Closed-loop-controlled and open-loop-controlled compensation of magnetic drift in the transformer of an isolated full-bridge DC-DC converter

Abstract

The invention relates to a method for actuating a transformer (2), which is fed by an inverter (1), for dynamically supplying power to an electrical load (3), wherein the inverter (1) is designed to convert a DC voltage (U) into an AC voltage (u1) and has controllable switching elements (Q1 to Q4) for this purpose, and wherein the method has the following steps: a) connecting a secondary side (2'') of the transformer (2) to a load (3); b) receiving a chronological progression of an input variable (α); c) actuating the switching elements (Q1 to Q4) of the inverter (1) for electrically supplying an alternating sequence of positive and negative voltage pulses (p) to a primary side (2') of the transformer (2), the actuation of the switching elements (Q1 to Q4) being defined by a drift-compensating control variable (zk) derived from the sampled input variable (αn), the drift-compensating control variable (zk) being calculated on the basis of the following steps: c1) applying a measure to the sampled input variable (αn) in order to ensure compliance with an idealized maximum reference magnetization; c2) detecting a drift (D=W(mactual)-W(m0)) of the magnetization; and c3) carrying out a conversion rule to reduce any drift (D) of the magnetization that is identified in step c2).

Description

[0001] P25086pct SWITCHING VOLTAGE REGULATOR: METHOD AND DEVICE The invention is based on the recognition that there are numerous manipulation measures for controlling an inverter intended to prevent the occurrence of saturation in a transformer supplied by the inverter, which in turn feeds a load. In addition, numerous measurement methods are known for detecting an offset in the transformer magnetization. r r nIt is known that the most symmetrical operation of a transformer is desirable in order to optimize the magnetic utilization of the transformer and thus keep the size and associated costs of transformers as low as possible. Another possibility to reduce the required size of transformers is to increase the switching frequencies of the inverter. This means that a lower inductance of the transformer is sufficient to transfer the same power as a transformer operated at a lower frequency. Increasing the switching frequency is therefore a popular measure for reducing the size and costs of transformers used in switching voltage regulators. However, there are limits to the technically possible switching frequency, which are defined in the .are generally determined by the maximum switching power and switching frequencies of the components used in the inverters. Recent developments have enabled a significant increase in switching frequencies. However, with increasing switching frequency, the tendency for the magnetizing current to drift increases, particularly because of technically unavoidable inaccuracies. s n in r d n The switching edges of the switching elements are playing an increasingly important role. The invention relates to semiconductor-based transformers, in particular inverters r Others from a traditional transformer. A transformer cannot produce DC voltage r G nThe magnetic flux is a time integral of the primary voltage, Ф = ∫u1·dt, and cannot exceed a certain saturation limit in the positive and negative directions. Outside the saturation limits, the primary and secondary sides of the transformer form coupled air coils, and the main inductance LH of the transformer is almost zero. .The magnetization requirement in the = Ф / LH grows beyond all limits, the transformer forms a P25086pct z 2 short circuit. The transformer is an unstable component, and the magnetization cannot be detected in a sufficiently timely manner to prevent a saturation short circuit. Magnetization, in short, refers to a standardized quantity that is only defined within the saturation limits and refers to both the flux and the magnetizing current. In other words, the invention relates to transformer switching voltage regulators, where control is achieved by setting the pulse duration of the alternating positive and negative voltage pulses applied to the transformer. The fundamental safety problem here is the sudden saturation of the transformer, which can occur even during steady-state operation without countermeasures.Known countermeasures detect a DC component in the magnetizing current, which is counteracted in an internal control loop using pulse correction. Pulse correction affects the ratio. ’ ratio of the duration of the positive to w the negative . ven pulses of the primary voltage, which is why the pulse width specified by an external control loop is not changed in the switching period average. However, the external control should not generate a DC component in the primary voltage, which is why several thousand switching periods are required, and ’ m to a new pulse width for a changed operating condition. Usually, an over-dimensioning of the transformer was required to accelerate the outer position of the pulse width. A significant acceleration of the switching frequencies occurred with the introduction of WBG transistors, whose switching frequencies i r m nCompared to previously used transistors, they can be 100 times higher, but this has exacerbated the problem of the very slowly changing (average) pulse duration. Accordingly, large energy storage devices are required, particularly in order to be short-circuit proof and, in the case of dynamic loads, to externally compensate for the slow pulse duration adjustment, which is only possible to a certain extent. The solutions EP3772166A1, AT511298B1, as well as DE19634713A and EP898360B, and EP2812987B1 are known for this purpose. The disadvantage is that no information about the nuclear magnetization is fed back to the pulse width modulators described in these solutions, and these solutions are limited to applications for power outputs below 10 kW, where almost identical power switches and reasonable additional expenditure, such as ..E.g., the P25086pct 3, so-called zero voltage switching, ensures that the nuclear magnetization does not migrate during steady-state operation. The solutions DE102017118296A1 and EP3840203A1 also require additional measures. The timely recording of the nuclear magnetization by measuring a voltage linked to the flux, e.g., the primary voltage u1, followed by integration, i m ∙ -L H= Ф = ∫u1∙dt, and pulse shutdown when the thus estimated magnetization reaches a limit value, is apparently simple and inexpensive, but is not scalable or becomes very expensive. Digitally, there is a quantization uncertainty, and in analog, the offset of an integrator circuit cannot be fully compensated. A growing long-term error arises between the actual magnetizing current occurring in the transformer and the estimated one, like a clock that runs fast or slow. Assume that in steady-state operation, no wandering or drifting of the nuclear magnetization actually occurs. The integral over a voltage linked to the flux wanders, i.e., deviates increasingly in a positive or negative direction from the actual nuclear magnetization over time.If the detected magnetization migrates in the positive direction and the control pulses for the positive primary voltage pulses are shortened, the core magnetization is driven in the opposite direction. Once a magnetization offset has been established, transient V leads to z Changes in the pulse width of the primary voltage quickly lead to a saturation short circuit that can no longer be interrupted or at least forces an interruption in operation so that the circuit breakers can cool down. Yao, P. et al., "Flux Balancing Control of Ungapped Nanocrystalline Core-Based Transformer in Dual Active Bridge Converters," IEEE Transactions on Power Electronics [online]. November 1, 2020 (01.11.2020). Vol. 35, No. 11, pages 11463–11474, and Ortiz, G. et al., "Flux Balancing of Isolation Transformers and Application of “The Magnetic Ear” for Closed-Loop Volt–Second Compensation," IEEE Transactions on Power Electronics [online]. ”wer Electronics [on 14 (08—.201 e4cline).August 20 )o. Vol. 29, No. 8, pages 4078–4090, concern the bidirectional but complicated to control DAB structure. However, the temporal relationship between the idealized input variable and the reference magnetization shown in Fig. 4 only applies if the primary leakage inductance can be neglected compared to an increased secondary coupling inductance (Fig. 5), as is the case in PANOVYURI ET ALL: “Novel transformer-flux --balancing control of dual-active-bridge bidirectional P25086pct ” 4 converters” 2015 IEEE APPLIED POWER ELECTRONICS CONFERENCE AND EXPOSITION (APEC), IEEE, March 15, 2015, pages 42-49. All solutions, generally for DAB, have in common that the control ’c and not now m ur slowly rbecause the mean value of the magnetizing current is regulated to zero and can only be measured with sufficient accuracy over several switching periods. Known and proven methods and circuits for the fixed linking of pulse and pulse integral are part of the inventive ’ procedure and will be described at the appropriate point. Because they are comparatively easy to stabilize – meaning that saturation short circuits are prevented and load short-circuit strength is achieved – alternative forward converters and flyback converters predominate in the power range below 1 kW. However, full utilization of the transformer is only possible with an inverter. At high power levels, the inverter is operated at full control for safety reasons, and the input current is adjusted with an upstream switching regulator. This involves a high level of effort, a relatively high zReduced efficiency and complicated control, but the power controller is short-circuit proof. In summary, the prior art can be said to be that previously known measures that manipulate an input variable to avoid transformer saturation prevent dynamic operation, and additional measures are necessary to disconnect the inverter from the input voltage in a timely manner in the event of a load short-circuit. The aim of the invention is to utilize the advantages of the forward converter principle—minimal component load, simplest control, highest efficiency—when the transformer is connected directly to the input voltage via an inverter, or to make the inverter cost-effectively short-circuit proof, as is the case with the forward converter. The present invention therefore relates to a method and a device in the context of a dynamic switching voltage regulator. P25086pct 5 D —The present invention provides a — An innovative combination of two measures is proposed, namely the combination of a test and, if necessary, manipulation rule according to step c1) with an interval-based temporal check of the actual magnetization to detect drift and obtain a drift-compensated control variable (according to step c3 in conjunction with step c2). The invention therefore fulfills the need to decouple the external actuating speed from the slow internal actuating speed. The invention functions with all possibilities for detecting the drift of the magnetizing current (direct pulse-pause value comparison, zero-crossing comparison with Hall voltage, evaluation of the currents in the transformer), and this while allowing high external dynamics as well as a pulse-to-pu n ls current limit. u rThe invention relates to a method for controlling a transformer fed by an inverter for dynamically supplying an electrical load, wherein the inverter is designed to convert a direct voltage into an alternating voltage and has controllable switching elements for this purpose, wherein the method comprises the following steps: a) connecting a secondary side of the transformer to a load, b) receiving a temporal profile of an input variable in the form of a temporal profile of a duty cycle and obtaining a sampled input variable by means of sampling —the input variable starting from an initial point in time continuously up to an actual point in time, wherein the input variable is provided for specifying a desired secondary-side output variable of the transformer, in particular an output voltage, to the load, c) controlling the switching elements of the inverter for electrically supplying a primary side of the transformer with an alternating sequence of positive and negative voltage pulses of the alternating voltage for outputting the desired output variable to the secondary side, wherein the control of the switching elements is determined by a drift-compensating control variable derived from the sampled input variable, wherein this drift-compensating control variable is calculated using the following steps: P25086pct 6c1) applying a measure to the sampled input variable to ensure the maintenance of an idealized and limited reference magnetization and obtaining a correspondingly tested,possibly manipulated, input variable, hereinafter referred to as idealized input variable, whereby the idealized input variable satisfies the following condition at each actual point in time, : : 0.5 c2) Comparing a —by measuring the actual value of the magnetization recorded at intervals during the operation of the transformer with an idealized value derived from the idealized input variable (i.e. a generated reference value) of the magnetization to detect a drift - W(m0)) of the magnetization, c3) executing a conversion rule to convert the idealized input variable into an - idealized control variable and - depending on the result of the comparison according to step c2) - manipulating the idealized control variable to obtain the drift-compensating control variable to reduce any drift of the magnetization determined in step c2). r Values ​​of the idealized input variable are to be understood as a temporal sequence according to the discrete steps n=1 to N, where n is a natural number and N increases with increasing time between the first measurement in a cThe difference between the initial time n=1 and the actual time n=N increases. They are included in the ongoing calculation of the idealized control variable z0. The duty cycle can be, for example, the duty cycle of a PWM, i.e., if the duty cycle is 100%, then the output voltage reaches a maximum value – and depending on the consumer, the output power also reaches a maximum. — value. The sampled input variable has values ​​between 0 and 1 and cannot assume negative values. Therefore, the following can be provided (see Fig. 3a):α 0,n = t on / T n P25086pct 7In particular, it can be provided that the switching elements are controlled according to the compensated control variable according to S ,,ch ”ritt c3). Incidentally, all variables referenced with the reference symbol “z” refer to r t become, impulse sszüge. Furthermore, it can be provided that for each discrete sampling time at least steps c1) and c3) are carried out and if step c2) is omitted, the steps r itt c3) last calculated manipulation up to a rn Time of a ern r current measurement by performing the step r itts c2). In particular, it can be provided that the drift of the magnetization detected by the measurement according to step c2) is stored and stored until a new measurement is made and compared with this new measurement, the result of the comparison being taken into account to the extent of the subsequent manipulation in step c3). . . Furthermore, it can be provided that an interval time duration of the interval-wise measurement according to step c2) is dynamically adapted to a speed wintensity of the drift, whereby with increasing drift and / or increasing drift speed the interval time is shortened and with decreasing drift and / or decreasing drift speed the interval time is extended . In particular, it can be provided that an interval time duration of the intervalw e egg s This measurement is carried out taking into account the operating conditions of the inverter. For example, a drift model can be derived from past experience and the compensating measures can be designed dynamically accordingly. For example, the drift rate can vary depending on the temperature, changes in the load current, and changes in the ' ing the pulse width bz 'w. Duty cycles or the associated jitter change. Such empirical values ​​can be taken into account in the drift model. Furthermore, it can be provided that an average magnetization value is recorded over a time interval whose interval duration is at least one switching period duration, and wherein the average value is formed over the same time interval using the idealized magnetization value, wherein any drift is determined by calculation by forming the difference between the average magnetization value and the average value of the idealized magnetization. P25086pct 8 In particular, it can be provided that the measure is applied to the sampled input variable in step c1) in such a way that a magnetization average value, which is obtained by a time integral of an idealized reference magnetization starting from the initial time up to any actual time, always lies below a predefinable average limit value.Further information on this measure has become known from EP3772166A1. Alternatively, it can be provided that a mean value is measured using a magnetization and a time mean value is provided for the reference magnetization, which has the value zero, wherein the drift is also calculated in step c2). In particular, it can be provided that the measure is applied to the sampled input variable in step c1) in such a way that the time integral of an idealized reference magnetization, starting from the initial time up to any actual time, has a zero crossing in each half period, so that a magnetization mean value determined over an integer number of switching periods is zero.Furthermore, it can be provided that the idealized control variable according to step c3) is manipulated by at least one of the following measures: manipulating the pulse lengths; by symmetrizing successive pulses; by blending successive sampled input variables according to the rule ^n,new=: (^n_-1+^n) / 2. Such measures can also be combined with one another. Further information on this can be found, for example, in the documents AT511298 B1 and WO2021019066A1. Various measures are known from the prior art to prevent the magnetization of a transformer from exceeding the standardized value 1 (above which saturation occurs). Some known measures are listed below as examples. 'hrt, which can be depicted in step c1) of the present application and can also be combined in connection with the present invention, but have not yet had the following steps c2) and c3) of the present invention: 1. Measure (AT511298): P25086pct 9 It r are α n from the control voltage α(t) sampled values ​​and as long as the above condition is fulfilled, is manipulated identically α0cm 0 , ergi,,n= α n .If m reaches the maximum value 1, in the case of positive equality (n=2k) we have m0,n + 0,,n 0 den o0,, mnn.-1 n aximaleno, value -1 n, ( _n=2k+1) holds m0,n-1 - 2,α, h α0 ,,n = -o0,n = (1 - m0,n-1) / 2.,-“12α = 1 where m ≤ 0 and the manipulation of α occurs nacReaching m 1, and it o, he —the manipulation of α follows an after α 0,n = (1 + m 0,n-1) / 2. According to this case distinction, when a magnetization upper limit is reached, the manipulation of αn takes place according to α0,,n = (1 + |m0,,n_-1|) / 2, see Fig. 6.2. Measure (DE19634713A, EP898360B): m 0,n := α0· (-1) n , such that |m |= α ≤ 1. 0 ,,n : ,n 0, - ( , 0,N 0,N m = m + 2 -1) n α , . with above he Settlement 0 o,,n ) o, = nn_-1 0,n α (-1 n α 0,n-1 (-1 n ) a n- o 1 , + 2(-1) n α 0,n α = 0,n -α0,n-1 + 2αr , gives a M o 0,n e ,n a _ nipul a a o, , tion of α n after α 0,n = (α 0,,n-1 + α 0 ,,n) / 23. Measure (EP 3772166 A1) It is known that ^ oc(t) can be manipulated in such a way that its time-sampled curve cannot generate a DC component in m0. (For example, when stopping, with a last (N+1)-th pulse, not only is m0,N+1 controlled to zero, but the instantaneous mean value from the 1st pulse,t, is continuously calculated as an integral over the reference magnetization m0 or double integral over the reference pulse train p '·zz0, generated by the idealized input variable, equal to zero..) In particular, it can be provided that during the pulse pauses, the actual value of the magnetization present at the respective time is adjusted by secondary-side current measurement. ’current and thereby the actual value of the magnetization is deduced, whereby the controllable switching elements of the inverter are switched off during the measurement period. The drift of the magnetization is calculated from the difference between the P25086pct 10 pulse-pause values ​​of the actual magnetization and the reference magnetization, whereby the accuracy is increased by averaging the differences from several half-periods. Furthermore, it can be provided that the zero crossing of the actual magnetization is temporally recorded by a Hall sensor and the drift is compared with the value of the reference occurring at this time. r z magnetization is calculated, taking into account a driver run-through time, which is the time until an edge of z k at the switching elements r ducks arrive. n In particular, it can be provided that the switching elements WBG frequencies can be up to a few MHz, e.g. up to 10 MHz, bet —-Tr ragra eannns .sistors are. The switching s rFurthermore, the invention relates to a pulse width modulator which is used to s sg r r r size (^ c ) and a characteristic value (W r is set up, the input ( nm,ist )) the nuclear magnetizer r ung (m ist ) and, according to the method according to one of the preceding claims, a drift-compensating control variable (e.g. k ) to reduce any drift (D) of the magnetization detected in step c2).Furthermore, the invention relates to an energy conversion device for dynamic V r eSupply of an electrical load, wherein the energy conversion device comprises the following: - an inverter with a voltage input for receiving a direct voltage, wherein the inverter is configured to convert a direct voltage into an alternating voltage and has controllable switching elements for this purpose, - a transformer fed on the primary side by the inverter, and - a computing unit with an interface for receiving an input variable, wherein the computing unit is configured to receive the input variable and W(mist) and, according to the method according to the invention, to calculate a drift-compensating control variable for reducing a magnetization drift possibly determined in step c2), and to control the switching elements by coupling the drift-compensating control variable to the inverter. P25086pct 11 The invention is i a m ' Consequence a nd using an example uand non-limiting embodiment, which is illustrated in the figures. Therein, Fig. 1 shows a schematic representation of an embodiment of an energy conversion device according to the invention, which is set up to carry out the method according to the invention, Fig. 2 shows a block diagram of the method according to the invention, Figs. 3a to 3e are the relationship between the duty cycles α0,,n (idealized input variable), the pulses of a pulse train z0 (also idealized control variable , called), and an idealized and normalized magnetization m0with the amplitudes m 0,n shown, Fig. 4a a further very general representation of an embodiment of the invention, and Fig. 5 describes an exemplary detailed representation of the representation according to Fig. 2. For a better understanding, the following figures are attached: Firg.6 Basic circuit diagram for the manipulative scanning according to c1), so that the test condition is met. Fig. 7 Time profiles of the signals in Fig. 6, Fig. 8 Flowchart of an exemplary method according to claim 8, Fig. 9 Further flowchart of an exemplary method. Figure 4b relates to further details and / or additions. In the following figures - unless otherwise stated - the same reference numerals designate the same features. P25086pct 12 Figure 1 shows a schematic representation of an embodiment of an energy conversion device according to the invention, which is set up to carry out the method according to the invention. Therein, the energy conversion device vThe device comprises an inverter 1, which feeds a transformer 2, via which in turn an electrical load 3 can be dynamically supplied. The inverter 1 is designed to convert a direct voltage U into an alternating voltage u1 and for this purpose has controllable switching elements Q1 to Q4. An input variable ^ can be specified, for example, by a user, whereby this input variable can be used, for example, to specify an output variable S out , in particular an output voltage U out , an output current I L2 or a related quantity can be used. Fig. 2 shows a block diagram of the method according to the invention, in particular with a pulse width modulator (PWM) 8 or the computing unit 5. The output variable zk is determined by two inputs, the input for receiving the input variable α and a further input for receiving the measured variable W(m ist).In step c1), the input variable α is sampled manipulatively or an already sampled input variable ^n is manipulated and the idealized input variable α00 is obtained, see e.g. Figs. 6 and 7. Reference magnetization m and the reference magnetization m0 is processed by function block 4' 0 and applied to the A ’ analogous to the recording of m ist to a comparable value W(m ) is provided. In step c2) W(m ist ) is compared with W(m0) and a drift D is output. In step c3) the idealized sampled values ​​α 0,,n a correction dependent on the drift D is added, see Fig.5, which, if control variable z k Primary voltage pulses of alternating polarity ’ ät p is suitable for detecting the course of the nuclear magnetization m istto the course of m0. In Fig. 3a to 3e the relationship between the duty cycles α0,n (idealized input variable), the pulses of a pulse train z0 (also idealized ’ rte tax rate , size), and an idealized and normalized magnetization m0 with the amplitudes m 0,n The pulse train z0 correlates with the idealized input variable α 0 , e switching period, consisting of two half cycles, where ,,n . In Fig.3a to 3d a . in this example the P25086pct 13 sampled input variable α n is chosen so low that already for α n = α0the following condition according to step c1) is fulfilled ,,n after , is: That is, the idealized input variable α 0,n can in this case be calculated with the sampled input variable α n be equated ,n, since no manipulation of the sampled input variable is required. The preservation of the idealized magnetization and the drift D is shown. Fig. 3a shows a periodic sawtooth z ancestor-shaped auxiliary voltage u h with amplitude 1 (in standardized representation) and the tax r voltage and ss , which is a tactile r hey a relationship α a (t) corresponds to the duration of one switching period. Shown are the (n-1)th and the n ''. — -th tooth of the auxiliary voltage u h over the normalized time t , The number n also denotes the starting time of the n-th half-cycle and this half-cycle is weighted with the sign (-1)n, see Fig. 3b. The n-th idealized and normalized primary voltage pulse u 10 n α 0,,n and is as p·z0 ers ,,n shows the normalized amplitude (-1) and the duty cycle. ‘ FIG. 3c z aeigt s the (n-1)-th and the n-th pulse of the generated pulse train z0, with the duty cycles α 0 ,,n-1 and α 0,,n . . In FIG. 3d, the time course of the normalized magnetizing current (m0) is shown, which in turn also corresponds to a normalized magnetization (provided there is no saturation and linear behavior of the inductances) of the transformer 3 .This curve, represented by a bold line, results from the activation of the switching elements Q1 to Q4, which can be derived from the pulse train z0, to supply current to the primary side of the transformer 3 according to the polarity shown in Fig. 3b. As an informative supplement, a dashed line is also shown, which would represent an idealized magnetization if the duty cycle were 1 and therefore the full magnetization swing from -1 to 1 would occur (assuming that the transformer) would have been magnetized to the value -1 at time n - -1. In this case, a polar pulse with duty cycle α 0,n a change of m 0,n to m umI'rlO0,n - ' m Irl-O0,n,n—-1 = ( (_-1) )n· '2 a^00,,n -. ra -1 0,nm n , o," o, P25086pct Zoo0m ,n (-1) . o 14 The pulse train z0 generates the curve m o0 with the amplitudes m o0,n = 0-2α 0,1 + 2α 0,2 -…+ 2α · n, ao, am ...For the i “dealized input variable α 0,,n for the impulses z 0,n the idealized control variable z0 is subject to the condition , ≤ bzw. " ' ' The invention will be discussed in detail below with reference to the figures: In other words, the following can be said about figures 3a to 3e: The beginning and end of a pulse of the idealized control variable z0 are manipulatively sampled and the pulse durations α0,n∙Tn are output, so that an ideal inverter supplies the primary voltage p∙U∙z0 to the transformer. , sf ' ormator would switch . .UB denotes the nominal voltage of the input voltage level U, which can exhibit fluctuations and in a half-period T n, in which a positive or a negative primary voltage pulse of a switching period of duration 2T n is generated, the value U n All values ​​shown are standardized. The duty cycle α 0,n is the duty cycleTon,n of the n —th pulse of pulse train z0 related to the maximum duty cycle Tn. The time t is also set to T n normalized and is the normalized time t' ’ plotted on the abscissa. The reference magnetization m0 increases or decreases during a duty cycle α 0,n ∙T n at a speed corresponding to the prevailing operating voltage U n proportional. By normalizing the time - e.g. by increasing the clock frequency of a digitally operating Pu —lswidth modulator (counter) is changed proportionally with the input voltage – the curves of the displayed quantities are unaffected by the input voltage U. The following formulas are mentioned in this context:0 ≤ α=:uS, α0 ,,n, ≤ 1 ; t '' = t / Tnn = t -·Un / (UB· 'TB ); p = (-1) nn ; αo0,,n = ton / Tnn,, u1 ≈ ~ p· -U·ton / Tn P25086pct 15m0 = im0 / Im, = Ф / Ф ; in the linear range -1 ≤ m ≤ 1 of the reference magnetisIm ,,max (Dma ax ,max 0 0,max 0 ationI = Ф / x / LH =UBTB / / (2 ZLH) = UnTn / / (2 ZLH)m ^^ o0, ே v = m ^^0 ,0 ∙ ∑ே − ^ ∙ ^^0 with ^^ = 0 (the magnetization at the time of switching-un'1g" w— ^ 00first input is assumed to be 0'n. With regard to Fig. 3e, it should be mentioned that it shows an example in which m has drifted in the positive direction because the amplification of zk to alternating positive and negative pulses of u1 was not carried out absolutely accurately. K is still zero, so that zk = z0. The drift results from the difference W(m) - W(m0), where W can be an instantaneous value function (sampling in particular of the pulse pause values) or a mean value function (filtering): D = m(tmess') - m0(tmess') = ms - m = m - m . The invention enables a function d ’ the — ser Drift' t,n 0,n AV 0,AVraphe Redu towards that — W , ert zero v d — through , G v Suitable choice of size z k.With regard to Fig. 1a and Fig. 1b, it should be mentioned that the invention relates to a method for controlling a transformer 2 fed by an inverter 1 for the dynamic supply of an electrical load 3. The inverter 1 is designed to convert a direct voltage U into an alternating voltage u1 and has controllable switching elements Q1 to Q4 for this purpose. The method comprises the following steps: r Please note: a) Connecting a secondary side 2' " ' of the transformer 2 with a load 3, b) receiving a time profile of an input variable ^ in the form of a time profile of a duty cycle and obtaining a sampled input variable ^n by sampling the input variable ^ starting from an initial time n=1 continuously up to an actual time n=N, whereby the input variable ^ is used to specify a desired secondary-side output variable S out of the transformer 2, in particular an output voltage Uout , to which load 3 is provided, c) controlling the switching elements ’ nte Q1 to Q4 of the inverter 1 for the electrical supply of a primary side 2' of the transformer 2 with an alternating sequence of positive and negative voltage pulses p of the alternating voltage u1 for outputting the desired output variable S out to the secondary side 2'', whereby the control of the switching elements Q1 to Q4 is carried out by a P25086pct 16 drift-compensating control variable z derived from the sampled “ input variable ^n k is determined, whereby this drift compensating control variable z k is calculated using the following steps :: c1) Applying a measure to the sampled input variable ^n to ensure the receipt of an idealized and limited maximum reference magnetization |m0≤1| and receiving a correspondingly tested, possibly manipulated, input variable, hereinafter referred to as idealized input variable ^0,n, where the idealized input variable ^0,n at each actual time n=N is , The following condition is met: , c2) Comparing a d —by interval-wise measurement during the operation of the transformer 2 recorded actual value W(mist) of the magnetization mist with an idealized value W(m0) of the magnetization derived from the idealized input variable ^0,,n to detect a drift D=W(m )-W(magnetization,rnist 0) (see Fig. 2) of the Ma s o c3) Execution of a conversion rule to convert the idealized input variable ^0,n into an idealized control variable z0 and - depending on the result of the comparison according to step c2) - manipulation of the idealized control variable z0 to obtain the drift-compensating control variable z k to reduce any drift D of the magnetization detected in step c2). At time t=0, n=0 and in the time t ''>0 to t ''=1, n=1. More precisely, the initial time n=1 describes the first period of an auxiliary voltage, the temporal course of which repeats periodically and at the actual time n=N the N — -th period. .AN denotes the test condition according to step c1. The manipulation of ^n to ^0,,n can be carried out by a reduction according to ^0, ,n < ^n, whereby corresponding measures for this are already known from the prior art and have already been discussed previously in this application. Advantageously, the manipulation only takes place P25086pct 17 if the test condition A is not met, and then again advantageously r in such a r N the extent that the bed r e nu ingung is just fulfilled. On this W e egg sThis way, excessive intervention in the control or regulation can be avoided. The switching elements Q1 to Q4 can be controlled according to the compensated control variable zk according to step c3). It can be provided that for each discrete sampling time n at least steps c1) and c3) are carried out and if step c2) is omitted, the manipulation last calculated in step c3) is applied up to a . m the time of a new measurement by performing step c2). In this way, it is possible to omit the measurement of the magnetization or a correlating current for predefined periods of time. The measurement time intervals can also be variable and can be longer for low drift and shorter for increasing drift. It can therefore be provided that an I wThe interval time duration of the interval-wise measurement according to step c2) is dynamically adapted to a speed of the drift D, whereby with increasing drift D and / or increasing drift speed the interval time duration is shortened and with decreasing drift D and / or decreasing drift speed the interval time duration is lengthened ' In particular, it can be provided that an interval time period for the interval assignment takes into account the operating states of the inverter. e sen Mes s It can be provided that the drift D of the magnetization detected by measurement according to step c2) is stored and stored until a new measurement and compared with this new measurement, wherein the result of the comparison is taken into account to the extent of the subsequent manipulation in step c3). It can be provided that over a time interval whose interval duration is at least one switching period duration 2T, an average value m ist,,avthe magnetization m ist is recorded, and over the same time interval the mean value m00 is determined, with an approximately ,,av The drift D formed over the idealized value W(m ) of the magnetization by calculation by taking the difference between the mean value of the magnetization m ist and the mean value of the idealized magnetization is determined D = mist,AV - m0,A .The measure on the sampled input variable ^n in step c1) can be such,” Trace, v d.assen magnetization mean value m 0,,AV , which is determined by a time integral of an idealized P25086pct 18 reference magnet — tization m 0, , n==1 starting from the initial time n=1 up to any actual time n=N, always below a predeterminable mean limit value M g lies . Alternatively, it can be provided that with a magnetization W(m) an average value(m AV) is measured and for the reference o , z magnetization W(m0), a time average value (m ) is provided which has the value zero, wherein the drift (D) is calculated in step c2) with D = mactual,,A mavV- ' m0,AV. In particular, it can be provided that the measure on the input variable ^ o .or an already sampled input variable (^n) to obtain the idealized input variable ^0, ,n in step c1) is carried out in such a way that the time integral of an idealized reference magnetization (m0,n=1) starting from the initial time (n=1) up to any actual time (n=N) has a zero crossing in each half period, so that a magnetization average value (m 0,AV v) is zero. Manipulation of the idealized control variable z0 according to step c3) can be carried out by at least one of these measures: Manipulation of the pulse lengths T; by symmetrizing successive pulses; by blending temporally successive sampled input variables according to the rule ^0, ,n=: ^0,,n- .1 + ^0,n, / 2It can be provided that during the duration of the pulse pauses Toff (see Fig. 3c) by secondary-side current measurement to the actual value W(Im,ist) of the magnetizing current im,ist present at the respective time and thereby to the actual value W(mist) of the magnetizing s ung m ist it is concluded that , during the measurement period sng the controllable switching elements Q1 to Q4 of the inverter 1 are switched off. Furthermore, it can be provided that the zero crossing of the actual magnetization mist is temporally detected by a Hall sensor and the drift D is compared with the value of the reference magnetization W(m )=m (t - taking into account a r the driver through r duration n To0 0 0 Tv) is calculated, namely v , the o e o the time until an edge of z k arrives at the switching elements Q1 to Q4. More precisely, the initial time n=1 describes the first period of an auxiliary voltage, the temporal progression of which repeats periodically and at the actual time n=N passes through the N-th period .. P25086pct 19It can be provided that the switching elements Q1 to Q4 are WBG transistors. Fig. 1 also shows an energy conversion device 6 for the dynamic supply of an electrical load 3, wherein the energy conversion device v Device 6 comprises: An inverter 1 with a voltage input E1, E2 for receiving a direct voltage U, wherein the inverter is designed to convert a direct voltage U into an alternating voltage u1 and hi v controllable switching elements Q1 to Q4. Furthermore, the energy conversion device 6 has a transformer 2, which is fed by the inverter 1 on the primary side, and a computing unit 5 with an interface for receiving an input variable ^. The computing unit 5 is configured to process the input variable ^ as well as an actual value W(Im,actual) or the magnetization m, detected by a detection device 4. ist of the magnetization v troms to receive and according, According to the method according to the invention, a drift-compensating control variable zk is calculated to reduce a drift D of the magnetization possibly determined in step c2, and the switching elements Q1 to Q4 are switched on by coupling in the drift-compensating control variable z kto the inverter 1. Fig. 4a and b show a further very general representation of an embodiment of the invention. Therein, it can be seen that a voltage source U feeds the aforementioned inverter 1. This in turn supplies a transformer 2, which is designed to supply a load 3. The load itself can also have a voltage source U2, which can, for example, act as a generator and supply power to the voltage source U. The voltage source U can be an energy storage device such as an accumulator, and the voltage source U2 can be, for example, a photovoltaic system or other energy sources. U2 can even be present as an alternating voltage. The invention enables energy to flow both from the voltage source U to the "load" or U2, and in the opposite direction, in which case the load has a negative ,, oak ”n and is actually active as a generator. U2 corresponds to the reference symbol Uout in Fig. 1, but with the difference that U2, as mentioned, can be active as a generator and not as a load. P25086pct 20 The invention enables in detail a highly dynamic control of the said energy flow, whereby the invention ensures at all times that the transformer 3 is operated in the linear range of magnetization and thus the risk of saturation occurring. unu To control the energy flow or the corresponding control of the switching elements .The circuit arrangement according to Fig. 4b comprises the controller 7, which derives the control voltages α and β required for the desired energy flow from the actual values ​​of the voltage levels U and U2 recorded, whereby the voltage at the transformer and thus its magnetization m0 is determined by the control variable ^. Fig. 5 describes an exemplary detailed representation of the representation according to Fig. 2. For step c1, a number "1" is shown, which ze should not be confused with the reference symbol 1 and is therefore placed in quotation marks. This is to clarify that a limitation for the normalized magnetization m0 is provided, namely to the maximum value "te in the transformer "1", which is selected so that just no saturation effects occur. Furthermore, it should be mentioned that Fig. 5 indicates a further development, namely by using a targeted selection of the polarity p, namely towards p*. With p*, this can be a sign-based preselection of the polarity, ie a manipulation of the sign type of the polar r aims at a pre it a ity. For example, exclusively positive or exclusively negative pulses can be manipulated in their duration. This makes corrections more precise. This approach is already known from the state of the art, see document " ,, N ” O vvel Transformer-Flux-Balancing Control of Dual-Active-Bridge Bidirectional Converters” by Yuri Panov et al, Delta Products Corporation. If a drift D is detected, a controller R m compensate for this drift D by outputting a value K that is suitable for compensating the drift D by ultimately deriving a suitable value zk from it. K represents the manipulation according to step c3), which can be kept constant, for example, if step c2) is omitted. Figs. 6-9 serve to describe the features of component c) of the invention in more detail. P25086pct 21 Fig. 6 shows a circuit diagram for generating the long-term error-free relationship between the idealized control variable (pulse train) z0, output by the RS flip-flop (RS), and the pulse durations T n - ∙α 0,,n , the idealized input variable α provided by intermediate storage ' z 0 0,,ncher (D), with the reference magnetization integral m , output by the integrator (INT2), depending on the input variable (α) for the realization of the test condition according to c1). The components D with input u h , RS with input (start), and INT2, with input (p) are controlled by the signals (start), output signal of K1, and (sampling), output signal of (O). An auxiliary voltage generator consists of an integrator (INT1), which constantly integrates a constant, e.g. "1", and a comparator (K1), which compares the integrator with a constant. , z set , exit with egg , n, ” e.g. also "1", and outputs the signal (start). The comparator output is connected to the re - , E ” input of INT1. Integrator INT1 outputs the sawtooth-shaped auxiliary voltage u hThe signal (start) is then connected to the start input of INT2, as well as to the input of the toggle flip-flop (T), which outputs the polarity signal (p). The signal (sample) causes the buffer D to be loaded with the current value of the auxiliary voltage u. h , resetting the RS flip-flop and stopping INT2. A sample is triggered either by the signal (ma n ,stop gt from comparator (K3 x x stop ), generated by comparator (K2), or by signal (α ), generated ), triggered by the input signals of (O). The inputs , of comparator (K2) are the output m of INT2 and nd a " ,, -1" ” . The inputs of K3 are the input variables o0 the switching thresholds “1” u ße α and the auxiliary voltage u h . ,, ”Fig. 7 shows the signal curves for switching on a load short circuit, where the controller could output the exemplary curve of duty cycle α. Plotted in Fig. 7a are , iter the auxiliary voltage u h and the sampled values ​​of the idealized input variable α 0,1 to α 0, ,4 . Fig.7b shows the signal (start), Fig.7c shows the polarity signal p, Fig.7d shows the reference magnetization m0. . P25086pct 22 In Fig.7e the signal (sampling) is shown, as well as its source max stop or α n,stop and in Fig.7f . the idealized control variable z0. xThis procedure can be implemented digitally. There is a constant relationship between numbers and the number of clock periods of a clock signal, which characterizes digital circuits. According to EP898360B1, an analog circuit could also have proven itself in harsh industrial practice. Fig. 8 shows the flowchart that is run through in controller 5 in each period of the auxiliary voltage. Steps c1) to c3) of component c) of claim 1 are shown. Fig. 9 shows the flowchart of a pulse width modulator 8 according to device claim 15 and method claim 9. The invention is not limited to the embodiments shown, but is defined by the entire scope of the claims. Individual aspects of the invention or the embodiments can also be taken up and combined with one another. r e reference symbol cn in the claims are exemplary and serve only to facilitate the reading of the claims without limiting them. ’ ä a nken.

Claims

P25086pct 23 claims 1. Method for controlling a transformer (2) fed by an inverter (1) for the dynamic supply of an electrical load (3), wherein the inverter (1) is designed to convert a direct voltage (U) into an alternating voltage (u1) and has controllable switching elements (Q1 to Q4) for this purpose, the method comprising the following steps: a) connecting a secondary side (2' "') of the transformer (2) with a load (3), b) receiving a temporal profile of an input variable (^) in the form of a temporal profile of a duty cycle and obtaining a sampled input variable (^n) by means of, in particular manipulated, sampling of the input variable (^) starting from an initial time (n=1) continuously up to an actual time (n=N), wherein the input variable (^) is provided for specifying a desired secondary-side output variable (Sout) of the transformer (2), in particular an output voltage (Uout), to the load (3) m oc) Controlling the switching elements (Q1 to Q4) of the inverter (1) for the electrical supply of a primary side (2') of the transformer (2) with an alternating sequence of positive and negative voltage pulses (p) of the alternating voltage (u1) for outputting the desired output variable (Sout) to the secondary side (2' "'), wherein the control of the switching elements (Q1 to Q4) is determined by a drift-compensating control variable (zk) derived from the sampled input variable (^n), wherein this drift-compensating control variable (zk) is calculated using the following steps: c1) Applying a measure to the sampled input variable (^n) to ensure the receipt of an idealized and limited reference magnetization (|m |≤1) and a correspondingly tested, possibly manipulated, input variable, hereinafter referred to as the idealized input variable (^0,n), wherein the idealized input variable (^0,n) at each actual time (n=N) fulfills the following condition' llt : : — P25086pct 24 0,5 c2) comparing an actual value (W(m0,n sist)) of the magnetization (measured value sst) recorded by interval measurement during operation of the transformer (2) with an idealized value (W(m0)) of the magnetization derived from the idealized input variable (^ ) to detect a drift (D=W(rmnsist)-W(m o0)) of the magnetization, c3) executing a conversion rule to convert the idealized input variable (^0,n) into an idealized control variable (z0) and - depending on the result of the comparison according to step c2) - manipulating the idealized control variable (z0) to obtain the drift-compensating control variable (zk) to reduce any drift (D) of the magnetization determined in step c2). Method according to claim 1, wherein the switching elements (Q1 to Q4) are controlled according to the compensated control variable (zk) according to step c3).3.Method according to one of the preceding claims, wherein for each discrete sampling time point (n) r at least the steps c1) and c3) are carried out and if step c2) is omitted, the manipulation last calculated in step c3) is retained until a time of a new measurement by carrying out step c2). .

4. Method according to one of the preceding claims, wherein the drift (D) of the magnetization detected by measurement according to step c2) is stored and stored until a new measurement and compared with this new measurement, the result of the comparison being taken into account in the extent of the subsequent manipulation in step c3).

5. Method according to one of the preceding claims, wherein an interval time duration of the interval-wise measurement according to step c2) is dynamically adapted to a speed of the drift (D), wherein with increasing drift (D) and / or increasing drift speed the interval time duration is shortened and with decreasing drift (D) and / or decreasing drift speed the interval time duration is lengthened. P25086pct 25 6. Method according to one of the preceding claims, wherein an interval time duration of the interval-wise measurement takes place taking into account the operating states of the inverter (1).

7. Method according to one of the preceding claims, wherein over a time interval, the interval duration of which is at least one switching period (2T), an average value (mist,,av) of the magnetization (mist) is recorded, and wherein over the same time interval the average value (m0a,av) of the reference magnetization (m0) is formed, wherein any drift (D) is determined by calculation by subtracting the difference between the average value of the magnetization (mest) and the average value of the idealized magnetization (D = mist,av -m o0,mav ). . n s, 8. Method according to one of claims 1 to 7, wherein the measure is applied to the scanned Eingang n rm sgröße (^n) in step c1) is carried out in such a way that a magnetization mean value (m 0a,avg), which is given by a time integral of an idealized RA nfanc efer gszeitpunkt (n=1) up to a rn belie ren b n zm i ge rn agnetisier nIst —-Zeitp rung (m point (n o0,, = nn=1 )starting from the =N) is always below a predefined mean limit (M g ) lie u t n .

9. Method according to one of claims 1 to 8, wherein a mean value (mAV) is measured with a magnetization W(m) and a one-time mean value (m0,av) is provided for the reference magnetization W(m0), which has the value zero, wherein the drift (D) in step c2) with D=m AV is calculated . .

10. Method according to claim 9, wherein the measure on the sampled input variable (^n) in step c1) is carried out in such a way that the time integral of an idealized reference mag —netization (m0,n=1) starting from the initial time (n=1) up to any actual time (n=N) has a zero crossing in each half period.

11. Method according to one of the preceding claims, wherein a manipulation of the idealized control variable (z0) according to step c3) is carried out by at least one of the measures: manipulation of the pulse length rgnean (T); by symmetrizing successive pulses; dur I'ch V I‘ers tc without temporally consecutive sampled input variables according to the rule ^0,mn=: ( a^0,,n- _11++^10,n,)n / 2 P25086pct 2612. Method according to one of claims 1 to 6, wherein during the duration of the pulse pauses (Toff) the pulse pause value of the magnetizing current (i2=mist) present at the respective time is deduced by means of secondary-side current measurement, wherein during the duration of the measurement the controllable elements (Q1 to Q4) of the inverter (1) are switched off (D=mmsist,,n '-m0,,n).

13. Method according to one of claims 1 to 6, wherein the zero crossing of the actual magnetization (mist) is temporally detected by a Hall sensor and the drift (D) is calculated with the value of the reference magnetization W(m0)=(m0(t0-Tv)) occurring at this time (t0), taking into account a driver run-through time (Tv), which is the time period until an edge of the drift-compensating control variable (zk) arrives at the switching elements (Q1 to Q4).

14. Method according to one of the preceding claims, wherein the switching elements (Q1 to Q4) WBG —-transistors.

15. Pulse width modulator (5) which is designed to measure the input value (^) and the actual value (W(I m,,ist ), , W(m ist )) of the magnetizing current (I m,ist ) or magnetization (m ist ) and to receive a drift-compensating control variable (z k) to reduce any drift (D) of the magnetization detected in step c2).

16. Energy conversion device (6) for the dynamic supply of an electrical load (3), wherein the energy conversion device (6) comprises the following: - an inverter (1) with a voltage input (E1, E2) for receiving a direct voltage (U), wherein the inverter is configured to convert a direct voltage (U) into an alternating voltage (u1) and has controllable switching elements (Q1 to Q4) for this purpose, - a transformer (2) fed on the primary side by the inverter (1), and - a computing unit (5), in particular in the form of a pulse width modulator according to claim 15, with an interface for receiving an input variable (^), P25086pct 27wobmei the computing unit (5) is configured to receive the input variable (^) and the actual value (W(Im,,ist),,W(m )) of the magnetization current (Imd according to the method according to a ,,ist) or of the magnetization (mist) m of the preceding claim a drift-compensating control variable (z k ) to reduce any drift (D) of the magnetization detected in step c2) and to control the switching elements (Q1 to Q4) by coupling the drift-compensating control variable (zk) to the inverter (1).

Images