ELECTROSURGICAL GENERATOR HAVING AN INVERTER WITH IMPROVED DYNAMIC RANGE

Abstract

An electrosurgical generator for an electrosurgical instrument includes DC voltage supply and high-voltage inverter that generates high-frequency AC voltage having variable voltage and frequency. Inverter is multilevel inverter controlled by reference signal and having at least two groups of series-connected inverter cells, wherein each group is supplied with different DC voltage and wherein the voltages output by the two groups are summed to be output at output. Group supplied with higher voltage enables fast and large voltage changes with its inverter cells, while inverter cells of other group supplied with lower voltage allow fine setting with high change speed. Dynamic range is improved both from temporal viewpoint and in terms of increased voltage span. The number of HVC cells to be switched may furthermore be varied by way of modulator, wherein further number of LVC cells are switched in an opposing manner for compensation purposes. Switching losses may be reduced.

Claims

1. An electrosurgical generator that is designed to output a high-frequency AC voltage to an electrosurgical instrument, comprising a DC voltage supply and a high-voltage inverter that is fed from the DC voltage supply and generates a high-frequency AC voltage having a variable voltage and frequency that is applied to an output for connection of the electrosurgical instrument, wherein the inverter is designed as a multilevel inverter controlled by a reference signal for the voltage to be output and having at least two groups of series-connected inverter cells, wherein each group is supplied with a different DC voltage and wherein voltages output by the groups are summed to be output at the output.

2. The electrosurgical generator as claimed in claim 1, wherein the inverter cells are combined into two groups, of which a first group are supplied with a lower DC voltage than a different, second group.

3. The electrosurgical generator as claimed in claim 1, wherein the inverter cells have a bipolar configuration and output at least three different output voltage levels, which are positive, negative or zero.

4. The electrosurgical generator as claimed in claim 1, wherein there is a fixed ratio between the absolute value of the voltage that is generated by the inverter cells of a second group and the absolute value of the voltage that is generated by the individual inverter cells of a first group.

5. The electrosurgical generator as claimed claim 1, wherein, in order to drive the inverter cells of the groups, provision is made for a modulator that is designed to reduce switching frequencies of high-voltage inverter cells by selectively replacing actuation of the high-voltage inverter cells with actuation of a plurality of de-low-voltage inverter cells.

6. The electrosurgical generator as claimed claim 1, wherein a tap changer interacts with a modulator, to which tap changer the reference signal is applied and which tap changer is designed to convert the reference signal into a voltage level signal that is applied to the modulator.

7. The electrosurgical generator as claimed in claim 5, wherein the modulator is actuated via an enable signal, and provision is made for a change detector that is designed to identify a change in the reference signal and/or voltage level signal and to apply the enable signal to the modulator.

8. The electrosurgical generator as claimed in claim 5, wherein the modulator is furthermore designed to vary the number of inverter cells of the second group to be switched based on at least one predefinable parameter and to determine a further number of the inverter cells of the first group to be switched and to switch these in an opposing manner for compensation purposes.

9. The electrosurgical generator as claimed in claim 8, wherein the predefinable parameter comprises a switching frequency of the inverter cells of the second group, and a modulator is designed to minimize this switching frequency.

10. The electrosurgical generator as claimed in claim 8, wherein the predefinable parameter comprises a metric for power loss of the inverter cells, and the modulator is designed to adapt power loss caused by actuating the inverter cells of the second group to the power loss caused by actuating the inverter cells of the first group.

11. The electrosurgical generator as claimed in claim 5, wherein alternative switching rules for voltage changes are implemented in the modulator, these both leading to the same voltage change but switching a different number of inverter cells of the second group.

12. The electrosurgical generator as claimed claim 11, wherein, in the event of a voltage increase, according to one of the alternative switching rules, the number of inverter cells of the second group remains the same and one of the inverter cells of the first group is activated, or according to the other of the alternative switching rules, the number of switched inverter cells of the second group is increased by one and a plurality of inverter cells of the first group are switched in an opposing manner, wherein this plurality corresponds to the voltage multiple minus one.

13. The electrosurgical generator as claimed in claim 11, wherein, in the event of a voltage decrease, according to one of the alternative switching rules, the number of inverter cells of the second group remains the same and one of the inverter cells of the first group is deactivated, or according to the other of the alternative switching rules, the number of switched inverter cells of the second group is reduced by one and a plurality of inverter cells of the first group are switched in an opposing manner, wherein this plurality corresponds to the voltage multiple minus one.

14. The electrosurgical generator as claimed in claim 11, wherein respective switching ranges are assigned to the switching rules, wherein the switching ranges are different for positive and negative output voltage polarity.

15. The electrosurgical generator as claimed in claim 14, wherein limits of the switching ranges are dynamically changeable during operation, magnetic flux and/or temperature.

16. The electrosurgical generator as claimed in claim 5, wherein the modulator is designed to block switching of the inverter cells of the second group, with additional inverter cells of the first group being switched when the voltage increase or decrease exceeds the voltage value of the inverter cells of the second group.

17. The electrosurgical generator as claimed in claim 1, wherein provision is made for a monitoring unit that is designed to ascertain and to store magnetic flux in the inverter cells of the second group and/or the inverter cells of the first group.

18. The electrosurgical generator as claimed in claim 17, wherein provision is made for a compensation unit that interacts with the monitoring unit and is designed such that, in the event of a voltage increase, it first switches inverter cells of the second group or inverter cells of the first group with a low magnetic flux and, in the event of a voltage decrease, first switches those of the inverter cells with a high magnetic flux.

19. The electrosurgical generator as claimed in claim 1, wherein provision is made for a control signal generator for the multilevel inverter, which is designed to generate a reference signal for driving the multilevel inverter, wherein the reference signal is a pattern for AC voltage to be output by the electrosurgical generator, wherein the curve form is able to be set freely as desired.

Description

[0041] FIG. 1 shows a schematic illustration of an electrosurgical generator according to one exemplary embodiment with a connected electrosurgical instrument;

[0042] FIG. 2A, 2B show block diagrams of exemplary embodiments for a multilevel inverter of the electrosurgical generator according to FIG. 1 with cascaded inverter cells;

[0043] FIG. 3 shows a schematic circuit diagram of two of the inverter cells;

[0044] FIG. 4A, 4B show block diagrams of examples of a selector with a modulator for driving high-voltage and low-voltage inverter cells;

[0045] FIG. 5 shows a simplified example of the switching of higher-voltage and low-voltage inverter cells for implementing a reference signal by voltage level;

[0046] FIG. 6A, 6B shows another, more complex example of the switching of higher-voltage and low-voltage inverter cells for implementing a reference signal;

[0047] FIG. 7 shows a table containing switching rules for the modulator based on a polarity of the output voltage and a rise or fall in the reference signal;

[0048] FIG. 8 shows a table containing changeable switching rules for the modulator as a variant to FIG. 7; and

[0049] FIG. 9A, 9B show exemplary switching profiles for high-voltage and low-voltage inverter cells, without and with considering magnetic saturation in the inverter cells.

[0050] An electrosurgical generator according to one exemplary embodiment of the invention is illustrated in FIG. 1. The electrosurgical generator, referenced in its entirety by the reference numeral 1, comprises a housing 11 that is provided with an output port 14 for an electrosurgical instrument 16; in the illustrated exemplary embodiment, this is an electrical scalpel. It is connected to the output port 14 of the electrosurgical generator 1 via a connection plug 15 of a high-voltage connection cable. The power output to the electrosurgical instrument 16 may be changed via a power controller 12.

[0051] In order to supply power to the electrosurgical generator 1, provision is made for a DC voltage supply 2, which is able to be connected, via a mains connection cable (not illustrated), to the public grid and is fed therefrom. The DC voltage supply 2 is a power supply unit in the illustrated exemplary embodiment. It comprises a rectifier and feeds a DC link circuit 20 with DC voltage, the value of which is preferably fixed and is for example 48 volts. However, it should not be ruled out that the DC voltage value is variable between 0 and around 400 volts, wherein the absolute value of the DC voltage may in particular depend on the set power, the type of electrosurgical instrument 16 and/or its load impedance, which in turn depends on the type of tissue being treated. However, an internal power supply unit is not necessary, meaning that the DC voltage supply may also be implemented by an external power supply unit, or provision is made for a direct DC feed, for example 24 volts in vehicles or 48 volts in stationary applications.

[0052] An inverter is fed by the DC link circuit 20 and generates, from supplied DC voltage, high-frequency AC voltage in the high-voltage range, at predefinable frequencies in the range between 200 kHz and 4 MHz. The inverter is designed in the structural form of a multilevel inverter 4, as will be explained in even more detail below. The frequency and curve form of the high-frequency AC voltage to be generated by the multilevel inverter 4 are in this case predefined by an inverter controller 41 on the basis of a reference signal 43 generated by a control signal generator 40 (see FIG. 4). The high-frequency AC voltage generated by the multilevel inverter 4 is routed via an output line 13, an output transformer 7 for stepping up the output voltage into the range of a few kilovolts and a low-pass filter 8 and, secured against undesirable DC current components by a blocking capacitor 17, output at the output port 14 for connection for the electrosurgical instrument 16. The voltage and current of the high voltage generated and output by the multilevel inverter 4 are furthermore measured by way of a combined voltage and current sensor 18, and the measured signals are supplied to a processing unit 19, which applies the corresponding data about the output voltage, current and power as feedback to an operating controller 10 of the electrosurgical generator 1, which for its part communicates with the control signal generator 40. The power controller 12 is also connected to the operating controller 10. The operating controller 10 is furthermore designed to set various what are known as modes, which are typically stored voltage/time profiles. Provision is made for a selection switch 12′ for the user to select the mode. The operating controller 10 furthermore interacts with the control signal generator 40, which is designed to generate the reference signal 43 for the AC voltage to be output, in particular with regard to amplitude, frequency, curve form and duty cycle and to output it to an inverter controller 41.

[0053] The multilevel inverter 4 comprises a plurality of series-connected inverter cells 5 that are driven by the inverter controller 41. The inverter cells 5 are divided into two groups I and II, which are fed, in groups, with DC voltage of different values. A first one is called “group I” and comprises low-voltage inverter cells (LVC), specifically three inverter cells 5-1, 5-2 to 5-3 in the example in FIG. 2. They are fed by a DC voltage source with a low DC voltage, 12 V in the example. A group of high-voltage inverter cells (HVC) is also formed, this being called “group II” and comprising two inverter cells 5-4, 5-5 in the example illustrated in FIG. 2, these being fed with a higher DC voltage, 48 V in the example. The voltages output by the two groups I and II are summed by way of a transformer 6 (see FIGS. 4A, 4B). Provision is furthermore made for a selector 3 that defines the inverter cells 5 to be driven, in particular the number “m” of high-voltage inverter cells (HVC) from group II to be driven and the number “n” of low-voltage inverter cells (LVC) from group I to be driven.

[0054] Reference is now made to FIG. 2A. In the exemplary embodiment illustrated there, a DC voltage source having a defined DC voltage is connected to the input (illustrated on the left in the drawing) of each of the inverter cells 5. The respective inverter cell 5 generates therefrom an AC voltage that is output at the output (illustrated on the right in the drawing) of the respective inverter cell 5 in the form of AC voltage. The number of inverter cells is not limited and is as desired per se. The inverter cells 5 are numbered consecutively in FIG. 2A with the designation “5-1”, “5-2” to “5-5”, wherein the number 5 is an example and any number of at least two inverter cells may be provided. The DC voltages applied at the input of the respective inverter cell 5 are optionally coupled in terms of potential via a busbar 50. The AC voltage output at the output of the respective inverter cell 5 is accordingly denoted “V_1”, “V_2” up to “V_5”. The series connection of the inverter cells 5 results in their output voltages being added, ultimately giving, as overall output voltage:

[00001]$V_{\mathrm{out}}=\underset{i=1}{\overset{N}{.Math.}}V\_i$

[0055] Compared with a simple structure having only one DC voltage source, the outlay in terms of DC voltage sources is increased according to the invention, because more (in the example: two) than just one are now required. However, to make up for this, the number of voltage levels is increased considerably, specifically starting from eleven voltage levels with only one DC voltage source to more than twice that with 23 voltage levels. The number of voltage levels thus able to be achieved follows the formula

2*(mHVc*r+nLVc)+1,

wherein mHVc represents the number of higher-DC-voltage inverter cells (in the above example m=2), nLVc represents the number of low-DC-voltage inverter cells (in the above example n=3) and r represents the ratio of higher to low DC voltage (in the above example r=4).

[0056] The two DC voltage sources do not need to be isolated from one another in terms of potential, but rather they may share a common reference potential, as implemented in FIG. 2A by way of the busbar 50. This also makes it possible to generate the lower DC voltage from the higher DC voltage, which may be for example the DC voltage in the link circuit 20, by way of a DC-to-DC converter 42, in particular a DC-to-DC buck converter. In the present example, this would be designed for a ratiometric supply with a step-down ratio of 4:1, as illustrated in FIG. 2B. One advantage of this configuration by way of a ratiometric supply is that changes or fluctuations in the higher DC voltage are then reflected proportionally in the lower DC voltage, meaning that the relative gradation is maintained. This makes it possible for example to increase the output voltage by 12 V in two different ways: one conventionally by activating a further 12 V inverter cell, or by activating a 48 V inverter cell in combination with deactivating three 12 V inverter cells.

[0057] The structure of the individual inverter cells 5 and their interaction are illustrated by way of example in the schematic circuit diagram according to FIG. 3. A total of two inverter cells 5-1 and 5-5 are shown there in a cascaded arrangement, in order thus also to illustrate their supply with DC voltage of different values. The common DC voltage source 2, having a supply voltage Vin of 48 volts, is illustrated on the left-hand edge of the image. It is assigned a stabilization capacitor 23. The two inverter cells 5-3 and 5-4 are thereby supplied with DC voltage. Reference is first of all made below to the switching of the inverter cell 5-5, which is supplied with 48 volts. Provision is made for four power switches that operate as current valves and are arranged in an H-bridge configuration. The power switches are power semiconductor switches, for example configured as IGBTs, MOSFETs or GaNFETs. The power switches 51, 53 are connected in series and form a first branch, and the power semiconductors 52, 54 are likewise connected in series and form a second branch. The center taps of the two branches are guided out and connected to both ends of a primary winding 61 of a first transformer 6-5. The transformer 6-5 furthermore has a secondary winding 62, wherein the transformation ratio is 1:1 (it is pointed out that another transformation ratio may be provided, in particular in order to achieve pre-amplification, for example with a transformation ratio of 1:2). An output line 13 is connected to the secondary winding 62 and leads to the output port 14 of the electrosurgical generator 1 (possibly via a low-pass filter 8, not illustrated in FIG. 3, and an output transformer 7, see FIG. 1).

[0058] The two power switches 51, 53 of the first branch are driven by a common signal C1.a, wherein this signal is supplied to the power switch 53 in inverted form. The two power switches 52, 54 of the second branch are accordingly likewise driven by a common signal C1.b, wherein this signal is supplied to the power switch 52 in inverted form. The signals C1.a and C1.b are generated in a manner known per se by the inverter controller 41. This means that, in the event of a HIGH signal of C1.a, the power switch 51 is put into the on state and the power switch 53 is put into the off state, that is to say the first power branch applies a positive potential to the upper connection of the primary winding 61 of the transformer 6-1. Accordingly, in the event of a HIGH signal of C2.b, the power switch 54 is put into the on state, while the power switch 52 is put into the off state in the second power branch. The second power branch thus applies a negative potential to the lower connection of the primary winding 61. In the event of a LOW signal of C1.a or C1.b, this accordingly applies vice versa, that is to say the polarity at the primary winding 61 is reversed. An AC voltage is thus generated by the inverter cell 5-1 and applied to the primary winding 61 of the transformer 6-1.

[0059] The second inverter cell 5-1 has an identical structure, but is fed from the DC voltage source 2 via the ratiometric DC-to-DC converter 42, which brings about a reduction to a quarter of the input voltage. It thus outputs a DC voltage of 12 volts, which is supplied to the second inverter cell 5-1 in a manner identical per se to the first inverter cell 5-5. The same reference numerals are therefore used for identical elements in the figure. It is driven by way of control signals C2.a and C2.b that are generated by the inverter controller 41, for example by way of pulse width modulation (PWM), which is known per se, in a manner corresponding to what has been described above. It thus likewise outputs, at its output, an AC voltage that is applied to the primary winding 61 of a second transformer 6-1. Depending on the design of the DC-to-DC converter 42, the two inverter cells 5-1 and 5-5 are connected in terms of potential. This means that the AC voltages output directly by the inverter cells 5-1 and 5-5 are not readily able to be added, since they are linked to one another in terms of their potential. However, since this output AC voltage is supplied to each of the transformers 6-1 and 6-5, the AC voltages output by the transformers 6-1 and 6-5 are each potential-free and are readily able to be added to one another to give a common output voltage that is applied to the output line 13. If the inverter cells 5-1 and 5-5 are however decoupled in terms of potential through an appropriate design of the DC-to-DC converter, then the output AC voltages may also be added directly through a series connection without this transformer.

[0060] The overall voltage generated and summed in this way (and by other inverter cells 5-2 to 5-4) is output via the output line 13, at the end of which the low-pass filter 8 is arranged. This may be designed for example as a second-order filter comprising an inductor and a capacitor. It is pointed out that stray inductances of the transformers 6-1 to 6-n also contribute to the inductance of the low-pass filter, and may possibly at least partially replace the inductor. The low-pass filter 8 is tuned such that interference in the generated AC voltage due to the switching frequency of the power switches in the inverter cells 5 of the multilevel inverter 4 is filtered out. The output of the low-pass filter 8 is applied to a primary winding of the output transformer 7, which brings about galvanic isolation of the output port 14 connected to the secondary winding. Provision is furthermore made for a blocking capacitor 17. This serves as a safety element for preventing the output of DC current components to the surgical instrument 16.

[0061] The control signal generator 40 generates, in particular on the basis of specifications for the operating voltage 10, a reference signal 43 for driving the multilevel inverter 4. This is an AC voltage signal that is typically sinusoidal and has a particular frequency and amplitude. Reference is now made to FIGS. 4A, 4B. The reference signal 43 is applied to the selector 3, which determines therefrom the number and the type (HVC or LVC) of the inverter cells 5 to be driven, specifically broken down by low-voltage inverter cells (LVC) from group I and higher-voltage inverter cells (HVC) from group II. For this purpose, the selector 3 comprises a tap changer 31 and a modulator 33. The tap changer 31 is designed to convert the typically continuous reference signal 43 into a voltage level signal. This is a discrete signal that is indicative of the number of voltage levels, specifically typically expressed in levels the value of which results from the voltage of the low-voltage inverter cells (LVC) from group I, that is to say in the present example in levels of 12 V. One example of such a continuous reference signal 43 and a level signal formed therefrom expressed in levels of 12 V is depicted in FIG. 6A. The graduated curve shows the voltage level signal and thus forms a discretization of the reference signal 43 represented by the continuous curve.

[0062] The tap changer 31 may furthermore already make a preliminary division as to what portion thereof is incumbent on the low-voltage inverter cells (LVC) from group I or on the higher-voltage inverter cells (HVC) from group II. In one embodiment, as illustrated in FIG. 4A, this may be achieved for example by minimizing the basic number of inverter cells required to achieve the voltage in accordance with the reference signal 43. Such a division may comprise two signals, a signal “h” for the basic number of higher-voltage inverter cells (HVC) from group II to be switched on and a signal “1” for the basic number of low-voltage inverter cells (LVC) from group I to be switched on. This may in particular be a purely numerical division, for example whereby, in order to generate a voltage of 84 V, exactly one higher-voltage inverter cell (HVC) from group II and three low-voltage inverter cells (LVC) from group I need to be driven.

[0063] However, these basic numbers “h” and “1” are not used directly for drive purposes, but rather are varied by way of the modulator 33. The modulator 33 is intended to reduce the switching frequencies of the high-voltage inverter cells (HVC) in group II. As a replacement for this, low-voltage inverter cells (LVC) in group I are switched instead. This is described in more detail below. The resultant division by the modulator 33 into the numbers “n” for the low-voltage inverter cells (LVC) from group I to be switched and “m” for the higher-voltage inverter cells (HVC) from group II to be switched differs depending on the situation and is ambiguous according to the invention.

[0064] The number “n”, which is thus varied, of low-voltage inverter cells (LVC) from group I to be switched and the varied number “m” are output as output signals from the modulator 33 and applied to subcontrollers 45, 46 for the inverter cells LVC from group I or HVC from group II. These control the respective inverter cells LVC in group I or HVC in group II in a manner known per se. The subcontrollers 45, 46 acquire switching data regarding the individual inverter cells 5 of the low-voltage inverter cells (LVC) from group I or the high-voltage inverter cells (HVC) from group II. These data comprise, inter alia, switch-on time, counts of the number of switching procedures and the magnetic flux through the individual inverter cells 5 and their transformer 6. They transmit corresponding state data via data lines 47, 48 to the modulator 33 and/or an adaptation module 36 upstream thereof.

[0065] The modulator 33 does not necessarily need to operate continuously. It may be enough for it to be actuated and to divide the voltage level signal into the number of low-voltage inverter cells (LVC) and high-voltage inverter cells (HVC) to be switched in particular when a change has resulted in the voltage level signal or in the reference signal 43. To this end, provision is optionally made for a change detector 32, which monitors the reference signal 43 and actuates the modulator 33 in the event of a change.

[0066] In one alternative embodiment, as illustrated in FIG. 4B, the tap changer 31′ has a different design, such that it outputs only a discretized reference signal (level signal) 44. The modulator 33 determines directly therefrom the number “m” of high-voltage inverter cells (HVC) to be switched and the number of low-voltage inverter cells (LVC) to be switched. This is explained with reference to a simplified example: The tap changer 31′ generates a discrete level signal 44 from the reference signal 43. The modulator 33 is designed to compare the level signal 44 with the last previous value of the level signal. The comparison may reveal that there is a rise, a fall or constancy. This is detected by way of the change detector 32′. The values for the numbers “m” and “n” are adjusted only in the event of a rise or a fall, that is to say only when a change has occurred in comparison with the previous level signal. This is achieved by way of the switching rules, as are explained further below with reference to the examples illustrated in FIGS. 7 and 8.

[0067] One example of the voltage generation using higher-voltage inverter cells (HVC) is illustrated in FIG. 5 by a dashed line, and the voltage generated by the low-voltage inverter cells (LVC) is illustrated by the solid line close to the zero line. Together, they give the desired sinusoidal profile, as illustrated by the quantized sinusoidal line. It may be seen that each of the two higher-voltage inverter cells (HVC) from group II needs to be switched on and off just once per half-wave, and the further adjustment is performed by frequently switching the low-voltage inverter cells (LVC) from group I. In this case, the LVCs from group I both increase the voltage (for example right at the start in the interval from 0 to 0.25 μs) and also reduce it through compensatory counter-switching in order to reduce the temporally excessively high voltage output by the inverter cells HVC (for example in the interval 0.25 to 0.65 μs and 0.87 to 0.98 μs). Switching procedures of the higher-voltage inverter cells (HVC) are thereby able to be avoided and the number thereof is thus able to be reduced, and the considerable switching power loss arising as a result of the switching of the HVC cells is thus also able to be reduced.

[0068] A more complex example of more inverter cells is illustrated in FIG. 6B. The values “m” resulting here for the number of higher-voltage inverter cells (HVC) from group II to be switched on are illustrated by a dashed line and for the number of low-voltage inverter cells (LVC) from group I to be switched on are illustrated by a solid graduated line “n”. It may clearly be seen on the profile of the line denoted “m” that the switching activity of the higher-voltage inverter cells (HVC) is considerably reduced, in particular in the region of the amplitude maxima of the reference signal and the zero crossing. By prioritizing switching activities of the low-voltage inverter cells (LVC), the higher-voltage inverter cells (HVC), having higher switching losses, are able to be spared.

[0069] To achieve this, switching rules 34 are implemented in the modulator 33. The switching rules are based on an exemplary configuration having two higher-voltage inverter cells (HVC) in group II and 4 low-voltage inverter cells in group I, as also illustrated in FIGS. 4A, 4B. Using the change detector 32, the switching states of the inverter cells are changed only when the reference signal 43 has also changed. The switching rules 34 provide two possible alternatives for the case of a rise: [0070] a) increasing the voltage output by the low-voltage inverter cells (LVC) by 1 level (corresponding to 12 V) without any change regarding the high-voltage inverter cells (HVC); or [0071] b) reducing the voltage output by the low-voltage inverter cells (LVC) by 3 levels (corresponding to −36 V) and increasing the voltage output by the high-voltage inverter cells (HVC) by one level (+48 V).

[0072] Both alternatives a), b) lead to the same voltage change by one level, specifically by +12 V. Alternative b) requires switching of one of the high-voltage inverter cells (HVC), which, due to the quadratic relationship with the voltage supply multiplied by four, means 16 times more switching losses in comparison with one of the low-voltage inverter cells (LVC). This is added to by another three switching procedures of the low-voltage inverter cells (LVC). Alternative b) thus means 19 times more energy loss than alternative a) of the switching rules 34.

[0073] These relationships are taken into consideration by the switching rules 34 illustrated in FIG. 7. Reference is now made to the left-hand column, which concerns the case of a voltage rise. This contains the two alternatives a) and b). Input parameters are the polarity of the output voltage and the voltage output by the low-voltage inverter cells (LVC) from group I, expressed in voltage levels of the LVCs. In this case, “1” represents an output voltage of +12 V, “4” represents an output voltage of +48 V, and accordingly “−4” represents an output voltage −48 V. The switching rules 34 implemented in the modulator 33 then state, for a positive polarity of the output voltage, that, in the event of a voltage level between −4 and 3 (corresponding to −48 V to +36 V) for group I, switching rule a) is used, that is to say the voltage output by the LVC converter cells from group I is increased by 12 V. If however voltage level 4 is already present in group I, then alternative switching rule b) is used, in which the HVC inverters from group II are switched up by one level, resulting in an increase by 48 V, and the LVC inverter cells from group I are switched down by three levels for compensation purposes, corresponding to −36 V, ultimately resulting in the desired increase by 12 V. If the polarity of the output voltage is negative, then appropriately adjusted switching ranges of −4 to −1 for alternative a) and 0 to 4 for alternative b) of the switching rules apply. The appropriate switching rule for the case of a voltage drop is illustrated in the right-hand column of FIG. 7. Alternatives a) and b) apply in this case too, but with adjusted ranges as may be seen from FIG. 7.

[0074] The modulator 33 may furthermore comprise a hysteresis module 35. This is designed to acquire the switching frequency in relation to the high-voltage inverter cells (HVC) from group II and to minimize their switching procedures in the event of excessive switching activity. For this purpose, the hysteresis module 35 acts for example on the switching rules 34 so as to change the range limits such that alternative b) becomes rarer.

[0075] The switching rules 34 and their switching ranges may be adapted by an adaptation module 36, in particular on the basis of operating conditions of the multilevel inverter 4 with its inverter cells 5. The adaptation module 36 comprises a monitoring unit having a compensation unit 38. It detects the magnetic flux in the individual inverter cells in each of groups I and II and thus acts on the switching ranges of the switching rules 34. The switching ranges of the switching rules 34 may thereby be changed dynamically. If the overall magnetic flux through the high-voltage inverter cells (HVC) is too high, then the switching ranges are changed such that these cells are switched on only later and that they are switched off again earlier. Parameters B and D may be changed for this purpose, as illustrated in the modified switching rules 34 according to FIG. 8. If by contrast, on the other hand, the magnetic flux is too low, then parameters A and C may be used to change the switching ranges such that the high-voltage inverter cells (HVC) are switched on earlier and switched off again only later. It is thereby possible to balance out the magnetic flux and avoid saturation.

[0076] The adaptation module 36 furthermore comprises an optional switch-on time monitor 39. This acquires, separately for the low-voltage inverter cells (LVC) and the high-voltage inverter cells (HVC), the duration of a positive or negative voltage output. If particular preset limit values are exceeded, then the switching ranges may be adjusted dynamically in a manner similar to that described above for magnetic saturation. Provision may however also be made that the corresponding highly loaded inverter cells are switched off for a certain time.

[0077] The effect of the adaptation module 36 with the dynamically changed switching ranges is illustrated in FIGS. 9A and 9B. FIG. 9A shows, as a starting point, the switching behavior in accordance with the switching rules 34 with unchanged switching ranges, as illustrated in FIG. 7. If the compensation unit 38 identifies that the magnetic flux through the high-voltage inverter cells (HVC) in group II is too low, then the adaptation module 36 shifts the parameter C, for example by a value of 3. This then results in appropriately changed switching ranges in accordance with the modified switching rules, as illustrated in FIG. 9B with the parameter C=3. This means that the high-voltage inverter cells (HVC), when they have been switched on once, remain switched on for longer, that is to say switch-off is delayed, as shown by the line m′. There is no resultant effect on switch-on here (this could be achieved by changing the parameter A). The switching activity of the low-voltage inverter cells (LVC) changes accordingly, as shown by the line n′. As a result, the switch-on and switch-off behavior of the high-voltage inverter cells (HVC) is thus asymmetric in the sense that they switch off considerably later. They are thus switched on for longer, which increases their magnetic flux. Dynamically changing the switching range of the switching rules 34 thus achieves the desired aim of increasing the magnetic flux in the high-voltage inverter cells (HVC).

[0078] Provision is furthermore made for an optional peak detector 37. The reference signal 43 is applied thereto. It is designed to identify the occurrence of signal peaks in the reference signal 43, for example when the amplitude reaches its maximum value. If this is identified, then the peak detector 37 may act on the modulator 33 such that switching, due per se in accordance with the switching rules 34, of high-voltage inverter cells (HVC) is blocked, and instead a surplus low-voltage inverter cell (LVC) is switched in order to achieve the last voltage levels. Such an optional surplus LVC inverter cell in group I is illustrated by a dashed line in FIG. 4. It is thereby possible to make use of the fact that the amplitude maximum is known and simple thanks to the reference signal 43. The peak detector 37 identifies this and acts on the modulator 33 so as to block or at least to minimize the switching procedures of high-voltage inverter cells (HVC) 5 close to the amplitude maxima.