DE102014206438A1 - Circuit arrangement for operating n loads - Google Patents

Abstract

The invention relates to a circuit arrangement for operating n loads, n greater than or equal to 2, comprising an input having a first (E1) and a second input terminal (E2) for coupling to a supply voltage (Vin), n outputs each having a first (A11; A21) and a second output terminal (A21; A22) for coupling to the respective i-th load, i = 1 to n, a converter with a converter switch (S0), at least one converter inductance (L1) and n converter diodes (D1, D2) , wherein a series circuit of the Wandlerinduktivität (L1) and the converter switch (S0) between the first and the second input terminal (E1, E2) is coupled, n strands comprising at least one series circuit each having an ith electronic switch (Si) with i = 1 to n, an i-th of the n converter diodes (Di) and the respective i-th output, wherein the n series circuits are connected in parallel and coupled to the Wandlerinduktivität (L1), that in the converter inductance (L1) stored energy in the discharge phase of the converter by Leitendschalten the respective i-th electronic switch (Si) at the i-th output is provided, a drive device (12) for controlling the converter switch (S0) and the n electronic switch ( Si), wherein the drive device (12) is adapted to drive the converter switch (S0) with a drive signal (Gate0), which consists of n, the sub-phases assigned to n strands (Ti), and the electronic switch (Si) with a respective Drive signal (Gate1, Gate2) to control, wherein a Conduction Mode of the circuit arrangement is defined with respect to the current flow through the Wandlerinduktivität (L1), wherein a Leitendphase of the converter switch (S0) is framed by the Leitendphasen two different of the n converter diodes, wherein the drive device (12), the drive signals (Gate0; Gate1, gate2) for the transducer switch (S0) and the switches (Si) such that at least one of the n strands within its assigned subphase (Ti) is operated in Continuous Conduction Mode (CCM) or Boundary Conduction Mode (BCM) , and the switches (Si) outside the subphase (Ti) assigned to them are switched non-conducting.

Description

Technical area

The present invention relates to a circuit arrangement for operating n loads, n ≥ 2, comprising an input having a first and a second input terminal for coupling to a supply voltage, n outputs each having a first and a second output terminal for coupling to the respective i-th Load, i = 1 to n, a converter with a converter switch, at least one converter inductor and n converter diodes, wherein a series circuit of the Wandlerinduktivität and the converter switch between the first and the second input terminal is coupled, n strands comprising at least one series circuit of one each i-th electronic string switch with i = 1 to n, an i-th of the n converter diodes and the respective ith output with i = 1 to n, wherein the n series circuits are connected in parallel and coupled to the Wandlerinduktivität such that in the Wandlerinduktivität stored energy in the discharge phase de s converter can be provided by Leitendschalten the respective i-th electronic string switch at the i-th output, and a drive device for controlling the converter switch and the n electronic strand switch, wherein the drive device is designed, on the one hand to control the converter switch with a drive signal resulting from n, on the other hand, the electronic string switches with a respective drive signal to control, wherein a Conduction Mode of the circuit is defined with respect to the current flow through the Wandlerinduktivität, wherein a Leitendphase the converter switch is framed by the Leitendphasen two different n converter diodes.

The invention will be illustrated below as an example of LEDs as loads. However, it can also be used for other types of loads.

State of the art

Modern LED lighting solutions often require a homogeneous light emission characteristic. Since an LED as such represents a point light source, therefore, an additional, lossy optics or the use of a larger amount of LEDs, possibly also in conjunction with a corresponding optics, such as a frosted glass, necessary. In order to achieve maximum system efficiency, good thermal distribution and at the same time sufficient homogenization of the light with regard to the emission characteristic, the use of many, distributed low-power LEDs (for example OSRAM TOPLED) compared to the use of only a few high-power LEDs ( For example, OSRAM OSLON) often proven.

In order to limit the circuit complexity for the energization of larger quantities of LEDs, in addition to the series connection, which is often only possible to a limited extent, since SELV limit values must be adhered to, the parallel connection of several low-power LED chains is also performed. But even with high-power LEDs, a parallel connection is required in certain cases.

Due to production tolerances, LEDs have slightly different forward voltages, which must be compensated for when supplying parallel strings. Thus, attempts are made by different circuits to balance the current per branch. Ideally, the current of each string would be identical here, since the luminous flux of each LED is, to a very good approximation, proportional to the electric current flowing through it.

However, this always means that the power converted in the LEDs per strand is slightly different. This is due to the different phase voltages in ideally identical currents. This can result in thermal inequalities on the LED module, which in the worst case lead to a significant difference in temperature of the components (LEDs) of a strand or multiple strands. Without any thermal or electrical balancing mechanisms between individual strings connected in parallel, these inequalities increase as the forward voltage of an LED decreases with temperature, whereby hotter LEDs tend to draw more current than cooler ones. Since the lifetime of LEDs is highly dependent on their junction temperature, an overstressed strand would thus be subject to increased wear (drop in luminous flux) or a higher probability of failure.

In this context is from the WO 2012/164511 A1 a circuit arrangement for operating a light source is known in which is assumed by an impressed, uniform current as a source for energizing the system under consideration. The converter stage which feeds in there should, if possible, have no capacity at its output, with this stage being followed by a completely separate balancing circuit, which in turn is derived from two known DC-DC converters. The current ratio between the LED strings connected to this balancing circuit corresponds to the duty cycle within the balancing circuit. The power supply to this separate converter stage takes place at an unusual place, namely neither at the entrance nor at the exit of this Stage, but in their midst. It follows that for the derivation of this separate converter stage only those DC-DC converters come into question, which are symmetrical with respect to the filtering at their inputs and outputs: the inverting converter and the Ćuk converter. Due to said center feed, these converter stages can have more than two inputs or outputs there. The balancing circuit may be integrated into the supply converter stage. The considered converter operates primarily in the Discontinuous Conduction Mode (DCM), in particular the supply converter stage. The balancing circuit operates completely asynchronous to the supply converter stage and requires an approximately ideally smoothed DC as input. The supply converter stage must therefore be regulated to its output current. The possibility of integrating the inductance of a supply converter stage with current output with the input filter of the separate balancing circuit into a single inductive component is purely academic. The inductive part of the output filter of the supply converter stage must have a sufficiently low value to ensure DCM at given voltages and frequencies, whereas the input inductance of the balancing circuit must be large enough to smooth the input current sufficiently even across the voltage fluctuations of the connected load strings. To decouple switching or filter properties of these two sections from each other, a filter capacitor with respect to the common ground line is required just in between.

A generic circuit arrangement is known from the US 2010/0295472 A1 , This document discloses a power supply that includes a power supply, a plurality of output channels, and a controller. Each of the output channels includes a load and an electronic string switch with a reference voltage. All string switches are based on the same reference voltage. The driving inverters used are those with current output, for example buck converters, uc, forward and half bridges, as well as those with a voltage output, such as boost, choke-inverter and SEPIC.

As mentioned, LEDs, as well as other electronic components, production tolerances on their electrical properties. Because of this results in a parallel connection of several components or groups of components at one and the same source an asymmetry of the respective strand outputs and currents. In the special case of the LEDs, there are relatively large deviations of the forward voltage, which, in conjunction with their highly non-linear characteristic, is particularly disturbing in parallel circuits: the smaller the strand voltage, the more disproportionately higher is the current. This leads to an inhomogeneous light output between the strands, to an increased heating of the affected LEDs with faster aging or faster failure as a result.

Frequently a strand current symmetry takes place. By nesting or clogging, such as in the DE102010002228A1 is shown, moreover, attempts to achieve a good and uniform thermal coupling of the strands, wherein alternately LEDs of different strands are arranged one after the other. A string current balancing is achieved by current mirror circuits between the strings at the same end of a string. Alternatively, complex transducer topologies may be considered, which contain at least one current-compensated choke, such as in DE102010041613A1 , in DE102010041632A1 or in EP1788850A1 represented, or from a DC-free high-frequency alternating current per strand always use only one direction such. In DE102010041618A1 proposed. It is also known to use a separate independent supply converter per string. By multi-stage transducer topologies, whose parallel power amplifiers are used only for balancing, a strand current symmetry can also be achieved. Finally, reference should also be made to the use of downstream differential voltage or power converters, as described, for example, in US Pat WO 2012/164511 A1 emerge. The current mirror circuits generate undesirable losses, the mentioned current compensated chokes are comparatively large and expensive, and the circuits utilizing only one current direction are ultimately limited to an even number of strings connected thereto and require particularly large output capacitors.

The power balancing of LED strings (or other loads), such as those mentioned US 2010/0295472 A1 is known, can be particularly easy to implement in flyback converters. Here, the transmitted energy per switching cycle is constant to a good approximation. In the cited document for this purpose each strand has a rectifier diode, a capacitance, the load and a switch. Alternately, a string is activated by the switch in each switching cycle. In this strand, the cached energy coming from the primary side is transmitted. The shunt capacitors reduce the ripple in the load currents, and the diode connected in series with each string prevents the exchange of energy between the individual strings. Thus, each strand receives the same amount of energy per unit time, is thus supplied with identical power.

Through this solution, the strand performances are balanced, whereby a thermal compensation of the load strands takes place to each other. This avoids hot spots and thus significantly reduces the risk of rapid aging or rapid failure of LEDs.

In the generic circuit arrangements known from the prior art, the ideal state is characterized in that the strand voltages in the parallel strands are the same. Power or current balancing attempts to compensate for deviations from this ideal state in terms of reducing wear. The LEDs used in the strands are of the same type, i. they ideally have the same forward voltage and emit radiation of the same wavelength. In another embodiment, the LEDs used are of different types, for example, by mixing different colored light to set the desired color location without color conversion. In this case, the individual strands are equipped such that their respective sum voltages are as similar as possible to each other, and that the intensity of the primary light colors generated at symmetrized power leads to the desired light color and light intensity.

Presentation of the invention

The object of the present invention is to develop a generic circuit arrangement such that an extension of the range of applications is made possible.

This object is achieved by a circuit arrangement having the features of patent claim 1.

The present invention is based on the finding that, given a given topology, different light effects can be generated by setting a control which generates the desired light effect between the two extreme states-power balancing or current balancing.

This opens up the possibility of consciously providing strands of different strand tension, for example a strand with a strand voltage of 37V and a strand with a strand voltage of 40V. For example, 40V can be generated by twelve blue LEDs. 37V can be realized by several red and white, serially connected LEDs. With current balancing, both strands are equally bright. With power balancing, the strand with the higher forward voltage is darker. If, as provided in the example just mentioned, LEDs with different colors in the strands are used, a color control can be realized by the present invention. If the voltage difference between the highest and the second highest phase voltage in each case is greater than the largest possible voltage drop at one of the string switches, the string switch of the string with the highest phase voltage can be omitted and replaced by a short circuit.

By means of the present invention, the color location of the radiation emitted by the entire circuit arrangement can thus be varied as desired only by different activation of the switches of the circuit device. A user interface coupled to the drive device can be provided, on which a user can specify a desired color location, in which case the drive device converts this color location into a suitable drive of the mentioned switches.

According to the invention, the drive device is therefore designed to provide the drive signals for the converter switch and the switches of the strands in such a way that at least one of the n strands within the subphase assigned to it is operated in Continuous Conduction Mode (CCM) or Boundary Conduction Mode (BCM), and the switches of the strands are switched non-conducting outside of their assigned subphase.

In contrast to the present invention, the balancing circuit operates according to the mentioned WO 2012/164511 A1 completely asynchronous with the supply transformer stage and requires an approximately ideally smoothed DC as input. The supply converter stage according to the mentioned WO 2012/164511 A1 must therefore be regulated to its output current. In contrast, in the present invention, the balancing circuit directly processes individual energy packets which are generated by the series circuit consisting at least of the converter switch and a converter inductance and are buffered in this inductance. Furthermore, in the present case, a converter inductance is used both for the conversion per se and for the distribution to a plurality of output strands. The above-mentioned series circuit represents the minimum topology for a supply conversion stage.

As will be explained in more detail below, switching between the different operating modes (CCM, BCM, DCM) can be carried out continuously from power balancing to current balancing and vice versa.

A preferred embodiment is characterized in that the drive device is designed to control the drive signals for the To provide transformer switches and for the switches of the strands such that at least one of the n strands is operated within the subphase assigned to it in the DCM and at least one of the n strands within the subphase assigned to it in the BCM or in the CCM; or all n strands within the sub-phases assigned to them are operated in the CCM or at most one will operate within the sub-phase assigned to it in the BCM; or at least one of the n strands within the sub-phase assigned to it in the DCM and at least one of the n strands within the sub-phase assigned to it is operated in the CCM. In the first variant, it is possible to achieve power balancing or even an overcompensation such that a strand with a higher strand voltage acquires a lower power, while the second variant allows the same current to flow in strands with different forward voltages. With the third variant, intermediate states or transition states can be achieved.

Each subphase is preferably composed of an on time t _on , in which the converter switch is turned on and leads to the magnetization of the converter inductance, and an off time t _offi , in which the converter switch is nonconductive, but the respective i th string switch is turned on and ends either with the demagnetization of the Wandlerinduktivität or with the re Leitendschalten the converter switch.

In this connection, the drive device can be designed to provide the drive signals for all involved switches such that the on-times t _{on are} constant and the off-times t _offi variable. It can thereby be achieved that at maximum overall switching frequency, ie optimum utilization of the converter, the advantages of the BCM can be utilized for each switching operation, that power balancing takes place at the same time, and thirdly that the interference spectrum of the overall arrangement is spread to at least three individual frequency lines, as a result of which the electromagnetic Compatibility of the overall arrangement significantly improved.

Finally, a simple control concept for power balancing can be realized for this mode of operation, in which the power is set _{per se} by the on-time t _on , and the endpoints of the off-times t _offi and the switchover points between the strand switches are _detected automatically by recognition of the Abmagnetization of the converter inductance result.

Alternatively, the drive device may be designed to provide the drive signals for all involved switches such that the on-times t _on and the off-times t _{off are} constant. This corresponds to the standard mode of operation for the generic circuit arrangements and allows all the above-described extreme and intermediate states at almost any power and two output lines. Finally, the drive device can be designed to provide the drive signals for all involved switches such that the sums of the respective on-times t _oni and off-times t _{offi are constant} for i = 1 to n. By this measure, a symmetrization can be achieved even with more than two output strands.

Particularly preferably, the drive device is designed to switch two electronic switches in opposite directions during the Leitendphase the converter switch.

The converter inductance is preferably realized as a transformer with at least one primary winding and one secondary winding. This allows the input to be galvanically isolated from the loads, allowing it to meet relevant SELV regulations.

To prevent flicker effects, an i-th capacitor can be connected in parallel with the respective i-th output.

In a preferred embodiment, a shunt resistor is serially coupled to the converter switch. This allows the measurement of the primary current and thus on the known quantities DC input voltage, on-time and value of the Wandlerinduktivität also the measurement of the input power.

The converter is preferably designed as a flyback converter, as an up-converter, as a SEPIC converter or as a choke inverter converter. These converters make it possible to implement the invention, whereas the invention can not be implemented with a buck converter, with a uc converter or with a zeta converter. The reason for this is that in all three latter converter topologies, energy packets are transmitted from the input to the currently activated output line, not only during the off-times, but also during the on-times, which can overlap or depend on the exact switching time between the line switches are, whereby the above-described intrinsic symmetry property of a generic circuit arrangement would be lost. In other words, these three topologies would require at least one separate, independent output inductance per string, which would cost them close to the most unfavorable variant outlined above, where each string is powered by a complete converter of its own.

The i-th load preferably comprises at least one LED, but more preferably a series connection of a plurality of LEDs. Parallel circuits of series circuits of LEDs can also be provided.

In particular, "n" can be equal to two. In this case, the drive device preferably comprises a D flip-flop whose input is coupled to the input of the converter switch, wherein the Q output of the D flip-flop with a first electronic switch and the Q Output of the D flip-flop is coupled to a second electronic switch, wherein the D input of the D flip-flops with the Q Output is coupled.

As already mentioned, the present invention is based on the setting of a desired state between power balancing and current balancing by different activation of the switches, in order thereby to achieve desired optical effects. Nevertheless, for safety reasons, the activation device can also be designed in particular to prevent premature wear, to vary the partial phases, in particular if one of the strands has exceeded or fallen below a predefinable temperature threshold value. In particular, in this context, the LEDs may all be of the same type, i. Radiate radiation of the same wavelength.

Further preferred embodiments emerge from the subclaims.

Short description of the drawing (s)

Hereinafter, embodiments of the present invention will be described in more detail with reference to the accompanying drawings. Show it:

1 a schematic representation to illustrate the principle of current balancing by means of string resistors;

2 a schematic representation of a flyback converter with secondary-side strand power balancing;

3 the temporal voltage curves at the gates of the switches S1 and S2 of 2 ;

4 one for the control of the switches of two strands, cf. 2 , suitable D flip-flop;

5 in a schematic representation of the time course of the currents in the transformer of the flyback converter of 2 ;

6 the time course of the currents in the first and second strand of the flyback converter of 2 ;

7 the time course of the power in the first and second train of the flyback converter of 2 ;

8th the time course of the magnetic flux related to the primary current in the converter inductance of the flyback converter of 2 ( 8a ) as well as the states of the converter switch and the strand switch ( 8b ) at DCM with t _off = const. and t _on = const .;

9 the time course of the magnetic flux related to the primary current in the converter inductance of the flyback converter of 2 ( 9a ) as well as the states of the converter switch and the strand switch ( 9b ) in mixed mode BCM-DCM with t _off = const. and t _on = const .;

10 the time course of the magnetic flux related to the primary current in the converter inductance of the flyback converter of 2 ( 10a ) as well as the states of the converter switch and the strand switch ( 10b ) at BCM with t _off not const. and t _on = const .;

11 the time course of the magnetic flux related to the primary current in the converter inductance of the flyback converter of 2 ( 11a ) as well as the states of the converter switch and the strand switch ( 11b ) in mixed operation CCM-DCM with t _off = const. and t _on = const .;

12 the time course of the magnetic flux related to the primary current in the converter inductance of the flyback converter of 2 ( 12a ) as well as the states of the converter switch and the strand switch ( 12b ) in mixed operation CCM-BCM with t _off = const. and t _on = const .;

13 the time course of the magnetic flux related to the primary current in the converter inductance of the flyback converter of 2 ( 13a ) as well as the states of the converter switch and the strand switch ( 13b ) at CCM with t _off = const. and t _on = const .;

14 a representation of the different Symmetriereigenschaften the example of the flyback converter of 2 as well as the inverted converter and the SEPIC different control of the converter switch and the strand switch;

15 a representation of the different symmetry properties similar 14 the example of an up-converter with different control of the converter switch and the strand switch;

16 a schematic representation of the structure of a generic inverting converter with two output strands;

17 a schematic representation of the structure of a generic boost converter with two output strands;

18 a schematic representation of the structure of a generic SEPICs with two output strands;

19 a schematic representation of the structure of a generic SEPICs with galvanic isolation of the two output strands.

In the different embodiments, the same reference numerals are used for the same and the same effect elements. These are introduced only once for the sake of clarity.

1 shows the basic circuit of a strand current symmetrization of LEDs using identical resistors R1 = R2 = R3 on the secondary side of a flyback converter. The first strand St1 comprises the LEDs D10 to D17, the second strand St2 comprises the LEDs D20 to D27 and the third strand St3 comprises the LEDs D30 to D37. The parallel-connected strands St1, St2 and St3, a capacitor C1 is connected in parallel. On the secondary side, a current flow I2 is generated by the primary side.

The resistors R1, R2 and R3 undesirably generate power loss.

To be at the in 1 circuit shown instead of the _phase currents I _St1 , I _St2 , I _St3 , which can be measured on the basis of the voltage drop across the resistors R1 to R3 to symmetrize the _{strand power} P _St1 , P _St2 , P _St3 , the use of a control structure would be necessary, which the individual _{Strangspannungen} V _St1 , V _St2 , V _St3 and the respective phase _current I _St1 , I _St2 , I _St3 linear or using a switching converter, such as a buck converter controls.

Regardless of the principle of the respective converter circuit used, power balancing can in principle be implemented in any switching converter or linearly controlled concept. In flyback converters, especially in flyback converters, this can be realized particularly easily.

In the following statements, the idea underlying the present invention will be described using the example of an embodiment of the present invention 2 illustrated flyback converter with two secondary-side LED strands St1, St2 explained. As will be apparent to one skilled in the art, the number of strands in practice may be arbitrarily extended. In addition to LEDs, the principle shown below can of course be applied to any other, real energy consuming loads. In addition to the series connection of loads, in particular LEDs, structures with parallel loads are also possible.

On the primary side, the flyback converter includes 2 the series connection of a primary inductance L1, a converter switch S0 and an ohmic resistance R0, which is connected between a voltage applied between input terminals E1 and E2 DC supply voltage V _in , for example, 400V, and a reference potential. The primary-side current is denoted by I1, the gate terminal of the converter switch S0 by Gate0.

The voltage drop across resistor R0 is applied to the sense input of a driver 12 coupled. On the secondary side, a secondary inductance L2 coupled to the primary inductance L1 is shown, which is coupled to the parallel connection of a first strand St1 and a second strand St2. The first string St1 comprises the series connection of a diode D1, a plurality of LEDs D10 to D15 and an armature switch S1 whose gate terminal is designated Gate1. The LEDs have a capacitor C1 connected in parallel. The second string St2 comprises the series connection of a diode D2, a plurality of LEDs D20 to D24 and a string switch S2 whose gate terminal is designated Gate2. The LEDs a capacitor C2 is connected in parallel.

It is assumed that the control bandwidth of the Gate0 drive is relatively low compared to the switching frequency, which may be 270 kHz, for example. That is, that of a respective PWM controller of the drive device 12 set duty cycle is changed only relatively slowly. Particularly with controllers that have been specially developed for the LED sector, a particularly slow regulation is often used. The bandwidth can be ≤500 Hz.

To simulate a strong asymmetry, the second strand St2 of 2 one less LED than the first strand St1. This results in the example, a difference of the phase _voltages V _St1 , V _St2 of 3.3 V. In practice, however, the asymmetry percentage is usually not so strong. The total secondary output current I2 in the example is 2.5 A.

In operation, the string switches S1 and S2 are alternately turned on by respective driving of gate1 and gate2 at each primary switching cycle of the flyback converter, compare to this 3 , In the example, the switching process occurs while the primary circuit of the flyback converter is active. Thus, secondary load-free switched, which also overlaps in the switching behavior of the string switches S1, S2 are completely unproblematic.

This results in a division of the primary switching frequency on the secondary side, i. only every second switching pulse goes through the same string. The energy transferred in the flyback converter per switching cycle is thereby alternately conducted to strand St1 or strand St2. Consequently, when using more than two strings, all strings must also be cyclically alternately switched on one after the other on the secondary side, resulting in a division of the primary switching frequency by the number of load strings from the perspective of a load string.

In order to prevent the energy of a strand from flowing out during the non-active periods and during the activity of the primary side, one diode must be connected in front of each strand, see the diodes D1 and D2 in FIG 2 , Here, if possible, low-loss Schottky diodes should be used due to their low forward voltage. In other words, these are the output diodes that are present anyway in the switching converter topology used, just one per string.

In order to keep the high-frequency current ripple (frequency = primary switching frequency / number of load strings) low on the LEDs, capacitors C1, C2 are required in parallel. The required value of these capacitors C1, C2 increases with the number of load strings, since the time to be bridged increases until the respective string is switched on again.

In practice, the control of Gate1 and Gate2 can be done for example via a digital ring counter module, which is triggered by the conceptual ripple on the secondary side. The triggering can also be effected by the primary-side gate signal Gate0, optionally also galvanically isolated via an optocoupler. The use of a microcontroller or other digital or analog circuit instead of a ring counter is conceivable.

Particularly in the case of only two strings St1, St2, the control of the string switches S1, S2 can be carried out particularly easily with a D flip-flop, see 4 , The triggering takes place again via the primary-side gate signal Gate0 or the secondary-side ripple. The D flip flop 10 acts as a simple frequency divider and provides the Gate2 signal at output Q and at the output Q the gate1 signal ready.

Since the control bandwidth of the Gate0 control is relatively low compared to the switching frequency, see above, even with changing load, ie a switching of the strands on the secondary side, always the same amount of energy transmitted. The duty cycle changes only very sluggishly, its value does not change due to load jumps with a frequency that corresponds to the primary sound frequency divided by the number of strands, whereby an average value is set in the long term in such a changing load. 5 shows this property based on the secondary-side current I2.

In the presentation of 6 Note that the time axis is not the zero line for the streams. Now consciously different _phase currents I _St1 , I _{St2 arise} . The strand with the lower forward voltage, string St2, now continuously receives more current than that with the higher forward voltage, string St1.

However, the powers P _St1 , P _St2 are almost identical, see here 7 , Now, in each of the principle arbitrary many strands the same performance is implemented. This leads to a thermal balance of all strands.

Would the two in the example of 2 shown strands St1, St2 are connected in parallel directly, ie without strand switch S1, S2 or without resistors R1, R2 or the like, so a very strong current unbalance would show up to a point at which no current flows in a strand at all. With the two strands considered here, whereby in one of them one LED is missing, with direct parallel connection in the other no current would flow. The extra LED in this other strand acted as a non-conductive electronic switch. In the present concept, therefore, the current is also - albeit not exactly - balanced. If the control slightly modified, as in 11 As can be seen, said conscious difference between the strand currents can be continuously reduced. In a drive according to 12 or 13 occurs complete current symmetry.

This is accomplished by always turning on only one strand, ensuring that each strand alternately flows. With multiple strings connected in parallel, the current would flow substantially through the strand having the lower resistance and the lowest forward voltage, respectively.

The current balancing factor k is defined as follows:

From the above-explained goal P _St1 = P _St2 , the _following results: V _St1 * I _St1 = V _St2 * I _St2 .

It follows:

The relative difference of the _phase currents I _St1 , I _St2 thus depends only on the ratio of the respective phase _voltages V _St1 , V _St2 to each other.

For practical applications in which the forward voltages of the parallelized strings differ only slightly, for example for k> = 0.95, the power balancing shown here can thus also show acceptable results in terms of current symmetry.

All properties shown so far apply, apart from possible deviations in the system dynamics, as well as for all SEPICs, see 18 and 19 , as well as for all inverting converters, see 16 , The effects of operation on the symmetry properties shown below apply equally to all SEPICs and inverting converters.

Achieving the same percent symmetry (factor k) with simple resistors or linear control solutions, such as current mirrors, would require a relatively high amount of power dissipation per resistor, typically the same amount of power consumed by a string. A small calculation for two LED strings whose LED performance is only to be fully balanced by two series resistors of the same size (it is not known at first which string has the greater forward voltage) shows that the series resistors would have to be chosen with such a high resistance, that across the first resistor, the forward voltage of the second LED strand drops and above the second resistor, the forward voltage of the first LED strand. The power loss would be at least as large as the net power. By employing the invention described heretofore and hereinafter, the efficiency of the overall arrangement can be approximately doubled.

The per primary-side switch-on time interval Ini, see for example 8a , in the converter topology _cached energy W _inx corresponds to the current-time integral defined for this section multiplied by the input voltage V _in , where I _ai corresponds to the current value of the primary-side current I at the beginning of this interval and I _{bi is} at its end (Ini the phase in the period Ti, in which the converter switch S0 is turned on to magnetize the Wandlerinduktivität):

The current increase in between corresponds to V _in / L1 and is assumed to be linear. Finally, as expected, the arithmetic mean of the starting and final currents multiplied by the input voltage V _in and the on-time interval ton: W _inx = V _in (I _ax + I _bx ) ton / 2.

The analogous consideration leads, for example, to the following diode conduction time interval in strand 1: W _St1 = V _St1 (I _b1 + I _a2 ) toff1 / 2.

In this case, W _{St1 denotes} the energy _package which is delivered to it during the partial phase assigned to the first strand.

If I _ax > 0, there is CCM (continuous conduction mode), the general case. The special case I _ax = 0 is called DCM (discontinuous conduction mode), the borderline case in which the current per period is zero only at a point at which the primary-side switch is turned on again, with BCM (boundary conduction mode ) designated.

For both DCM and BCM: W _inx = V _in I _bx ton / 2.

In the 8th to 10 In each case, the magnetization of the converter inductance relative to the input current I is shown in the three modes of operation of the topologies considered so far, in which power balancing between the two illustrated strings St1, St2 applies. This becomes clear because, in each sub-phase T1, T2 ... Tx, see 8a , the current becomes zero at least at one point and because W _inx = W _St1 = W _St2 = W _Stx . DCM power balancing works with theoretically any number of output channels.

The converter must be designed for the occurring peak values. The actual output per string is calculated by: P _Stx = W _Stx / T = W _inx / T With
T = T ₁ + T ₂ + ... + T _x .

It follows that the total period T and thus the current zero pauses in the representation of 8a must be selected as short as possible in order to transmit the maximum power at a given transducer dimensioning.

8a shows the time course of the magnetization of the transformer inductance related to the primary current during the partial phases for the first and the second strand at DCM with toff = const. and ton = const. In1 and Out1 designate the magnetization phases during the partial phase T1 for the first strand St1, In1 being initiated by the conduction of the converter switch S0 and the converter inductance being magnetized. Out1 designates the demagnetization of the transducer inductance and the transfer of the corresponding energy into the capacitor C1 after the switch S0 is not turned on until the transducer inductance is completely demagnetized. In this respect, the period t _{off1 is} dependent on the _strand voltage V _St1 , because only this voltage determines the speed of demagnetization and thus the gradient of the graph. The same applies with respect to the quantities In2, Out2, t _off2 for the strand St2. Accordingly, In1 results from turning off the converter switch S0, while Out1 represents the current resulting from the magnetization of the converter inductance, which current flows through the strand St1 due to the line switch S1 which is turned on in this phase.

8b shows the corresponding switching states of the converter switch S0 and the string switch S1, S2. Throughout each _toffi section and in its full length, there are solid lines in the S1 and S2 diagrams, each in opposite _directions : In the areas where t _{off1 is} drawn, S1 = "on", corresponding to t _off2 then S2 = "on". During the t _on sections, it is insignificant in which switching state the switches S1 and S2 are due to the existing in each strand flow diodes D1, D2. Therefore, in these sections in the diagrams of S1 and S2, all lines are dashed. The exact switching time between S1 and S2 does not matter, not even its existence: in the "dashed sections" even several string switches can be switched on simultaneously or all switched off. The same applies to those sections by which t _{off is} longer than t _off1 or t _off2 , ie for those sections in which the converter switch S0 is indeed switched off, but more current is fed into none of the strings due to the complete demagnetization of the converter inductance. The latter sections are also commonly called power gaps.

However, it is advisable to switch exactly complementary, as already in connection with the control with the aid of a D flip-flop, see 4 , executed. The exact switching time between S1 and S2 or vice versa is irrelevant as long as the switching takes place during a t _on section. During each t _on section S1 and S2 have a dashed staircase (counter-clockwise) with a vertical step in the middle. This is shown in dashed lines, as already explained above, the exact switching point does not matter. Underneath there are, as indicated, light gray areas bounded by dashed lines. As in 8b can be seen, any such gray area may even be extended to the left before the beginning of a t _oni section: This indicates that it does not matter what the switch Si does when t _{offi has} expired, but t _off yet. Not permitted are switching states that are not shown. Everything is possible, which is gray or black. Only sharply defined switching pattern exists for the transformer switch S0. For the switch-on sections of the string switches Si, the lines raised by the associated t-axes are recommended, as they result from the flip-flop control already proposed in connection with Fig. 4 with a possible small trigger delay. This results in two strands St1, St2 for the switches S1, S2 an opposite drive with the same length switching states with exactly half the frequency of the converter switch S0.

Without the constancy of frequency and duty cycle, so t _on = const. and t _off = const., give up, results in a particularly advantageous constellation in the control according to 9 : "T _off " is shortened to such an extent that after the period in which the strand with the lowest flux voltage was active (indicated by the flattest drop of I (t)), BCM occurs.

T can be further shortened for any number of strands without loss of power symmetry if BCM is selected after all strands, as in 10 shown. t _on remains constant, but t _off becomes variable, which requires a different control strategy in the window of a total period T: A higher and _lower regulator controls t _{on the} basis of the current demand, t _{offi reconnects} each time by detecting the current Zero events. The frequency gidder (T1 ≠ T2), which adjusts itself, facilitates the radio interference suppression. The newly emerging subharmonic f / n = 1 / T should be above 20 kHz and below 150 kHz in normal applications.

Starting from the illustration in 9 is further shortened at constant frequency and constant duty cycle t _off , so that a mixed operation of DCM and CCM arises, as determined by the course of the converter magnetization in 11a is shown. The CCM, which is already in 9 BCM suggests that occurs always after the discharge phase of the strand with the lowest forward voltage, ie the strand with the slowest current drop. Since at each switching between output diode Di and converter switch S0 in the CCM increased switching losses occur because S0 at quasi full blocking voltage first carry the reverse recovery current of the diode Di and only then has to short-circuit its own storage charge before he finally turned on, takes the Efficiency of the overall arrangement when entering the CCM slightly from, but this can be compensated by the further increased converter utilization. The negative effects of hard switching on electromagnetic compatibility are also known. Will t _off so far shortened that the DCM off 11 arrives at the BCM, results in the 12 illustrated situation. This case is the lower limit case of what is in 13 is shown as a general CCM for two output strands: Immediately it is noticeable that now, despite t _on = const. and despite the same current increases due to the different starting currents I _a1 and I _a2 in the sub-cycles T1 and T2 different energies E1 and E2 are stored in the converter topology.

It follows that power balancing can no longer exist. As before - as in the 8th . 9 . 11 . 12 . 13 - applies next to t _on = const. also t _off1 = t _off2 = t _off = const., ie f = const. and duty cycle D = const. What is new is that now, over the entire period, T must always be balanced.

For the period "In1" the following applies: W _in1 = V _in (I _a1 + I _b1 ) t _on / 2; t _on V _in = (I _{b1 -I a1} ) L ₁ ; both inserted into each other gives: W _in1 = L1 ( _I2 / b1 -I2 / a1) / 2; analogously results for Out1: W _St1 = L1 (I 2 / _{b 1} - I 2 / a 2) / 2 and for the other two time periods "In2" and "Out2": W _in2 = L1 ( _I2 / b2 -I2 / a2) / 2, W _St2 = L1 (I 2 / _b 2 - I 2 / a 1) / 2.

For the two output voltages, due to their demagnetizing effect described above: V _St1 = L1 (I _{b1 -I a2} ) / toff, V _St2 = L1 (I _b2 - I _a1 ) / toff.

For the two output powers work per time applies, thus: P _St1 = W _St1 / T = L1 (I 2 / _{b 1 -I} 2 / a 1) / 2T = L1 (I _{b1 -I a2} ) (I _b1 + I _a2 ) / 2T P _St2 = W _St2 / T = L1 (I 2 / b2 - I 2 / a1) / 2T = L1 (I _b2 - _a1 I) (I + I _{b2 a1)} / 2T.

The above-mentioned "third binomial formula" greatly facilitates the following, because power through voltage gives the current: I _St1 = P _St1 / V _St1 = (I _b1 + I _a2 ) toff / 2T, I _St2 = P _St2 / V _St2 = (I _b2 + I _a1 ) toff / 2T.

The equation seen from the entrance I _b1 - I _a1 = I _b2 - I _a2 can also be represented as follows: I _b1 + I _a2 = I _b2 + I _a1 .

This is used in both of the above equations for I _St1 and I _St2 , giving: I _St1 = I _St2 .

This proves that with two alternately switched output strands in a SEPIC, flyback or inverting transducer, at f = const., At D = const. and in general CCM, there is an inherent current balance between the two strings, no matter how different their fluence voltages are.

12 turns out 13 if I _a2 = 0, and thus is the lower limit of 13 , This also applies to the situation 12 the just proven inherent strand current symmetry.

Are the current forms of the considered converter from the 8th . 9 . 11 and 12 by frequency adjustment with constant on-time t _on - if at least one "Stom-Null-Event" is ensured per total period T even without current monitoring - that changes Rule concept in general CCM of 13 The primary current I describing the magnetization of the converter inductance must be monitored and fed back into the control loop as a further variable in order to reliably avoid saturation of the converter inductance. In the CCM, it is also possible to counteract by means of duty-cycle adjustment instead of by frequency adjustment.

Furthermore, it can be 13 It is also easy to see why this inherent current symmetry in the general CCM works so well for two output _channels : The _phase currents I _St1 and I _St2 are determined from the mean value of the current summits belonging to the phases "Out1" and "Out2", as is well _understood by their formulas , Because the current summits of the phases "In1" and "In2" are always equally far apart, and because the circle is already closed again by "Out2", the corner points I _b1 and I _a2 are exactly covered by the next corner points I _b2 and I _a1 why the average values defining the output currents are also the same.

It does not need to be described in more detail that for 11 a balance between power balancing and current balancing applies, depending on how long the remaining DCM current gap, ie the time difference between t _off and t _off1 , is. The strand with the smaller current, so with the higher forward voltage gets the greater power. This relationship emerges 14 , which represents the symmetry properties of the flyback converter, the inverse converter and the SEPIC.

If the durations of t _off between the individual strings are varied, resulting in a CCM with a constant t _on but variable frequency, the ratio of the individual phase currents can be controlled to one another and not equal to one. The same effect is achieved by an additional variation of t _on between the individual partial phases Ti, whereby a constant frequency of the total converter can result, but the duty cycle changes in the partial phases. Finally, if only t _{on is} varied and t _off kept constant, the phase currents can also be consciously controlled "askew" to one another.

15 shows the symmetry properties for a boost converter. The in 17 shown up converter or boost converter or boost converter can be achieved by connecting several output diodes D1 and D2 in parallel, and by adding a series switch S1, S2 per strand as well as the converter, which in the 2 . 16 such as 18 and 19 are shown, operate with multiple output strands.

All these converters have in common that they have a so-called simple voltage smoothing at their output. With regard to the symmetry properties, however, the boost converter deviates in part from what has been described hitherto. 13 can also describe the boost coil current in a boost converter, which is now assumed here. The formulas for the output energies are the same as above: W _St1 = U _St1 (I _b1 + I _a2 ) toff / 2, W _St2 = U _St2 (I _b2 + I _a1 ) toff / 2.

From the rule "electricity = work by voltage through time" results: I _St1 = (I _b1 + I _a2 ) toff / 2T, I _St2 = (I _b2 + I _a1 ) toff / 2T.

Since the current increase is the same during the up-converter switch during t _on , the same set of equations apply here as well: I _b1 - I _a1 = I _b2 - I _a2 ; I _b1 - I _a2 = I _b2 - I _a1 ; I ₁ = I ₂ .

Thus, the up-converter also generates inherent output current balancing with exactly two parallel, alternately active output lines in general CCM and the other conditions known from above. Conversely, all possibilities and restrictions from above apply here as well. Since pure boost converters never have isolated transducer inductances, they can be operated more easily than the previously described topologies, in particular their isolated forms, in the CCM. Indeed, the leakage inductance of each isolated transducer inductor significantly amplifies the above-described difficulties in CCM.

For a boost converter with multiple outputs in general DCM, as already stated 8th read, I _ax = 0 and I _b1 = I _b2 = I _b .

In contrast to the first considered up and down converters, such as the flyback converter, the input voltage source V is deprived of energy _{in the} boost converter even if the converter switch S0 is already turned off, but still flows, namely over D _x during t _offx . The equation valid for the flyback converter E _x-flyback = U _in I _bx ton / 2 must be modified. For the up-converter, the following applies: E _x-boost = U _in I _bx (ton + toffx) / 2.

Since only slightly different output voltages and the D _x -Leitzeitertauern t _{offx are} similar, an approximate power _balancing between the strands can be assumed for the up-converter with multiple parallel and sequentially active output strands in the general DCM.

On closer examination, however, the more energy is stored in the converter topology per subphase T _x , the longer t _offx lasts, that is, the slower the current drops or the magnetic flux in the boost coil demagnetizes. Longer t _offx correlates with smaller output voltage V _Stx , as with the flyback converter, so that for the up-converter at DCM for two or more output lines a kind of power _{overcompensation} applies:
The strand with the lowest flux voltage is supplied with the most power.

In 14 is shown the current or power curve for two output lines in the flyback converter, while in 15 the same is shown in the boost converter, depending on the mode of operation.

Out 15 It will be seen that also the boost converter has a certain point at which power balancing occurs, viz. the point Y in that transition region whose associated converter current is in 11 is shown.

When operating in CCM and t _off = const. results in a circuit that is inherently insensitive to short circuit in a strand, provided that a primary-side peak current monitoring intervenes and the primary-side switch-on after a freewheeling phase (in which degenerate the depletion phase at shorted strand degenerate) accordingly shortened.

Regardless of the number of strands and operation, primary-side voltage _monitoring during t _offx , which is related to the highest possible value of all the line _voltages , can detect all lines damaged by interruption and avoid destruction of the converter by connecting the next line switch. The "position" of the open string can be saved, the setpoint of the primary current can be reduced accordingly, the corresponding string switch can be skipped during the next circulation.

At the in 19 shown isolated SEPIC and the flyback converter shown at the beginning 2 Furthermore, it is possible to provide a separate secondary winding on the isolated transformer inductance or the power transformer per output line. Then, the reference voltages or sources of the switches S0, S1, S2, Sx can be at different potentials, provided that their gate drives are designed to be electrically isolated. Furthermore, results in the circuits after 2 and 19 the possibility of not galvanically isolating all the output strings from the input, if this is not required for all strings, or if for some reason a fixed potential relation between the input and some of the outputs is advantageous.

Finally, by deliberately different numbers of turns per secondary strand or by intermediate taps of the secondary winding in still valid power balancing (DCM at flyback, SEPIC and buck-boost) set significantly more divergent strand currents, if so require the connected LEDs, if, for example, the white LEDs require the threefold current like a few colored ones for color locus correction.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• WO 2012/164511 A1 [0007, 0010, 0020, 0020]
• US 2010/0295472 A1 [0008, 0011]
• DE 102010002228 A1 [0010]
• DE 102010041613 A1 [0010]
• DE 102010041632 A1 [0010]
• EP 1788850 A1 [0010]
• DE 102010041618 A1 [0010]

Claims (15)

Circuit arrangement for operating n loads, n greater than or equal to 2, comprising: - an input having a first (E1) and a second input terminal (E2) for coupling to a supply voltage (V _in ); - n outputs each having a first (A11; A21) and a second output terminal (A21; A22) for coupling to the respective ith load, i = 1 to n; A converter with a converter switch (S ₀ ), at least one converter inductor (L1) and n converter diodes (D1, D2), wherein a series connection of the converter inductance (L1) and the converter switch (S ₀ ) between the first and the second input terminal ( E1, E2) is coupled; - n strands comprising at least one series circuit each having an ith electronic switch (S _i ) with i = 1 to n, an i-th of the n converter diodes and the respective i-th output with i = 1 to n, wherein the n series circuits are connected in parallel and are coupled to the Wandlerinduktivität (L1) that stored in the Wandlerinduktivität (L1) energy in the discharge phase of the converter by Leitend switching the respective i-th electronic switch (S _i ) at the ith output is available , as well as - a driving device ( 12 ) for driving the converter switch (S ₀ ) and the n electronic switch (S _i ), wherein the drive device ( 12 ) is designed: - to drive the converter switch (S ₀ ) with a drive signal (Gate0), which is composed of n partial phases (T _i ) assigned to the n strings, and - the electronic switches (S _i ) with a respective drive signal (Gate1 , Gate2) to control; wherein a Conduction Mode of the circuit arrangement is defined with respect to the current flow through the Wandlerinduktivität (L1), wherein a Leitendphase the converter switch (S ₀ ) is framed by the Leitendphasen two different of the n converter diodes, characterized in that the drive device ( 12 ) is arranged to provide the drive signals (Gate0, Gate1, Gate2) for the converter switch (S ₀ ) and the switches (S _i ) such that at least one of the n strands within the sub-phase (T _i ) assigned to it in the continuous conduction mode ( CCM) or in the Boundary Conduction Mode (BCM) is operated, and the switches (S _i ) outside of their assigned sub-phase (T _i ) are switched non-conducting. Circuit arrangement according to Claim 1, characterized in that the drive device ( 12 ) Is adapted to the drive signals (Gate0; Gate1, Gate2) (for the switches S0, S _i) providing such a way that - at least one of the n strands within its allocated sub-phase (T _i) in the discontinuous conduction mode (DCM), and at least one of the n strands within the sub-phase assigned to it is operated in the BCM or in the CCM; or - all n strands within the sub-phases assigned to them are operated in the CCM or at most one will operate within the sub-phase assigned to it in the BCM; or - at least one of the n strands within the sub-phase (T _i ) assigned to it in the DCM and at least one of the n strands within the sub-phase (T _i ) assigned to it is operated in the CCM. Circuit arrangement according to one of claims 1 or 2, characterized in that each partial phase (T _i ) is composed of an on-time (t _on ), in which the converter switch (S0) is turned on and leads to the magnetization of the Wandlerinduktivität (L1) , and an off-time (t _offi ), in which the transducer switch (S0) non-conductive, but the respective i-th electronic switch (Si) is turned on, and with either the demagnetization of the Wandlerinduktivität (L1) or with the renewed Leitendschalten the converter switch (S0) ends. Circuit arrangement according to Claim 3, characterized in that the drive device ( 12 Is designed), the drive signals (Gate0; Gate1, Gate2) (for all involved switch S0, provide S _i) such that the on-time (t _on) are constant and the off-times (t _offi) is variable. Circuit arrangement according to one of claims 3 or 4, characterized in that the drive device ( 12 Is designed), the drive signals (Gate0; Gate1, Gate2) (for all involved switch S0, provide S _i) such that the on-time (t _on) and the off-periods (t _off) are constant. Circuit arrangement according to one of claims 3 to 5, characterized in that the drive device ( 12 Is designed), the drive signals (Gate0; Gate1, Gate2) (for all involved switch S0; S _i) to provide in such a way that the sums of the respective on-times (t _oni) and off-times (t _offi) for i = 1 to n are constant. Circuit arrangement according to one of the preceding claims, characterized in that the drive device ( 12 ) is designed to switch two electronic switches (Si, Sj) in opposite directions during the Leitendphase the converter switch (S0). Circuit arrangement according to one of the preceding claims, characterized in that the Wandlerinduktivität (L1) is implemented as a transformer having at least one primary winding (L1) and a secondary winding (L2). Circuit arrangement according to one of the preceding claims, characterized in that an i-th capacitor (C1; C2) is connected in parallel with the respective i-th output (A11, A12; A21, A22). Circuit arrangement according to one of the preceding claims, characterized in that a shunt resistor (R0) is coupled in series with the converter switch (S0). Circuit arrangement according to one of the preceding claims, characterized in that the converter is designed as a flyback converter, as an up-converter, as a SEPIC or as a choke-inverted converter. Circuit arrangement according to one of the preceding claims, characterized in that the i-th load comprises at least one LED (D10 to D15, D20 to D24), in particular a series circuit of a plurality of LEDs (D10 to D15, D20 to D24) represents. Circuit arrangement according to one of the preceding claims, characterized in that n = 2. Circuit arrangement according to claim 12, characterized in that the drive device ( 12 ) a D flip flop ( 10 ) whose input is coupled to the input of the converter switch (S0), the Q output of the D flip-flop ( 10 ) with a first electronic switch (S1) and the Q Output of the D flip-flop ( 10 ) is coupled to a second electronic switch (S2), wherein the D input of the D flip-flop ( 10 ) with the Q Output is coupled. Circuit arrangement according to one of the preceding claims, characterized in that the drive device ( 12 ) is designed to vary the partial phases (T _i ), in particular if one of the strands has exceeded or fallen below a predefinable temperature threshold.

Images