −p ₂ = 0. Consider the case where p ₁ −p ₂ = 0 holds. In this case, since p ₁ = p ₂ , the following equation (13) is established.

  When this is modified, the following equation (14) is obtained.

The case where the formula (14) is satisfied is a case where Eθ [n] and Eθ [n−1] are equal. That is, the current Eθ [n] is equal to the previous Eθ [n−1]. Therefore, none of the _{S 1} [n] and _{S 2} [n] is not zero, and none of the denominator _{S 1} of _{p 1 [n-1]} and the denominator _{S 2} [n-1] of _{p 2} is zero If the condition that p ₁ −p ₂ = 0 is satisfied, Eθ [n−1] calculated last time is used as the current Eθ [n].

The third and fourth arithmetic expressions from the top of Table 1 are arithmetic expressions applied when the denominator S ₁ [n−1] of p ₁ becomes zero. Since S ₁ [n-1] = A ₁ sinEθ [n-1], when sinEθ [n-1] = 0, S ₁ [n-1] = 0. That is, when Eθ [n−1] is 0 degree or 180 degrees, S ₁ [n−1] is zero. Since S ₂ [n-1] = A ₂ sin (Eθ [n-1] +60), when Eθ [n-1] is 0 degree, S ₂ [n-1]> 0, and Eθ [n− When 1] is 180 degrees, S ₂ [n−1] <0. Therefore, when S ₁ [n-1] = 0 and S ₂ [n-1]> 0, Eθ [n-1] = 0, and S ₁ [n-1] = 0 and S ₂ [n− When 1] <0, Eθ [n−1] = 180.

When Eθ [n−1] = 0, the expressions (6d) and (6e) are expressed by the following expressions (15d) and (15e), respectively.
S ₂ [n] = A ₂ sin (Eθ [n] +60) (15d)
S ₂ [n-1] = A ₂ sin 60 = √3 / 2 · A ₂ (15e)
From the equation (15e), the following equation (16) is obtained.

A ₂ = (2 / √3) · S ₂ [n-1] (16)
Substituting the equation (16) into the equation (15d) yields the following equation (17).
sin (Eθ [n] +60) = (√3 / 2) · (S ₂ [n] / S ₂ [n−1]) (17)
Therefore, Eθ [n] can be calculated by the following equation (18).

That is, as shown third from the top in Table 1, neither S ₁ [n] nor S ₂ [n] is zero and the denominator S ₂ [n−1] of p ₂ is not zero. When the condition that the denominator S ₁ [n-1] of p ₁ is zero and S ₂ [n-1]> 0 is satisfied, the arithmetic expression represented by the above equation (18) is obtained. Based on this, Eθ [n] is calculated.
On the other hand, when Eθ [n−1] = 180, the expressions (6d) and (6e) are expressed by the following expressions (19d) and (19e), respectively.

S ₂ [n] = A ₂ sin (Eθ [n] +60) (19d)
S ₂ [n-1] = A ₂ sin 240 = −√3 / 2 · A ₂ (19e)
From the equation (19e), the following equation (20) is obtained.
A ₂ = (− 2 / √3) · S ₂ [n−1] (20)
Substituting the equation (20) into the equation (19d) yields the following equation (21).

sin (Eθ [n] +60) = (− √3 / 2) · (S ₂ [n] / S ₂ [n−1]) (21)
Therefore, Eθ [n] can be calculated by the following equation (22).

That is, as shown in the fourth from the top of Table 1, neither S ₁ [n] nor S ₂ [n] is zero and the denominator S ₂ [n−1] of p ₂ is not zero. When the condition that the denominator S ₁ [n-1] of p ₁ is zero and S ₂ [n-1] <0 is satisfied, the arithmetic expression represented by the above equation (22) is satisfied. Based on this, Eθ [n] is calculated.
The fifth and sixth arithmetic expressions from the top of Table 1 are arithmetic expressions applied when S ₂ [n] = 0. Since S ₂ [n] = A ₂ sin (Eθ [n] +60), S ₂ [n] = 0 when sin (Eθ [n] +60) = 0. That is, S ₂ [n] = 0 when Eθ [n] is −60 degrees or 120 degrees. Since S ₁ [n] = A ₁ sin Eθ [n], S ₁ [n] <0 when Eθ [n] is −60 degrees, and S ₁ [n]> when Eθ [n] is 120 degrees. 0. Accordingly, Eθ [n] = 120 when S ₂ [n] = 0 and S ₁ [n]> 0, and Eθ [n] when S ₂ [n] = 0 and S ₁ [n] <0. ] = − 60.

That is, as shown fifth from the top in Table 1, S ₁ [n] is not zero, the denominator S ₂ [n−1] of p ₂ is not zero, and S ₂ [n] = 0. When there is a condition that S ₁ [n]> 0, Eθ [n] is calculated as 120 degrees. Further, as shown in Table 6 from the top, S ₁ [n] is not zero, the denominator S ₂ [n−1] of p ₂ is not zero, and S ₂ [n] = 0. If there is a condition that S ₁ [n] <0, Eθ [n] is calculated as −60 degrees.

The seventh and eighth arithmetic expressions from the top of Table 1 are arithmetic expressions applied when the denominator S ₂ [n−1] of p ₂ is zero. Since S ₂ [n-1] = A ₂ sin (Eθ [n-1] +60), when sin (Eθ [n-1] +60) = 0, S ₂ [n-1] = 0 Become. That is, when Eθ [n−1] is −60 degrees or 120 degrees, S ₂ [n−1] is zero. Since S ₁ [n-1] = A ₁ sinEθ [n-1], when Eθ [n-1] is −60 degrees, S ₁ [n-1] <0, and Eθ [n-1] is At 120 degrees, S ₁ [n-1]> 0. Therefore, when S ₂ [n-1] = 0 and S ₁ [n-1]> 0, Eθ [n-1] = 120, S ₂ [n-1] = 0 and S ₁ [n− When 1] <0, Eθ [n−1] = − 60.

When Eθ [n−1] = 120, the equations (6a) and (6b) are represented by the following equations (23a) and (23b), respectively.
S ₁ [n] = A ₁ sin Eθ [n] (23a)
S ₁ [n-1] = A ₁ sin 120 = √3 / 2 · A ₁ (23b)
From the equation (23b), the following equation (24) is obtained.

A ₁ = (2 / √3) · S ₁ [n-1] (24)
Substituting the equation (24) into the equation (23a) yields the following equation (25).
sinEθ [n] = (√3 / 2) · (S ₁ [n] / S ₁ [n-1]) (25)
Therefore, Eθ [n] can be calculated by the following equation (26).

That is, as shown in Table 7 from the top, S ₁ [n] is not zero, the denominator S ₂ [n-1] of p ₂ is zero, and S ₁ [n-1]> When the condition of 0 is satisfied, Eθ [n] is calculated based on the arithmetic expression expressed by the expression (26).
On the other hand, when Eθ [n−1] = − 60, the expressions (6a) and (6b) are expressed by the following expressions (27a) and (27b), respectively.

S ₁ [n] = A ₁ sin Eθ [n] (27a)
S ₁ [n-1] = A ₁ sin (−60) = − √3 / 2 · A ₂ (27b)
From the equation (27b), the following equation (28) is obtained.
A ₁ = (− 2 / √3) · S ₁ [n−1] (28)
Substituting the equation (28) into the equation (27a) yields the following equation (29).

sinEθ [n] = (− √3 / 2) · (S ₁ [n] / S ₁ [n−1]) (29)
Therefore, Eθ [n] can be calculated by the following equation (30).

That is, as shown in Table 8 from the top, S ₁ [n] is not zero, the denominator S ₂ [n−1] of p ₂ is zero, and S ₁ [n−1] < When the condition of 0 is satisfied, Eθ [n] is calculated based on the arithmetic expression expressed by the expression (30).
The ninth and tenth arithmetic expressions from the top of Table 1 are arithmetic expressions that are applied when S ₁ [n] = 0. Since S ₁ [n] = A ₁ sinEθ [n], when sinEθ [n] = 0, S ₁ [n] = 0. That is, when Eθ [n] is 0 degree or 180 degrees, S ₁ [n] = 0. Since it is _{S 2 [n] = A 2} sin (Eθ [n] +60], Eθ [n] is _{S 2} [n] at the time of 0 degree> 0, E.theta [n] is at the time of 180 ° _{S 2} [ n] <0 Therefore, if S ₁ [n] = 0 and S ₂ [n]> 0, Eθ [n] = 0, and S ₁ [n] = 0 and S ₂ [n] If <0, Eθ [n] = 180.

That is, as shown in the ninth from the top in Table 1, when the condition that S ₁ [n] is zero and S ₂ [n]> 0 is satisfied, Eθ [n] is 0 degree. Is calculated as As shown in the tenth from the top in Table 1, when the condition that S ₁ [n] is zero and S ₂ [n] <0 is satisfied, Eθ [n] is 180 degrees. Is calculated as
When Eθ [n] is calculated, the amplitude A ₁ can be calculated based on the equation (6a), and the amplitude A ₂ can be calculated based on the equation (6d). That is, E, θ [n], A _1, A ₂ can be calculated in the first calculation mode.
[2] Second Calculation Mode The second calculation mode is applied when the second and third magnetic sensors 72 and 73 detect the same magnetic pole for three sampling periods (three calculation periods) continuously. This is a calculation mode that may be In the second calculation mode, the rotation angle θ is calculated based on output signals for three samplings in the second and third magnetic sensors 72 and 73.

When the output signals S ₂ and S ₃ of the second magnetic sensor 72 and the third magnetic sensor 73 are expressed using the angular width error correction value E, the output signal S ₂ [n] of the second magnetic sensor 72 is S ₂ [ n] = A ₂ · sin (E ₂ θ [n] +60), and the output signal S ₃ [n] of the third magnetic sensor 73 is S ₃ [n] = A ₃ · sin (E ₃ θ [ n] +120). Here, E ₃ is an angle width error correction value corresponding to the magnetic pole detected by the third magnetic sensor 73. When the second magnetic sensor 72 and the third magnetic sensor 73 detect the same magnetic pole, E ₂ and E ₃ are equal to each other. Therefore, when these are represented by E, the output signal S ₂ of the second magnetic sensor 72 is output. [n] is represented by S ₂ [n] = A ₂ · sin (Eθ [n] +60), and the output signal S ₃ [n] of the third magnetic sensor 73 is S ₃ [n] = A ₃ · sin It is expressed as (Eθ [n] +120).

When (Eθ [n] +60) is replaced with EΘ [n], the second output signal S ₂ [n] is represented by S ₂ [n] = A ₂ · sinEΘ [n], and the third output signal S ₃ [ n] is represented by S ₃ [n] = A ₃ · sin (EΘ [n] +60). Therefore, EΘ [n] and E can be calculated using the second output signal S ₂ and the third output signal S ₃ in the same manner as described above. Since EΘ [n] = Eθ [n] +60, θ [n] = (EΘ [n] −60) / E. Therefore, the rotation angle θ [n] of the input shaft 8 is calculated by substituting the calculated EΘ [n] and E into the equation θ [n] = (EΘ [n] −60) / E. Can do. When EΘ [n] is calculated, the amplitude A ₂ and the amplitude A ₃ can be calculated. That is, E, θ [n], A _2, A ₃ can be calculated in the second calculation mode.

When the second and third output signals for three samplings used for the rotation angle calculation in the second calculation mode are expressed by the following equations (31a) to (31f) following the equations (6a) to (6f), EΘ The basic arithmetic expression and the E arithmetic expression can be expressed by the following expressions (32) and (33), respectively.
S ₂ [n] = A ₂ sin (Eθ [n] +60) (31a)
S ₂ [n-1] = A ₂ sin (Eθ [n-1] +60) (31b)
S ₂ [n-2] = A ₂ sin (Eθ [n-2] +60) (31c)
S ₃ [n] = A ₃ sin (Eθ [n] +120) (31d)
S ₃ [n-1] = A ₃ sin (Eθ [n-1] +120) (31e)
S ₃ [n-2] = A ₃ sin (Eθ [n-2] +120) (31f)

Note that the definitions of q _{1 to} q ₆ , t in the E calculation formula (33) in the second calculation mode are different from the definitions of q _{1 to} q ₆ , t in the E calculation formula (8) in the first calculation mode. .
[3] Third Calculation Mode The third calculation mode is applied when the first and third magnetic sensors 71 and 73 detect the same magnetic pole together for three sampling periods (three calculation periods) continuously. This is a calculation mode that may be In the third calculation mode, the rotation angle θ is calculated based on output signals for three samplings in the first and third magnetic sensors 71 and 73.

When the first and third output signals for three samplings used for the rotation angle calculation in the third calculation mode are expressed according to the above equations (6a) to (6f), the following equations (34a) to (34f) are obtained. .
S ₁ [n] = A ₁ sin (Eθ [n]) (34a)
S ₁ [n-1] = A ₁ sin (Eθ [n-1]) (34b)
S ₁ [n-2] = A ₁ sin (Eθ [n-2]) (34c)
S ₃ [n] = A ₃ sin (Eθ [n] +120) (34d)
S ₃ [n-1] = A ₃ sin (Eθ [n-1] +120) (34e)
S ₃ [n-2] = A ₃ sin (Eθ [n-2] +120) (34f)
Considering Eθ [n] as one unknown, it solves simultaneous equations consisting of four equations (34a), (34b), (34d) and (34e) among the six equations (34a) to (34f). Thus, Eθ [n] is expressed by the following equation (35) (hereinafter referred to as “Eθ basic operation equation (35)”).

Further, by solving the simultaneous equations composed of the above six equations (34a) to (34f), the angular width error correction value E is expressed by the following equation (36) (hereinafter referred to as “E operation equation (36)”). Represented. The definition of q _{1 to} q ₆ , t in the E operation expression (36) in the third operation mode is different from the definition of q _{1 to} q ₆ , t in the E operation expression (8) in the first operation mode. .

Therefore, θ [n] is obtained by dividing Eθ [n] calculated by the Eθ basic calculation formula (35) by the angular width error correction value E calculated by the E calculation formula (36). Can do. That is, θ [n] can be obtained by the following equation (37).
θ [n] = Eθ [n] / E (37)
However, when any denominator of the fraction included in the E calculation formula (36) is zero, the angle width error correction value E cannot be calculated based on the E calculation formula (36). Therefore, in this embodiment, when any denominator of the fraction included in the E equation (36) becomes zero, the angle width error correction value E calculated last time is used as the current angle width error correction. The value E is used.

It should be noted that the case where any denominator of the fraction included in E arithmetic expression (36) is zero means that when the denominator of t in E arithmetic expression (36) is zero, the numerator of t is zero. Or (S ₁ [n] · S ₃ [n] −S ₁ [n−1] · S ₃ [n−1]) becomes zero.
In addition, when any denominator of the fraction included in the Eθ basic calculation formula (35) is zero, Eθ [n] cannot be calculated based on the Eθ basic calculation formula (35). Therefore, in this embodiment, when any denominator of the fraction contained in the Eθ basic arithmetic expression (35) becomes zero, Eθ [n] is calculated by an arithmetic expression different from the Eθ basic arithmetic expression (35). Is calculated. Further, in this embodiment, Eθ [n] can be calculated based on the Eθ basic calculation formula (35). However, when Eθ [n] can be calculated using a simpler calculation formula, the Eθ basic calculation formula is used. Eθ [n] is calculated by an arithmetic expression different from (35). In this embodiment, the case where Eθ [n] can be calculated more easily than the Eθ basic calculation formula (35) is the case where S ₁ [n] = 0 or S ₃ [n] = 0.

  In this embodiment, 10 types of calculation formulas including the Eθ basic calculation formula (35) are prepared as calculation formulas for calculating Eθ [n]. Table 2 shows 10 types of arithmetic expressions and conditions to which the arithmetic expressions are applied. When calculating Eθ [n], it is determined whether or not the condition is satisfied in order from the top of Table 2. If it is determined that the condition is satisfied, the subsequent condition determination is not performed. Eθ [n] is calculated by an arithmetic expression corresponding to the condition.

The first arithmetic expression from the top of Table 2 is the Eθ basic arithmetic expression (35). In the Eθ basic arithmetic expression (35), neither S ₁ [n] nor S ₃ [n] is zero, and any denominator of the fraction included in the Eθ basic arithmetic expression (35) is not zero. Applicable when conditions are met. The condition that any denominator of the fraction included in the Eθ basic arithmetic expression (35) is not zero is that p ₁ −p ₃ ≠ 0, p ₁ ² + p ₁ p ₃ + p ₃ ² ≠ 0, and S It is satisfied when ₁ [n-1] ≠ 0 and S ₃ [n-1] ≠ 0. Note that S ₁ [n-1] is the denominator of p ₁ and S ₃ [n-1] is the denominator of p ₃ .

However, p ₁ ² + p ₁ p ₃ + p ₃ ² = 0 holds only when p ₁ = p ₃ = 0, but the first magnetic sensor 71 and the third magnetic sensor 73 have a phase of 120. The sensor values S ₁ and S _{3 of} both magnetic sensors 71 and 73 do not become zero at the same time because they are deviated. Therefore, p ₁ ² + p ₁ p ₃ + p ₃ ² = 0 does not hold. Therefore, the condition that any denominator of the fraction included in the Eθ basic arithmetic expression (35) is not zero is that p ₁ −p ₃ ≠ 0, S ₁ [n−1] ≠ 0, and S ₃ [ It is satisfied when n−1] ≠ 0.

The second arithmetic expression from the top of Table 2 is an arithmetic expression applied when p ₁ −p ₃ = 0. Consider the case where p ₁ −p ₃ = 0 holds. In this case, since p ₁ = p ₃ , the following equation (38) is established.

  When this is transformed, the following equation (39) is obtained.

The case where the equation (39) is satisfied is a case where Eθ [n] and Eθ [n−1] are equal. That is, the current Eθ [n] is equal to the previous Eθ [n−1]. Therefore, none of the _{S 1} [n] and _{S 3} [n] is not zero, and none of the denominator _{S 3} of the denominator _{S 1 [n-1]} and _{p 3} of _{p 1 [n-1]} is zero If the condition that p ₁ −p ₃ = 0 is satisfied, Eθ [n−1] calculated last time is used as the current Eθ [n].

The third and fourth arithmetic expressions from the top of Table 2 are arithmetic expressions applied when the denominator S ₁ [n−1] of p ₁ is zero. Since S ₁ [n-1] = A ₁ sinEθ [n-1], when sinEθ [n-1] = 0, S ₁ [n-1] = 0. That is, when Eθ [n−1] is 0 degree or 180 degrees, S ₁ [n−1] is zero. Since S ₃ [n-1] = A ₃ sin (Eθ [n-1] +120), when Eθ [n-1] is 0 degree, S ₃ [n-1]> 0, and Eθ [n− When 1] is 180 degrees, S ₃ [n−1] <0. Therefore, when S ₁ [n-1] = 0 and S ₃ [n-1]> 0, Eθ [n-1] = 0, and S ₁ [n-1] = 0 and S ₃ [n− When 1] <0, Eθ [n−1] = 180.

When Eθ [n−1] = 0, the expressions (34d) and (34e) are expressed by the following expressions (40d) and (40e), respectively.
S ₃ [n] = A ₃ sin (Eθ [n] +120) (40d)
S ₃ [n-1] = A ₃ sin 120 = √3 / 2 · A ₂ (40e)
From the equation (40e), the following equation (41) is obtained.

A ₃ = (2 / √3) · S ₃ [n-1] (41)
Substituting the equation (41) into the equation (40d) yields the following equation (42).
sin (Eθ [n] +120) = (√3 / 2) · (S ₃ [n] / S ₃ [n−1]) (42)
Therefore, Eθ [n] can be calculated by the following equation (43).

That is, as shown third from the top in Table 2, neither S ₁ [n] nor S ₃ [n] is zero, and the denominator S ₃ [n−1] of p ₃ is not zero. When the condition that the denominator S ₁ [n-1] of p ₁ is zero and S ₃ [n-1]> 0 is satisfied, the arithmetic expression represented by the above equation (43) is satisfied. Based on this, Eθ [n] is calculated.
On the other hand, when Eθ [n−1] = 180, the expressions (34d) and (34e) are expressed by the following expressions (44d) and (44e), respectively.

S ₃ [n] = A ₃ sin (Eθ [n] +120) (44d)
S ₃ [n-1] = A ₃ sin 300 = −√3 / 2 · A ₃ (44e)
From the equation (44e), the following equation (45) is obtained.
A ₃ = (− 2 / √3) · S3 [n−1] (45)
Substituting the equation (45) into the equation (44d) yields the following equation (46).

sin (Eθ [n] +120) = (− √3 / 2) · (S ₃ [n] / S ₃ [n−1]) (46)
Therefore, Eθ [n] can be calculated by the following equation (47).

That is, as shown in the fourth from the top of Table 2, neither S ₁ [n] nor S ₃ [n] is zero and the denominator S ₃ [n−1] of p ₃ is not zero. And when the condition that the denominator S ₁ [n-1] of p ₁ is zero and S ₃ [n-1] <0 is satisfied, the arithmetic expression represented by the equation (47) is obtained. Based on this, Eθ [n] is calculated.
The fifth and sixth arithmetic expressions from the top in Table 2 are arithmetic expressions that are applied when S ₃ [n] = 0. Since S ₃ [n] = A ₃ sin (Eθ [n] +120), S ₃ [n] = 0 when sin (Eθ [n] +120) = 0. That is, when Eθ [n] is −120 degrees or 60 degrees, S ₃ [n] = 0. Since S ₁ [n] = A ₁ sin Eθ [n], S ₁ [n] <0 when Eθ [n] is −120 degrees, and S ₁ [n]> when Eθ [n] is 60 degrees. 0. Therefore, Eθ [n] = 60 when S ₃ [n] = 0 and S ₁ [n]> 0, and Eθ [n] when S ₃ [n] = 0 and S ₁ [n] <0. ] = − 120.

That is, as shown fifth from the top in Table 2, S ₁ [n] is not zero, the denominator S ₃ [n−1] of p ₃ is not zero, and S ₃ [n] = 0. If there is a condition that S ₁ [n]> 0, Eθ [n] is calculated as 60 degrees. Further, as shown in Table 6 from the top, S ₁ [n] is not zero, the denominator S ₃ [n−1] of p ₃ is not zero, and S ₃ [n] = 0. If the condition that S ₁ [n] <0 is satisfied, Eθ [n] is calculated as −120 degrees.

The seventh and eighth arithmetic expressions from the top of Table 2 are arithmetic expressions applied when the denominator S ₃ [n−1] of p ₃ is zero. Since S ₃ [n-1] = A ₃ sin (Eθ [n-1] +120), when sin (Eθ [n-1] +120) = 0, S ₃ [n-1] = 0 Become. That is, when Eθ [n−1] is −120 degrees or 60 degrees, S ₃ [n−1] is zero. Since S ₁ [n-1] = A ₁ sinEθ [n-1], when Eθ [n-1] is −120 degrees, S ₁ [n-1] <0, and Eθ [n-1] is When it is 60 degrees, S ₁ [n-1]> 0. Therefore, when S ₃ [n-1] = 0 and S ₁ [n-1]> 0, Eθ [n-1] = 60, S ₃ [n-1] = 0 and S ₁ [n− When 1] <0, Eθ [n−1] = − 120.

When Eθ [n−1] = 60, the equations (34a) and (34b) are expressed by the following equations (48a) and (48b), respectively.
S ₁ [n] = A ₁ sin Eθ [n] (48a)
S ₁ [n-1] = A ₁ sin 60 = √3 / 2 · A ₁ (48b)
From the equation (48b), the following equation (49) is obtained.

A ₁ = (2 / √3) · S ₁ [n-1] (49)
Substituting the equation (49) into the equation (48a) yields the following equation (50).
sinEθ [n] = (√3 / 2) · (S ₁ [n] / S ₁ [n-1]) (50)
Therefore, Eθ [n] can be calculated by the following equation (51).

That is, as shown in the seventh from the top in Table 2, S ₁ [n] is not zero, the denominator S ₃ [n-1] of p ₃ is zero, and S ₁ [n-1]> When the condition of 0 is satisfied, Eθ [n] is calculated based on the arithmetic expression expressed by the above equation (51).
On the other hand, when Eθ [n−1] = − 120, the equations (34a) and (34b) are expressed by the following equations (52a) and (52b), respectively.

S ₁ [n] = A ₁ sin Eθ [n] (52a)
S ₁ [n-1] = A ₁ sin (−120) = − √3 / 2 · A ₁ (52b)
From the equation (52b), the following equation (53) is obtained.
A ₁ = (− 2 / √3) · S ₁ [n−1] (53)
Substituting the equation (53) into the equation (52a) yields the following equation (54).

sinEθ [n] = (− √3 / 2) · (S ₁ [n] / S ₁ [n−1]) (54)
Therefore, Eθ [n] can be calculated by the following equation (55).

In other words, as shown in Table 8 from the top, S ₁ [n] is not zero, the denominator S ₃ [n-1] of p ₃ is zero, and S ₁ [n-1] < When the condition of 0 is satisfied, Eθ [n] is calculated based on the arithmetic expression represented by the expression (55).
The ninth and tenth arithmetic expressions from the top of Table 2 are arithmetic expressions applied when S ₁ [n] = 0. Since S ₁ [n] = A ₁ sinEθ [n], when sinEθ [n] = 0, S ₁ [n] = 0. That is, when Eθ [n] is 0 degree or 180 degrees, S ₁ [n] = 0. Since it is _{S 3 [n] = A 3} sin (Eθ [n] +120], Eθ [n] is _{S 3} [n] when the 0 degree> 0, E.theta [n] is at the time of 180 _{S 3} [ n] <0 Therefore, if S ₁ [n] = 0 and S ₃ [n]> 0, Eθ [n] = 0, and S ₁ [n] = 0 and S ₃ [n] If <0, Eθ [n] = 180.

That is, as shown in the ninth from the top in Table 2, when the condition that S ₁ [n] is zero and S ₃ [n]> 0 is satisfied, Eθ [n] is 0 degree. Is calculated as Further, as shown in the tenth from the top of Table 2, when the condition that S ₁ [n] is zero and S ₃ [n] <0 is satisfied, Eθ [n] is 180 degrees. Is calculated as
When Eθ [n] is calculated, the amplitude A ₁ can be calculated based on the equation (34a), and the amplitude A ₃ can be calculated based on the equation (34d). That is, E, θ [n], A _1, A ₃ can be calculated in the third calculation mode.

  In the first calculation mode, the second calculation mode, or the third calculation mode, the rotation angle θ [of the input shaft 8 based on the output signals for three samplings in two of the three magnetic sensors 71, 72, 73. Since n] is calculated, a highly accurate rotation angle can be calculated. Further, in the first to third calculation modes, even if the number of mathematical expressions used for the calculation of the rotation angle θ [n] of the input shaft 8 is smaller than the number of original unknowns included in these mathematical expressions, the input is performed. Since the rotation angle θ [n] of the shaft 8 can be calculated, the number of sensor values required for calculating the rotation angle θ [n] of the input shaft 8 can be reduced.

  In the first to third calculation modes, it is considered that the amplitudes of the output signals of the same magnetic sensor are the same between the three samplings. The amplitude of the output signal of the same magnetic sensor during three samplings may be different depending on the influence of temperature change. However, when the sampling interval is small, the temperature change between the three samplings is very small, so that the amplitude of the output signal of the same magnetic sensor during the three samplings can be regarded as equal. Therefore, in the first to third calculation modes, it is possible to compensate for the amplitude variation due to the temperature change between the three samplings. In the first to third calculation modes, the amplitude between the two magnetic sensors used for the calculation of the rotation angle is treated as a separate unknown, and therefore the temperature characteristics of the two magnetic sensors vary. Can be compensated for. Thereby, a highly accurate rotation angle can be detected.

Further, in the first to third calculation modes, variations in the angular width (pitch width) of the magnetic poles M1 to M8 of the magnet 61 can be compensated with high accuracy, so that a rotation angle with a smaller error can be detected. .
[4] Fourth operation mode 4 arithmetic mode, the angular width error correction value amplitude A ₁ of E ₁ and the first output signals S ₁ to the first magnetic sensor 71 corresponds to the magnetic pole are detected by the first operational mode Alternatively, the calculation mode may be applied when already calculated in the third calculation mode and stored in the memory. In the fourth calculation mode, the rotation angle θ is calculated mainly based on the output signal S ₁ of the first magnetic sensor 71.

  For example, as shown in FIG. 8A, when the magnet 61 (input shaft 8) rotates in the direction indicated by the arrow, the first and second magnetic sensors 71 and 72 detect the same magnetic pole (M1 in this example). A case is assumed where the magnetic pole detected by the second magnetic sensor 72 changes from the state in which the magnetic field is detected. In the state immediately after the change, both the second magnetic sensor 72 and the third magnetic sensor 73 do not detect the same magnetic pole continuously for three sampling periods (three calculation periods), so the second calculation mode cannot be applied. In such a case, the fourth calculation mode may be applied.

With number n of the angular width error correction value E and calculation cycle, the output signals S ₁ of the first magnetic sensor 71 sampled at the present calculation cycle is represented by the following formula (56).
S ₁ [n] = A ₁ [n] sin (E ₁ θ [n]) (56)
E ₁ is an angle width error correction value corresponding to the magnetic pole detected by the first magnetic sensor 71.
From the equation (56), the rotation angle θ [n] is expressed by the following equation (57).

θ [n] = (1 / E ₁ ) sin ⁻¹ (S ₁ [n] / A ₁ ) (57)
Assuming that E ₁ and A ₁ corresponding to the magnetic pole detected by the first magnetic sensor 71 are stored in the memory, θ [n] is calculated by substituting them into the equation (57). Can do. However, when the rotation angle θ [n] is calculated by the equation (57), since two rotation angles θ [n] are calculated, it is necessary to determine which rotation angle is the actual rotation angle. There is. This determination method will be described with reference to FIG. FIG. 9 shows waveforms for one cycle of the first output signal S ₁ , the second output signal S _2, and the third output signal S ₃ . The rotation angle [deg] on the horizontal axis in FIG. 9 represents an electrical angle.

As shown in FIG. 9, when the first output signal S ₁ [n] has a positive value, for example, the rotation corresponding to (1 / E ₁ ) sin ⁻¹ (S ₁ [n] / A ₁ ) The angle θ [n] is two rotation angles: a rotation angle in the region R1 of 0 to 90 degrees and a rotation angle in the region R2 of 90 to 180 degrees. Further, when the first output signal S ₁ [n] is, for example, a negative value, the rotation angle θ [n] corresponding to (1 / E ₁ ) sin ⁻¹ (S ₁ [n] / A ₁ ). Are two rotation angles: a rotation angle in the region U1 of 180 to 270 degrees and a rotation angle in the region U2 of 270 to 360 degrees.

In this embodiment, based on one of the output signals S ₂ and S ₃ of the _two magnetic sensors 72 and 73 other than the first magnetic sensor 71, out of the two rotation angles calculated by the equation (57). Which is the actual rotation angle is determined. When one of the second magnetic sensor 72 and the third magnetic sensor 73 is out of order, the determination is made based on the output signal of one normal magnetic sensor.

First, the case where the determination is performed based on the second output signal S ₂ [n] will be described. 1/2 of the amplitude A ₂ of the second output signal S ₂ is set as a threshold value a (a> 0). The threshold value a, for example, a amplitude A ₂ of the second output signal S ₂ which is stored in the memory, be determined based on the amplitude A ₂ of the first magnetic sensor 71 corresponds to the magnetic pole is detected it can. Note that ½ of the amplitude A ₁ of the first output signal S ₁ may be set as the threshold value a (a> 0).

The rotation angle θ [n] of the input shaft 8 that can be taken when the second output signal S ₂ [n] is greater than or equal to a is in the range of 0 degrees to 90 degrees and in the range of 330 degrees to 360 degrees. The rotation angle θ [n] of the input shaft 8 that can be taken when the second output signal S ₂ [n] is less than −a is in the range of 150 degrees to 270 degrees. The rotation angle θ [n] of the input shaft 8 that can be taken when the second output signal S ₂ [n] is greater than or equal to −a and less than a is in the range of 90 ° to 150 ° and in the range of 270 ° to 330 °.

Therefore, it can be determined based on the second output signal S ₂ [n] which of the two rotation angles calculated by the equation (57) is the actual rotation angle. Specifically, when the first output signal S ₁ [n] is a positive value, if the second output signal S ₂ [n] is greater than or equal to a, 2 calculated by the equation (57) is used. Of the two rotation angles, the rotation angle in the region R1 is determined to be the actual rotation angle. On the other hand, if the second output signal S ₂ [n] is less than a, it is determined that the rotation angle in the region R2 is the actual rotation angle among the two rotation angles calculated by the equation (57). .

When the first output signal S ₁ [n] is a negative value, if the second output signal S ₂ [n] is less than −a, the two rotation angles calculated by the equation (57) are used. Of these, the rotation angle in the region U1 is determined to be the actual rotation angle. On the other hand, if the second output signal S ₂ [n] is −a or more, it is determined that the rotation angle in the region U2 is the actual rotation angle among the two rotation angles calculated by the equation (57). The

Next, a case where the determination is performed based on the third output signal S ₃ [n] will be described. A half of the amplitude A ₃ of the third output signal S ₃ is set as a threshold value a (a> 0). The threshold value a, for example, a amplitude A ₃ of the third output signal S ₃ which is stored in the memory, be determined based on the amplitude A ₃ of the first magnetic sensor 71 corresponds to the magnetic pole is detected it can. Note that ½ of the amplitude A ₁ of the first output signal S ₁ may be set as the threshold a (a> 0).

When the first output signal S ₁ [n] is a positive value, if the third output signal S ₃ [n] is greater than or equal to −a, the two rotation angles calculated by the equation (57) are used. Of these, the rotation angle in the region R1 is determined to be the actual rotation angle. On the other hand, if the third output signal S ₃ [n] is less than −a, it is determined that the rotation angle in the region R2 is the actual rotation angle among the two rotation angles calculated by the equation (57). The

When the first output signal S ₁ [n] is a negative value, if the third output signal S ₃ [n] is less than a, the two rotation angles calculated by the equation (57) are used. It is determined that the rotation angle in the region U1 is an actual rotation angle. On the other hand, if the third output signal S ₃ [n] is greater than or equal to a, it is determined that the rotation angle in the region U2 is the actual rotation angle among the two rotation angles calculated by the equation (57). .
[5] Fifth operation mode fifth operation mode, amplitude A ₂ of the angular width error correction value E ₂ and the second output signal S ₂ of the second magnetic sensor 72 corresponds to the magnetic pole are detected by the first operational mode Alternatively, the calculation mode may be applied when the calculation is already performed in the second calculation mode and stored in the memory. In the fifth operation mode, the rotation angle θ is calculated mainly on the basis of the output signal S ₂ of the second magnetic sensor 72.

  For example, as shown in FIG. 8B, when the magnet 61 (input shaft 8) rotates in the direction indicated by the arrow, the first, second, and third magnetic sensors 71, 72, 73 have the same magnetic pole (this example Let us assume a case where the magnetic pole detected by the first magnetic sensor 71 changes from the state in which M1) is detected. When the third magnetic sensor is out of order, the second calculation mode cannot be applied in the changed state. In such a case, the fifth calculation mode may be applied.

With number n of the angular width error correction value E and calculation cycle, the output signal S ₂ of the second magnetic sensor 72 sampled at the present calculation cycle is represented by the following formula (58).
S ₂ [n] = A ₂ [n] sin (E ₂ θ [n] +60) (58)
E ₂ is an angle width error correction value corresponding to the magnetic pole detected by the second magnetic sensor 72.
From the equation (58), the rotation angle θ [n] is expressed by the following equation (59).

θ [n] = (1 / E ₂ ) {sin ⁻¹ (S ₂ [n] / A ₂ ) −60} (59)
Assuming that E ₂ and A ₂ corresponding to the magnetic pole detected by the second magnetic sensor 72 are stored in the memory, θ [n] is calculated by substituting them into the equation (59). be able to.
When the rotation angle θ [n] is calculated by the equation (59), two rotation angles θ [n] are calculated. Therefore, based on one of the output signals S ₁ and S ₃ of the two magnetic sensors 71 and 73 other than the second magnetic sensor 72, the two rotation angles θ [n] calculated by the equation (59) are calculated. It is determined which of these is the actual rotation angle. Note that, when one of the first magnetic sensor 71 and the third magnetic sensor 73 is out of order, the determination is made based on the output signal of one normal magnetic sensor.

First, the case where the determination is performed based on the first output signal S ₁ [n] will be described. The threshold a (a> 0) is set to 1/2 of the amplitude A ₁ of the first output signal S ₁ or 1/2 of the amplitude A ₂ of the second output signal S ₂ stored in the memory.
When the second output signal S ₂ [n] is a positive value, if the first output signal S ₁ [n] is less than a, the two rotation angles calculated by the equation (59) are used. It is determined that the rotation angle in the region of 0 degrees to 30 degrees or 300 degrees to 360 degrees is the actual rotation angle. On the other hand, if the first output signal S ₁ [n] is greater than or equal to a, the rotation angle in the region of 30 to 120 degrees out of the two rotation angles calculated by the equation (59) is the actual rotation angle. It is determined that there is.

When the second output signal S ₂ [n] is a negative value, if the first output signal S ₁ [n] is greater than or equal to −a, the two rotation angles calculated by the equation (59) are used. Of these, the rotation angle within the range of 120 to 210 degrees is determined to be the actual rotation angle. On the other hand, if the first output signal S ₁ [n] is less than −a, the rotation angle in the region of 210 ° to 300 ° out of the two rotation angles calculated by the equation (59) is the actual rotation angle. It is determined that

Next, a case where the determination is performed based on the third output signal S ₃ [n] will be described. The threshold a (a> 0) is set to 1/2 of the amplitude A ₃ of the third output signal S ₂ or 1/2 of the amplitude A ₂ of the second output signal S ₂ stored in the memory.
In the case where the second output signal S ₂ [n] is a positive value, if the third output signal S ₃ [n] is greater than or equal to a, of the two rotation angles calculated by the equation (59). It is determined that the rotation angle in the region of 0 degrees to 30 degrees or 300 degrees to 360 degrees is the actual rotation angle. On the other hand, if the third output signal S ₃ [n] is less than a, the rotation angle in the region of 30 ° to 120 ° out of the two rotation angles calculated by the equation (59) is the actual rotation angle. It is determined that there is.

When the second output signal S ₂ [n] is a negative value, if the third output signal S ₃ [n] is less than −a, the two rotation angles calculated by the equation (59) are used. Of these, the rotation angle within the range of 120 to 210 degrees is determined to be the actual rotation angle. On the other hand, if the third output signal S ₃ [n] is greater than or equal to −a, the rotation angle in the region of 210 to 300 degrees out of the two rotation angles calculated by the equation (59) is the actual rotation angle. It is determined that
[6] Sixth operation mode sixth operation mode, the angular width error correction value E ₃ and the third amplitude A ₃ of the output signal S ₃ of the third magnetic sensor 73 corresponds to the magnetic pole are detected by the second operation mode Alternatively, the calculation mode may be applied when already calculated in the third calculation mode and stored in the memory. In the sixth calculation mode, the rotation angle θ is calculated mainly based on the output signal S ₃ of the third magnetic sensor 73.

  For example, as shown in FIG. 8C, when the magnet 61 (input shaft 8) rotates in the direction indicated by the arrow, the second and third magnetic sensors 72 and 73 detect the same magnetic pole (M2 in this example). It is assumed that the magnetic pole detected by the second magnetic sensor 72 changes from the state in which the magnetic field is detected. In the state immediately after the change, both the first magnetic sensor 71 and the second magnetic sensor 73 do not detect the same magnetic pole for three sampling periods (three calculation periods) continuously, so the first calculation mode cannot be applied. In such a case, the sixth calculation mode may be applied.

  Further, as shown in FIG. 8B described above, when the magnet 61 rotates in the direction indicated by the arrow, the first, second, and third magnetic sensors 71, 72, 73 have the same magnetic pole (M1 in this example). When the magnetic pole detected by the first magnetic sensor 71 changes from the state where the first magnetic sensor 71 is detected, the sixth calculation mode may be applied even when the second magnetic sensor 72 is out of order. is there.

With number n of the angular width error correction value E and calculation cycle, the output signal S ₃ of the third magnetic sensor 73 sampled at the present calculation cycle is represented by the following formula (60).
S ₃ [n] = A ₃ [n] sin (E ₃ θ [n] +120) (60)
E ₃ is an angle width error correction value corresponding to the magnetic pole detected by the third magnetic sensor 73.
From the equation (60), the rotation angle θ [n] is expressed by the following equation (61).

θ [n] = (1 / E ₃ ) {sin ⁻¹ (S ₃ [n] / A ₃ ) −120} (61)
Assuming that E ₃ and A ₃ corresponding to the magnetic pole detected by the third magnetic sensor 73 are stored in the memory, θ [n] is calculated by substituting these into the equation (61). be able to.
When the rotation angle θ [n] is calculated by the equation (61), two rotation angles θ [n] are calculated. Therefore, based on one of the output signals S ₁ and S ₂ of the two magnetic sensors 71 and 72 other than the third magnetic sensor 73, the two rotation angles θ [n] calculated by the equation (61) are calculated. It is determined which of these is the actual rotation angle. In addition, when any one of the 1st magnetic sensor 71 and the 2nd magnetic sensor 72 has failed, the said determination is performed based on the output signal of one normal magnetic sensor.

First, the case where the determination is performed based on the second output signal S ₂ [n] will be described. The threshold a (a> 0) is set to 1/2 of the amplitude A ₂ of the second output signal S ₂ or 1/2 of the amplitude A ₃ of the third output signal S ₃ stored in the memory.
When the third output signal S ₃ [n] is a positive value, if the second output signal S ₂ [n] is less than a, the two rotation angles calculated by the equation (61) are used. It is determined that the rotation angle within the range of 240 degrees to 330 degrees is the actual rotation angle. On the other hand, if the second output signal S ₂ [n] is greater than or equal to a, the rotation angle in the region of 330 ° to 360 ° or 0 ° to 60 ° among the two rotation angles calculated by the equation (61). Is determined to be the actual rotation angle.

When the third output signal S ₃ [n] is a negative value, if the second output signal S ₂ [n] is equal to or greater than −a, the two rotation angles calculated by the equation (61) are used. Of these, the rotation angle in the region of 60 degrees to 150 degrees is determined to be the actual rotation angle. On the other hand, if the second output signal S ₂ [n] is less than −a, the rotation angle in the region of 150 to 240 degrees out of the two rotation angles calculated by the equation (61) is the actual rotation angle. It is determined that

Next, a case where the determination is performed based on the first output signal S ₁ [n] will be described. 1/2 of the amplitude _{A 3} 1/2 or third output signal _{S 3} of the first output signals _{S 1} amplitude A1 stored in the memory as the threshold a (a> 0).
When the third output signal S ₃ [n] is a positive value, if the first output signal S ₁ [n] is less than −a, the two rotation angles calculated by the equation (61) are used. Of these, the rotation angle in the region of 240 to 330 degrees is determined to be the actual rotation angle. On the other hand, if the first output signal S ₁ [n] is −a or more, the rotation within the region of 330 ° to 360 ° or 0 ° to 60 ° out of the two rotation angles calculated by the equation (61). It is determined that the angle is an actual rotation angle.

When the third output signal S ₃ [n] is a negative value, if the first output signal S ₁ [n] is greater than or equal to a, the two rotation angles calculated by the equation (61) It is determined that the rotation angle in the region of 60 degrees to 150 degrees is the actual rotation angle. On the other hand, if the first output signal S ₁ [n] is less than a, the rotation angle in the region of 150 to 240 degrees out of the two rotation angles calculated by the equation (61) is the actual rotation angle. It is determined that there is.
[7] Seventh operation mode The seventh operation mode is applied when the second magnetic sensor 72 is out of order and any of the third operation mode, the fourth operation mode, and the sixth operation mode cannot be applied. Calculation mode. In the seventh calculation mode, the rotation angle θ [n] is calculated using the same calculation expressions (35), (36), and (37) as in the third calculation mode. That is, in the seventh calculation mode, the first and third magnetic sensors 71 and 73 both do not satisfy the condition that the same magnetic pole is continuously detected for three sampling periods (three calculation periods). The rotation angle θ [n] is calculated using the same calculation formula as in the third calculation mode.

FIG. 10 is a flowchart showing the operation of the first rotation angle calculator 77A.
When the torque calculation ECU 77 is turned on, the first rotation angle calculation unit 77A performs a failure determination process to determine whether or not each of the magnetic sensors 71, 72, 73 has failed (step S1). Details of this processing will be described later.
In the failure determination process of step S1, if it is determined that all three magnetic sensors 71, 72, 73 are not broken or if it is determined that only one magnetic sensor is broken, 1 rotation angle calculation unit 77A proceeds to step S2.

In step S2, the first rotation angle calculation unit 77A determines whether or not the value of the forced rotation flag F _AK is 1. The forced rotation flag F _AK is a flag for storing that the rotation angle calculation process based on the forced rotation in step S3 described later was performed when the torque calculation ECU 77 was turned on. The initial value of the forced rotation flag F _AK is 0, and the value of the forced rotation flag F _AK is set to 1 when the rotation angle calculation process based on the forced rotation is performed.

When the value of the forced rotation flag F _AK is 0 (step S2: NO), the first rotation angle calculation unit 77A performs a rotation angle calculation process based on the forced rotation (step S3). This process is a process of rotating the input shaft 8 (output shaft 9) by temporarily forcibly rotating the electric motor 18 to calculate the rotation angle θ of the input shaft 8. Details of this processing will be described later.
In the first calculation mode, the second calculation mode, or the third calculation mode described above, the output signals of the two magnetic sensors used for calculating the rotation angle θ [n] change between the previous sampling time and the current sampling time. If not, the previously calculated Eθ [n] (or EΘ [n]), E and θ [n] are the current Eθ [n] (or EΘ [n]), E and θ [n]. ] (See the second arithmetic expression from the top of Table 1 and Table 2). However, when the ignition key is turned on to turn on the torque calculation ECU 77, the previously calculated Eθ [n] (or EΘ [n]), E and θ [n] do not exist. . For this reason, when the output signals of the two magnetic sensors used for calculating the rotation angle θ [n] have not changed after the torque calculation ECU 77 is turned on, the first, second, or third The rotation angle θ [n] cannot be calculated depending on the calculation mode. Therefore, in order to create the previous values of Eθ [n] (or EΘ [n]), E, and θ [n], rotation angle calculation processing based on forced rotation is performed.

When the rotation angle computing process based on forced rotation is completed, the first rotation angle computation unit 77A, after setting the value of the forced rotation flag F _AK 1 (step S4), and proceeds to step S6.
In step S6, the first rotation angle calculation unit 77A determines whether or not an ignition off command is input. If the ignition off command has not been input (step S6: NO), the first rotation angle calculation unit 77A returns to step S1.

When it is determined in step S2 that the value of the forced rotation flag F _AK is 1 (step S2: YES), the first rotation angle calculation unit 77A performs a normal rotation angle calculation process ( Step S5). Details of this processing will be described later.
When the rotation angle is calculated by the normal rotation angle calculation process, the first rotation angle calculation unit 77A proceeds to step S6 and determines whether or not an ignition-off command is input. If the ignition off command has not been input (step S6: NO), the first rotation angle calculation unit 77A returns to step S1.

If it is determined in step S6 that an ignition off command has been input (step S6: YES), the first rotation angle calculation unit 77A ends the rotation angle calculation process.
FIG. 11 is a flowchart showing the procedure of the failure determination process in step S1 of FIG.
First, the first rotation angle calculation unit 77A acquires the sensor values S ₁ [n], S ₂ [n], and S ₃ [n] of the magnetic sensors 71, 72, and 73 (step S11). The memory in the torque calculation ECU 77 stores sensor values for a plurality of times (three or more times) from the sensor value acquired a predetermined number of times to the latest acquired sensor value.

Then, the first rotation angle calculation unit 77A, based on the acquired sensor values S ₁ [n], S ₂ [n], S ₃ [n], the first magnetic sensor 71, the second magnetic sensor 72, and the second 3. It is determined whether or not the magnetic sensor 73 is out of order (steps S12, S14, S16). When the magnetic sensor fails, the output signal is fixed to a predetermined value. Therefore, for example, when the sensor value of the first magnetic sensor 71 does not change despite the sensor values of the second and third magnetic sensors 72 and 73 changing, the first It is determined that the magnetic sensor 71 has failed.

When it is determined that the first magnetic sensor has failed (step S12: YES), the first rotation angle calculation unit 77A sets the value of the first sensor failure flag F _A1 to 1 (step S13). ). The initial value of the first sensor failure flag F _A1 is zero.
If it is determined that the second magnetic sensor has failed (step S14: YES), the first rotation angle calculation unit 77A sets the value of the second sensor failure flag F _A2 to 1 (step S15). ). The initial value of the second sensor failure flag F _A2 is zero.

When it is determined that the third magnetic sensor has failed (step S16: YES), the first rotation angle calculation unit 77A sets the value of the third sensor failure flag F _A3 to 1 (step S17). ).
Thus, when the failure determination for each of the magnetic sensors 71, 72, 73 is completed, the first rotation angle calculation unit 77A determines whether or not two or more magnetic sensors have failed (step S18). . When two or more magnetic sensors are out of order (step S18: YES), the first rotation angle calculation unit 77A performs an abnormality process (step S19). That is, the first rotation angle calculation unit 77A stops the rotation angle calculation process and transmits a command for stopping the control of the electric motor 18 to the motor control ECU 12. Thereby, the electric motor 18 is not driven.

When two or more magnetic sensors have not failed (step S18: NO), that is, when all the magnetic sensors 71, 72, 73 are normal or only one magnetic sensor has failed. The first rotation angle calculation unit 77A proceeds to step S2 in FIG.
12A, 12B and 12C are flowcharts showing the procedure of the rotation angle calculation process based on the forced rotation in step S3 of FIG.

  When the rotation angle calculation process based on forced rotation is started, the magnetic pole number detected by one normal magnetic sensor is used as a reference magnetic pole, and the relative magnetic pole number is assigned to each magnetic pole. It is defined as a pole number. The relative pole number of the magnetic pole detected by the first magnetic sensor 71 (hereinafter referred to as “first relative pole number”) is represented by a variable r1, and the relative pole of the magnetic pole detected by the second magnetic sensor 72 is represented. The number (hereinafter referred to as “second relative pole number”) is represented by a variable r2, and the relative pole number of the magnetic pole detected by the third magnetic sensor 73 (hereinafter referred to as “third relative pole number”). Let it be represented by a variable r3. Each of the relative pole numbers r1, r2, and r3 takes an integer of 1 to 8, and the relative pole number that is 1 less than 1 is 8, and the relative pole number that is 1 greater than 8 is 1.

The memory in the torque calculation ECU 77 is provided with areas indicated by e1 to e4 as shown in FIG. In the area e1, the value of the angle width error correction value E is stored for each of the relative magnetic pole numbers 1 to 8. The area e2, the amplitude _{A 1} of the first output signal _{S 1} is stored for each 8 relative pole number. The area e3, the amplitude _{A 2} of the second output signal _{S 2} is stored for each 8 relative pole number. The area e4, the amplitude _{A 3} of the third output signal _{S 3} is stored for each 8 relative pole number.

  Returning to FIG. 12A, in the rotation angle calculation process based on forced rotation, the steering wheel 2 is driven to rotate for a short time. For this reason, the driver may misunderstand that some kind of failure has occurred. Therefore, the first rotation angle calculation unit 77A issues a warning to the driver (step S21). Specifically, the first rotation angle calculation unit 77A is connected to a video / audio control device (not shown) for controlling a display device (not shown), an audio output device (not shown) and the like provided in the vehicle. Send a warning output command. When receiving the warning output command, the video / audio control device displays a message such as “the steering wheel is forcibly rotated but is not in failure” on the display device, or outputs the sound by the audio output device.

  Next, the first rotation angle calculation unit 77A rotates the electric motor 18 in the first direction (step S22). Specifically, the first rotation angle calculator 77A transmits a first forced rotation command for rotating the electric motor 18 in the first direction to the motor control ECU 12. When receiving the first forced rotation command, the motor control ECU 12 rotates the electric motor 18 in the first direction.

The first rotation angle calculator 77A acquires the sensor values S ₁ [n], S ₂ [n], S ₃ [n] of the magnetic sensors 71, 72, 73 (step S23). The process in step S23 is repeatedly executed at predetermined calculation cycles until YES is determined in step S30, step S32, or step S34 described later. The memory in the torque calculation ECU 77 stores sensor values for a plurality of times (three or more times) from the sensor value acquired a predetermined number of times to the latest acquired sensor value.

The first rotation angle calculation unit 77A determines whether or not the current process is the first process after the start of the rotation angle calculation process based on the forced rotation (step S24). When the current process is the first process after the start of the rotation angle calculation process based on the forced rotation (step S24: YES), the first rotation angle calculation unit 77A performs a relative pole number setting process ( Step S25).
FIG. 14 is a flowchart showing a detailed procedure of the relative pole number setting process.

Here, all the magnetic sensors 71, 72, 73 are normal, and the magnetic pole numbers when the relative numbers are assigned to the respective magnetic poles using the magnetic pole detected by the first magnetic sensor 71 as a reference magnetic pole are relative to each other. The case where it is defined as a target pole number will be described as an example.
First rotation angle computation section 77A, first, the first output signals _{S 1} to determine whether greater than zero or not (step S61). If the first output signal _{S 1} is greater than 0 (step S61: YES), determines that the first rotation angle computation unit 77A includes pole first magnetic sensor 71 is detected by a magnetic pole of N-pole Then, the first relative pole number r1 is set to 1 (step S64). Then, the process proceeds to step S66.

On the other hand, if the first output signal _{S 1} is equal to or smaller than 0 (step S61: NO), the first rotation angle computation section 77A, the first output signal _{S 1} is determined whether 0 less than or ( Step S62). If the first output signal _{S 1} is smaller than 0 (step S62: YES), determines that the first rotation angle computation unit 77A includes pole first magnetic sensor 71 is detecting is the magnetic pole of the S pole Then, the first relative pole number r1 is set to 2 (step S65). Then, the process proceeds to step S66.

In the step S62, if the first output signals _{S 1} is determined to be 0 or more (step S62: NO), that is, if the first output signal _{S 1} is 0, the first rotating corner arithmetic unit 77A, the rotation angle of the input shaft 8 is 0 to determine a whether a is either 180 ° °, the second output signal S ₂ is to determine whether greater than zero or not (step S63). If the second output signal _{S 2} is greater than 0 (step S63: YES), the first rotation angle computation unit 77A determines that the rotation angle of the input shaft 8 is 0 °, the first relative pole The number r1 is set to 1 (step S64). Then, the process proceeds to step S66.

On the other hand, when the second output signal _{S 2} is equal to or smaller than 0 (step S63: NO), the first rotation angle computation unit 77A, the rotation angle of the input shaft 8 is determined to be a 180 °, the first The relative pole number r1 is set to 2 (step S65). Then, the process proceeds to step S66.
In step S66, the first rotation angle calculation unit 77A determines whether or not the condition of “S ₁ ≧ 0 and S ₂ > 0” or “S ₁ ≦ 0 and S ₂ <0” is satisfied. When this condition is satisfied (step S66: YES), the first rotation angle calculator 77A detects the pole number of the magnetic pole detected by the second magnetic sensor 72 by the first magnetic sensor 71. It is determined that it is the same as the pole number of the magnetic pole, and the same number (r2 = r1) as the first relative pole number r1 is set as the second relative pole number r2 (step S67). Then, the process proceeds to step S69.

  On the other hand, when the condition of step S66 is not satisfied (step S66: NO), the first rotation angle calculator 77A indicates that the pole number of the magnetic pole detected by the second magnetic sensor 72 is the first magnetic field. It is determined that the number is 1 larger than the pole number of the magnetic pole detected by the sensor 71, and a number (r2 = r1 + 1) larger by 1 than the first relative pole number r1 is set in the second relative pole number r2. (Step S68). Then, the process proceeds to step S69.

In step S69, the first rotation angle calculation unit 77A determines whether or not the conditions of “S ₁ ≧ 0 and S ₃ > 0” or “S ₁ ≦ 0 and S ₃ <0” are satisfied. When this condition is satisfied (step S69: YES), the first rotation angle calculation unit 77A detects the pole number of the magnetic pole detected by the third magnetic sensor 73 by the first magnetic sensor 71. It is determined that it is the same as the pole number of the magnetic pole, and the same number (r3 = r1) as the first relative pole number r1 is set as the third relative pole number r3 (step S70). And it transfers to step S28 of FIG. 12A.

  On the other hand, when the condition of step S69 is not satisfied (step S69: NO), the first rotation angle calculator 77A indicates that the pole number of the magnetic pole detected by the third magnetic sensor 73 is the first magnetic field. It is determined that the number is 1 larger than the pole number of the magnetic pole detected by the sensor 71, and a number (r3 = r1 + 1) larger by 1 than the first relative pole number r1 is set in the third relative pole number r3. (Step S71). And it transfers to step S28 of FIG. 12A.

The reason why the second relative pole number r2 is determined based on the condition of step S66 and the reason why the third relative pole number r3 is determined based on the condition of step S69 will be described. For example, the signal waveforms of the first, second and third output signals S ₁ , S ₂ and S ₃ when the magnetic pole pair of the magnetic pole M1 and the magnetic pole M2 in the magnet 61 passes through the first magnetic sensor 71 are schematically shown. Expressed as shown in FIG. 15, (a), (b), and (c).

  In FIG. 15, in the areas indicated by Q1, Q2, Q4 and Q5, the pole numbers of the magnetic poles detected by the second magnetic sensor 72 are the same as the pole numbers of the magnetic poles detected by the first magnetic sensor 71. is there. On the other hand, in the regions indicated by Q3 and Q6, the pole number of the magnetic pole detected by the second magnetic sensor 72 is one greater than the pole number of the magnetic pole detected by the first magnetic sensor 71.

In the regions Q1 and Q2, the sensor values S ₁ and S ₂ satisfy the first condition of S ₁ ≧ 0 and S ₂ > 0. In the region Q3, both sensor values _S 1, _{S 2 is, S} 1> 0 and the second condition is satisfied in _{S 2} ≦ 0. In the regions Q4 and Q5, both sensor values S ₁ and S ₂ satisfy the third condition of S ₁ ≦ 0 and S ₂ <0. In the region Q6, the sensor values S ₁ and S ₂ satisfy the fourth condition of S ₁ <0 and S ₂ ≧ 0. Therefore, when the first rotation angle calculator 77A satisfies one of the first condition and the third condition, the first magnetic sensor 71 detects the pole number of the magnetic pole detected by the second magnetic sensor 72. It is discriminated that it is the same as the pole number of the magnetic pole. On the other hand, when neither the first condition nor the third condition is satisfied, the first rotation angle calculation unit 77A indicates that the pole number of the magnetic pole detected by the second magnetic sensor 72 is the first magnetic sensor. 71 is determined to be one larger than the pole number of the magnetic pole detected.

  In FIG. 15, in the areas indicated by Q <b> 1 and Q <b> 4, the pole number of the magnetic pole detected by the third magnetic sensor 73 is the same as the pole number of the magnetic pole detected by the first magnetic sensor 71. On the other hand, in the regions indicated by Q2, Q3, Q5, and Q6, the pole number of the magnetic pole detected by the third magnetic sensor 73 is one greater than the pole number of the magnetic pole detected by the first magnetic sensor 71.

In the region Q1, the sensor values S ₁ and S ₃ satisfy the fifth condition of S ₁ ≧ 0 and S ₃ > 0. In the regions Q2 and Q3, both sensor values S ₁ and S ₃ satisfy the sixth condition of S ₁ > 0 and S ₃ ≦ 0. In the region Q4, both sensor values S ₁ and S ₃ satisfy the seventh condition of S ₁ ≦ 0 and S ₃ <0. In the regions Q5 and Q6, the sensor values S ₁ and S ₃ satisfy the eighth condition of S ₁ <0 and S ₂ ≧ 0. Therefore, when the first rotation angle calculation unit 77A satisfies one of the fifth condition and the seventh condition, the first magnetic sensor 71 detects the pole number of the magnetic pole detected by the third magnetic sensor 73. It is discriminated that it is the same as the pole number of the magnetic pole. On the other hand, when neither the fifth condition nor the seventh condition is satisfied, the first rotation angle calculation unit 77A indicates that the pole number of the magnetic pole detected by the third magnetic sensor 73 is the first magnetic sensor. 71 is determined to be one larger than the pole number of the magnetic pole detected.

Returning to FIG. 12A, when it is determined in step S24 that the current process is not the first process after the start of the rotation angle calculation process based on the forced rotation (step S24: NO), the process proceeds to step S26.
In step S26, the first rotation angle computation unit 77A, based on the sensor values _{S 1,} S 2, _{S 3} stored in the memory, for each sensor value _{S 1,} S 2, _{S 3,} the sensor value It is determined whether or not a zero cross whose sign is inverted is detected. When the zero cross is not detected (step S26: NO), the first rotation angle calculation unit 77A proceeds to step S28.

If a zero cross is detected for any of the sensor values S ₁ , S ₂ , S _{3 in} step S26 (step S26: YES), the first rotation angle calculator 77A determines the relative pole number. Is updated (step S27). Specifically, the first rotation angle calculator 77A uses the relative pole number r1, r2, or r3 currently set for the magnetic sensor in which the zero cross is detected in step S26 as the input shaft 8 (magnet 61) The number is changed to a number larger by 1 or a number smaller by 1 depending on the rotation direction.

  When the rotation direction of the input shaft 8 is the positive direction (the direction indicated by the arrow in FIG. 6), the first rotation angle calculation unit 77A is currently set for the magnetic sensor in which the zero cross is detected in step S26. The relative pole number r1, r2 or r3 being updated is updated to a number larger by one. On the other hand, when the rotation direction of the input shaft 8 is the reverse direction, the first rotation angle calculation unit 77A has the relative pole numbers r1, r2 currently set for the magnetic sensor in which the zero cross is detected, or Update r3 to a number smaller by one. However, as described above, the relative pole number that is smaller by 1 than the relative pole number of “1” is “8”. Further, the relative pole number that is larger by 1 than the relative pole number of “8” is “1”.

In addition, the rotation direction of the input shaft 8 can be determined based on, for example, the previous value and the current value of the output signal in which the zero cross is detected, and the current value of the other output signal. Specifically, when the output signal in which the zero cross is detected is the first output signal S ₁ , “the previous value of the first output signal S ₁ is greater than 0 and the current value is less than or equal to 0, The condition that the second output signal S ₂ is smaller than 0 (the third output signal S ₃ is smaller than 0), or “the previous value of the first output signal S ₁ is less than 0 and its current value is greater than or equal to 0; when the second output signal S ₂ is (a third output signal S ₃ greater than 0) greater than 0 satisfies the condition that "the direction of rotation If it is positive (the direction indicated by the arrow in FIG. 6) Determined.

On the other hand, “the previous value of the first output signal S ₁ is 0 or more and the current value is less than 0, and the second output signal S ₂ is greater than 0 (the third output signal S ₃ is greater than 0)”. The condition, or “the previous value of the first output signal S ₁ is less than or equal to 0 and the current value is greater than 0 and the second output signal S ₂ is less than 0 (the third output signal S ₃ is less than 0)” If the condition is satisfied, it is determined that the rotation direction is the reverse direction.

When the output signal in which the zero cross is detected is the second output signal S ₂ , “the previous value of the second output signal S ₂ is greater than 0 and the current value is less than or equal to 0, and the first output signal S ₁ Is greater than 0 (the third output signal S ₃ is less than 0), or “the previous value of the second output signal S ₂ is less than 0 and its current value is greater than or equal to 0, and the first output signal S _{when one} meets a condition that 0 is less than (the third output signal S ₃ is greater than zero) ", the rotation direction is determined to be a positive direction (the direction indicated by the arrow in FIG. 6). On the other hand, “the previous value of the second output signal S ₂ is 0 or more and the current value is less than 0, and the first output signal S ₁ is less than 0 (the third output signal S ₃ is greater than 0)”. The condition, or “the previous value of the second output signal S ₂ is less than or equal to 0 and the current value is greater than 0 and the first output signal S ₁ is greater than 0 (the third output signal S ₃ is less than 0)” If the condition is satisfied, it is determined that the rotation direction is the reverse direction.

When the output signal in which the zero cross is detected is the third output signal S ₃ , “the previous value of the third output signal S ₃ is greater than 0 and its current value is less than or equal to 0, and the second output signal S ₂ Is greater than 0 (the first output signal S ₁ is greater than 0), or “the previous value of the third output signal S ₃ is less than 0 and its current value is greater than or equal to 0, and the second output signal S _When the condition “ ₂ is smaller than 0 (first output signal S ₁ is smaller than 0)” is satisfied, the rotation direction is determined to be the positive direction (the direction indicated by the arrow in FIG. 6).

On the other hand, “the previous value of the third output signal S ₃ is 0 or more and the current value is less than 0, and the second output signal S ₂ is less than 0 (the first output signal S ₁ is less than 0)”. The condition, or “the previous value of the third output signal S ₃ is less than or equal to 0 and the current value is greater than 0 and the second output signal S ₂ is greater than 0 (the first output signal S ₁ is greater than 0)” If the condition is satisfied, it is determined that the rotation direction is the reverse direction.

When the relative pole number update process ends, the first rotation angle calculation unit 77A proceeds to step S28. In step S28, the first rotation angle calculation unit 77A determines whether or not the value of the second sensor failure flag F _A2 is zero. If the value of the second sensor failure flag F _A2 is 0 (step S28: YES), that is, if the second magnetic sensor 72 has not failed, the first rotation angle calculation unit 77A The process proceeds to step S31 of 12B.

  In step S31, the first rotation angle calculation unit 77A sets a condition that the first and second magnetic sensors 71 and 72 are normal and both detect the same magnetic pole for three calculation cycles continuously. It is determined whether or not it is satisfied. In each calculation cycle, the relative numbers of the magnetic poles detected by the first and second magnetic sensors 71 and 72 can be recognized by the first relative magnetic pole number r1 and the second relative magnetic pole number r2, respectively. Therefore, by storing the relative magnetic pole numbers r1 and r2 for a plurality of calculation periods from the predetermined calculation period to the current calculation period in the memory, the first and second magnetic sensors 71 and 72 can be continuously operated for three calculation periods. It can be determined whether or not the same magnetic pole is detected.

  When the condition of step S31 is not satisfied (step S31: NO), the first rotation angle calculation unit 77A has one magnetic pole in which the second and third magnetic sensors 72 and 73 are normal and both are the same. It is determined whether or not the condition that is detected continuously for three calculation cycles is satisfied (step S33). In each calculation cycle, the relative numbers of the magnetic poles detected by the second and third magnetic sensors 72 and 73 can be recognized by the second relative magnetic pole number r2 and the third relative magnetic pole number r3, respectively. Accordingly, by storing the relative magnetic pole numbers r2 and r3 for a plurality of calculation periods from a predetermined calculation period before the current calculation period in the memory, the second and third magnetic sensors 72 and 73 can be continuously operated for three calculation periods. It can be determined whether or not the same magnetic pole is detected.

When the condition of step S33 is not satisfied (step S33: NO), the first rotation angle calculation unit 77A returns to step S23 of FIG. 12A.
When it is determined in step S31 that the condition of step S31 is satisfied (step S31: YES), the first rotation angle calculation unit 77A uses the Eθ basic calculation formula (7 ) And E arithmetic expression (8), it is determined whether or not the condition that any denominator of the fraction is not zero is satisfied (step S32). When the condition of step S32 is not satisfied (step S32: NO), the first rotation angle calculation unit 77A returns to step S23 of FIG. 12A. On the other hand, when the condition of step S32 is satisfied (step S32: YES), the first rotation angle calculation unit 77A proceeds to step S35.

  If it is determined in step S33 that the condition of step S33 is satisfied (step S33: YES), the first rotation angle calculation unit 77A uses the EΘ basic calculation expression (32) used in the second calculation mode. ) And E arithmetic expression (33), it is determined whether or not the condition that any denominator of the fraction is not zero is satisfied (step S34). When the condition of step S34 is not satisfied (step S34: NO), the first rotation angle calculation unit 77A returns to step S23 of FIG. 12A. On the other hand, when the condition of step S34 is satisfied (step S34: YES), the first rotation angle calculation unit 77A proceeds to step S35.

  In step S35, the first rotation angle calculator 77A rotates the electric motor 18 in a second direction that is opposite to the first direction. Specifically, the first rotation angle calculator 77A transmits a second forced rotation command for rotating the electric motor 18 in the second direction to the motor control ECU 12. When receiving the second forced rotation command, the motor control ECU 12 rotates the electric motor 18 in the second direction.

Thereafter, the first rotation angle calculation unit 77A acquires the sensor values S ₁ [n], S ₂ [n], and S ₃ [n] of the magnetic sensors 71, 72, and 73 (step S36). The process of step S36 is repeatedly executed at predetermined calculation cycles until YES is determined in step S41, S45, or step S49 described later. Then, the first rotation angle calculation unit 77A calculates the sensor value sign for each of the sensor values S ₁ , S ₂ , S ₃ based on the sensor values S ₁ , S ₂ , S ₃ stored in the memory. It is determined whether or not an inverted zero cross is detected (step S37). When the zero cross is not detected (step S37: NO), the first rotation angle calculation unit 77A proceeds to step S39.

If a zero cross is detected for any of the sensor values S ₁ , S ₂ , S _{3 in} step S37 (step S37: YES), the first rotation angle calculator 77A determines the relative pole number. Is updated (step S38). The relative pole number update process is the same as the relative pole number update process in step S27 described above (see FIG. 12A). When the relative pole number update processing in step S38 is completed, the first rotation angle calculation unit 77A proceeds to step S39.

In step S39, the first rotation angle calculation unit 77A determines whether or not the value of the second sensor failure flag F _A2 is zero. When the value of the second sensor failure flag F _A2 is 0 (step S39: YES), that is, when the second magnetic sensor 72 has not failed, the first rotation angle calculation unit 77A The process proceeds to step S40 of 12C.
In step S40, the first rotation angle calculation unit 77A sets a condition that the first and second magnetic sensors 71 and 72 are normal and both detect the same magnetic pole for three calculation cycles continuously. It is determined whether or not it is satisfied. When the condition of step S40 is not satisfied (step S40: NO), the first rotation angle calculation unit 77A has one magnetic pole in which the second and third magnetic sensors 72 and 73 are normal and both are the same. It is determined whether or not the condition that is detected continuously for three calculation cycles is satisfied (step S44). When the condition of step S44 is not satisfied (step S44: NO), the first rotation angle calculation unit 77A returns to step S36 of FIG. 12B.

  If it is determined in step S40 that the condition of step S40 is satisfied (step S40: YES), the first rotation angle calculation unit 77A uses the Eθ basic calculation formula (7 ) And E arithmetic expression (8), it is determined whether or not the condition that any denominator of the fraction is not zero is satisfied (step S41). When the condition of step S41 is not satisfied (step S41: NO), the first rotation angle calculation unit 77A returns to step S36 of FIG. 12B.

When it is determined that the condition of step S41 is satisfied (step S41: YES), the first rotation angle calculation unit 77A calculates θ [n], E, A ₁ , A _{2 according} to the first calculation mode. Calculation is performed (step S42). Then, the first rotation angle calculator 77A associates the calculated E, A ₁ and A ₂ with the relative pole numbers of the magnetic poles detected by the first and second magnetic sensors 71 and 72 in the memory. Store (step S43). The relative pole numbers of the magnetic poles detected by the first and second magnetic sensors 71 and 72 have the same value as the currently set first relative pole number r1 or second relative pole number r2. Specifically, the first rotation angle calculator 77A calculates the calculated E, A in the storage location corresponding to the currently set first relative pole number r1 in the memory areas e1, e2, e3. ₁ and _{a 2} to be stored. Thereafter, the first rotation angle calculation unit 77A proceeds to step S52.

  If it is determined in step S44 that the condition of step S44 is satisfied (step S44: YES), the first rotation angle calculation unit 77A uses the EΘ basic calculation expression (32) used in the second calculation mode. ) And E arithmetic expression (33), it is determined whether or not the condition that any denominator of the fraction is not zero is satisfied (step S45). When the condition of step S45 is not satisfied (step S45: NO), the first rotation angle calculation unit 77A returns to step S36 of FIG. 12B.

If the condition of step S45 is satisfied (step S45: YES), the first rotation angle calculator 77A calculates θ [n], E, A ₂ , A ₃ in the second calculation mode (step S45). S46). Then, the first rotation angle calculator 77A associates the calculated E, A ₂ and A ₃ with the relative pole numbers of the magnetic poles detected by the second and third magnetic sensors 72 and 73 in the memory. Store (step S47). The relative pole numbers of the magnetic poles detected by the second and third magnetic sensors 72 and 73 have the same value as the currently set second relative pole number r2 or third relative pole number r3. Specifically, the first rotation angle calculator 77A calculates the calculated E, A in the storage location corresponding to the currently set second relative pole number r2 in the areas e1, e3, e4 of the memory. ₂ and _{a 3} are respectively stored. Thereafter, the first rotation angle calculation unit 77A proceeds to step S52.

  In step S52, the first rotation angle calculation unit 77A stops driving the electric motor 18 and stops the warning to the driver. Specifically, the first rotation angle calculation unit 77A transmits a drive stop command for the electric motor 18 to the motor control ECU 12, and transmits a warning stop command to the video / audio control device. When the motor control ECU 12 receives a drive stop command for the electric motor 18, the motor control ECU 12 stops the drive of the electric motor 18. When receiving the warning stop command, the video / audio control device stops warning display, warning voice output, and the like. Thereby, the rotation angle calculation process based on forced rotation ends.

Returning to FIG. 12A, if it is determined in step S28 that the value of the second sensor failure flag F _A2 is not 0 (step S28: NO), that is, the second magnetic sensor 72 is turned on immediately after the power is turned on. If it is determined that a failure has occurred, the first rotation angle calculation unit 77A proceeds to step S29. In step S29, the first rotation angle calculation unit 77A determines whether or not the first and third magnetic sensors 71 and 73 both satisfy the condition that the same magnetic pole is detected continuously for three calculation cycles. Determine. In each calculation period, the relative numbers of the magnetic poles detected by the first and third magnetic sensors 71 and 73 can be recognized by the first relative magnetic pole number r1 and the third relative magnetic pole number r3, respectively. Therefore, by storing the relative magnetic pole numbers r1 and r3 for a plurality of calculation periods from the predetermined calculation period to the current calculation period in the memory, the first and second magnetic sensors 71 and 73 can be continuously operated for three calculation periods. It can be determined whether or not the same magnetic pole is detected.

  When the condition of step S29 is not satisfied (step S29: NO), the first rotation angle calculation unit 77A returns to step S23. On the other hand, when it is determined that the condition of step S29 is satisfied (step S29: YES), the first rotation angle calculation unit 77A uses the Eθ basic calculation formula (35) and E used in the third calculation mode. It is determined whether or not the condition that any denominator of the fraction contained in the arithmetic expression (36) is not zero is satisfied (step S30). When the condition of step S30 is not satisfied (step S30: NO), the first rotation angle calculation unit 77A returns to step S23. On the other hand, when the condition of step S30 is satisfied (step S30: YES), the first rotation angle calculator 77A proceeds to step S35 of FIG. 12B and drives the electric motor 18 in the second direction.

In step S39 of FIG. 12B, when it is determined that the value of the second sensor failure flag _FA2 is not 0 (step S39: NO), that is, the second magnetic sensor 72 has failed immediately after the power is turned on. If it is determined that the rotation angle is calculated, the first rotation angle calculation unit 77A proceeds to step S48 in FIG. 12D.
In step S48, the first rotation angle calculator 77A determines whether or not the first and third magnetic sensors 71 and 73 both satisfy the condition that the same magnetic pole is detected continuously for three calculation cycles. Determine. When the condition of step S48 is not satisfied (step S48: NO), the first rotation angle calculation unit 77A returns to step S36 of FIG. 12B.

  If it is determined in step S48 that the condition of step S48 is satisfied (step S48: YES), the first rotation angle calculation unit 77A uses the Eθ basic calculation formula (35) used in the third calculation mode. ) And E arithmetic expression (36), it is determined whether or not the condition that any denominator of the fraction is not zero is satisfied (step S49). When the condition of step S49 is not satisfied (step S49: NO), the first rotation angle calculation unit 77A returns to step S36 of FIG. 12B.

When it is determined that the condition of step S49 is satisfied (step S49: YES), the first rotation angle calculation unit 77A sets θ [n], E, A ₁ , A ₃ in the third calculation mode. Calculation is performed (step S50). Then, the first rotation angle calculator 77A associates the calculated E, A ₁ and A ₃ with the relative pole numbers of the magnetic poles detected by the first and third magnetic sensors 71 and 73 in the memory. Store (step S51). The relative pole numbers of the magnetic poles detected by the first and third magnetic sensors 71 and 73 have the same value as the currently set first relative pole number r1 or third relative pole number r3. Specifically, the first rotation angle calculator 77A calculates the calculated E, A in the storage location corresponding to the currently set first relative pole number r1 in the memory areas e1, e2, e4. ₁ and _{a 3} are respectively stored. Thereafter, the first rotation angle calculation unit 77A proceeds to step S52.

16A, 16B, and 16C are flowcharts showing the procedure of the normal rotation angle calculation process in step S5 of FIG.
In the normal rotation angle calculation process, the first rotation angle calculation unit 77A uses the sensor value S ₁ [n] acquired in the failure determination process in step S1 of FIG. 10 (more specifically, step S11 of FIG. 11). , S ₂ [n], S ₃ [n], the rotation angle θ of the input shaft 18 is calculated.

First rotation angle computation unit 77A, based on the sensor values _{S 1,} S 2, _{S 3} stored in the memory, for each sensor value _{S 1,} S 2, _{S 3,} the sign of the sensor value is inverted It is determined whether a zero cross has been detected (step S81). When the zero cross is not detected (step S81: NO), the first rotation angle calculation unit 77A proceeds to step S83.
If a zero cross is detected for any of the sensor values S ₁ , S ₂ , S _{3 in} step S81 (step S81: YES), the first rotation angle calculator 77A determines the relative pole number. Is updated (step S82). The relative pole number update process is the same as the relative pole number update process in step S27 of FIG. 12A described above. When the relative pole number update process in step S82 ends, the first rotation angle calculation unit 77A proceeds to step S83.

In step S83, the first rotation angle calculation unit 77A determines whether or not the value of the second sensor failure flag F _A2 is zero. If the value of the second sensor failure flag F _A2 is 0 (step S83: YES), that is, if the second magnetic sensor 72 has not failed, the first rotation angle calculator 77A The process proceeds to S84.
In step S84, the first rotation angle calculation unit 77A sets a condition that the first and second magnetic sensors 71 and 72 are normal and both detect the same magnetic pole for three calculation cycles continuously. It is determined whether or not it is satisfied. If the condition of step S84 is satisfied (step S84: YES), the first rotation angle calculator 77A calculates θ [n], E, A ₁ , A ₂ in the first calculation mode (step S84). S85). When calculating the rotation angle θ [n] in the first calculation mode, the first rotation angle calculation unit 77A determines whether or not the denominator of the fraction included in the Eθ basic calculation formula (7) is zero. It is determined whether or not the denominator of the fraction included in E arithmetic expression (8) is not zero, and θ [n], E, A ₁ , A ₂ are calculated according to the determination results.

When the first rotation angle calculation unit 77A calculates θ [n], E, A ₁ , A ₂ , any denominator of the fractions included in the Eθ basic calculation formula (7) and the E calculation formula (8) is calculated. It is determined whether or not the condition that is not zero is also satisfied (step S86). If met the condition of step S86 (step S86: YES), the first rotation angle computation unit 77A is computed E, _{A 1} and _{A 2,} the first and second magnetic sensors 71 and 72 Is stored in the memory in association with the relative pole number of the magnetic pole detected by (step S87). The relative pole numbers of the magnetic poles detected by the first and second magnetic sensors 71 and 72 have the same value as the currently set first relative pole number r1 or second relative pole number r2. Specifically, the first rotation angle calculator 77A calculates the calculated E, A in the storage location corresponding to the currently set first relative pole number r1 in the memory areas e1, e2, e3. Store ₁ and A2, respectively. Note that memory areas e1, e2, e3 already E in the storage location, _{if A 1} and _{A 2} are stored, this time computed E, _{A 1} and _{A 2} are overwritten. Thereafter, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.

If it is determined in step S86 that the condition of step S86 has not been satisfied (step S86: NO), the first rotation angle calculation unit 77A does not perform the process of step S87, but at this normal time. This completes the rotation angle calculation process. Therefore, in this case, the calculated E in step S85, _{A 1} and _{A 2} are not stored in the area e1, e2, e3 of the memory.

If it is determined in step S84 that the condition in step S84 is not satisfied (step S84: NO), the first rotation angle calculator 77A indicates that the second and third magnetic sensors 72 and 73 are normal. Then, it is determined whether or not the condition that the same magnetic pole is detected for three consecutive calculation periods is satisfied (step S88). If the condition of step S88 is satisfied (step S88: YES), the first rotation angle calculator 77A calculates θ [n], E, A ₂ , A ₃ in the second calculation mode (step S88). S89). When calculating the rotation angle θ [n] in the second calculation mode, the first rotation angle calculation unit 77A determines whether the denominator of the fraction contained in the EΘ basic calculation expression (32) is not zero, It is determined whether or not the fractional denominator included in the E equation (33) is non-zero, and θ [n], E, A ₂ , A ₃ are calculated according to the determination results.

When the first rotation angle calculation unit 77A calculates θ [n], E, A ₂ , A ₃ , any denominator of the fractions included in the EΘ basic calculation expression (32) and the E calculation expression (33) is calculated. It is also determined whether or not the condition that is not zero is satisfied (step S90). When the condition of step S90 is satisfied (step S90: YES), the first rotation angle calculation unit 77A converts the calculated E, A ₂ and A ₃ into the second and third magnetic sensors 72 and 73. Are stored in the memory in association with the relative pole numbers of the magnetic poles detected (step S91). The relative pole numbers of the magnetic poles detected by the second and third magnetic sensors 72 and 73 have the same value as the currently set second relative pole number r2 or third relative pole number r3. Specifically, the first rotation angle calculator 77A calculates the calculated E, A in the storage location corresponding to the currently set second relative pole number r2 in the areas e1, e3, e4 of the memory. ₂ and _{a 3} are respectively stored. If E, A ₂ and A ₃ are already stored in the storage locations of the memory areas e1, e3 and e4, the currently calculated E, A ₂ and A ₃ are overwritten. Thereafter, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.

If it is determined in step S90 that the condition of step S90 is not satisfied (step S90: NO), the first rotation angle calculation unit 77A does not perform the process of step S91, and the current normal time This completes the rotation angle calculation process. Therefore, in this case, E, A ₂ and A ₃ calculated in step S89 are not stored in the memory areas e1, e3 and e4.

  If it is determined in step S88 that the condition of step S88 is not satisfied (step S88: NO), the first rotation angle calculation unit 77A proceeds to step S92 of FIG. 13B. In step S92, the first rotation angle calculation unit 77A stores in the memory an angle width error correction value E corresponding to the magnetic pole detected by the first magnetic sensor 71 when the first magnetic sensor 71 is normal. It is determined whether or not. Whether or not the angular width error correction value E corresponding to the magnetic pole detected by the first magnetic sensor 71 is stored in the memory is determined based on the currently set first relative pole number in the area e1 of the memory. This is performed based on whether or not the angular width error correction value E is stored in the storage location corresponding to r1.

  When the first magnetic sensor 71 is normal and the angle width error correction value E corresponding to the magnetic pole detected by the first magnetic sensor 71 is stored in the memory (step S92: YES), the first magnetic sensor 71 is normal. The rotation angle calculator 77A calculates the rotation angle θ [n] in the fourth calculation mode (step S93). Then, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.

  If it is determined in step S92 that the first magnetic sensor 71 has failed or the angle width error correction value E corresponding to the magnetic pole detected by the first magnetic sensor 71 is not stored in the memory. If it is determined (step S92: NO), the first rotation angle calculator 77A proceeds to step S94. In step S94, the first rotation angle calculation unit 77A determines whether or not the angular width error correction value E corresponding to the magnetic pole detected by the second magnetic sensor 72 is stored in the memory. This determination is made based on whether or not the angular width error correction value E is stored in the storage location corresponding to the currently set second relative pole number r2 in the area e1 of the memory.

When the angle width error correction value E corresponding to the magnetic pole detected by the second magnetic sensor 72 is stored in the memory (step S94: YES), the first rotation angle calculation unit 77A performs the fifth calculation. The rotation angle θ [n] is calculated according to the mode (step S93). Then, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.
If it is determined in step S94 that the angular width error correction value E corresponding to the magnetic pole detected by the second magnetic sensor 72 is not stored in the memory (step S94: NO), the first rotation The angle calculation unit 77A calculates the rotation angle θ [n] in the sixth calculation mode (step S96). Then, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.

Returning to FIG. 16A, if it is determined in step S83 that the value of the second sensor failure flag F _A2 is not 0 (step S83: NO), that is, if the second magnetic sensor 72 has failed. The first rotation angle calculation unit 77A proceeds to step S97 in FIG. 16C. In step S97, the first rotation angle calculation unit 77A determines whether or not the first and third magnetic sensors 71 and 73 both satisfy the condition that the same magnetic pole is detected continuously for three calculation cycles. Determine. When the condition of step S97 is satisfied (step S97: YES), the first rotation angle calculation unit 77A calculates θ [n], E, A ₁ , A ₃ in the third calculation mode (step S97). S98). When calculating the rotation angle θ [n] in the third calculation mode, the first rotation angle calculation unit 77A determines whether the fractional denominator included in the Eθ basic calculation formula (35) is not zero, It is determined whether or not the denominator of the fraction included in the E arithmetic expression (36) is not zero, and θ [n], E, A ₁ , A ₃ are calculated according to the determination results.

When the first rotation angle calculation unit 77A calculates θ [n], E, A ₁ , A ₃ , any denominator of the fractions included in the Eθ basic calculation formula (35) and the E calculation formula (36) is calculated. It is determined whether or not the condition that is not zero is also satisfied (step S99). When the condition of step S99 is satisfied (step S99: YES), the first rotation angle calculation unit 77A converts the calculated E, A ₁ and A ₃ into the first and third magnetic sensors 71 and 73. Are stored in the memory in association with the relative pole numbers of the magnetic poles detected (step S100). The relative pole numbers of the magnetic poles detected by the first and third magnetic sensors 71 and 73 have the same value as the currently set first relative pole number r1 or third relative pole number r3. Specifically, the first rotation angle calculator 77A calculates the calculated E, A in the storage location corresponding to the currently set first relative pole number r1 in the memory areas e1, e2, e4. ₁ and _{a 3} are respectively stored. When E, A ₁ and A ₃ are already stored in the storage locations of the memory areas e1, e2 and e4, the currently calculated E, A ₁ and A ₃ are overwritten. Thereafter, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.

If it is determined in step S99 that the condition of step S99 is not satisfied (step S99: NO), the first rotation angle calculation unit 77A does not perform the process of step S100, and the current normal time This completes the rotation angle calculation process. Therefore, in this case, the calculated E in step S98, _{A 1} and _{A 3} are not stored in the area e1, e2, e4 of the memory.

  If it is determined in step S97 that the condition of step S97 is not satisfied (step S97: NO), the first rotation angle calculator 77A corresponds to the magnetic pole detected by the first magnetic sensor 71. It is determined whether or not the angular width error correction value E to be stored is stored in the memory (step S101). This determination is performed based on whether or not the angular width error correction value E is stored in the storage location corresponding to the currently set first relative pole number r1 in the memory area e1.

When the angle width error correction value E corresponding to the magnetic pole detected by the first magnetic sensor 71 is stored in the memory (step S101: YES), the first rotation angle calculation unit 77A performs the fourth calculation. The rotation angle θ [n] is calculated according to the mode (step S102). Then, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.
If it is determined in step S101 that the angular width error correction value E corresponding to the magnetic pole detected by the first magnetic sensor 71 is not stored in the memory (step S101: NO), the first rotation The angle calculation unit 77A proceeds to step S103. In step S103, the first rotation angle calculator 77A determines whether or not the angular width error correction value E corresponding to the magnetic pole detected by the third magnetic sensor 73 is stored in the memory. This determination is made based on whether or not the angular width error correction value E is stored in the storage location corresponding to the currently set third relative pole number r3 in the area e1 of the memory.

When the angle width error correction value E corresponding to the magnetic pole detected by the third magnetic sensor 73 is stored in the memory (step S103: YES), the first rotation angle calculation unit 77A performs the sixth calculation. The rotation angle θ [n] is calculated according to the mode (step S104). Then, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time.
If it is determined in step S103 that the angular width error correction value E corresponding to the magnetic pole detected by the third magnetic sensor 73 is not stored in the memory (step S103: NO), the first rotation The angle calculation unit 77A calculates the rotation angle θ [n] in the seventh calculation mode (step S105). That is, the calculation angle θ [n] is calculated using the same calculation formulas (35), (35), and (37) as in the third calculation mode. Then, the first rotation angle calculation unit 77A ends the normal rotation angle calculation process at this time. The process of step S105 is executed when NO is determined in step S103, but it is unclear whether or not the situation determined as NO in step S103 actually occurs. Moreover, even if the situation determined as NO in step S103 occurs, the occurrence frequency is considered to be very low. In this embodiment, the process of step S105 is prepared just in case.

Note that step S86 of FIG. 16A may be omitted, and the process may proceed to step S87 when the process of step S85 is completed. Similarly, step S90 of FIG. 16A may be omitted, and when the process of step S89 is completed, the process may proceed to step S91. Similarly, step S99 of FIG. 16C may be omitted, and the process may proceed to step S101 when the process of step S98 ends. In the above-described embodiment, the process between the second magnetic sensor 72 and the first magnetic sensor 71 may be performed. The sum of the angular interval and the angular interval between the second magnetic sensor 72 and the third magnetic sensor 73 is 120 ° in electrical angle, and is set smaller than the angular width of one magnetic pole (180 ° in electrical angle). ing. For this reason, even if any one of the three magnetic sensors 71, 72, 73 fails, the two normal magnetic sensors both use the same magnetic pole for three sampling periods (three calculation periods). ) A state of continuous detection may occur.

  If the two normal magnetic sensors satisfy the condition that the same magnetic pole is detected continuously for three sampling periods (three calculation periods), the output of three normal magnetic sensors for three samplings By calculating the rotation angle of the rotating body based on the signal, it is possible to calculate a highly accurate rotation angle. As a result, even when any one of the three magnetic sensors fails, it is possible to calculate a highly accurate rotation angle when the above condition is satisfied.

  Further, in this embodiment, even when the condition that both normal two magnetic sensors detect the same magnetic pole continuously for three sampling periods (three calculation periods) is not satisfied, The rotation angle is calculated in any one of the fourth, fifth and sixth calculation modes. For this reason, even when the above conditions are not satisfied, a highly accurate rotation angle can be calculated.

  As mentioned above, although one Embodiment of this invention was described, this invention can also be implemented with another form. For example, in the above-described embodiment, the angular interval between the second magnetic sensor 72 and the first magnetic sensor 71 is 60 °, and the angular interval between the second magnetic sensor 72 and the third magnetic sensor 73 is 60 °. °. However, the sum of the angular interval between the second magnetic sensor 72 and the first magnetic sensor 71 and the angular interval between the second magnetic sensor 72 and the third magnetic sensor 73 is less than 180 ° in electrical angle. For example, the angle interval may be an angle other than 60 °. In this case, the angular interval between the second magnetic sensor 72 and the third magnetic sensor 73 and the angular interval between the second magnetic sensor 72 and the third magnetic sensor 73 may be different.

  In the above-described embodiment, in the first calculation mode, the second calculation mode, and the third calculation mode, when two predetermined magnetic sensors detect the same magnetic pole for three consecutive sampling periods, The rotation angle θ is calculated on the basis of output signals for three samplings in these two magnetic sensors. However, in the first calculation mode, the second calculation mode, and the third calculation mode, when two predetermined magnetic sensors detect the same magnetic pole for two consecutive sampling periods, these two magnetic sensors The rotation angle θ may be calculated based on the output signal for two samplings in FIG.

  In this case, one of the amplitude and angle width error correction values of the two magnetic sensors used for the calculation of the rotation angle θ is set as a fixed value. As a result, the rotation angle θ can be calculated based on the output signals for two samples from the two magnetic sensors. When the amplitudes of the two magnetic sensors are fixed values, the information stored in relation to the relative pole number is only the angle width error correction value. On the other hand, when the angle width error correction value is a fixed value, the information stored in relation to the relative pole number is only the amplitude of each magnetic sensor.

In addition, in the first calculation mode, the second calculation mode, and the third calculation mode, when two predetermined magnetic sensors detect the same one magnetic pole continuously for a predetermined number of sampling periods of 4 or more, these The rotation angle θ may be calculated based on the predetermined number of sampling output signals in the two magnetic sensors.
The present invention can be modified in various ways within the scope of the matters described in the claims.

  DESCRIPTION OF SYMBOLS 1 ... Electric power steering apparatus, 6 ... Steering shaft, 8 ... Input shaft, 9 ... Output shaft, 11 ... Torque sensor, 12 ... Motor control ECU, 18 ... Electric motor, 41 ... Current command value setting part, 61, 62 ... Magnets, 71 to 76 ... Magnetic sensors, 77 ... ECU for torque calculation, 77A ... First rotation angle calculation unit, 77B ... Second rotation angle calculation unit

Claims (3)

A rotation angle calculation device for calculating the rotation angle of a rotating body,
An annular multipolar magnet that rotates according to the rotation of the rotating body and has a plurality of magnetic poles;
Three magnetic sensors arranged side by side in the circumferential direction of the multipolar magnet and each outputting a sine wave signal having a phase difference in accordance with the rotation of the multipolar magnet, the center of the three magnetic sensors The sum of the angular interval between one magnetic sensor and the other magnetic sensor and the angular interval between the central magnetic sensor and the other magnetic sensor is less than 180 degrees in electrical angle 3 Two magnetic sensors,
Failure determination means for determining whether or not a failure has occurred in each of the magnetic sensors;
When at least two magnetic sensors including the central magnetic sensor among the three magnetic sensors are normal, the two normal magnetic sensors including the central magnetic sensor continuously use the same magnetic pole for a predetermined plurality of sampling periods. A first calculating means for calculating a rotation angle of the rotating body based on output signals for the predetermined plurality of samplings of the two magnetic sensors when the first condition that the two magnetic sensors are detected is satisfied;
When only the central magnetic sensor of the three magnetic sensors is out of order, two normal magnetic sensors excluding the central magnetic sensor both detect the same magnetic pole continuously for a predetermined sampling period. Rotation angle calculation, including second calculation means for calculating the rotation angle of the rotating body based on the output signals for the predetermined plurality of samplings of the two magnetic sensors when the second condition is satisfied apparatus. When the first calculation means satisfies the first condition, the output signals for the predetermined plurality of samplings of the two normal magnetic sensors including the central magnetic sensor always satisfy a certain requirement. Sometimes, means for calculating and storing information relating to the magnetic pole width of the magnetic pole detected by the two magnetic sensors and / or information relating to the amplitude of the output signals of the two magnetic sensors in association with the magnetic poles is included. And
When the second calculation means satisfies the second condition, the output signals for the predetermined plurality of samplings of two normal magnetic sensors excluding the central magnetic sensor always or when satisfying the second condition satisfy a certain requirement Sometimes, means for calculating and storing information relating to the magnetic pole width of the magnetic pole detected by the two magnetic sensors and / or information relating to the amplitude of the output signals of the two magnetic sensors in association with the magnetic poles is included. And
When at least two magnetic sensors including the central magnetic sensor among the three magnetic sensors are normal and the first condition is not satisfied, information on the magnetic pole width of the normal magnetic sensors and / or Using output signals for one sampling of two magnetic sensors including one magnetic sensor that detects magnetic poles associated with information related to amplitude, and the information stored by the first calculation means Third calculating means for calculating the rotation angle of the rotating body;
When only the central magnetic sensor of the three magnetic sensors is out of order, the normal two magnetic sensors excluding the central magnetic sensor that do not satisfy the second condition are detected. When information on the magnetic pole width and / or information on the amplitude is stored in association with at least one of the magnetic poles, the output signals for one sampling of two normal magnetic sensors, and the second The rotation angle calculation device according to claim 1, further comprising fourth calculation means for calculating a rotation angle of the rotating body using the information stored by the calculation means.   When the three magnetic sensors are a first magnetic sensor, a second magnetic sensor, and a third magnetic sensor, and the second magnetic sensor is the central magnetic sensor, the second magnetic sensor, the first magnetic sensor, The rotation angle according to claim 1 or 2, wherein an angular interval between the second magnetic sensor and the third magnetic sensor is 60 degrees in electrical angle, and an angular interval between the second magnetic sensor and the third magnetic sensor is 60 degrees in electrical angle. Arithmetic unit.

Images