DE202025107004U1 - Three-phase harmonic power line filter - Google Patents

Abstract

Three-phase harmonic mains filter (110; 210; 260; 310; 360; 410; 460; 510), comprising: - a first filter branch (120) between a first filter input (L1) and a first filter output (L11), wherein the first filter branch (120) comprises a first series connection of three inductors (IND1, IND2, IND3) connected between the first filter input (L1) and the first filter output (L11), and wound on three different legs (132, 134, 136) of a three-leg filter core (130); - a second filter branch (122) between a second filter input (L2) and a second filter output (L12), wherein the second filter branch (122) comprises a second series connection of three inductors (IND4, IND5, IND6) connected between the second filter input (L2) and the second filter output (L12), and are wound on three different legs (132, 134, 136) of the three-legged filter core (130),- and a third filter branch (124) between a third filter input (L3) and a third filter output (L13), wherein the third filter branch (124) has a third series connection of three inductors (IND7, IND8, IND9) which are connected between the third filter input (L3) and the third filter output (L13), and which are wound on three different legs of the three-legged filter core (130), wherein the input inductors (IND1, IND4, IND7) or the output inductors (IND3, IND6, IND9) of the three filter branches (120, 122, 124) are on different legs (132, 134, 136) of the three-legged filter core, wherein the first filter branch (120) comprises a first transverse inductance (IND10), the second filter branch (122) comprises a second transverse inductance (IND11), and the third filter branch (124) comprises a third transverse inductance (IND12), wherein a node to which two inductances (IND1, IND2; IND2, IND3) of the first series connection are connected is coupled via the first transverse inductance (IND10) to a first terminal (152) of a capacitive energy storage device (150), wherein a node to which two inductances (IND4, IND5; IND5, IND6) of the second series connection are connected is coupled via the second transverse inductance (IND11) to a second terminal (154) of the capacitive energy storage device (150), wherein a node to which two inductances (IND7, IND8; IND8, IND9) of the third series circuit are connected, via the third transverse inductance (IND12) is coupled to a third terminal (156) of the capacitive energy storage device (150), and wherein the three transverse inductances (IND10, IND11, IND12) are each arranged on one of the three legs of the three-legged filter core, characterized in that the respective inductances IND_i, i e [1;12] have a respective number of turns Ni, i ∈ [1;12], wherein the following ratios apply: - N1/N8> 8, - N1/N6> 8, - N4/N2> 8, - N4/N9> 8, - N7/N5> 8, - N7/N3> 8, - N1/N10> 2.5, - N4/N11> 2.5, - N7/N12> 2.5.

Description

The present invention relates in general to a three-phase harmonic network filter, and in particular to a passive three-phase harmonic network filter as a front end for non-linear loads.

The rapidly increasing proportion of power electronics in the energy supply network, especially in the field of drive technology, leads to a growing distortion of the supply voltage due to the high harmonic content of the current. To avoid harmonic currents in the supply networks, standards have been issued in Europe in recent years that prescribe specific guidelines for manufacturers of electrical and electronic equipment.

To comply with the established standards, guidelines, and recommendations, various active and passive solutions exist from different manufacturers worldwide. Depending on the performance and application of the devices, or their use by the end customer, these solutions can have advantages or disadvantages. Essentially, the currently available active or passive devices and filters for reducing power harmonics are not particularly attractive in terms of size or cost and are therefore only used to a limited extent.

For electronic devices with an internal B2 or B6 rectifier circuit, the following conventional methods are used to reduce current harmonics: AC and DC chokes, higher-pulse rectifier circuits via B12, B18, or B24, resonant circuits, low-pass filters for 50 Hz or 60 Hz, special harmonic filters, active sinusoidal current pickup devices (so-called active front ends), and active harmonic filters. The active harmonic filters are operated in parallel with the mains supply.

The following section examines specific harmonic filters in more detail. Currently available harmonic filters have numerous disadvantages, sometimes requiring a very large physical size compared to the device being used, or generating immense costs that often exceed the actual cost of the connected device.

Since the circuit arrangement of special harmonic filters generally consists of inductive and capacitive components, three main problems arise during filter operation. High inductance values in the series branch of a filter lead to load-dependent voltage drops and can result in a reduced DC link voltage (DC voltage after a rectifier). Although this effect is partially offset by adding capacitors, as capacitors raise the voltage again, a load-dependent voltage change remains.

Furthermore, capacitors connected in a cross branch generate a capacitive reactive current that flows into the harmonic filter even under no-load conditions. The proportion of capacitive reactive current should generally be kept very low, as this so-called overcompensation is undesirable for energy suppliers. Some manufacturers of specialized harmonic filters therefore offer the option of partially or completely switching off the capacitors under partial load conditions using a contactor. This, in turn, increases costs and complexity, as such a contactor should have contacts suitable for capacitive current and the filter must be integrated into a control sequence.

Another disadvantage of conventional, specialized harmonic filters is the resonance behavior of LC circuits. Fundamentally, all circuits consisting of inductive and capacitive components have at least one resonant frequency. While the filters are designed to minimize the frequency range of these resonant frequencies, this can be difficult to predict under dynamic load changes, especially when combined with load changes in the power grid or the switching on and off of compensation systems connected to the grid.

It is therefore evident that conventional special harmonic filters have serious technical and economic disadvantages that make their use more difficult or expensive.

The WO 2006/048161 A1 The diagram already shows a three-phase mains filter designed as a special harmonic filter. Several series inductors are provided for each phase; a cross branch leads from an intermediate node via another inductor to star-connected capacitors. The aim of the WO 2006/048161 A1 The aim is to guide the fundamental frequency current to the load almost unimpeded and to preferentially divert high-frequency current components into the cross branches.

In practice, however, the information presented there proves insufficient, as a conflict exists between a small capacitor current at mains frequency, a low voltage drop and loss in the series path, and simultaneously stable harmonic attenuation. Even moderate component tolerances or fluctuations in the mains frequency can lead to this. Excessive impedance can cause the 50 Hz current across the neutral point to rise, the voltage drop to become unacceptable, or the damping in critical harmonic bands to decrease, potentially leading to undesirable resonance effects. Consequently, capacitors must be oversized or made switchable, increasing size, cost, and complexity.

The invention is therefore based on the objective of providing a three-phase harmonic mains filter that keeps the capacitor current low at mains frequency, keeps the voltage drop in the series path small and at the same time enables stable harmonic attenuation without switchable capacitor banks and enables a low THDI value.

The inventors have surprisingly discovered that the problems according to the invention can be solved by having the first coil arranged on each of the three legs of the three-legged core have a significantly higher inductance than the two subsequent inductances of each leg, since this optimizes the magnetic flux in the core. The effect is further enhanced by the fact that this first coil has a significantly higher inductance than the coil of the respective secondary leg. This shifts the frequency-dependent impedance of the filter into an extremely robust range.

The problem according to the invention is solved in particular by a three-phase harmonic mains filter, comprising a first filter branch between a first filter input and a first filter output, wherein the first filter branch has a first series connection of three inductors which are connected between the first filter input and the first filter output and which are wound on three different legs of a three-legged filter core, a second filter branch between a second filter input and a second filter output, wherein the second filter branch has a second series connection of three inductors which are connected between the second filter input and the second filter output and which are wound on three different legs of the three-legged filter core, and a third filter branch between a third filter input and a third filter output, wherein the third filter branch has a third series connection of three inductors which are connected between the third filter input and the third filter output and which are wound on three different legs of the three-legged filter core (130). The input or output inductors of the three filter branches are wound on different legs of the three-legged filter core. The first filter branch includes a first transverse inductor, the second filter branch includes a second transverse inductor, and the third filter branch includes a third transverse inductor. A node where two inductors of the first series connection are connected is coupled via the first transverse inductor to a first terminal of a capacitive energy storage device; a node where two inductors of the second series connection are connected is coupled via the second transverse inductor to a second terminal of the capacitive energy storage device; and a node where two inductors of the third series connection are connected is coupled via the third transverse inductor to a third terminal of the capacitive energy storage device, with the three transverse inductors each being arranged on different legs of the three-legged filter core.

• - N₁/N₈ > 8,
• - N₁/N₆ > 8,
• - N₄/ N₂ > 8,
• - N₄/N_g > 8,
• - N₇/N₅ > 8,
• - N₇/N₃ > 8,
• - N₁/N₁₀ > 2,5,
• - N₄/N₁₁ > 2,5,
• - N₇/N₁₂ > 2,5.

The three-phase harmonic mains filter is characterized by the fact that the respective inductances IND_i, i ∈ [1;12] have a respective number of turns N _i , i ∈ [1;12], where the following ratios apply:

• - N ₁ /N ₈ > 8,
• - N ₁ /N ₆ > 8,
• - N ₄ / N ₂ > 8,
• - N ₄ /N _g > 8,
• - N ₇ /N ₅ > 8,
• - N ₇ /N ₃ > 8,
• - N ₁ /N ₁₀ > 2.5,
• - N ₄ /N ₁₁ > 2.5,
• - N ₇ /N ₁₂ > 2.5.

This achieves the technical effect that the series path remains low-impedance at the fundamental frequency, while its high-frequency impedance increases sufficiently to reliably disperse harmonics into the shunt branch. Simultaneously, the shunt branch remains sufficiently high-impedance at 50 Hz, keeping the capacitor current low and effectively preventing the series resonance resulting from the shunt branch inductance and the star capacitor from affecting the fundamental frequency. In conjunction with the star connection, the high-frequency component is buffered with a phase shift and absorbed by the other phases, while the fundamental frequency current flows to the load largely undisturbed. Measurements and simulations show that precisely these inductance ratios reliably manage the aforementioned conflict and significantly reduce sensitivity to component tolerances and mains fluctuations. This allows for the construction of compact, low-loss filters without switching. bare capacitor banks are implemented, which also have a very low THDI value.

The core concept of the present invention is to distribute the series inductances of a filter branch between the filter input and filter output across different legs of a multi-leg filter core, while defining the number of windings in these longitudinal windings in fixed, distinctly staggered ratios to each other and to the respective transverse winding. It has been recognized that such a distribution, in conjunction with the defined minimum spacing of the winding numbers, leads to a significant reduction in the effective voltage drop across the longitudinal branch. By assigning the windings of a filter branch to three legs according to the invention, and by the pronounced increase in the input-side series winding compared to the downstream series windings and the transverse winding, individual flux components in the filter core are reduced or partially compensated. This reduces the magnetic energy stored in the core and allows for a reduction in the overall size.

The reduction of field energy is based on the fact that the currents in the filter branches of a three-phase harmonic mains filter have a fixed phase relationship to each other, and the magnetic fluxes they generate superimpose in the multi-limbed core. With the described winding staggering, the frequency-dependent impedance is adjusted so that the fundamental frequency current is passed through to the load in the series path with minimal loss, while high-frequency current components preferentially flow into the respective cross-branch. These high-frequency components are temporarily stored out of phase via the star-connected capacitive energy storage devices and absorbed by the other phases. In this way, the current through the capacitors remains low at mains frequency, critical resonance effects are avoided, and the desired harmonic attenuation is reliably achieved.

As a result, the voltage drop across a three-phase harmonic network filter according to the invention is lower than in conventional arrangements. The load dependency of the output voltage at the filter output or at the output of a downstream rectifier is reduced. The filter size can be reduced, and losses can be lowered. Transverse inductances can be implemented with smaller wire cross-sections because the total energy to be stored is smaller due to the distribution and winding staggering according to the invention. The same applies to the capacitive energy storage devices, which can be dimensioned smaller due to the reduced energy demand. This leads to a reduced capacitive reactive current at no load and under partial load. Switching off the capacitive energy storage devices in these operating conditions is not necessary. This simplifies filter and system control, and the construction of a system with the network filter according to the invention becomes faster and more cost-effective.

For the purposes of this application, the three-phase harmonic network filter can also be referred to simply as a network filter.

In a preferred embodiment, the three-phase harmonic network filter is designed such that useful AC currents with a predetermined fundamental frequency are passed from the three filter inputs to the three filter outputs, while interference currents with frequencies deviating from the fundamental frequency, which arise at the filter output due to the connected load, are effectively attenuated towards the filter inputs. The starting point is the understanding that the interference currents originate from the load and should ideally not be coupled into the supply network connected to the filter inputs. The network filter according to the invention therefore focuses primarily on suppressing these disturbances, in particular harmonic currents and other network feedback, while the useful AC current with fundamental frequency is passed through to the load largely unimpeded. A corresponding filter design allows for even highly distorted current waveforms at the filter output, up to approximately block-shaped waveforms, while simultaneously maintaining a substantially sinusoidal current draw at the filter inputs. The higher-frequency current components required for this are provided in the filter by inductive and capacitive energy storage devices. In this way, loads with virtually any current waveform can be operated without high-frequency interference currents flowing back into the upstream grid.

According to the invention, each of the three phase-related filter branches comprises a first series inductance between the respective filter input and an inner node of the phase filter branch, a second series inductance between this inner node and the respective filter output, and a third inductance in a cross branch, which is connected to the inner node of the phase filter branch. The two series inductances of a phase filter branch are distributed on different legs of a common three-legged filter core. The distribution of the series windings of the three phases is such that each phase is represented on all three legs, thereby creating targeted magnetic coupling between the phases. This arrangement reduces the T- Each phase structure enables the frequency-selective effect of the filter and is connected via a defined magnetic coupling between the phases, eliminating the need for additional windings. The inductance of the cross branch can be wound onto a suitable leg of the filter core, allowing the arrangement to be adapted to specific boundary conditions and design objectives.

Furthermore, a capacitive energy storage device is connected in the cross branch of each phase filter branch via the aforementioned cross inductance, with the three capacitive energy storage devices being connected to each other in a star configuration. The capacitive energy storage provides current during periods when the energy stored in the inductors is low and absorbs current when high-frequency components are discharged. In this way, the filter can provide or absorb an additional current component beyond the supply current received at the filter inputs, thus achieving non-sinusoidal output current waveforms while reliably keeping high-frequency current components away from the grid. The use of the star-connected capacitive energy storage devices is particularly advantageous in conjunction with the inventive distribution of the series windings across the three legs of the common filter core and the resulting magnetic coupling of the phases. The magnetic coupling reduces the energy to be stored in the core and thus decreases the required reactive energy in the cross branches. This allows for smaller wire cross-sections in the transverse inductors and smaller dimensions for the capacitive energy storage devices. The capacitive reactive current at no load and under partial load decreases, eliminating the need to switch off the capacitive energy storage in these operating conditions. This simplifies filter and system control and enables a more compact and cost-effective design for the mains filter.

According to the invention, a three-phase filter core is used. There are three filter branches, each with a series connection of at least two inductors between the filter input and filter output. The input-side and output-side inductors in this branch are each wound around different legs of the three-phase filter core. A three-phase design offers the significant advantage that the three-phase harmonic network filter can thus be used in conjunction with conventional three-phase power grids. Furthermore, a three-phase design has the advantage that the phase relationship between the individual phases, and thus between the currents in the individual filter branches, as determined by the three-phase power grid, enables a particularly advantageous coupling of the filter branches and a reduction or cancellation of flux components in the three-phase filter core. In a three-phase filter, there are three phases that are each shifted in phase by 120 degrees relative to one another. This applies to the voltages, the currents, and the magnetic fields generated by the currents. Superimposing all three phases can lead to a reduction in the magnetic field or magnetic flux. A three-phase design is particularly well-suited for reducing the magnetic flux in a three-phase harmonic network filter, thereby reducing the energy stored in the filter and its physical size. Power loss also decreases with a sufficiently significant reduction in magnetic flux.

According to the invention, each of the three phase-related filter branches comprises three inductors connected in series between the respective filter input and the respective filter output, the windings of which are distributed over all three legs of a common three-legged filter core. This magnetically couples each filter branch to both other filter branches simultaneously, resulting in a highly symmetrical arrangement. In the three-phase configuration, this double coupling is particularly advantageous because the outer conductors are phase-shifted by 120 degrees relative to each other: Coupling only between two branches would adversely affect the phase relationship and reduce the magnetic flux only to a limited extent, whereas the uniform coupling of one branch to both other branches reduces the magnetic flux significantly more without shifting the phase relationship. The strength of the coupling is advantageously selected to be the same with respect to both other branches; this maintains symmetry and avoids phase deviations.

To support the frequency-dependent operation, the input-side series winding of each leg has a significantly higher number of turns than the two downstream series windings of the same leg and is also significantly over-coupled compared to the corresponding shunt winding. This keeps the series path for the fundamental frequency current low-impedance, while high-frequency components preferentially flow into the shunt branch. The shunt inductances of the three phases are each arranged on different legs and connected to capacitive energy storage devices, which are connected in a star configuration. This allows high-frequency current components to be temporarily stored out of phase and absorbed by the other phases, while the current... The voltage drop across the capacitors remains small at mains frequency. Overall, the combination of distributed winding arrangement, uniform coupling of all three phases, and pronounced winding staggering results in a low voltage drop, reduced field energy in the core, robust harmonic damping, and stable operating behavior even with component tolerances and mains fluctuations.

Preferably, the inductances of the three filter branches within a branch and also between branches have the same winding direction. Such a design facilitates the transmission of the desired AC current from the filter inputs to the filter outputs with low voltage drop and low losses, while interference currents, especially harmonics, are preferably carried away via the shunt branch. When an interference current is coupled in from the output, the output-side series inductances are traversed first, followed by the shunt inductance in the opposite direction, which significantly reduces the effective inductance from the perspective of the interference, provided the winding direction is suitable. In contrast, a desired current coupled in from the input traverses the input-side series inductance and the shunt inductance in the same direction, so that the shunt branch appears high-impedance at the fundamental frequency, and the desired current is passed on to the output. Therefore, specifying a uniform winding direction represents a significant degree of freedom in the design.

It is particularly preferred to select the number of turns of the series inductors distributed across the three legs of each filter branch, as well as their distribution, such that the magnetic flux in the legs of the three-leg filter core is significantly reduced compared to an arrangement with windings bundled on a single leg. This is achieved by the inventive combination of distributing the series windings across all three legs and a pronounced staggering of the number of turns, in which the input-side series winding of each leg is significantly larger than the two downstream series windings of the same leg and compared to the associated transverse winding. A reduced flux density allows for smaller core cross-sections, lowers losses, and enables a more compact design overall.

In each filter branch, a capacitive energy storage device is coupled to a node located between the series inductors, preferably via a transverse inductor. The three capacitive energy storage devices are connected in a star configuration. The capacitive energy storage provides or absorbs current in suitable phases, thus enabling non-sinusoidal current waveforms at the output, while high-frequency components are kept away from the grid. The aforementioned winding staggering supports the frequency-dependent distribution: The fundamental frequency current is carried in the series branch, while high-frequency components are diverted to the transverse branch; the capacitor current at grid frequency therefore remains small. The magnetic coupling of the three phases via the common core reduces the energy to be stored in the core, thus also reducing the reactive power required in the transverse branches.

A star connection of capacitors is preferred as a capacitive energy storage device. Alternatively, a delta connection can be used. Both circuit configurations are common in power engineering and can be implemented with reasonable effort; the star connection offers a defined reference point, while the delta connection requires a lower capacitance value with appropriately adjusted voltage rating.

The mains filter is designed such that the current flowing through each terminal of the capacitive energy storage device at the operating frequency is less than a fraction of the rated current per phase. This is achieved by the distribution of the longitudinal inductances across the three legs of the common filter core according to the invention, in conjunction with the described winding staggering and the star-shaped connection of the capacitors. The lower currents in the cross-branches allow the use of smaller wire cross-sections and smaller capacitors. This reduces the capacitive reactive current at no load and under partial load, eliminating the need to disconnect the capacitors in these operating conditions. System control is simplified, and the filter's size and cost are reduced.

According to a further feature of the invention, it is provided that the number of turns of the three input inductors is identical, and/or that the number of turns of the three output inductors is identical, and/or that the number of turns of the three middle inductors is identical, and/or that the number of turns of the three transverse inductors is identical.

The equality of the number of turns within each of the four winding groups results in a high degree of symmetry in the magnetic coupling across all three legs. This avoids phase deviations, improves flux compensation, reduces the voltage drop in the longitudinal branch, and makes harmonic damping reproducible. At the same time, the symmetry simplifies the design. Positioning the star point currents increases tolerance robustness against network and component variations and reduces manufacturing and storage costs.

The "input inductances" are the first inductances of each leg. "Middle inductances" are the second series inductances of the same leg. "Output inductances" are the third series inductances of a leg. "Transverse inductances" are the inductances of the cross branches that lead from the inner node of a filter branch to the capacitive energy storage device, which is connected in a star configuration to the cross branches of the other two phases.

"Identical" means that the relevant number of turns within a winding group is the same in pairs; this includes a tolerance range of ± 5%, preferably ± 2%, and more preferably ± 1%, in each case based on the number of turns. The assessment of identity is carried out under identical measurement and operating conditions (in particular, the same reference frequency).

According to a further feature of the invention, the number of turns N _i of the three input inductors (IND1, IND4, IND7) is identical and corresponds to the first number of turns N _A , the number of turns of the three middle inductors (IND2, IND5, IND8) is identical and corresponds to the second number of turns N _Bα , the number of turns of the three output inductors (IND3, IND6, IND9) is identical and corresponds to the third number of turns N _Bβ , and the number of turns of the three transverse inductors (IND10, IND11, IND12) is identical and corresponds to the fourth number of turns N _C , wherein the turns ratio r is defined by the ratio N _A / N _C , where the following formula applies: $$N_{B\alpha }=N_C*\frac{r-3}{4\sqrt{2}}.$$

In this way, two opposing requirements are met simultaneously and robustly: Firstly, the isolation between the fundamental frequency and harmonics is stably maintained because the effective high-frequency impedance of the main branch relative to the cross branch is forced into a narrow, defined range. Secondly, the 50 Hz stress on the star-connected capacitors is kept low without requiring any switching logic or forced oversizing of the capacitors. This dual effect is not achieved by simply increasing or decreasing the size of individual inductances, but rather results from the specific coupling of the total number of windings in the rear branch to the ratio of the input and cross windings. This reliably keeps resonance effects of the neutral buffer branch away from the fundamental frequency, keeps the neutral currents at mains frequency low, and ensures that the harmonic attenuation remains within its target range even with component tolerances and mains impedance fluctuations.

Compared to the one from the WO 2006/048161 A1 With this known network filter, an unexpectedly robust "self-locking" of the filter impedances is achieved, in which the aforementioned conflicting objectives are resolved not sequentially, but simultaneously. This results in a particularly low THDI value.

Preferably or alternatively, it is provided that the following formula relationship applies to the number of windings of the three output inductors: $$N_{B\beta }=N_C*\frac{r-3}{4\sqrt{2}}.$$

The "input inductances" are the three windings on the three legs, each carrying the input-side series winding (A position); their identical number of turns is designated N _{_A} . The "middle inductances" are the three windings on the three legs, each carrying the middle series winding (B-α position); their identical number of turns is designated N_{_Bα} . The "output inductances" are the three windings on the three legs, each carrying the output-side series winding (B-β position); their identical number of turns is designated N_{_Bβ} . The "transverse inductances" are the three windings of the transverse branches (C position), each located on one of the three legs and leading from the inner node to the star-connected capacitor; their identical number of turns is designated N_{_C} . The winding ratio r is defined as the ratio of the number of turns in the input winding to the number of turns in the cross winding, i.e., r := N _A /N _C . Where the term "identical" is used, this encompasses a tolerance range of ± 5%, preferably ± 2%, and more preferably ± 1% with respect to the number of turns; this interpretation also applies accordingly to the identity of the groups mentioned here.

In a preferred embodiment, the two rear windings of each leg are identical, such that N _Bα = N _Bβ . This maximizes leg symmetry and simplifies manufacturing.

In another preferred embodiment, the above relationship applies to N _Bα and N _Bβ. in conjunction with the lower limits of the number of windings according to claim 1, such that in particular r = N _A / N _C is significantly larger than 1 and the input-side series winding on each leg is significantly increased compared to the rear series windings and compared to the transverse winding. This ensures that the parameterization is effectively implemented in the claimed robust operating range without having to make additional assumptions about the magnetic core constant.

In a particularly preferred embodiment, the number of turns of the central series winding N _{_Bα} increases linearly with the excess of the ratio r = N_{_A/}N_{_C} above a threshold value r _₀. Below this threshold, N _{_Bα} is not required in this preferred embodiment and remains at a minimum value N _{_min}, in particular at 1. In the symmetrical three-legged case, r _₀ = 3, the scaling factor κ is determined by the winding ratios and the winding position and can $\mathrm{¼}*\sqrt{2}$ In a preferred embodiment, N _Bα = κ N _C (r - r ₀ ) for r ≥ r ₀ , and N _Bα = N _min for r ≤ r ₀ .

In particular, the inventors surprisingly discovered that particularly efficient and low-loss filtering can be achieved when the following relationship applies: $$\mathrm{0,25}\le \frac{L_A*\left(L_B-L_C\right)+L_C^2}{L_B{L}_C}\le \mathrm{0,4166}$$

In a preferred embodiment of the above formula, the lower limit is 0.3 and the upper limit is 0.366. In a particularly preferred embodiment of the above formula, the lower limit is 0.315 and the upper limit is 0.348. More preferably, the upper and lower limits are identical and are 0.333.

According to a further feature of the invention, the inductances of the three input inductors (IND1, IND4, IND7) are identical and correspond to the first inductance L _{_A}, the inductances of the three middle inductors (IND2, IND5, IND8) are identical and correspond to the second inductance L _{_Bα} , the inductances of the three output inductors (IND3, IND6, IND9) are identical and correspond to the third inductance L _{_Bβ} , and the inductances of the three transverse inductors (IND10, IND11, IND12) are identical and correspond to the fourth inductance L_{_C}, wherein the capacitance of the capacitive energy storage device is designated as C_{_Y} , wherein the following formula applies: $$C_Y=\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$80$}\right.*\frac{\sqrt{3}}{{\mathrm{\omega }}_{01}^2*\left(L_A+L_C\right)},$$ where ω ₀₁ denotes the angular frequency of the network's fundamental frequency.

This achieves two things simultaneously: Firstly, the current at the terminals of the star connection remains low at the fundamental frequency. Secondly, the star point buffer branch, consisting of the shunt inductor and capacitor, operates at a safe distance from series resonance. As a result, the mains filter reliably separates the fundamental frequency from the harmonic band, without any switching logic for capacitors, with a low voltage drop in the series branch, and with robust damping even in the face of component tolerances and mains fluctuations. This effect only arises from the functional relationship between the inductance ratios and the capacitance limit. Simply increasing or decreasing the size of individual components would not reliably establish this balance.

The capacitance of a capacitor in a star-connected capacitive energy storage system is denoted as C _Y. It can be measured in farads. The angular frequency of the mains fundamental frequency is denoted as w ₀₁ .

For the purposes of this application, the mains fundamental frequency is understood to be the frequency of the fundamental oscillation of the three-phase supply voltage applied to the filter inputs. It is the frequency of the first harmonic in the voltage Fourier spectrum and is denoted by f in Hertz; the corresponding angular frequency is _ω₀ = 2πf. Harmonics, intermediate frequencies, clock or switching frequencies of power electronic devices, and any periodic modulation components are not included in the mains frequency.

When supplied from a public three-phase grid, the grid fundamental frequency is the nominally specified fundamental frequency. When supplied from an islanding source (e.g., converter, generator), the grid fundamental frequency is the fundamental frequency set at the AC output of this source at the filter terminal; it is not identical to the internal switching or carrier frequency of the converter. The nominal grid frequency is used for component design. For evaluating the frequency-dependent relationships specified in this application, the fundamental frequency value present at the time under consideration is decisive.

Preferably, the capacitance of the star point buffer branch is chosen such that it scales inversely proportional to the square of the angular frequency of the mains fundamental frequency ω ₀₁ and inversely proportional to the sum of the input-side series inductance L _A and the transverse inductance L _C.

Preferably, the capacitance of the neutral-point buffer branch is selected such that the product of the square of the angular frequency of the network's fundamental frequency, the capacitance of the neutral-point buffer branch, and the sum of the input-side series inductance and shunt inductance is greater than one twenty-fifth and less than one-ninth. This places the series resonance of the L _C -C _Y branch between the third and fifth harmonics of the network's fundamental frequency, without actually hitting either of these two orders.

The inventors recognized that in the targeted applications, the fifth harmonic is particularly pronounced due to the typical spectrum of nonlinear three-phase loads. This is especially true for loads with a six-pulse rectifier bridge B6 and a downstream DC link, whose line currents characteristically contain harmonics of the 6k±1 order. In practice, the fifth and seventh orders dominate, with the fifth acting as a counter-rotating component of the rotating field and, based on experience, causing increased copper and iron losses, torque pulsations, and additional network feedback. A resonance at this point would further amplify these effects.

The third harmonic belongs to the triple oscillations with in-phase propagation in all phase conductors. In star-connected configurations, these components add up at the neutral point. In a filter structure with star-connected capacitors, a defined return path is available for this purpose. Even slight asymmetries, manufacturing tolerances, or fluctuating network impedances can therefore trigger increased neutral point currents and cause unfavorable magnetization of the core if the series resonance of the cross branch lies precisely in this frequency range.

By deliberately avoiding the third and fifth order frequencies and positioning the series resonance between these two frequencies, peaks at critical points are specifically prevented. The capacitance current at the fundamental frequency remains low, attenuation in the harmonic band sets in early, and the frequency-dependent current distribution between the series branch and the neutral buffer branch remains stable even with tolerances and temperature drift. At the same time, thermal and electrical stresses on the capacitors and the choke are reduced, which has a positive impact on the required installation size and power loss.

By deliberately positioning the series resonance between the third and fifth orders, peaks at precisely these two critical frequencies are avoided. At the same time, the capacitance current at the fundamental network frequency remains low, and the frequency-dependent current distribution between the series branch and the neutral-point buffer branch becomes robust and insensitive to tolerances. This tuning rule is independent of absolute component values and scales with power and size, as it is based solely on the described tuning factor.

Preferably, it is provided that at the angular frequency of the mains fundamental frequency, the capacitance of the star-connected capacitor is selected such that the condition $$\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$25$}\right.<{\omega }_{01}^2*C_Y*\left(L_A+L_C\right)<\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$9$}\right.$$ This condition is met. This ensures that the current at the terminals of the capacitive energy storage remains small at the fundamental frequency and that the star point buffer branch consisting of L _A , L _C , and C _Y is operated with sufficient distance from the series resonance, especially from the critical third and fifth harmonics. In this way, the filter does not act like a resonant circuit.

This design supports the frequency-dependent distribution of currents, keeps the capacitor current low at the fundamental frequency, and at the same time ensures robust damping of harmonics without switching logic for the capacitors.

• 1 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer ersten Ausführungsform;
• 2 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer zweiten Ausführungsform;
• 3 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer dritten Ausführungsform;
• 4 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer vierten Ausführungsform;
• 5 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer fünften Ausführungsform;
• 6 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer sechsten Ausführungsform;
• 7 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer siebten Ausführungsform;
• 8 einen Schaltplan eines erfindungsgemäßen dreiphasigen Oberschwingungs-Netzfilters in einer achten Ausführungsform; und
• 9 ein Oszillogramm der Stromverläufe an dem Netzeingang und dem Ausgang eines erfindungsgemäßen Dreiphasen-Oberschwingungs-Netzfilters.

Further advantages and features of the invention will become apparent from the following description with reference to the figures. These show

• 1 a circuit diagram of a three-phase harmonic mains filter according to the invention in a first embodiment;
• 2 a circuit diagram of a three-phase harmonic mains filter according to the invention in a second embodiment;
• 3 a circuit diagram of a three-phase harmonic mains filter according to the invention in a third embodiment;
• 4 a circuit diagram of a three-phase harmonic mains filter according to the invention in a fourth embodiment;
• 5 a circuit diagram of a three-phase harmonic mains filter according to the invention in a fifth embodiment;
• 6 a circuit diagram of a three-phase harmonic mains filter according to the invention in a sixth embodiment;
• 7 a circuit diagram of a three-phase harmonic mains filter according to the invention in a seventh embodiment;
• 8 a circuit diagram of a three-phase harmonic mains filter according to the invention in an eighth embodiment; and
• 9 an oscillogram of the current waveforms at the mains input and output of a three-phase harmonic mains filter according to the invention.

1 Figure 110 shows a three-phase harmonic mains filter. The three filter inputs L1, L2, and L3 are connected to the three filter outputs L11, L12, and L13. The mains filter is designed so that currents at the mains fundamental frequency are passed from the inputs to the outputs, while interference currents with frequencies deviating from the mains fundamental frequency, which arise at the outputs due to the connected load, are effectively attenuated towards the inputs. The voltage drop across the mains filter should be kept as low as possible.

The mains filter has a common three-legged filter core 130 with legs 132, 134, and 136. The series inductances of the three phases are distributed across these three legs, so that each phase is represented on all three legs. This distribution creates targeted magnetic coupling between the phases. The phase shift of the conductor currents by 120 degrees causes the magnetic fluxes in the legs to partially cancel each other out. This reduces the magnetic energy that needs to be stored in the core and contributes to a low voltage drop.

An internal node lies between the two series inductors of each phase. From this node, a cross branch leads via a cross inductor to a capacitive energy storage device whose three capacitors are connected in a star configuration. The cross branch, together with the neutral point of the capacitor, forms a neutral buffer branch. This branch exhibits a high impedance at the fundamental frequency of the mains and preferentially carries away high-frequency current components. In this way, the current at the capacitor terminals remains low at the fundamental frequency, while harmonics are selectively decoupled from the series branch.

The star-shaped connection of the capacitors creates a phase-shifted buffering of the high-frequency current components. These are guided from the three phases to the star point via the transverse inductances and absorbed by the other phases. This ensures a stable separation between the fundamental frequency and the harmonic band, keeps resonance effects of the transverse branch away from the fundamental frequency, and maintains robust damping even with component tolerances and mains fluctuations.

The input-side longitudinal winding of each leg has a significantly higher number of turns than the two subsequent longitudinal windings of the same leg and than the corresponding transverse winding. This winding staggering promotes the frequency-dependent distribution of currents, keeps the longitudinal branch low-impedance at the fundamental frequency, and increases its impedance in the harmonic range compared to the transverse branch.

The windings are preferably configured with the same winding direction. The symmetry across all three legs and the uniform winding direction improve flux compensation, reduce voltage drop, lower reactive current at no load and under partial load, and enable compact cross branches and capacitors without switching logic.

The mains filter 110 comprises three phase-related filter branches 120, 122, and 124 between inputs L1, L2, and L3 and outputs L11, L12, and L13. The three-legged filter core 130 has legs 132, 134, and 136. Filter branch 120 includes the series-connected inductances IND1, IND2, and IND3 between L1 and L11. IND1 is wound on the first leg 132, IND2 on the second leg 134, and IND3 on the third leg 136. Filter branch 122 includes the series inductances IND4, IND5, and IND6 between L2 and L12. IND4 is located on the second leg 134, IND5 on the third leg 136, and IND6 on the first leg 132. The filter branch 124 comprises the series inductances IND7, IND8, and IND9 between L3 and L13. IND7 is wound on the third leg 136, IND8 on the first leg 132, and IND9 on the second leg 134.

An internal node lies between the first and second series inductors of each filter branch. Internal node 24 of filter branch 120 connects IND1 and IND2. Internal node 34 of filter branch 122 connects IND4 and IND5. Internal node 44 of filter branch 124 connects IND7 and IND8. From these nodes, the transverse inductors IND10, IND11, and IND12 lead to terminals 152, 154, and 156 of the capacitive energy storage device 150. The energy storage device 150 consists of three capacitors C1, C2, and C3 connected in a star configuration.

The mechanical arrangement of the windings is fixed. On the first leg 132, inductors IND1, IND8, and IND6 are arranged sequentially. On the second leg 134, inductors IND4, IND2, and IND9 are located. On the third leg 136, inductors IND7, IND5, and IND3 are arranged. All windings preferably have the same winding configuration. lung sense. The assignment of the inner nodes to the transverse inductances and to the terminals of the star connection is in 1 evident.

Inputs L1, L2, and L3 serve as mains connections. Outputs L11, L12, and L13 form the device connection points. A symmetrical design is assumed for the following description. The three input-side series inductances IND1, IND4, and IND7 are collectively referred to as inductance L_{_A. The three middle series inductances IND2, IND5, and IND8 are collectively referred to as inductance L_Bα}. The three output-side series inductances IND3, IND6, and IND9 are collectively referred to as inductance L _{_Bβ} . The transverse inductances IND10, IND11, and IND12 are collectively referred to as inductance L_{_C} .

The star-connected capacitor, together with the transverse inductors, forms the neutral buffer branch. This branch absorbs high-frequency current components from the three phases and releases them out of phase. At the network's fundamental frequency, the neutral buffer branch provides a high impedance, so the fundamental frequency current is carried in the series branch, and the current through the capacitor remains small.

The in 1 The arrangement shown can be part of larger filter systems. The distribution of the longitudinal windings across the three legs, the symmetrical magnetic coupling of the three phases, and the design of the neutral buffer branch can be readily transferred to scaled power levels and adapted component values.

The following describes the basic operating principle and calculation of a harmonic filter. This is illustrated using the three-phase mains filter 110 according to the 1 shown. Of course, it is possible to replicate or recalculate various versions of the circuit arrangement analogous to the circuit described here.

First, the calculation basis for the filter is outlined. The starting point is the voltage drop across the input-side series inductances L_{_A} and L_{_B}, expressed as the relative short-circuit voltage U_K. The value of U_K is chosen so that the voltage drop in the series branch remains low while simultaneously achieving effective attenuation of the harmonics generated by the load. Increasing L_{_A} generally reduces harmonic distortion at the input, but it affects the overall voltage drop of the filter and the required capacitance values; therefore, the design requires a carefully considered selection of parameters.

From the specified rated current I_r of the three-phase harmonic mains filter 110, and taking into account the desired short-circuit voltage U_K, the inductance L _A of the input-side series winding is calculated. The output-side series inductance of the series branch is formed by the two downstream inductances L _Bα and L _Bβ , the sum of which is represented as L _B in the equivalent circuit diagram. The inductance of the shunt branch is designated as L _C. The values mentioned each refer to one leg; in a symmetrical design, the three L _A , the three L _Bα , the three L _Bβ , and the three L _C inductances are identical.

From L _{_A}, L _{_B} = L _{_Bα} + L _{_Bβ} , and L_{_C}, the required magnetic energy content of the three-phase iron-core inductor can be determined. According to the invention, this energy content is reduced by two effects: Firstly, the distribution of the longitudinal windings across the three legs with phase-correct magnetic coupling causes a partial compensation of the fluxes, so that for the longitudinal branch, an effective difference arises between the contribution from L_{_A} and the contribution from L_{_B}. Secondly, at the fundamental frequency of the network, a significantly lower current flows in the neutral-point buffer branch consisting of L_{_C} and the star-connected capacitor than in the longitudinal branch. The resulting energy contribution of the shunt branch therefore scales with a significantly smaller current component than the contribution of the longitudinal branch.

The aforementioned reduction in energy content is a direct consequence of the winding distribution according to the invention across all three legs and the resulting coupling of all three phases. If L_{_A} and L_{_B} of a branch were located on the same leg, the compensating difference formation would be eliminated, the core would have to be larger, and losses would increase. In contrast, the arrangement according to the invention reduces the core cross-sectional area required and lowers the field energy during operation.

The core size is selected based on the inductance values determined in this way. The determination of the magnetic characteristic (AL value) and the conversion between inductance and number of turns are carried out according to standard choke design rules and are not described in detail here. It is essential that the inductance values correspond to the number of turns according to the invention and thus fall within the robust operating range of the filter.

A further advantage of the design according to the invention lies in the dimensioning of the transverse inductance L_{_C} . Due to the magnetic coupling of the three phases and the winding staggering at Due to the input-side series winding, the number of turns N_{_C} of the shunt inductance can be chosen to be comparatively high without resulting in a disproportionately large overall size. This is because the RMS current in the shunt branch remains significantly below the rated current I_r at the fundamental frequency of the mains. Therefore, a smaller conductor cross-section can be used for L_{_C} , and the physical size of the shunt inductance remains small despite the high number of turns.

The star-connected capacitor simultaneously provides phase-shifted buffering of the high-frequency current components. The coupling of the inductors according to the invention positions the frequency-dependent impedance such that the fundamental frequency current is carried in the series branch, while harmonics preferentially flow via L_{_C} to the star point. This keeps the current at the capacitor terminals low at the mains fundamental frequency and keeps critical series resonance effects of the L _{_C}-C branch away from the fundamental frequency range.

Surprisingly, it was found that when the winding ratios according to the invention are maintained between the input-side series winding, the two downstream series windings, and the shunt winding, the current in the neutral buffer branch remains within a narrow, tolerance-robust range at the fundamental frequency of the mains, which is significantly below the rated current I_r. This behavior occurs without additional switching logic for capacitors and is maintained even with component variations and typical fluctuations in the mains impedance. Consequently, the required capacitance value decreases, the reactive current at no load and under partial load is reduced, and the harmonic attenuation at the input remains stable over the relevant frequency range.

To enhance this effect, the capacitance of the star-connected capacitors can preferably be selected such that the L-C condition specified in this description is met at the mains fundamental frequency. This ensures that the star point buffer branch operates at a sufficient distance from a series resonance and that the current at the capacitor terminals remains low.

The specific numerical determination of the component values is based on the known equivalent circuit diagrams of the filter and taking into account the relationships between the inductances and the number of windings specified in this application. The rated current I_r, the target relative short-circuit voltage u_K across L(A), and the target specifications regarding harmonic feedback at the input serve as input data for the design.

The reduction of the cross-branch current at the mains fundamental frequency and the simultaneous increase in the high-frequency impedance of the main branch according to the invention result in the total harmonic distortion at the input (THDI) being within the desired quality range, without unduly increasing the voltage drop of the main branch. This balance does not result from an isolated change in individual inductances, but from the functional coupling of the number of turns ratios and the resulting position of the impedance-determining parameters.

Due to the significantly lower current in the neutral buffer branch during fundamental mode operation, the shunt inductors can be designed with smaller wire cross-sections and the capacitors can be dimensioned smaller. This reduces the overall size and costs, lowers the reactive current at no load and under partial load, and eliminates the need to disconnect the capacitors in these operating conditions.

In summary, the inventive combination of series winding distribution across the three legs, magnetic coupling of all three phases, pronounced winding staggering, and optimized dimensioning of the neutral point buffer branch allows for a compact, low-loss, and tolerance-robust filter design. The fundamental frequency current is routed to the load with minimal loss, high-frequency components are reliably diverted to the cross branch, and input distortions remain within the target range without the need for switchable capacitance logic.

The input current of the three-phase harmonic mains filter 110 is almost sinusoidal during normal operation, which corresponds to the filter's intended use. The output current can be highly distorted, ranging from almost block-like, depending on the connected load (see [reference]). 9 The current flowing through the neutral buffer branch is derived from the known relationship between input and output current. This branch consists of the shunt inductance L_{_C} and the star-connected capacitance for each phase, and carries two essential current components: firstly, the capacitive component at the mains fundamental frequency, and secondly, the non-sinusoidal differential current between input and output under load.

During idle and partial load operation, the current in the neutral buffer branch is characterized by the capacitive component at the network's fundamental frequency. Under load, this is superimposed by the high-frequency differential current, which accounts for the gaps in the time intervals of the The output current is bridged on the energy side. According to the invention, it has been found that, by adhering to the described winding staggering (significantly increased number of turns of the input-side series winding compared to both downstream series windings and compared to the associated shunt winding), the current at the terminals of the star-connected capacitor remains within a narrow, tolerance-robust range at the mains fundamental frequency, while high-frequency components preferentially shift into the star point buffer branch. This achieves a stable isolation between the fundamental frequency and the harmonic band without the need for switching logic.

For the capacitance design, a purely component-related condition is preferably used, which limits the 50 Hz current (or the current at the fundamental frequency of the grid) at the capacitance terminals and simultaneously ensures a safe distance from a series resonance of the L _C - C branch. For this purpose, the capacitance C _Y is selected such that at the angular frequency of the fundamental frequency of the grid, with ω ₀₁ = 2πf, the following holds: $$\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$25$}\right.<{\omega }_{01}^2*C_Y*\left(L_A+L_C\right)<\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$9$}\right.$$ This setting keeps the capacitance current small at the fundamental frequency, positions the impedance-determining quantities in a robust range, and avoids approaching the series resonance.

In a preferred embodiment, the aforementioned LC condition is additionally linked to the ratio of the inductances of the input-side series winding to the shunt inductance. This results in a closed upper limit for C _Y as a function of the inductance ratio k = L(A)/L(C). This functional coupling has proven particularly advantageous because it specifically fixes the high-frequency impedance of the series branch relative to the neutral buffer branch and simultaneously keeps the 50 Hz capacitance current low – even with component variations and typical network impedance fluctuations.

While a further increase in capacitance can reduce harmonic distortion at the input in specific areas, experience has shown that this leads to disadvantages: The capacitive reactive current at the mains fundamental frequency increases, the approach to series resonance intensifies, and additional requirements arise regarding protection and physical size. The condition for C _Y specified according to the invention avoids these conflicting objectives and allows for a compact, low-loss design without switchable capacitor banks.

Fine-tuning the relationship between the individual inductances of the multiple choke and the capacitance – for example, through simulations and measurements – can further optimize the filtering effect. This does not affect the fundamental effect of the arrangement according to the invention: The distribution of the series windings across the three legs with symmetrical coupling of all three phases, as well as the winding staggering, results in a reduction of the energy to be stored in the core and a stable separation between the fundamental frequency and harmonics; the capacitance condition ensures that this effect is reproducibly achieved in series operation.

A further advantage becomes apparent in conjunction with typical loads, especially devices with a rectifying input section and DC link capacitance: The current waveform provided by the filter at the output results in a low ripple current in the internal smoothing capacitors of the load. This reduces the thermal stress on these components and can increase the service life of the connected device.

The aforementioned relationships can be analytically determined using known equivalent circuit diagrams (single-phase equivalent network per phase). T- and π-equivalent circuits are interconvertible. However, the decisive factor for the present design is not the choice of a specific equivalent circuit diagram, but rather adherence to the winding staggering specified by the invention and the LC condition for C _Y , which achieves the described isolation effect and the robustness of the filter function.

In the embodiments shown, the minimum ratios of the number of windings per leg specified in claim 1 apply.

2 Figure 1 shows a circuit diagram of a three-phase mains filter according to a second embodiment of the present invention. The mains filter is designated 210 in its entirety. The structure and function of the mains filter 210 differ only slightly from the structure and function of the one described in Figure 210. 1 The network filter 110 shown is described here, so only its different features are described. It should be noted in particular that identical reference symbols here and in the following figures refer to the same devices.

Shown in 2 In particular, the geometric arrangement of the transverse inductances IND10, IND11, IND12 on the legs of the filter core. The transverse inductance IND10 belonging to the first filter branch 120 is wound on the first leg 132. The transverse inductance IND11 belonging to the second filter branch 122 is wound on the second Leg 134 of the filter core is wound. The transverse inductance IND12 belonging to the third filter branch 124 is wound on the third leg 136 of the filter core. Such a winding results in the transverse inductances IND10, IND11, IND12 being strongly coupled to the input-side series inductances IND1, IND4, IND7 of the respective filter branch.

Since the transverse inductances IND10, IND11, IND12 have the same winding direction as the corresponding input-side series inductances IND1, IND4, IND7, the input-side series inductances IND1, IND4, IND7 and the transverse inductances IND10, IND11, IND12 are connected in series with respect to an input current flowing into the filter at the filter inputs L1, L2, L3, and therefore represent a high inductance. This reduces the leakage of the input current via the transverse branch and thus reduces the generation of reactive currents in the mains filter 210. The further operation of the filter 210 is described in relation to the one shown in 1 The filter shown, 110, remains unchanged, so a description is omitted here.

3 Figure 1 shows a circuit diagram of a three-phase mains filter according to a third embodiment of the present invention. This is again very similar to the one shown in the 1 and 2 The filter shown is used, so only the differences are described here. The description is based on the following: 2 The mains filter 210 shown is used. The present mains filter is designated 260. The same reference numerals again indicate the same devices as in the previously described embodiments.

The structure of the mains filter 260 is unchanged compared to the mains filter 210. The only difference is the mechanical position of the output-side inductors IND2, IND5, IND8 and IND3, IND6, IND9 on the legs 132, 134, 136 of the filter core. The order of the inductors with respect to the current flow from the filter input to the filter output is unchanged for the filter 260 compared to the filter 210. Thus, for example, in the first filter branch 120, inductors IND1, IND2, and IND3 are located in precisely this order between the filter input L1 and the filter output L11. A similar arrangement applies to the second filter branch 122 and the third filter branch 124. However, the mechanical arrangement of the inductors on the filter legs is different for the mains filter 260 compared to the mains filter 210. However, inductances IND1 and IND10 remain on the first filter leg 132, inductances IND4 and IND11 on the second filter leg, and inductances IND7 and IND12 on the third filter leg. The arrangement of the output inductances, however, has changed. Inductance IND2 of the first filter branch is now located on the third filter leg 136, and inductance IND3 of the first filter branch 120 is now located on the second filter leg 134. Furthermore, the arrangement of inductance IND5 of the second filter branch 122, which in filter 260 is wound on the first leg 132, and inductance IND6 of the second filter branch 122, which is now wound on the third leg 136, has also changed. Finally, the inductance IND8 of the third filter branch 124 is wound on the second leg 134 and the inductance IND9 on the first leg 132.

A modified mechanical arrangement of the inductors on the filter legs leaves the properties of the mains filter 260 essentially unchanged, but represents a further embodiment which may be mechanically advantageous depending on the circumstances.

4 Figure 1 shows a circuit diagram of a three-phase mains filter according to the fourth embodiment of the present invention. The mains filter shown is designated 310 in its entirety. The mains filter 310 is also similar in structure and function to the one shown in Figure 2. 2 The network filter shown is very similar to the one shown (210). Therefore, only the differences will be discussed here. The same reference symbols denote the same devices.

In the mains filter 310, the coupling of the shunt branch is not between the first and second inductors IND1, IND2; IND4, IND5; IND7, IND8 of each filter branch (counting from the filter input), but between the second and third inductors IND2, IND3; IND5, IND6; IND8, IND9. For further explanation, let's consider the first filter branch 120. The shunt inductor IND10 of the first filter branch 120 is now coupled between inductor IND2 and inductor IND3. Regarding the further circuitry, in particular the distribution of the inductances across the branches, there are no differences between the mains filters 210 and 310.

The mains filters 210 and 310 do not differ significantly in their fundamental properties. However, differences can arise in their dimensioning, i.e., the design of the inductances and/or capacitances. Depending on the requirements and the mechanical conditions, a filter arrangement 210 can therefore be configured according to... 2 or a filter assembly 310 according to 4 be more advantageous.

5 shows a circuit diagram of a three-phase mains filter according to the invention. fifth embodiment of the present invention. The filter corresponds in its basic structure and function to those described in the 1 , 2 , 3 until 4 The filters shown are described below, so reference is made here again to their descriptions. The same reference symbols indicate the same devices as in the previously described network filters. This is based on... 5 The power line filter shown is designated 360 in its entirety. The distribution of inductances across the filter cores corresponds to that shown in... 3 The network filter 260 shown. However, the cross branches branch off, similarly to the one based on... 4 described mains filter 310, between the second and third inductance IND2, IND3; IND5, IND6; IND8, IND9 of each filter branch 120, 122, 124.

Such an embodiment, in turn, represents an alternative to the one based on 3 shown filter 260 as well as the one based on 4 The filter shown is 310. The properties are essentially unchanged, however, a different dimensioning of the inductances and capacitances is required.

6 Figure 1 shows a circuit diagram of a three-phase mains filter according to a sixth embodiment of the present invention. The filter is designated as 410 in its entirety and is based on the circuit shown in Figure 2. 2 The filter shown is 210. Identical reference symbols denote identical devices. Features of filter 410 that are unchanged compared to filter 210 are not described again here. Instead, reference is made to the descriptions of filter 210 and filter 110, respectively.

Filter 410 is extended compared to filter 210 by the introduction of a second cross-branch. This branch comprises inductors IND13, IND14, and IND15, as well as a second capacitive energy storage device 420, which includes three capacitors C4, C5, and C6. The second capacitive energy storage device 420 has a first terminal 422, a second terminal 424, and a third terminal 426. It should also be noted that the inductors of the first cross-branch are now collectively designated L(C1), while the inductors IND13, IND14, and IND15 of the second cross-branch are collectively designated L(C2). The inductor IND13 of the second cross branch is connected to the junction between the second inductor IND2 and the third inductor IND3 of the first filter branch 120, as well as to the first terminal 422 of the second capacitive energy storage device 420. The inductor IND13 of the second cross branch of the first filter branch 120 is wound on the first leg 132. The winding direction is the same as that of all other inductors.

In a manner analogous to the inductance IND13 of the first filter branch, the inductances IND14 and IND15 of the second and third filter branches are also connected and wound on the second and third legs 134, 136 of the three-phase filter core, respectively. The details of the connection are described in the 6 to be taken.

A mains filter 410 with a second cross-branch can be designed to achieve better filtering performance than a mains filter with only one branch. In particular, the cross-branchs can be dimensioned to suppress two unwanted frequencies. Overall, there are more degrees of freedom in the filter design because the filter is of a higher filter order. This also increases the complexity of implementing a mains filter with two cross-branchs, as additional cross-branch inductances IND13, IND14, IND15 and additional capacitors C4, C5, C6 are required. Depending on the requirements, it is therefore advantageous to use either a filter with only one cross-branch or a filter 410 with two cross-branchs.

7 Figure 1 shows a circuit diagram of a three-phase mains filter according to a seventh embodiment of the present invention. This filter essentially corresponds to the one shown in Figure 2. 6 shown mains filter 410, wherein the inductances in the longitudinal branch are as in the one based on 3 filter 260 shown, instead of as in the one based on 2 The filters shown, 210, are interconnected. Filter 460 therefore represents only another alternative that can be used depending on the requirements and mechanical conditions.

8 Figure 510 shows a circuit diagram of a three-phase mains filter according to an eighth embodiment of the present invention, which is designated in its entirety by 510. The filter essentially corresponds to the one shown in the 2 and 3 The network filters 210 and 260 shown are described in detail, so that unchanged devices are not described again. Instead, reference is made to the previous description. In particular, identical reference numerals also indicate identical devices. Filter 510 is modified compared to filter 210 in that the energy storage device 150' here comprises a delta connection of capacitors C1', C2', and C3'. A delta connection of capacitors offers the advantage over a star connection, as shown in network filter 210, that the capacitors need to have a lower capacitance. However, it is necessary that the capacitors of a Delta-connected capacitors have a higher voltage rating than those in a star connection. Furthermore, it is not possible to ground one of the capacitor terminals when using a delta connection.

Therefore, whether a star connection of capacitors or a delta connection of capacitors is more advantageous depends on the application and the requirements.

The mains filters shown can be modified extensively without deviating from the core concept of the invention. For example, it is possible to use only one series inductor (e.g., IND2, IND5, and IND8) on the output side of each filter branch and to omit the second inductor (e.g., IND3, IND6, IND9). While complete symmetry is no longer guaranteed with such a filter, it still offers advantages over a conventional filter in which all inductors of a filter branch are arranged on the same leg of the filter core.

Furthermore, it is possible to wind the transverse inductances IND10, IND11, IND12 and, if applicable, IND13, IND14, IND15 of a filter branch onto a different leg 132, 134, 136 of the filter core than the input-side inductance IND1, IND2, IND3. Such an exchange provides an additional degree of freedom in the design and dimensioning of a mains filter.

Furthermore, it is also easily possible to supplement a mains filter with additional filter stages to obtain a higher-order filter. While such a filter is more complex to manufacture, it offers improved filtration characteristics if properly designed. This may be necessary when high filtration performance is required.

Furthermore, it is possible to add additional capacitors or inductors to the filter. For example, several cross-branches can be coupled to a junction between two series inductors located between the filter input and the filter output. A cross-branch can comprise not only a series connection of an inductor and a capacitive energy storage element, but also a capacitor itself. This can be helpful in suppressing high-frequency interference, provided the capacitor is designed such that the capacitive reactive current at the nominal frequency of the mains filter is sufficiently small.

Furthermore, the filter can include switching devices that allow it to be adapted to different operating conditions. For example, it can be advantageous to switch off cross-capacitances. It can also be desirable to bridge individual inductances. This allows the voltage drop across the filter or a reactive current component generated by the filter to be influenced. This can be advantageous if very strong load changes can occur or if the filter needs to be configurable for a wide variety of operating conditions.

Finally, there is a high degree of flexibility in the design of the multiphase filter core. In principle, all available core types can be used, for example, cores made of iron or iron powder.

9 shows an oscillogram of the current waveforms at the mains input and output of a mains filter according to the invention as described in 1 or 2 The filter shown. The oscillogram in its entirety is labelled 610. It shows a first curve 620, which represents the current waveform at the input of the mains filter according to the invention. The abscissa t represents time, while the ordinate I represents the input current. Similarly, the oscillogram shows a second curve 630, which represents the output current at the output of the mains filter according to the invention. Again, the abscissa t represents time, while the ordinate I represents the current.

For the measurement, a mains filter according to the invention is connected to a three-phase load, which has internal B6 rectification and capacitor smoothing. The input current of the mains filter, described by the waveform 620, is essentially sinusoidal. The output current, described by the waveform 630, is approximately block-shaped. The current waveform at the filter output shows a very steep rise and a very steep fall of the current, while the current is almost constant for large current values. In the vicinity of the zero crossing, the current changes only slightly over time, so that the current flow is almost constant for a time interval of about 2 ms (with a period of 20 ms).

It should also be mentioned that the current waveform shown has a period of approximately 20 ms, which corresponds to a frequency of 50 Hz. The amplitude of the current is approximately 250 amperes.

It turns out that the current flowing into the connected device, due to its block-like waveform, can lead to a very low ripple current in the device's internal capacitors. This can result in an increased lifespan for the electrolytic capacitors in the device and thus a longer service life. lifespan of the connected consumer device.

In summary, the invention relates to a passive multiphase harmonic mains filter consisting of a multi-arm distributed multiple-winding choke and a capacitive energy storage device. The filter is connected between the power supply and a non-linear load and serves to significantly reduce current harmonics at the mains input. The invention is particularly suitable for devices with a rectifying input section and downstream smoothing, such as drive systems with B6 rectification.

The unique feature of the invention lies not only in the topology, but also in the interlocking design comprising the winding distribution, minimum ratios of the number of turns per leg, and a component-specific capacitance requirement for the neutral buffer branch. Three series windings are provided for each phase, distributed across the three legs of a common three-leg core. From an inner node, a transverse branch with transverse inductance leads to the star-connected capacitance for each phase. The input-side series winding has a significantly higher number of turns per leg than the two subsequent series windings and the associated transverse winding. In a preferred embodiment, the number of turns within groups A, Bα, Bβ, and C are symmetrically identical across the legs, and the subsequent windings fulfill the functional relationship with the ratio of A to C windings specified in the description. This staggering inevitably results in the impedance of the longitudinal and transverse branches being set in such a way that the fundamental frequency current is carried in the longitudinal branch, while high-frequency components escape into the star point buffer branch.

The star-connected capacitor, together with the shunt inductance in each phase, forms the neutral buffer branch. Preferably, the capacitor is selected such that the LC condition specified in the description is met at the mains fundamental frequency. This keeps the 50 Hz current at the capacitor terminals low and safely avoids the series resonance of the _LC -C branch. The interaction of the winding ratios and the capacitor condition results in a tolerance-robust self-locking of the frequency-dependent impedances: the separation between the fundamental frequency and the harmonic band is established stably, without the need for switchable capacitor banks or switching logic.

The distribution of the series windings across all three legs and the uniform winding direction result in phase-matched magnetic coupling of the three phases. Due to the 120-degree phase angle, flux components in the core partially cancel each other out. This reduces the magnetic energy to be stored in the core and the effective voltage drop in the series branch. Simultaneously, the required energy contribution of the transverse branch decreases because the capacitance current remains small during the fundamental frequency. This allows the transverse inductances to be implemented with smaller conductor cross-sections and the capacitances with smaller values. The no-load and partial-load reactive currents are low, and switching off the capacitances is generally unnecessary. Overall, this results in a compact design, lower losses, and a marketable solution for reducing network disturbances.

The filter's effect is also evident on the load side: The nearly block-shaped output current results in low ripple currents in the load's internal smoothing capacitors, thus reducing their thermal stress and aging. The invention is not limited to a single detailed embodiment. As an alternative to the star connection, the capacitive energy storage can also be connected in a delta configuration, provided the voltage rating is appropriate. However, the star connection with a defined reference point is preferred. Similarly, additional filter stages or a second cross branch can be provided to attenuate additional frequency bands. Regardless of such variations, the core of the invention remains: the three-legged distribution of the series windings, the leg-symmetrical coupling of all three phases, the pronounced winding staggering with the specified ratios, and the capacitance condition of the star point buffer branch, which resolves the aforementioned conflicting objectives simultaneously and robustly.

Reference sign

110

Three-phase harmonic power line filter

120

First filter branch

122

Second filter branch

124

Third filter branch

130

Three-pronged filter core

132

First limb (of the filter core)

134

Second leg (of the filter core)

136

Third limb (of the filter core)

150

Capacitive energy storage device

152

First connection (of the capacitive energy storage device)

154

Second connection (of the capacitive energy storage device)

156

Third connection (of the capacitive energy storage device)

210

Three-phase harmonic power line filter

260

Three-phase harmonic power line filter

310

Three-phase harmonic power line filter

360

Deriphase harmonic power line filter

410

Three-phase harmonic power line filter

460

Three-phase harmonic power line filter

510

Three-phase harmonic power line filter

CY

Capacity (of the capacitive energy storage device)

IND1

First input inductance

IND2

First medium inductance

IND3

First output inductance

IND4

Second input inductance

IND5

Second middle inductance

IND6

Second output inductance

IND7

Third input inductance

IND8

Third medium inductance

IND9

Third output inductance

IND10

First transverse inductance

IND11

Second transverse inductance

IND12

Third transverse inductance

k

Inductance ratio

L1

First filter inlet

L2

Second filter inlet

L3

Third filter inlet

L11

First filter output

L12

Second filter output

L13

Third filter output

LA

First inductance

LBα

Second inductor

LBβ

Third inductance

LC

Fourth inductor

N1

Number of windings of the first input inductor

N2

Number of turns of the first middle inductor

N3

Number of turns of the first output inductor

N4

Number of windings of the second input inductor

N5

Number of turns of the second middle inductor

NC

Number of windings of the second output inductor

N7

Number of windings of the third input inductor

NC

Number of turns of the third middle inductor

N9

Number of turns of the third output inductor

N10

Number of turns of the first transverse inductance

N11

Number of turns of the second transverse inductance

N12

Number of turns of the third transverse inductor

N/A

First winding number (number of windings of the input winding)

NB

Output inductance (sum of N _Bα and N _Bβ )

NBα

Second number of windings (first rear inductor)

NBβ

Third winding number (second rear inductor)

NC

Fourth winding number (number of turns of the transverse winding)

r

winding ratio

QUOTES INCLUDED IN THE DESCRIPTION

This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions.

Cited patent literature

• WO 2006/048161 A1 [0010, 0038]

Claims (4)

Three-phase harmonic mains filter (110; 210; 260; 310; 360; 410; 460; 510), comprising: - a first filter branch (120) between a first filter input (L1) and a first filter output (L11), wherein the first filter branch (120) comprises a first series connection of three inductors (IND1, IND2, IND3) connected between the first filter input (L1) and the first filter output (L11), and wound on three different legs (132, 134, 136) of a three-leg filter core (130); - a second filter branch (122) between a second filter input (L2) and a second filter output (L12), wherein the second filter branch (122) comprises a second series connection of three inductors (IND4, IND5, IND6) connected between the second filter input (L2) and the second filter output (L12), and are wound on three different legs (132, 134, 136) of the three-legged filter core (130), - and a third filter branch (124) between a third filter input (L3) and a third filter output (L13), wherein the third filter branch (124) has a third series connection of three inductors (IND7, IND8, IND9) which are connected between the third filter input (L3) and the third filter output (L13), and which are wound on three different legs of the three-legged filter core (130), wherein the input inductances (IND1, IND4, IND7) or the output inductances (IND3, IND6, IND9) of the three filter branches (120, 122, 124) are on different legs (132, 134, 136) of the three-legged filter core, wherein the first filter branch (120) comprises a first transverse inductance (IND10), the second filter branch (122) comprises a second transverse inductance (IND11), and the third filter branch (124) comprises a third transverse inductance (IND12), wherein a node to which two inductances (IND1, IND2; IND2, IND3) of the first series connection are connected is coupled via the first transverse inductance (IND10) to a first terminal (152) of a capacitive energy storage device (150), wherein a node to which two inductances (IND4, IND5; IND5, IND6) of the second series connection are connected is coupled via the second transverse inductance (IND11) to a second terminal (154) of the capacitive energy storage device (150), wherein a node to which two inductances (IND7, IND8; IND8, IND9) of the third series circuit are connected, is coupled via the third transverse inductance (IND12) to a third terminal (156) of the capacitive energy storage device (150), and wherein the three transverse inductances (IND10, IND11, IND12) are each arranged on one of the three legs of the three-legged filter core, characterized in that the respective inductances IND_i, ie [1;12] have a respective number of turns N _i , i ∈ [1;12], wherein the following ratios apply: - N ₁ /N ₈ > 8, - N ₁ /N ₆ > 8, - N ₄ /N ₂ > 8, - N ₄ /N ₉ > 8, - N ₇ /N ₅ > 8, - N ₇ /N ₃ > 8, - N ₁ /N ₁₀ > 2.5, - N ₄ /N ₁₁ > 2.5, - N ₇ /N ₁₂ > 2.5. Three-phase harmonic mains filter (110; 210; 260; 310; 360; 410; 460; 510) according to Claim 1 , characterized in that the number of turns of the three input inductors (IND1, IND4, IND7) are identical, and/or that the number of turns of the three output inductors (IND3, IND6, IND9) are identical, and/or that the number of turns of the three middle inductors (IND2, IND5, IND8) are identical, and/or that the number of turns of the three transverse inductors (IND10, IND11, IND12) are identical. Three-phase harmonic mains filter (110; 210; 260; 310; 360; 410; 460; 510) according to one of the Claims 1 or 2 , characterized in that the number of turns N _i of the three input inductors (IND1, IND4, IND7) are identical and correspond to the first number of turns N _A , and that the number of turns of the three middle inductors (IND2, IND5, IND8) are identical and correspond to the second number of turns N _Bα , and that the number of turns of the three output inductors (IND3, IND6, IND9) are identical and correspond to the third number of turns N _Bβ , and that the number of turns of the three transverse inductors (IND10, IND11, IND12) are identical and correspond to the fourth number of turns N _C , wherein the winding ratio r is defined by the ratio N _A /N _C , where the following formula relationship applies: $$N_{B\alpha }=N_C*\frac{r-3}{4\sqrt{2}}.$$ Three-phase harmonic mains filter (110; 210; 260; 310; 360; 410; 460; 510) according to one of the Claims 1 until 3 , characterized in that the inductances of the three input inductors (IND1, IND4, IND7) are identical and correspond to the first inductance L _A , and that the inductances of the three middle inductors (IND2, IND5, IND8) are identical and correspond to the second inductance L _Bα , and that the inductances of the three output inductors (IND3, IND6, IND9) are identical and correspond to the third inductance L _Bβ , and that the inductances of the three Transverse inductances (IND10, IND11, IND12) are identical and correspond to the fourth inductance L _C , where the capacitance of the capacitive energy storage device is referred to as C _Y , where the following formula relationship applies: $$C_Y=\raisebox{1ex}{$3$}\!\left/ \!\raisebox{-1ex}{$80$}\right.*\frac{\sqrt{3}}{{\mathrm{\omega }}_{01}^2*\left(L_A+L_C\right)},$$ where ω ₀₁ denotes the angular frequency of the network's fundamental frequency.