Method for controlling a power converter

Abstract

The invention relates to a method for controlling a power converter whose regulation is ensured by power electronic components used in switching at a switching frequency, the converter being controlled by at least one control variable, the control variable being calculated at a control frequency, the switching frequency being an integer multiple of the control frequency, characterized in that the method comprises providing an estimate of the control variable at the switching frequency, outside the times at which the control variable is calculated. Figure for the abstract: Fig. 4

Description

Title of the invention: Method for controlling a power converter Technical field

[0001] The present invention relates to the field of power electronics, and in particular to a method of controlling a power converter.

[0002] Among the various topologies of AC / DC or DC / DC converters, DAB (Dual Active Bridge) converters are currently attracting great interest, particularly in the field of electric vehicles. A DAB converter is typically composed of two H-bridges that transfer power through a transformer.

[0003] DAB converters have a high power density, which is advantageous for battery charging, and they can operate at high switching frequencies, which allows for smaller components.

[0004] In a switching power converter, comprising a plurality of controlled switches, the transition from the on state to the off state, and vice versa, of each of the switches, must be carried out according to a certain sequence, and a certain delay between each switching, in order to transfer the maximum power during the conversion, while ensuring switching of the converters in ZVS mode ("Zero Voltage Switching" or zero voltage switching).

[0005] Thus, methods for controlling the switches have been designed, making it possible to obtain the desired level of performance for each converter topology.

[0006] Examples of methods for controlling a converter comprising two H-bridges and connected by a transformer have been described in [1], [2] and [3].

[0007] The computers used to control the switching of the switches must be able to calculate the switching commands in real time, based on the control variables measured at the converter level. These computers are generally embedded in the charger for electric vehicles.

[0008] However, in the context of the development of three-phase power supplies for electric vehicles, which makes it possible to triple the power supplied, the integration of three computers with constant charger volume is currently difficult to envisage.

[0009] Thus, it was considered to sub-sample the control of the switches. By performing a sub-sampling of the command, the computer has a real-time calculation constraint which is alleviated, which provides a gain in volume of the computer.

[0010] The principle of command subsampling in a DAB converter is

[0011]

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020] described in [4]. Undersampling involves blocking the commands until the next control calculation time. Since the input voltage varies (for an AC voltage to be converted), as does the power setpoint, the converter will deviate from the nominal operation between two command calculation times, especially since the decimation factor is large. There is therefore a need to mitigate the effects of command subsampling, at reduced (or even negligible) computational cost compared to nominal sampling. Summary of the invention An object of the invention is therefore a method for controlling a power converter whose regulation is ensured by electronic power components used in switching at a switching frequency, the converter being controlled by at least one control variable, the control variable being calculated at a control frequency, the switching frequency being an integer multiple of the control frequency, characterized in that the method comprises providing an estimate of the control variable at the switching frequency, outside the times of calculation of the control variable. Advantageously, the estimation of the control variable is carried out with a first-order prediction model. Advantageously, the power converter comprising two H-bridges connected by a transformer, the estimation of the control variable is defined by the relations: gh' 1 if i- [tv, [itN p [ î 1 — Otherwise ct & H corresponding respectively to an estimate of the control variable at time iTsw and at time (i - l)rsw, where Tsw is the inverse of the switching frequency (f ] g[ / ] corresponding to a calculated value of the control variable (g) at time zTsw v corresponding to the ratio between the switching frequency (y J and the control frequency (y ], 7j [ / ] and [ / - 1 ] corresponding respectively to the values ​​of the voltage at the terminals of the transformer input at time iT^ and at time (i- l)rsw, / ] corresponding to a parameter calculated as a function of the last two calculated values ​​g [ i ] and g [ i - y ], g [ i - v ] corresponding to a calculated value of the variable

[0022]

[0023]

[0024]

[0025]

[0026]

[0027]

[0028]

[0029]

[0030]

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] control (g) at time > and the two values ​​of the voltage across the terminals of the transformer input measured at the same times. Advantageously, the parameter âr[ z] is determined by the following relation: z-1] otherwise Advantageously, the power converter comprising two H-bridges connected by a transformer, the estimation (g) of the control variable is defined by the relations: [ Sa [ * ] if i - fiv, / / GN I-1]+aa_v[i(v![i: ] - Vi[i' - 1 ]) otherwise g [ z] corresponding to an analytical expression of the control variable (g) v corresponding to the ratio between the switching frequency yj and the control frequency yj; g [ / ] and g [ i -1 ] corresponding respectively to an estimate of the control variable at time iTsw and at time (z - 1)TSW, where Tsw is the inverse of the switching frequency (f ); aa v[ i ] corresponding to an adaptive parameter evaluated by an analytical expression; Vj[ z ] and V| [ i - 1 ] corresponding respectively to the values ​​of the voltage at the terminals of the transformer input at time z'Tsw and at time (i- 1)TSW. Advantageously, the control variable includes a phase shift variable between the bridges. Advantageously, the phase shift variable between the bridges comprises a phase shift between the falling edge of the voltage across the transformer input and the falling edge of the voltage across the transformer output. Advantageously, the whole multiple is equal to 2 or 4. The invention also relates to a DAB type power converter, capable of implementing the aforementioned method. The invention also relates to a three-phase charger, comprising a aforementioned power converter. Description of figures Other characteristics, details and advantages of the invention will emerge from reading the description given with reference to the appended drawings given by way of example. [Fig.l] illustrates a topology of a DAB FB-FB converter (for "Full Bridge-Full Bridge") without losses.

[0038] [Fig.2] illustrates a type 1 modulation waveform in boost mode for AC / DC power transfer.

[0039] [Fig.3] illustrates a type 2 modulation waveform in boost mode for AC / DC power transfer.

[0040] [Fig.4] illustrates the different stages of the method according to the invention.

[0041] [Fig.5] illustrates control variable curves.

[0042] [Fig.6] illustrates transferred power curves.

[0043] [Fig.7] illustrates control variable and power error curves transferred.

[0044] The method according to the invention can be applied to any type of power converter whose regulation is ensured by electronic power components used in switching at a switching frequency. As an example, the method is described for a DAB FB-FB converter, illustrated by [Fig.l].

[0045] The converter CONV comprises an HP bridge^, receiving the input voltage V\. By H bridge, we mean a parallel association of at least two branches (branches BR^ and ^^for the H bridge Py) between two nodes or terminals (nodes ND{ and ND2 for the HP bridge\).

[0046] Each bridge branch is defined by an association of two switches electrically in series between the terminals of the branch. In the HP bridge, the branch BRj comprises switches K and K2 in series. The branch BR2 comprises switches KJ, and in series.

[0047] Node ND{ is located between switches K and K2 in series, and node ND2 is located between switches K3 and K4 in series.

[0048] Similarly, in the H-bridge Pi, branch BR3 includes switches K5 and K6 in series. Branch BR4 includes switches K7 and K8 in series.

[0049] Node ND3 is located between switches K5 and in series, and node ND4 is located between switches K2 and Ks in series.

[0050] The converter further comprises a capacitive element connecting the terminals B} and B2 of the bridge in HP\. The voltage L» is applied between the terminals B{ and B2 through an impedance Z^ The capacitive element C1 and the impedance Zx form a filter making it possible to limit variations, at each switching, of the voltage and / or the current supplied to the converter.

[0051] In the same way, the converter comprises a capacitive element C2 connecting the terminals B3 and B^ of the H-bridge P2. The voltage between the terminals B3 and B^ is supplied by the converter through an impedance Z2. The capacitive element C2 and the impedance Z2 form a filter making it possible to limit variations, at each com- mutation, of the voltage vout and / or of the current supplied by the converter.

[0052] The first bridge in HP\ and the second bridge in HP^ are connected by a transformer TR. The transformer TR comprises a winding EN iconnected to the nodes ND} and ND2, and a winding EN ^connected to the nodes NIT and ND4.

[0053] The transformer TR has a transformation ratio n (n: 1 by convention). The transformation ratio corresponds to a ratio between a number of turns of the second winding and a number of turns of the first winding.

[0054] The converter is considered to operate in step-up mode (boost mode), i.e. Vj < nV2 (m = / nV2 < 1). In buck mode (step-down, m > 1), the same results can be used by making the following change of variable: Vt nV2, V2-V1 / n.

[0055] The transformer TR comprises a leakage inductance C which connects a terminal of the winding EN^ to the node ND^. The inductance k is crossed by a current it, and the voltage across the inductance L is called vl.

[0056] Thus, in each bridge, the switching of the branches forms a switching sequence. The switching sequences of the two branches are repeated at the switching frequency f . At each repetition of the switching sequences, the leakage inductance has the role of storing / removing energy, so as to circulate this energy from one bridge to the other of the converter.

[0057] The leakage inductance L has at its terminals a variable voltage equivalent to a voltage vl at the terminals of the leakage inductance shown.

[0058] The switching sequences of the two branches are repeated at the switching frequency. At each repetition of the switching sequences, the leakage inductance L has the role of storing / removing energy, so as to circulate the energy from one bridge to the other of the converter.

[0059] The converter further comprises a control circuit (not visible in the figures), which receives values ​​representative of the respective voltages Vj and V2.

[0060]

[0100] The control circuit (or computer) provides control signals switches Kb K2, K3K4, K5, K6, K7, and Kg, based on modeled values ​​of power transferred from the first bridge Pi to the second bridge A, and from the values ​​of voltages Vj and V2.

[0061] Those skilled in the art may refer to one of the documents [1], [2] or [3], which describe in more detail the topology and control of a DAB FB-FB converter.

[0062] As illustrated in Figures 2 and 3, the voltage vaci on the primary side of transformer TR consists of positive and negative pulses of amplitude Similarly, the voltage ^acz on the secondary side of transformer TR consists of positive and negative pulses of amplitude v2.

[0063] In the modulation example of [Fig.2] (so-called type 1 modulation in mode

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083] boost for an AC / DC conversion), the transferable power is given by the relation: VjHVt z _ T P= 2Ü"(g'2g2+W'2w2) 8 corresponds to the delay between the falling edge (instant L) of the voltage on the primary side and the falling edge (instant t3 = Tsw / 2) of the voltage n-vAC2 on the secondary side. w corresponds to the delay between the rising edge (instant tp and the falling edge (instant 12) of the voltage on the primary side. By imposing the constraint iL(tj ) = Îl( h) (this 9^ amounts to imposing ZVS conditions of zero voltage switching identical to these switchings), we obtain: V[(l-2g) / nc \ w =.....=m(0.5-g) with m _ vi 111 ~ nV2 Thus, the converter control includes the determination of the control variable S, so that the transferable power P can be calculated. Which gives: p = ( ( -4-4m 2 ) g 2 + (4m 2 -2m + 2)g+mm 2 ) When fsw is fixed, S is given by: -b-Jb“-4ac g=........2a........ with a = 2- 2m2 > O b = -1 + m - 2m2 c — _ m + llL + 2fswLP 2 2 nm2Vf In the modulation example of [Fig.3] (type 2 modulation in boost mode for an AC / DC conversion), the transferable power is given by the relation: g corresponds to the delay between the falling edge (instant 12) of the voltage vxci on the primary side and the falling edge (instant t3 - Tsw / 2) of the voltage n-vAC2 on the secondary side. w corresponds to the delay between the rising edge (instant tj) and the falling edge (instant C) of the voltage vAci on the primary side. By imposing the constraint iL ( t? ) = Îl ( to ) (this 9^ amounts to imposing ZVS conditions of zero voltage switching identical to these switchings), we obtain: Vt(l-2g) x w _______ - m (0.5-g)

[0085] where m _ JO.

[0086] We therefore have: 100871 P= 4Cl ( l-2g)( lm( l-2g)-4g)

[0088] When fsw is fixed, » is given by:

[0089] a _ —jl.--- » 2a

[0090] with

[0091] a = 4(2-m)>0

[0092] b = 4m - 6

[0093] c=_4Mi.m+1 iïmPV;

[0094] The method according to the invention is not limited to the aforementioned modulation schemes, which are described only by way of example.

[0095] The developments will be illustrated for a triangular type modulation (see current at the bottom of Figure 3) with a constant switching frequency fsw. The method is applicable for other types of modulations, such as phase-shift and trapezoidal type modulations (see Figure 2), or with a variable switching frequency fsw.

[0096] It is essential, in the context of the invention, that the converter is controlled by at least one control variable, which may be, for example, the delay S between the falling edge of the voltage vaci on the primary side and the falling edge of the voltage ftvAC2 on the secondary side. Any other control variable may be used, provided that it corresponds to a delay between the voltage vaci on the primary side and the voltage n-vAC2 on the secondary side.

[0097] The method according to the invention is based on a sub-sampling of the control of the switches. Let f be the frequency of calculation of the control, i.e. the frequency of calculation of the control variables of the converter, where f = f / v with v the decimation factor, a strictly positive integer, e.g. v = 2, 4, 8, etc.

[0098] The fact of not calculating the control variable at the switching frequency fsw makes it possible to alleviate the real-time calculation constraint, thus making it possible to use a less bulky control circuit.

[0099] The larger the multiple 17, the greater the effect of the control subsampling. In Figures 5 and 6 described below, v = 8, in order to obtain illustrative graphs. In practice, a division of the switching frequency by 2 or 4 would be more desirable, so as not to excessively degrade the stability margins of the closed-loop system.

[0100] Instead of blocking commands until the next calculation time, the method according to the invention comprises a provision of an estimation g of the control variable at the switching frequency fsw, outside the times of calculation of the control variable 8; the missing values ​​of the control at the switching frequency fsw between two times of calculation of the control are estimated.

[0101] For the modulation considered, the relationship which links the interval S between two falling edges to the voltage V] between terminals B, and B2 is a non-linear algebraic relationship. The voltage Vj is assumed to be known, and can be measured or estimated, for example with a phase-locked loop (PLL) for an alternating voltage vin

[0102] It is assumed that the other quantities involved in the calculation of S are constant or slowly variable (for example the voltage V2 between terminals B3 and B4, so that the first-order Taylor development is written, for the function g(Vj):

[0103] / \ / of I »(V l + AV1j«g(V1) + ^.AV!

[0104] where AVj is a small variation around the voltage Vj for a given operating point.

[0105] The control variable differential Ag is defined by the following relation:

[0106] Ag^-^AV^ctAV! 0 dVj ii

[0107] The estimation g of the control variable is carried out with a first-order prediction model, which gives very satisfactory results and simple calculations. The developments remain valid for higher orders and increased precision, at the cost of a greater computational load.

[0108] The method according to the invention proposes to estimate a and to calculate g at the control frequency fc, and to estimate g at the frequency fsw (outside the instants of calculation of g to fc).

[0109] The method is described in more detail in the flowchart illustrated by [Fig.4].

[0110] In the following, we will note A ^ / 7"^ (& £ ^C) for a sampled variable x dependent on time.

[0111] The control variable 8 is calculated every v periods Tsw: i = f.iv, E

[0112] Thus, a computer counter increments at the period Tc = vTsw

[0113] At the first step SI, we fix:

[0114] / / = 1, i = fiv, / 7 E

[0115] ^[0toM]-g[0]

[0116] The test Tl consists of determining whether i - / / V.

[0117] If the result of test T1 is negative, step S5 is implemented:

[0118] The estimated value at v [ i ] corresponds to the value estimated at the last increment of P.

[0119] âv[z] = âv[z-1] (that is to say âv[z] = )

[0120] Furthermore, if the result of the test Tl is negative, the estimate g[z] of the control variable is calculated as follows:

[0121] g[f] =g[z- 1] - VJz- 1])

[0122] Vji] and V] [i - 1] correspond respectively to the value of the voltage at the terminals of the transformer input at time iTsw and at time (i - 1)TSW.

[0123] The calculator then increments / to the next value; i — i + l (step S6).

[0124] If the result of test T1 is positive, step S2 is implemented. It consists of calculating the control variable g[z], in a manner known to those skilled in the art. For example, with the relationship defined previously for type 1 and 2 modulations:

[0125] -b-Jb2-4ac ë 2a

[0126] Thus, the calculation of » is performed at a control frequency / = fsw / v, which lightens the computational load by a factor v.

[0127] Step S2 also includes updating the estimate gp]:

[0128] £[<]=<,]

[0129] The estimate gp -1] is used for step S5,.

[0130] The next step S3 consists of determining the parameter âv[ z ] as a function of the last two calculated values ​​g [ i ] and g [ i - y ], g [ i - v ] corresponding to a calculated value of the control variable 8 at the instant (j _ , and of the two values ​​of the voltage across the transformer input measured at the same times, gp - y ] corresponds to the calculation of S at the previous increment of 101311

[0132] The next step S4 consists of an incrementation, by the calculator, of the value of A. The calculator then increments i to the following value: i = i + 1 (step S6).

[0133] After the execution of step S6, the method returns to the test TL II. It is understood that each iteration of the test T1 is executed at the switching frequency fsw.

[0134] Figures 5, 6 and 7 illustrate the benefits provided by the method according to the invention.

[0135] For this example, the operating point considered for the converter is Vj = 230 V RMS, V2 = 400V, 16 A RMS, n= 1, L = 4.5gF with an open-loop lossless model and switches considered perfect. The converter is driven at fsw = 100 kHz in modulation 2 (triangular type) in step-up mode (“boost” mode).

[0136] The left part of Figure 5 represents the control variable S which represents the phase shift between the falling edge of the voltage across the input of the transformer and the falling edge of the voltage across the transformer output, over a period of 10 ms. The left part of Figure 5 also represents the estimated control variable g using the method according to the invention, as well as the control variable 8^ obtained by blocking the control between two calculation times, without estimation.

[0137] The right part of Figure 5 represents an enlargement of the left part, over the interval [4 - 4.3 ms]. The estimated control variable £ is almost the same as the calculated value of the control variable 8 (the differences are very small), whereas the control variable 8}w[d obtained by blocking the control between two calculation times tends to deviate from the calculated value 8 at each time.

[0138] The same effect is visible in Figure 6, which represents the transferred power P from the first bridge P] to the second bridge Pi. On the right-hand part, which represents an enlargement of the left-hand part, over the interval [4 - 4.3 ms], the estimated power p is confused with the calculated value of the power P (see above for the calculation of the power), whereas the power Phold obtained by blocking the control between two calculation instants tends to deviate from the calculated value P at each instant.

[0139] Figure 7 illustrates the errors made on the control variable 8 (graphs on the left of Figure 7, error g_hold corresponds to the error between the theoretical control variable 8 and the sub-sampled control variable, according to the state of the art; err°rg—est corresponds to the error between the theoretical control variable 8 and the control variable with the method according to the invention) and on the transferred power P (graphs on the right of Figure 7 error p_hold corresponds to the error between the theoretical transferred power P and the sub-sampled transferred power P, according to the state of the art; error p_est corresponds to the error between the theoretical transferred power P and the transferred power with the method according to the invention), in units and as a percentage.

[0140] A change of scale, on the four graphs at the bottom, allows for a better evaluation of the estimation errors.

[0141] It appears that the maximum error on the control variable 8 is 0.07% with the method according to the invention, whereas it is 5.2% with the blocking between two calculation instants, according to the state of the art. It can be noted that the prediction model being linear, the error increases with the curvature of g — f(y j.

[0142] It also appears that the maximum error on the transferred power P is 0.14% with the method according to the invention, whereas it is 46% with the blocking between two calculation instants, according to the state of the art. In [Fig.7], top right, the power is cancelled at the ends of the interval [0, 10] ms. Thus, the calculation of the percentage error is restricted to powers greater than 10 W, i.e. in the interval [0.12 - 9.88] ms.

[0143] Thus, the method according to the invention makes it possible to apply a sub-sampling of the calculation of the command for the switches of each of the converter bridges, while significantly reducing errors generated by subsampling. The process thus makes it possible to propose high-frequency switching of switches, with smaller calculator components, or in a dual way, for a load of fixed calculation, to increase the switching frequency.

[0144] According to an alternative embodiment, when an analytical expression of the control variable 8 is available, the parameter a can be evaluated and calculated, at place to be estimated, as is the case in the previously cited embodiment. Md - So this is an advantage because the prediction model is more accurate.

[0145] The estimate g of the control variable is defined by the relations: sii = ^v, -1] + aa_y [i] (V [i] - V. [i -1]) otherwise

[0146] #aU] corresponds to an analytical expression of the control variable g

[0147] v corresponds to the ratio between the switching frequency fsw and the control frequency fc.

[0148] v[i] corresponds to an adaptive parameter evaluated by an analytical expression.

[0149] V] [i] and Vf[i- 1] correspond respectively to the values ​​of the voltage at the terminals of the transformer input at the instant iTsw and at the instant (i - 1)TSW.

[0150] In 2 boost modulation, the analytical expressions of 8a and aa are given below:

[0151]

[0152] 4 V -6 V, n+V, n 1 2 - V____________vÇù____________ 8Vrl6V.n » (2 Vf V2+V}2 n p+32 L V2f^ p) 4>. ' 2 V

[0153] Where:

[0154] Ùv\Vt2-64Lf PV,n+V28LV,f pt£ J 2 i / .ÇÇV * 1 ZJ 1

[0155] The analytical expressions could be determined for other modulation schemes.

[0156] These calculations can be complex, inducing a significant computational load. Nevertheless, the determinations of the analytical expressions can be simplified by using lookup tables, in which the values ​​of the analytical expressions of 8 eta would be stored.

[0157] References cited:

[0158] [1] EP 3 910 774 Al

[0159] [2] EP 3 910 775 Al

[0160] [3] EP 3 910 776 Al

[0161] [4] AAP Machado and IA Pires, “Undersampling Control for a DC-DC Boost and Forward Dual Active Bridge for a Single-Phase Grid-Connected PV Converter”, J. Control Autom. Electr. Syst., vol. 29, no. 5, p. 650-659, Oct. 2018, doi: 10.1007 / s40313-018-0404-9.

Claims

Claims ^Claim 1] Method for controlling a power converter whose regulation is ensured by electronic power components used in switching at a switching frequency (f ), the converter being controlled by at least one control variable (g), the control variable (g) being calculated at a control frequency (f ), the switching frequency (f ) being an integer multiple of the control frequency (f J, characterized in that the method comprises a provision of an estimate (g) of the control variable at the switching frequency (f outside the instants of calculation of the control variable Cg). ^Claim 2] Method according to claim 1, in which the estimate (g) of the control variable is carried out with a first-order prediction model. ^Claim 3] 3. Method according to one of the preceding claims, in which,the power converter comprising two H-bridges connected by a transformer, the estimate (£) of the control variable is defined by the relations: g[z] ifi = gv, g[z-1] -Vjz-1]) otherwise and - 1] corresponding respectively to an estimate of the control variable at time iTsw and at time (i - 1)TSW, where is the inverse of the switching frequency (fjg[z] corresponding to a calculated value of the control variable Cg) at time zTsw v corresponding to the ratio between the switching frequency (f ) and the control frequency (y J, Vj f] and Vj i - 1 ] corresponding respectively to the values ​​of the voltage across the transformer input at time ÎT^, and at time (i - âv[ i ] corresponding to a parameter calculated as a function of the last two calculated values ​​g [ i] and g [ i - v ],g [ i - v] corresponding to a calculated value of the control variable (g) at time (i_ v)T \ • SW and the two values ​​of the voltage across the input of the trans-, trainer measured at the same times.

4. 4. Method according to claim 3, in which the parameter âv[ j] is determined by the following relation: «[t] -,?[«'] , ri vrivr i ^11= PV âv[z] = âv[zl] otherwise

5. 5. Method according to claim 1, in which, the power converter comprising two H-bridges connected by a transformer, the estimation (g) of the control variable is defined by the relations: g [ i 1 if i - / / V, fl GN ÿ -1] + aa_vU ](Vj [ z ] - [ z -1 ]) otherwise g [z] corresponding to an analytical expression of the control variable (g), v corresponding to the ratio between the switching frequency ( f J and the control frequency ( / ); g [ z ] and g [ z - 1 ] corresponding respectively to an estimation of the control variable at the instant iTsw and at the instant (i- 1)TSW, where Tsw is the inverse of the switching frequency (f J ; aa v[ i ] corresponding to an adaptive parameter evaluated by an analytical expression; yj f] and Vj z - 1] corresponding respectively to the values ​​of the voltage across the transformer input at time iTsw and at time (z - 1)TSW.

6. Method according to one of the preceding claims, in which the control variable (g) comprises a phase shift variable between the bridges.

7. 7. The method of claim 6, wherein the phase shift variable between the bridges comprises a phase shift between the falling edge of the voltage across the transformer input and the falling edge of the voltage across the transformer output.

8. 8. Method according to one of the preceding claims, in which the integer multiple is equal to 2 or 4.

9. DAB type power converter, characterized in that it is capable of implementing the method according to one of the preceding claims.

10. 10. Three-phase charger, characterized in that it comprises a power converter according to claim 9.

Images