ON-THE-FLY TOPOLOGY MORPHING FOR FREQUENCY-MULTIPLIER HALF-BRIDGE OPERATION

Abstract

Full bridge converter-circuit having a H-bridge, comprising a control circuit for operating the H-bridge in a first operation mode or a second operation mode, wherein the control circuit is configured that between switching from the first operation mode to the second operation mode a morphing operation mode is applied in which a positive inverter voltage pulse length on a first half-bridge of the H-bridge is determined on the basis of a negative inverter voltage pulse length on a second half-bridge of the H-bridge.

Claims

1. A full bridge converter-circuit having an H-bridge, comprising: a control circuit for operating the H-bridge in a first operation mode or a second operation mode, wherein the control circuit is configured that between switching from the first operation mode to the second operation mode a morphing operation mode is applied in which a positive inverter voltage pulse length on a first half-bridge of the H-bridge is determined on the basis of a negative inverter voltage pulse length on a second half-bridge of the H-bridge.

2. The full bridge converter-circuit according to claim 1, wherein the first operation mode is a full bridge operation of the H-bridge and the second operation mode is a frequency multiplier half bridge operation of the H-bridge wherein switches are pulsed with a switching frequency that is at least twice a frequency of a resonant current or the first operation mode is a frequency multiplier half bridge operation of the H-bridge and the second operation mode is a full bridge operation of the H-bridge.

3. The full bridge converter-circuit according to claim 1, wherein the positive inverter voltage pulse length and/or the negative inverter voltage pulse length are adapted such that a duty cycle of a positive inverter equals to a duty cycle of an inverter and the duty cycle of a negative inverter equals to the duty cycle of a morphing operation.

4. The full bridge converter-circuit according to claim 1, wherein the control circuit is configured to increase or decrease the negative inverter voltage pulse length linearly during the morphing operation mode.

5. The full bridge converter-circuit according to claim 1, wherein the full bridge converter-circuit comprises a look up table, LUT, for looking up the positive inverter voltage pulse length on the basis of the negative inverter voltage pulse length.

6. The full bridge converter-circuit according to claim 1, wherein the positive inverter voltage pulse length is adjusted based on measurements of a magnetic field sensor or the positive inverter voltage pulse length is adjusted based on online calculations.

7. The full bridge converter-circuit according to claim 5, wherein a functional graph representing a dependency of the positive inverter voltage pulse length on the negative inverter voltage pulse length has a minimum.

8. The full bridge converter-circuit according to claim 7, wherein the minimum is between a value of 0.2 and 0.3 of the negative inverter voltage pulse length and/or between a value of 0.2 and 0.3 of the positive inverter voltage pulse length.

9. The full bridge converter-circuit according to claim 7, wherein the functional graph is V-shaped.

10. The full bridge converter-circuit according to claim 1, wherein to a value of 0 of the negative inverter voltage pulse length a value of 0.5 of the positive inverter voltage pulse length is assigned and/or to a value of 0.5 of the negative inverter voltage pulse length a value of 0.5 of the positive inverter voltage pulse length is assigned.

11. The full bridge converter-circuit according to claim 1, wherein the full bridge converter-circuit is an LLC resonant converter.

12. A converting method for a full bridge converter-circuit having an H-bridge, comprising the steps: operating the H-bridge in a first operation mode; operating the H-bridge in a morphing operation mode in which a positive inverter voltage pulse length on a first half-bridge of the H-bridge is determined on the basis of a negative inverter voltage pulse length on a second half-bridge of the H-bridge; and operating the H-bridge in a second operation mode.

13. The converting method according to claim 12, wherein the negative inverter voltage pulse length is increased or decreased linearly during the morphing operation mode.

14. The converting method according to claim 12, wherein the positive inverter voltage pulse length is stored in a look up table, LUT, on the basis of the negative inverter voltage pulse length.

15. The converting method according to claim 14, wherein the LUT is stored in a digital memory of a control circuit.

16. The converting method according to claim 14, wherein the LUT is derived from simulations in which a pulse length is adjusted to control a magnetizing current.

17. The converting method according to claim 12, wherein the first operation mode is a full bridge operation of the H-bridge and the second operation mode is a frequency multiplier half bridge operation of the H-bridge where switches are pulsed with a switching frequency that is at least twice a frequency of a resonant current or the first operation mode is a frequency multiplier half bridge operation of the H-bridge and the second operation mode is a full bridge operation of the H-bridge.

18. The converting method according to claim 12, wherein the positive inverter voltage pulse length and/or the negative inverter voltage pulse length are adapted such that a duty cycle of a positive inverter equals to a duty cycle of an inverter and the duty cycle of a negative inverter equals to the duty cycle of a morphing operation.

19. The converting method according to claim 12, wherein the positive inverter voltage pulse length is adjusted based on measurements of a magnetic field sensor or the positive inverter voltage pulse length is adjusted based on online calculations.

20. The converting method according to claim 12, wherein a functional graph representing a dependency of the positive inverter voltage pulse length on the negative inverter voltage pulse length has a minimum.

Description

BRIEF DESCRIPTION OF DRAWINGS

[0026] The drawings used to explain the embodiments show:

[0027] FIG. 1 is a circuit diagram of an LLC resonant converter;

[0028] FIG. 2A to FIG. 2D are four possible half bridge modulations for an LLC resonant converter;

[0029] FIG. 3A and FIG. 3B are frequency multiplied half bridge modulations employed on the LLC converter;

[0030] FIG. 4 is a loss analysis assuming a constant on-state resistance;

[0031] FIG. 5A and FIG. 5B are experimental measurement results of the frequency multiplied half bridge modulation;

[0032] FIG. 6A and FIG. 6B are half-bridge configuration with half bridge modulations and frequency multiplied half bridge modulation;

[0033] FIG. 7A to FIG. 7C are on-the-fly morphing transition from full-bridge modulation to half-bridge configuration;

[0034] FIG. 8A and FIG. 8B are experimental measurement results of an alternating asymmetrical morphing transition;

[0035] FIG. 9A to FIG. 9C are experimental measurement results of a symmetrical morphing transition;

[0036] FIG. 10 is an improved morphing operation from full-bridge to FD-HBM;

[0037] FIG. 11 is a functional graph representing the dependency of duty cycles;

[0038] FIG. 12 is a conventional morphing transition with a constant duty cycle;

[0039] FIG. 13 is an improved morphing transition with a dynamic duty cycle; and

[0040] FIG. 14 is a block diagram of a converting method.

[0041] In the figures, the same components are given the same reference symbols.

DESCRIPTION OF EMBODIMENTS

[0042] FIG. 1 shows a circuit diagram of an LLC resonant converter as a full bridge converter-circuit 100 having an H-Bridge 101. The H-Bridge 101 comprises a first half bridge 109-1 having the switches S.sub.1 and S.sub.3 and a second half bridge 109-2 having the switches S.sub.2 and S.sub.4. A control circuit 103 serves for controlling each of the switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4.

[0043] The switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 are implemented by suitable Metal-Oxide-Semiconductor Field-Effect Transistors (MOSFET).

[0044] Conventional LLC converters are not suitable for a wide voltage-transfer ratio as it results in a wide switching frequency range. To ensure a sufficient gain, the inductance ratio L.sub.r/L.sub.m needs to be designed small leading to large resonant currents that have a detrimental effect on the efficiency. Therefore, the LLC resonant converter is best applied near resonance. However, to cover a wide voltage-transfer ratio, an LLC full-bridge converter can be operated in half-bridge operation mode instead to a full bridge operation mode. In this case one of the switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 is turned on continuously, whereas the other corresponding switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 of the respective bridge leg is permanently turned off. This, however, leads to large conduction losses of the permanently turned-on switch (no switching losses) while the switch that is turned off exhibits no losses. Each switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 that is pulsed, on the other hand, exhibits half the conduction losses of the switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 that is permanently turned on and additional switching losses.

[0045] The full-bridge converter 100 with a blocking capacitor in the resonant path is operated in half-bridge mode to decrease the gain of this specific mode and, thus, increase the gain range of the converter. While in the conventional full-bridge operation the inverter voltage is pulsed between v.sub.AB∈{−V.sub.in;+V.sub.in}, in half-bridge operation mode the inverter voltage is pulsed between v.sub.AB∈{−V.sub.in;0} or v.sub.AB∈{0;V.sub.in}. The resonant capacitor Cr hereby acts as a blocking capacitor as the average voltage is half the input voltage. As the blocking capacitor is charged to half the input voltage, the effective input voltage of the transformer is reduced by a factor of two. This reduces the gain by a factor of two making this operation mode especially useful for small voltage-transfer ratios whereas for large voltage-transfer ratios the conventional full-bridge operation can be employed.

[0046] For intermediate gains and loads, phase-shift operation is beneficial to reduce the switching frequency and increase the efficiency. To change the operation mode from full-bridge operation to half-bridge operation and reverse, an on-the-fly topology morphing operation mode is applied between switching from the one mode to the other.

[0047] If the LLC converter 100 is operated in half-bridge operation mode, one switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 conducts the full resonant current while exhibiting no switching losses. The respective other switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 of this half-bridge 109-1 or 109-2 is permanently turned off such that this switch is not stressed at all. The switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 of the pulsing half bridge 109-1 or 109-2 conduct the resonant current for one half of a period such that they are stressed with half the conduction losses compared to the switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 that is permanently turned on. However, they exhibit turn-off losses.

[0048] This leads to imbalanced inverter losses. For dominant conduction losses, the switch that is permanently turned on is stressed with much larger losses than the pulsing switch.

[0049] If the switching losses are dominant, the pulsing switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 exhibit larger losses compared to the switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 that is permanently turned on.

[0050] The frequency-multiplier modulation (FMM) is a beneficial alternative to the conventional half-bridge modulation (HBM) to reduce the maximum junction temperature. While both half bridges 109-1 or 109-2 are pulsed with half the operating frequency, the resulting inverter voltage has a multiple integer of the frequency of the half bridges 109-1 or 109-2. The FMM can be applied for the conventional full-bridge LLC converter 100. As this modulation emulates the HBM it will be referred to as frequency-multiplier half-bridge modulation (frequency-multiplier half-bridge modulation, FM-HBM). This modulation is an alternative to the conventional HBM.

[0051] By employing the FM-HBM, the maximum junction temperature can be reduced significantly and should be the preferred modulation when employing half-bridge configuration. Furthermore, a simple-to-implement modulation is used to morph in operation on-the-fly from full-bridge to FM-HBM and back.

[0052] FIG. 2A to 2D show four possible HBMs for an LLC resonant converter 100. The switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 of a full-bridge inverter are switched to achieve four different switching states: in two energy-transfer (ET) states energy may be transferred from the input to the resonant tank or the output. In the other two freewheeling (FW) states, the inverter voltage is zero such that no energy is transferred from the input to the resonant tank or the output. The following list shows all of the four switching states with the appropriate switches that are turned on: [0053] ET+: S.sub.1 and S.sub.4.fwdarw.v.sub.AB=V.sub.in [0054] ET−: S.sub.2 and S.sub.3.fwdarw.vAB=−Vin [0055] FW+: S.sub.1 and S.sub.2.fwdarw.vAB=0 [0056] FW−: S.sub.3 and S.sub.2.fwdarw.vAB=0

[0057] When employing the conventional HBM, one leg of the inverter is not pulsed—one switch is permanently turned on while the other switch is permanently turned off. All potential HBM schemes are depicted in FIG. 2A to FIG. 2D. Intervals with v.sub.AB=0 are depicted in blue for FW.sup.+-intervals and in red for FW.sup.−-intervals.

[0058] FIG. 2A and FIG. 2B show modulations that result in a positive (or zero) inverter voltage (the average resonant capacitor voltage is v.sub.res=½V.sub.in), whereas FIG. 2C and FIG. 2D show modulations that result in only negative (or zero) inverter voltages (the average resonant capacitor voltage is v.sub.res=−½V.sub.in). Modulations in FIG. 2B and FIG. 2C cannot be achieved with bootstrap drivers since the high-side switches cannot be permanently turned on with this driver supply. The modulations in FIG. 2A and FIG. 2B as well as modulations in FIG. 2C and FIG. 2D result in an equal inverter output voltage v.sub.AB and, thus, in equal resonant currents. Assuming that MOSFETs are employed as the inverter switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4, the losses of the individual switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 that are pulsed can be calculated as:

[00001]$\begin{array}{cc}{P}_{\mathrm{compl}}^{\mathrm{HBM}}=\text{?}& \left(1\right)\end{array}$$\text{?}\text{indicates text missing or illegible when filed}$

[0059] To evaluate the distribution of the losses, the losses of these switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 are divided into the conduction losses P.sub.cond,compl.sup.HBM and the switching losses P.sub.sw,compl.sup.HBM. The switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 that is permanently turned on is stressed with the full resonant current such that the losses can be calculated as:

P.sub.p-on.sup.HBM=I.sub.res.sup.2R.sub.ds,on(ϑ.sub.j)   (2)

[0060] Assuming that the same MOSFET type is employed for all switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 and neglecting temperature dependencies of the on-state resistance R.sub.ds,on, the overall inverter losses can be calculated as P.sub.inv=2I.sub.res.sup.2R.sub.ds,on+2f.sub.swE.sub.off(i.sub.off). The conduction losses of the switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 that is permanently turned on are twice as large as the conduction losses of the pulsed switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 (P.sub.p-on.sup.HBM=2P.sub.cond,compl.sup.HBM). This may be a problem for converters that employ switches where conduction losses are the dominant loss contributor—in this case the switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 are designed for a lower on-state resistance, which results in larger conversion costs.

[0061] From the afore described FIG. 2 it is evident that two modulations achieve exactly the same inverter voltage. Therefore, it is possible to interchange the freewheeling intervals FW.sup.+ and FW.sup.− as depicted in FIG. 3A and FIG. 3B without modifying the inverter voltage v.sub.AB. By including both FW.sup.+ and FW.sup.− intervals into the modulation, the switching frequency of S.sub.1 and S.sub.2 reduces while the operating frequency of the LLC remains the same. The operating frequency is, thus, multiplied compared to the switching frequency of the individual switches. Therefore, these modulations are labeled frequency-multiplier modulation or frequency-multiplier half-bridge modulation.

[0062] FIG. 3A and FIG. 3B show a frequency-doubler half-bridge modulation FD-HBM with a switching frequency of half the operating frequency. The FD-HBM is a specific example of the FM-HBM that is employed on the LLC converter 100. The modulation consists of intervals of the length 3 T/2 and T/2, which are phase shifted for one bridge leg by T. The pulses of the switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 thus repeat their pattern every 2T.

[0063] Employing this concept, asymmetrical pulse patterns result for both half bridges 109-1 and 109-2, which are phase-shifted by T. Considering FIG. 3A and FIG. 3B, the modulation utilizes an up-down counter to achieve the pulse pattern. It is equally possible to employ any other counter. However, an up-down counter is beneficial in an on-the-fly morphing operation and is also employed to achieve balanced losses for an operation in phase-shift mode. This avoids the necessity of a PWM reconfiguration.

[0064] Two distinct types of switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 can be distinguished: the switches that are turned on 75% of the time, resulting in high losses and the switches that are turned on for the remaining 25% of the time, resulting in lower losses. The high losses P.sub.high.sup.FD-HBM of the switches that are turned on 75% of the time can be calculated as

[00002]$\begin{array}{cc}{P}_{\mathrm{high}}^{\mathrm{FD}-\mathrm{HBM}}=\text{?}& \left(3\right)\end{array}$$\text{?}\text{indicates text missing or illegible when filed}$

[0065] The overbrace puts these losses in relation to the conventional HBM while neglecting temperature dependencies of the on-state resistance. The switches are turned off every two periods T of the resonant current while the switches conduct for three quarters of a period. The lower losses P.sub.low.sup.FD-HBM of the switches that are only turned on for 25% of the time on the other hand can be calculated as

[00003]$\begin{array}{cc}{P}_{\mathrm{low}}^{\mathrm{FD}-\mathrm{HBM}}=\text{?}& \left(4\right)\end{array}$$\text{?}\text{indicates text missing or illegible when filed}$

[0066] Thus, the overall losses P.sub.inv of this modulation are the same compared to the conventional HBM. To evaluate the performance of this modulation, the HBM and the FM-HBM are evaluated for different loss distributions of the conduction losses and the switching losses. The reference is the conventional HBM.

[0067] FIG. 4 shows a loss analysis assuming a constant on-state resistance R.sub.ds,on. The losses of the switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 are depicted over the ratio of the switching losses p.sub.sw,compl.sup.HBM to the overall losses of the operation P.sub.inv. Obviously, the FM-HBM distributes the losses more equally among all switches S.sub.1, S.sub.2, S.sub.3 and S.sub.4 compared to the conventional HBM. While for P.sub.sw,compl.sup.HBM/P.sub.inv<⅙ dominant conduction losses in P.sub.compl.sup.HBM), the switch that is permanently turned on experiences the largest losses, for P.sub.sw,compl.sup.HBM/P.sub.inv>⅙ (dominant switching losses in P.sub.compl.sup.HBM), the switches that are switched face the largest losses.

[0068] The switch S.sub.1, S.sub.2, S.sub.3 or S.sub.4 being subject to the largest losses of the FM-HBM (P.sub.high.sup.FM-HBM) always shows lower or identical overall losses than the most-stressed switch S.sub.1, S.sub.2, S.sub.3 and S.sub.4 of the conventional HBM. Only for P.sub.sw,compl.sup.HBM/P.sub.inv=⅙, where the conduction and switching losses are equal in P.sub.compl.sup.HBM(P.sub.sw,compl.sup.HBM=P.sub.cond,compl.sup.HBM), the losses of the maximum stressed switches S.sub.1, S.sub.2, S.sub.3 or S.sub.4 in the HBM are the same as the maximum stressed switches S.sub.1, S.sub.2, S.sub.3 or S.sub.4 in the FM-HBM. The HBM and FM-HBM can for example be both employed on an LLC converter 100 utilizing silicon-carbide (SiC) MOSFETs of type SCT3120AW7 (Rohm, 650 V, 120 mΩ@25° C.) in the operating point V.sub.in=350 V, V.sub.out=11.5 and I.sub.out=125 A.

[0069] FIGS. 5A and 5B shows experimental measurement results of the HBM and FM-HBM. Depicted are the low-side gate voltages v.sub.GS,S3 and v.sub.GS,S4, the resonant current i.sub.res and the inverter voltage v.sub.AB. Both modulations result in about equal inverter voltages and resonant currents. The measured efficiencies are identical.

[0070] FIG. 6A shows an LLC half-bridge configuration with HBM and FIG. 6B shows a FM-HBM in a thermal analysis. FIG. 6A depicts the modulation of FIG. 2B, while FIG. 6B depicts the modulation of FIG. 3A. From top to bottom the switches (shown by the marker) are: S.sub.1, S.sub.2, S.sub.3 and S.sub.4. The FM-HBM reduces the maximum temperature by about 17 K. For the depicted operation, conduction losses are the dominant loss contributor. The results are at thermal steady state at the same operating conditions.

[0071] FIG. 7A to FIG. 7C show an on-the-fly morphing transition full-bridge to half-bridge mode and back. FIG. 7A shows a transition to the conventional half-bridge configuration and FIG. 7B a transition to the FM-HBM. For FIG. 7B, one compare value is shifted to achieve the morphing modulation. FIG. 7C includes the triangle modulation for determining the duty cycles.

[0072] To change the operating mode from full-bridge to half-bridge mode and back, an intermediate on-the-fly topology morphing method is used. For this transition, the lowside duty cycle of one half-bridge is slowly increased from 50% to 100% whereas the high-side switch of the respective half bridge 109-1 and 109-2 is switched complimentary. For a transition from half-bridge to full-bridge operation, this method is reversed.

[0073] While the duty cycle of all switches changes only by 25% when transitioning between the full-bridge modulation and the FM-HBM, the average resonant capacitor voltage is adjusted from v.sub.res=0 to v.sub.res=V.sub.inv/2 (or to v.sub.res=−V.sub.inv/2 if the other half-bridge configuration is employed). The sudden change from half-bridge configuration to full bridge configuration results in current or voltage breakdowns or overshoots. Considering the change of the switching frequency and resonant capacitor voltage a large overshoot/undershoot of the (photovoltaic) voltage and current is observed. A sudden transition from full-bridge to FM-HBM and back leads to an output voltage overshoot of 44% when transitioning from FM-HBM to full-bridge modulation and to an undershoot of 36% in the reverse direction. A feed-forward control adjusting the switching frequency is able to reduce this over-/undershoot.

[0074] However, since the switching frequency depends on the components of the resonant tank, component deviations affect the system behavior such that this type of control may be problematic if the exact component values are not matched. Furthermore, the switching frequency depends on the output load, which requires that the load must be known at the instance of the transition.

[0075] Employing the FM-HBM, it is equally possible to control the morphing from full-bridge to FM-HBM modulation and back (visualized for the Frequency Multiplier operation in FIG. 7B. The operation is hereby achieved also with an up-down counter. During the morphing operation mode one of the reference values is slowly adjusted to achieve the morphing modulation while the frequency is adjusted to achieve a steady output voltage/current. Since the modulation does not alter the shape of the inverter voltage, the traits of the morphing modulation remain the same: the magnetizing current still needs to be addressed in the design process and may still increase dynamically during the morphing transition. Furthermore, similar to the conventional morphing method, this modulation also results in a hard turn-on of the switches S.sub.1, S.sub.2, S.sub.3 or S.sub.4 close to the FMHBM.

[0076] FIG. 8A and FIG. 8B show experimental measurement results of the alternating asymmetrical (a2) morphing transition. FIG. 8A shows a transition from FD-HBM as an example of the FM-HBM to FBM, the intersections (A)-(C) show the intervals that are depicted in FIG. 9A to FIG. 9C. FIG. 8B shows a transition from FBM to FM-HBM. The alternating asymmetrical (a2) morphing as depicted in FIG. 7B in both directions can be implemented on an LLC converter 100 for the operating point V.sub.in=350 V, V.sub.out=10 V and I.sub.out=100 A. The transition results in an output voltage (V.sub.out) over/undershoot of about 6%.

[0077] FIG. 9A to FIG. 9C show experimental measurement results of the a2 morphing transition. Different time intervals of the transition process in FIG. 8A are depicted in FIG. 9A to FIG. 9C. The gate voltages v.sub.s3 and v.sub.s4, the inverter voltage v.sub.AB and the resonant voltage v.sub.res and current i.sub.res at different times in the FM-HBM to full-bridge transition of FIG. 8B are visualized.

[0078] FIG. 10 shows an improved morphing operation from full-bridge to FD-HBM. The inverter voltage can, however, be equally transferred to the conventional morphing process. In his morphing operation mode, the positive inverter voltage pulse length d.sub.pos*T on the first half-bridge 109-1 of the H-bridge 101 is determined on the basis of the negative inverter voltage pulse length d.sub.neg*T on the second half-bridge 109-2 of the H-bridge 101.

[0079] The positive inverter voltage pulse length d.sub.pos*T can be equal to the switching time d.sub.inv*T for switching a switch S.sub.1, S.sub.3 on the first half-bridge of the H-bridge, where d.sub.inv is the duty cycle and T the time length of the switching period. The negative inverter voltage pulse length d.sub.neg*T can be equal to the switching time d.sub.morph*T for switching a switch S.sub.2, S.sub.4 on the second half-bridge of the H-bridge, where d.sub.morph is the duty cycle and T the time length of the switching period.

[0080] This morphing operation mode differs from the above in that the constant duty cycle d.sub.inv is replaced by a dynamic duty cycle d.sub.inv the value of which is dependent on the value of the duty cycle d.sub.m. In this morphing operation mode, the magnetizing current can be effectively limited to prevent a saturation of the transformer. The relationship between the duty cycle d.sub.inv and duty cycle d.sub.m can be stored in a lookup table (LUT). The lookup table can be stored in a digital memory 105 of the control circuit 103. The lookup table can be derived using simulation results where a controller adjusts the pulse length to control the magnetizing current.

[0081] FIG. 11 shows a functional graph 107 representing the dependency of the duty cycle d.sub.inv on the second duty cycle d.sub.m. The functional graph 107 is stored as discrete values in the lookup table. The functional graph 107 is obtained for the resonant tank parameters L.sub.r=30 μH, L.sub.m=200 μH and C.sub.r=90 nF. The functional graph 107 is V-shaped and has a minimum between a value of 0.2 and 0.3 of the duty cycle d.sub.m and between a value of 0.2 and 0.3 of the first duty cycle d.sub.inv.

[0082] FIG. 12 shows a conventional morphing transition with a constant duty cycle. The on-the-fly topology transition is achieved with a constant positive voltage pulse length of t.sub.inv=0.5 T, where T is the pulse period. During the transition, the magnetizing current may increase dynamically such that the transformer core can be saturated. The magnetizing current increases dynamically by 70% compared to the steady-state value.

[0083] FIG. 13 shows the improved morphing transition with a dynamic duty cycle. The duty cycle d.sub.inv is dynamically adjusted according to the lookup table as depicted in FIG. 11. The magnetizing current does not increase dynamically and stays below the steady-state maximum. Thus, by adjusting the positive inverter voltage pulse, the dynamic increase of the magnetizing current can be limited to prevent transformer saturation.

[0084] FIG. 14 shows a block diagram of a converting method for the full bridge converter-circuit 100. The converting method comprises the step S101 of operating the H-bridge 101 in the first operation mode. In step S102 the H-bridge 101 is operated in a morphing operation mode in which the positive inverter voltage pulse length d.sub.pos*T on the first half-bridge 109-1 of the H-bridge 101 is determined on the basis of the negative inverter voltage pulse length d.sub.neg*T on the second half-bridge 109-2 of the H-20 bridge 101. In step S103 the H-bridge 101 is operated in a second operation mode.

[0085] The positive inverter voltage pulse length d.sub.pos*T and/or the negative inverter voltage pulse length d.sub.neg*T are adapted such that the duty cycle d.sub.pos equals to the duty cycle d.sub.inv and the duty cycle d.sub.neg equals to the duty cycle d.sub.m.

[0086] By employing the frequency-multiplier half-bridge modulation (FM-HBM), the junction temperature of the most-stressed inverter switch can be reduced significantly compared to the conventional HBM. The benefit depends on the system configuration and operating point. While for dominant conduction or switching losses, the benefit is remarkable, for nearly identical switching and conduction losses of the complementary switched MOSFETs, the junction temperature is similar. To maximally exploit the primary semiconductors in half-bridge configuration, the FM-HBM is the superior choice compared to the conventional HBM. Especially if conduction losses are the dominant loss contributor and the semiconductors are designed for full-bridge configuration, the FM-HBM offers an easy and effective loss reduction of the most stressed semiconductor.

[0087] An on-the-fly morphing transition between FM-HBM and full-bridge operation can be easily achieved using an up-down counter for the modulation. Experimental results prove that the undershoot/overshoot of the transition was limited to only 6%. Finally, EMI implications of the modulation must be considered as additional noise frequencies that result from the reduced switching frequency at odd integers of half the operating frequency. While this operation was able to reduce oscillations of the output significantly, it can result in partial saturation of the transformer core because the magnetizing current increases dynamically during the morphing process.

[0088] Through the frequency-multiplier modulation over at least two resonant periods, thermal imbalances are reduced or remain the same as with conventional half-bridge modulation, resulting in lower junction temperatures. If the operating mode, however, is changed suddenly from full-bridge operation to frequency multiplier modulation during the operation, the output voltage may experience significant oscillations that can destroy the converter. Through the presented easy-to-implement on-the-fly topology morphing, it is possible to morph during the operation from full-bridge modulation to frequency-multiplier half-bridge operation and reverse. During the morphing process, it is possible to slowly adjust the frequency to prevent significant output oscillations. By controlling the pulse lengths during the morphing mode operation, saturation of the transformer core is prevented and a constant output power during the morphing is maintained.

[0089] In summary, it is to be noted that the disclosure enables to provide a method as well as a circuit for limiting a dynamic increase of the magnetizing current to prevent a transformer saturation.