DE102019112823A1 - Method for providing a precision multi-level converter - Google Patents

Abstract

The invention relates to a method for providing a precision multi-level converter which is configured to output a total output voltage in a high-voltage system, the precision multi-level converter being implemented by a modular multi-level converter with serial and parallel connectivity, the modular multi-level converter being formed by a plurality of modules, with a respective module has two inputs and two outputs, at least four power transistors and at least one energy storage device, the modules being arranged in at least one two-wire string, with the power transistors either blocking or passing a current in at least one module of the at least one string and thus a digital one Module is formed, wherein the power transistors are operated resistively in at least one further module of the at least one string and thereby an analog module is formed, the at least a digital module arranged per line forms a quantized voltage level and the at least one analog module arranged per line simulates a remaining voltage difference to a reference voltage. Furthermore, the precision multi-level converter and a high-voltage system are claimed.

Description

The present invention relates to a method for providing a precision multi-level converter which combines digital and analog modules in strings. Furthermore, the precision multi-level converter provided by the method is claimed.

An AC motor, such as that used as a traction motor in an electric vehicle, requires at least one converter that converts the DC voltage from a DC voltage source, for example a battery, into an AC voltage. Conventional converters use so-called bridge circuits, i. H. Circuits that alternately connect the output terminals to the positive and negative pole (more rarely also to a zero position) of the DC voltage source and select the dwell time in each state so that a desired AC voltage is created on average over time (so-called pulse width modulation or other switching modulation methods). The disadvantage here is a very poor voltage quality and distortion. EMC problems too, i. H. Electromagnetic interference from radiation in electrical components and energy losses come to light.

Another problem is that a desired continuous voltage or current profile is never achieved by a quantized simulation with voltage levels. Instead, only a curve that is similar in terms of time average is generated. An unavoidable deviation is usually referred to as a distortion and is often determined in the frequency domain (as the distortion spectrum), since many types of modulation, i.e. forms for quantizing a continuous course, even use regular switching patterns and therefore show harmonic components in the distortion spectrum.

For various power electronics applications, however, such distortion caused by switched power electronics is not acceptable. For example, electrical machines or motors for drives and other applications on test benches are usually tested on test benches using power electronics that form an inverter. In the event of unexpected or otherwise unsatisfactory behavior, however, when testing the electrical machine with the inverter, the source of the behavior cannot be identified if the power electronics used have distortions and, in particular, switching harmonics. Although distortions in current or voltage usually do not contribute to an active power, they do generate reactive currents which, in addition to the expected ohmic losses in the machine, can cause additional high-frequency effects in the winding, iron and permanent magnets. Due to the non-linearity of various effects, the distortion can lead to significantly different behavior, increase the losses disproportionately or excessively heat components (e.g. magnets), which should experience a low thermal load according to the respective machine design. To identify power losses in electrical machines, ideal, distortion-free inverters are therefore required, which are not available with the current state of the art.

Furthermore, a test inverter should allow any other inverter to be emulated and, for example, to simulate their distortions (usually starting at 5-15 kHz) in order to test the different behavior of the machine in combination with several different inverter candidates. Other areas of application for distortion-free inverters are speaker driver circuits and power sources for materials testing and equipment test systems, as well as some scientific sources.

The prior art provides a large number of universal or ideal inverters, for example multilevel converters with a large number of voltage levels and high switching speeds, multi-phase inverters and fast-switching inverters with downstream low-pass filters. On the one hand, corresponding circuits continue to generate distortions, even if they are reduced, on the other hand, filters lead to increased losses and / or phase shifts in the output current and / or voltage.

Such a multilevel converter is that of R. Marquardt, for example in the publication US 2018/0109202 disclosed modular multilevel converter, also referred to as MMC or M2C for short. The so-called Marquardt macro topology implemented there uses strings assigned to a respective alternating current phase, which have at least two modules and, strictly speaking, only have to represent one voltage polarity. However, such a modular multilevel converter also has significant distortions, that is to say deviations from the desired reference signal (for example a voltage or current curve), which is mostly present as a reference voltage.

Typical circuits in power electronics use semiconductors almost exclusively as switches and avoid lossy resistive or linear operation. They just create quantized ones Output voltages (in current-controlled circuits, typically with thyristors, e.g. in rail drives according to the current). Intermediate levels are generated on average over time through rapid switching (switching modulation, switch-mode power supplies as an English catchphrase) between adjacent levels.

Building on this, in the above-mentioned modular multilevel converter, as in the document EP 2 928 055 A1 described, deviations between the quantized levels and the reference voltage can be reduced by individual linearly operated modules. However, the modules implemented there are limited to a serial connection.

The latter limitation is overcome by an MMSPC, abbreviated for modular multilevel converter with serial and parallel module connectivity, which is described, for example, in “Goetz, SM; Peterchev, AV; Weyh, T., "Modular Multilevel Converter With Series and Parallel Module Connectivity: Topology and Control," Power Electronics, IEEE Transactions on, vol.30, no.1, pp.203,215, 2015. doi: 10.1109 / TPEL.2014.2310225 .

Against this background, it is an object of the present invention to provide a method for providing a precision multi-level converter in which distortions with respect to the reference voltage are avoided. It should be possible to generate both a desired AC voltage and a DC voltage with high precision, which should extend with high power in a wide frequency range. Furthermore, a precision multi-level converter provided by the method is to be presented.

To achieve the above-mentioned object, a method for providing a precision multi-level converter is proposed, the precision multi-level converter being configured to output a total output voltage in a high-voltage system. The precision multi-level converter is implemented by a modular multi-level converter with serial and parallel connectivity, the modular multi-level converter being formed by a plurality of modules. Each module has two inputs and two outputs, at least four power transistors and at least one energy store. Through the at least four power transistors, the at least one energy store of a respective module is connected either in series-plus or series-minus or in parallel or not at all electrically to at least one energy store of a directly adjacent module. The modules are arranged in at least one two-wire line, the respective lines forming a respective common connection at their first two-wire end. These respective common connections can either be connected to a common neutral point or a double neutral point. At their respective second end of the respective strings, these form a respective pole of the total output voltage. In the case of at least one module of the at least one string in which the power transistors either block or pass a current, a digital module is thus formed. In the case of at least one further module of the at least one string in which the power transistors are operated resistively, an analog module is thereby formed. The at least one digital module arranged per string forms a quantized voltage level and the at least one analog module arranged per string simulates a remaining difference to a reference voltage.

A modular multilevel converter with serial and parallel connectivity, abbreviated to MMSPC and, for example, described in the publication DE 10 2015 112 512 A1 , differs e.g. B. from a conventional modular multilevel converter in that an additional parallel state exists. An MMSPC can generate almost any electrical serial-parallel circuit configuration of module-integrated energy storage devices in one line and dynamically change it. The at least one energy store of the respective module is connected in series-plus if it is in series with a passage of the current in the respective module and increases an output voltage of the respective module compared to an input voltage. It is connected in series minus if it is in series with the transmission of the current in the respective module and the output voltage of the respective module is lower than the input voltage. The at least one energy store of the respective module is connected in parallel if it is located in the respective module parallel to the passage of the current. Finally, a bypass state, also referred to as a bypass, is usually available in order to bypass the at least one energy store of a module. Depending on the embodiment, this bypassing state also allows a polarity reversal between the two inputs and the two outputs.

While the at least one digital module in the respective string creates a base as a digital output voltage - for example, the quantized next to a string voltage required by the reference voltage, i.e. with a total of the digital modules arranged in a respective string voltage level that can be generated - the analog module only needs to generate the (positive or negative) difference between the quantized digital output voltage and the desired reference voltage. In this way, significantly lower losses are advantageously possible than with a conventional amplifier. The at least one analog module thus represents a type of correction stage for the quantized voltage stage formed by the at least one digital module, which only roughly approximates the reference voltage. The solution is therefore based on the concept of the multilevel converter or a multilevel converter in order to reduce distortion, and supplements it with a correction stage in order to eliminate this reduced distortion.

If the precision multi-level converter provided by the method according to the invention has, for example, N digital modules and M analog modules in a respective string, where N and M each denote whole positive numbers, then 2N + 1 voltage levels can be implemented dynamically as digital output voltage. As a linearly controllable and adjustable correction stage, the M analog modules form the respective remaining voltage difference to the reference voltage. In general, the number of modules per line can also vary. It is also conceivable to combine different modules or to operate a respective power transistor on the one hand digitally or on the other hand in the linear range.

In one embodiment of the method according to the invention, a phase of an alternating voltage is formed by the respective pole of a strand.

In a further embodiment of the method according to the invention, an adjustable DC voltage is output by connecting individual modules in parallel in a string. The voltage level of the direct voltage is set using the reference voltage. Due to the possibility of parallel connection of the at least one energy store of the respective module, this embodiment can even output direct voltage and / or direct current.

In yet another embodiment of the method according to the invention, a capacitor is selected as the respective energy store.

In a continued further embodiment of the method according to the invention, the respective capacitor of the respective digital module of the at least one strand is recharged either by a parallel connection with charged capacitors of adjacent modules of the at least one strand or by a series connection of all capacitors of the modules of the at least one strand.

The parallel mode, e.g. also between digital and analog modules, allows charge to be transported from module to module. In particular, charge can be transported from the digital modules to the analog module or modules in this way, since the analog modules, due to their greater losses (due to the resistance operation of semiconductors, which generate voltage differences accordingly, to the desired output voltage or -strom) can also be discharged much faster and require additional energy.

The energy store of the at least one analog module discharges significantly faster than the energy store of a digital module due to the high losses in resistance operation and must be supplied with energy accordingly. The capacitor of the at least one analog module, like the respective digital module, can be recharged via the load current through the respective string as well as by regular parallel connection with charged neighboring modules. For a high charging throughput, parallel charging is preferably carried out with fully connected power transistors on the parallel side. Reloading from the load current through the string takes place in series.

In a further embodiment of the method according to the invention, at least one module of the modular multilevel converter comprises eight power transistors.

In yet another embodiment of the method according to the invention, at least one power transistor in at least one digital module is formed by a MOSFET.

In a further embodiment of the method according to the invention, at least one power transistor is formed in at least one analog module by a bipolar transistor or a field-effect transistor. Both transistor types have a wide resistance operation. Because they are used in amplifier circuits, they are also referred to as amplifier bipolar transistors or amplifier field effect transistors.

The power transistors are operated, for example, in complementary stages, preferably composed of a high-side n-channel transistor or a high-side npn transistor and a low-side p-channel transistor or a low-side pnp transistor Transistor in which both transistors are predominantly both operated in resistance mode and the voltage of the electrical node connecting the two transistors is predominantly in the range between 1/5 and 4/5 between the lower and the upper supply voltage of the bridge. Such drivers can, for example, also include operational amplifiers (which, after an input-side differential stage, often use a complementary output stage themselves).

This is in contrast to a half-bridge output stage that is usual with binary switching gate drivers, either consisting of two n-channel transistors or npn transistors connected in series (controlled by complementary binary control voltages or control currents) or consisting of a low-side n- Channel or low-side npn transistor and a high-side p-channel or high-side pnp transistor, with both low and high-side transistors by only the threshold voltage (or double this) shifted control voltage, whereby alternately only one of the two transistors (with maximum conductivity) is switched on, while the other blocks.

In yet another embodiment of the method according to the invention, a respective module is implemented as a two-quadrant module or as a four-quadrant module in a respective topology. This embodiment is shown in 2 further explained.

The strings resulting from the common arrangement of at least one digital module and at least one analog module can be combined in various macro topologies of the MMSPC. Thus, in yet another embodiment of the method according to the invention, a plurality of strands are arranged in a so-called Marquardt macro topology as half bridges or in a so-called matrix macro topology. In the Marquardt macro topology, for example described in the publication US 2018/0109202 , for example, direct voltage of the energy store can be converted into a two-, three- or multi-phase voltage or into a two-, three- or multi-phase current. Another example is the MMSPC, in which two series-connected strings are generated for each output phase to be generated, which in turn are connected at the ends to which direct voltage is preferably applied for generation. Alternatively, the strands can also be arranged in the matrix macro topology in order to convert single-phase or multiphase alternating voltage of a first type of properties, e.g. amplitude, frequency component, phase, etc., into a single-phase or multiphase alternating voltage of a second type of changed properties. For example, a matrix topology is known from the prior art, in which a respective string is positioned from each of at least two inputs to each of the at least two outputs, with any DC and AC voltages being applied or generated at the inputs and outputs can.

In yet another embodiment of the method according to the invention, a voltage level of the total output voltage of the precision multi-level converter predetermined by the reference voltage is generated by breaking down the predetermined voltage level into a nearest quantized voltage level formed by the at least one digital module arranged per string and a voltage level formed by the at least one analog module remaining differential voltage divided, it happens that the amount of the nearest quantized voltage level is greater than the voltage level specified by the reference voltage and to be realized in a respective string. In such a case, a negative connectivity is switched in the respective analog module of the respective line with serial minus. The MMSPC used according to the invention advantageously has this serial-minus connection option.

The total output voltage is generated by breaking down the reference voltage, for example from a regulation, into quantized steps for the at least one digital module and into a continuous signal for the at least one analog module. The more digital modules are arranged, the higher the number of intermediate levels, which is preferably greater than ten, but is also conceivable with greater than 50. This division is advantageously carried out by generating a base amount by rounding it to an adjacent quantization level (next higher, next lower or absolutely closest) and then generating the remainder, which is formed by the at least one analog module, e.g. in equal parts.

A precision multi-level converter is also claimed, which the method according to the invention provides. This reveals an almost distortion-free power inverter that uses common power transistors in one, two, three or more outputs or phases to generate almost ideal voltage and / or current curves with high power over a wide frequency range (e.g. from 0 Hz to 5 kHz) can. In particular, the precision multi-level converter according to the invention can even permit the generation of high spectral components (for example up to 50 kHz, preferably up to over 100 kHz) in order to be able to simulate, for example, distortions of different switching inverters.

Finally, a high-voltage system is claimed which comprises an energy reservoir, a control unit and a precision multi-level converter according to the invention, at least one energy store of the precision multi-level converter recharging from the energy reservoir, the control unit being configured to provide a reference voltage which comprises either a DC voltage signal or an AC voltage signal, wherein the control unit is additionally configured to operate a respective digital module and a respective analog module as intended, and wherein the high-voltage system is configured to output a total output voltage according to a specification by the reference voltage.

The high-voltage system according to the invention is particularly suitable for generating three-phase current with several phases in fixed phase relationships in order to provide low-distortion alternating voltage for electrical machines and networks. In addition, a high degree of utilization is advantageously achieved; H. the high-voltage system generates a maximum possible output voltage amplitude for a given input voltage and comparatively inexpensive components.

While space vector modulation is mostly used in the prior art for this field of activity in the quantized area, the precision multi-level converter according to the invention can be used to generate an equivalent of analog-digital space vector modulation. Similar to the conventional space vector modulation to multilevel converters, the closest states of a desired vector are determined according to an instantaneous vector peak. However, instead of switching between several neighbors (usually three) with a temporal dwell time in each neighbor, as in conventional switching modulation, so that a desired vector is created in a temporal average, only the closest state is selected, output as a quantized voltage level and the differential voltage through the at least one Analog module generated, which prevents switching distortion.

Further advantages and configurations of the invention emerge from the description and the accompanying drawings.

It goes without saying that the features mentioned above and those yet to be explained below can be used not only in the respectively specified combination, but also in other combinations or on their own, without departing from the scope of the present invention.

• 1 zeigt einen Strang von Modulen in einer Ausgestaltung eines erfindungsgemäßen Präzisionsmultilevelkonverters.
• 2 zeigt Module aus dem Stand der Technik als Beispiele für Modulimplementierungen in einer Ausführungsform des erfindungsgemäßen Präzisionsmultilevelkonverters.
• 3 zeigt eine Zerlegung eines Referenzsignals durch Module einer weiteren Ausführungsform des erfindungsgemäßen Präzisionsmultilevelkonverters.
• 4 zeigt ein Schaltbild zur Zerlegung des Referenzsignals.

The figures are described coherently and comprehensively; the same components are assigned the same reference symbols.

• 1 shows a string of modules in an embodiment of a precision multilevel converter according to the invention.
• 2 shows modules from the prior art as examples of module implementations in an embodiment of the precision multi-level converter according to the invention.
• 3 shows a decomposition of a reference signal by modules of a further embodiment of the precision multi-level converter according to the invention.
• 4th shows a circuit diagram for decomposing the reference signal.

In 1 becomes a strand 100 of modules 110 , 120 shown in one embodiment of a precision multi-level converter according to the invention. The two-core strand 100 is at its first end 101 interconnected to a string input connection, which in the precision multi-level converter according to the invention can be connected, for example, to further strings to a common neutral point. At a second end 102 is the two-core strand 100 joined together to provide a total output voltage 102 to provide. Both a digital module 110 as well as an analog module 120 has a first entrance 103 , a second entrance 104 , a first exit 105 and a second exit 106 . In the digital module 110 a respective power transistor is only switched between an “on” and an “off” state and is therefore either conductive or non-conductive. In the analog module 120 a respective power transistor is modulated in the intermediate linear mode (with a gate voltage between a threshold voltage and a fully switched control voltage). A plurality N of digital modules 110 allows 2N + 1 voltage levels, while the at least one analog module 120 applies a differential voltage to a reference voltage.

In 2 will be two modules 210 , 220 from the prior art shown as examples of module implementations in an embodiment of the precision multi-level converter according to the invention. A four-quadrant module 210 , e.g. described in “Goetz, SM; Peterchev, AV; Weyh, T., "Modular Multilevel Converter With Series and Parallel Module Connectivity: Topology and Control," Power Electronics, IEEE Transactions on, vol.30, no.1, pp.203,215, 2015. doi: 10.1109 / TPEL.2014.2310225 , comprises a total of eight power transistors “T1” 211, “T2” 212, “T3” 213, “T4” 214, “T5” 215, “T6” 216, “T7” 217, “T8” 218 and a capacitor 219 as an energy storage device with a voltage potential 201 can be interconnected in the switching states listed in Table 1. Table 1: Switching states for a four-quadrant module with eight power transistors as switches Switching status Activated switches serial plus left T2 & T4 serial plus right T5 & T6 serial minus left T1 & T3 serial minus right T6 & T8 parallel left T1 & T4 parallel right T6 & T7 Bypass left T2 & T4 Bypass right T6 & T8 Zero left none Zero right none

The four-quadrant module 210 can be used both as a digital module with either conducting or non-conducting power transistors, and as an analog module with power transistors that are at least partially no longer used as switches, but as controllable resistors 211 , 212 , 213 , 214 , 215 , 216 , 217 , 218 operate. When used as an analog module, the two outputs 105 and 106 or at the two entrances 103 and 104 Continuous voltages with both polarities and in both current directions are generated, with at least bridges on one side being in a complementary amplifier mode in the case of serial states, while the other side can either be switched in binary mode (either high-side transistors 215 , 217 turned on and low-side transistors 216 , 218 switched off or vice versa) or is also activated in an intermediate state. In the case of parallel states, whether they are double-sided or only one-sided, control is the same as with the digital module, i.e. two bridges on one side are controlled inversely so that one output is temporarily connected to the positive and an output to the negative connection of the Capacitor 219 connected is. The other side can be operated in serial resistance mode. The bypass status is the same as with the digital module.

A two-quadrant module 220 , for example. Described in the publication DE 10 2016 112 250.5 , comprises a total of four power transistors “T1” 221, “T2” 222, “T3” 223, “T4” 224, and a capacitor 229 as an energy storage device with a voltage potential 202 can be interconnected in the switching states listed in Table 2. In this embodiment only one polarity is possible. Table 2: Switching states for a two-quadrant module with four power transistors as switches Switching status Activated switches serial left T2 serial right T3 parallel left T1 parallel right T4 Bypass left none Bypass right T4 zero none

Also the two-quadrant module 220 can be used both as a digital module with either conducting or non-conducting power transistors, and as an analog module with power transistors that are at least partially no longer used as switches, but as controllable resistors 221 , 222 , 223 , 224 operate. When used as an analog module, the states are similar to those of the analog four-quadrant module, except that only half of the power transistors are available and therefore only voltages in one polarity direction via the two-quadrant module 220 can be generated away.

In 3 becomes a decomposition 300 a reference signal 310 shown by modules of a further embodiment of the precision multi-level converter according to the invention. A timeline 302 runs to the right and has a stress axis 304 up. At a given point in time, that is determined by the reference voltage 312 specified voltage level of the total output voltage of the precision multi-level converter through the digital modules arranged per string as the quantized voltage level closest at the given point in time 322 as the output signal of the digital modules 320 educated. An additional implementation of switching modulation is also conceivable. The at least one analog module forms a remaining differential voltage as a correction, which preferably comprises only one voltage level 332 as output signal 330 .

In 4th becomes a schematic 400 to decompose the reference signal 402 and to the formation of a multilevel converter control signal 404 shown. The reference input signal 310 becomes a quantizer 410 which outputs the quantized voltage levels that are used to control the digital modules and the digital module output voltage 320 to lead. The differential voltage 330 between reference input signal 310 and the quantized pedestal signal 406 receives the correction level 412 , which the continuous control signal to the at least one analog module for forming the analog module signal 330 supplies.

QUOTES INCLUDED IN THE DESCRIPTION

This list of the documents listed by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent literature cited

• US 2018/0109202 [0007, 0028]
• EP 2928055 A1 [0009]
• DE 102015112512 A1 [0013]
• DE 102016112250 [0041]

Non-patent literature cited

• “Goetz, S.M .; Peterchev, A.V .; Weyh, T., "Modular Multilevel Converter With Series and Parallel Module Connectivity: Topology and Control," Power Electronics, IEEE Transactions on, vol.30, no.1, pp.203,215, 2015. doi: 10.1109 / TPEL.2014.2310225 [ 0010, 0039]

Claims (13)

A method for providing a precision multi-level converter (100) which is configured to output a total output voltage (102) in a high-voltage system, the precision multi-level converter (100) being implemented by a modular multi-level converter (100) with serial and parallel connectivity, the modular multi-level converter ( 100) is formed by a plurality of modules (110, 120, 210, 220), with a respective module (110, 120, 210, 220) having two inputs (103, 104) and two outputs (105, 106), at least four Has power transistors (211, 212, 213, 214, 215, 216, 217, 218, 221, 222, 223, 224) and at least one energy store (219, 229), the at least four power transistors (211, 212, 213, 214, 215, 216, 217, 218, 221, 222, 223, 224) of the at least one energy store (219, 229) of a respective module (110, 120, 210, 220) either serial-plus or serial-minus or parallel or not at all electric with de m at least one energy store of a directly adjacent module (110, 120, 210, 220) is interconnected, the modules (110, 120, 210, 220) being arranged in at least one two-wire line, in which the respective lines at their first two-wire end (101) form a common connection and at their respective second end (102) form a respective pole of the total output voltage (102), with the power transistors (211, 212, 213) in at least one module (110, 210, 220) of the at least one string , 214, 215, 216, 217, 218, 221, 222, 223, 224) either block or pass a current and thus a digital module (110) is formed, with at least one further module (120, 210, 220) of the at least one string, the power transistors (211, 212, 213, 214, 215, 216, 217, 218, 221, 222, 223, 224) are operated resistively and thereby an analog module (120) is formed, the at least one a digital module (110) arranged per line e forms a quantized voltage stage (322) and the at least one analog module (120) arranged per string simulates a remaining differential voltage (332) to a reference voltage (312). Procedure according to Claim 1 , in which a phase of an alternating voltage is formed by the respective pole of a strand. Procedure according to Claim 1 , in which an adjustable DC voltage is output by connecting individual modules in parallel in a string. Method according to one of the preceding claims, in which a capacitor (219, 229) is selected as the respective energy store. Procedure according to Claim 4 , in which the respective capacitor (219, 229) of the respective digital module (110) of the at least one string either by a parallel connection with charged capacitors of adjacent modules of the at least one string or by a series connection of all capacitors (219, 229) of the modules ( 110, 120) of the at least one strand is reloaded. Method according to one of the preceding claims, in which eight power transistors (211, 212, 213, 214, 215, 216, 217, 218) are included in at least one module of the modular multilevel converter (100). Method according to one of the preceding claims, in which at least one power transistor (211, 212, 213, 214, 215, 216, 217, 218, 221, 222, 223, 224) is formed in at least one digital module by a MOSFET. Method according to one of the preceding claims, in which at least one power transistor (211, 212, 213, 214, 215, 216, 217, 218, 221, 222, 223, 224) in at least one analog module by a bipolar transistor or a Field-effect transistor is formed. Method according to one of the preceding claims, in which in a respective topology a respective module is implemented as a two-quadrant module (220) or as a four-quadrant module (210). Method according to one of the preceding claims, in which a plurality of strands are arranged in a so-called Marquardt macro topology as half bridges or in what is known as a matrix macro topology. Method according to one of the preceding claims, in which a voltage level of the total output voltage (102) of the precision multi-level converter (100) predetermined by the reference voltage (312) is achieved by breaking down the predetermined voltage level into the at least one arranged per strand a digital module (110) formed nearest quantized voltage level (322) and the remaining differential voltage (332) formed by the at least one analog module (120) is divided, the closest quantized voltage level (322) being greater in magnitude than that by the Reference voltage (312) predetermined voltage level to be realized in a respective string, and in which at least one analog module (120) of the respective string is connected with a serial minus negative connectivity. Precision multi-level converter (100) which is provided by a method according to one of the preceding claims. High-voltage system, which includes an energy reservoir, a control unit and a precision multi-level converter (100) Claim 12 comprises, wherein at least one energy store (219, 229) of the precision multi-level converter (100) recharges from the energy reservoir, the control unit being configured to provide a reference voltage (312) which comprises either a DC voltage signal or an AC voltage signal, the control unit being additionally configured to control a respective digital module (110) and a respective analog module (120) as intended, and wherein the high-voltage system is configured to output a total output voltage (102) according to a specification by the reference voltage (312).

Images