A SYSTEM AND METHOD FOR PROVIDING N BIPOLAR AC PHASE VOLTAGES

Abstract

Disclosed herein is a system (20) for providing N bipolar AC phase voltages U.sub.Vj, with j=1 . . . N, said system (20) comprising N modular energy storage direct converter systems (MESDCS) (22) and a control system (20), wherein the first ends (24) of each MESDCS (22) are connected to a common floating connection point (28), and wherein the j-th MESDCS (22) is controllable to output at its second end (26) a star voltage Us.sub.j with respect to the floating connection point (28), with j=1, . . . , N, wherein said system (20) is configured to provide each of said phase voltages Uv.sub.j as voltage differences between two of said star voltages, such that Uv.sub.j=Us.sub.j+1−Us.sub.j, or Uv.sub.j=Us.sub.j−Us.sub.j+1 for each j between 1 and N−1, and Uv.sub.N=Us.sub.1−Us.sub.N, or Uv.sub.N=Us.sub.N−Us.sub.1, respectively, wherein said control system (30) is configured to control each MESDCS (22) to output a corresponding unipolar star voltage Us.sub.j that can be decomposed into a periodic bipolar AC function P.sub.j(t) and a unipolar offset U.sub.off(t) that is common to each star voltage Us.sub.j, such that Us.sub.j(t)=P.sub.j(t)+U.sub.off(t), wherein the absolute value of said common unipolar offset U.sub.off(t) is at all times t sufficiently high that Us.sub.j (t) is unipolar,

wherein the periodic bipolar AC functions P.sub.j(t) associated with different star voltages Us.sub.j are phase-shifted copies of each other such that for each integers i, j chosen from [1, . . . , N] and k chosen from [1, . . . , N−1], P.sub.i(t)=P.sub.j(t+k.Math.T/N), wherein T is the period of said periodic bipolar AC function P.sub.j(t), wherein in particular, P.sub.i(t)=P.sub.j(t+(i−j).Math.T/N).

Claims

1. A system for providing N bipolar AC phase voltages U.sub.Vj, with j=1 . . . N, said system comprising: N modular energy storage direct converter systems and a control system, wherein each MESDCS comprises a converter arm having a first end and a second end and comprising a plurality of sequentially interconnected modules, wherein each module comprises at least one first terminal and at least one second terminal, a storage element for electrical energy or an energy conversion element, and a plurality of module switches, wherein in each two adjacent modules, the at least one first terminal of one module is connected either directly or via an intermediate component to the at least one second terminal of the other module, wherein under control of said control system, said plurality of module switches are configured to connect energy storage elements or energy conversion elements of adjacent modules in series, and selectively deactivate and or bypass the energy storage element or energy conversion element of each module and connect energy storage elements or energy conversion elements of adjacent modules in parallel, wherein the first ends of each MESDCS are connected to a common floating connection point, and wherein the j-th MESDCS is controllable, by said control system, to output at its second end a star voltage Us.sub.j with respect to the floating connection point, with j=1, . . . , N, wherein said system is configured to provide each of said phase voltages Uv.sub.j as voltage differences between two of said star voltages, such that Uv.sub.j=Us.sub.j+1−Us.sub.j, or Uv.sub.j=Us.sub.j−Us.sub.j+1 for each j between 1 and N−1, and Uv.sub.N=Us.sub.1−Us.sub.N, or Uv.sub.N=Us.sub.N−Us.sub.1, respectively, wherein said control system is configured to control each MESDCS to output a corresponding unipolar star voltage Us.sub.j that can be decomposed into a periodic bipolar AC function P.sub.j(t) and a unipolar offset U.sub.off(t) that is common to each star voltage Us.sub.j, such that Us.sub.j(t)=P.sub.j(t)+U.sub.off(t), wherein the absolute value of said common unipolar offset U.sub.off(t) is at all times t sufficiently high that Us.sub.j(t) is unipolar, wherein the periodic bipolar AC functions P.sub.j(t) associated with different star voltages Us.sub.j are phase-shifted copies of each other such that for each integers i, j chosen from [1, . . . , N] and k chosen from [1, . . . , N−1], P.sub.i(t)=P.sub.j(t+k.Math.T/N), wherein T is the period of said periodic bipolar AC function P.sub.j(t.

2. The system of claim 1, wherein U.sub.off(t) is a constant offset U0, wherein U0 is chosen such that each star voltage Us.sub.j is equal to zero once per period T, or that an absolute value of each star voltage differs from zero once per period T by less than 15% of the maximum absolute voltage value which that star voltage acquires during each period T.

3. The system of claim 1, wherein U.sub.off(t) is a time-dependent offset chosen such that at each instance in time, one of said star voltages Us.sub.j is equal to zero, or that an absolute value of each star voltage differs from zero by less than 15 of the maximum absolute voltage value which that star voltage acquires during each period T.

4. The system of claim 2, wherein P.sub.j(t)=U0.Math.sin(ωt+(j−1).Math.2π/N), with U0 being a voltage amplitude and ω=2π/T.

5-7. (canceled)

8. The system of claim 1, wherein each module has two first terminals and two second terminals, wherein each energy storage element or energy conversion element has a first pole connected or connectable with one of the first terminals and a second pole connected or connectable with the other one of the first terminals, and wherein said plurality of module switches comprise a first switch having a first switch terminal connected with one of the second terminals of the module and a second switch terminal connected or connectable with the first pole of said energy storage/energy conversion element, a second switch having a first switch terminal connected with the other one of the second terminals of the module and a second switch terminal connected or connectable with the second pole of said energy storage/energy conversion element, and a third switch provided in a connection line connecting the first switch terminal of the first switch and the second switch terminal of the second switch, said module configured to connecting its energy storage/energy conversion element in series with the energy storage/energy conversion element of an adjacent module connected to its second terminals by opening the first and second switches and closing the third switch, and configured to connect its energy storage/energy conversion element in parallel with the storage/energy conversion element of an adjacent module connected to its second terminals by closing the first and second switches and opening the third switch.

9. The system of claim 8, wherein each module further comprises a fourth switch connected either between the first pole of said storage/energy conversion element and the second switch terminal of the first switch, or between the second pole of said storage/energy conversion element and the second switch terminal of the second switch, said forth switch configured to deactivate or bypass the storage/energy conversion element of a given module n a series or parallel connection of storage/energy conversion elements of adjacent modules connected to the first terminals and second terminals of said given module, respectively.

10. The system of claim 1, wherein the energy storage elements of the modules are batteries having a nominal voltage, wherein the nominal battery voltages are equal to or less than 25 V and wherein the batteries within a module do not comprise a battery management system of their own.

11. The system of claim 1, wherein said module switches are semiconductor switches having a gate, and wherein a gate voltage of at least one semiconductor switch of each module is provided by a battery of said module forming said energy storage element.

12. The system of claim 1, wherein said module switches are semiconductor switches having a gate, and wherein a gate voltage of at least one semiconductor switch of each module is provided by a battery of another one of the modules, said battery forming said energy storage of said another module.

13. The system of claim 11, wherein in a given module, a first group of switches among said module switches have their gate voltages provided by the battery of said given module, and a second group of switches among said module switches have their gate voltages provided by the battery of one or more different modules wherein each of the first and second groups comprises one or more switches and wherein said first and second groups are chosen such that in case the battery in the given module fails and any switch of said first group is no longer operative, the battery of said given module can be bypassed in one or both of a series connection or a parallel connection of modules to both sides of said given module.

14. The system of claim 13, wherein said first group of switches comprises the third switch and said second group of switches comprises the first and second switches.

15. A system for providing N bipolar AC phase voltages U.sub.Vj, with j=1 . . . N, said system comprising: N modular energy storage direct converter systems (MESDCS) and a control system, wherein each MESDCS comprises a converter arm having a first end and a second end and comprising a plurality of sequentially interconnected modules, wherein each module comprises at least one first terminal and at least one second terminal, a storage element for electrical energy or an energy conversion element, and a plurality of module switches, wherein in each two adjacent modules the at least one first terminal of one module is connected either directly or via an intermediate component to the at least one second terminal of the other module, wherein under control of said control system, said plurality of module switches are configured to connect energy storage elements or energy conversion elements of adjacent modules in series, and for one or both of selectively deactivate or bypass the energy storage element or energy conversion element of each module and connect energy storage elements or energy conversion elements of adjacent modules in parallel, wherein the first ends of each MESDCS are connected to a common floating connection point, and wherein the j-th MESDCS is controllable, by said control system, to output a star voltage Us.sub.j with respect to the floating connection point, with j=1, . . . , N, wherein said system is configured to provide each of said phase voltages Uv.sub.j as voltage differences between two of said star voltages, such that Uv.sub.j=Us.sub.j+1−Us.sub.j, or Uv.sub.j=Us.sub.j−Us.sub.j+1 for each j between 1 and N−1, and Uv.sub.N=Us.sub.1−Us.sub.N, or Uv.sub.N=Us.sub.N−Us.sub.1, respectively, wherein said control system is configured to control each MESDCS to output a corresponding bipolar star voltage Us.sub.j that can be decomposed into a periodic bipolar AC function P.sub.j(t) and a bipolar offset U.sub.off(t) that is common to each star voltage Us.sub.j, such that Us.sub.j=P.sub.j(t)+U.sub.off(t), wherein the periodic bipolar AC functions P.sub.j(t) associated with different star voltages Us.sub.j are phase-shifted copies of each other such that for each integers i, j chosen from [1, . . . , N] and k chosen from [1, . . . , N−1], P.sub.i(t)=P.sub.j(t+k.Math.T/N), wherein T is a period of said periodic bipolar AC function P.sub.j(t), and wherein preferably, P.sub.i(t)=P.sub.j(t+(i−j).Math.T/N), and wherein U.sub.off(t) is such that for each star voltage Us.sub.j, a difference between the maximum and minimum values acquired during a full period T is less than without such bipolar offset.

16. The system of claim 15, wherein P.sub.j(t)=A.Math.sin(ωt+(j−1).Math.2π/N), with A being a voltage amplitude and ω an angular frequency, and wherein U.sub.off(t) is such that for each phase voltage Us.sub.j, the difference between the maximum and minimum values acquired during a full period 2π/ω is less than 2.Math.A.

17. The system of claim 15, wherein U.sub.off(t)=−(Max (P.sub.j(t))+Min(P.sub.j(t)))/2, where Max(P.sub.j(t)) is defined as the maximum value among each of the functions P.sub.j(t) at a given time instance t, and Min(P.sub.j(t)) is defined as the minimum value among each of the functions P.sub.j(t) at a given time instance t.

18. The system of claim 15, wherein
U.sub.off(t)=L−Max(P.sub.j(t)), if Max(P.sub.j(t))>L, and
U.sub.off(t)=−L−Min(P.sub.j(t)), if Min(P.sub.j(t))<−L, wherein Max(P.sub.j(t)) is defined as the maximum value among each of the functions P.sub.j(t) at a given time instance t, and Min(P.sub.j(t)) is defined as the minimum value among each of the functions P.sub.j(t) at a given time instance t, and L is a constant.

19. The system of claim 18, wherein P.sub.j(t)=A.Math.sin(ωt+(j−1).Math.2π/N), N=3 and L=A.Math.√{square root over (3)}/2.

20. The system of claim 15, wherein said plurality of module switches are configured to connect energy storage elements or energy conversion elements of adjacent modules in anti-series, wherein the anti-series connection of an energy or energy conversion element corresponds to a series connection with reversed polarity.

21-23. (canceled)

24. The system of claim 15, wherein said system is connected to a load comprising N impedances Zj, j=1, 2, . . . N, connected in a ring configuration, wherein the k-th impedance Zk is connected between the second ends of the k-th and (k+1)-th MESDCS, such that the k-th phase voltage Uv.sub.k is applied across the k-th impedance Zk, with k=1, . . . , N−1, and the N-th impedance ZN is connected between the second ends of the N-th and the first MESDCS, and wherein the load is an electric motor.

25. A method for providing N bipolar AC phase voltages Uv.sub.j, with j=1 . . . N using N modular energy storage direct converter systems (MESDCS), wherein each MESDCS comprises a converter arm having a first end and a second end and comprising a plurality of sequentially interconnected modules, wherein each module comprises at least one first terminal and at least one second terminal, a storage element for electrical energy or an energy conversion element, and a plurality of module switches, wherein in each two adjacent modules, the at least one first terminal of one module is connected either directly or via an intermediate component to the at least one second terminal of the other module, wherein said method comprises operating said plurality of module switches for connecting energy storage elements or energy conversion elements of adjacent modules in series, and for one or both of selectively deactivating or bypassing the energy storage element or energy conversion element of each module and connecting energy storage elements or energy conversion elements of adjacent modules in parallel, wherein the first ends of each MESDCS are connected to a common floating connection point, and wherein the j-th MESDCS is controlled to output at its second end a star voltage Us.sub.j with respect to the floating connection point, with j=1, . . . , N, wherein the method comprises providing each of said phase voltages Uv.sub.j as voltage differences between two of said star voltages, such that Uv.sub.j=Us.sub.j+1−Us.sub.j, or Uv.sub.j=Us.sub.j−Us.sub.j+1 for each j between 1 and N−1, and Uv.sub.N=Us.sub.1−Us.sub.N, or Uv.sub.N=Us.sub.N−Us.sub.1, respectively, wherein each MESDCS is controlled to output a corresponding unipolar star voltage Us.sub.j that can be decomposed into a periodic bipolar AC function P.sub.j(t) and a unipolar offset U.sub.off(t) that is common to each star voltage Us.sub.j, such that Us.sub.j(t)=P.sub.j(t)+U.sub.off(t), wherein the absolute value of said common unipolar offset U.sub.off(t) is at all times t sufficiently high that Us.sub.j (t) is unipolar, wherein the periodic bipolar AC functions P.sub.j(t) associated with different star voltages Us.sub.j are phase-shifted copies of each other such that for each integers i, j chosen from [1, . . . , N] and k chosen from [1, . . . , N−1], P.sub.i(t)=P.sub.j(t+k.Math.T/N), wherein T is the period of said periodic bipolar AC function P.sub.j(t), wherein P.sub.i(t)=P.sub.j(t+(i−j).Math.T/N).

26-30. (canceled)

31. The system of claim 1, wherein P.sub.i(t)=P.sub.j(t+(i−j).Math.T/N).

Description

SHORT DESCRIPTION OF THE FIGURES

[0120] FIG. 1 shows a four-quadrant-module for use in an MESDCS having eight switches.

[0121] FIG. 2 shows a four-quadrant-module for use in an MESDCS having nine switches.

[0122] FIG. 3 shows on the left a two-quadrant-module for use in an MESDCS having three module switches, and on the right a further two-quadrant-module for use in an MESDCS having four module switches.

[0123] FIG. 4 shows four cascaded two-quadrant modules having three switches (top) and four switches (bottom), respectively.

[0124] FIG. 5 is a schematic representation of a system according to a first aspect of the invention.

[0125] FIG. 6 shows three MESDCS connected in a star configuration that could be used in the system of FIG. 5.

[0126] FIG. 7 shows two star voltages U.sub.S1, U.sub.S2 and a corresponding phase-to-phase voltage U.sub.V1=U.sub.S1-U.sub.S2, assuming idealized sinusoidal voltage waveforms.

[0127] FIG. 8 shows voltage waveforms similar to those of FIG. 7 but as generated in a step-like manner by means of two MESDCS.

[0128] FIG. 9 shows the power of one of the MESDCS in operation as a function of time.

[0129] FIG. 10 shows three sinusoidal, monopolar star voltages U.sub.S1, U.sub.S2, U.sub.S3 generated using a constant offset voltage U.sub.o.

[0130] FIG. 11 shows the corresponding phase-to-phase voltages U.sub.V1, U.sub.V2, and U.sub.V3.

[0131] FIG. 12 shows the MESDCS powers associated with the three star voltages, similar to FIG. 11.

[0132] FIG. 13 shows optimized star voltages U.sub.S1, U.sub.S2 and U.sub.S3 using a time-dependent common offset U.sub.off(t).

[0133] FIG. 14 shows the power losses associated with the optimized star voltages of FIG. 13.

[0134] FIG. 15 shows the power loss associated with the star voltages of FIG. 10 when a constant offset is used.

[0135] FIG. 16 shows the power of the MESDCS when using the optimized time-dependent common offset to the star voltages.

[0136] FIG. 17 schematically shows a system using three MESDCS that are only capable of generating unipolar star voltages.

[0137] FIG. 18 schematically shows a system with three MESDCS capable of generating bipolar star voltages, based on two-quadrant modules.

[0138] FIG. 19 schematically shows a system with three MESDCS capable of generating bipolar star voltages, based on four-quadrant modules.

[0139] FIG. 20 shows optimized star voltages that can be employed in the systems of FIG. 18 or 19.

[0140] FIG. 21 shows a module 10 where the module battery voltages match the gate voltages of semiconductor switches employed therein.

[0141] FIG. 22 shows two complete modules and part of a third module of the type shown in FIG. 21, illustrating how module with a dysfunctional battery can be bypassed.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0142] FIG. 5 shows a system 20 for providing three AC phase voltages. The system comprises three modular energy storage direct converter systems (MESDCS) 22, each comprising a converter arm having a first end 24 and a second end 26 and comprising a plurality of sequentially interconnected modules 10 which in FIG. 5 are only symbolically represented. Each of the first ends 24 of the MESDCS 22 are connected to a common floating connection point 28. In other words, the MESDCS 22 are connected in a star configuration, but the common floating connection point 28 is not a “star point” in the usual sense, because its potential can vary upon operation of the system 20, and can hence not necessarily be used as the “star point” of an electrical machine.

[0143] FIG. 6 shows a more detailed example of three MESDCS converter arms 22 arranged in star configuration, where each converter arm is shown to have four modules to, which are of the type shown in FIG. 3. Note that in practical applications, the number of modules 10 would typically be much larger than this.

[0144] With reference again to FIG. 5, the system 20 further comprises a control system 30 which is configured to control the switches within each of the modules 10 of the MESDCS 22 such as to generate a corresponding star voltage at the second end 26 of each MESDCS converter arm 22 (note that in the description of FIG. 5, the terms “MESDCS and “converter arm” may be used synonymously, since each MESDCS 22 comprises only one converter arm). The control system 30 can be provided by hardware, by software or a combination of both. For example, the control system 30 may comprise one or more microprocessors carrying out the control of the MESDCS 22, and in particular the control of the operation of the module switches 18 comprised in the modules 10, under control of a corresponding software code. Additionally or alternatively, the control system 30 may comprise one or more ASICs or FPGAs. The control system 30 may comprise a single control unit, or a plurality of control units which are in data or signal communication with each other. Signal links are provided to connect the control system 30 with the MESDCS 22 and their respective modules 10, which are, however, not shown in the figures for clarity. The signal links could be formed by wired or wireless connections.

[0145] FIG. 5 shows a three phase system, in which the phases are conventionally designated by letters R, S, and T, and the star voltages would therefore often be referred to as U.sub.R, U.sub.S and U.sub.T in the art. However, since the invention is not limited to systems with three phases, in the summary of the invention, a more general terminology was introduced, according to which the star voltages are designated as U.sub.Sj. wherein the index j=1, . . . , N designates the corresponding phase. Accordingly, using the more general terminology for this specific three-phase case, one obtains U.sub.S1=U.sub.R, U.sub.S2=U.sub.3 und U.sub.S3=U.sub.T.

[0146] The system 20 of FIG. 5 provides three by polar AC phase voltages as phase-to-phase voltages, which are each generated as voltage differences between two of said three star voltages:

U.sub.RS=U.sub.V1=U.sub.S1−U.sub.S2;

U.sub.ST=U.sub.V2=U.sub.S2−U.sub.S3; and

U.sub.TR=U.sub.V3=U.sub.S3−U.sub.S1.

[0147] Moreover, in the system of FIG. 5, the load is formed by three resistors connected in a triangular configuration, where the star voltages U.sub.S1 (corresponding to U.sub.R), U.sub.S2 (corresponding to U.sub.S) and U.sub.S3 (corresponding to U.sub.T) are applied to the vertices of the triangle. In preferred embodiments, the load could be formed by an electric motor. In the embodiment shown, the MESDCS 22 comprises only two-quadrant-modules and the converter arm of the MESDCS 22 does not have a polarity reversing circuit, such that the star voltages U.sub.Sj are by construction unipolar. However, the phase-to-phase voltages U.sub.Vj, which are based on differences between star voltages U.sub.Sj+1, U.sub.Sj will be bipolar.

[0148] For example, in one embodiment, the star voltages U.sub.Sj may be sinusoidal voltages oscillating between 0V and a maximum voltage U0, and with a mutual phase shift of T/3, where T is the period of the oscillation:

[00001]$U_{S1}\left(t\right)=U_0.Math.\left(\frac{1}{2}+\frac{1}{2}.Math.\sin \left(\omega t\right)\right);$$U_{S2}\left(t\right)=U_0.Math.\left(\frac{1}{2}+\frac{1}{2}.Math.\sin \left(\omega t+\frac{2\pi }{3}\right)\right);$$U_{S3}\left(t\right)=U_0.Math.\left(\frac{1}{2}+\frac{1}{2}.Math.\sin \left(\omega t+\frac{4\pi }{3}\right)\right)$

[0149] In this case, the first phase-to-phase voltage U.sub.V1 is given as:

[00002]$\begin{array}{c}{U}_{V1}\left(t\right)=U_0.Math.\left(\frac{1}{2}+\frac{1}{2}.Math.\sin \left(\omega t\right)\right)-\left(\frac{1}{2}+\frac{1}{2}.Math.\sin \left(\omega t+\frac{2\pi }{3}\right)\right))\\ =\frac{1}{2}.Math.U_0.Math.\left(\sin \left(\omega t\right)-\sin \left(\omega t+\frac{2\pi }{3}\right)\right)\\ =\frac{\sqrt{3}}{2}.Math.U_0.Math.\sin \left(\omega t+\frac{\pi }{6}\right)\end{array}$

[0150] It is seen that the first phase-to-phase voltage U.sub.V1 is a bipolar sinusoidal waveform having the same period T as the star voltages U.sub.Sj, and an amplitude that is increased by a factor v. The corresponding curves U.sub.S1, U.sub.S2 and U.sub.v1 are shown in FIG. 7.

[0151] In the above description, it was assumed for simplicity that the star voltages U.sub.Sj were ideal sinus functions. FIG. 8 shows the corresponding voltage waveforms as generated by a typical MESDCS 22. Since the MESDCS 22 generates the star voltage U.sub.Sj by connecting a finite number of modules 10 in series, the output voltages can only acquire discrete values, which leads to the step-like behavior seen in FIG. 8. However, as used herein, a step-like waveform approximating a sinus function will be referred to as a sinus function herein as well, and the same understanding applies for the appending claims.

[0152] Note that in view of the symmetry of the star voltages U.sub.Sj and assuming a symmetric load, the sinusoidal phase-two-phase voltages will lead to sinusoidal currents in the MESDCS 22. This means that corresponding module batteries forming the “energy storage elements” 16 will also be periodically subjected to transient currents in “opposite direction”, i.e. in a direction which leads to charging of the battery. The power Power1(t) as a function of time for the first MESDCS 22 (assuming an ohmic load) is hence obtained as:

[00003]$\begin{array}{c}{\mathrm{Power}}_1\left(t\right)=U_0.Math.\left(\frac{1}{2}+\frac{1}{2}.Math.\sin \left(\omega t\right)\right).Math.I_0.Math.\sin \left(\omega t\right)\\ =\frac{1}{2}.Math.U_0.Math.I_0.Math.{\left(\sin \left(\omega t\right)+\sin \left(\omega t\right)\right)}^2)\end{array}$

[0153] The corresponding waveform is shown in FIG. 9. Note, however, that the “opposite currents” only occur at time intervals where the corresponding star voltage U.sub.Sj is small, and hence at time intervals where the modules 10 are either bypassed or connected in parallel, such that the additional battery losses are in fact very small. Note that the fact that the battery is charged by “opposite currents” of course per se does not imply that the corresponding energy is lost, since the energy is stored in the battery. However, overall transferring energy to the battery and back to the load involves net losses.

[0154] In the discussion so far, it was assumed that the star voltages U.sub.Sj are sinusoidal but with a constant common offset that ensures that the star voltages are always positive:

U.sub.S1(t)=U.sub.0.Math.(1+sin(ωt))

[0155] FIG. 10 summarizes all three star voltages U.sub.S1, U.sub.S2, and U.sub.S3, which are seen to oscillate between zero and 2U0, and FIG. 11 shows the corresponding phase-to-phase voltages U.sub.V1, U.sub.V2, and U.sub.V3. Finally, FIG. 12 shows the corresponding powers Power.sub.1, Power.sub.2, Power.sub.3 of the MESDCS 22. Herein, the power associated with the first MESDCS 22 is shown with a thick line to better discern its waveform, while the powers associated with the second and third to MESDCS are simply phase-shifted copies thereof.

[0156] The skilled person will appreciate that the constant common offset U0 in this case is the smallest possible constant offset that ensures that the star voltages U.sub.S1, U.sub.S2, and U.sub.S3. are larger than zero at all times, as they must be, since the MESDCS 22 are only capable of generating positive voltages in the shown embodiment. However, since the phase-to-phase voltages U.sub.V1, U.sub.V2, and U.sub.V3 are based on differences between star voltages, any common offset to the star voltages does not have any effect on the phase-to-phase voltages, such that a higher constant offset would likewise be possible and lead to the same phase-to-phase voltages U.sub.V1, U.sub.V2, and U.sub.V3. However, for reasons presented above, when keeping both, the average and the peak star voltages low, the efficiency both with regard to the use of hardware as well as the energy efficiency can be increased. Accordingly, if a constant offset U0 is employed, it should indeed be chosen such that each star voltage U.sub.Sj is equal to 0 once per period T, or differs from zero once per period T by less than 15%, preferably less than 10%, and most preferably less than 5% of the maximum voltage value it acquires during each period.

[0157] Note that although reference has so far been made to sinusoidal star voltages and phase-to-phase voltages, the invention is not limited to this. Instead, according to the present invention, the star voltages U.sub.Sj can be any waveforms, as long as they can be decomposed into a periodic bipolar AC function P.sub.j(t) and a unipolar offset U.sub.off(t) that is common to each star voltage Us.sub.j, such that Us.sub.j(t)=P.sub.j(t)+U.sub.off(t), wherein the absolute value of said common unipolar offset U.sub.off(t) is at all times t sufficiently high that Us.sub.j(t) is unipolar, and wherein the periodic bipolar AC functions P.sub.j(t) associated with different star voltages Us.sub.j are phase-shifted copies of each other such that for each integers i, j, chosen from [1, . . . , N] and k chosen from [1, . . . , N−1], P.sub.i(t)=P.sub.j(t+k.Math.T/N), wherein T is the period of said periodic bipolar AC function P.sub.j(t), and in particular, P.sub.i(t)=P.sub.j(t+(i−j).Math.T/N).

[0158] In a further embodiment, the common offset is not a constant, but a time-varying function U.sub.off(t). This U.sub.off(t) can then be chosen such that the star voltages U.sub.j never become negative, and that, for a desired phase-to-phase voltage U.sub.Vj, the star voltages U.sub.Sj, U.sub.Sj+1 reach a maximum value that is as low as possible. In the following, it is shown how such an optimum U.sub.off(t) can be derived for a sinusoidal three phase system.

[0159] The starting points of the derivation are the periodic biopolar AC functions P.sub.j(t) which contribute to the corresponding unipolar star voltages U.sub.Sj, and which are chosen to give the desired phase-to-phase voltages U.sub.Vj. In the present case, we obtain P.sub.j(t)=U.sub.o.Math.sin(ωt+(j−1).Math.2π/3). In a first step, a constant offset is determined, such that U.sub.Sj(t) is always larger than 0, and this constant offset is again U.sub.o, leading to the star voltages U.sub.Sj as shown in FIG. 10.

[0160] Then, in each time interval, one subtracts a further time-dependent component from the common offset, which sets the star voltage that currently has the lowest voltage to 0. In other words, for each point in time, one determines the star voltage (with constant shift U.sub.o) which has the lowest value, and subtracts this value, referred to as U.sub.s,min, from the constant common offset. Accordingly, a time dependent offset U.sub.off(t)=U.sub.o−U.sub.s,min(t) is obtained. Herein, U.sub.s,min(t) is a function that is always larger than or equal to 0.

[0161] An analytical derivation for the resulting voltage for the first star voltage U.sub.S1 will be presented next. With reference to FIG. 10, it is seen that in a first interval I (bounded by dashed vertical lines) ranging from ωt=−π/6 to π/2, U.sub.S3 is the star voltage with the lowest value, i. e. U.sub.S,min=U.sub.S3. Accordingly, in the range −π/6<ωt=π/2, we obtain:

[00004]$\begin{array}{c}{U}_{S1\left(t\right)}\left(t\right)/U_0=\left(1+\sin \left(\omega t\right)\right)-\left(1+\sin \left(\omega t+\frac{4\pi }{3}\right)\right)\\ =\sin \left(\omega t\right)-\sin \left(\omega t+\frac{4\pi }{3}\right)\\ =\sqrt{3}.Math.\sin \left(\omega t+\frac{\pi }{6}\right)\end{array}$

[0162] In a second region II ranging from

[00005]$\omega t=\frac{\pi }{2}\mathrm{to}7\frac{\pi }{6},$

we obtain:

[00006]$\begin{array}{c}{U}_{S1\left(t\right)}\left(t\right)/U_0=\left(1+\sin \left(\omega t\right)\right)-\left(1+\sin \left(\omega t+\frac{2\pi }{3}\right)\right)\\ =\sin \left(\omega t\right)-\sin \left(\omega t+\frac{2\pi }{3}\right)\\ =\sqrt{3}.Math.\sin \left(\omega t-\frac{\pi }{6}\right)\end{array}$

[0163] Finally, in the third region III from

[00007]$\omega t=7\frac{\pi }{6}\mathrm{to}11\frac{\pi }{6},$

we obtain:

[00008]$\begin{array}{c}{U}_{S1\left(t\right)}\left(t\right)/U_0=\left(1+\sin \left(\omega t\right)\right)-\left(1+\sin \left(\omega t+\frac{0\pi }{3}\right)\right)\\ =\sin \left(\omega t\right)-\sin \left(\omega t+\frac{0\pi }{3}\right)\\ =0\end{array}$

[0164] In other words, in interval III, U.sub.S1=0. The result is shown in FIG. 13, where U.sub.S1(t) has again been highlighted by larger line thickness. The waveforms for U.sub.S2 (t) und U.sub.S3 (t) can be obtained in a similar manner, likewise shown in FIG. 13, and are simply phase-shifted copies thereof. Accordingly, by comparison of FIGS. 13 and 10, it can be seen that the amplitude of the star voltages U.sub.Sj can be reduced from 2U.sub.o to U.sub.o √{square root over (3)}, or in other words to 86.6% of the voltage amplitude that is necessary when using a constant of said U.sub.o only. This means that a lower number of modules to is necessary to obtain the same phase-to-phase voltages. In addition, the average star voltage U.sub.Sj that is to be generated by the MESDCS 22 per cycle is lowered.

[0165] For the same MESDCS 22 design, this means that there are more often possibilities to connect modules in parallel, which means that losses, and in particular battery losses can be reduced.

[0166] The corresponding MESDCS power values Power.sub.j(t) using the optimized common offset U.sub.off(t) as shown in FIG. 16. Comparing this with the power diagrams of FIG. 12, it is seen that the negative portions of the power is reduced, which is indicative of reduced battery losses.

[0167] A closer inspection reveals that reducing the peak voltage as compared to the constant offset case by 13.4% actually leads to a power saving of close to ⅓ as compared to the constant offset case, as will be explained next. It is assumed that the loads are purely ohmic and that the losses in the semiconductor switches are negligible as compared to the battery losses, and it is further assumed that the batteries themselves can be regarded as having ohmic resistance. Moreover, as was mentioned before, the battery losses can be reduced if batteries are connected in parallel. The smaller the star voltage U.sub.Sj needed at any point in time, the more modules to can be connected in parallel. Rather than basing the analysis on any finite number of modules, one can to a good approximation assume that the internal resistance of the converter arm is inversely proportional to the square of the voltage applied across the converter arm. Numerical simulations carried out by the inventors show that by using the optimized common offset voltage U.sub.off(t), as compared to the constant offset U.sub.o, but leaving everything else the same, the battery losses can be reduced to 77.3%, simply due to the fact that modules can be connected in parallel more often.

[0168] However, when adapting the design of the MESDCS to the reduced star voltage amplitude, the losses can be reduced even further. Since the peak star voltages are reduced, the number of modules may likewise be reduced by the same factor √{square root over (3)}/2. However, since the total capacity should be kept constant, the capacity of the individual module battery would be increased by the factor 2/√{square root over (3)}, and the internal resistance of the converter arm and the battery losses is reduced by the same factor. It follows that by using the optimized star voltages U.sub.Sj with a time dependent common offset U.sub.off(t) as described above, the losses can be reduced by a factor

[00009]$F=0,773.Math.\frac{\sqrt{3}}{2}=0,6694$

[0169] to 66.9% of the losses that would be obtained when using a constant offset U.sub.o.

[0170] FIG. 14 shows the power loss using the optimized common offset U.sub.off(t), while FIG. 15 shows the losses when using the constant offset U.sub.o. As was mentioned above, FIG. 16 shows the power values Power.sub.j(t) of each MESDCS in operation using the optimized common offset to U.sub.off(t). In comparing this to the power values of FIG. 12, it is seen that in this case, the size of the regions with negative power are significantly reduced for the optimized common offset U.sub.off(t).

[0171] Obviously, best results can be obtained if indeed, the time dependent common offset U.sub.off(t) is chosen such that at each instance in time, one of said star voltages U.sub.Sj is equal to 0. However, improvements can also be obtained if this criterion is relaxed. In preferred embodiments, however, U.sub.off(t) is chosen such that at each instance in time, one of said star voltages differs from 0 by less than 15%, preferably less than 10% and most preferably less than 5% of the maximum absolute voltage value it requires during each period T.

[0172] So far, only a system 22 has been described in which each MESDCS 22 can generate unipolar star voltages U.sub.Sj only. This situation is again summarized schematically in FIG. 17, where the unipolar voltages are symbolized by arrows which are all pointing upward.

[0173] However, the considerations presented above can also be used in applications where the MESDCS 22, according to a second aspect of the present invention, are capable of generating bipolar star voltages U.sub.Sj.

[0174] One such system is schematically shown in FIG. 18, where each MESDCS 22 is comprised of two parts, 22a and 22b, which can each only generate a unipolar star voltage contribution, but with opposite polarities: The upper part 22a in FIG. 18 can only generate positive star voltages while the lower part 22b can only generate negative star voltages. Accordingly, if a negative star voltage is required, all modules of the upper part 22a would be in bypass mode, and if a positive star voltage is required, all modules of the lower part 22b would be in bypass mode. The individual parts 22a and 22b of the MESDCS 22 could then still be made up of two-quadrant modules, similar to those shown in FIGS. 3 and 4.

[0175] Moreover, FIG. 19 shows a case where each MESDCS 22 is capable of generating bipolar star voltages U.sub.Sj, as is symbolically indicated by the arrows showing in both directions. Such MESDCS could be based on four-quadrant modules, such as the ones shown in FIGS. 1 and 2, or by two-quadrant-modules, which are provided with an additional polarity reversing circuit as discussed in the introductory part of the specification.

[0176] According to this second aspect, in one embodiment it is assumed that the star voltages U.sub.Sj are phase-shifted periodic functions, for example phase-shifted sinus functions, but provided with a time-dependent common offset U.sub.off(t) which allows for reducing the overall amplitude of the star voltages while resulting in the same phase-to-phase voltages U.sub.Vj that would be obtained without such time-dependent common offset U.sub.off(t). An example for this is shown in FIG. 20, where the star voltages are given as:

U.sub.S1(t)=U.sub.o.Math.sin(ωt)−(U.sub.MAX−U.sub.MIN)/2

U.sub.S2(t)=U.sub.o.Math.sin(ωt+2π/3)−(U.sub.MAX−U.sub.MIN)/2

U.sub.S3(t)=U.sub.o.Math.sin(ωt+4π/3)−(U.sub.MAX−U.sub.MIN)/2

[0177] Herein, U.sub.MAX is at every point in time the maximum of the three phase shifted sinus functions, and U.sub.MIN is the absolute value of the minimum of the three phase shifted sinus functions.

[0178] Accordingly, with the more general terminology introduced in the introductory portion of the specification, one can define a common offset U.sub.f(t)=−(Max(P.sub.j(t))+Min(P.sub.j(t)))/2, where Max(P.sub.j(t)) is defined as the maximum value among each of the functions P.sub.j(t) at a given time instance t, and Min(P.sub.j(t)) is defined as the minimum value among each of the functions P.sub.j(t) at a given time instance t.

[0179] It is seen that in the case of three phases and sinus waveforms, the maximum voltage to be provided by each MESDCS is reduced by a factor of ½.Math.√{square root over (3)}=0.866, as compared to the case without the time-dependent offset U.sub.off(t). This means that again, the MESDCS can be used with a smaller number of modules, or can be operated with reduced module voltages.

[0180] Moreover, without changing the design of the MESDCS, modules can be connected in parallel more often, such that only by this measure, the battery losses can be reduced by 11%. It is further emphasized that this second aspect of the invention can be very easily applied by only modifying the control provided by the control system 30.

[0181] Clearly, there are many possibilities to generate time-dependent offset function U.sub.off(t) that allow for reducing the voltage amplitudes of the MESDCS to be less than twice the amplitude of the sinus function.

[0182] A further suitable definition of U.sub.off(t) for this purpose was present in the summary of the invention as follows:

U.sub.off(t)=L−Max(P.sub.j(t)), if Max(P.sub.j(t))>L, and

U.sub.off(t)=−L−Min(P.sub.j(t)), if Min(P.sub.j(t))<−L,

[0183] wherein Max(P.sub.j(t)) is defined as the maximum value among each of the functions P.sub.j(t) at a given time instance t, and Min(P.sub.j(t)) is defined as the minimum value among each of the functions P.sub.j(t) at a given time instance t, and L is a constant. In the important case N=3 and P.sub.j(t)=A.Math.sin(ωt+(j−1).Math.2π/N), L=A.Math.√{square root over (3)}/2.

[0184] It is however emphasized that due to the fact that using the MESDCS 22, practically any star voltage waveforms can be generated, it is easily possible to provide other offset functions, different from the ones described herein, which can readily be implemented without any modifications to the hardware, but only by revising the control of the control unit 30 accordingly.

[0185] In FIG. 21, a module to similar to that of FIG. 3 is shown in detail. As shown therein, the energy storage element 16 is formed by a battery having in this case a nominal voltage of 3.7 V only. The battery 16 has a first pole 16a connected with one of the first terminals 12, referenced as “A” in FIG. 21, and a second pole 16b connected with the other one of the first terminals 12 (“B”). The module comprises [0186] a first switch S1 having a first switch terminal connected with one of the second terminals 14 (“D”) of the module 10 and a second switch terminal 36 connected with the first pole 16a of said battery 16, [0187] a second switch S2 having a first switch terminal 34 connected with the other one of the second terminals 14 (“C”) of the module 10 and a second switch terminal 36 connected with the second pole 16b of said battery 16, and [0188] a third switch S3 provided in a connection line 38 connecting the first switch terminal 34 of the first switch S1 and the second switch terminal 36 of the second switch S2.

[0189] Herein, the term “switch terminal” was introduced merely to define “points” to either side of the switch such as to better describe the structure of the module. The two terminals of the switch define points which are conductively connected when the switch is closed. However, the term “switch terminal” does not imply any specific physical entity or structural element. The module 10 allows for connecting its battery 16 in series with the battery 16 of an adjacent module 10 connected to its second terminals 14 by opening the first and second switches S1, S2 and closing the third switch S3, and for connecting its battery 16 in parallel with the battery 16 of an adjacent module 10 connected to its second terminals 14 by closing the first and second switches S1, S2 and opening the third switch S3.

[0190] Different from what is shown in FIG. 21, the module 10 may further comprise a fourth switch connected either between the first pole 16a of said battery 16 and the second switch terminal 36 of the first switch S1, or between the second pole 16b of said battery 16 and the second switch terminal 36 of the second switch S2. Such forth switch allows for deactivating or bypassing the battery 16 of a given module 10 in a series or parallel connection of batteries 16 of adjacent modules 10 connected to the first terminals 12 and second terminals 14 of said given module 10, respectively.

[0191] Importantly, in the embodiment shown, the gate voltages of the switches S1, S2 and S3 are, at least in part, taken directly from the batteries 16, which in the embodiment shown have an unusually low voltage. In prior art MESDCS 22, the gate voltages for the switches would be provided via an external voltage source, but not from the batteries included in the modules. One reason why the skilled person would not have considered such a design is that in prior art, one would usually use modules 10 having batteries 16 with considerably higher voltages, for example having a nominal voltage of 50 V, such as to keep the total number of modules necessary for obtaining a desired total output voltage low. However, this means that in prior art modules, DC-DC converters would have to be used for down-converting the battery voltage to a suitable gate voltage for the semiconductor switch, such as a MOSFET switching device. Such DC-DC converters would add significantly to the costs of the module. A module having three switches 18 would require two DC-DC converters per module, and the eight-switch-module shown in FIG. 1 would even require five DC-DC converters.

[0192] In contrast to this, when the battery voltages are in the same range as the required gate voltages, one can almost completely dispense with DC-DC converters. In this case, the gate voltages can be supplied either directly from the battery, or via a voltage doubling device 40, which however is significantly cheaper than a DC-DC converter. The three-switch module shown in FIG. 21 requires only a single voltage doubling device 40, while the eight-switch-module of FIG. 1 would require for voltage doubling devices 40.

[0193] When using the module batteries 16 to supply the gate voltages, an obvious design would be to provide all of the switches 18; S1, S2 and S3 within a given module 10 by the same module's battery 16. However, in the embodiment shown in FIG. 21, it is seen that the switches S1 and S2 are actually powered by the battery 16 of a different module to, in this case the neighboring module 10 to the right. This leads to a fail-proof converter design, in which the converter arm as a whole remains functional even if one of the batteries 16, and all of the switches powered thereby fail completely, as is apparent from FIG. 22. FIG. 22 shows two full and part of a third module 10 of the type shown in FIG. 21, where it is assumed that the battery 16 in the module 10 in the middle breaks completely. This means that switch S3 of this middle module to, but also the switches S1 and S2 of the module 10 to the left are nonconducting, as is indicated by the crosses. However, since the switch S3 of the leftmost module remains active, and since the S1 in the middle module is powered by the battery 16 of the rightmost module, the module 10 in the middle can be effectively bypassed by the conductive path shown therein with the thick line.

LIST OF REFERENCE SIGNS

[0194] 10 module [0195] 12 first terminal [0196] 14 second terminal [0197] 16 energy storage device [0198] 18 switch [0199] 20 system for providing N bipolar AC phase voltages [0200] 22 MESDCS [0201] 24 first end of MESDCS [0202] 26 second end of MESDCS [0203] 28 floating connection point [0204] 30 control system [0205] 32 system for providing N bipolar AC phase voltages [0206] 34 first switch terminal [0207] 36 second switch terminal [0208] 38 connection line [0209] 40 voltage doubling device