Converter and method for determining a current space vector

Abstract

Method for determining a current space vector and an offset of a current measurement, where the motor current of an inverter-fed electric motor is measured, with means arranged in the bridge branches of the inverter for scanning the currents flowing in the bridge branches, where the motor current space indicator is non-vanishing, wherein, in a bridge branch that is not energized at the time of measurement, the current detected by the means arranged in that bridge branch is recorded as an offset value for the current detection in that bridge branch, wherein the respective switched-off bridge branch, in which the respective current offset value is determined, is chosen differently depending on the average output voltage space vector, by not determining an offset value in the bridge branch for which the expected time interval for a vanishing current sample value is too short to perform interference-free acquisition, and only determining an offset value when the mean output voltage space vector is correspondingly changed, i.e., when the time interval for a vanishing current sample value is then sufficiently long for interference-free acquisition. wherein the current space vector is determined from the current sampling values ​​determined in switched-on bridge branches, wherein the mean output voltage space vector is determined by the differences of the output potentials of the three output phases averaged over one pulse width modulation period, where - to control the half-bridges, a pulse pattern is generated within a time interval equal to one pulse width modulation period, such that at a first sampling time lying within the pulse width modulation period, at least one upper or one lower bridge branch, equipped with means for sampling the current, is switched off. wherein in a further switched-off bridge branch current sampling values ​​lying symmetrically before and after the first sampling time are recorded in order to determine the offset value.

Description

[0001] The invention relates to a converter and a method for determining a current space vector.

[0002] In a digitally controlled, pulse-width modulated (PWM) output voltage converter, achieving sufficient control accuracy requires determining a current space vector for each PWM period that corresponds to the average value of the actual current space vector calculated over that PWM period. With an inductive load, such as an electric motor, the PWM output voltage results in a current ripple. The current space vector must be determined without being affected by this current ripple. In some state-of-the-art converters, the current sensors are located in the motor leads, resulting in continuous current signals. These signals are filtered analogously to remove the current ripple and can then be used for control.

[0003] From the book “Practice of Field-Oriented Three-Phase Drive Controls” by Nguyen Phung Quang and Jörg-Andreas Dittrich, second edition, from 1999, pulse-width modulated inverters are known from pages 110-113, in which current sensing is carried out using three ( Fig. .1) or even just two ( Fig. .3, 5.5, 5.6, 5.9, 5.11) are provided with potential-isolating current sensors in the output branches. A current space vector with two degrees of freedom is determined from the current measurements ( Fig. .3, 5.5, 5.6, 5.9, 5.11). Furthermore, pages 76 to 77 explain that to detect the fundamental frequency of the motor currents without harmonics, i.e., without current ripple, the current measurements must be taken at suitable times. The deviation from the fundamental frequency is referred to as the differential current space vector. The current measurements are then taken at the zero crossing of the differential current space vector. A disadvantage of these known converters is that at least two expensive, galvanically isolated current measuring devices are required.

[0004] From DE 196 81 189 T1 is a pulse-width modulated converter with an intermediate circuit capacitor ( Fig. 1, reference numeral 3) and a connected output stage are known, which includes only a current sensor for detecting the intermediate circuit current. Depending on the angle of the output voltage phasor, the current detection thus provides information about a single phase current or sums of phase currents. A disadvantage is that, at certain angles of the output voltage phasor, determining a current space vector is not satisfactorily feasible.

[0005] From the publication by Francesco Parasiliti, "Low cost current sensing in DSP Based Drives," Industrial Electronics, 1999, ISIE'99, Proceedings of the IEEE, Volume 3, 1999, International Symposium, Volume 3, 1999, a converter is known in which shunt resistors are arranged in all lower branches of the half-bridges as a means of sensing the respective currents. All three current measurements are taken, with the measurement impulses occurring in the middle of the time interval within a pulse-width modulation period in which the discrete switching state (000), i.e., the zero vector v0, is present (page 1287, left column). Fig. 7) In the actual implementation, there is a slight time offset from this center point due to the consideration of dead times, signal propagation delays, and the like. The current space vector is then determined from the corresponding measured values. A disadvantage of this, however, is that in the shaded areas of the hexagon (page 1287, left column, last section and Fig. 8) No or only an inaccurate determination of the current space vector is possible, since either no zero vector is used in these shaded areas or only for a very short time. If the zero vector is not present or is only present for a very short time, then automatically no or only a very short current measurement signal is present in a phase, which is therefore not available for evaluation or only available inaccurately. This results in the following: in the case of an output voltage space vector averaged over a pulse width modulation period from the shaded area (page 1287, Fig. 8) at least one current measurement is falsified.

[0006] From US 5 815 391 A, a converter is known in which means for detecting the respective currents are arranged in all lower branches of the half-bridges ( Fig. 2A). In the corresponding table ( Fig. Section 2B) explains that the current in phase A is measurable depending on the switching states (for example, line 1: measurable;) or can be calculated from the two other current measurements (line 5: calculable;). However, if the lower switch of phase A and another lower switch are open, the current measurement of phase A cannot be determined. Such switching states are marked as "unknown" for phase A in this document.

[0007] The aforementioned Table 2B and the described problem relate to the switching states, i.e., the instantaneous state of the inverter. Table 2B refers only to phase A. However, corresponding tables can be compiled for phases B and C. It is readily apparent that for some switching states, namely (111), (110), (101), and (011), no current measurement or only one current measurement can be determined, and therefore a current space vector cannot be calculated for these instantaneous states. US 5,815,391 does not explain how to determine a current space vector that corresponds to the average value of the actual current space vector calculated over a pulse-width modulation period. In particular, it does not describe how to proceed so that the current space vector is not distorted by current ripple.

[0008] The duration of a "known" switching state within a pulse width modulation period depends on the following factors: Carrier frequency of pulse width modulation, The type of pulse width modulation method. If the duration of a "known" switching state is too short, the current space vector cannot be determined.

[0009] How the pulse-width modulation method is implemented is not explained in detail in this document; however, it is clear to those skilled in the art that the document teaches that these "unknown" states pose a problem. Column 1, lines 40-44, mentions a "backward calculation," but this is not described in a feasible manner. The document teaches those skilled in the art that the problem of the "unknown" states can be circumvented by following the procedure described below. Fig. 5 and Fig. 6. Measuring instruments, in particular potential-free measuring instruments such as Rogowski coils or the like, are used in both the upper and lower branches of the half-bridges. By cleverly combining the measurement signals, especially adding the measurement signals from the lower and upper measuring instruments, a continuous current measurement signal is achieved and the "unknown" states no longer pose a problem.

[0010] This principle from US 5,815,391 A, to use measuring instruments in both the upper and lower branches of half-bridges, would also be applicable to the aforementioned IEEE document. In this way, current measurement in the shaded areas would be easily and accurately feasible. However, this would require complex and therefore expensive measuring instruments, particularly those for overcoming the potential barrier. The solutions would thus be very complex and costly.

[0011] From DE 102 48 375 C2, a method for determining a current space vector is known as the closest prior art.

[0012] From DE 11 2006 000 786 T5 a current replacement correction method for a current control unit is known.

[0013] The invention is therefore based on the objective of further developing a converter in which as many costly parts as possible can be dispensed with and, in particular, cost-effective manufacturing can be achieved in mass production, and yet a method for determining a current space vector can be carried out.

[0014] According to the invention, the problem is solved in a method according to the features specified in claim 1 or 24 or 27 and in a converter according to the features specified in claim 28.

[0015] Important features of the invention in the method are that, for determining a current space vector for a pulse-width modulated inverter, comprising signal electronics and a power output stage, which includes power switches arranged in three half-bridges, each comprising a lower and an upper branch, where - to control the half-bridges, a pulse pattern is generated within a time interval equal to one pulse width modulation period, such that at a first sampling time lying within the pulse width modulation period, at least one upper or one lower bridge branch, equipped with means for sampling the current, is switched off. - in the switched-off bridge branch, at least one current sample value is determined using the means for sampling the current, and a current offset value for current detection is determined and taken into account from this, - the switched-off bridge branch is chosen differently depending on the average output voltage space vector and - the current space vector is determined from the current sampling values, whereby the mean output voltage space vector is determined by the differences of the output potentials of the three output phases averaged over a pulse width modulation period.

[0016] An advantage of this approach is that a current offset value is determined, thus enabling offset-free current measurement and resulting in improved control behavior of the inverter. Even in the event of temperature drift in the measuring amplifier's offset value, a consistently valid value is always determined. Since the drift in typical industrial operating conditions exhibits a characteristic time constant that is larger than the pulse width modulation period corresponding to a pulse width modulation frequency in the range of 1 kHz to 100 kHz, it is not necessary to determine an offset value in every pulse width modulation period; rather, it suffices to perform this at longer intervals. Furthermore, the invention eliminates the need for a sample-hold circuit, even though current measurements for multiple phases can be acquired and only one analog-to-digital converter is required.

[0017] In an advantageous design, - to control the half-bridges, a symmetrical pulse pattern is generated within a time interval equal to one pulse width modulation period, such that at a further sampling point located in the middle of the pulse width modulation period, one pair of the three upper or one pair of the three lower bridge branches, equipped with means for sampling the current, are switched on. - in each of the two switched-on bridge branches, at least one current sample value is determined using the means for sampling the current, - the pair of switched-on bridge branches is chosen differently depending on the mean output voltage space vector and - the current space vector is determined from the current sampling values, whereby the mean output voltage space vector is determined by the differences of the output potentials of the three output phases averaged over a pulse width modulation period.

[0018] Advantageously, to determine the current space vector, only the two most suitable of the three possible current measurements are selected for each pulse-width modulation period. This is based, in particular, on the surprising finding that the information about the current space vector is contained in the time course of the two most suitable current measurement signals. It is even permissible for switching states to occur temporarily within this pulse-width modulation period that, on their own, would not allow for the determination of the current space vector; these states are therefore considered "unknown" states in the sense of US 5,815,391 A.

[0019] Since, in the variant with current detection devices arranged in all three lower branches of the half-bridges, the phase in which the corresponding pulse-width modulation signal is shorter than a minimum duration in the LOW state is always excluded, reliable determination of the current space vector is ensured using a suitable pulse-width modulation method, in particular a symmetrical pulse-width modulation method. In the respective LOW state, the lower power switch is closed and the upper one is open. The opposite is true in a HIGH state.

[0020] Furthermore, as many costly components as possible can be eliminated, especially potential-isolating current sensors, since the currents are measured not in the output branches, but in the lower or upper branches of the half-bridges. Mass production allows for cost-effective manufacturing.

[0021] In particular, with two of the three current measurement methods, the corresponding current samples are not acquired simultaneously within the pulse width modulation period and / or the corresponding measurement impulses are not executed simultaneously. An advantage of this is that the inverter can use a microcontroller with only a single analog-to-digital converter and no additional sample-hold circuit, i.e., no external sample-hold circuit.

[0022] In particular, at least one of the two means is used to record a current sample value more than once per pulse width modulation period and / or to provide more than one measurement impulse within a pulse width modulation period.

[0023] In particular, a current sample is acquired exactly twice per pulse-width modulation period using the first of the two means, and the measurement trigger for the second of the two means lies in the middle of the pulse-width modulation period. Advantageously, the two measurement triggers of the first of the two means are equidistant from the middle of the pulse-width modulation period, meaning that the time interval between the first measurement trigger of the first means and the first measurement trigger of the second means is equal to the time interval between the measurement trigger of the second means and the second measurement trigger of the first means, with the latter measurement trigger of the second means lying in the middle of the pulse-width modulation period.

[0024] In the variant with current sensing devices arranged in all three lower branches of the half-bridges, the midpoint of the pulse-width modulation period corresponds to the midpoint of the respective LOW states. Current measurement signals are present at the current sensing devices during these states.

[0025] To determine the current space vector, the current measurements from the phases belonging to the two aforementioned means are sufficient. In the present invention, all three current measurements from the two means are always available because the invention employs an advantageous pulse-width modulation method, such as a symmetrical pulse-width modulation method, and the output voltage space vector is limited to the inscribed circle of the hexagon. This also means, for example, that no corner vectors of the hexagon occur continuously over an entire pulse-width modulation period.

[0026] To ensure the current space vector remains unaffected by the current ripple caused by pulse-width modulation, the measurement impulses must be triggered at suitable times. Ideally, the measurement impulses of both devices would be triggered simultaneously and in the middle of the pulse-width modulation period. However, this would require either two analog-to-digital converters or external sample-hold circuits, which would be costly. According to the invention, however, two measurement impulses are executed with the first device at different times. The measurement impulse of the second device is triggered in the middle of the pulse-width modulation period. The timing of both measurement impulses of the first device is the same time interval from the middle of the pulse-width modulation period.Thus, as a result of the averaging and because of the symmetry of the current ripple, which is symmetrical to the middle of the pulse width modulation period due to the symmetrical pulse width modulation method, a current measurement value is determined using the first mean, which is equal to the current measurement value that could be measured with the first mean at the time of the middle of the pulse width modulation period.

[0027] In particular, the current space vector is determined for each pulse width modulation period. An advantage of this is that even at low pulse width modulation frequencies, the current space vector can be determined as frequently as possible to improve the control performance of the inverter's control method.

[0028] In an advantageous embodiment, the first sampling point lies in the middle of the pulse-width modulation period or in the middle of a time interval during which the respective bridge branch is switched off. Advantageously, any current occurring in the preceding time interval has thus decayed and no residual currents remain.

[0029] In an advantageous embodiment, several initial sampling points occur within a single time interval, and the average of the acquired current measurements is used to determine the offset value. A key advantage is that a multiplexer allows the measured values ​​from only one analog-to-digital converter to be acquired, which sequentially acquires the current measurements from different bridge branches.

[0030] In an advantageous embodiment, current sampling values ​​are acquired in a further switched-off bridge branch in a time-symmetrical manner, both before and after the first sampling time, to determine the offset value. The advantage here is that the acquired values ​​for different switched-off bridge branches essentially correspond to a time-synchronous acquisition.

[0031] In preferred embodiments, the current measurement signals are generated at the lower DC link potential, with current values ​​assigned to the respective half-bridges being determined from current sampling values ​​derived from the current measurement signals. In particular, the reference potential of the signal electronics, which includes a control and regulation unit for the inverter, corresponds to the reference potential at which the current measurement signals are generated. An advantage of this is that optocouplers for galvanic isolation can be omitted. The reference potential of the signal electronics also corresponds to the lower DC link potential. A significant advantage is that the control signals for the lower power switches in the half-bridges can be generated by the signal electronics without a large voltage difference that would necessitate complex galvanic isolation.Only the control signals of the upper circuit breakers need to be controlled via optocouplers or other potential-isolating devices. Particularly with a converter that has no connections for an encoder, such as a speed or position encoder, or for other potential-isolating devices, this represents a significant step towards cost savings and a reduction in the number of parts.

[0032] Advantageously, power semiconductor switches, such as npn type IGBTs, can be used as lower power switches in the half-bridges, which can be controlled with control voltages that have the lower DC link potential as their reference potential.

[0033] When using complementary power semiconductor switches, such as IGBT type pnp, the upper DC link potential must be selected as the reference potential for current measurement and for the signal electronics, and the inverter must be designed accordingly.

[0034] In a preferred implementation, the pair is selected differently depending on the angle of the output voltage space vector and not on the pulse-width modulation pattern. An advantage of this is that the programmatic implementation is particularly simple.

[0035] In another implementation, a time interval of two or more pulse-width modulation periods is used instead of a single pulse-width modulation period. The advantage here is that the current space vector can be determined even at high switching frequencies. While this does introduce a slight distortion, it is minimal at high switching frequencies.

[0036] In a preferred embodiment of the invention, the acquisition of the current sample value acquired by the first current measuring device (i.e., the means for sensing the currents) is performed midway through the pulse-width modulation period. Specifically, a second of the two current measuring devices acquires a first and a second corresponding current sample value symmetrically before and after the acquisition of a current sample value acquired by the first current measuring device. An advantage of this is that current measurements can be determined with only one analog-to-digital converter without external sample-hold circuits, and these measurements are not distorted by current ripple caused by pulse-width modulation. Therefore, determining the current measurements is particularly simple and requires minimal computational effort.

[0037] By using additional external sample-hold circuits, the corresponding current samples could be acquired simultaneously using two of the three current acquisition methods, and / or the corresponding measurement impulses could be executed simultaneously. In particular, the measurement impulses would be executed in the middle of the pulse-width modulation period. The advantage of this approach is that determining the current measurements is particularly simple, yet there are no distortions caused by current ripple.

[0038] In another preferred embodiment of the invention, the first of the two current sensing means acquires an associated current sample value at least one pulse-width modulation period after acquiring a current sample value associated with the second current sensing means. In particular, the measurement impulses associated with the two current sensing means occur at different pulse-width modulation periods. An advantage of this is that even at high switching frequencies, a current space vector can be determined using only a single analog-to-digital converter and without additional sample-hold circuits, especially without distortions caused by current ripple.

[0039] In a preferred embodiment, the first current sample of the pair is acquired more than once per pulse-width modulation period, and an interpolated value and / or average is calculated from the acquired values ​​according to the times of the respective acquisitions. Specifically, a first current sample of the pair, assigned to a first half-bridge, is acquired before and after the second current sample of the pair, assigned to a second half-bridge. An advantage of this approach is that, through interpolation or averaging, a fictitious current measurement can be determined with good accuracy, which is available simultaneously with the other acquired current sample.

[0040] In a preferred embodiment, the measured current space vector corresponds to the mean value of the current space vector over one pulse-width modulation period. An advantage of this is that the current ripple resulting from the pulse-width modulation does not distort the result and therefore also the control procedures.

[0041] Essential features of the invention of the inverter, comprising signal electronics and a power output stage, which includes power switches arranged in three half-bridges, each comprising a lower and an upper branch, wherein the inverter is pulse-width modulated, are that - means for recording the respective currents are arranged either in all three lower branches or in all three upper branches of the half-bridges, - current measurement signals acquired by the three means of current acquisition are fed and / or can be fed to only a single analog-to-digital converter via a multiplexer.

[0042] The advantage here is that costly potential-separating means can be saved and the size of the inverter can be reduced.

[0043] In a preferred embodiment, the current sensing means comprise resistors, in particular shunt resistors. An advantage of this is that the current sensing is extremely cost-effective.

[0044] In a preferred embodiment, the current sensing means are arranged in the half-bridges such that they are connected to either the upper or the lower DC link potential. An advantage of this arrangement is that galvanic isolation means are eliminated. In a further preferred embodiment, the signal electronics have a reference potential that also serves as the reference potential for the current sensing means. If this reference potential is Uz-, then galvanic isolation means for the control signals of the lower circuit breakers of the half-bridges are eliminated. If this reference potential is Uz+, then galvanic isolation means for the control signals of the upper circuit breakers of the half-bridges are eliminated.

[0045] In a preferred embodiment, the signal electronics have a reference potential that also serves as the reference potential for the current sensing devices. This is advantageous because costly galvanic isolators can be omitted. If the current sensing devices were located in the motor's supply lines, costly galvanic isolators would be necessary.

[0046] In a preferred embodiment, any current sensing means can be assigned to a single analog-to-digital converter by means of a multiplexer and / or switch. An advantage of this is that the converter can be implemented cost-effectively, particularly by eliminating the need for additional analog-to-digital converters.

[0047] In principle, it would also be possible to assign more than one analog-to-digital converter (ADC) to the current sensing means. The advantage here would be that, with the use of appropriately arranged multiplexers, the selected pair of current sensing means could be assigned to, for example, two ADCs, thus eliminating the need for additional sample-hold circuits. Specifically, each current sensing means could be assigned its own ADC. The advantage here is that no additional sample-hold circuits are necessary and synchronous acquisition of the current samples can be ensured. Preferably, a sample-hold circuit could also be placed between each current sensing means and its assigned ADC. The advantage here would be that the current space vector could be determined without distortion from current ripple, and a single ADC would suffice.Preferably, the signal electronics could include a microcontroller with a single analog-to-digital converter and a sample-hold circuit not integrated into the microcontroller, i.e., an additional external component. The advantage here would be the use of a cost-effective microcontroller that only includes a single analog-to-digital converter.

[0048] In a preferred embodiment, the signal electronics comprise means for generating pulse-width modulated control signals for the circuit breakers, and the signal electronics have a reference potential that also serves as the reference potential for the current sensing means. An advantage of this is that means for potential isolation can be eliminated.

[0049] In an advantageous embodiment of the invention, a single analog-to-digital converter can be assigned a specific current sensing device by means of a multiplexer and / or switch. This eliminates the need for costly additional analog-to-digital converters.

[0050] In another advantageous embodiment of the method according to the invention, the essential features are that - means for recording the respective currents are arranged either in all three lower branches or in all three upper branches of the half-bridges, - the pulse width modulation frequency is greater than a minimum frequency, - within a time interval of two pulse width modulation periods, at least one corresponding current sample value is determined using any two of the three means of current detection, and the same mean output voltage space vector is output in both pulse width modulation periods, - a first current sample value is acquired with a first means of the pair in the middle of the first of the two pulse width modulation periods, and a corresponding current sample value is acquired with a second means of the pair in the middle of the second, i.e. immediately following, pulse width modulation period, - a current space vector is formed from the current sampling values ​​determined with this pair of means, or the current values ​​in the output branches are formed, in particular for use in a control and / or regulation procedure, and - this pair of means is selected differently depending on the mean output voltage space vector, - and where the mean output voltage space vector is determined by the differences of the output potentials of the three output phases averaged over one pulse width modulation period.

[0051] An advantage of this is that even at high pulse-width modulation frequencies, a single analog-to-digital converter is sufficient and no external sample-hold circuits are necessary. If the frequency falls below the minimum threshold, which is, for example, 10 kHz, the system can switch to the previously described method according to the invention. Thus, the advantages are present at all pulse-width modulation frequencies. Further advantageous embodiments are described in the dependent claims.

[0052] The invention will now be explained in more detail with the help of illustrations: In the Fig. Figure 1 shows a schematic diagram of a power output stage for a three-phase inverter. The power switches with associated freewheeling diodes are represented by six switch symbols S. Ro , S So , S To , S Ru S Su , and S Tu designated. In the lower branches of the half-bridges, shunt resistors R are used as a means of current detection.R , R S and

[0053] R T These are thus arranged with the lower reference potential U. Z_ connected, which is also simultaneously the reference potential of the signal electronics of the inverter comprising a control and regulation device 1. The motor-side output potentials of the inverter are U R , U S and U T ; the motor currents are denoted by I R , I S and I T designated.

[0054] The resistors are connected to amplifier circuits V R , V S and V T connected, each outputting a current measurement signal I RM , I SM and I TMThe three current measurement signals are used to generate current samples, which are then derived from analog-to-digital conversion. Current measurements are then generated from these samples. The current space vector is determined from these current measurements, with the half-bridges corresponding to the current measurements being selected based on the angle of the voltage space vector. Since the motor currents are determined according to Kirchhoff's law I R + I S + I TIf the current space vector is zero, it has two degrees of freedom. It therefore has two independent parameters. In principle, it is sufficient to use only the current measurements assigned to two branches to determine the current space vector. However, it must be noted that pulse-width modulation signals from the inverter do not have to be constant within a pulse-width modulation period; the switching states can change in such a way that switching states may occur which, on their own, would not allow for the determination of the current space vector!

[0055] For the control procedure of the inverter, it is sufficient to achieve adequate control quality by determining a current space vector as a calculated variable once per pulse width modulation period, which represents the mean value of the physical current space vector formed over one pulse width modulation period.

[0056] The pulse width modulation signals (PWM) generated by the control device 1 R (t), PWM S (t) and PWM T (t) determine the state of the circuit breakers S Ro , S So , S To , S Ru , S Su and S Tu The pulse width modulation signals are provided as follows: If the respective pulse width modulation signal, for example PWM, R (t), 1, the associated upper circuit breaker, i.e., for example S Ro , closed and the associated lower circuit breaker, for example S Ru , opened. In this case, the corresponding output voltage potential is U. Z+ If the pulse width modulation signal is 0, i.e., in the LOW state, the corresponding power switches are in the opposite state and the corresponding output voltage potential is U. Z_The so-called dead time included in the practical implementation, during which the upper and lower switches are open, is not relevant to the basic function of the invention.

[0057] When a pulse-width modulation signal belonging to a half-bridge is in the LOW state, the corresponding motor current flows in the lower branch of the associated half-bridge and thus through the respective shunt resistor. For accurate current measurements, the corresponding motor current must flow through the shunt resistor for a minimum duration. This minimum duration depends on the filtering effect of the measuring amplifier circuits, comprising amplifier circuits V. R , V S and V TA filtering effect is achieved through appropriate circuitry to suppress noise or interference components in the current measurements. For example, with a pulse width modulation frequency of 16 kHz, a filter time constant in the range of 0.5 µs to 2 µs is advantageous.

[0058] The present invention advantageously employs a so-called symmetrical pulse-width modulation (PWM) method. In such symmetrical PWM methods, the time-averaged value calculated from the time of the first switching state change of the PWM signal in a first phase and the time of the subsequent corresponding switching state change in the same phase has the same value as the correspondingly calculated averages of the other two phases. Therefore, the switching state changes from HIGH to LOW and back are symmetrical about the center of the PWM period in all three phases.

[0059] In this invention, the amplitude of the averaged output voltage space vector is limited to U2 / √3, so that a sinusoidal rotating voltage system can always be generated. In this invention, this limitation denotes the maximum output level. Overdriving, i.e., exceeding this operating range defined by the limitation, is always avoided.

[0060] The invention therefore relates only to those inverters that are operated with pulse-width modulation in such a way that no pulse-width modulation period results in such a high output level that only a single active switching state occurs for a full pulse-width modulation period. In such an active switching state, the pulse-width modulation signals (PWM) R (T), PWM S (t) and PWM T(t) the values ​​(110), (101), (100), (001), (010) or (011), which are also referred to as discrete active output voltage space vectors. Active switching states therefore do not include switching states (111) and (000). The latter two switching states are also referred to as discrete zero voltage space vectors or zero vectors.

[0061] So-called boundary vectors are the output voltage space vectors averaged over a pulse-width modulation period, which would theoretically result if a single active switching state were applied over an entire pulse-width modulation period. In the invention, the output voltage space vector averaged over a pulse-width modulation period never assumes the value of such a boundary vector due to the aforementioned limitation of the operating range.

[0062] The image shows the aforementioned IEEE document by Francesco Parasiliti, according to the one there. Fig. As discussed in section 8 on page 1287, the limitation of the output voltage space vector averaged over a pulse-width modulation period, as described above, means that it lies within the maximum incircle of the hexagon. The averaged output voltage space vector never assumes the value of a boundary vector, and therefore, in the invention, never lies at the corners of the hexagon.

[0063] Since the output voltage space vector is limited, various discrete switching states alternate. By carefully selecting the timing of the measurement impulses within a pulse-width modulation period, a current space vector can be determined for each pulse-width modulation period. The alternating states even include states for which determining a current space vector would not be possible if they were continuously present over a pulse-width modulation period, such as state (101) according to the last row of Table 2B from US 5,815,391. Specifically, for example, a measurement impulse is executed at a first time point in a first phase, a second measurement impulse at a second time point in a second phase, and a third measurement impulse at a third time point in the first phase.By averaging the first and third measured values, the determinable current space vector is not distorted by current ripple. This is achieved in the pulse-width modulation method permissible for the invention, particularly symmetrical pulse-width modulation methods, where the output voltage space vector is limited to the inner circle of the hexagon. Thus, only a single analog-to-digital converter is necessary, especially without the need for external sample-hold circuits.

[0064] In the Fig. 2 is a feasible variant for part of the control and regulating device 1 from Fig. Figure 1 shows a sample-hold circuit 21, which receives the three current measurement signals, and a microcontroller 22 with an integrated analog-to-digital converter. The microcontroller 22 controls the sampling times, also called measurement triggers, via the signals S / H1, S / H2, and S / H3. At these times, the sample-hold circuit 21 holds current measurement signals until the analog-to-digital conversion. The microcontroller 22 then only needs to include a single analog-to-digital converter along with an analog multiplexer or switch 23. The advantage of this is that simultaneous acquisition of current samples is achieved with only a single analog-to-digital converter, even though a sequential analog-to-digital conversion takes place. The microcontroller 22 then determines current measurements or a current space vector from the current samples.The microcontroller 22 controls the sample hold circuit via the signals S / H1, S / H2 and S / H3, and thus the measurement impulses according to the procedures described below.

[0065] In another embodiment according to the invention, according to Fig. 3 in contrast to the variant after Fig. 2. The sample-hold circuit 21 can be omitted. However, it is necessary to acquire the current sample values ​​sequentially according to the procedures described below.

[0066] In the Fig. Figure 4 shows exemplary time courses of the motor-side output potentials U, averaged over one pulse width modulation period. R , U S and U T of the inverter over one output voltage period, where the motor-side output potentials U R , U S and U TThe values ​​are shown in a normalized form, and the output voltage angle α extends over the range from 0 to 2π. According to the normalization, the potential value U corresponds to... Z_ the value -1 and U Z+ the value +1.

[0067] In the Fig. Figure 4 shows that the mean values ​​of the output potentials contain a third harmonic.

[0068] Further are in Fig. Four angular ranges 1 to 6 of the output voltage pointer are shown, in which different means for current detection are used according to the method according to the invention.

[0069] For an output voltage angle of α = π / 6, the mean value of the output potential U is S Zero. The mean value of the initial potential U. R lies close to the positive maximum value of 1. The mean value of the initial potential U Tlies close to the minimum value of -1. The mean values ​​of the output potentials therefore almost reach the maximum output level of the inverter at this output voltage angle and with the exemplary chosen amplitude.

[0070] In the Fig. 5 are for the initial potentials from Fig. 4 again at the output voltage angle α = π / 6 corresponding time profiles of the pulse width modulation signals PWM R (t), PWM S (t) and PWM T (t) sketched. From these, the control signals for the circuit breakers S are derived. Ro , S So , S To , S Ru , S Su and S Tu derived. If the respective pulse width modulation signal, for example PWM, R (T), 1, the associated upper circuit breaker, i.e., for example S Ro , closed and the associated lower circuit breaker, for example S Ru, opened. In this case, the corresponding output voltage potential is U. R = U Z+ If the pulse width modulation signal is 0, the corresponding power switches go to the opposite state and the corresponding output voltage potential is +U. Z+ The so-called dead time incorporated into the practical implementation, during which the upper and lower switches are open, is fundamentally irrelevant to the function of the invention. Fig. Figure 5 shows two pulse width modulation periods in the abscissa direction.

[0071] The output voltage potential U averaged over one pulse width modulation period S = 0 is according to Fig. 5 through a PWM S (t) - Signal reached that is 1 for 50% of the pulse width modulation period and 0 for 50% of the pulse width modulation period. The pulse width modulation ratios of the other pulse width modulation signals (PWM) R (t) and PWM T(t) are distributed according to their output voltage potential averaged over a pulse width modulation period.

[0072] In the Fig. 6 are the exemplary motor currents I corresponding to the two mentioned pulse width modulation periods. R (t), I S (t) and I T (t) is shown in the output branches. The current ripple resulting from the pulse-width modulated output potentials is also indicated.

[0073] In the Fig. Figure 7 is the idealized time course of the current measurement signals I corresponding to the two pulse width modulation periods mentioned. RM (t), I SM (t) and I TM (t) shown without the filtering effect of the amplifier circuits V R , V S and V TA current reading that deviates from zero only occurs during those time intervals when the pulse-width modulation (PWM) signal of the corresponding half-bridge is zero; that is, when the lower switch is closed. When the upper switch is closed, the motor current does not flow through the shunt resistor of the half-bridge, so during this time the current measurement signal is zero, regardless of the actual current. Therefore, the current measurement signals do not continuously represent the motor currents, but only those time intervals in which the PWM signals are zero. R (t), PWM S (t) and PWM T (t) are zero. In the shown curve, the current measurement signal I RM (t) is only different from zero for a very short period of time. At full output voltage, this period would disappear entirely. Fig. 7 It is therefore evident that determining a current space vector fails under such operating conditions if the current space vector is formed using current measurement signals that are only different from zero for such a short time as in Fig. 7 examples I RM (t) that they are significantly distorted by the filter effect at the time of acquisition, i.e., at the measurement trigger. However, a minimum filter effect is necessary to suppress measurement noise and interference signals.

[0074] Therefore, in the present invention, for each pulse-width modulation period, the "optimal" pair exhibiting the wider pulse-width modulation signals is selected from the three half-bridges; that is, only current measurement signals from these two specific half-bridges are used. The selection is made such that the current measurement signal whose lower switch remains closed for a shorter time within a given pulse-width modulation period than the other two lower switches is not used.

[0075] In the Fig. Figure 7 shows an operating state for two pulse width modulation periods, in which the selected pair I SM (t) and I TM (t). The times at which the current sampling values ​​are recorded, also referred to as measurement impulses, are indicated by jagged arrows.

[0076] According to the invention, two basic types of methods can be used.

[0077] For a first method, the measurement points are indicated by solid, not dashed, jagged arrows. The current sampling values ​​for the selected pair I are then recorded. SM (t) and I TM (t) simultaneously acquired. Several implementations are possible. In one implementation, each current sample value is generated by a separate analog-to-digital converter. In a second implementation, the three current measurement signals are processed according to Fig. Two signals are each fed to a sample-hold amplifier of the sample-hold circuit 21, with the sample-hold amplifiers switching to the hold state at the time of sampling. Thus, the analog-to-digital converter can sequentially convert the held current measurement signals using the multiplexer or switch. In a third embodiment, two analog-to-digital converters and corresponding switches are used. Further embodiments also allow for hybrid configurations.

[0078] The first method advantageously uses a symmetrical pulse-width modulation (PWM) technique, where the switching frequency is not higher than 20 kHz and the maximum modulation level described above is not exceeded. Thus, within each PWM period, there is always a pulse pattern in which a PWM signal with a LOW state shorter than the minimum duration occurs in at most one output branch. This means that both PWM signals are never simultaneously in the LOW state for less than the minimum duration in two phases. In the event that the LOW state of one of the PWM signals, for example, PWM, is R Even if (t) falls below the minimum duration, it is still ensured that the current space vector is determined correctly. This is because the two other pulse-width modulation signals, in this example PWM, S (t) and PWM T(t) do not fall below the minimum duration. Thus, with the present invention, a correct determination of the current space vector is always ensured, provided that, in the respective type of pulse width modulation method, it is ensured that at most only one pulse width modulation signal remains in the LOW state for a shorter duration than the minimum duration.

[0079] In a second method according to the invention, the measurement impulse lying in the middle of the pulse width modulation period is used for I TM The first method according to the invention is replaced by two measurement impulses, each with a time offset Δt. The first measurement impulse for I is performed sequentially. TM , then the one for I SM and then the second measurement impulse for I TM It is advantageous to use the measurement impulse for I SMto place in the middle of the pulse-width modulation period. The time offset Δt should be as small as possible, but larger than the analog-to-digital converter time. From the first and second current samples for I TM The average value is calculated and used to determine the current space vector. This average value corresponds to the fictitious current sampling value for I. TM , which simultaneously serves to record the current sampling value I SM would be detectable. Thus, the current space vector determined in this way is free from the influence of the current ripple. This is because, when using symmetrical pulse-width modulation methods, the current ripple also exhibits a point-symmetrical course relative to the respective current function value at t=0.5, around the middle of the pulse-width modulation period, as is also the case in Fig. 6 is recognizable as an example. Fig. 6. The midpoint of the first pulse width modulation period lies at t=0.5 and the midpoint of the second pulse width modulation period at t=1.5. Therefore, since the two current sampling values ​​I TM (t = 0.5 - Δt) and I TM (t = 0.5 + Δt) point-symmetric to the current function value I TM (t = 0.5) is the average of the two current sampling values ​​I. TM (t = 0.5 - Δt) and I TM (t = 0.5 + Δt) the current function value I TM (t = 0.5).

[0080] This method makes it possible to determine the current space vector without distortion from current ripple using only a single analog-to-digital converter. Since all current samples can be performed sequentially, additional sample-hold circuits are unnecessary. These additional sample-hold circuits are not the same as those that are standard features in conventional analog-to-digital converters.

[0081] The second method also advantageously employs a symmetrical pulse-width modulation (PWM) method, whereby the switching frequency is advantageously not higher than 20 kHz and the maximum modulation level described above is not exceeded. Thus, within each PWM period, a pulse pattern is always present in which a PWM signal whose LOW state lasts shorter than the minimum duration occurs in at most one output branch. In the PWM method used in the second method according to the invention, at least one PWM signal is always present whose LOW state lasts longer than the sum of the minimum duration and twice the time offset, i.e., 2*Δt.Thus, the current space vector is always correctly determined according to the present invention, provided that the pulse width modulation signals have a sufficient pulse width for the respective type of pulse width modulation method. While flat-top methods are generally symmetrical pulse width modulation methods, they are not advantageously suited for use with low output voltages.

[0082] In the exemplary embodiment according to Fig. Furthermore, in those time periods where, under idealized conditions, a vanishing current sample value would be expected, one or more current samples are recorded. Idealized conditions here mean that the switching behavior is assumed to be ideal and the settling time of the measuring amplifiers is neglected.

[0083] For example, in Fig. 7 showed that for phase R the measured value I RMoffis recorded at two time points, using the mean of the two recorded measurements, and for phase S the measured value I SMoff . In phase T, the time range for an interference-free acquisition of a current offset measurement is too short, and therefore no current offset measurement is acquired in those time periods in which, under non-idealized consideration, the output value of the measuring amplifier has not yet settled to the actual current offset measurement.

[0084] If, in subsequent pulse-width modulation periods, the voltage values ​​to be generated in the individual phases change, and thus a time interval occurs in phase T during which the current offset measurement for phase T can be determined, a current offset measurement is also recorded here. Particularly when generating a rotating field, a sufficiently current current offset measurement is recorded in each of the three phases over a corresponding period.

[0085] In this way, the offset of the individual measuring amplifiers in the sample-hold circuit can be determined. This offset is taken into account during current measurement by subtracting the current offset value from the measured value. This results in an improved measured value and thus a correspondingly improved motor control. Advantageously, this offset determination is possible online, i.e., during the ongoing operation of the control system. Only additional sampling of the phase current is performed, which does not cause any significant additional load on the control unit or slow down the control process. A further advantage is that any drift in the offset caused by temperature increases is compensated for.

[0086] The sampling time for the current offset measurement is at least 6 µs, preferably at least 8 µs, apart from the preceding time interval in which the corresponding bridge branch was switched on. This ensures that any oscillations have ceased and the current measurement signal has stabilized at the current offset value.

[0087] Preferably, the measured values ​​in phase R are recorded symmetrically around the central sampling point of phase S.

[0088] In a further embodiment of the invention, instead of the two measurements in phase R, only a single measured value is acquired, which is not acquired simultaneously with the measured value of phase S. This is because, unlike the current space vector acquisition described above, where a current ripple must be taken into account when selecting the sampling times, no current ripple is present when determining the current offset measured values, and therefore the sampling times for the current offset measured value are not restricted.

[0089] In continuing education, a moving average of some recently recorded measurements is used as the currently recorded offset value.

[0090] In the present invention, only pulse width modulation methods are used which have a switching state in the middle of the pulse width modulation period such that at least two lower switches are closed.

[0091] In a further embodiment of the invention, the current samples are not determined in the same pulse-width modulation period. Instead, the first current sample is acquired in a first pulse-width modulation period, and the second current sample is acquired in one of the subsequent pulse-width modulation periods. Again, only a single analog-to-digital converter is required without an additional sample-hold circuit. This is particularly advantageous at high switching frequencies and thus short pulse-width modulation periods. No distortion due to current ripple occurs. Only the change in the angle of the current space vector occurring during this time can cause distortion. The selection of the pair is carried out according to Fig. 8.

[0092] In Fig. Figure 4 shows angle ranges 1 to 6, corresponding to the different angle ranges of the output voltage phasor. Fig. In section 8, the pair to be used for each of these angle ranges is indicated by jagged arrows. For example, in angle range 1, only S and T are used, i.e., I. SM and I TM , to use. In angle range 2, only R and T are used, i.e., I. RM and I TM , to be used. In those angular ranges where no current measurement is performed in a given phase, the measurement of the respective offset value is provided. This is in Fig. 8 with entry I RMoff , I SMoff and I TMoff designated.

[0093] As described above, in the second method according to the invention, the measurement impulse for a current sample value of the first method according to the invention, which lies symmetrically in the middle of the pulse width modulation period, is replaced by two measurement impulses, each of which has a time offset Δt. Fig. Figure 9 shows how the measurement pulses are to be performed depending on the angular ranges of the output voltage space vector. It specifically shows in which half-bridge two measurement pulses are to be performed and in which half-bridge the single, centrally located measurement pulse is to be performed. Thus, in angular range 1, a first measurement pulse for I is performed. TM , then a single, centrally located measuring impulse for I SM and finally a second measurement impulse for I TM The measurement impulses have a time interval of Δt.

[0094] In angle range 3, a first measurement impulse for I therefore takes place. RM , then a single, centrally located measuring impulse for I TM and finally a second measurement impulse for I RM .

[0095] For each angular range, there is a phase with a solid flash, indicating a current sample taken in the center of that phase, and another phase with two dashed flashes, indicating a double current sample taken in that phase by the time offset Δt before and after the center point. The present invention uses only pulse-width modulation methods whose pulse pattern is always such that the lower switches are closed in the respective angular range of the phases marked with flashes for a period of time around the respective current samples. In the phase with two dashed flashes, the corresponding lower switch is closed from before the first measurement pulse until after the second measurement pulse.

[0096] A characteristic of the usable pulse width modulation methods is that at least at the time of the middle of the pulse width modulation period, two lower switches are always closed, with the one of these two lower switches that is assigned to the phase with the double current sampling remaining closed for at least a time interval 2Δt.

[0097] Again, in those angular ranges where no current measurement is performed in a given phase, the respective measurement of the offset value is provided. This is in Fig. 9 with entry I RMoff , I SMoff and I TMoff marked. Thus, the otherwise unused areas are utilized for determining the offset value of the current measurement, i.e., all the means for this purpose, such as measuring amplifiers, multiplexers and analog / digital converters.

[0098] In other embodiments according to the invention, Fig.7 the two current sampling values ​​of the current measurement signal I TM not symmetrical around the measurement impulse for capturing the current sample value of the current measurement signal I SM around, i.e. not with a time interval Δt before and after the detection of I SM , but with different time intervals. Then, instead of the average value, an interpolated value is calculated that takes the corresponding time intervals into account, although motor-characteristic parameters as well as the type and duration of the switching states must be considered during interpolation.

[0099] In other embodiments of the invention, more current samples are acquired instead of the two mentioned. Furthermore, each current sample can, in principle, be replaced by multiple current samples. Thus, a further reduction in measurement noise is foreseeable.

[0100] In further embodiments according to the invention, the time response of the amplifier circuits V R , V S and V T This is taken into account by delaying all measurement impulses by the filter time constant of the amplifier circuits. The filter time constant is less than half the minimum duration.

[0101] The method also works at pulse width modulation frequencies higher than 20 kHz. Reference symbol list 1 Control and regulating device 21 Sample Hold Circuit 22 microcontrollers 23 multiplexers

Claims

[1] Method for determining a current space vector and an offset of a current measurement, where the motor current of an inverter-fed electric motor is measured, with means arranged in the bridge branches of the inverter for scanning the currents flowing in the bridge branches, where the motor current space indicator is non-vanishing, wherein, in a bridge branch that is not energized at the time of measurement, the current detected by the means arranged in that bridge branch is recorded as an offset value for the current detection in that bridge branch, wherein the respective switched-off bridge branch, in which the respective current offset value is determined, is chosen differently depending on the average output voltage space vector, by not determining an offset value in the bridge branch for which the expected time interval for a vanishing current sample value is too short to perform interference-free acquisition, and only determining an offset value when the mean output voltage space vector is correspondingly changed, i.e., when the time interval for a vanishing current sample value is then sufficiently long for interference-free acquisition. wherein the current space vector is determined from the current sampling values ​​determined in switched-on bridge branches, wherein the mean output voltage space vector is determined by the differences of the output potentials of the three output phases averaged over one pulse width modulation period, where - to control the half-bridges, a pulse pattern is generated within a time interval equal to one pulse width modulation period, such that at a first sampling time lying within the pulse width modulation period, at least one upper or one lower bridge branch, equipped with means for sampling the current, is switched off. wherein in a further switched-off bridge branch current sampling values ​​lying symmetrically before and after the first sampling time are recorded in order to determine the offset value. [2] Method for determining a current space vector, especially for use in a tax and / or regulatory procedure, for a pulse-width modulated inverter comprising signal electronics, a power output stage with power switches arranged in three half-bridges, each having a lower and an upper branch, and means for sampling the respective currents arranged in the three upper or lower branches of the half-bridge, where - to control the half-bridges, a pulse pattern is generated within a time interval equal to one pulse width modulation period, such that at a first sampling time lying within the pulse width modulation period, at least one upper or one lower bridge branch, equipped with means for sampling the current, is switched off. - in or in at least one switched-off bridge branch, at least one current offset sampling value is determined using the means for sampling the current, and from this a current offset value for the current sensing associated with the bridge branch is determined and then taken into account, wherein the respective switched-off bridge branch, in which the respective current offset value is determined, is chosen differently depending on the average output voltage space vector, by not determining an offset value in the bridge branch for which the expected time interval for a vanishing current sample value is too short to perform interference-free acquisition, and only determining an offset value when the mean output voltage space vector is correspondingly changed, i.e., when the time interval for a vanishing current sample value is then sufficiently long for interference-free acquisition. wherein the current space vector is determined from the current sampling values ​​determined in switched-on bridge branches, wherein the mean output voltage space vector is determined by the differences of the output potentials of the three output phases averaged over one pulse width modulation period, wherein in a further switched-off bridge branch current sampling values ​​lying symmetrically before and after the first sampling time are recorded in order to determine the offset value. [3] Method according to claim 1, characterized by , that - to control the half-bridges, a symmetrical pulse pattern is generated within a time interval equal to one pulse width modulation period, such that at a further sampling point located in the middle of the pulse width modulation period, one pair of the three upper or one pair of the three lower bridge branches, equipped with means for sampling the current, are switched on. - in each of the two switched-on bridge branches, at least one current sample value is determined using the means for sampling the current, - the pair of switched-on bridge branches is chosen differently depending on the mean output voltage space vector and - the current space vector is determined from the current sampling values, whereby the mean output voltage space vector is determined by the differences of the output potentials of the three output phases averaged over a pulse width modulation period. [4] Method according to at least one of the preceding claims, characterized by , that the first sampling time lies in the middle of the pulse width modulation period or in the middle of a time interval during which the respective bridge branch is switched off. [5] Method according to at least one of the preceding claims, characterized bythat several initial sampling times lie within a time period and the mean of the recorded current measurements is used to determine the offset value. [6] Method according to at least one of claims 1 to 3, characterized by , that the first sampling time lies in the end region of the pulse width modulation period or in the end region of the time interval during which the respective bridge branch is switched off. [7] Method according to claim 6, characterized by , that the sampling times of the different phases are chosen at different times. [8] Method according to at least one of the preceding claims, characterized by , that with the means of the second bridge branch of the pair, a first and a second associated current sample value are provided temporally before and after the acquisition of a current sample value acquired with the means of the first bridge branch of the pair. [9] Method according to at least one of the preceding claims, characterized by , that the pair is selected differently depending on the angle of the output voltage space vector. [10] Method according to at least one of the preceding claims, characterized by that the current sampling values ​​are determined using means connected to the lower or upper intermediate circuit potential. [11] Method according to at least one of the preceding claims, characterized by , that a reference potential of the signal electronics corresponds to a reference potential on which the current measurement signals lie and / or the current sampling values ​​are recorded. [12] Method according to at least one of the preceding claims, characterized by , that the acquisition of a current sample value occurs in the middle of the state of the pulse width modulation period in which the power switch located in the same branch as the associated current measuring device is conductive. [13] Method according to at least one of claims 1 to 5, characterized by , that the acquisition of a current sample value is offset by the filter time constant to the middle of the state of the pulse width modulation period in which the power switch located in the same branch as the associated current measuring device is conductive. [14] Method according to at least one of the preceding claims, characterized by , that the second means of the pair is the means in which the longest conducting state of the three lower circuit breakers occurs during the respective pulse width modulation period, provided that the current measuring means are arranged in the three lower branches, and that the first means of the pair is the means in which the second longest lasting conducting state of the three lower power switches occurs during the respective pulse width modulation period. [15] Method according to at least one of claims 1 to 5, characterized by , that the second means of the pair is the means in which the longest lasting conducting state of the three upper circuit breakers occurs during the respective pulse width modulation period, provided that the current measuring means are arranged in the three upper branches, and that the first means of the pair is the means in which the second longest lasting conducting state of the three upper power switches occurs during the respective pulse width modulation period. [16] Method according to at least one of the preceding claims, characterized by , that the measurement impulses for recording the current sampling values ​​belonging to a mean within a pulse width modulation period each have the same time interval from the center of the pulse width modulation period. [17] Method according to at least one of the preceding claims, characterized by that both methods capture a current sample value more than once per pulse width modulation period and / or that more than one measurement impulse is provided within a pulse width modulation period. [18] Method according to at least one of the preceding claims, characterized by , that from the recorded current sampling values ​​an interpolated value and / or average value is formed according to the times of the respective recording as a current measurement value to determine the current space indicator. [19] Method according to at least one of the preceding claims, characterized by , that the first and second current samples of the second means of the pair are provided for simultaneously to capture the captured current sample, in particular replaced by a single sample. [20] Converter for carrying out the method according to any of the preceding claims, comprising a signal electronics and a power output stage, which includes power switches arranged in three half-bridges, each comprising a lower and an upper branch, where the inverter is pulse width modulated, characterized by , that - means for recording the respective currents are arranged either in all three lower branches or in all three upper branches of the half-bridges and - current measurement signals acquired by the three means of current acquisition are fed and / or can be fed to only a single analog-to-digital converter via a multiplexer (23). - where the means for recording the currents are also intended as means for determining the offset values ​​of the current recording assigned to the respective bridge branches. [21] Inverter according to claim 20, characterized by , that the means for current sensing include resistors, in particular shunt resistors and / or The means for current detection are arranged in the half-bridges in such a way that they are connected either to the upper or to the lower intermediate circuit potential. [22] Inverter according to at least one of the preceding claims, characterized by , that the signal electronics include means for generating pulse-width modulated control signals for the circuit breakers. [23] Converter according to at least one of the preceding claims, characterized by that the signal electronics have a reference potential that is also the reference potential for the means of current detection.

Images