WO1998031946A1

WO1998031946A1 - Method and device for regulating a sensor-free magnetic bearing

Info

Publication number: WO1998031946A1
Application number: PCT/CH1998/000014
Authority: WO
Inventors: Ladislav Kucera
Original assignee: Eidgenössische Technische Hochschule Eth-Zentrum
Priority date: 1997-01-16
Filing date: 1998-01-16
Publication date: 1998-07-23

Abstract

The invention relates to a sensor-free magnetic bearing actuated according to the modulation method. The position information required for regulating the magnetic bearing is obtained from the air-gap dependent properties of the impedance of the electromagnets. This impedance is measured at high-frequency, with regulating structures, filters and limiters being provided to prevent the power actuator of the magnetic bearing from disturbing the measurement of impedance. The invention further relates to devices corresponding to said method, in which both the input of the high-frequency modulation signal (60') and the analysis of the impedance of the electromagnet (50) are effected with an external measuring inductance (51) or at the midpoint tap of the electromagnet. This has the advantage of producing a high-quality measurement signal (62') at low modulation currents.

Description

METHOD AND DEVICE FOR CONTROLLING A SENSORLESS MAGNETU.GER

The invention relates to methods and devices with which the distance between the electromagnets and the rotor of a magnetic bearing system can be regulated by means of a high-frequency impedance measurement in accordance with patent claims 1 and 8.

Magnetic bearings enable contactless storage of rotors and work completely free of wear, maintenance and lubricants. In various applications, they therefore offer significant advantages over conventional storage methods. The sensorless magnetic bearing is a special form of magnetic bearing in which no external position sensor is required. The position information is obtained through the air gap-dependent properties of the magnetic bearing.

The main advantage of the sensorless magnetic bearing is the reduction in manufacturing costs. It also has other advantages that make it interesting for solving technical problems. The lack of a sensor carrier simplifies the design, assembly and maintenance of the magnetic bearing system. The rotor can be made more compact, which increases its natural frequencies. Because there are no cables to the sensor carrier, the probability of failure of the system is also reduced. Two basic procedures are known for sensorless magnetic bearing operation:

Modulation method: The structure of an electromagnet has the same structure as an inductive displacement sensor, and can also be operated as such. With a suitable spectral separation, in which the power controller is assigned the low-frequency range and the sensor system the high-frequency range, these two tasks can be carried out simultaneously with the same physical element. The sensorless magnetic bearing with modulation method is known from DE-OS 2537597. There is suggest superimposing the low-frequency control current with a high-frequency sensor current and measuring the air gap-dependent sensor voltage at the center taps of the magnetic bearing coils. MD Noh and EH Maslen use the high-frequency carrier frequency of the competitor's amplifier to obtain a position signal (Self-Sensing Magnetic Bearings, Fifth Int. Symp. On Magnetic Bearings, Kanazawa, Japan 1996, pp. 95-100).

Linear controller method: If the rotor moves in the air gap, voltage is induced across the coil of the electromagnet, similar to a rotating DC motor. This changes the course of the current through the coil. If the current is now measured and the voltage is controlled, the missing track conditions, position and speed of the rotor can be determined with an observer and then stabilized with a controller. Patent specification CH-678090 describes the modeling of the controlled system and the design of the controller in detail. EP-Al-0549912 describes a sensorless magnetic bearing system which combines the linear controller with the modulation method. The modulation signal is used to obtain a static position signal that removes the negative stiffness of the sensorless magnetic bearing.

Although the two methods of the sensorless magnetic bearing extract the position information from the air gap-dependent impedance of the magnetic bearing, they differ in the frequency ranges assigned to the sensors. The modulation method measures the impedance well above the frequency bandwidth of the controller, while the method with a linear controller implements the properties of the magnetic bearing impedance within the controller structure, ie sensors and controls have the same frequency range. The combination of sensors and control in the sensorless magnetic bearing with linear controller can be implemented in hardware with the simplest of means and at low cost. However, because there is no directly measurable position information in such a magnetic bearing arrangement, the commissioning of this method proves to be extremely difficult, especially since the overall system is very sensitive to stretching. kenparameter behaves. In addition to the commissioning problems, the parameter sensitivity also has a negative impact on the system robustness. The loss of robustness is considerable in comparison to a magnetic bearing system with an external position sensor. Since there is no way to further increase the robustness of this method, there is great interest in using the advantages of the sensorless magnetic bearing when using the modulation method despite the slightly increased hardware complexity.

The functionality of the modulation method is based on the property that an electromagnet can not only be used to generate forces (force controller), but also for non-contact distance measurements (position sensor). The execution of such a contact-free position measurement is linked to the structure and the functioning of inductive position sensors. With inductive position sensors, the distance to a metallic body is measured when using a measuring coil. Comparable to an electromagnet, the measuring coil consists of a core and one or more windings. The impedance of the measuring coil changes with the distance to the object to be measured and is converted into a position signal with the corresponding evaluation circuit. The structure of the modulation method typically comprises an evaluation circuit based on the method of inductive position sensors. This provides the position signal for the controller, which controls the electromagnets and keeps the rotor in the working point.

Such an evaluation circuit generally consists of a modulator, a measured variable and a demodulator. The task of the modulator is to excite the measuring system at high frequency, which must be fed in at the appropriate point. The modulation signal actually corresponds to a high-frequency voltage source, which preferably has a sinusoidal shape. The basic frequency of the modulator is referred to as the modulation frequency. The measurement variable or the measurement signal is set as a function of the transfer function "feed point of the modulation signal to the tap". The distance between the electromagnet and the rotor changes the amplitude of the measurement signal, which is evaluated by the demodulator. There are diverse implementation options for a demodulator. Examples are synchronous rectification and peak value rectification. The modulation frequency does not necessarily have to be constant. For example, measuring arrangements can be implemented using hysteresis switches or PLLs (Phase Locked Loop) resonant circuits, the oscillating frequency of which represents the measured variable. When realizing a sensor, it is important to aim for a measurement signal with the greatest possible change in amplitude or frequency. In the following, this circumstance is referred to as the path sensitivity of the position signal. In addition to the path sensitivity, each real position signal also has an interference sensitivity. This also generates an amplitude. Frequency change of the measurement signal. Of course it is important that the path sensitivity clearly outweighs the disturbance sensitivity.

1A shows how known sensorless magnetic bearings with modulation methods integrate the evaluation circuit for the position signal. The modulation signal 60 'of the modulation source 60 is added to the output signal 61 of the controller and fed into the power amplifier 30. The power amplifier can be designed as either a linear or switched amplifier. In the case of a switched amplifier, the switching frequency can be used as a modulation signal. A terminal 70 of the winding 69 of the electromagnet 50 is attached to the output of the power amplifier 30. The other connection 71 of the winding 69 is connected to a measuring inductance 51, which in turn is connected to ground, and forms the tap 62 for the measuring signal 62 '. The measurement signal 62 'is then fed to the demodulator 65, which converts it to a position signal x. The position signal describes the deviation of the rotor 31 from the operating point xO. The distance xl from the electromagnet 50 to the rotor 31 corresponds to the air gap of the magnetic bearing and is xl = xO - x. The impedances and signals within this description are understood in the frequency domain. Accordingly, they are complex quantities, which contain amplitude and phase information.

Although the implementation of the modulation scheme seems to be can be solved by simply incorporating known evaluation circuits of inductive position sensors, so far no magnetic bearing systems are known whose system robustness would have exceeded that of the method with a linear controller. The reasons for the actually unexpected lack of system robustness are on the one hand the inadequate measurement precision of the modulation arrangement, and on the other hand the peculiarities that an electromagnet dimensioned for the power actuator operation brings with it for a sensor task. The peculiarities of the electromagnets are described below.

The size of an electromagnet is given by the forces that it has to generate. For inductive position sensors, a ratio of the size of the measuring coil to the measuring range of 3: 1 is recommended. Transferred to the electromagnet, this means that the recommended air gap should be very large. There are applications with a large air gap, but as a rule the air gap is small so that the magnetic bearing can generate large bearing forces. As a result of the unfavorable ratio of the size of the electromagnet to the measuring distance, the maximum possible change in impedance in the permissible air gap area is reduced for the electromagnet, which manifests itself in a deterioration in displacement sensitivity.

In addition to the size, the electromagnet also differs in the core material. The core of the electromagnet typically consists of laminated cores that are optimized for motors and transformers. They are suitable up to frequencies of a few 100 Hz. At higher frequencies, the eddy current losses in the sheets increase significantly. The eddy current losses are used in various inductive position sensors to generate the measured variable. In the case of electromagnets, however, again as a result of the unfavorable ratio of the size of the electromagnet to the measuring range, the eddy current losses cause the path sensitivity to deteriorate with increasing modulation frequency.

In addition to their self-inductance, real coils also have a coil capacity. This is caused by the potential differences that occur between the adjacent windings. The coil capacitance is independent of the air gap and leads to a complete loss of the measurement signal at high modulation frequencies (approx. 100-500 kHz), ie the impedance of the electromagnet no longer changes despite the changing air gap. The coil capacity of an inductive position sensor is comparatively much smaller and accordingly allows significantly higher modulation frequencies.

Since the electromagnet causes a high magnetic flooding of the iron in the power controller mode, there is a noticeable change in the permeability of the iron when the electromagnet is controlled during the adjustment processes. A change in permeability creates a change in the impedance of the electromagnet and is included in the position signal as an additional sensitivity to interference. M.D. Noh and E.H. Maslen describe a compensation circuit that counteracts this effect (Self-Sensing Magnetic Bearings - Effects of Saturation, Fifth Int. Symp. On Magnetic Bearings, Kanazawa, Japan 1996, pp. 113-118). Since the compensation circuit must take into account very complex effects such as the nonlinear and hysteresis-superimposed permeability characteristic, a permeability-related portion of the interference sensitivity will always be retained in real systems.

The explanations about the peculiarities of the electromagnets make it clear that the path sensitivity in sensorless magnetic bearings is worse than that of inductive position sensors. In addition, the electromagnets have an interference sensitivity that is many times greater than that of inductive position sensors. The ratio of path sensitivity to interference sensitivity, hereinafter referred to as the signal-to-noise ratio, is low in principle for sensorless magnetic bearings. It is therefore of great importance to strive for evaluation circuits that utilize the maximum achievable path sensitivity without increasing the sensitivity to interference.

In the known arrangement according to FIG. 1A, there is a large loss of signal-to-noise ratio due to the modulation by means of power amplifier 30. At a modulation frequency of 10-100 kHz, the position-related change in amplitude of the measurement signal is only a few percent of the amplitude at the operating point. The power amplifier would therefore have to generate a modulation signal whose amplitude accuracy is far below the percentage range in order not to unnecessarily increase the interference sensitivity. Real power amplifiers cannot achieve the required amplitude precision because, in addition to the modulation, they also have to deliver high currents for power controller operation.

Among all modulation arrangements, only a method is listed in DE-OS 2537597 that does not modulate with the power amplifier. It uses a high frequency power source to generate the modulation signal. In practice, however, this is problematic since a current source is a regulated element which, depending on the load on the current source, tends to produce undesirable vibrations. It is difficult to realize a stable current source, since the system to be controlled has path non-linearities at the high frequencies, which result from the long leads to the electromagnets and the eddy current impedance of the electromagnets.

It is the object of the present invention to specify methods and devices with which the distance between the electromagnets and the rotor of a magnetic bearing system can be regulated by means of a high-frequency impedance measurement.

According to the invention, this object is achieved with a method according to the wording of claim 1 and a device according to the wording of claim 8. The invention is explained in more detail below with reference to the drawings. Show it:

Fig. 1A Known arrangement for an impedance measurement with measuring inductance

1B arrangement according to the invention for an impedance measurement with measuring inductance

Fig. 2 arrangement for an impedance measurement in several magnetic circuits

3 representation of a radial bearing with four feed put

Fig. 4A representation of a known thrust bearing in section

Fig. 4B representation of an axial bearing according to the invention in section with two magnetic circles in the bearing cross section

5A block diagram of a sensorless magnetic bearing system in a one-sided arrangement

5B block diagram of a sensorless magnetic bearing system in a two-sided arrangement

6A block diagram of an amplifier unit

6B basic circuit diagram of an electromagnet with center tap, a current measurement and a feed of the modulation signal

Fig. 6C block diagram of a demodulator for magnetic bearing systems in a one-sided arrangement

6D basic circuit diagram of a demodulator for magnetic bearing systems in a two-sided arrangement

Fig. 6E block diagram of a controller for magnetic bearing systems in a one-sided arrangement

Fig. 6F Block diagram of a controller for magnetic bearing systems in a two-sided arrangement

7A arrangement for feeding the modulation signal and for tapping the measured variable when using a measuring inductance

7B arrangement for feeding the modulation signal and for tapping the measured variable when using the center tap of a magnetic bearing with more than one magnetic circuit

7C-7E arrangement for feeding the modulation signal and for tapping the measured variable when using a transmitter

Fig. 8A Representation of a measured linearity curve of the position signal

8B shows a measured absolute deviation of the position signal from a reference sensor

FIG. 1B shows a separate feed in accordance with the invention of a precise modulation signal 60 'with low power. The output 61 of a controller is converted into a power amplifier 30 fed. The output of the power amplifier 30 is connected to the winding connection 70 of the winding 69 of an electromagnet 50. The other winding connection 71 is connected to a measuring inductance 51 and to a coupling impedance 63, and forms the tap 62 of the measuring signal 62 '. The measurement signal 62 'is then fed to a demodulator 65, which then converts it to a position signal x. The distance xl from the electromagnet 50 to the rotor 31 corresponds to the air gap of the magnetic bearing and is xl = xO - x. The coupling impedance 63 is complex and can be constructed in the form of a passive network. At modulation frequencies of 10-100 kHz, the amount of the coupling impedance is in the kΩ range. It can be seen from this that the modulation source 60 only has to deliver small currents in comparison to the power amplifier 30, and can accordingly be constructed very precisely with commercially available operational amplifiers.

The measuring inductance 51 is generally large and expensive, and also requires the power amplifier 30 to be oversized. In addition, it causes a reduction in the path sensitivity, which should be avoided for the reasons already mentioned. A waiver of the measuring inductance 51 would therefore be advantageous from several perspectives.

2 shows an arrangement for an impedance measurement in a plurality of magnetic circuits, as a result of which the above-mentioned problem points are avoided. Again, the output 61 of the controller is fed into the power amplifier 30. An electromagnet 52 no longer has the shape of a U-magnet with one winding, but rather the shape of an E-magnet with the two windings 69 'and 69''. This design change is necessary because an electromagnet structure with at least two magnetic circuits 66 ', 66''and 67 is required. The magnetic flux in the circles 66 'and 66''corresponds to the power flow, which is guided through a rotor 31 and which is required to generate the bearing forces. It is generated by connecting the output of the power amplifier 30 to the connection 72 of the coil windings 69 ′, 69 ″ connected in series, while the connection 74 is connected to ground (earth). The modulation signal 60 'is via the coupling impedance 63 at the center tapped tap 73 of the two windings, and generates in the magnetic circuit 67 the modulation flow, which superimposes the power flow. The center tap 73 also serves as a tap 62 of the measurement signal 62 ', which is converted by the demodulator 65 to the position signal x. It should be noted that an arrangement with a U-magnet with two windings does not offer the same possibilities, since without the second magnetic circuit either the power flow or the modulation flow is canceled, ie either the generation of the force or the measurement signal would be 0.

The type of modulation described is possible with a wide variety of winding structures in the corresponding electromagnetic structures. It is essential that at least one second magnetic circuit exists.

3 describes a modulation arrangement for a radial bearing, which is used to support the radial degrees of freedom of a rotor 31. The radial bearing has a stator 52 and is characterized by a common return yoke of the magnetic poles. Together with the backflow through the rotor 31, it has several magnetic circuits. Thus, no constructive or geometrical adjustments are necessary to implement a modulation arrangement comparable to FIG. 2. With both the power flow and the modulation flow, care must be taken that the polarities of the magnetic field alternate. N, S denote the north and south poles of the power flow; n, s the north and south poles of the modulation flow. The embodiment shown in FIG. 3 can be achieved by connecting the power amplifiers, which determine the polarity of the power flow, to the connections 80, 81, 82 and 83, while the connections 80 ', 81', 82 'and 83' with Earth. The power amplifiers deliver currents that flow through the coils according to the arrow directions. The modulation signals 60 'of the modulation sources 60 are fed in via the coupling impedances 63 at the center taps 84, 84', 85 and 85 '. In order to obtain the desired polarities of the modulation flow, the modulation signals 60 'are fed in at the center taps 85 and 85' with a negative sign. The opposite lying measurement signals 84 and 84 'or 85 and 85' are subtracted from each other, and then fed to the demodulators, which convert these difference signals to the position signals x and y.

4A shows a cross section through a known, industrially customary design of an axial bearing. This serves to stabilize the axial degree of freedom of a rotor 31. The axial bearing is divided into the core 33 and the winding 34 of the electromagnet and into the counterpart 32, which transmits the force to the rotor 31, and thus forms a so-called pot magnet.

FIG. 4B shows a cross section through an axial bearing design according to the invention, which permits modulation according to FIG. 2. The rotor 31 and the counterpart 32 correspond to those in Fig. 4A. A core 35 has the two windings here

36 and 37 of the electromagnet. The two windings 36 and

37 are connected in series, and generate the center tap at the connection point, which is equivalent to FIG. 2 for feeding the modulation signal 60 'via the coupling impedance 63 and as a measuring tap 62.

5A shows the block diagram of a magnetic bearing in a one-sided arrangement. The one-sided magnetic bearing is characterized in that the magnetic bearing only has an electromagnet that can only generate forces F1 in one direction. In order to achieve the equilibrium of forces, this arrangement requires a counterforce Fg, which must be generated either by the weight of the rotor 31 or by some other type of force source. The one-sided arrangement consists of an amplifier unit Gal, an electromagnet Gml including the modulation arrangement, a demodulator Gdl, a compensation circuit Gel and a regulator unit Grl. The output of the amplifier unit Gal controls a voltage ul across the coil of the electromagnet, which results in the current flow il through the coil. The measurement signal ul is converted by the demodulator Gdl to a position signal xn. Any non-linearities of the position signal xn can be partially linearized by the compensation circuit Gel depending on the current il become. The linearized position signal x is processed together with the current il by the controller unit Grl to an actuating signal vl and fed to the amplifier unit Gal. This closes the control loop that stabilizes the rotor at the operating point.

5B shows the block diagram of a magnetic bearing in a two-sided arrangement. The two-sided magnetic bearing is characterized in that the magnetic bearing has two electromagnets opposite the rotor 31, which generate opposing forces F1 and F2. The two-sided arrangement is the usual industrial version. It consists of two amplifier units Gal and Ga2, two electromagnets Gml and Gm2 including the modulation arrangement, a demodulator Gd2, a compensation circuit Gc2 and a controller unit Gr2. The outputs of the amplifier units Gal and Ga2 control the voltages ul and u2 across the coils of the two electromagnets Gml and Gm2. As a result, the currents il and i2 pass through the coils. The measurement signals um1 and um2 are converted by the demodulator Gd2 to a position signal xn. Any non-linearities of the position signal xn can be partially linearized by the compensation circuit Gc2 depending on the currents il or i2. The linearized position signal x is processed together with the currents il and i2 by the controller unit Gr2 to the control signals vl and v2, respectively, and fed to the amplifier units Gal and Ga2. This closes the control loop, via which the rotor is stabilized at the working point.

6A shows the basic circuit diagram of an amplifier unit Gal or Ga2. The power amplifier 30 supplies a current which is converted by the electromagnet into a force which attracts the rotor. Since the electromagnet is predominantly inductive, the current must be built up using the voltage of the power amplifier. For economic reasons, the power amplifier is usually not dimensioned so strongly that it could regulate every load case without amplifier saturation. Accordingly, the amplifier saturation belongs to an expected operating case and must be within the arrangement of the sensorless one Magnetic bearing are taken into account. When the level control of the power amplifier reaches amplifier saturation, higher harmonic interference signals are fed into the electromagnets. These interference signals affect the demodulation and thus undesirably get into the position information. Without appropriate measures, the voltage of the power amplifier begins to oscillate in high frequency during transient processes, the system recovering from this state after a certain time. This undesirable behavior must be prevented for industrial applications. This is done by taking over the voltage and current limits from special circuits outside the power amplifier, which respond earlier than the internal limits of the power amplifier. The voltage limitation is part of the amplifier unit Gal, while the current limitation is part of the regulator unit Grl and will be described later. 6A is divided into a limiter or limiter G12, a low-pass filter Gt2 and a commercially available power amplifier 30 with input 1 and output 2. Limiter G12 limits the input of the amplifier unit at a value slightly below the maximum possible output voltage of the Amplifier. The limiter generates spectrally high frequencies that must be damped by the low-pass filter Gt2. With this measure, the coupling of frequencies in the range of the modulation frequency is suppressed.

Fig. 6B shows the principle circuit diagram of an electromagnet with a center tap, a current measurement ^'and a feed of Modulationssignalε what has been shown as a function block Gml in Fig. 5A, or as functional blocks GML and Gm2 in Fig. 5B. The windings 53 and 53 'of an electromagnet 54 are connected to ground via a current measuring resistor 66. The modulation signal 60 'of a modulation source 60, which is grounded on one side, is connected to the center tap 67' of the windings 53 and 53 'via a coupling impedance 63 and a capacitance 67. The output voltage of the power amplifier is present at input 3 of winding 53. At the output 4 of the current measurement resistor 66, the current measurement signal and at the output 5, which is the connection point of the coupling impedance 63 and Capacitance 67 forms, the measurement signal tapped. In addition to the position of the rotor, a magnetic bearing system also regulates the current through the electromagnets. For this purpose, the current must be measured, as is done here when using a current measuring resistor 66. The current signal is then amplified and filtered. A capacitance 67 for potential isolation is connected to the center tap 67 'of the electromagnet 54. The purpose of this capacitance is to prevent low-frequency signals from the power amplifier and to largely transmit the modulation signals. The capacitance 67 is to be dimensioned such that the modulation source, which has a low supply voltage, is not driven at too low a frequency by the power amplifier, since saturation of the modulation source would lead to the loss of the position signal and thus to instability of the magnetic bearing system. In view of a simple spectral separation, it is desirable to choose the modulation frequency as high as possible. On the other hand, the sensor should have a good path sensitivity, but this is no longer the case when the frequencies are too high. The high-frequency impedance of the magnetic bearing is mainly characterized by the eddy current-induced coil inductance and the path-independent coil capacitance. With increasing frequency, the position-related change in the magnetic bearing impedance decreases with magnetic bearings due to the eddy currents and the coil capacity, the interference caused by the operation of the force controller increasing. At the resonance frequency of the magnetic bearing impedance, formed from the parallel resonant circuit, the coil inductance and the coil capacitance, laminated magnetic bearings have a high path sensitivity, so that this range is suitable for the modulation frequency. In magnetic bearings, where there are unbleached parts within the magnetic flux (e.g. axial bearings), this resonance range is strongly damped by the eddy currents, which is also transferred to the path sensitivity. For this type of magnetic bearing, it is essential to reduce the resonance frequency by means of an additional capacitance connected in parallel with the electromagnet, and to select the modulation frequency equal to the reduced resonance frequency.

6C shows a demodulator Gdl for a one-sided magnetic bearing arrangement in a schematic representation. As already mentioned, there are various demodulation arrangements. The present one is based on synchronous rectification. The measurement signal is present at input 6 of the demodulator. The offset is subtracted from this measurement signal by subtracting half the modulation signal 40 '. The modulation signal 40 'is obtained from the modulation signal 60' of the modulation source 60 by halving. It is assumed here that the coupling impedance 63 has been optimally designed. With an adjustable factor 40, the offset deduction can be set so that the output 7 of the demodulator is 0 V at the operating point of the rotor. After the offset has been subtracted from the measurement signal, the remaining signal 45 is fed to a bandpass Gb. The Bandpass Gb has two functions. On the one hand, it serves to keep frequencies outside the modulation frequency away from the demodulator. It should be noted here that the modulation frequency shifts depending on the movement of the rotor. The bandpass must neither dampen nor shift the modulation frequency. Both would worsen the dynamic position measurement. On the other hand, it is the task of the bandpass filter Gb to prevent the demodulator from being overdriven. In the case of transient control processes, the power amplifier controls the magnetic bearing with high voltages which, if the bandpass is dimensioned incorrectly, reach the demodulator and drive it to saturation. The result would be the loss of the position signal and thus instability of the magnetic bearing system. The demodulation takes place with an analog multiplier 43, which multiplies the output of the bandpass with the demodulation signal 64 'of a demodulation source 64. The demodulation signal is out of phase with the modulation signal. The adjustable gain factor 42 determines the slope of the position signal. The low-pass filter Gt3 connected downstream of the multiplier 43 removes from the position signal the signal generated by the synchronous rectification at twice the modulation frequency. The output 7 of the demodulator corresponds to the non-linearized position signal.

6D shows a demodulator Gd2 for a two-sided magnetic bearing arrangement in a schematic representation. This consists of the two inputs 14 and 15 of the measurement signals and the output 16 of the non-linearized position signal. The subtraction of the two measurement signals eliminates the offset, which had to be deducted separately for the one-sided magnetic bearing arrangement. Any offset error that could arise due to a slight asymmetry of the two-sided arrangement can be compared with the coefficient 41, after which an adjusted measurement signal 41 'is fed to the subtraction. After subtraction, the remaining signal 45 is fed to a bandpass Gb. The remaining part of the demodulator corresponds to the already described demodulator for one-sided magnetic bearings (FIG. 6C).

6E shows a controller unit Grl for magnetic bearings in a one-sided arrangement in a schematic representation. The controller inputs consist of the position signal 12 and the current through the electromagnet 11. The controller output 13 corresponds to the target voltage of the power amplifier. The speed of the rotor is required by the state control. It is derived from the position signal with the differentiator Dl. In order not to unnecessarily amplify higher-frequency interference signals, the position signal is filtered with a low-pass filter Gtl before the differentiation. For the design of the controller coefficients kx, kv and kc, reference is made to the relevant literature in the field of control engineering. As already mentioned, the current limitation is a component of the controller structure and thus prevents the current saturation of the power amplifier. According to FIG. 6E, the current limitation can be integrated into a controller structure with cascaded current and position control. The target current of the position controller output is limited with the limiter GH. The limited setpoint current is added to the quiescent current iO for the premagnetization. This is followed by the current control, which ensures that the current through the electromagnet matches the current specification.

6F shows a controller unit Gr2 for magnetic bearings in a two-sided arrangement in a schematic representation. The controller inputs consist of the position signal 22 and the currents through the electromagnets 21 and 23. The controller outputs 24 and 25 speak the target voltages of the two power amplifiers. The position control and the limitation of the target current agree with the description of the controller unit Grl for magnetic bearings on one side (Fig. 6E). The limited setpoint current is added to the quiescent current OK in the first current regulator and subtracted from the quiescent current OK in the second current regulator. This is followed by the current control, which ensures that the currents through the two opposite electromagnets match the current specifications.

The distance measuring method is non-linear. Accordingly, it may be necessary to correct the non-linearity of the position signal. In addition, the non-linearity and the hysteresis of the magnetization characteristic curve produce a position error that can be taken into account to a certain extent by including the currents of the electromagnets (MD Noh and EH Maslen, Self-Sensing Magnetic Bearings - Effects of Saturation, Fifth Int. Symp. On Magnetic Bearings, Kanazawa, Japan 1996, pp. 113-118). The linearization characteristic is particularly important for magnetic bearings in a one-sided arrangement. In the technically usual two-sided arrangement, the linearity of the position signal is usually so good due to the formation of differences that a linearization characteristic curve can be dispensed with.

The compensation circuit gel for one-sided magnetic bearings (Fig. 5A) consists of a non-linearized position signal xn (input 9), which is linearized depending on the current through the electromagnet, and appears at the output 10 as a linearized position signal x.

The compensation circuit Gc2 for two-sided magnetic bearings (Fig. 5B) also consists of a non-linearized position signal xn (input 18), which is linearized depending on the currents through the two opposite electromagnets (inputs 17 and 19), and at output 20 as linearized position signal x appears.

7A to 7E illustrate that the modulation arrangement shown in FIG. 6B is only one of many possibilities. FIGS. 7A to 7E are as far as compared to FIG. 6B further simplified that a representation of the current measurement with the current measuring resistor 66 and the capacitance 67 for potential isolation has been dispensed with. For this purpose, both the measurement tap and the reference potential were drawn in to clearly define the measurement signal 5. The connection 3 of the electro-magnet is connected to the output of the power amplifier.

7A describes a modulation arrangement in which the center tap, formed from the electromagnet 50 and the measuring inductance 51, serves as a measuring tap 5 and as a feed point for the modulation signal 60 'of the modulation source 60 via the coupling impedance 63. The advantage of this arrangement is that no additional connections are used except for those that would already be available for the power controller operation. Accordingly, this arrangement is suitable for applications where a low probability of failure is required. For example, a magnetic bearing used in a gas turbine would be exposed to high temperatures and vibrations, which is why efforts would be made to dispense with as many magnetic bearing elements as possible in the direct vicinity of the gas turbine. The sensorless magnetic bearing with the modulation arrangement according to FIG. 7A is very well suited for such fields of application, since the number of electrical leads to the electromagnet is minimal, and since no external position sensors are required. The disadvantages of this arrangement are obvious: the measuring inductance has the same size as the electromagnet and therefore causes the power amplifier to be oversized. It is therefore a cost factor.

7B shows a further arrangement which is particularly advantageous from an economic point of view. The modulation source 60, the modulation signal 60 'and the coupling impedance 63 correspond to FIG. 7A. The use of two magnetic circuits enables center tapping within the electromagnet 54, as a result of which an additional measuring inductance is no longer required. Apart from the additional connection to the electromagnet, this does not result in any further costs and the amplifier does not need to be oversized. External measuring inductors reduce the path sensitivity by at least a factor of 2 in the equal to the path sensitivity of the present arrangement, which thus generates a position signal with the best signal-to-noise ratio.

FIG. 7C shows an arrangement comparable to FIG. 7A, but here the measuring inductance is replaced by a transformer 55. The low-resistance winding 59 of the transformer is connected to the winding 53 of the electromagnet 50 and has the same functions as the measuring inductance. Accordingly, the transformer has the same size as the measuring inductance. The measurement signal 5 is tapped at the high-resistance winding 59 'of the transformer. The modulation signal 60 'of the modulation source 60 is fed in at the center tap, formed from the winding of the electromagnet and the low-ohmic winding of the transformer, via the coupling impedance 63. The transformer enables impedance conversion and potential isolation of the measuring tap. However, both properties do not offer any noteworthy advantages.

FIG. 7D shows an arrangement comparable to FIG. 7A, with the measuring inductance being replaced by a transformer 55 here too. The low-resistance winding 59 of the transformer is connected to the winding 53 of the electromagnet 50 and forms the tap for the measurement signal 5. The modulation signal 60 'of the modulation source 60 is fed in via the coupling impedance 63 to the high-resistance winding 59' 'of the transformer. Depending on the winding ratio, the transformer effects an impedance conversion from the low-resistance to the high-resistance winding. In this way, the signal power can be increased while the voltage of the modulation signal source is kept constant, in order to generate an increased path sensitivity.

FIG. 7E shows a further transformer arrangement, which combines that of FIGS. 7C and 7D, in that the transformer 56 is equipped with two high-resistance windings 59 ′ and 59 ″, and also a low-resistance winding 59, which is connected to the electromagnet 50. The measurement signal 5 is tapped at one high-omnid winding 59 'and the modulation signal 60' from the modulation source 60 is fed in at the other 59 '' via the coupling impedance 63. The suitability of the present modulation method was checked by means of impedance measurements of various types of electromagnets and by means of an exemplary embodiment. The aim of the impedance measurements was to quantify the signal-to-noise ratio and to determine the optimal modulation frequency. The signal-to-noise ratio of a laminated magnetic bearing in a one-sided arrangement is approximately 10-20. In the case of one-sided axial bearings, the signal-to-noise ratio is approximately 1-2, given the same modulation frequency, due to the larger eddy currents in the unblanked core parts. With magnetic bearings in a two-sided arrangement (eg radial bearings), a large part of the interference sensitivity is eliminated due to the common-mode rejection, which is why the signal-to-noise ratio improves by a factor of about 5-10. Ultimately, this leads to a signal-to-noise ratio for radial bearings of 50-200 and for axial bearings of 5-20. The values of the signal-to-noise ratios refer to a modulation frequency of 100 kHz. For thrust bearings, the signal-to-noise ratio can be increased to around 15-60 if the modulation frequency is reduced to 10 kHz. A signal-to-noise ratio of 5-10 is sufficient for the implementation of sensorless magnetic bearings with modulation methods. In addition to the one-sided axial bearings, all types of magnetic bearings are suitable for the modulation process up to 100 kHz.

An exemplary embodiment with one degree of freedom is described below. The experimental setup contains a magnetic bearing in a two-sided arrangement, consisting of two electromagnets with an E-shape (see FIG. 2) and an evaluation circuit largely based on FIG. 5B. The most important characteristics of the evaluation circuit are listed below:

Modulation signal: 100 kHz, 20 V _pp

Coupling impedance: 5 kΩ

Power amplifier: +/- 50 V, 2 A

Phase position of the demodulator: 90 degrees

Position signal bandwidth: 1 kHz

Gb: 1st order bandpass with corner frequencies from 30 kHz and 300 kHz Gt3: 2nd order low pass with a corner frequency of 2 kHz Gt2: 3rd order low pass with a corner frequency of 2 kHz 8A shows the linearity curve of the position signal x measured using the exemplary embodiment. The deviation from x to a straight line is so small that the use of a linearization characteristic Gc2 has been dispensed with. The measurement took place up to the mechanical stop of the magnetic bearing, which is located in half of the air gap. A calibrated inductive position sensor served as the reference xref of the measurement.

8B shows the difference x-xref from the position signal to the signal from the reference sensor. The RMS error (Root Mean Square), standardized to the maximum possible range of movement of the rotor, is 0.13% and corresponds to an absolute error of lμm. It can be seen that x-xref runs on different tracks depending on the direction of movement of the rotor. This is a consequence of the disturbing influence caused by the hysteresis of the magnetization characteristic.

Commissioning the modulation process is easy. Apart from the adjustment of the zero offset 40 or 41 and the slope 42 of the position signal (FIGS. 6C and 6D), no further adjustments are necessary. Since the sensor signal can be tested independently of the power controller operation, there is no difficulty in commissioning the controller if there is sufficient spectral separation between the controller and sensors.

It can be seen from the measurement results of the exemplary embodiment that a robust sensorless magnetic bearing system can be implemented with the present modulation method. The structure of the modulation process is slightly more complex than that of the sensorless magnetic bearing with a linear controller. However, this additional effort is compensated for with much better system robustness.

Claims

claims

1. A method for regulating the distance between the rotor and the electromagnets of a magnetic bearing system, which comprises at least one electromagnet, wherein each electromagnet is assigned sensor electronics, a regulator and a power amplifier, characterized in that at least one modulation signal (60 ') via at least one Coupling impedance (63) is fed in that at least one air gap-dependent impedance of the electromagnets is measured at high frequency and converted to at least one position signal, that the position signal is fed to at least one controller unit, in which controller signals are processed to at least one controller output, that the controller output of an amplifier unit is supplied, in which the controller output is processed, and that the signal of the amplifier output is fed to at least one winding of the electromagnets, whereby the control loop is closed.

2. The method according to claim 1, characterized in that the modulation signal (60 ') in the presence of a measuring inductance

(51) is fed to the center tap (62) of the inductive voltage divider, which is formed by the measuring inductance (51) and by the impedance of the electromagnet.

3. The method according to claim 1, characterized in that the modulation signal (60 ') is fed in the presence of an electromagnet with at least two magnetic circuits at the center tap (62), which is located at the connection of 2 pole shoe windings.

4. The method according to claim 1, characterized in that the modulation signal (60 ') is fed in the presence of a transformer (55) at the center tap (62) of the inductive voltage divider, which by the impedance of the electromagnet and by one of the low-resistance windings of the transformer ( 55) is formed.

5. The method according to claim 1, characterized in that the modulation signal (60 ') in the presence of a transformer (55) is fed via a high-resistance winding of the transformer (55).

6. The method according to any one of claims 1-5, characterized in that the prevention of higher-frequency interference signals in the case of current saturation of the power amplifier is effected in the controller unit by limiting the target current specification is.

7. The method according to any one of claims 1-6, characterized in that the conditioning of the controller output of the controller unit in the amplifier unit to prevent interference signals when the amplifier is saturated by a limiter (G12) in amplitude and by a low pass (Gt2) in the bandwidth can be limited.

8. Device for carrying out a method according to one of claims 1-7, characterized in that a rotor (31), at least one electromagnet (50), and for each electromagnet a modulation source (60), a coupling impedance (63), an amplifier unit (30), a controller unit, and sensor electronics are provided.

9. The device according to claim 8, characterized in that the electromagnet is formed with at least two windings, so that at least a second magnetic circuit and a center tap are formed, the latter for feeding the modulation signal (60 ') and for tapping the measurement signal (5) serves.

10. The device according to claim 9, characterized in that the electromagnet is designed as a pot magnet.

11. The device according to any one of claims 8-10, characterized in that a limiter (G12) and a downstream low-pass filter (Gt2) are provided in the amplifier unit (30).

12. Device according to one of claims 8 - 11, characterized in that a current limitation is provided in the controller unit.