Low load flying capacitor balancing method and power converter

Abstract

A control method for controlling a multi-level flying capacitor power converter is disclosed, said power converter being configured to perform a predetermined control function, said function comprising preferably regulating power to a load and / or correcting a power factor, said power converter comprising a plurality of multi-level flying capacitor legs connected together, wherein each multi-level flying capacitor leg comprises at least one flying capacitor, a plurality of controllable switches and an inductor, the control method comprising continuously calculating current components to be injected for controlling the voltages of the flying capacitors over the whole range of leg currents, and controlling the switching of the plurality of controllable switches to obtain leg currents each comprising a respective first current component associated with the predetermined control function and a respective injected current component associated with the respective calculated current component, wherein the injected current components cancel each other out within the power converter.

Description

[0001] LOW LOAD FLYING CAPACITOR BALANCING METHOD AND POWER

[0002] CONVERTER

[0003] The present invention relates to a flying capacitor balancing method and the associated power converters, in particular a method and associated devices adapted for an efficient low load operation.

[0004] Multi-level Flying Capacitor Converters (FCCs) are advanced power electronic devices designed for applications requiring high voltage and high power. They utilize a topology based on multiple levels of voltage to improve efficiency and reduce switching ripple compared to traditional two- level converters.

[0005] These converters consist of several voltage levels formed using capacitors, which are charged and discharged in a controlled manner. They are primarily used in high-voltage and high-power applications, such as renewable energy systems (like solar inverters), electric drives, and grid integration. Their ability to handle high voltage and in a lesser extent current levels makes them suitable for medium to high voltage applications. As the demand for efficient power conversion continues to grow, these technologies play a vital role in the evolution of power electronics.

[0006] FCCs can be used in many applications, including DC-AC drives for electric motors, amplifiers for powering loads that require high precision voltage and / or current regulation (e.g. gradient coils of MRI scanners and linear motors of lithography machines), DC-DC interfaces for photo-voltaic installations, AC-DC interfaces for electrolyzers etc. They offer an attractive alternative to conventional two-level converter topologies, particularly in high-voltage applications, due to the higher number of voltage levels which enables the use of semiconductor switches with lower voltage rating which are generally cheaper and have relatively higher performance (i.e. they are more efficient) compared to switches with high voltage rating. Moreover, the higher number of voltage levels results in an increased effective switching frequency leading to smaller, cheaper (passive) output filters and increased control bandwidth of the output voltage and / or current. Compared to other multi-level converters, the FCC offers additional advantages, i.e., due to the employment of a single DC source no isolated voltage sources are required as is the case for the cascaded H-bridge converters and no high-power clamping diodes, which are oftentimes accompanied with anti-parallel synchronously operated semiconductor switches, like in the Neutral Point Clamped (NPC) topologies. Moreover, FCCs can operate in bidirectional DC-DC, AC-DC, and DC-AC mode and can be easily be paralleled to support high output current and output power levels which makes them perfectly suited for the use of new fast-switching Silicon Carbide (SiC) MOSFETs and Gallium Nitride (GaN) FETs which cannot be easily paralleled due to the need for ultra-precise gate-drive timing to avoid unequal current distribution and excessive switching losses. Lastly, interleaving of paralleled FCC stages enables an even higher effective switching frequency and a modular (scalable) design.

[0007] One of the key challenges in multi-level flying capacitor converters is balancing the voltages across the flying capacitors. This is crucial to ensure reliable operation and prevent capacitor failure. More in particular, balancing the Flying Capacitor (FC) voltages is required in practice to avoid large increases of the harmonics in the output voltage as well as over-voltages across the semiconductor switches. Since performance and reliability (related to over-voltages of the semiconductor switches) depend on the balancing of the individual FC voltages, FCCs are seldom employed in industry applications, despite the many advantages.

[0008] “Classical”, modulation-inherent natural / passive, FC balancing techniques which rely on the natural balancing of the FC voltages resulting from the applied PWM modulation of the semiconductor switches in normal operation of the FCC and the currents correspondingly flowing in the converter. However, as has been extensively described in literature, these ‘self-balancing’ methods cannot guarantee balancing of the FC voltages in many practical applications.

[0009] Balancing is especially challenging when powering loads that require arbitrary output voltage or current waveforms, particularly output current waveforms having large intervals with low current value that does not allow to perform active balancing of the FCs based on compensation of the Pulse- Width Modulated drive signals of the switches.

[0010] EP3836376B1A discloses a voltage balance control method for a flying-capacitor multilevel converter. If the amplitude of the resultant current of the inductor currents from a plurality of output inductors is lower than or equal to a threshold current value, the flowing direction of the inductor current of at least one flying-capacitor multilevel branch circuit is controlled to be changed. When the inductor current is low, the voltage of the flying capacitor is correspondingly controlled. In this way, the voltage balance of the flying capacitor of the flying-capacitor multilevel converter is achieved for low current.

[0011] Yet, such a prior art solution has drawbacks linked to the hysteretic nature of its control, creating stability and performance issues as well as being less flexible in handling varying load conditions, output requirements and fast transient responses. In addition, the above prior art solution only addresses the challenge of balancing flying capacitor voltages at low load in DC / DC mode. There is thus a need for a method and associated devices addressing these drawbacks and limitations. An object of the invention, next to other objects, is to provide a control method for a multi-level flying capacitor converter suitable for controlling the voltages of the flying capacitors in an improved manner over the whole range of leg ( / load) currents and for all modes of application (bidirectional DC-DC, AC-DC, and DC-AC mode).

[0012] This object, next to other objects, is met by a control method according to claim 1. Specifically, this is met by a method for controlling a power converter where the power converter is configured to perform a predetermined control function, The function comprises preferably regulating power / voltage / current to a load and / or correcting a power factor. The power converter comprises a plurality of multi-level flying capacitor legs connected together, wherein each multi-level flying capacitor leg comprises at least one flying capacitor, a plurality of controllable switches and an inductor. The control method comprises continuously calculating current components to be injected for controlling the voltages of the flying capacitors over said whole range of leg currents, and controlling the switching of the plurality of controllable switches to obtain leg currents each comprising a respective first current component associated with the predetermined control function and a respective injected current component associated with the respective calculated current component, wherein the injected current components cancel each other out within the power converter.

[0013] In this way, a continuous and hence stable control of the voltages of the flying capacitors is offered over the whole range of leg (load) currents. The injection of the calculated current components is to be understood as an additional step within the normal control steps of the predetermined control function. This additional step is introduced to extend the range of control of the voltages of the flying capacitors. At the same time, the injection of said current components does not affect the predetermined control function, since the injected current components cancel each other out within the power converter. In other words, the additional control of the flying capacitors voltages may be rendered transparent from the perspective of the predetermined control function but available over the whole range of leg (load) currents (low load, transients but also steady state) in a continuous manner, addressing thus the limitations of the prior art (hysteretic control). Continuous control by allowing smooth and gradual changes leads to better performance in terms of stability and responsiveness.

[0014] By “continuously” is meant here that the calculated current components are adjusted continuously over time, in particular at each clock cycle (rather than in discrete steps and only for certain periods of time). “Continuously” also means that the calculating step is not limited to certain load conditions, but will always / permanently be performed, even, for instance, in steady (permanent) state when the leg currents created by the predetermined control function are per se sufficient to control the FC voltages. The whole range of leg currents is meant for example to refer to the whole range of load currents when the power converter is powering a load.

[0015] The terms “over the whole range of leg currents” may be further understood as meaning “at any point of operation of the power converter”, whether in a transient / dynamic operation (including for instance start up, and / or shutdown, and / or load variations) or in a steady state operation.

[0016] By “current component” of a leg current is meant a distinct portion of a leg current that can be analyzed separately based on its characteristics. In other words, a leg current may be broken down at least into the first current component and the injected current component. By “first current components associated with the predetermined control function” is meant currents components that would be obtained by a control method for performing solely the predetermined control function. In that sense, the first current components may be regarded as current components obtained by prior art control methods. A control function may regulate over time one or more control variables of the power converter, like for instance output current, output voltage, output power, input power factor, etc. A typical control function may be abstracted as one or more control loops for regulating said one or more control variables, for instance a current regulation loop cascaded with a FC voltage regulation loop.

[0017] By “current components to be injected” is meant target current component values used as references in the control loop configured to obtain corresponding “actual” current component values (actual here in the sense that they may be sensed or estimated in the power converter). According to a preferred embodiment, continuously calculating the current components to be injected comprises calculating each respective current component to be injected from a respective sensed (or estimated) average leg current. In this way, the injected current components and the average leg currents related to the first current components may be interrelated. In particular, the current components to be injected may be derived from the average leg currents to allow a continuous change of the injected current components over the whole range of average leg currents which are related to the output load situation. This means that the injected components may compensate for the low values of average leg currents, which would not suffice on their own to control the flying capacitors voltage due to a low load situation.

[0018] According to a preferred embodiment, for a power converter with the plurality of multi-level flying capacitor legs connected together in parallel, wherein the injected current components circulate between the multi-level flying capacitor legs for balancing the voltages of the flying capacitors and cancel each other out among legs connected in parallel. By “balancing the voltages of the FCs” is meant that the voltages of the FCs of different legs evolve over time similarly towards a steady state identical value, typically half of the bus voltage in case of a 3 level FC topology.

[0019] According to a preferred embodiment, the injected currents are DC currents. In this way, for DC- DC applications, the control method may be kept quite simple. For instance, the DC currents may have a sign and amplitude such that their sum amounts to zero

[0020] According to a preferred embodiment, the injected currents are AC currents. DC currents have limited uses since a negative injected DC current may be cancelled out by a low leg current at low load. In contrast AC currents offer more options to (on average) over time have a current present to balance the flying capacitors. In particular, more variables of control (amplitude, phase, frequency) may be introduced when injecting AC currents. It is here noted that the use of AC currents need not be limited to continuous control and may be envisaged for prior art hysteretic control in a non- obvious and advantageous manner. To that extent, the present disclosure encompasses a control method for controlling a multi-level flying capacitor power converter, said power converter being configured to perform a predetermined control function, said function comprising preferably regulating power to a load and / or correcting a power factor, said power converter comprising a plurality of multi-level flying capacitor legs connected together, wherein each multi-level flying capacitor leg comprises at least one flying capacitor and a plurality of controllable switches, the control method comprising calculating AC current components to be injected for controlling the voltages of the flying capacitors over the whole range of leg currents, and controlling the switching of the plurality of controllable switches to obtain leg currents each comprising a respective first current component associated with the predetermined control function and a respective injected current component associated with the respective calculated current component, wherein the injected current components cancel each other out within the power converter.

[0021] According to a preferred embodiment, the injected currents are substantially phase-shifted to each other, such as to cancel each other out. In this way, the cancellation of injected components within the power converter may be easily obtained.

[0022] According to a preferred embodiment, the injected currents have an amplitude dimensioned to control the voltage of flying capacitors given the size of the flying-capacitors and the control range of the duty cycles of the controllable switches. A minimum amplitude may be required to be able to charge / discharge the flying capacitors. The injected currents may thus to be dimensioned accordingly to achieve the desired control effect. Although the capacitance of the capacitors and the control range of duty cycles may be preferred dimensioning parameters, a skilled person would understand that other considerations may be taken into account including a nominal load current, etc.

[0023] According to a preferred embodiment, the respective injected currents have a controllable amplitude that continuously changes in accordance with the respective first current components, preferably in a linear manner. In this way, gradual control may be achieved improving the stability of the control. This gradual control may be particularly superior when dealing with transient conditions compared to hysteretic control.

[0024] According to a preferred embodiment, the respective injected currents have a controllable phase and optionally a frequency to be set in accordance with the respective first current components, preferably having a phase set to avoid the injected currents of parallel connected legs being out of phase of a sum of the first current components of said parallel connected legs and optionally a frequency set to be a multiple of the frequency of the first current components. In this way, in case the leg currents may be AC currents; by controlling the phase and frequency, it may be avoided that injected current components are in opposition of phase with the first current components, improving thus the efficiency of the method. Alternatively, a combination of both DC and AC currents may be envisaged. It is further noted that the frequency need not to be fixed. A fixed frequency may be selected when the current components have themselves a fixed frequency, for instance when the converter is used as a Power Factor corrector on a mains with 50 / 60 Hz frequency.

[0025] According to a preferred embodiment, controlling the switching of the plurality of controllable switches comprises controlling duty ratios and / or phase shifts of the controllable switches. Although Pulse Width Modulation (PWM) and phase shift modulation are the most widely used control techniques, it is here noted that other control techniques (pulse frequency modulation, etc..) may be used in as far as the principle of a continuous injection of additional current components for the purpose of FC voltage balancing may be applied in combination with said other control techniques.

[0026] According to a preferred embodiment, controlling the switching of the plurality of controllable switches comprises: obtaining target duty ratios, obtaining sensed flying capacitor voltage values, calculating duty ratios of the controllable switches based on the obtained target duty ratios, the sensed flying-capacitor voltage values and a flying capacitor voltage balancing control function. In this way, a control loop for regulating the FC voltages may be achieved. In particular the FC balancing control loop may aim at regulating the FC voltages to a portion of the DC bus voltage, typically half of the DC bus voltage for 3-level FCs (connected to the DC side of the converter). It is hereby noted that estimation may be used instead of sensing in any embodiments of the present invention whenever suitable and / or more appropriate.

[0027] According to a preferred embodiment, obtaining target duty ratios comprises: obtaining target first current components to perform the predetermined power conversion control function, adding a respective obtained first current component to a respective calculated current component to be injected to calculate a respective target leg current, obtaining sensed leg current values, and calculating current errors between the respective calculated leg current and the respective sensed leg current value, calculating the target duty ratios based on the calculated current errors and the predetermined control function. In this way, current components to be injected may be introduced into the current regulation control loop. The control method according to an embodiment may differ in that sense from steps of a prior art control method using a typical current regulation loop (only) in that current components to be injected are calculated and added onto the reference current to ensure a continuous control of the FC voltage for any leg (load) current.

[0028] According to a preferred embodiment, calculating current components to be injected comprises: obtaining a Cutoff Peak Amplitude value, wherein when an average leg current is equal to said Cutoff Peak Amplitude value the target injected current of that leg is zero, and obtaining a Maximum Injection Peak Amplitude value, wherein when an average leg current is zero the target injected current of that leg has an amplitude equal to the Maximum Injection Amplitude value. In this way, the injected current components may be gradually controlled in amplitude to ensure smooth control without discontinuities nor discrete steps. The CutOffPeak Amplitude value may in that sense not create any hysteretic nor disruption of the control contrary to the threshold of the prior art.

[0029] According to a preferred embodiment, calculating current components to be injected comprises: obtaining sensed average leg currents, comparing said sensed average leg currents with the obtained Cutoff Amplitude value, and calculating an amplitude of a respective current component to be injected by applying to said comparison a continuous transformation, preferably a linear transformation, up to a saturation value set to be the Maximum Injection Amplitude value. In this way, a smooth variation of the current components may be calculated. The linear saturation of the difference between the sensed leg current and the CutOffAmplitude creates a gradual decrease of the current components to be injected inversely proportionally to the increase of the average leg current towards the CutOffAmplitude. Despite boundary conditions, the control may in this way remain continuous within the regulation loops (no change of control principle at boundaries). Although typically a linear transformation may be used, other continuous transformation options (exponential for instance) may be envisaged within the scope of the customary practice. In this way the amount of injection can be controlled to allow fading out the injection such as to avoid thresholds effects and stability issues.

[0030] According to a preferred embodiment, calculating over the whole range of leg currents current components to be injected further comprises:

[0031] - obtaining respective phases, and optionally a frequency, of the current components to be injected,

[0032] - generating the respective current components to be injected based on the respective obtained phases, and optionally frequency, and the respective calculated amplitudes. In this way, care may be taken that the injected current components combine well with the first current components. In particular phases and frequency may be adapted to avoid the injected current components to be out of phase of the first current components which would be detrimental to the purpose of controlling the FC voltages. For instance, the frequency of the injected current components may be set to differ from the one of the first current components.

[0033] According to a preferred embodiment, the method being for upon start charging the flying capacitors of all legs in a balanced manner using the injected circulating currents, and preferably for progressively increasing the leg currents up to nominal values. In this way, a smooth and fast charging and balancing of the FCs can be achieved.

[0034] According to a preferred embodiment, when the leg currents are higher than a predetermined low load current threshold, the calculated second currents to be injected are zero and the first current components are such as to control the voltages of the flying capacitors based on regulating a phase shift in the duty ratios of the switches of an FC leg. It is noted that when the leg currents are lower than the predetermined low load current threshold, the calculated second currents to be injected are used for the regulation of the voltages of the flying capacitors based on regulating a phase shift in the duty ratios of the switches of an FC leg. In other words, the regulation of the charging of the FCs is achieved using the same balancing control scheme (namely using phase shifting of the duty ratios) continuously both at low load and normal load.

[0035] According to another aspect, a multi-level flying capacitor power converter configured to perform a predetermined control function is provided. The function comprises preferably regulating power to a load and / or correcting a power factor, said power converter comprising:

[0036] - a plurality of multi-level flying capacitor legs connected together, wherein each multi-level flying capacitor leg comprises at least one flying capacitor, a plurality of controllable switches and an inductor, and

[0037] - a controller configured to continuously calculate (over the whole range of leg currents) current components to be injected for controlling the voltages of the flying capacitors over the whole range of leg currents, the controller being further configured to control the switching of the controllable switches to obtain leg currents each comprising a respective first current component associated with the predetermined control function, and a respective injected current component associated with the respective calculated current component, wherein the injected current components (circulate between the multi-level flying capacitor legs and) cancel each other out within the power converter.

[0038] According to a preferred embodiment, the controller is further configured to continuously calculate each respective second current component to be injected from a respective sensed average leg current.

[0039] According to a preferred embodiment, the plurality of multi-level flying capacitor legs are further connected in parallel, and the controller is configured such that the injected current components circulate between the multi-level flying capacitor legs for balancing the voltages of the flying capacitors and cancel each other out among legs connected in parallel.

[0040] According to a preferred embodiment, the controller is configured such that the injected current components are DC currents.

[0041] According to a preferred embodiment, the controller is configured such that the injected current components are AC currents.

[0042] According to a preferred embodiment, the controller is configured such that the injected current components are substantially phase-shifted to each other such as to cancel each other out. According to a preferred embodiment, the controller is configured such that the injected current components have an amplitude dimensioned to control the voltage of flying capacitors given the size of the flying-capacitors and the control range of the duty cycles of the controllable switches. According to a preferred embodiment, respective injected currents have a controllable amplitude that gradually changes in accordance with their respective first current components, preferably in a linear manner.

[0043] According to a preferred embodiment, the respective injected currents have a controllable phase and optionally a frequency to be set in accordance with their respective first current components, preferably having a phase set to avoid the injected currents of parallel connected legs being out of phase of a sum of the first current components of said parallel connected legs and optionally a frequency set to be a multiple of the frequency of the first current components.

[0044] According to a preferred embodiment, the converter is an AC-DC converter, preferably a multiphase rectifier for power rectification and / or power factor correction. Preferably, the multi-phase rectifier may have for each phase several branches in parallel wherein the injected current components may circulate between parallel branches of the same phase. Yet, the principle of the present invention may be used between phases of a multiphase rectifier (not necessarily having several branches in parallel per phase) such that the injected current components would then circulate between branches of different phases. Reactive current from a mains may then be used to balance the FC voltages.

[0045] According to a preferred embodiment, the converter is a DC-DC converter for regulating power to a load, the converter being preferably a buck converter. Preferably, the DC-DC converter may comprise a plurality of legs connected in parallel. However in some applications, the DC-DC converter may comprise several branches connected to the same (output) common connection node, each branch being however connected to a different input source, such that the injected currents may circulate between legs from their common connection node.

[0046] According to a preferred embodiment, the converter is a DC- AC converter for regulating power to a load, preferably a multi-phase inverter, typically a three-phase inverter. Preferably, the multiphase inverter may have several branches in parallel for each phase wherein the injected current components may circulate between parallel branches of the same phase. Yet, the principle of the present invention may be used between phases of a multiphase inverter (not necessarily having several branches in parallel per phase) such that the injected current components would then circulate between branches of different phases.

[0047] According to a preferred embodiment, the multi-level flying-capacitor legs are 3 -level flying capacitor legs, comprising four switches and one flying capacitor, said four switches being connected in series with each other across the input while the flying capacitor is connected in parallel to the middle two switches, a middle point between the middle two switches forming an output point. Alternatively, higher levels FC arrangements known in the art may be used without limitation. A skilled person would be able to adapt accordingly the control method.

[0048] According to a preferred embodiment, the power converter is for powering a pulsed load, preferably an electrolyser.

[0049] According to another aspect of the invention, a computer readable medium is provided which is configured to store instructions which, when executed by a controller of a power converter according to any of the above power converter claims, cause said power converter to carry out the steps of any of the above method claims.

[0050] This and other aspects of the present invention will now be described in more detail, with reference to the appended drawings showing currently preferred embodiments of the invention, wherein:

[0051] Figure 1 illustrates the working principle of a 3 level FCC according to the prior art.

[0052] Figure 2 illustrates the principle of balancing FC voltages according to a phase shifting of the respective duty cycles of Figure 1.

[0053] Figure 3 illustrates a cascaded AC-DC DC-DC arrangement comprising two FCCs according to embodiments of the present invention.

[0054] Figures 4a and 4b illustrate schematic waveforms of the FC voltages and leg currents of a DC stage of Figure 3 having three parallel branches in a scenario from no load to load.

[0055] Figure 5 illustrates a schematic bloc diagram of an injection current generator for calculating the current components to be injected according to an embodiment of the invention.

[0056] Figures 6a and 6b illustrate waveforms of a DC stage according to an embodiment of Figure 3 for a CutoffAmplitude and a Maxlnj Amplitude of both 5A. Figure 6a illustrates the total current versus the injected current components of each branch while Figure 6b illustrates the leg currents of each branch versus the average leg current.

[0057] Figures 7a and 7b illustrate, similar to Figures 6a and 6b, waveforms of a DC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 4A and a Maxlnj Amplitude of 6A. Figures 8a and 8b illustrate, similar to Figures 6a and 6b, waveforms of a DC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 6A and a Maxlnj Amplitude of 4A. Figures 9a and 9b illustrate waveforms of an AC stage of Figure 3 having three parallel branches in a scenario from no load to load for a CutoffAmplitude and a Maxlnj Amplitude of both 5 A. Figure 9a illustrates the total current versus the injected current components of each branch while Figure 9b illustrates the leg currents of each branch versus the average leg current.

[0058] Figures 10a and 10b illustrate, similar to Figures 9a and 9b, waveforms of an AC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 4A and a MaxInjAmplitude of 6A.

[0059] Figures 1 la and 1 lb illustrate, similar to Figures 9a and 9b, waveforms of an AC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 6A and a MaxInjAmplitude of 4A.

[0060] Figure 12 illustrates a schematic block diagram of a flying capacitor balancing controller for an FCC according to an embodiment.

[0061] Figure 13 illustrates a schematic block diagram of a complete control structure for an FCC according to an embodiment.

[0062] Figure 14 illustrates the steps of a method according to an embodiment of the presentation invention.

[0063] Figure 15 illustrates sub-steps of the first step of Figure 14 related to the calculation of current components to be injected.

[0064] Figure 16 illustrates sub-steps of the second step of the Figure 14 related to the voltage balancing of the FC voltages.

[0065] Figure 17 illustrates sub-steps of the first step of Figure 16 related to the current regulation.

[0066] Figure 1 shows the working principle of a three-level FCC. A three-level FCC typically comprises four controllable switches S1-S4 and one flying capacitor CFC- The controllable four switches S1-S4 are connected in series with each other across an input voltage VIN while the flying capacitor CFC is connected in parallel to the middle two switches S2 and S3, a middle point between the middle two switches S2 and S3 forms then an output point seeing Vsw- An output inductor L is connected in between the middle point and an output capacitor Cour for powering a load. The load current is the current L flowing through the output inductor L, while the load sees the voltage VOUT across the output capacitor Cour- The switches S1 / S4 form a first complementary pair controlled by a duty cycle du and the switches S2 / S3 form a second complementary pair controlled by a duty cycle di?- The carriers of S14 and S23 (not shown) are 180 degrees phase shifted. Figure 1 also shows for duty cycles lower than 0,5 the conduction of the switches during each switch state and the resulting switch node voltage Vsw- The flying capacitor CFC is charged or discharged when either SI and S3 are conducting or S2 and S4 are conducting. When the duty cycles du and di ? are equal and VFC=VIN / 2, the FCC is balanced. When the flying capacitor CFC is equally charged and discharged, then on average the voltage VFC stays constant. For duty cycles higher than 0,5, an extra state where SI and S2 are both conducting is present.

[0067] Figure 2 shows that, by changing the duty cycles du and d23, the time of charging vs. discharging can be changed, which is typically used to actively charge or discharge the flying capacitor. It is noted that the dl / dt of the inductor current will also change when VFC is not equal to VIN / 2. This prior art “classical” balancing control method is typically used in permanent regime and when the load current is significant enough. However without load current or at very low load current, this classical balancing method for charging or discharging the capacitor CFC is not possible.

[0068] Figure 3 illustrates a cascaded AC-DC DC-DC arrangement 1000 comprising two FCCs according to embodiments of the present invention forming respectively an AC stage 100 and a DC stage 200. The AC stage 100 may be a Flying Capacitor (FC) rectifier 100 for rectification and / or power factor correction, connected between an input three-phase network X (with input currents io,a, io,b, io.c and input phase voltages vo, a, Vo, b, Vo, c ) and a DC link formed across series connected capacitors CDC,Pand Coc,n (M referenced as middle point). The rectifier 100 may comprise, for each input phase a, b and c, q FC branches (or legs) 10ai- 10aq, respectively 10bi - 10bq, respectively 10ci- 10cq, in parallel with each other, on both (ac and de) sides (common ac terminals mi, resp. m2, resp. m3; common DC terminals hi, resp. I12, resp. I13; h, resp. I2, resp. I3) , where q is an integer bigger than 1. The DC stage 200 may be a Flying Capacitor (FC) DC-DC converter for regulating power to a load (including current regulation), connected in cascade with the FC rectifier 100 between the DC link and the load. The DC-DC converter 200 may comprise r FC branches (or legs) 101- 10rconnected in parallel with each other, both on the input (common terminals h and 1) and the output (common terminal m), r being an integer bigger than 1. Each branch / leg 10 (whether for the rectifier 100 or the DC-DC converter 200, hence generalised in notation as a FC leg 10) may have a three-level configuration as previously discussed with respect to Figure 1 with four controllable switches, a (flying) capacitor and an inductor. A common output capacitor Cout may be connected on the de output in parallel with the load. An output voltage vomay be present across the output capacitor Cout- An output current iomay flow to the load.

[0069] However note the number of levels, controllable switches and flying capacitors is not to be limited to the simple practical example of three-level and a skilled person would without inventive step extrapolate the teachings of the present disclosure to any number of level adapted to circumstances. Although a cascaded AC-DC DC-DC arrangement is shown, the insights behind the present application do not require cascaded stages and may be applied to any power converter of any nature whether DC-DC, AC-DC or DC-AC, whether unidirectional or bidirectional, as long as a plurality of multi-level FC legs are connected together, each multi-level flying capacitor leg comprising at least one flying capacitor and a plurality of controllable switches.

[0070] A power converter is meant typically to perform a predetermined power conversion control function. The power conversion control function may be for example correcting a power factor and / or regulating a current / voltage to a load. Other control functions come within the scope of the customary practice of the art. The power conversion control function may further include the classical voltage balancing method. To perform said power conversion control function, a converter typically comprises a controller generating control signals for the controllable switches according to a control method of the art.

[0071] Figures 4a and 4b illustrate schematic waveforms of the FC voltages and leg currents of a DC stage of Figure 3 having three parallel branches when controlled according to the invention in a scenario starting from no output load to a nominal output load. The waveforms of Figures 4a and 4b illustrate the control method behind the invention. According to an embodiment, the control method comprises controlling of the plurality of controllable switches such as to obtain leg currents each comprising a respective first current component associated with a predetermined control function, and a respective injected current component for controlling the voltages of the flying capacitors over the whole range of leg currents. This means that in addition to the first current components typically obtainable by a control method of the art, according to the control method of the invention, other current components are injected for the specific purpose of controlling the voltages of the flying capacitors over the whole range of leg currents, and in particular under low load. The additional injected current components allow charging and discharging the flying capacitors even in the absence of a first current component or even if said first current component is too low to effectively control the voltage of the flying capacitors. To do so, without affecting the control function of the art, the injected current components circulate between the multi-level flying capacitor legs and cancel each other out within the power converter. Such a controlling of the switching of the controllable switches comprises continuously calculating (over the whole range of leg currents) current components to be injected. However, note that although always calculated, the calculated current components may be zero if the first current component is sufficiently high to achieve voltage control using the classical balancing method.

[0072] Figure 4b will now be described in more detail in view of said explained control method. Figure 4b shows the waveforms of the three leg currents of converter 200, when controlled to start from no load to nominal load. The converter 200 is typically meant for regulating the load current or voltage. The operation may be subdivided into three phases A, B, C. In the first injection (start-up) phase A, the injection of interleaved current components is visible as the three leg interleaved currents show each an interleaved AC component in addition to a common DC steadily increasing first current component. The AC component may correspond there to the injected current component for controlling the FC voltages over the whole range, while the DC component may be associated with a (first) current component for regulating the load. In this first phase A, the first current component is too low for the “classical” balancing method to work efficiently, hence the need for the injected (also referred to as second) current components. This first phase A is followed by a dynamic “normal” (start-up) phase B in which the leg currents are superposed and follow an increasing trendline. It is here noted that although the example here shows startup operation with an increasing output current, the method is not to be understood as limited to start-up operation and may be envisaged for any arbitrary dynamic operation with any arbitrary waveform. In the second phase B, the first current component is high enough for the “classical” balancing method to function while ramping up the load current. This is the reason why an AC component is no longer visible, which illustrates that the control method calculates zero second current components during that phase. Finally when the nominal load current is obtained, the “classical” balancing method and current regulation are performed while zero second current components are calculated. In this steady state phase C, the leg currents correspond thus entirely with the first current components for performing the “classical” balancing method and the current regulation. Although for clarity Figure 4 has been described as showing three phases A, B and C, namely an injection phase A, a normal dynamic (start-up) phase B and a steady state (also called permanent regime) phase C, note that the control method of this embodiment is the same over the whole operation (whether during A, B or C) and that the intrinsic control method does not use different control steps during said phases. Even in steady state, current components to be injected are calculated and then injected in the control. Simply their amplitude may be zero rendering them invisible. In that sense, the injected current components are continuously calculated and used by the control method. This continuous calculation renders said control method smooth and stable as illustrated by the waveforms Figure 4. There are no non-linear, nor sudden changes in the waveforms shown created by the control method itself ( i.e., that would not originate from a sudden load current change) .

[0073] Figure 5 illustrates a schematic bloc diagram of an injection current generator for calculating the current components to be injected according to an embodiment of the invention. As illustrated in Figure 5, the injection current generator 20 generates the current components to be injected based on a sensed (or estimated) average leg currents. As previously noted, estimation may be an obvious alternative to sensing depending on circumstances and may be envisaged in the present invention for all variables, As indicated by the number three present above each signal line, the present block diagram applies for each phase respectively. By sensed average leg current is meant the mean value of a sensed leg current. A mean operator 21 is first applied to said sensed average leg current, before dividing said result by a variable named CutOffAmplitude (peak) at a divider 22. Then a subtractor 24 subtracts the output of divider 22 from a value of 1 obtained by block 23. The mean(norm) operator 21 providing an absolute value, the divider 22, the subtractor 24 are such when an average leg current is equal to said CutOff Amplitude value the target injected current of that leg will be zero. In other words the variable CutOffAmplitude determines the amplitude of the average leg current at which the current components will become null. Although the injection current generator 20 for average leg current values higher than CutOffAmplitude will generate a zero output, this element 20 continues operating in a continuous manner over the whole range of leg currents and thus differs from an hysteretic element. The element 20 may thus follow in an efficient and stable manner any load variations over time. The injection current generator 20 further comprises a saturation element 25 receiving the output of subtractor 23. The saturation element 25 saturates the output of the substractor between 0 and 1. The output minimum value of the saturation element is set to 0 while the maximum value of the saturation element 25 is set to 1. A gain 26 receives the output of the saturation element 25. The gain 26 has a value named Maxlnj Amplitude. The variable Maxlnj Amplitude determines the maximum amplitude of an inject current component at any time. The combination of elements 25 and 26 are such that when an average leg current is zero the target injected current of that leg has an amplitude equal to the Maximum Injection Amplitude value. The output of element 26 defines the amplitude of the current component to be injected, also called Injection Amplitude. A phase block 27 receives said Injection Amplitude and a variable Angle prior to generating dq current components. The variable Angle in block 27 amounts to the relative phase of the current components to be injected with regard to the clock or synchronization signal. The variable Angle may typically be different for each leg current phase and in particular may be set interleaved with each other. A dq-abc trandforming block 28 receives said dq current components and a Phase. The variable Phase , which may also be noted as (w*t), in this block 28 changes with time and is proportional to the frequency of the current components to be injected. Block 28 then outputs InjectionCurrentSetPoints, which amount to current components having the phases, frequency and amplitudes as set within the injection current generator 20. The characteristics of the current components may be optimized to a use case scenario by setting optimized values for the variables CutOffAmplitude, MaxInjAmplitude, Angle and Phase. The effects of at least some of these variables on the output response of the system will be shown in Figures 6a- 1 lb.

[0074] Figures 6a and 6b illustrate waveforms of a DC stage according to an embodiment of Figure 3 for a CutOffAmplitude and a MaxInjAmplitude of both 5A. Figure 6a illustrates the total current I-rotai versus the injected current components linji , Iinj2, Iinj3 of each branch. The DC stage comprises three legs in parallel. By total current is thus meant the sum of the three leg currents, lTotai= Iregi+ Ireg2+ lLeg3- As can be seen, Dotal is a DC current linearly varying from zero to its nominal value (outside of the window of Figure 6a) during dynamic (startup) operation. The injected current components linji , Iinj2, Lnj3, on the other hand are AC currents with the same frequency but the currents are interleaved with each other such that any point in time Iinji+ Iinj2+Iinj3 =0. The amplitude of the injected currents is maximum at start and zero at the end of period A of injection (start up). Figure 6b illustrates more in particular the leg currents Legi , Ireg2, lLeg3, of each branch versus the average leg current iLegaverage- Figure 6b is a zoom of Figure 4b showing in an enlarged manner the behavior during the injection (start-up) phase of Figure 4b. By average leg current is meant the average value of the three leg currents, lLegaverage= lTotai / 3=(lLegi+ lLeg2+ lLeg3) / 3. As can be seen the leg currents Legi , lLeg2, lLeg3, of each branch during period A, each comprise a (DC) current component equal to ILegaverage, and a second (AC) current component amounting to the injected current components linji , Iinj2, Iinj3 of Figure 6a. The maximum amplitude of the leg currents is set at 5 A, by setting the variable Maxlnj Amplitude at 5A. The AC current component becomes zero when the average leg current reaches 5A by setting the variable CutOffAmplitude at 5A as well. Figures 7a and 7b illustrate, similar to Figures 6a and 6b, waveforms of a DC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 4A and a Maxlnj Amplitude of 6A. Compared to Figures 6a and 6b, the period A is shorter in figures 7a and 7b due to the relatively lower value of CutoffAmplitude.

[0075] Figures 8a and 8b illustrate, similar to Figures 6a and 6b, waveforms of a DC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 6A and a Maxlnj Amplitude of 4A. Compared to Figures 6a and 6b, the period A is longer in figures 8a and 8b due to the relatively higher value of CutoffAmplitude.

[0076] The duration of period A may thus be set by the CutoffAmplitude. At the same time, the value of Maxlnj Amplitude may be set relative to the one of CutoffAmplitude to ensure enough injected current to achieve the balancing of the FC voltages in the set period A. A skilled person would obtain suitable values for CutoffAmplitude and / or Maxlnj Amplitude based on circumstances of use. Figures 6a- 8b show multiple ways to implement a continuous injection of currents, with different DC average currents (over time) to control the flying capacitor voltages.

[0077] Figures 9a and 9b illustrate waveforms of one phase of an AC stage of Figure 3 having three parallel branches for each input AC phase in a scenario from no load to load for a CutoffAmplitude and a Maxlnj Amplitude of both 5A. Figure 9a illustrates the total current I-rotai associated with one AC input phase (a, b or c of Figure 3) versus the injected current components linji , Iinj2, Iinj3 of each branch of this parallel phase arrangement. The AC stage comprises three legs in parallel for each input AC phase (a, b or c of Figure 3). By total current is thus meant, for each of said input AC phase, the sum of the three leg currents of that input AC phase, lTotai= lLegi+ lLeg2+ lLeg3- In that sense, the notation Dotal in Figures 9a or 9b corresponds to any one of the notations io,a, io, b or io,cof Figure 3. As can be seen, I-rotai is an AC current with a first frequency and an amplitude linearly varying from zero to its nominal value (outside of the window of Figure 9a) during startup. The injected current components linji , Iinj2, Iinj3, are AC currents with a second frequency, typically the same as the first one, where the current are interleaved with each other such that any point in time Iinji+ Iinj2+Iinj3 =0. The amplitude of the injected currents may be maximum at start and zero at the end of period A of injection start up.

[0078] Figure 9b illustrates more in particular the leg currents Legi , Ireg2, lLeg3, of each branch versus the average leg current iLegaverage- By average leg current is meant, for each input AC phase, the average value of the three leg currents of that input AC phase, lLegaverage= lTotai / 3=(lLegi+ lLeg2+ lLeg3) / 3. The leg currents Legi , lLeg2, lLeg3, of each branch during period A, each comprise a (AC) current component equal to ILegaverage, and a second (AC) current component amounting to the injected current components linji , Iinj2, Iinj3 of Figure 9a. The maximum amplitude of the leg currents is set at 5 A, by setting the variable Maxlnj Amplitude at 5A. The AC current component becomes zero when the average leg current reaches 5A by setting the variable CutOffAmplitude at 5A as well. Figures 10a and 10b illustrate, similar to Figures 9a and 9b, waveforms of an AC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 4A and a MaxInjAmplitude of 6A. Compared to Figures 9a and 9b, the period A is shorter in figures 10a and 10b due to the relatively lower value of CutoffAmplitude.

[0079] Figures 1 la and 1 lb illustrate, similar to Figures 9a and 9b, waveforms of an AC stage according to yet another embodiment of Figure 3 for a CutoffAmplitude of 6A and a MaxInjAmplitude of 4A. Compared to Figures 9a and 9b, the period A is longer in figures I la and 1 lb due to the relatively higher value of CutoffAmplitude.

[0080] Whether for a DC-DC converter as in Figures 6a-8b or for an AC-DC converter as in Figures 9a- 1 lb, or for a DC- AC converter (not illustrated), the principle of operation of injecting current components that cancel each other out but enable to balance the voltages of the FC even at low load or no load applies thus in a similar manner differing only in the first component associated with the primary control function being specific to primary control function. It is further noted that although said injected current components have been illustrated as AC components, De components may also be used. DC components may be designed as having amplitudes and sign that cancel each other out. For instance, one or more current components in one or more branches may be flowing in an opposite direction to the rest of the current components, such that the sum of the amplitudes of said current components is zero.

[0081] Figure 12 illustrates a schematic block diagram of a prior art flying capacitor balancing controller 40 which may be used in an FCC according to an embodiment. The FC balancing controller 40 may form an additional control loop on top of the control loop of the predetermined control function (for regulation the power to a load and / or controlling a power factor for instance). The FC balancing controller 40 may enable a “classical” balancing control for each flying capacitor. The FC balancing controller 40 may receive respectively for each branch, as inputs a target duty ratio Duty_In of that branch, a sensed DC bus voltage V_Bus, a sensed FC voltage V_FC of the FC of that branch and a sensed leg / branch current I_L (in other words Legi, or Legz or Legs). Based on these inputs, element 40 may generate the duty ratios for the pair of switches S1 / S4 and the pair of switches S2 / S3, Duty_Out_sl4 and Duty_Out_s23. The target duty ratio Duty_In is obtained from a leg current controller 30 which will be further described in Figure 13 and generates a reference duty ratio for obtaining a reference leg current calculated to perform the predetermined control function and incorporating the injected current components output by the Injection Current generator 20.

[0082] In the FC balancing controller 40, the sensed DC bus voltage V_Bus is input to a gain 41 with a fixed Yi value to half said sensed DC bus voltage V_Bus. Note that the ratio Yi comes from the three-level topology and that a skilled person would extrapolate other values and corresponding calculations for other FC topologies. This Figure is in that sense only meant as an exafor illustrative purposes. The halfed V_Bus is then fed to a comparator 42 forming an error between the halfed V_Bus and the sensed FC voltage V_FC. The obtained error is then input into a controller 43 for regulating said error to reduce said error. The control signal output by controller 43 is then multiplied by multipler 45 with the sign of the leg current I_L obtained by feeding the current I_L to a sign operator block 44. The signal output by multipler 45 is then input to a limiter 46 to obtain a signal Delta representing a duty ratio variation for balancing the FC voltage. The sensed FC voltage V_FC is further divided by the sensed DC bus voltage V_Bus at divider 47. The output of divider 47 is then multiplied with the signal Delta at multiplier 48. The output of multiplier 48 is then added to the target duty ratio Duty_In at adder 49 to obtain the duty ratio for the pair of switches S1 / S4 Duty_Out_sl4. To obtain the duty ratio for the other pair of switches S2 / S3, the Duty_Out_sl4 is compared to the signal Delta by subtracting Delta from Duty_Out_sl4 at comparator / subtractor 51. The comparator 51 outputs then the duty ratio for the pair of switches S2 / S3, Duty_Out_s23.

[0083] Figure 13 illustrates a schematic block diagram of a complete control structure 50 for an FCC according to an embodiment where the predetermined function is current regulation. The block diagram of Figure 13 shows the interconnections between the Injection Current Generator 20 of Figure 5 (labelled in Figure 13 Reactive PowerGen), the FC balancing controller 40 of Figure 12 (labelled in Figure 13 FCCtrl) and a current loop comprising a current controller 30 for current regulation. The complete control structure 50 may receive, for each phase respectively, a target( / reference) total current I_TOT_REF. I_TOT_REF may represent a reference total current for a current regulation loop. A closed feedback current regulation loop may be present indeed to perform the function of regulating the current to the output load. The feedback current regulation loop may comprise a comparator 54 generating an error between a target leg current and a sensed leg current I_IL. The error is input to a current controller 30 to generate the target duty ratio Duty_In. The target leg current received at one input of the comparator 54 is obtained by adding the injected current component InjectionCurrentSetPoint generated by the Injection current generator 20 to one third of the reference total current I_TOT_REF obtained by applying gain 55 with 1 / 3 value. Note that the ratio 1 / 3 derives from the presence of three legs being used in parallel in the chosen illustration but is not limitative. A skilled person would adapt the ratio to the chosen arrangement and associated number of connected legs.

[0084] Figure 14 illustrates the steps of a method according to an embodiment of the presentation invention for one phase. As illustrated in Figure 14, for any given phase, the method comprises a first S10 of continuously calculating, over the whole range of leg currents, current components to be injected for controlling the voltages of the flying capacitors over said whole range of leg currents, followed by a steps S20 of controlling the switching of the plurality of controllable switches to obtain leg currents each comprising a respective first current component associated with the predetermined control function and a respective injected current component associated with the respective calculated current component, wherein the injected current components circulate between the multi-level flying capacitor legs and cancel each other out within the power converter.

[0085] Figure 15 illustrates sub-steps of the first step S10 of Figure 14 related to the calculation of current components to be injected. The first step S10 may be subdivided into a succession of steps SI 1- S 15 to be executed for any given phase. The first step Si l may comprise obtaining a Cutoff Peak Amplitude value, a Maximum Injection Peak Amplitude value, a Phase and optionally a frequency (when AC currents are injected). This step corresponds to receiving the above-mentioned parameters from either a memory or a user interface. These parameters may then be used as illustrated in Figure 5 by the block diagram of the Injection Current Generator 20 and communicated to blocks 22, 26, 27 and 28. A following step S12 may comprise obtaining a sensed average leg current. This step may comprise sensing an average leg current from a dedicated current sensor or alternatively estimating an average leg current from an estimator. As illustrated in Figure 5, the sensed average leg current is indeed an input of the Injection Current Generator 20. A following step S13 may comprise comparing said sensed average leg current with the obtained Cutoff Amplitude value. This step corresponds to the operations of mean operator 21, divider 22, gain 23 and comparator 24 in Figure 5. A following step S14 may comprise calculating an amplitude of a current component to be injected by applying to said comparison a continuous transformation, preferably a linear transformation, up to a saturation value set to be the Maximum Injection Amplitude value. This step corresponds to the operation of saturation element 25 and gain 26. A final step S15 may comprise generating the current component to be injected based on the obtained phase, the calculated amplitude, and optionally the obtained frequency. This step corresponds to the operations of elements 27 and 28 of Figure 5 in which the amplitude, the phase and angle are used to generate the Injection CurrentSetPoint output (of Figures 5 and 13), also referred to as the current component to be injected.

[0086] Figure 16 illustrates sub-steps of the second step of the Figure 14 related to the voltage balancing of the FC voltages. The second step S20 of Figure 14 may be subdivided into a succession of steps S40-S42 to be executed for any given phase. The first step S41 may then comprise obtaining a target duty ratio. By target duty ratio is meant the variable Duty_In of Figure 13. This step S41 may correspond to receiving the output of controller 30 for current regulation. In other the target duty ratio may be output by the current loop regulation. A following step S41 may comprise obtaining a sensed flying capacitor voltage value. This steps S41 may comprise sensing a FC voltage value V_FC using a voltage sensor or estimating it using an estimator. A flowing step S42 may comprise calculating duty ratios Duty_Out_sl4 and Duty_Out_s23 based on the obtained target duty ratio Duty_In, the sensed FC voltage value V_FC and a FC voltage balancing control function. The FC voltage balancing control function corresponds to transfer function illustrated in Figure 12

[0087] Figure 17 illustrates sub-steps of the first step of Figure 16 related to the current regulation. The first step of Figure 16 about obtaining a target duty ratio Duty_In may be further subdivided into a plurality steps S30-S34. The first steps S30 may comprise obtaining a target first current component to perform the predetermined power conversion control function. The target first current component of step S30 may also be referred to as the target / (reference) average leg current and corresponds to I_TOT_REF / 3 of Figure 13. This variable may be derived from I_TOT_REF (via gain) which may be received from a memory or may be output from a further not illustrated control loop related to the predetermined control function. For instance, in case of power factor correction, a further power factor correction control loop may output said I_TOT_REF. A following step S31 may comprise adding the previously obtained first current component I_TOT_REF (multiplied by a gain) to the calculated current component to be injected to calculate a target leg current. This step corresponds to the operation of comparator 53 in Figure 13. A following step S32 may comprise obtaining a sensed leg current value, I_L in Figure 13. This step may comprise sensing said variable or estimating it. A following step S33 may comprise calculating a current error I_ERROR between the calculated target leg current and the sensed leg current value I_L. This step may correspond to the operation of comparator 54 of Figure 13. A following step S34 may comprise calculating a target duty ratio Duty_In based on the calculated current error I_ERROR and the predetermined control function being for instance a current regulation function, comprising for example any kind of well known, P / PI or PID function for closed loop regulation. Whilst the principles of the invention have been set out above in connection with specific embodiments, it is understood that this description is merely made by way of example and not as a limitation of the scope of protection which is determined by the appended claims.

Claims

22CLAIMS1. A control method for controlling a multi-level flying capacitor power converter, said power converter being configured to perform a predetermined control function, said function comprising preferably regulating power to a load and / or correcting a power factor, said power converter comprising a plurality of multi-level flying capacitor legs connected together, wherein each multi-level flying capacitor leg comprises at least one flying capacitor, a plurality of controllable switches and an inductor, the control method comprising: continuously calculating current components to be injected for controlling the voltages of the flying capacitors over the whole range of leg currents, and controlling the switching of the plurality of controllable switches to obtain leg currents each comprising a respective first current component associated with the predetermined control function and a respective injected current component associated with the respective calculated current component, wherein the injected current components cancel each other out within the power converter.

2. The control method of claim 1 , wherein continuously calculating the current components to be injected comprises calculating each respective second current component to be injected from a respective sensed average leg current.

3. The control method according to claim 1 or 2, for a power converter with the plurality of multi-level flying capacitor legs connected together in parallel, wherein the injected current components circulate between the multi-level flying capacitor legs for balancing the voltages of the flying capacitors and cancel each other out among legs connected in parallel.

4. The control method according to any of the above claims, wherein the injected currents are DC currents.

5. The control method according to any of the above claims, wherein the injected currents are AC currents.

6. The control method according to the previous claim, wherein the injected currents are substantially phase-shifted to each other such as to cancel each other out.

7. The control method according to any of the above claims, wherein the injected currents have an amplitude dimensioned to control the voltage of flying capacitors given the size of the flying-capacitors and the control range of the duty cycles of the controllable switches.

8. The control method according to any of the above claims, wherein the respective injected currents have a controllable amplitude that gradually continously changes in accordance with the respective first current components, preferably in a linear manner.

9. The control method according to any of the above claims, wherein the respective injected currents have a controllable phase, and optionally a frequency, to be set in accordance with the respective first current components, preferably having a phase set to avoid the injected currents of parallel connected legs being out of phase of a sum of the first current components of said parallel connected legs and optionally a frequency set to be a multiple of the frequency of the first current components.

10. The control method according to any of the above claims, wherein controlling the switching of the plurality of controllable switches comprises controlling duty ratios and / or phase shifts of the controllable switches.

11. The control method according to any of the above claims, wherein controlling the switching of the plurality of controllable switches comprises: obtaining target duty ratios, obtaining sensed flying capacitor voltage values, calculating duty ratios of the controllable switches based on the obtained target duty ratios, the sensed flying-capacitor voltage values and a flying capacitor voltage balancing control function.

12. The control method according to the previous claim, wherein obtaining target duty ratios comprises: obtaining target first current components to perform the predetermined power conversion control function, adding a respective obtained first current component to a respective calculated current component to be injected to calculate a respective target leg current, obtaining sensed leg current values, and calculating current errors between the respective calculated leg current and the respective sensed leg current value,calculating the target duty ratios based on the calculated current errors and the predetermined control function.

13. The control method according to any of the previous claims, wherein calculating current components to be injected comprises: obtaining a Cutoff Peak Amplitude value, wherein when an average leg current is equal to said Cutoff Peak Amplitude value the target injected current of that leg is zero, and obtaining a Maximum Injection Peak Amplitude value, wherein when an average leg current is zero the target injected current of that leg has an amplitude equal to the Maximum Injection Amplitude value.

14. The control method of the previous claim, wherein calculating current components to be injected further comprises: obtaining sensed average leg currents, comparing said sensed average leg currents with the obtained Cutoff Amplitude value, and calculating an amplitude of a respective current component to be injected by applying to said comparison a continuous transformation, preferably a linear transformation, up to a saturation value set to be the Maximum Injection Amplitude value.

15. The control method of any of the above claims, wherein calculating current components to be injected further comprises:- obtaining respective phases, and optionally a frequency, of the current components to be injected,- generating the respective current components to be injected based on the respective obtained phases, and optionally frequency, and the respective calculated amplitudes.

16. The control method according to any of the above claims, for upon start charging the flying capacitors of all legs in a balanced manner using the injected circulating currents, and preferably for progressively increasing the leg currents up to nominal values.

17. The control method according to any of the above claims, wherein when the leg currents are higher than a predetermined low load current threshold, the calculated second currents to be injected are zero and the first current components are such as to control the voltages of the flying capacitors based on regulating a phase shift in the duty ratios of the switches of an FC leg.2518. A multi-level flying capacitor power converter configured to perform a predetermined power conversion control function, said function comprising preferably regulating a current to a load and / or correcting a power factor, said power converter comprising: a plurality of multi-level flying capacitor legs connected together, wherein each multi-level flying capacitor leg comprises at least one flying capacitor, a plurality of controllable switches and an inductor, and characterized in that it comprises: a controller configured to continuously calculate current components to be injected for controlling the voltages of the flying capacitors over the whole range of leg currents, the controller being further configured to control the switching of the controllable switches to obtain leg currents each comprising a respective first current component associated with the predetermined control function, and a respective injected current component associated with the respective calculated current component, wherein the injected current components cancel each other out within the power converter.

19. The converter according to the previous claim, wherein the controller is further configured to continuously calculate each respective second current component to be injected from a respective sensed average leg current.

20. The converter according to any of the last two previous claims, wherein the plurality of multilevel flying capacitor legs are further connected in parallel, and the controller is configured such that the injected current components circulate between the multi-level flying capacitor legs for balancing the voltages of the flying capacitors and cancel each other out among legs connected in parallel.

21. The converter according to any of the above converter claims, wherein the controller is configured such that the injected current components are DC currents.

22. The converter according to any of the above converter claims, wherein the controller is configured such that the injected current components are AC currents.

23. The converter according to the previous claim, wherein the controller is configured such that the injected current components are substantially phase-shifted to each other such as to cancel each other out.

24. The converter according to any of the above converter claims, wherein the controller is configured such that the injected current components have an amplitude dimensioned to26 control the voltage of flying capacitors given the size of the flying-capacitors and the control range of the duty cycles of the controllable switches.

25. The converter according to any of the above converter claims, wherein the respective injected currents have a controllable amplitude that continuously changes in accordance with their respective first current components, preferably in a linear manner.

26. The converter according to any of the above converter claims, wherein the respective injected currents have a controllable phase and optionally a frequency to be set in accordance with their respective first current components, preferably having a phase set to avoid the injected currents of parallel connected legs being out of phase of a sum of the first current components of said parallel connected legs and optionally a frequency set to be a multiple of the frequency of the first current components.

27. The converter according to any of the above converter claims, wherein the converter is an AC- DC converter, preferably a multi-phase rectifier for power rectification and / or power factor correction.

28. The converter according to any of the above converter claims, wherein the converter is a DC- DC converter for regulating power to a load, preferably a buck converter.

29. The converter according to any of the above converter claims, wherein the converter is a DC- AC converter for regulating power to a load and / or correcting power factor, preferably a multiphase inverter, typically a three-phase inverter.

30. The converter according to any of the above converter claims, wherein the multi-level flyingcapacitor legs are 3 -level flying capacitor legs, comprising four switches and one flying capacitor, said four switches being connected in series with each other across the input while the flying capacitor is connected in parallel to the middle two switches, a middle point between the middle two switches forming an output point.

31. The converter according to any of the above converter claims, wherein the power converter is for powering a pulsed load, preferably an electrolyzer.2732. A computer readable medium configured to store instructions which, when executed by a controller of a converter according to any of the above converter claims, cause said power converter to carry out the steps of any of the above method claims.

Images