Active adaptive filter with a switch signal-based pre-control path and a switch signal delay function in the main path

Abstract

The invention relates to a method for suppressing at least one electromagnetic interference signal (d) in a power electronics system, having the following steps: a) detecting the input signal (x), b1) delaying the input signal (x) by means of a delay device (2) and forwarding the delayed input signal to a comprehensive primary path (P), said input signal running through the delay device (2) and the primary path (P) over a prolonged primary path propagation time, b2) forwarding the input signal (x) to the active filter (7) in a parallel manner, said input signal running through the active filter (7) over a filter propagation time, c1) detecting an interference signal (d) present on the primary path (P), c2) generating a preliminary counter interference signal (y) by means of the main filter (1) and forwarding the preliminary counter interference signal (y) to a secondary path (S), wherein the preliminary counter interference signal (y) runs through the secondary path and is output as a final counter interference signal (y'); d) superimposing the interference signal (d) with the final counter interference signal (y') in order to obtain a remaining interference signal (e), the length of the delay of the input signal in step b1) being selected such that the prolonged primary path propagation time is greater than or equal to the filter propagation time. In this manner, a method is provided with which broadband interference signals in power electronics systems can be suppressed.

Description

[0001] ACTIVE ADAPTIVE FILTER WITH SWITCHING SIGNAL-BASED

[0002] FEEDBACK PATH AND SWITCHING SIGNAL DELAY IN THE MAIN PATH

[0003] The invention relates to a method for suppressing at least one electromagnetic interference signal in a power electronic system comprising a power electronic circuit by means of an active filter, wherein at least one input signal is received at the power electronic system.

[0004] Power electronics is a branch of electrical engineering that deals with the conversion of electrical energy using switching electronic components. Typical applications include frequency converters for electric drives, inverters, switching regulators, and switching power supplies.

[0005] To increase efficiency and energy density, modern power electronic systems increasingly use high-frequency switching power semiconductors with short switching times. This results in high-frequency electromagnetic interference, which can be emitted both in conducted and radiated form and can thus adversely affect the functionality of other electronic systems. Typical examples of potential interference sinks include broadcasting and communications systems or safety-relevant systems for semi-autonomous driving in electric vehicles.

[0006] Passive filter circuits are commonly used to reduce such interference emissions. However, depending on the required interference reduction and the system's performance, these contain large and heavy components, which limits the space and weight savings achieved by modern semiconductors. The concept of active EMI filters has already become established. In this case, interference is measured, processed, and injected as counter-interference with the goal of destructive interference. However, the unavoidable signal propagation times between the interference measurement point and the injection of the counter-interference significantly limit the frequency range within which interference can be successfully suppressed. The devices or methods known from the state of the art do not yet allow the above-described principle of counter-interference to be applied to suppress interference signals for broadband frequency ranges.

[0007] Based on this, the object of the invention is to provide a method with which interference signals with a broad frequency spectrum can be suppressed in power electronic systems.

[0008] This problem is solved by the subject matter of patent claim 1. Preferred developments can be found in the subclaims.

[0009] According to the invention, a method for suppressing at least one electromagnetic interference signal in a power electronic system comprising a power electronic circuit by means of an active filter comprising a main filter, wherein at least one input signal is input to the power electronic system, comprising the following method steps: a) detecting the input signal, b1) delaying the input signal by means of a delay device and forwarding the delayed input signal to a primary path comprising the power electronic circuit, wherein the input signal passes through the delay device and the primary path with a predetermined extended primary path propagation time, b2) forwarding the input signal in parallel to the active filter, wherein the input signal passes through the active filter with a predetermined filter propagation time, c1) detecting an interference signal present on the primary path,c2) generating a preliminary counter-interference signal by means of the main filter and forwarding the preliminary counter-interference signal to a secondary path encompassed by the active filter, wherein the preliminary counter-interference signal passes through the secondary path and is output as the final counter-interference signal; d) superimposing the final counter-interference signal on the interference signal to obtain a residual interference signal, wherein the length of the delay of the input signal in step bl) is selected such that the extended primary path propagation time of the input signal through the delay device and the primary path is greater than or equal to the filter propagation time through the active filter.

[0010] The method is designed so that the input signal, which is sufficiently correlated with the interference, passes through a primary path and, in parallel, through the active filter. The primary path is primarily formed by the power electronic circuit. On the primary path, or in the power electronic circuit, the input signal results in an interference signal to be suppressed. The parallel passage through the active filter is necessary to generate a suitable counter-interference signal, which can then be added to the interference signal. The active filter comprises a main filter and a secondary path to generate a preliminary counter-interference signal. The preliminary counter-interference signal generated in the main filter then passes through the secondary path, where it is further modified and output as the final counter-interference signal.The suppression of the interference signal is carried out in particular according to the principle of destructive interference, in which the interference signal is minimized, i.e. suppressed, or ideally canceled, by adding the counter-interference signal. To achieve optimal suppression, it is necessary that the counter-interference signal and the interference signal are the same at all times. This can only be guaranteed if the signal propagation times of the primary path and the active filter are accordingly coordinated. From this, the causality condition is derived, which states that the propagation time of the primary path is at least as long as the propagation time of the active filter including the secondary path. This can be explained by the fact that if the propagation time in the active filter is longer than in the primary path, the counter-interference cannot be formed in time.

[0011] A key aspect of the invention is therefore that the input signal is delayed before passing through the power electronic circuit of the primary path. The time delay is adjusted such that the interference signal of the desired signal, after passing through the primary path, meets the counter-interference signal in phase, allowing the interference signal to be optimally suppressed.

[0012] According to a preferred embodiment of the invention, the main filter comprises an adaptive filter, in particular a finite impulse response filter. An adaptive filter in digital signal processing is a special filter that can independently change its transmission characteristics during operation.

[0013] In addition to finite impulse response filters (FIR filters), infinite impulse response filters (IIR filters) are available as adaptive filters on digital media. A finite impulse response filter (FIR filter) is a discrete, usually digitally implemented filter. While IIR filters generally require less computational effort while maintaining the same adaptive filter quality, they are not always stable due to their feedback characteristics, which can lead to problematic filter instability during adaptation. Therefore, an adaptive FIR filter is preferably used for noise suppression.

[0014] According to a preferred development of the invention, the method comprises the following further method steps: e) feeding the residual interference signal back to the active filter, f) repeating steps c2) and d), wherein the counter-interference signal is adapted by changing filter weights of the main filter in such a way that the residual interference signal is minimized.

[0015] When we talk about "minimizing" in this case, we mean that the transfer characteristics of the main filter are adjusted in such a way that when the final counter-interference signal is superimposed on the interference signal, the interference signal is virtually canceled out by destructive interference. The filter weights of the main filter are therefore continuously adjusted during operation, with the filter weights being continuously updated and determined for the next time step from the filter weight of the previous time step and an added correction term. Ideally, the signals are adjusted to one another in such a way that the interference signal is completely canceled out by destructive interference and no residual interference signal remains. In this way, the active filter can independently change its transfer characteristics during operation, so that the suppression of the interference signal can be continuously optimized.In addition, the active filter can account for any changes in the transmission characteristics of the primary path during operation and provide an optimal counter-interference signal here as well. According to a preferred development of the invention, the input signal comprises at least one control signal for controlling the power electronics system. If, for example, an interference current or an interference voltage of the power electronics system is used directly as the input signal, this causality condition explained above is hardly fulfillable, since the interference propagates through the primary path at approximately two-thirds the speed of light. If the signal processor of the active filter required, for example, 100 ns to output the counter-interference signal, the primary path would have to include a corresponding delay, which is technically difficult to implement. Instead, the control signals of the power electronics system are preferably used directly.These directly trigger the switching processes of the power electronic circuit and are thus directly related to the interference, thus correlating well with the interference. Since the control signals are logical signals, they can be delayed with little effort and, in the example above, only passed to the circuit 100 ns later. This gives the counter-interference system sufficient time to form and output the counter-interference signal.

[0016] According to a preferred development of the invention, the main filter uses a filtered-×-LMS algorithm to generate the counter-interference signal. An efficient algorithm is required for real-time implementation and simultaneous adaptation of the required FIR filter. Therefore, the least mean squares algorithm (LMS algorithm) is preferably used here. However, due to the influence of the secondary path, the stability of the LMS algorithm cannot usually be guaranteed. If the secondary path causes a frequency-dependent phase shift of more than 90°, the algorithm becomes unstable, causing the filter weights to diverge and assume values ​​that, after a certain period of time, violate the value range of the digital system. As a result, a suitable counter-interference signal cannot be formed.

[0017] Therefore, the Filtered-x-LMS algorithm is still preferred. In this algorithm, the input signal x(n) is filtered through a model of the secondary path before being multiplied by the residual noise signal e(n) and the step size p. Since the model is not an exact replica, but rather an estimate of the secondary path, it is denoted by S(z), which, in the ideal case, models the secondary path S(z) sufficiently accurately. The calculation rule of the Filtered-

[0018] X-LMS algorithm for the filter weights w(n) is composed as follows: w(n+l) = w(n) + pe(n) x'(n) x'(n) = x(n) * s(n)

[0019] Accordingly, the input signal x(n) is convolved with the impulse response s(n) of the estimate of the secondary path to produce the filtered input signal x'(n).

[0020] To generate the counter-interference signal using the Filtered-x-LMS algorithm, the secondary path is preferably modeled in advance. An efficient and precise implementation of this modeling must be achieved in the digital system. The preferred approach is as follows:

[0021] For appropriate modeling, a measurement of the transfer characteristics of the secondary path is first required. For this purpose, the following procedural steps are carried out:

[0022] A) Detecting an interference signal present on the counter-interference system,

[0023] B) Identifying several interference-relevant frequencies in the detected interference signal,

[0024] C) Generation and injection of a narrowband test signal for each interference-relevant frequency.

[0025] D) Measurement of the test signal at the original point of the feedback of the residual noise signal, whereby the test signal has passed through the secondary path.

[0026] E) Comparison of the injected test signal with the measured test signal and determination of the behavior of the secondary path for each frequency at which a test signal was injected in step C).

[0027] This allows the behavior of the secondary path to be determined or identified for individual frequencies. The identified values ​​are stored in the values ​​s o and SJ for each frequency fj. The frequencies fj are chosen so that they correspond to the switching harmonics of the fundamental frequency f Qof the power electronic system, since the resulting disturbance is expected to also have these frequencies. After this identification, the individual frequency points can be combined into a coherent, non-parametric model (S fj'). Finally, this data can be used as a starting point for an optimization procedure to parameterize an infinite impulse response filter or a simple passive filter so that it optimally models the secondary path.

[0028] The key advantage of the infinite impulse response filter described here is that, due to its feedback characteristics, it can simulate poles of the transfer function in the secondary path, thus modeling a typical frequency response, for example, second-order, with minimal computational and resource expenditure. Since the secondary path is expected to exhibit not only a frequency-dependent component but also a delay time, for example, due to the digital-to-analog converters and amplifiers along the path, the secondary path is simulated primarily by a variable dead time, which is not feasible with a simple passive filter.

[0029] According to a preferred development of the invention, the method comprises the following further method steps: b') filtering the input signal using a pre-filter and forwarding the pre-filtered input signal to the main filter, wherein the input signal passes through the pre-filter and the main filter with a total filter propagation time and the length of the delay of the input signal in step bl) is selected such that the extended primary path propagation time of the input signal through the delay device and the primary path is greater than or equal to the total filter propagation time, i.e. the previous filter propagation time extended by the pre-filter propagation time. The pre-filter can be implemented in analog or digital form. The combination of the pre-filter with the main filter serves to reduce the computational effort of the main filter. The time delay is then selected such that the propagation time through the pre-filter is also taken into account.

[0030] Alternatively or additionally, according to a preferred development of the invention, the method further comprises the following method steps: e') Limiting the residual interference signal by means of a residual interference signal filter, which can again be implemented in analogue or digital form.

[0031] When we talk about “limiting” in this case, we do not mean pre-filtering as in section b'), since the criterion here is not the best possible approximation of the primary path, but the successful band limitation of the input and the residual noise signal.

[0032] In addition to filtering the input signal to simplify the active filter, the residual noise signal can also be filtered. This has the advantage of limiting the frequency of the measured residual noise signal. Such a limitation allows the input and residual noise signals, which are fed to the counter-noise system, to be placed within a specific frequency range. This can be advantageous if, for example, only certain frequencies need to be suppressed. In this case, the pre-filter should not replicate the primary path, but rather adjust the effect of the main filter in the frequency domain.

[0033] According to a preferred development, the primary path comprises N switching states and the active filter comprises N subfilters. In addition, each of the N switching states is preferably assigned to one of the N subfilters, and steps bl) to f) are carried out for each pair of switching state and subfilter. In cases in which the system behavior of the power electronic system differs fundamentally between the switching states of the semiconductors, the relationship between the control signal and the interference signal varies over time. Therefore, in particular, several active subfilters are implemented. Switching between the N active subfilters can take place synchronously with the N switching states of the power electronic system, so that each subfilter outputs a suitable counter-interference signal for each switching state of the interference source.

[0034] According to the invention, a device for carrying out a method described above is further provided. The device comprises: a signal input for detecting an input signal, a delay device for temporally delaying the input signal, a primary path comprising the power electronic (interference) circuit, an active filter for generating a counter-interference signal comprising a

[0035] A main filter and a secondary path, wherein the secondary path comprises an injector path and / or a sensor path, and the delay device and the primary path are arranged parallel to the active filter. The device also comprises a superposition point for superimposing the interference signal with the counter-interference signal. At the superposition point, the interference signal coming from the primary path and the counter-interference signal coming from the secondary path are superimposed to obtain a resulting signal with sufficiently small interference.

[0036] The input signal thus reaches the signal input and is forwarded to the delay device. After the time delay, the input signal passes through the primary path. The active filter is arranged parallel to this, so that the generation of the counter-interference signal and its passage through the secondary path occur parallel to the time delay.

[0037] In particular, the power electronic circuit is already present. The device according to the invention is constructed around the power electronic circuit and connected to it.

[0038] According to a preferred development of the invention, the injector path comprises a digital-to-analog converter and / or an amplifier and / or an electronic circuit for coupling the counter-interference. The counter-interference signal passes through an injector path with a specific transmission behavior until it is actually superimposed with the interference signal. In an acoustic application, this comprises, for example, a loudspeaker; in the case of the electromagnetic application described here, the injector path comprises a digital-to-analog converter and / or an amplifier and / or an electronic circuit for coupling the counter-interference.

[0039] Analogously, the residual interference signal, in particular, passes through a sensor path with a predetermined transmission behavior. In acoustics, this includes, for example, a microphone. In the case of the electromagnetic application described here, the sensor path includes, in particular, a circuit for measuring the residual interference and / or an analog-to-digital converter for processing the residual interference signal in a processor.

[0040] According to a preferred embodiment of the invention, the signal processing of the active filter is implemented on a field-programmable gate array, a digital signal processor, or an application-specific integrated circuit. These digital systems serve to implement the algorithm used to operate the active filter, in particular the Filtered-×-LMS algorithm.

[0041] It should be noted that only a small amount of resources are required and that the process steps a') to e') already take into account all influences that arise in particular from a circuit-technical implementation.

[0042] When modeling, it should be noted that implementing a higher-order IIR filter in FPGA systems is a significant challenge. Specifically, it is planned to combine multiple second-order IIR filters to model a higher-order secondary path.

[0043] The invention is explained in more detail below using a preferred embodiment with reference to the drawings.

[0044] The drawings show

[0045] Fig. 1 schematically shows a structure of a device for carrying out a

[0046] Method for suppressing an interference signal according to a preferred embodiment of the invention in a block diagram,

[0047] Fig. 2 shows schematically a structure of a device for carrying out a

[0048] Method for suppressing an interference signal according to a further preferred embodiment of the invention in a block diagram,

[0049] Fig. 3 shows schematically a structure of a device for carrying out a

[0050] Method for suppressing an interference signal according to a further preferred embodiment of the invention in a block diagram,

[0051] Fig. 4 shows schematically a structure of a device for carrying out a

[0052] Method for suppressing an interference signal according to a further preferred embodiment of the invention in a block diagram,

[0053] Fig. 5 schematically shows a sequence of a method for suppressing a

[0054] Interference signal according to another preferred embodiment of the invention in a flow chart.

[0055] Fig. 1 shows a schematic block diagram of the structure of a device for carrying out a method for suppressing an interference signal according to a preferred embodiment of the invention. From the signal input 5, the signal path is split into two parallel paths. On one path, the input signal x passes through the delay device 2 and the primary path 5. On the other path, the input signal x passes through the active filter 7, here consisting of a main filter 1 and the secondary path S. In the primary path P, an interference signal d is created from the delayed input signal. In the adaptive filter 1, a counter-interference signal y is generated. This passes through the secondary path S and leaves it as the counter-interference signal y'. After the primary path P and the secondary path S, the interference signal d is combined with the counter-interference signal y' in the superposition point 6.It can be seen that by using the control signal as input signal x it is possible to delay the control signals and thus reduce the propagation time Ö. P of the primary path P until the causality condition Ö P > Ö A + Ö s is fulfilled. Since the control signals are used as input signals x, the primary path P comprises the entire power electronic circuit with the transistors controlled by the signals.

[0056] The block diagram in Fig. 2 differs from the block diagram in Fig. 1 by the additional pre-filter 3, which pre-filters the input signal x before the filtered input signal x is passed on to the main filter 1.

[0057] While the described method with the setup shown in Fig. 1 can in principle be used for active noise suppression, its performance is limited due to the limited resources of an FPGA. Since, ideally, the adaptive filter exactly replicates the frequency response P(z) of the primary path and the inverse 1 / S(z) of the secondary path, W(z) = P(z) / S(z) should apply. However, the complexity that the main filter W(z) can replicate is limited by the computing capacity of the FPGA. To reduce the computational effort of the main filter 1, it can be combined with a pre-filter (3) with fixed parameters, which is explained in process step b'). This results in the transfer function W ges (z) from the adaptive part W(z) and the pre-filtered part H x(z). The pre-filter 3 is preferably implemented digitally on the same FPGA, so that its characteristics remain reconfigurable and no additional circuits or processors are required. Preferably, the static filter 3 is arranged before the main filter 1, which places it outside the feedback loop of the residual noise signal and thus does not count towards the secondary path S, which begins at the output of the main filter. This eliminates the need to consider the pre-filter 3 when modeling the secondary path S.

[0058] The block diagram in Fig. 3 differs from the block diagrams in Fig. 1 and 2 by the additional second residual interference signal filter 4, which is used to limit the residual interference signal e before it is fed back to the main filter 1. In addition to the filtering in Fig. 2, the residual interference signal e can also be filtered. The purpose of this is to limit the frequency of the measured residual interference signal e. Together with filtering the input signal x, the input and residual interference signals x, e, which are fed to the counter-interference system, can be placed in a specific frequency range. This can be advantageous if, for example, only the suppression of certain frequencies is necessary. In this case, the filters 3, 4 should not simulate the primary path, but rather adjust the effect of the adaptive filter 1 in the frequency domain.This procedure must be distinguished from pre-filtering, since the criterion here is not the best possible approximation of the primary path P, but rather the successful band-limiting of the input signal x. A combination of both methods in the form of a band-limiting filter and an additional pre-filter is combined in the filter H. x (z), is also possible.

[0059] The block diagram in Fig. 4 differs from the block diagram in Fig. 1 in that there are multiple switching states P1, P2 of the primary path P. Accordingly, the active filter also has multiple active sub-filters 1A, 1B in order to be able to react individually to the switching states P1, P2. The embodiments described so far have been based on the general arrangement shown in Figure 1. This describes the interference source via the primary path P with the transfer function P(z), which by definition is assumed to be a linear and time-invariant system. This assumption must therefore be transferred to the power electronics system, with the control signal of the power semiconductors as the input signal x and the interference as the output signal. This is the case, for example, with the common-mode interference in a drive system.In other cases, however, the system behavior of the power electronic system can differ fundamentally between the switching states of the semiconductors, whereby the relationship between the control signal x and the interference signal d varies over time. Therefore, in the event that linear and, in particular, time-invariant behavior is not present, an implementation of several adaptive filters 1A, 1B is provided. Switching between these can then take place synchronously with the switching states Pl, P2 of the power electronic system, so that each sub-filter 1A, 1B outputs a suitable counter-interference signal y for each switching state Pl, P2 of the interference source. By appropriately switching the input signal x, the residual interference signal e, and the output signal, several active (sub-)filters 1A, 1B can be used to simulate the varying behavior of the primary path P depending on the switching state Pl, P2 and to generate a suitable counter-interference signal y.

[0060] Figure 5 shows a method for suppressing an interference signal d in a flowchart. First, in a first step a), an input signal x is detected and then passes through two parallel paths. In steps b1) and c1), the input signal x is delayed in time and passes through the primary path P, which includes the power electronic circuit. In the process, an interference signal d is generated. In parallel, in step b2), a counter-interference signal y is generated from the input signal x using the main filter 1, which then passes through the secondary path S. However, the input signal x is pre-filtered in step b'). In step d), the interference signal d and the counter-interference signal y' are superimposed. The residual interference signal e is limited in step e') and fed back to the active filter 1. Subsequently, in step f), steps c2) to e) are repeated to further minimize the residual interference signal.

[0061] 1 main filter

[0062] 2 Delay device

[0063] 3 pre-filters

[0064] 4 residual noise filters

[0065] 5 Signal input

[0066] 6 Overlay point

[0067] 7 Active filter x Input signal x Pre-filtered input signal x* Band-limited input signal d Interference signal y Preliminary counter-interference signal before the secondary path y' Final counter-interference signal e Residual interference signal e* Band-limited residual interference signal

[0068] P Primary path

[0069] S Secondary path

[0070] W filter path w filter weights

[0071] 8p runtime through the primary path

[0072] 8 A Running time through the main filter with pre-filter if necessary

[0073] 8 S Running time through the secondary path n Time step z Variable of the z-transformation

[0074] N Number of active filter units in time-variant systems

Claims

Patent claims 1. A method for suppressing at least one electromagnetic interference signal (d) in a power electronic system comprising a power electronic circuit by means of an active filter (7) comprising a main filter (1), wherein at least one input signal (x) is input to the power electronic system, comprising the following method steps: a) detecting the input signal (x), b1) delaying the input signal (x) by means of a delay device (2) and forwarding the delayed input signal to a primary path (P) comprising the power electronic circuit, wherein the input signal passes through the delay device (2) and the primary path (P) with a predetermined extended primary path propagation time, b2) forwarding the input signal (x) in parallel to the active filter (7), wherein the input signal passes through the active filter (7) with a predetermined filter propagation time,cl) detecting an interference signal (d) present on the primary path (P), c2) generating a preliminary counter-interference signal (y) by means of the main filter (1) and forwarding the preliminary counter-interference signal (y) to a secondary path (S) encompassed by the active filter (7), wherein the preliminary counter-interference signal (y) passes through the secondary path and is output as the final counter-interference signal (y'); d) superimposing the final counter-interference signal (y') on the interference signal (d) to obtain a residual interference signal (e), wherein the length of the delay of the input signal in step bl) is selected such that the extended primary path propagation time of the input signal (x) through the delay device (2) and the primary path (P) is greater than or equal to the filter propagation time through the active filter (7).

2. The method according to claim 1, wherein the main filter (1) comprises an adaptive filter, in particular a finite impulse response filter.

3. Method according to claim 1 or 2, with the following further method steps: e) feeding the residual interference signal (e) back to the active filter (7), f) Repeating steps c2) and d), wherein the final counter-interference signal (y') is adapted by changing filter weights of the main filter (1) such that the residual interference signal (e) is minimized.

4. Method according to one of the preceding claims, wherein the input signal (x) comprises at least one control signal for controlling the power electronic system.

5. Method according to one of the preceding claims, wherein the main filter (1) uses a filtered-x-LMS algorithm to generate the counter-interference signal (y).

6. Method according to one of the preceding claims, with the following further Method step: b') Pre-filtering the input signal (x) by means of a pre-filter (3) and forwarding the filtered input signal (x*) to the main filter (1), wherein the input signal (x) passes through the pre-filter (3) and the main filter (1) with a total filter propagation time and the length of the delay of the input signal (x) in step bl) is selected such that the extended primary path propagation time of the input signal (x) through the delay device (2) and the primary path (P) is greater than or equal to the total filter propagation time.

7. Method according to one of the preceding claims, with the following further Method step: e') Limiting the residual interference signal (e) by means of a static analog or digital residual interference signal filter (4) and feeding the filtered residual interference signal (e*) back to the main filter (1).

8. Method according to one of the preceding claims, wherein the primary path (P) comprises N switching states and the active filter (1) comprises N active sub-filters, each of the N switching states is assigned to one of the N active sub-filters (1A, 1B), and steps bl) to f) are carried out for each pair of switching state and active sub-filter (1A, 1B).

9. Device for carrying out a method according to one of the preceding claims, comprising a signal input (5) for detecting an input signal (x), a delay device (2) for delaying the input signal (x), a primary path (P) comprising the power electronic circuit, an active filter (7) for generating a counter-interference signal (y') comprising a main filter (1) and a secondary path (S), and a superposition point (6) for superimposing an interference signal (d) with the counter-interference signal (y'), wherein the secondary path comprises an injector path and / or a sensor path, and the delay device (2) and the primary path (P) are arranged parallel to the active filter (7).

10. Device according to claim 9, wherein the injector path comprises a digital-to-analog converter and / or an amplifier and / or a device for coupling in the counter-interference (y').

11. Device according to claim 9 or 10, wherein the sensor path comprises an analog-to-digital converter and / or a device for measuring the residual interference signal (e).

12. Device according to one of claims 9 to 11, wherein the main filter (1) is implemented on a field programmable gate array, a digital signal processor or an application-specific integrated circuit.

Images