Patent Application
20110144953
This application claims benefit under 35 U.S.C. §119(a) of German Patent Application No. 10 2009 024 826.9-32, filed Jun. 13, 2009, the entire contents of which are incorporated herein by reference.
1. Field of the Invention
The invention relates generally to a system for compensating electromagnetic interfering fields, and in particular to a system for magnetic field compensation having two sensors and a digital processor.
2. Description of Related Art
For compensating electromagnetic interfering fields, in particular magnetic interfering fields, feedback control systems are used in the very most cases, whereby one, or more sensors measure the amplitude of the interfering field for all three Cartesian space axes. The measuring signals of the sensors are fed to a control loop, which calculates control, or actuator signals from the measuring signals of the sensors, for devices generating magnetic fields.
The magnetic field to be compensated may be the terrestrial magnetic field, or may be generated by other current-carrying devices being in the surrounding.
Magnetic field compensation systems are for example used in connection with imaging systems using magnetic fields, for example in the case of scanning electron microscopes (SEM).
In case of the mentioned devices for generating magnetic fields, it may be a matter of a current-carrying conductor, in the easiest case. Generally, one assumes interfering fields having far field characteristics, i.e. such fields, whose field amplitude does not essentially change within the range of 5 m. This assumption for example is true for interferences by rail vehicles. If the interfering fields are homogeneous in the range of interest, the compensation fields should be homogeneous, also.
Pairs of so-called Helmholtz coils are preferably used for generating homogeneous compensation fields. At this, it is about two coils each being connected in the same direction, and having a distance to each other being equal to the half length of the edge (=coil diameter) (so-called Helmholtz condition).
Furthermore, pairs of Helmholtz coils are used, whose distance to each other is equal to one length of the edge. If one pair of Helmholtz coils is used for each of the three space axes, the pairs of coils form a cube-shaped cage around the location, at which one, or more interfering fields shall be compensated. In case of such a coil arrangement, there indeed are field inhomogeneities in the interior of the cage, but these are acceptable in the most cases of application.
A device for compensating magnetic fields is disclosed in U.S. Publication No 2005/019555A1 and has three coil pairs in a cage. The magnetic field to be compensated is measured and compensated, where an analog controller is used.
Systems are also available, with which only one coil per space axis is used for generating the compensation field, however the compensation region, i.e. the region in which a good compensation is achieved, is considerably smaller than in the case of Helmholtz coils.
Generally, one single magnetic field sensor is used for measuring the magnetic field at the place of interest. As an exception, there is a second sensor which is, however, used for diagnosis purposes. A single magnetic field sensor does not allow to detect, whether the magnetic field to be compensated is homogeneous, or inhomogeneous at the location of the object to be protected.
It is a further problem when compensating electromagnetic interfering fields that it cannot be measured directly at the location at which the interfering field is to be compensated, since the object to be protected against interfering fields generally is at this location.
A further problem arises, if two magnetic field compensation systems are arranged directly adjacent to one another. Then, undesired feedback effects may occur between the two systems.
There are problems with the control systems in that these control systems can generally be optimized to single application. An adjustment to control tasks that are quite different, such as upon changes in the control configuration, is as a rule not possible or only in a restricted manner possible and/or is to be implemented with great difficulties. Furthermore non-linear control systems which may have a better interference field compensation than linear control systems, generally can only be implemented with high costs. When control circumstances change, the whole control circuit or the control loop would have to be newly calculated, designed and/or changed. In most cases, the direct user is not a position to do so.
Therefore, it is an object of the invention to provide a system for compensating electromagnetic interfering fields with which system homogeneous as well as inhomogeneous magnetic fields may be compensated.
It is a further object of the invention to perform a simulation of measuring electromagnetic interfering fields at the location of the object to be protected.
It is a still further object of the invention to equalize potentially arising feedback effects in the case of using two magnetic field compensation systems in immediate vicinity.
In detail, a system for compensating electromagnetic interfering fields is provided, which has two real triaxial magnetic field sensors, three pairs of compensation coils, and one control unit in order to protect an object against influences of an interfering field. It is preferred to design the control unit as a control processor such as a Digital Signal Processor DSP or a field programmable gate array FPGA.
The six in total output signals of the two real sensors may be combined to three output signals of a virtual sensor, by means of a freely definable kind of averaging. By choosing the averaging algorithm properly, it can be achieved that the output signals of the virtual sensor represent the amplitude of the interfering field at the location of the object to be protected.
The averaging takes place by means of the control system, which receives the six output signals of the two real magnetic field sensors via six inputs.
For every sensor, the output signals of the two magnetic field sensors may be represented by a three-dimensional vector. These two vectors may be combined to six-dimensional vector, i.e. a 6×1 matrix. The averaging over the output signals of the two real sensors, i.e. calculating the output signals of the virtual sensor, may be described by a matrix multiplication:
V=M·S
The now available output signals (=virtual input signals of the control system) of the virtual sensor are used as an input for independent control loops operating in parallel. These control loops may be broadband, selective concerning a frequency range, or selective concerning a frequency, also. The control loops have control algorithms transforming the virtual input signals V into changed signals {circumflex over (V)}. At this, {circumflex over (V)} is a 6×1 matrix representing the in total six changed input signals of the control system. The control algorithm is described by an operator Ω. There are no limitations concerning the control algorithm being used. Accordingly, the operator Ω may not be a matrix so that nonlinear algorithms may also be used. Therefore, the transition to the modified signals {circumflex over (V)} is described by
{circumflex over (V)}=Ω(V)
The matrix {circumflex over (V)} is multiplied by a 6×6 matrix L, in order to obtain control signals for the six coils, i.e.
O=L·{circumflex over (V)}
with:
L: 6×6 matrix for calculating the control signals O from the modified signals O=L·{circumflex over (V)}.
Therefore, the algorithm used by the control system may overall be described as follows:
O=L·Ω(M·S)
The more inhomogeneous the compensation field is in case of homogeneous interference, and the more homogeneous the compensation field is in case of inhomogeneous interference, the smaller is the region around the feedback sensor having a good compensation effect.
If the interference field is inhomogeneous, it is not purposeful to generate a homogeneous compensation field. In this case, it is also purposeful to use a single actuator coil instead of a pair of Helmholtz coils.
Only a single compensation system is used in this case, i.e. only three virtual signals are used for processing virtual sensor positions, and for generating gradient fields so that M may be a 3×6 matrix, and L may be a 6×3 matrix. Alternatively, the “not used” elements of the 6×6 matrices may also be equal to zero.
In case of a Helmholtz coil arrangement, only one coil of the pair is actively actuated, and that depending on the gradient of the interfering field below the compensation region, or above the compensation region. Therefore, a rearrangement for changing the position of the single coil is not necessary besides a new parametrisation of the control loops, in case of a change of the structure of the interfering field.
If two compensation systems are operated directly beside each other, this results in mutual interferences. The feedback between the two systems my be described by means of a 6×6 feedback, or crosscoupling matrix C. C represents the feedback of a control signal Oi with a virtual signal Vi.
For avoiding interferences, the feedback system will not deliver optimal results. As a rule, an overcompensation, or an under compensation is only feasible for digital control systems, and also in this case for systems not operating in broadband. The position of the sensor would have to be fitted for all other systems. Such a change of position may it make it necessary that the sensors for the three space axes have to be positioned at different positions in space. But because one single system for all kinds of applications is not aimed for, overcompensation or undercompensation respectively is not an appropriate method.
When doing so, the matrix S of the output signals of the real sensors is enlarged to a 6×1 matrix Ŝ. Therefore, it is true over all:
O=L·Ω(M·(S−C·O))
In the following, the invention is described in more detail referring to the attached figures by means of exemplary embodiments, wherein same reference signs refer to same components.
The amplitude of the interfering field 1 is measured by two real magnetic field sensors 3, and 4. The first real sensor 3 provides an output signal {right arrow over (S)}1=[x1(t), y1(t), z1(t)], and the second real sensor 4 provides an output signal {right arrow over (S)}2=[x2(t), y2(t), z2(t)]. These two output signals are fed in a digitised form to the control unit 7 shown in
The control unit 7 has six inputs for the six signals in total, corresponding to 2×3 space axes. Furthermore, the control unit 7 has six outputs for outputting control signals for six coils 6.
The two vectors {right arrow over (S)}1, and {right arrow over (S)}2 are combined to a 6-vector S=(S1, S2, S3, S4, S5, S6). S is processed by the control unit 7 according to the algorithm schematically shown in
V=M·S
The virtual signals V correspond to the amplitude of the interfering field at the location of the object 2 to be protected. Therefore M describes the geometry of the whole arrangement, and how the signals of the two real sensors 3, and 4 are combined.
The virtual signals V generated in such a manner are fed to independent control loops operating in parallel, and processed further. These control loops as part of the control unit 7 may be broadband, selective concerning a frequency range, or selective concerning a frequency. The control loops change the virtual signals V to modified signals {circumflex over (V)}. The transition from V to {circumflex over (V)} is described by an operator Ω. Therefore, it applies:
{circumflex over (V)}=Ω(V)
Since there are no limitations concerning the used control algorithms, the modification of the signals V is generally described by the operator Ω, which is not necessarily a matrix so that nonlinear algorithms may be used, also.
For gaining control signals for the coils 6, the modified signals {circumflex over (V)} are converted into real control signals O. O again is a 6×1 matrix, therefore containing six single signals, which are used for controlling the six coils 6. The transition from the modified signals {circumflex over (V)} to the control signals O is therefore described by
O=L·{circumflex over (V)}
or over all:
O=L·Ω(M·S)
Here, L is a 6×6 matrix. The precise values of its elements depend on the nature of the interfering field to be compensated, and on the geometry of the coils 6 generating the compensation field. If, for example, a gradient field acting in x direction shall be compensated, the two coils acting in direction get differently strong signals so that the two coils generate differently high magnetic fields so that the compensation field also is a gradient field, whose direction of field intensity is inverse to the direction of the interfering field.
The algorithm described up to now is used as long as one single compensation system is only used. For this case, three virtual signals are needed, only. When doing so, virtual sensor positions are calculated, and gradient fields are generated. For this purpose, it is sufficient, if M is a 3×6 matrix, and L is a 6×3 matrix. Alternatively, the “not used” elements of the 6×6 matrices may also be equal to zero.
Also, two compensation system being placed directly beside each other may be operated by means of the control unit 7. This can make sense, if two objects to be protected are directly placed beside each other, and shall, or may not be protected by a large compensation system. This implicates that, due to the two compensation systems being used, the regions to be protected have a significantly smaller volume. Therefore, no gradient fields are needed for compensation. With such an installation, generating gradient fields for compensation, however, is also not possible, because the six output signals of the control unit 7 are given to six pairs of coils, which are only able to generate a homogeneous magnetic field in each of the directions in space. The pairs of coils may be connected in series, in parallel, or depending on the impedance. These pairs of coils are each placed around the object 2 to be protected, and each of the corresponding systems is each arranged inside the cage formed by the three pairs of coils each. This configuration is shown in
Two compensation systems may also be arranged directly beside each other. This case is shown in
If two compensation systems are used in direct vicinity, feedback effects may arise between the two systems. This is accounted for by providing a 6×6 back coupling matrix C, which computationally eliminates the parts of the signals, which are crosstalks from an output signal Oi to a virtual signal Vi. Therefore, C describes the kind of feedback between the two compensation systems installed directly beside each other.
According to the invention, the 6×1 matrix of the real sensor signals is expanded by the feedback part. If the 6×1 matrix of these expanded signals is denominated by Ŝ, it applies
Ŝ=S−C·O
The 6×1 matrix with the virtual sensor signals is calculated from the signals Ŝ expanded by the feedback part, obtained in this manner. Therefore, it applies:
V=M·Ŝ
finally yielding control signals according to the following relation:
O=L·Ω(M·(S−C·O))
In the following, a standard installation of the systems shall be assumed, i.e. only one system is installed. Therefore, no feedback effects occur, which means that the matrix C is equal to the zero matrix. Furthermore, it shall be assumed that the virtual sensor signal in x direction shall be composed of the arithmetic mean of the two real sensor signals in x direction, because the gradient of the interfering field proceeds in x direction. The virtual sensor signal in y direction shall be equal to the signal in y direction of the second real sensor, because, for example, the signal in y direction of the first real sensor contains unwanted components caused by a local interferer. Due to averaging/noise suppression reasons, the virtual sensor signal in z direction shall be equal to the arithmetic mean of the two real sensor signals in z direction. Under these assumptions, the matrix M has the following form:
If the compensation coils are formed as pairs, and if a homogeneous compensation field shall be emitted in y, and in z direction, which field has a gradient in x direction, the matrix L has the following form:
A double installation is considered in the following example, i.e., two systems for compensating electromagnetic fields are operated directly beside each other.
Since the output signals for both compensation cages are known inside the control unit 7 in this case, now also feedback parts can be taken into consideration in the control structure. This takes place, as already is described, by using a feedback, or crosscoupling matrix C. This matrix C or its elements, respectively, may experimentally be determined in a comparably easy manner, by applying a signal to an output of the first compensation system, and measuring at the second system, which components are absorbed by the sensors of the second system, and which fraction of the amplitude, in comparison with the sensor of the first system. Then, these signals parts are the elements of the feedback matrix C. When doing so, this measuring method has to be done for all coils.
If, for example, the output O5 still radiates onto the sensor input Ss with 40%, the matrix element has to be C25=0.4.
| Number | Date | Country | Kind |
|---|---|---|---|
| 102009024826.9-32 | Jun 2009 | DE | national |
