JP7259675B2 - Vacuum pump and controller - Google Patents

Description

The present invention relates to vacuum pumps and controllers.

In a magnetically levitated vacuum pump, a pump rotor is magnetically levitated and supported by a magnetic bearing (see, for example, Patent Document 1). Magnetic bearings in vacuum pumps generally employ five-axis control in which a pump rotor is supported by four radial magnetic bearings and one axial magnetic bearing. In the magnetic bearing control, the target current value calculation of the magnetic bearing based on the position information of the pump rotor detected by the position sensor and the control amount calculation based on the actual bearing current value and the calculated target current value are performed for each axis. is performed on

In order to ensure sufficient controllability of bearing control, it is necessary to acquire position information of each axis at the same time and quickly reflect it in the magnetic bearing current. Therefore, in Patent Document 1, an FPGA (Field Programmable Gate Array) whose configuration can be set by a designer after manufacturing is adopted as a digital computing unit. By using FPGA, it is possible to parallelize the control of 5 axes.

JP 2019-60274 A

However, when FPGA is used, it has the drawback that FPGA is expensive and it is troublesome to change the program. Therefore, it is desired to employ an inexpensive digital calculator to replace the FPGA. In that case, not only is it inexpensive, but controllability similar to that of FPGA is required.

A vacuum pump according to a first aspect of the present invention includes: a magnetic bearing device having a plurality of magnetic bearings corresponding to a plurality of control shafts; a bearing drive unit that supplies current, a current value detection unit that detects the current value of the exciting current, and a position detection unit that detects the floating position of the pump rotor, wherein the bearing control unit is a multi-core processor a plurality of first arithmetic processes for calculating a target current value of the excitation current for each of the plurality of control axes based on the target floating position and the floating position detected by the position detection unit; a plurality of CPU cores sharing a plurality of second arithmetic processes for calculating an excitation current control signal for each of the plurality of control axes based on the target current value and the current value detected by the current value detection unit; Prepare.
A control device according to a second aspect of the present invention is a control device for a vacuum pump in which a rotationally driven pump rotor is magnetically levitated and supported by a plurality of magnetic bearings corresponding to a plurality of control shafts, wherein the magnetic bearings are excited A bearing control unit that controls current, a current value detection unit that detects a current value of the excitation current, and a position detection unit that detects a floating position of the pump rotor, wherein the bearing control unit is a multi-core processor. a plurality of first arithmetic processes for calculating a target current value of the excitation current for each of the plurality of control axes based on the target floating position and the floating position detected by the position detection unit; A plurality of CPU cores sharing a plurality of second arithmetic processes for calculating an excitation current control signal for each of the plurality of control axes based on the target current value and the current value detected by the current value detection unit. .

ADVANTAGE OF THE INVENTION According to this invention, cost reduction can be aimed at, suppressing the fall of bearing controllability.

FIG. 1 is a diagram showing an example of a vacuum pump equipped with a magnetic bearing device. FIG. 2 is a block diagram showing a schematic configuration of a motor control system and a magnetic bearing control system. FIG. 3 is a diagram showing a control block for the X1 axis. FIG. 4 is a diagram showing an example of a timing chart according to the embodiment. FIG. 5 is a diagram showing a timing chart of a comparative example. FIG. 6 is a timing chart for explaining the first modification. FIG. 7 is a timing chart for explaining the second modification. FIG. 8 is a timing chart for explaining the third modification. FIG. 9 is a diagram illustrating a multicore processor according to an embodiment; FIG. 10 is a diagram illustrating a modification of the multicore processor according to the embodiment; FIG. 11 is a diagram illustrating a multi-core processor of the first modified example. FIG. 12 is a diagram illustrating a multi-core processor of the second modification. FIG. 13 is a diagram illustrating a multi-core processor of a third modified example. FIG. 14 is a diagram illustrating a single-core processor of a comparative example; FIG. 15 is a schematic diagram showing the arrangement of electromagnets in a 5-axis control type magnetic bearing.

Embodiments for carrying out the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an example of a vacuum pump provided with a magnetic bearing device, and is a sectional view showing a schematic configuration of a pump main body 1A of a magnetically levitated turbo-molecular pump. Although not shown in FIG. 1, the turbo-molecular pump 1 includes a controller 1B (see FIG. 2) for driving and controlling the pump body 1A.

The magnetic bearings employed in the pump main body 1A are 5-axis control type magnetic bearings, comprising radial magnetic bearings 4x1 and 4y1 related to mutually orthogonal X1 and Y1 axes, and radial magnetic bearings related to mutually orthogonal X2 and Y2 axes. 4x2, 4y2 and an axial magnetic bearing 4z for the Z axis.

FIG. 15 is a diagram schematically showing the arrangement of the electromagnets of the magnetic bearings 4x1, 4y1, 4x2, 4y2 and 4z. The X1-axis magnetic bearing 4x1 has a pair of electromagnets 41p and 41m arranged opposite to each other with the rotor shaft 3 interposed therebetween. The Y1-axis magnetic bearing 4y1 has a pair of electromagnets 42p and 42m arranged opposite to each other with the rotor shaft 3 interposed therebetween. The X2 and Y2 axes of the magnetic bearings 4x2 and 4y2 are different from the X1 and Y1 axes in the axial support position with respect to the rotor shaft 3. The X2-axis magnetic bearing 4x2 has a pair of electromagnets 43p and 43m arranged opposite to each other with the rotor shaft 3 interposed therebetween. The Y2-axis magnetic bearing 4y2 has a pair of electromagnets 44p and 44m arranged opposite to each other with the rotor shaft 3 interposed therebetween. The Z-axis axial magnetic bearing 4z has a pair of electromagnets 45u and 45l arranged opposite to each other with a thrust disk 300 fixed to the rotor shaft 3 interposed therebetween.

Radial displacement sensors 5x1 and 5y1 are provided corresponding to the respective axes of the radial magnetic bearings 4x1 and 4y1. Similarly, radial displacement sensors 5x2 and 5y2 are provided corresponding to the respective axes of the radial magnetic bearings 4x2 and 4y2. An axial displacement sensor 5z is provided corresponding to the axial magnetic bearing 4z. A rotor shaft 3 to which the pump rotor 2 is fixed is magnetically levitated by magnetic bearings, and the floating position of the rotor shaft 3 is detected by radial displacement sensors 5x1, 5y1, 5x2, 5y2 and an axial displacement sensor 5z.

The rotor shaft 3 rotatably magnetically levitated by the magnetic bearings 4x1, 4y1, 4x2, 4y2 and 4z is rotationally driven by the motor 6. FIG. A brushless DC motor or the like is used as the motor 6, for example. When the magnetic bearings 4x1, 4y1, 4x2, 4y2 and 4z are not operating, the rotor shaft 3 is supported by emergency mechanical bearings 7a and 7b. Magnetic bearings 4x1, 4y1, 4x2, 4y2, 4z, displacement sensors 5x1, 5y1, 5x2, 5y2, 5z, motor 6 and mechanical bearings 7a, 7b are arranged on base 8. As shown in FIG.

The pump rotor 2 is formed with a plurality of stages of rotor blades 2a and a cylindrical portion 2b, which constitute a rotation-side exhaust function portion. On the fixed side, on the other hand, a fixed blade 8a and a screw stator 8b, which are fixed side exhaust function units, are provided. A plurality of stages of fixed blades 8a are stacked via spacers 9, and arranged alternately with the rotary blades 2a in the axial direction. The screw stator 8b is provided on the outer peripheral side of the cylindrical portion 2b with a predetermined gap therebetween.

FIG. 2 is a block diagram showing a schematic configuration of a motor control system and a magnetic bearing control system of the turbo-molecular pump 1. As shown in FIG. The controller 1B includes an AC/DC converter 20, a DC/DC converter 21, a DC power supply 23, an inverter 24, an excitation amplifier 25, a sensor circuit 26, a motor controller 30 and a bearing controller 31. An AC input from outside is converted to a DC output (DC voltage) by the AC/DC converter 20 . The DC voltage output from the AC/DC converter 20 is input to the DC/DC converter 21, and the DC/DC converter 21 generates a DC voltage for driving the motor and a DC voltage for driving the magnetic bearings.

A DC voltage for driving the motor is input to the inverter 24 . A DC voltage for driving the bearings is input to a DC power supply 23 . As described above, the magnetic bearings 4x1, 4y1, 4x2, 4y2 and 4z are 5-axis control type magnetic bearings, and each axis is provided with a pair of electromagnets. Each electromagnet is provided with an excitation amplifier 25 for supplying an excitation current. That is, ten excitation amplifiers 25 are provided. A sensor circuit 26 is provided for each of five pairs of displacement sensors 5x1, 5y1, 5x2, 5y2 and 5z provided corresponding to each of the five axes.

The motor control unit 30 drives and controls the motor 6 . The bearing control unit 31 drives and controls the magnetic bearings 4x1, 4y1, 4x2, 4y2 and 4z. A processor including a CPU, memory (RAM and ROM), and peripheral circuits is used for each of the motor control unit 30 and the bearing control unit 31 . As will be described later, at least the bearing control unit 31 is a so-called multi-core processor. Motor controller 30 is a single-core processor or a multi-core processor.

A signal 302 relating to the phase voltage and phase current of the motor 6 is input from the inverter 24 to the motor control unit 30 . A PWM control signal 301 is output from the motor control unit 30 to the inverter 24 for on/off controlling the switching element provided in the inverter 24 . The bearing controller 31 outputs a sensor carrier signal (carrier wave signal) 305 to each sensor circuit 26 . A floating position signal 306 modulated by a change in floating position is input from each sensor circuit 26 to the bearing control unit 31 . Further, the bearing control unit 31 outputs a PWM control signal 303 to each excitation amplifier 25 for controlling on/off switching elements provided in the excitation amplifier 25 . A current signal 304 relating to the excitation current of the magnetic bearing is input from each excitation amplifier 25 to the bearing control section 31 .

FIG. 3 shows the X1-axis control block for the magnetic bearing 4x1 of FIG. As shown in FIG. 15, the electromagnets 41m and 41p of the magnetic bearing 4x1 are arranged on the X1 axis with the rotor shaft 3 interposed therebetween. A displacement sensor 5x1 arranged near the magnetic bearing 4x1 is also arranged in the X1-axis direction so as to sandwich the rotor shaft 3 therebetween. The displacement sensor 5x1 detects displacement of the rotor shaft 3 in the X1-axis direction, and the sensor circuit 26 outputs a floating position signal 306 based on the displacement signal.

A difference value ε1 between the target floating position signal X1s and the floating position signal 306 from the sensor circuit (that is, floating position deviation) is input to the target current value computing section 311 of the bearing control section 31 . The target current value calculator 311 calculates a target current value Isx1 of the exciting current to be supplied to the magnetic bearing 4x1 based on the difference value ε1. A difference value ε2 between the target current value Isx1 and the current signal 304 detected by the current detection unit 25a of the excitation amplifier 25 is input to the control amount calculation unit 312 . The current detection unit 25a is, for example, a shunt resistor provided in the excitation amplifier 25, or the like. The control amount calculator 312 generates a control signal to the excitation amplifier 25 based on the difference value ε2. The excitation amplifier 25 is driven by the control signal, and an excitation current is supplied from the excitation amplifier 25 to the magnetic bearing 4x1.

In this manner, the bearing control unit 31 executes target current value calculation and control amount calculation in each axis control. In the case of a 5-axis control type magnetic bearing, 10 calculation processes, ie, 5 target current value calculations and 5 control amount calculations, are performed. In this embodiment, a multi-core processor having a plurality of CPU cores is used as the bearing control unit 31, and 10 arithmetic processes are shared by the plurality of CPU cores.

In the following explanation, X1-axis target current value calculation processing is X1A, Y1-axis target current value calculation processing is Y1A, X1-axis control amount calculation processing is X1B, Y1-axis control amount calculation processing is Y1B, Z-axis calculation processing is Let ZA be the target current value calculation process, and ZB be the Z-axis control amount calculation process.

FIG. 9 is a diagram showing the configuration of the multicore processor MP of the embodiment. The multi-core processor MP implements ten CPU cores 1-10. Target current value calculation and control amount calculation for two axes (X1 axis and Y1 axis) of magnetic bearings 4x1 and 4y1, that is, four calculation processes X1A, Y1A, X1B, and Y1B are performed by four CPU cores 1 to 4. Four CPU cores 5 to 8 perform target current value calculation processing X2A and Y2A and control amount calculation processing X2B and Y2B for two axes (X1 axis and Y1 axis) of the magnetic bearings 4x2 and 4y2. The two CPU cores 9 and 10 share the target current value calculation processing ZA and the control amount calculation processing ZB for the Z axis.

The multi-core processor MP includes 10 CPU cores 1 to 10, RAM, ROM, timing generator, peripheral circuits, etc., and these are connected by bus lines. Each of the CPU cores 1 to 10 has an arithmetic logic unit (ALU), an instruction register that temporarily stores the read instruction, an instruction decoder that decodes the instruction stored in the instruction register and sends it to the control unit, and an interrupt function. It consists of an interrupt control circuit for control and a program counter for managing the memory address of the instruction to be executed next.

FIG. 4 shows, for example, target current value calculation and control amount calculation for two axes (X1 axis and Y1 axis) of the magnetic bearings 4x1 and 4y1 in FIGS. 4 is a diagram showing an example of a timing chart when processing is shared by 4. FIG. On the other hand, FIG. 5 is a diagram showing a comparative example, and is a diagram showing an example of a timing chart when one CPU core processes four arithmetic processes for two axes.

First, a comparative example will be described in detail with reference to FIGS. 14 and 5. FIG. FIG. 14 is a diagram showing the configuration of a single-core processor DP of a comparative example, which includes one CPU core. The CPU core performs target current value calculation processing X1A, Y1A, X2A, Y2A, ZA and control amount calculation processing X1B, Y1B, X2B, Y2B, ZB.

FIG. 5 shows timing charts of X1-axis target current value calculation processing X1A, Y1-axis target current value calculation processing Y1A, X1-axis control amount calculation processing X1B, and Y1-axis control amount calculation processing Y1B. It is. Here, the target current value calculation process is performed at an interval of about Δt1=100 μs, and the control amount calculation process is performed at a shorter interval (for example, about Δt2=10 μs). Generally, target current value calculation in magnetic bearing control is more complicated than control amount calculation, and target current value calculation takes longer than control amount calculation.

At time t1, the X1-axis displacement signal ΔX1 and the Y1-axis displacement signal ΔY1 are acquired, and the X1-axis target current value calculation process X1A is started based on the acquired displacement signal ΔX1. At time t2, which is the control amount calculation processing timing, the target current value calculation processing X1A started at time t1 is temporarily interrupted, and the control amount calculation processing X1B for the X1 axis and the control amount calculation processing for the Y1 axis are performed. Execute Y1B in order. Then, when the control amount calculation process Y1B ends at time t3, the interrupted target current value calculation process X1A is resumed. As described above, Δt2 is about 1/10 of Δt1, and the control amount calculation process has a higher priority than the target current value calculation process. Therefore, the target current value calculation process X1A is temporarily interrupted and the control amount calculation processes X1B and Y1B are performed.

When the target current value calculation process X1A ends at time t4, the target current value used in the control amount calculation process X1B is updated at time t4 by the calculation result (target current value Isx1) obtained at time t4. When the target current value calculation process X1A ends at time t4, the Y1-axis target current value calculation process Y1A is started based on the displacement signal ΔY1 acquired at time t1. After that, when the interval Δt2 elapses from time t2 and reaches time t5, the target current value calculation process Y1A is interrupted, and the control amount calculation processes X1B and Y1B are executed in order. The control amount calculation process X1B is executed based on the target current value Isx1 updated at time t4, but the control amount calculation process Y1B is performed based on the target current value updated before time t4, that is, the previous target current value It is executed based on the calculation result of the calculation process Y1A.

When the control amount calculation process Y1B ends at time t6, the interrupted target current value calculation process Y1A is resumed. Then, at time t7 when the interval Δt2 has passed from time t5, the target current value calculation process Y1A is interrupted again, and the control amount calculation processes X1B and Y1B are executed in order. When the control amount calculation process Y1B ends at time t8, the interrupted target current value calculation process Y1A is resumed. When the target current value calculation process Y1A ends at time t9, the target current value used in the control amount calculation process Y1B is updated by the calculation result (target current value Isy1) obtained at time t9.

Thereafter, control amount calculation processes X1B and Y1B are sequentially executed from time t10. The control amount calculation process Y1B here is performed based on the target current value Isy1 updated at time t9. At time t11 when the interval Δt1 of the target current value calculation process has elapsed from time t1, the displacement signals ΔX1 and ΔY1 are acquired, and the X1-axis target current value calculation process X1A is started based on the acquired displacement signal ΔX1.

In this way, when four arithmetic processes for two axes are processed by one CPU core, two arithmetic processes cannot be executed in parallel. Therefore, as shown in FIG. Processed at a timing different from the current value calculation process Y1A, and when the target current value calculation processes X1A and Y1A are not performed, or when the target current value calculation processes X1A and Y1A are temporarily interrupted, the control amount calculation process X1B and Y1B are performed in order.

In general, it is preferable to simultaneously perform the processing for correcting the deviation of the rotor shaft 3 from the target floating position for all of the multiple shafts. Specifically, acquisition of the sensor displacement signal of each axis, update of the target current value, and calculation of the control amount are performed at the same timing. However, in the example of FIG. 5 in which one CPU core performs four arithmetic processes, the displacement signals ΔX1 and ΔY1 are acquired simultaneously for two axes, but the target current value arithmetic processes X1A and Y1A cannot be processed in parallel. is shifted. Therefore, it takes a long time to update the target current value Isy1 after obtaining the displacement signals ΔX1 and ΔY1.

The update timing of the target current value based on the displacement signals ΔX1 and ΔY1 acquired at time t1 is time t4 for the X1 axis and time t9 for the Y1 axis. Since the floating position of the rotor shaft 3 changes even during the calculation, the target current values Isx1 and Isy1 updated at time t4 and time t9 deviate from the target current values for the displacement signals at time t4 and time t9. become. As a result, the vibration of the rotor shaft 3 cannot be sufficiently reduced, which is an obstacle to the reduction of pump vibration.

On the other hand, in the example shown in FIG. 4, four arithmetic processes on two axes are processed in parallel by four CPU cores. Specifically, displacement signals ΔX1 and ΔY1 are acquired at time t1, and based on the displacement signals ΔX1 and ΔY1, target current value calculation processes X1A and Y1A are simultaneously started at time t1, and at the same time, control amount calculation processes X1B and Y1B are also started. have started. Therefore, the target current values Isx1 and Isy1 calculated in the target current value calculation processes X1A and Y1A are reflected in the control amount calculation processes X1B and Y1B which start at time t5 immediately after the calculation ends. That is, the time from detection of floating position deviation to reflection in the control current is shortened compared to FIG. 5, and vibration reduction can be improved. Furthermore, by using a multi-core processor having a plurality of CPU cores 1 to 4, the cost can be reduced compared to using an expensive FPGA.

FIG. 4 shows timing charts for four CPU cores on the two axes X1 and Y1, but another four CPU cores for the two axes X2 and Y2 and another two CPU cores for the Z axis. core is used.

In the embodiment shown in FIG. 4, the most preferable case has been described in which four arithmetic operations on two axes are individually processed in parallel by four CPU cores. If such a 2-axis-4-core configuration is applied to a 5-axis control type magnetic bearing, 10 CPU cores are required. In that case, a configuration in which ten CPU cores are provided in one multi-core processor is preferable in terms of simultaneously starting ten arithmetic processes.

Of course, in a configuration in which 10 CPU cores provided in a plurality of multi-core processors share processing, that is, a configuration in which multiple arithmetic processing is shared by a plurality of multi-core processors, it is necessary to take measures to match the start timing of processing. but it is possible. For example, as shown in FIG. 11, each of CPU cores 1-4, CPU cores 5-8, and CPU core 9 may be implemented in multi-core processors MP1-MP3.

As described above, in the example shown in FIG. 4, one CPU core is provided for one arithmetic process. all can be done at the same time. That is, the processing of the target current value calculation processes X1A and Y1A, which take a long time, can be completed simultaneously in a short time without being hindered by the control amount calculation processes X1B and Y1B. In the first to third modifications described below, it is not possible to perform deviation correction processing from the target levitation position at the same timing for all of the multiple axes, but a smaller number of CPU cores share the multiple arithmetic processing. The configuration will be explained.

(First modification)
In the first modified example, a case will be described in which four CPU cores share processing of 10 arithmetic processes on 5 axes, taking a 5-axis control type magnetic bearing as an example. FIG. 6 shows a timing chart in that case. FIG. 11 shows an example in which four CPU cores 1 to 4 are mounted on one multi-core processor MP. The two axes of the radial magnetic bearings 4x1 and 4y1 in FIGS. 1 and 2 are the X1 axis and the Y1 axis, the two axes of the radial magnetic bearings 4x2 and 4y2 are the X2 axis and the Y2 axis, and the axial magnetic bearing 4z is the Z axis. X1A, Y1A, X2A, Y2A and ZA in FIG. 6 represent the target current value calculation processing of the X1 axis, Y1 axis, X2 axis, Y2 axis and Z axis, and X1B, Y1B, X2B, Y2B and ZB represent the X1 axis. , Y1-axis, X2-axis, Y2-axis and Z-axis control amount calculation processing. CPU core 1 performs arithmetic processing of X1A and X2A, CPU core 2 performs arithmetic processing of Y1A, Y2A, and ZA, CPU core 3 performs arithmetic processing of X1B and X2B, and CPU core 4 performs arithmetic processing of Y1B. , Y2B and ZB are calculated.

In the example shown in FIG. 6, the X1-axis target current value calculation process X1A and the X2-axis target current value calculation process X2A are performed by the first CPU core 1 with different timings, and the Y1-axis target current value calculation process Y1A, The Y2-axis target current value calculation process Y2A and the Z-axis target current value calculation process ZA are performed by the second CPU core 2 with different timings, and the X1-axis control amount calculation process X1B and the X2-axis control amount calculation process X2B is performed by the third CPU core 3 with different timings, and the Y1-axis control amount calculation processing Y1B, the Y2-axis control amount calculation processing Y2B, and the Z-axis control amount calculation processing ZB are performed by the fourth CPU core 4. Do it staggered.

In FIG. 6, by shifting the processing timings of the target current value calculation processes X2A and Y2A and the control amount calculation processes X2B and Y2B with respect to the target current value calculation processes X1A and Y1A and the control amount calculation processes X1B and Y1B, , 8 processes are performed by four CPU cores 1-4. Furthermore, for the target current value calculation processing ZA and the control amount calculation processing ZB of the Z axis, the target current value calculation processing ZA and the control amount calculation processing ZB are performed during the idle time between the processing timing of the target current value calculation processing Y1A and the processing timing of the control amount calculation processing Y2B in the CPU core 2. The current value calculation process ZA is performed, and the control amount calculation process ZB is performed in the idle time between the processing timing of the control amount calculation process Y2B in the CPU core 4 and the control amount calculation process Y1B.

As described above, in the first modified example, between two axes having the same support position in the axial direction, such as the X1 axis and the Y1 axis, and the X2 axis and the Y2 axis, in order to prevent the deviation of the control timing. , acquisition of the sensor displacement signal, update of the target current value, and calculation of the control amount are performed at the same timing. On the other hand, by shifting the arithmetic processing timing between the control axes with different support positions in the axial direction, the processing is performed with a smaller number of CPU cores. Specifically, the target current value calculation process is performed by shifting the timing in the order of the X1 axis and Y1 axis, the Z axis, the X2 axis and the Y2 axis, and the control amount calculation process is performed on the X1 axis and Y1 axis, the X2 axis and Y2 axis. The timing was shifted in the order of the axis and the Z axis. As a result, two target current value calculation processes that require long processing times can be completed simultaneously in a short period of time without being hindered by the control amount calculation process between two shafts having the same axial support position.

At time t1, the displacement signals ΔX1 and ΔY1 are acquired, and based on the displacement signals ΔX1 and ΔY1, target current value calculation processing X1A for the X1 axis in the first CPU core 1 and Y1 axis calculation processing in the second CPU core 2 are performed. A target current value calculation process Y1A is started. The target current value calculation processing X1A, Y1A ends at time t4, and the target current values of the X1-axis and Y1-axis are updated to the calculated Isx1, Isy1, respectively. The control amount calculation process X1B in the third CPU core 3 and the control amount calculation process Y1b in the fourth CPU core 4, which are started at time t5, are performed based on the target current value updated at time t4. done.

Next, at time t20, the displacement signal ΔZ is obtained, and the second CPU core 2 starts the Z-axis target current value calculation processing ZA based on the displacement signal ΔZ. When the target current value calculation processing ZA is completed, the target current value of the Z axis is updated to Isz, and from the control amount calculation processing ZB in the fourth CPU core 4 starting at time t22, the updated target current value Isz A control amount calculation is performed. At time t21 after the target current value calculation process ZA is completed, X2-axis and Y2-axis displacement signals ΔX2 and ΔY2 are obtained, and X2-axis displacement signals in the first CPU core 1 are adjusted based on the displacement signals ΔX1 and ΔY1. Target current value calculation processing X2A and Y2-axis target current value calculation processing Y2A in the second CPU core 2 are started. When the target current value calculation processes X2A and Y2A are completed, the target current values for the X2-axis and Y2-axis are updated to Isx2 and Isy2, respectively. At time t23, the third and fourth CPU cores 3 and 4 start the control amount calculation processes X2B and Y2B based on the updated target current values Isx2 and Isy2.

(Second modification)
In the second modification, a case will be described in which three CPU cores share the processing of 10 calculations in 5 axes, taking a 5-axis control type magnetic bearing as an example. FIG. 7 is a timing chart for explaining the second modification. FIG. 12 shows an example in which three CPU cores 1 to 3 are mounted on one multi-core processor MP. The CPU core 1 performs calculation processing of X1A and X2A, the CPU core 2 performs calculation processing of Y1A, Y2A and ZA, and the CPU core 3 performs calculation processing of X1B, X2B, Y1B, Y2B and ZB.

In the example shown in FIG. 7, the X1-axis target current value calculation process X1A and the X2-axis target current value calculation process X2A are performed by the first CPU core 1 with different timings, and the Y1-axis target current value calculation process Y1A, The Y2-axis target current value calculation process Y2A and the Z-axis target current value calculation process ZA are performed by the second CPU core 2 with different timings. All of the value calculation processes X1B, Y1B, X2B, Y2B and ZB are performed by the third CPU core 3 with different timings.

In FIG. 7, as in the case of FIG. 6, the processing timings of the target current value calculation processes X2A and Y2A are shifted with respect to the processing timings of the target current value calculation processes X1A and Y1A, and the Z-axis target current value calculation process ZA is shifted. is performed in the idle time between the processing timing of the target current value calculation processing Y1A and the processing timing of the control amount calculation processing Y2B in the CPU 2, so that the two CPU cores 1 and 2 perform the five calculation processing. ing. Furthermore, the five control amount calculation processes X1B, Y1B, X2B, Y2B, and ZB are sequentially performed by one CPU core 3. FIG. As a result, in the second modification as well, between two shafts having the same support position in the axial direction, the two target current value calculation processes that take a long time can be performed in a short time without being hindered by the control amount calculation process. and can end at the same time. The second modification differs from the first modification in that the five control amount calculation processes X1B, Y1B, X2B, Y2B, and ZB are sequentially performed by one CPU core. The number of CPU cores can be reduced from 3 to 2, although the difference in arithmetic processing becomes large.

(Third modification)
FIG. 8 is a timing chart for explaining the third modification. FIG. 13 shows an example in which three CPU cores 1 to 3 are mounted on one multi-core processor MP. The CPU core 1 performs calculation processing of X1A and ZA, the CPU core 2 performs calculation processing of Y1A, and the CPU core 3 performs calculation processing of X1b, Y1B and ZB. In the third modified example, a case will be described where the present invention is applied to a three-axis control type magnetic bearing including two radial magnetic bearings and one axial magnetic bearing. In the example shown in FIG. 8, three CPU cores 1 to 3 share the processing of six calculation processes of the three-axis control type magnetic bearing.

In the case of a 3-axis control type magnetic bearing, it is composed of 3 axes, the X1 and Y1 axes in the radial direction and the Z axis in the axial direction. Six arithmetic operations with X1B, Y1B, and ZB are performed. The target current value calculation process X1A is performed by the first CPU core 1, and the target current value calculation processes Y1A and ZA are performed by the second CPU core 2 with different processing timings. is performed by the third CPU core 3 by shifting the processing timing, so that the three CPU cores share the processing of the six arithmetic processing.

At time t1, the displacement signals ΔX1 and ΔY1 are acquired, and based on the displacement signals ΔX1 and ΔY1, target current value calculation processing X1A for the X1 axis in the first CPU core 1 and Y1 axis calculation processing in the second CPU core 2 are performed. A target current value calculation process Y1A is started. The target current value calculation processing X1A, Y1A ends at time t4, and the target current values of the X1-axis and Y1-axis are updated to Isx1, Isy1, respectively. The third CPU core 3 starts the control amount calculation process X1B based on the target current value Isx1 at time t5, and when the control amount calculation process X1B ends at time t24, starts the control amount calculation process Y1B based on the target current value Isy1. do.

At time t20, the Z-axis target current value calculation process ZA is started in the second CPU core 2, and when the target current value calculation process ZA ends, the Z-axis target current value is updated to Isz1. After that, at time t21, the third CPU core 3 performs control amount calculation processing ZB based on the target current value Isz1.

In the case of a three-axis control type magnetic bearing, six CPU cores are required if each of the target current value calculation processing and the control amount calculation processing for each axis is processed by individual CPU cores as in the case of FIG. In the third modification, two target current value calculation processes Y1A and ZA are performed by the second CPU core 2, and three control amount calculation processes X1B, Y1B and ZB are performed by the third CPU core 3. . As a result, the target current value calculation processes X1A and Y1A, which take a long time, can be completed simultaneously in a short time without being disturbed by the control amount calculation processes X1B, Y1B, and ZB. can be reduced to three.

It will be appreciated by those skilled in the art that the multiple exemplary embodiments and variations described above are specific examples of the following aspects.

[1] A vacuum pump according to one aspect includes a magnetic bearing device that has a plurality of magnetic bearings corresponding to a plurality of control shafts, magnetically levitates and supports a rotationally driven pump rotor, and controls an excitation current to the magnetic bearings. a current value detection unit that detects the current value of the excitation current; and a position detection unit that detects the levitation position of the pump rotor, wherein the bearing control unit is a multi-core processor and has a target levitation a plurality of first arithmetic processes for calculating a target current value of the excitation current for each of the plurality of control axes based on the position and the levitation position detected by the position detection unit; and the calculated target current value. and the current value detected by the current value detection unit, a plurality of CPU cores sharing a plurality of second calculation processes for calculating an excitation current control signal for each of the plurality of control axes.

For example, as shown in FIG. 4, two first calculation processes (target current value calculation processes X1A and Y1A) for calculating the target current values Isx1 and Isy1 of the two control axes X1 and Y1, and the excitation current control signal Four CPU cores (CPU1 to CPU4) share two second calculation processes (control amount calculation processes X1B and Y1B) for calculating two control amounts. As a result, the detection results of the displacements ΔX1 and ΔY1 can be reflected in the excitation current with a time delay of about "(target current value calculation time) to (target current value calculation time) + Δt2", thereby reducing costs and Controllability of the degree of identification can be obtained as in the case of parallel processing by FPGA.

[2] In the vacuum pump described in [1] above, the magnetic bearing device includes a first radial magnetic bearing and a second radial magnetic bearing for radially supporting the pump rotor at a first position in the axial direction; A five-axis control type magnetic bearing device having a third radial magnetic bearing and a fourth radial magnetic bearing for radially supporting the rotor at a second position in the axial direction, and an axial magnetic bearing for axially supporting the pump rotor. The bearing control unit is a multi-core processor, and includes a first CPU core that performs the first arithmetic processing relating to the first radial magnetic bearing and the third radial magnetic bearing, the second radial magnetic bearing, the A second CPU core that performs the first arithmetic processing on the fourth radial magnetic bearing and the axial magnetic bearing, respectively; and a third CPU core that performs the second arithmetic processing on the first radial magnetic bearing and the third radial magnetic bearing, respectively. and a fourth CPU core that performs the second arithmetic processing on the second radial magnetic bearing, the fourth radial magnetic bearing, and the axial magnetic bearing, respectively.

In the case of a 5-axis control type magnetic bearing device, as shown in FIG. The total number of X1B, Y1B, X2B, Y2B and ZB is ten. In this case, as shown in FIG. 6, the target current value calculation processes X1A and X2A are performed by the first CPU core 1, the target current value calculation processes Y1A, Y2A and ZA are performed by the second CPU core 2, and the control amount The number of CPU cores can be reduced to four by performing the arithmetic processing X1B, X2B in the third CPU core 3 and performing the control amount arithmetic processing Y1B, Y2B, ZB in the fourth CPU core 4. FIG.

In this configuration, target current value calculations and control amount calculations for the X1-axis and Y1-axis having the same axial support position are processed at the same timing by different CPU cores. Similarly, target current value calculations and control amount calculations for the X2-axis and Y2-axis having the same axial support position are also processed at the same timing by different CPU cores. As a result, there is no timing deviation between the control axes with the same axial support position, and the reflection of the detected displacement to the control can be shortened as much as possible. Therefore, it is possible to reduce costs while suppressing deterioration in bearing controllability.

[3] In the vacuum pump described in [1] above, the magnetic bearing device includes a first radial magnetic bearing and a second radial magnetic bearing for radially supporting the pump rotor, and axially supporting the pump rotor. a three-axis control type magnetic bearing device having an axial magnetic bearing that performs the following: the bearing control unit is a multi-core processor; a first CPU core that performs the first arithmetic processing related to the first radial magnetic bearing; a second CPU core that performs the first arithmetic processing relating to the second radial magnetic bearing and the axial magnetic bearing; and the second arithmetic processing relating to the first radial magnetic bearing, the second radial magnetic bearing and the axial magnetic bearing. and a third CPU core for each processing.

In the case of a three-axis control type magnetic bearing device, as shown in FIG. The total number with ZB is six. In this case, as shown in FIG. 8, target current value calculation processes X1A and X2A are performed by the first CPU core 1, target current value calculation process ZA is performed by the second CPU core 2, and control amount calculation processes X1B and X2A are performed. By performing Y1B and ZB with the third CPU core 3, the number of CPU cores can be reduced to three.

In this configuration, for the X1 and Y1 axes that have the same axial support position, the target current value calculations are processed by different CPU cores 1 and 2 at the same timing, and the control amount calculations are processed by another CPU core 3. be. As a result, there is no timing deviation between the control axes with respect to the X1 axis and the Y1 axis, and the reflection of the detected displacement to the control can be shortened as much as possible. Therefore, it is possible to reduce costs while suppressing deterioration in bearing controllability.

[4] A bearing control device according to one aspect is a control device for a vacuum pump in which a rotationally driven pump rotor is magnetically levitated and supported by a plurality of magnetic bearings corresponding to a plurality of control shafts, wherein the magnetic bearings are excited A bearing control unit that controls current, a current value detection unit that detects a current value of the excitation current, and a position detection unit that detects a floating position of the pump rotor, wherein the bearing control unit is a multi-core processor. a plurality of first arithmetic processes for calculating a target current value of the excitation current for each of the plurality of control axes based on the target floating position and the floating position detected by the position detection unit; a plurality of CPU cores sharing a plurality of second arithmetic processes for calculating an excitation current control signal for each of the plurality of control axes based on the target current value and the current value detected by the current value detection unit; , provided.

For example, as shown in FIG. 2, a plurality of magnetic bearings 4x1, 4y1, 4x2, 4y2, 4z that support the pump rotor of the turbomolecular pump 1 by magnetic levitation are controlled by a controller 1B. As shown in FIG. 4, two first calculation processes (target current value calculation processes X1A and Y1A) for calculating the target current values Isx1 and Isy1 of the two control axes X1 and Y1, and two Four CPU cores 1 to 4 share the two second calculation processes (control amount calculation processes X1B and Y1B) for calculating the control amount. As a result, the detection results of the displacements ΔX1 and ΔY1 can be reflected in the excitation current with a time delay of about "(target current value calculation time) to (target current value calculation time) + Δt2", thereby reducing costs and Controllability of the degree of identification can be obtained as in the case of parallel processing by FPGA.

In the above-described embodiment, each CPU core is provided for each arithmetic processing of each control axis, and in the first to third modifications, the case of 5-axis 4-core, 5-axis 3-core, and 3-axis 3-core was explained. However, the number of control axes of the magnetic bearing device, the number of CPU cores for each arithmetic process of each control axis, and the combination of arithmetic processes processed by the same CPU core are not limited to the above examples.

Although various embodiments and modifications have been described above, the present invention is not limited to these contents. Other aspects conceivable within the scope of the technical idea of the present invention are also included in the scope of the present invention.

For example, in the above-described embodiments, a turbo-molecular pump was described as an example, but any vacuum pump supported by magnetic bearings is not limited to a turbo-molecular pump. Further, with regard to the magnetic bearing, the present invention is not limited to the magnetic bearing using the independent displacement sensor as described above, but can also be applied to a vacuum pump equipped with a sensorless type (also called self-sensing type) magnetic bearing with a different rotor displacement detection method. can be applied as well. In the case of a sensorless magnetic bearing, a sensing current component is superimposed on an exciting current, and rotor displacement is detected by detecting modulation of the sensing current component due to rotor displacement. Although the displacement detection method is thus different, target current value calculation and control amount calculation based on the detected rotor displacement are performed in the same manner as in the above-described embodiment.

REFERENCE SIGNS LIST 1 turbomolecular pump 1A pump body 1B controller 2 pump rotor 3 rotor shaft 4x1, 4y1, 4x2, 4y2 radial magnetic bearing 4z axial magnetic bearing 5x1, 5y1, 5x2, 5y2 Radial displacement sensor 5z Axial displacement sensor 6 Motor 25a Current detection unit 31 Bearing control unit 311 Target current value calculation unit 312 Control amount calculation unit

Claims (5)

a magnetic bearing device having a plurality of magnetic bearings corresponding to a plurality of control axes and supporting a rotationally driven pump rotor by magnetic levitation;
a bearing control unit that controls excitation currents to the plurality of magnetic bearings;
a current value detection unit that detects the current value of the excitation current;
a position detection unit that detects the floating position of the pump rotor,
The bearing control unit is a multi-core processor comprising a plurality of CPU cores ,
Some CPU cores among the plurality of CPU cores calculate a target current value of the excitation current for each of the plurality of control axes based on the target levitation position and the levitation position detected by the position detection unit. Execute the first arithmetic processing that
The remaining CPU core performs second arithmetic processing for calculating an excitation current control signal for each of the plurality of control axes based on the calculated target current value and the current value detected by the current value detection unit. Execute,
Vacuum pump. A vacuum pump according to claim 1,
The magnetic bearing device includes a first radial magnetic bearing and a second radial magnetic bearing for radially supporting the pump rotor at a first axial position, and a second radial magnetic bearing for radially supporting the pump rotor at a second axial position. A five-axis control type magnetic bearing device having three radial magnetic bearings, a fourth radial magnetic bearing, and an axial magnetic bearing for axially supporting the pump rotor,
the bearing control unit is a multi-core processor;
a first CPU core that respectively performs the first arithmetic processing relating to the first radial magnetic bearing and the third radial magnetic bearing;
a second CPU core that performs the first arithmetic processing on the second radial magnetic bearing, the fourth radial magnetic bearing, and the axial magnetic bearing;
a third CPU core that performs the second arithmetic processing on the first radial magnetic bearing and the third radial magnetic bearing;
and a fourth CPU core that performs the second arithmetic processing on the second radial magnetic bearing, the fourth radial magnetic bearing, and the axial magnetic bearing, respectively. A vacuum pump according to claim 1,
The magnetic bearing device is a three-axis control type magnetic bearing having a first radial magnetic bearing and a second radial magnetic bearing for radially supporting the pump rotor, and an axial magnetic bearing for axially supporting the pump rotor. a device,
the bearing control unit is a multi-core processor;
a first CPU core that performs the first arithmetic processing related to the first radial magnetic bearing;
a second CPU core that respectively performs the first arithmetic processing regarding the second radial magnetic bearing and the axial magnetic bearing;
A vacuum pump, comprising: a third CPU core that performs the second arithmetic processing regarding the first radial magnetic bearing, the second radial magnetic bearing, and the axial magnetic bearing, respectively. 4. The vacuum pump according to claim 1, wherein said first arithmetic processing and said second arithmetic processing are simultaneously executed by a plurality of CPU cores. A control device for a vacuum pump in which a rotationally driven pump rotor is magnetically levitated and supported by a plurality of magnetic bearings corresponding to a plurality of control shafts,
a bearing control unit that controls the excitation current of the magnetic bearing;
a current value detection unit that detects the current value of the excitation current;
a position detection unit that detects the floating position of the pump rotor,
The bearing control unit is a multi-core processor comprising a plurality of CPU cores ,
Some CPU cores among the plurality of CPU cores calculate a target current value of the excitation current for each of the plurality of control axes based on the target levitation position and the levitation position detected by the position detection unit. Execute the first arithmetic processing that
The remaining CPU core performs second arithmetic processing for calculating an excitation current control signal for each of the plurality of control axes based on the calculated target current value and the current value detected by the current value detection unit. run ,
Control device.

Images