Two-stage operating device with an isolatable clocked converter as a power factor corrector and control method for the operating device
Patent Information
- Application Number
- EP2022809799
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-01-17
- Filing Date
- 2022-10-28
- Publication Date
- 2024-11-27
AI Technical Summary
Conventional power factor correctors and post-regulators in lighting technology face challenges with high power losses, inefficient regulation of output current, and the need for oversized intermediate circuit capacitors due to mains ripple and temperature fluctuations, which affect the service life and efficiency of lighting systems.
A two-stage control gear with an isolable switched converter as a power factor corrector and a step-down post-regulator, utilizing optocouplers for digital communication across the isolation barrier, and a control circuit that dynamically adjusts the margin voltage to minimize losses and regulate the output current, while maintaining a stable output voltage for LED lighting.
This solution reduces power losses, enhances the quality of the output current, extends the service life of intermediate circuit capacitors, and provides efficient and stable lighting with minimal light modulation, even under high temperatures.
Smart Images

Figure 1.1
Abstract
Description
[0001] TWO-STAGE OPERATING DEVICE WITH ISOLABLE SWITCHED
[0002] CONVERTER AS POWER FACTOR CORRECTOR AND CONTROL METHOD FOR THE OPERATING DEVICE
[0003] DESCRIPTION
[0004] Technical area
[0005] The invention relates to a circuit arrangement for operating a load, which requires only a margin voltage of a post-regulator as a feedback signal to adjust a load current. An operating device comprising the circuit arrangement, which is intended for supplying power to and regulating light-emitting diodes for general lighting technology, comprises an isolable first clocked converter as a power factor corrector and the post-regulator connected to and powered by it, which, as its load, in turn supplies the at least one light-emitting diode with the necessary power. To enable its operation, the post-regulator has a margin between an intermediate circuit voltage and an output voltage, which margin can also include mains hum.The output voltage corresponds to the forward voltage of at least one LED connected to the control gear, or the sum of the forward voltages of all connected LEDs if, as is usually the case, they are connected in series. The intermediate circuit voltage is the output voltage of the power factor corrector, which is usually higher than the forward voltage and can be the same during certain times or periods. At these times, the margin is eliminated and the post-regulator is inactive. The main task of the post-regulator is to compensate for mains hum. Therefore, it includes an analog current regulator located on the output side of the control gear, whose setpoint is transmitted in pulsed form via a first optocoupler.A second optocoupler transmits binary information indicating whether the margin is greater or smaller than a predefined limit from the output side back to the input side of the operating device, which houses a control circuit. This not only provides punctual feedback of the output voltage, since the predefined limit and the intermediate circuit voltage are known, but also, through the presence of the margin, indicates the activity of the post-regulator itself. In the event of excess temperature, the periods with existing margin are shortened, causing the post-regulator to remain inactive for longer and, of course, reducing the output current to protect the operating device. Further protection for the operating device is provided by reducing losses in the post-regulator. The post-regulator can be designed as a linear regulator or as a second clocked converter, particularly advantageously as a buck converter.Particularly good cooling of the output side can shift an overtemperature threshold to higher temperatures.
[0006] background
[0007] Switched-mode converters are widely used as power factor correctors, which are mandatory in lighting technology starting with a single input power of 5 W and, for example, in entertainment and computer technology starting with a single input power of 80 W. The output terminals of such a power factor corrector, which can also be referred to as an input stage, are in most cases connected by a large intermediate circuit capacitor, in which a DC supply voltage is buffered for subsequent switching stages or directly for a load, also known as the intermediate circuit voltage. Buffering is essential because, by design, the power factor corrector, as the input stage, only generates recharging currents for the intermediate circuit capacitor that are fully modulated and pulsate at twice the frequency of an AC mains supply, i.e., even have zero points at this double mains frequency.The result is a so-called mains hum on the DC supply voltage buffered in the DC link capacitor at twice the mains frequency. In the case of a mains frequency of 50 Hz, this hum occurs at 100 Hz, unless the DC link capacitor has virtually infinite capacitance, such as "supercaps" or rechargeable batteries. Electrolytic capacitors or large film capacitors of finite capacitance, commonly used as DC link capacitors, are therefore confronted with the above-mentioned mains hum, which is superimposed on the DC supply voltage or DC link voltage, and for which they must be designed. The lower the capacitance of a DC link capacitor compared to the current drawn, the higher the amplitude of the mains hum.In a higher-level device considered here, a first clocked converter is always the input stage, which is terminated by the intermediate circuit capacitor and which produces and maintains an intermediate circuit voltage there.
[0008] Power factor correctors as input stages often have the property of generating intermediate circuit voltages that are higher than the voltage peaks of the associated AC mains supply. In the case of a standard 230 V mains voltage, intermediate circuit voltages or DC supply voltages of at least 400 V often result. These voltages are too high for many loads, especially those with low power. Therefore, switch-mode converters are increasingly being used as power factor correcting input stages, which can also produce and maintain smaller intermediate circuit voltages. Four such topologies for switch-mode converters are known from the literature, each comprising only one active power transistor and one fast power diode: the choke converter, the 'Cuk converter, the Zeta converter, and the SEPIC (Single-Ended Primary Inductor Converter).The first two, at least in their original non-insulating form, have the problem that their respective output voltage, i.e., the intermediate circuit voltage, is negative relative to a circuit ground that runs between the input and output. Alternatively, a choke converter can be designed such that its output voltage is positive, but the output ground is on the positive pole of its input voltage, so that the output voltage figuratively "rides on the input voltage." In short, with the choke converter and the 'Cuk' converter, either the output voltage is negative or the ground "doesn't pass through."
[0009] This problem is solved in the other two topologies above. The SEPIC was therefore developed from a 'Cuk' converter to combine its advantages—an input inductance, which is particularly advantageous for power factor correctors, and the freely selectable output voltage—with a continuous circuit ground and a positive output voltage. Since every SEPIC requires capacitive termination, which is already provided by the DC link capacitor, this converter topology is particularly well-suited as a voltage-level-free power factor corrector. The choke converter can only be correctly polarized on the output side by replacing its inverting choke with a counter-coupled transformer, thus giving rise to the well-known flyback converter.Although it does not include an input inductance, this converter is also suitable as a voltage-level-free power factor corrector and, due to its minimal number of components and its equally good suitability for recharging a DC link capacitor, is also the most commonly used power factor corrector in this function or as a first switch-mode converter. Due to its output inductance, a 'Cuk' converter as a power factor correcting input stage is particularly well-suited for loads that require a current link instead of a voltage link, such as discharge lamps, welding equipment, or electric drives, and is therefore not suitable as an input stage. A zeta converter has the same limitation and, in addition, has no inductance at its input, which is why it is not even suitable as an input stage, power factor corrector, or first switch-mode converter.Because it is just as complicated as a SEPIC or a 'Cuk converter and therefore, without any advantages, only combines their disadvantages.
[0010] Regardless of the voltage or current intermediate circuit and the polarity of the output voltage, the four converter topologies above for voltage-free outputs have the further immense advantage of being galvanically isolated by an isolation transformer between their input and output. The flyback converter, which, as mentioned above, is created by isolating a choke converter, illustrates the principle: The inductance with a ground connection in the original converter topology is replaced by the transformer, breaking the ground connection. In a SEPIC, its storage inductance connected to ground on the output side is replaced by the isolation transformer; in a zeta converter, the same would happen with its input storage inductance.A 'Cuk' converter can only be isolated by splitting its internal capacitor and inserting a transformer between them. While this is complicated, it offers the enormous advantage that the isolation transformer is only loaded with pure alternating current on both sides. It therefore requires no air gap. All of this has recently led to a significant shift in tasks between the input stage and a post-regulator within a higher-level device, particularly when a galvanic isolation barrier must extend through the entire higher-level device. While this barrier was previously primarily the responsibility of the post-regulator, which then often had to be designed as a complex full- or half-bridge or zeta converter, it is now increasingly the responsibility of the power factor correcting input stage or the first switch-mode converter.
[0011] In addition to the power required for an electrical load, which is handled by the topology of the first switch-mode converter, various measurement signals must also be transmitted as analog data across the same isolation barrier. Furthermore, it is recommended to concentrate all the logic for a higher-level device on one side of this barrier. Typically, this side is the output side of a higher-level device, to which its electrical load is connected. This is because the parameters of this load must be measured continuously and in real time in order to control one of them at a time. The resulting control signals for the at least one power transistor of a switch-mode converter, which is always located on the input side, can be easily transmitted across a galvanic barrier because they are essentially pulse-shaped.However, with active power factor correction using the first switch-mode converters described above, measurements must also be taken on the input side of a higher-level device, to which the AC mains supply is connected. For this purpose, at least part of the logic is also required there, which means that isolating higher-level devices often have two logic centers: one for active power factor correction on their input side and a second for load control and monitoring on their output side. The problem with this is that logic connections often only use the current DC link voltage if the transmission of additional logic signals is to be avoided. Such a reduced logic connection proves inadequate, particularly for highly dynamic operation or for data networking of a higher-level device.The current trend is therefore to concentrate the control circuit on the input side of a higher-level device, so that this control circuit can also fully take over the control of the power factor correction, while simultaneously avoiding complex analog data transmission across the isolation barrier. With the migration of a galvanic isolation barrier into the power factor corrector, the concentrated control circuit often moves to the input side of a higher-level device.
[0012] This higher-level device comprising the specified circuit arrangement can be, among other things, a plug-in power supply, a simple power supply, a switched-mode power supply, a laboratory power supply, a DC-DC converter, a DC-DC converter, an active current source, an active voltage source, or another operating device, e.g., for light sources. Therefore, as already mentioned at the beginning, when a higher-level device is meant, we will refer only to an operating device.
[0013] Its electrical load here consists in particular of a series connection of several light-emitting diodes intended for general lighting technology.
[0014] The more precisely this electrical load is characterized, the more the operating device designed for it can be optimized. In the case described here, the load, which preferably consists of light-emitting diodes, impresses an almost ideally constant output voltage, and a post-regulator between this load and the intermediate circuit capacitor must bridge the difference between the intermediate circuit voltage and the fixed output voltage in such a way that an output current, which is also the load current through the at least one light-emitting diode, can flow through it continuously and, in particular, without gaps. The simplest "post-regulator" is an ohmic resistor connected to the intermediate circuit capacitor in series with the load. For this to happen, the intermediate circuit voltage must always be higher than the output voltage; otherwise, the load current would have gaps.Mains hum is also a component of the voltage difference mentioned above, which, as mentioned above, is also referred to as headroom. The load current with this ohmic series resistor acting as a post-regulator has at least the same relative ripple as the mains hum based on the time-averaged value of the intermediate circuit voltage, but usually a significantly higher one. This problem is solved by replacing the ohmic series resistor with a linear regulator, which keeps the output current constant regardless of the voltage difference. This regulator therefore has an almost infinitely high differential series resistance as long as it can actively regulate. For this, too, the intermediate circuit voltage must always be higher than the output voltage; this headroom must therefore never become negative.A zero or no margin leads to an interruption of the active regulation of the mains hum to maintain a constant output current through the linear regulator as a post-regulator, which then acts as a rigid and low-impedance connection from the DC link capacitor to the output of the operating device. This advantageously results in a smaller output current flowing, defined by a first-order low-pass filter formed from the LED load perspective by the capacitance of the DC link capacitor and the differential resistance of all participating LEDs, augmented by a current measuring resistor. This low-pass filter can only passively filter the sinusoidal recharging current for the DC link capacitor, which is generated thanks to power factor correction, and can no longer actively regulate it. Therefore, without margin, the ripple of the output current through at least one LED increases noticeably.With an ohmic series resistor as a “post-regulator,” this ripple would be even higher because, without any margin, it would simply no longer allow any load current to flow through at least one LED.
[0015] Both of the post-regulators mentioned so far generate high losses, which depend on the level of the required output current and the level of the headroom. It follows that this headroom, while always positive, should only be as high as is necessary. This headroom optimization or minimization, also known as "headroom control," is the subject of countless patent applications, including EP 1-372-359 B1. A mitigation of the loss problem by periodically reducing or eliminating the headroom in sections and periods, with the post-regulator temporarily interrupting its activity, allowing the output current to pass unregulated and thus smaller than desired, while generating no losses at all, is disclosed, for example, in DE 10-2014-205-469 A1 and DE 10-2014-218-422 A1.The disadvantage of the solutions there is that the ratio of time periods with margin and those without margin, or in other words the level of the load current ripple, is designed into the operating device via component values, and in particular that it or it always has a considerable level and cannot be close to zero.
[0016] If a second clocked converter is used as a post-regulator instead of a series resistor or a linear regulator, the loss problem is almost completely eliminated, and headroom control is simplified. This is because the need to minimize headroom is eliminated, and only the requirement for a consistently positive headroom remains if the second clocked converter is a buck converter. Because a boost converter makes no sense as a post-regulator, and because a buck converter is the only one of the five remaining clocked converters that behaves externally like a linear regulator, especially without headroom and thus without clocking its power transistor, the post-regulator—if clocked—is always formed by a buck converter.
[0017] Numerous studies have shown that the loss-optimizing effect of minimizing the headroom always outweighs the countervailing effect of reducing the current in the DC link capacitor by increasing its voltage and thus also any headroom. Only at the extremes of low DC link voltages, e.g., below 10 V, can it be more advantageous to slightly increase the headroom in order to slightly reduce the currents in the DC link capacitor at constant power due to the resulting overall higher DC link voltage.
[0018] Task
[0019] The invention proposes a circuit and an associated control method for an operating device comprising a first switch-mode converter as a power factor corrector and a step-down post-regulator. The device has a galvanic isolation barrier or SELV barrier extending through the power factor corrector. The device has a control circuit concentrated on its input side, which is designed to be powered from a standard AC mains supply. The data for a complete logic connection between the input and output sides is to be transmitted only via two optocouplers. The device's output side is configured for connecting a DC load, in particular a light-emitting diode or a series circuit comprising several light-emitting diodes, intended for general lighting applications.Both optocouplers are to be operated digitally, i.e. pulsed, in order to compensate for their nonlinearity, temperature drift, aging and noise.
[0020] A SELV barrier is an insulation between the primary and secondary sides of the switch-mode converter, usually with a transformer, which prevents contact with parts carrying mains voltage. Since heat sinks for LED loads in particular often have a certain potential in the circuit that drives them, compliance with the criteria for safety extra-low voltage according to DIN EN 61140 (VDE 0140-1) is mandatory in such cases. The German term Sicherheitsschutzkleinspannung is referred to in English as Safety Extra Low Voltage (SELV). A first optocoupler transmits from the input to the output side, a second in the opposite direction. On its output side, the operating device should include an analog current regulator so that only the setpoint for the output current needs to be transmitted via the first optocoupler.The control of the margin between the DC link voltage and the output voltage should be linked to the return transmission of the output voltage as a "load = OK" signal, and the margin should vary depending on the temperature of the operating device. In particular, the margin should be smaller the warmer the operating device is.
[0021] In the following, the headroom or headroom voltage refers to the headroom of the post-regulator, i.e., the ability to regulate a linear regulator or buck converter. The headroom voltage is the potential between the negative output (= negative pole for the load connection / for the LEDs) and GND or circuit ground. The headroom swing is defined as the swing of the headroom, i.e., twice the amplitude of the modulation of the DC link voltage by the mains hum. The headroom swing is thus the differential voltage from the minimum to the maximum of the headroom.
[0022] A hysteresis band controller is particularly well-suited for controlling switch-mode converters that have an inductance through which a continuous or non-continuous current flows. Due to the converter's clocking, the continuous current is sawtooth-shaped, and the difference between the resulting maxima and minima of the continuous current waveform is determined by the width of the hysteresis band. A setpoint for the continuous current is usually located in the middle of the hysteresis band. Then, using a symmetrical Schmitt trigger, which compares the continuous current waveform with the corresponding setpoint, the control for the converter's clocking can be generated directly. The narrower the hysteresis band in the Schmitt trigger, the higher the clock frequency.If the current exceeds the upper limit of the hysteresis band, an active power transistor in the clocked converter switches off. If the current falls below the lower limit of the hysteresis band, the same power transistor switches on again. As can be seen from this brief explanation, the same result can also be achieved by using a simple comparator to compare the continuous current waveform with the corresponding setpoint, and by delaying both switching actions triggered by the comparison—power transistor on when current is below the setpoint, and power transistor off when current is above the setpoint. The resulting clock frequency is indirectly proportional to the time delay.
[0023] A special form of this is a constant-off-time control, which again works with a simple comparator as a clock generator, and in which only the switch-on command is delayed, resulting in an approximately constant switch-off time for the active power transistor. After this switch-off time, the power transistor is switched on, and as soon as the sawtooth current waveform exceeds its setpoint, the power transistor is switched off again. This control is similar to switch-off upon detection of a peak current, but is also a hysteresis band control, with its lower limit defined by the constant off-time and its upper limit corresponding to the setpoint for the continuous current. A static difference between an average value of the sawtooth current waveform and its peak value must be compensated by an integrating component of a controller connected upstream of this control.
[0024] In the event of excess temperature, the headroom, which is amplitude-modulated by the mains hum, should also be able to dry up in sections and periodically to interrupt the activity of the post-regulator, thereby initiating a natural current reduction that protects the operating device. At the same time, the operating device is then relieved of some of the losses in the post-regulator. In contrast to the literature cited above, however, this should only occur here depending on the situation, in particular temperature. Thus, operating points without interruptions in the activity of the post-regulator should be possible, especially under normal conditions, thereby enabling very smooth and thus very small or very deeply dimmed output currents. Temperature measurement should be performed directly by the control circuit; if necessary, the sensors should be isolated from the object to be measured by the circuit board.Due to particularly good cooling of the output side of the operating device, its working range should be expandable.
[0025] In an overtemperature range, the post-regulator should periodically interrupt its activity, and the control circuit should be able to calculate the differential load resistance by comparing several different operating points over an extended operating period in order to adjust the length of the periods without margin, i.e., with interrupted post-regulator activity, to a permissible ripple of the output current. In particular, it should be possible to determine how far the output current drops at a given temperature and the resulting interruption time, what average value for the output current results from this, and whether it might not be better to lower the setpoint toward this average value in order to prevent heating of the operating device, shorten the interruption times, and thus reduce the ripple of the output current again.
[0026] Another parameter for the modulation of the headroom by the mains hum, i.e., the magnitude of the voltage amplitude at twice the mains frequency over the headroom, is the capacitance of the intermediate circuit capacitor, which decreases by up to 40% over its lifetime. In particular, the energy stored there must be related to the instantaneous power drawn, which initially shows that the lower the power drawn and largely passed on to the device output, the smaller the mains hum is. And the smaller the capacitance of the intermediate circuit capacitor is, the greater the mains hum is. According to the formula CU 2the stored energy decreases quadratically with a decrease in the intermediate circuit voltage and with it the mains hum even increases quadratically, provided that the same power is always drawn. Even if the output power decreases with a decrease in the intermediate circuit voltage, as is the case with resistive loads, for example, there is still a linear increase in the mains hum. The higher the mains hum, the faster a DC link capacitor ages due to its self-heating caused by losses in the dielectric. Over very long periods of time, the entire system is unstable because an increasing mains hum caused by a decreasing DC link capacitance also increases the losses and thus the power drawn at the time, which in turn increases the mains hum and accelerates aging, ultimately causing the DC link capacitance to decrease even faster.Conventional designs with fixed DC link voltages and large margins require heavily oversized DC link capacitors to meet the required service life because the condition at the end of the service life must be taken into account in the calculation.
[0027] A further objective of the invention is to cleverly control the operating range, either by using a smaller DC link capacitor for a constant service life, or by achieving a longer service life for the operating device with the same DC link capacitor. Intermediate solutions in the form of a slightly smaller DC link capacitor with a somewhat longer service life are also being pursued. In particular, the properties of an unused DC link capacitor are to be better utilized by deliberately underloading it.
[0028] Description of the invention
[0029] The object is achieved according to the invention with a circuit arrangement for operating a load having an input for inputting a mains power, and an output with a positive output terminal and a negative output terminal for operating a load with an output current and an associated output voltage, a first clocked converter with a SELV barrier between its primary side and its secondary side for generating a converter output current in order to generate an intermediate circuit voltage with a ripple on the secondary side and thereby regulate it to an adjustable mean value of the intermediate circuit voltage, wherein the circuit arrangement is designed to regulate the first clocked converter such that the mean value of the intermediate circuit voltage can follow the output voltage with a distance corresponding to the margin voltage,a post-regulator connected downstream of the first clocked converter for regulating the ripple on the intermediate circuit voltage generated by the first clocked converter, a first optocoupler for transmitting information from the primary side to the secondary side about a current setpoint of the output current, a second optocoupler for transmitting information about a margin voltage from the secondary side to the primary side to enable regulation of the intermediate circuit voltage by the first clocked converter, wherein the margin voltage corresponds to the potential difference between the negative output terminal and a circuit ground, precisely one control circuit on the primary side for regulating the circuit arrangement, wherein the control circuit is designed to regulate the margin voltage to the smallest possible value,This allows for a low power loss in the circuit arrangement, especially in the post-regulator, while simultaneously keeping the output current ripple as low as possible. This measure enables precise control of the output current without establishing a closed control loop by simply transmitting the setpoint of the output current from the primary side of the SELV barrier to the secondary side, and simple information about the currently prevailing margin voltage is transmitted from the secondary side to the primary side. The circuit arrangement is advantageously designed such that the first switch-mode converter advantageously only adjusts the intermediate circuit voltage according to the information about the currently prevailing margin voltage, and the post-regulator advantageously adjusts the output current and compensates for the ripple of the intermediate circuit voltage.
[0030] In a preferred embodiment, the control circuit regulates the value of the headroom voltage and finely adjusts it depending on the output current and / or the output voltage and / or the temperature at the circuit components and / or the number of operating hours. The headroom voltage is increased within the fine-adjustment range up to an upper value, so that the power loss in the circuit arrangement, and especially in the post-regulator, is increased up to a predetermined limit in order to minimize the ripple of the output current. This measure advantageously provides a very high quality of the output current and thus, when LEDs are connected as the light source, a very good light quality without significant light modulation.
[0031] In another alternative embodiment, the control circuit regulates the value of the headroom voltage and finely adjusts it depending on the output current and / or the output voltage and / or the temperature at the circuit components and / or the number of operating hours. In such a way that the headroom voltage is reduced to a lower value in the fine-adjustment range, so that the ripple is increased up to a predetermined limit in order to set the power dissipation in the circuit arrangement, and especially in the post-regulator, to an even lower value. This measure advantageously reduces the power dissipation of the circuit arrangement to a minimum, which is particularly advantageous where the circuit arrangement is installed at high temperatures, because it protects the circuit arrangement from premature failure.
[0032] In a preferred embodiment, the control circuit is configured to operate the first optocoupler purely digitally and to operate the second optocoupler either in a mixed mode, wherein the second optocoupler transmits a periodically recurring analog signal with gaps greater than zero between the sections of the analog signal, or the control circuit is configured to operate the second optocoupler purely digitally, wherein the second optocoupler transmits a digital signal having a low level that corresponds in time to the gaps in the analog signal and otherwise has a high level. This measure advantageously ensures that the optocouplers are operated in such a way that the known aging phenomena do not have a negative impact on the function of the circuit arrangement, since these play no role in digital operation, and neither do they play a role in mixed mode operation.
[0033] In a particularly preferred embodiment, the first switch-mode converter also functions as a power factor corrector. This advantageously represents a particularly simple, cost-effective, and space-optimized circuit solution for compliance with the required standards for grid operation.
[0034] In another embodiment, a microcontroller is part of the control circuit. This represents a cost-effective and highly flexible solution for designing the control circuit.
[0035] In a particularly preferred embodiment, the post-regulator is a linear regulator. This is a particularly cost-effective solution that ensures maximum output current quality with manageable losses, as linear regulators can regulate very quickly, thus effectively compensating for ripple in the output current.
[0036] In another embodiment, the post-regulator is a second clocked converter in the form of a buck converter. This represents an alternative that advantageously generates less power loss, but is more complex and expensive to manufacture. The buck converter can be controlled using hysteresis band control, which is particularly simple and cost-effective to implement. As a special version, the buck converter can also be controlled using constant off-time control after peak current detection, which is even simpler to implement.
[0037] In a further preferred embodiment, the post-regulator is configured to regulate the output current to a setpoint without transmitting any actual output current value back to the control circuit. This has the significant advantage of significantly simpler feedback, since no analog value needs to be transmitted, which could be corrupted by the aging of the optocoupler and then result in an incorrect current being delivered at the output of the circuit arrangement as the number of operating hours increases.
[0038] In a further related embodiment, the control circuit is configured to transmit the setpoint purely digitally via the first optocoupler to the post-regulator. As already mentioned above, this has the great advantage that the aging phenomena of the optocoupler cannot negatively impact the quality of the current regulation. Particularly preferably, the control circuit is configured to transmit the setpoint as a duty cycle of a pulse width modulation via the first optocoupler to the post-regulator, wherein the associated frequency is 30 Hz to 1 kHz, advantageously 60 Hz to 170 Hz, and particularly advantageously 140 Hz or 145 Hz. This measure offers the great advantage of simple and cost-effective generation of the pulse width modulation, whereby the control can operate very precisely and the signal, as a digital signal, is not subject to degeneration by the optocoupler.
[0039] In a particularly preferred embodiment, the circuit arrangement is configured to measure the headroom voltage in the post-regulator and to evaluate the measurement result by a detection circuit, wherein the detection circuit outputs a signal to the second optocoupler as long as the headroom voltage exceeds a limit, wherein the signal contains the amount by which the headroom voltage limit is exceeded, and wherein the limit is stored in the detection circuit. This measure offers the very significant advantage that no quantitative signal, but only a qualitative signal, needs to be generated and transmitted. This means that aging phenomena and component tolerances play no role here either, since the purely qualitative nature of the signal allows optimal results in regulating the circuit arrangement even with minor deviations.
[0040] The problem is also solved with a control method for a circuit arrangement described above, characterized by the following steps:
[0041] - Receiving the signal from the control circuit,
[0042] - Control of the first clocked converter to an intermediate circuit voltage according to the value of the signal
[0043] - Transferring the setpoint of the output current via the first optocoupler to the post-regulator,
[0044] - Regulating the ripple of the intermediate circuit voltage and regulating the output current of the circuit arrangement by the post-regulator,
[0045] - Transmitting the signal from the detection circuit to the control circuit. This method establishes a feedback loop control system, which is significantly simpler than conventional control methods, since a quantitatively precise signal does not need to be transmitted in every direction, but rather only a qualitative signal, which places significantly simpler demands on the components involved and thus enables a circuit arrangement that can be manufactured more cost-effectively while maintaining the same level of quality.
[0046] In a preferred embodiment of the control method, the minimum margin voltage is 0.0V to 0.5V, and particularly advantageously 0.0V to 0.2V, and the range of the margin voltage is 1V to 2.5V, and particularly advantageously 1.3V to 1.8V if the post-regulator is a linear regulator. This advantageously ensures particularly efficient operation with minimal losses of the post-regulator.
[0047] In an alternative preferred embodiment of the control method, the minimum of the margin voltage is 0.0V to 10V, and particularly advantageously 0.0V to 2.5V, and the swing of the margin voltage is 2V to 12V, and particularly advantageously 5V to 8V if the post-regulator is a buck converter. This measure also ensures efficient operation with optimal buck converter performance.
[0048] In a further preferred embodiment, the control circuit measures a temperature of the circuit arrangement, and the intermediate circuit voltage is reduced if the measured temperature exceeds a predetermined temperature. This advantageously ensures that the circuit arrangement cannot be overloaded and thus has a very long service life.
[0049] Preferred embodiments can be found in further dependent claims and the entire disclosure, whereby the presentation does not always distinguish in detail between device and use aspects; in any case, the disclosure is implicitly to be read with regard to all claim categories.
[0050] Brief description of the drawings. Further advantages, features, and details of the invention will become apparent from the following description of exemplary embodiments and from the drawings, in which identical or functionally equivalent elements are provided with identical reference numerals. In the drawings:
[0051] Fig. 1 an overall situation with a large fixed limit for the margin and with its regulation symmetrical around this limit,
[0052] Fig. 2 a second overall situation with the same size limit and regulation of the margin from this limit asymmetrically downwards,
[0053] Fig. 3 a third overall situation with the same limit and regulation of the margin to a minimum possible value,
[0054] Fig. 4 a fourth overall situation with a small fixed limit for the margin and with its regulation symmetrical around this limit,
[0055] Fig. 5 a fifth overall situation with an equally small limit and regulation of the margin from this limit asymmetrically downwards, whereby the margin periodically dries up, at the same time the work of the post-regulator is interrupted, and therefore the output current collapses,
[0056] Fig. 6 shows a possibility of regulating higher margins using the specified method,
[0057] Fig. 7 the block diagram of an operating device,
[0058] Fig. 8 a simple circuit variant for the output side of the operating device with a linear post-regulator,
[0059] Fig. 9 shows an improved circuit variant for the same output stage according to Fig. 8,
[0060] Fig. 10 shows a first circuit variant for the output side of the operating device with a clocked buck converter as post-regulator,
[0061] Fig. 11 shows a second circuit variant for the same output side according to Fig.
[0062] 10, Fig. 12 a third circuit variant for the same output side according to Fig.
[0063] 10, and
[0064] Fig. 13 shows a temperature drift of a clocked optocoupler at low and at higher clock frequencies.
[0065] Preferred embodiment of the invention
[0066] FIG 1 shows a possible overall situation within a specified operating device, described by five electrical quantities. The constant output current l_LED in the lower graph corresponds to the time average of the rectified sinusoidal current l~, which corresponds to the input current converted to the output side. This input current is usually taken from a public AC mains supply and must therefore be sinusoidal. If the time value of the current l~ is greater than the output current l_LED, there is excess power, which is used to recharge an intermediate circuit capacitor that terminates each first clocked converter acting as a power factor correcting input stage of the operating device. Conversely, the same intermediate circuit capacitor acts as a buffer as long as the time value of the current l~ is below the output current l_LED, i.e., as long as there is a power shortage.If the intermediate circuit capacitor has only a finite capacitance, as is usually the case, the alternation between excess power and lack of power causes mains hum, visible in the ripple of the voltage SR in the middle graph of the figure, which oscillates between its MAX and MIN values. The voltage SR describes the operating margin for the post-regulator and corresponds exactly to the intermediate circuit voltage reduced by the constant output voltage (not shown). A measuring voltage M, if positive, moves at a constant distance G below the margin SR. This voltage is shown again in the top graph of the figure and leads directly to an evaluation signal DIGI: DIGI is "one" or at a high level if the measuring voltage M is greater than zero. The constant distance G corresponds to the aforementioned fixed limit for the output voltage-dependent regulation of the margin SR.
[0067] FIG 2 uses the same five electrical variables from the previous figure to illustrate a first method for reducing the actual margin while maintaining a fixed, predefined limit G. It also illustrates the general principle underlying the control method for the specified operating device. The duty cycle of the DIGI signal is reduced. This occurs indirectly by reducing the power of the first switch-mode converter operating as a power factor correcting input stage, while DIGI is continuously monitored. The power is reduced until DIGI displays a desired duty cycle. The principle works the same in the other direction: If the duty cycle in DIGI increases because more power flows through the input stage, the intermediate circuit voltage and thus the margin SR also increase, which is undesirable.In short, DC-link voltage regulation is reduced to the bare minimum, and its characteristic, its ripple, is additionally exploited. Only the presence of a margin SR is checked. Its approximate size is guaranteed by the limit G, and the output voltage—constant and arbitrary within a given range—intrinsically "inflates" the very small margin SR to a true DC-link voltage, as currently required by the connected load.
[0068] Its maximum MAX and thus simultaneously the maximum of the margin SR is temporally located at the end of each phase with excess power, i.e., when the sinusoidal current l~ crosses the output current l_LED on a falling slope. Its minimum consequently occurs at the end of each phase with a lack of power, including the current zero crossing of l~, i.e., when the sinusoidal current l~ crosses the output current l_LED on a rising slope. At the time of this current zero crossing of l~, the margin SR has its greatest decrease and, at the same time, its average value. The same applies at the time of the maxima of l~: there, the margin SR has its greatest increase and, again, its average value. This is because the total decrease between MAX and MIN must correspond to the total increase, and because of the sinusoidal current l~ causing all this, the points of the average value of SR also lie, temporally speaking, exactly between the respective times of MAX and MIN.Because the maxima of the sinusoidal current l~ are closer to the output current l_LED than its zero crossings, the largest increase in the margin SR is smaller in magnitude than its largest decrease. This ultimately means that the durations of increases in an intermediate circuit voltage generated by a power factor correcting input stage are always longer than the durations of decreases in the same intermediate circuit voltage as soon as active power is drawn from the intermediate circuit. The same applies to the margin SP and, along with the exact doubling of the frequency of an AC mains supply, is the second typical characteristic of any mains hum originating from a power factor corrector.
[0069] In addition to the above-mentioned OB observation, the measurement voltage M enables a finer resolution for measuring the margin SR. Compared to the corresponding time curve in the previous figure, the positive sections of the M time curve are not only shorter, but also flatter or lower. Low-pass filtering this M voltage time curve produces a very finely resolved signal for the amount of margin.
[0070] In FIG. 3, the limit G has been consumed. The minimum MIN of the margin SR is now just above zero. The evaluation signal DIGI only shows short peaks, and the measurement voltage M has disappeared. This makes it clear that to apply the control method associated with the specified operating device, both signals—DIGI and M—should always be evaluated, but in any case, the pulse-shaped signal DIGI, since this contains the exact information about when the margin SR has actually reached its minimum.
[0071] Secondly, it becomes clear that such minimization or consumption of the margin SR, which must still be consistently positive to maintain functionality, is only possible if the fixed limit G is at least as large as the ripple in SR, i.e., the voltage difference between MAX and MIN, which is load-dependent. Normally, the maximum power with the minimum possible DC link voltage is used to determine the ripple. At the same time, there are many possible special cases behind this, deliberately deviating from this normal design principle in order to achieve more complex optima. Another important boundary parameter here is the capacitance of the DC link capacitor. This is because it also influences the ripple in the margin SR; the ripple is indirectly proportional to the capacitance.
[0072] Thirdly, if the amount of headroom is also a measure of losses, as is the case with linear regulators as post-regulators, it becomes clear that significantly larger DC link capacitors are required for linear regulators than, for example, for clocked post-regulators, which generate significantly lower losses than linear regulators almost regardless of their headroom SR. Large DC link capacitors enable significantly reduced MAX values of the headroom for a given MIN value, thus correspondingly lower limits G and a greater minimization of the overall headroom SR with consequent reduced losses in the linear regulator. In the case investigated, a lack of space actually led to the switch from a linear regulator to a clocked post-regulator in order to be able to install a significantly smaller DC link capacitor - also mechanically.
[0073] Fourthly, a fuzzy character inherent in the control method for the specified operating device is clearly evident here, and it focuses on the essentials. The unsightly looping effects that frequently occur, particularly in analog control loops, are elegantly avoided by simply not regulating at all in the looping ranges, and only proportionally controlling in between with an extra-high loop gain. In other words, the input stage runs unregulated at its maximum power when no DIGI signal is measured at all, because the DC link voltage is much too low. Conversely, the input stage is completely shut down if the DIGI signal is continuously measured at its high level, because in this case the DC link voltage is much too high. Only the detection of a level change due to the mains hum activates the DC link voltage and thus the headroom control.The duty cycle in DIGI covers its entire observable range from 1% to 99%, or from 0.1% to 99.9% if the margin SP changes by only 2 V, for example. Thus, a large part of the above-mentioned extra-high loop gain is already included in the measurement.
[0074] Overall, for loss-insensitive, i.e. clocked, post-regulators, the formulation G > MAX - MIN results as a rule of thumb for the fixed limit G. The minimum MIN of the margin SR can be between 0 V and 10 V, particularly advantageously between 0 V and 2.5 V. A swing of the margin SR, given by the term MAX - MIN, can be 2 V to 12 V and particularly advantageously 5 V to 8 V, which also specifies the limit G for MIN = 0 V. If minima with a positive value are to be maintained, their value must be added to G. The maximum MAX of the margin SR can thus be in a range from 2 V to 22 V. The limit G is ultimately defined by the value of MAX.
[0075] Or, for linear post-regulators, as shown in Figure 3, the control is set to a possible minimum of the SR margin, so that MIN is just positive. With this control to the minimum, the mean value of the SR margin is automatically reduced at low output currents. This is advantageous because flicker is also optimized by reducing the mean value of the margin at low currents.
[0076] FIG 4 illustrates a second method for reducing the margin SP. With otherwise identical boundary parameters as in all three previous figures, the fixed limit G is selected to be smaller. This reduction results in a similar reduction of the margin SR even at all times, as a direct comparison of this figure with Figure 1 reveals. All other time profiles are identical. If, as in Figure 4, the limit G is selected to be slightly larger than half the ripple in SR, i.e. at least as large as half the voltage difference between MAX and MIN, the decision as to whether a continuous margin SR is present or not is shifted to a duty cycle of approximately 50% on the signal DIGI, as can be seen from the
[0077] FIG 5 becomes clear. Here the duty cycle in DIGI is only slightly under 50%, and the SR margin dries up periodically, illustrated by the negative sections in the SR time curve, which are not actually possible, but do illustrate the curve of the associated intermediate circuit voltage. They also illustrate how much the output voltage is too low, because the zero line of the SR margin represents the "positive end" of the output voltage. In these negative sections of the SR time curve, the post-regulator is inactive or interrupts its activity or stops, but remains connected with low resistance, so that the load, primarily a series connection of light-emitting diodes intended for general lighting technology, remains supplied, but with a periodically reduced output current l_LED.
[0078] This reduces the output power, even though the specified operating device does not provide for this reduction, because an output current control loop is periodically interrupted. This also reduces the input power accordingly, visible in the reduced amplitude of the current l~, which corresponds to the input current. Its time characteristic, however, remains unchanged sinusoidal, because, firstly, the DC link capacitor has a very large capacitance, and, secondly, the inherent slowness of any power factor correcting controller means that this controller does not even notice such brief drops in the output current l_LED, which would manifest as a "hump" in the DC link voltage.
[0079] The extent to which the output current l_LED drops depends on the internal forward voltage of all series-connected LEDs, which acts as a counter voltage on the output of the specified control gear and is normally constantly smaller than its output voltage, and on a differential load resistance R_diff, which, together with the normally regulated output current l_LED, causes this constant, small difference. The actual output voltage, above which the margin is located, is calculated according to the formula
[0080] Actual output voltage = internal forward voltage + R_diff * l_LED.
[0081] During the periodic periods without regulation and therefore with a drop in output current l_LED, this formula is reversed to l_LED = (DC link voltage - internal forward voltage) / R_diff, taking into account that with the post-regulator inactive, the actual output voltage corresponds to the DC link voltage. At first glance, it appears as if the lower the impedance of the differential load resistance, the higher the output current can remain even during its drops. In fact, however, it is the other way around. The lower the impedance of the differential load resistance R_diff, the smaller the difference between output voltage and internal forward voltage, so the value of R_diff cancels out. Therefore, the shape of the load current drops in the I_LED curve corresponds almost exactly to the shape of the margin SR in the periods in which it is "negative."The above voltage difference is consumed more significantly, visible as a drop in the output current l_LED, the smaller it is, and even more so the lower the resistance of the differential load. Thus, if there is no margin SR or if the post-regulator fails, the lower the resistance of the differential load, the more the output current l_LED drops. If it were zero, the load current would even drop despite the clearly present, but slightly too low, output voltage.
[0082] This consideration, in turn, offers the possibility of monitoring the load more closely during normal operation, i.e., with constant regulated values of l_LED, if at least two different levels l_LED1 and l_LED2 > l_LED1 have been reached within a defined period of time. The actual output voltage is also precisely known via the margin defined by the fixed limit G. The internal forward voltage is assumed to be current-independent. At the value l_LED2, an actual output voltage U_A2 will be established that is somewhat higher than an actual output voltage U_A1 , which is established at the lower output current l_LED1. The previously unknown but important value R_diff is calculated as
[0083] R_diff = (U_A2 - U_A1) / (l_LED2 - l_LED1) according to the general definition of a differential resistance. If the control circuit of the specified operating device has a processor with non-volatile memory in which some past values for the intermediate circuit voltage and the output current are stored, it can perform the above calculation after each setpoint change and then has a picture of the currently effective value of a differential load resistance. If the calculation result is zero or negative, for example, a load change has occurred, and the last calculation is discarded.
[0084] A more advanced control circuit program can calculate the time periods during which a post-regulator is interrupted or inactive, taking into account additional boundary parameters from the applicable limit G and the duty cycle in the DIGI signal. If R_diff is known, this can be used to calculate the extent to which the load current l_LED drops. As will be shown later, the control circuit does not know the actual load current profile. Using the above drop type, the interruption period of the post-regulator, and the regulated value of the output current l_LED, a newly resulting average output current can be calculated, which can then be used, for example, as the new setpoint. Or a value between the previous setpoint for l_LED and the calculated average can be used as the new setpoint for the output current.
[0085] Periodically failing post-regulators are not the rule in the specified control gear, but the exception, which goes beyond the state of the art cited at the beginning. Furthermore, a post-regulator should only fail if it is a linear regulator, since such a regulator can get very hot. Its temperature can be measured and is therefore known to the control circuit. The higher this temperature, the smaller the margin should be, and the more the duty cycle in DIGI is reduced below 50% by reducing power, which alone counteracts further heating of the control gear.According to FIG. 5, the post-regulator's interruptions soon occur, which in themselves further reduce losses and thus reduce further heating. Their durations and the resulting load current curves l_LED can be calculated by the further developed program for the control circuit, resulting in a lower setpoint for the output current l_LED that is more favorable for the current temperature. If this is set instead of the actually specified value, the heating of the operating device can be completely contained, thus very effectively protecting the operating device against overheating. Overall, the formulation G > * (MAX - MIN) results as a rule of thumb for the fixed limit G when using linear regulators as post-regulators.Firstly, the decision limit as to whether the margin SR is still present throughout shifts to the optimally detectable range around 50% of the duty cycle in the evaluation signal DIGI, thus allowing very precise control down to the "limit of existence" of the margin. Secondly, even with dropouts in the post-regulator, a measurable duty cycle still remains in DIGI. This can be advantageously used to estimate the dropout durations. With linear regulators as post-regulators, the minimum MIN of the margin SR should be in the range between 0 V and 0.5 V, particularly advantageously in the range between 0 V and 0.2 V. A swing of the margin SR, given by the term MAX - MIN, should be between 1 V and 2.5 V, particularly advantageously between 1.3 V and 1.8 V. The maximum MAX of the margin SR thus lies in a range between 1 V and 3 V.If the duty cycle of the DIGI signal is to be exactly 50% while maintaining the exact margin, exactly half of the stroke defines the limit G using the equation G = * (MAX - MIN).
[0086] FIG. 6 expands the space depicted so far, both in terms of time and voltage. It depicts dozens of system periods, countable by the small waves along the range SR. The amplitude of these small waves, which often consume the entire range above, reveals that the range SR itself is also significantly higher here. This mode of the specified control method is useful for very low output voltages, i.e., those below 5 V or below 10 V, for example, because if the DC link voltage is tracked to just above such values, as shown in all the preceding figures, the currents in the DC link capacitor can become exorbitant. In such cases, it makes more sense to reduce losses, and in particular, aging, by increasing the DC link voltage and instead allow somewhat more losses in post-regulator 1 or 1'.This mode is particularly suitable when a post-regulator 1' contains a clocked buck converter. Furthermore, this mode requires analog feedback on the state of the margin SR via the differential signal M, as the lack of data in the DIGI signal clearly demonstrates.
[0087] Since only the margin SR is measured, both the intermediate circuit voltage and the output voltage are unknown. To correct this deficiency, the margin SR is slowly reduced over a period of time t_ramp until it is completely used up, as indicated by the drop in the DIGI signal. At this point, the intermediate circuit voltage corresponds to the output voltage. The following now occurs: Based on the duty cycle averaged over one mains period in the first clocked converter, which forms the power factor correcting input stage, its instantaneous voltage ratio is calculated. Using the measured mains voltage averaged over the same period, the intermediate circuit voltage is determined.Or the peak value of the mains voltage is detected and measured. At the same time, the current voltage ratio of the first switch-mode converter is determined based on the current duty cycle in the power factor correcting input stage. Multiplying the two results in the current value of the intermediate circuit voltage. This value is stored as the output voltage. The margin is then increased again, preferably for the same time period t_ramp, and the device knows that the intermediate circuit voltage corresponds exactly to the sum of the margin SR and the stored output voltage. This mode is possible because, when LEDs are used as the load, the output voltage either does not change at all or changes very slowly.
[0088] This entire procedure is repeated every 0.1 seconds to 60 seconds, and depending on this, the duration t_ramp is particularly advantageously between 0.02 and 15 seconds. Since the gradient of the ramp must not be too high to reliably exclude flicker and other aliasing effects, the larger the ramp SR is desired, the less frequent the repetitions must be, and the duration t_ramp must be longer.
[0089] Without the procedure shown in Figure 6, i.e., in the situations shown in Figures 1 to 5, the intermediate circuit voltage can be calculated in the same way and then saved as such. The output voltage is determined by subtracting the known limit G from the calculated intermediate circuit voltage. Knowing the current output voltage of the operating device is important because this output voltage is limited to certain values by the SELV (Safety Extra Low Voltage) standard, for example, 60 V, which can be quickly reached or exceeded during idle operation or load shedding.
[0090] FIG. 7 focuses on the technical implementation of the specified operating device, which is configured for a control method as described so far. It shows a block diagram of a possible embodiment, characterized by an input side and an output side, which are connected to each other via a first optocoupler OK1 and a second optocoupler OK2. The first optocoupler OK1 transmits data signals from the input side to the output side of the specified operating device, and the second optocoupler OK2 transmits other data signals from the output side back to the input side of the operating device. Both optocouplers can be operated digitally, i.e., in pulsed mode, to neutralize their nonlinearity, temperature drift, aging, and noise.The connecting line between the two optocouplers, represented by a double line, marks the required galvanic isolation barrier 99, which should extend through the specified operating device. Apart from the two optocouplers, the same isolation barrier 99 is spanned only by at least one radio interference suppression capacitor (not shown) – a so-called Y capacitor – and the power transformer 30 of the first clocked converter 10, which forms the power factor correcting input stage. Its topology is only indicated here for the sake of general familiarity. Its output termination, an intermediate circuit capacitor 11, acts as the energy source for a post-regulator 1 and thus as an indispensable component thereof; in the case of a clocked post-regulator 1', even as an obligatory component according to converter theory.The intermediate circuit capacitor 11 is connected to an output-side circuit ground 21, often referred to as safety ground, SELV ground, SELV_GND, SEC_GND, or similar, and the positive terminal 31 of the intermediate circuit voltage VZ, which is buffered in the intermediate circuit capacitor 11 and whose determination is explained above. The first clocked converter, operating as a power factor correcting input stage, consequently feeds its output power into the input consisting of the two terminals 31 and 21. The result of this feed, the output current l_LED, is measured in the post-regulator 1 or 1' and fed back in the form of a measuring voltage on a first feedback line 41.
[0091] The output current flows via a positive output 161, which is directly connected to the positive pole of the intermediate circuit voltage VZ, through a load 8 back to a negative output 61 of the post-regulator 1 or T. According to a static resistance and / or a forward voltage of the load 8, the output current l_LED creates an output voltage V_LED between the positive output 161, which is intended for connecting the positive pole of the load, and the negative output 61, which is intended for connecting the negative pole of the load.
[0092] The post-regulator 1 or 1' is controlled by a controller 2, comprising an operational amplifier 12 as the only active part of the controller with output 22. The entire controller is advantageously located on the output side of the specified operating device and particularly advantageously has a finite proportional integrating gain, i.e. is a PI controller, effected by its negative feedback surrounding the operational amplifier 12 from the output 22 to the negative input of the operational amplifier, which essentially consists of a series circuit of an integrating capacitor 32 and a negative feedback resistor 42.Equally important, particularly for the quantitative level of the gain of controller 2, is a feedback resistor 52, which is connected to the feedback line 41 and, together with the negative feedback resistor 42, either directly coupled to the negative input of the operational amplifier 12 or connected to a first small inductance 62, which can be a ferrite bead choke, serves to block very high-frequency interference signals, which can originate, for example, from mobile telephone networks, and is in turn connected to the negative input of the operational amplifier 12. A second small inductance 72 is also particularly advantageously connected between the positive input of the operational amplifier 12 and the setpoint input 63 of the entire controller 2.Optionally, and for even more effective blocking of high-frequency interference, the two small inductors 62 and 72 can also be coupled by means of a common bead 82 in such a way that a common-mode choke is created in the direction of the two inputs of the operational amplifier 12. Figuratively speaking, both input lines for the amplifier 12 then exit the bead 82 on the same end face.
[0093] If a periodic signal is to be subtracted from the controller setpoint, this periodic signal can be input as a voltage with any reference potential at a point 112. From this point, a high-pass filter comprising at least a high-pass capacitor 92 and a series resistor 102 leads to the first small inductor 62 or directly to the negative input of the operational amplifier 12.
[0094] Controller 2 receives its setpoint 63 from a two-stage low-pass filter 3. Why a single-stage low-pass filter is insufficient is explained towards the end. An input voltage divider consisting of divider resistors 13 and 23 attenuates a pulsed input signal 24 and immediately smooths it using the first filter capacitor 33. A subsequent second low-pass filter comprising at least one filter resistor 43 and a second filter capacitor 53 maps the approximately smooth average of the voltage across the first filter capacitor 33 as the voltage across the second filter capacitor, which is passed to the controller as setpoint 63.
[0095] Low-pass filter 3 therefore has the task of passing on a time average of the pulse-shaped input signal 24, weighted according to its input voltage divider, to the controller as setpoint 63.
[0096] If a periodic signal is to be added to the setpoint for the controller, this periodic signal can be input as a voltage with any reference potential at point 113. From this point, a high-pass filter comprising at least a high-pass capacitor 93 and a series resistor 83 leads to the setpoint output 63 of the low-pass filter 3.
[0097] The precision of signal 24 at the input of low-pass filter 3 increases significantly when only one parameter varies, thus representing the signal value. In this case, a pulse-pause ratio is particularly advantageous due to the optocoupler as the transmission path. The voltage of signal 24 during the pulses, however, should be constant, and that during the pauses should be constant zero. Both of these are accomplished by a voltage shaper 4 connected upstream of low-pass filter 3. Its input signal corresponds to the output signal of the first optocoupler OK1 on a receive line 14, whose transmitting diode is connected via a current-limiting resistor 19 to an auxiliary voltage 3V3 and the output SOLL of a control circuit 9 located on the input side of the specified operating device. The input side includes the inverter side of the first clocked and simultaneously power factor-correcting converter as well as a control circuit 9.
[0098] If the SOLL output is connected to ground or to the reference potential GND of control circuit 9 with low impedance, which occurs periodically with a pulse-pause ratio to be transmitted, the transmitting diode of the first optocoupler OK1 also lights up periodically, and the receiving transistor of the same optocoupler also becomes periodically conductive, causing the receiving line 14 to also periodically assume a high level, essentially with a pulse-pause ratio identical to the signal to be transmitted. The only distortion is caused by the dynamic and other properties of the optocoupler OK1.
[0099] This fully outlines a key characteristic of the specified control method and a control device designed for it. The actual control of the output current l_LED is outsourced to a fast and precise analog circuit on the output side of the control device, comprising at least the voltage shaper 4, the two-stage low-pass filter 3, the PI controller 2, and the post-regulator 1. This entire arrangement is precise and fast enough that the control circuit 9 on the input side of the control device does not even need to know the exact I_LED time characteristic, but can rely on the information "l_LED is proportional to the mark-to-space ratio of the low-impedance of SOLL." A constant proportionality factor in between, due to the actual design of the control device, must of course be taken into account when generating the mark-to-space ratio in the control circuit 9.For the described "outward path" to function, i.e., for an output current l_LED to flow, the intermediate circuit voltage VZ buffered in the intermediate circuit capacitor 11 must be sufficiently large. For this to happen, the primary side of the first clocked converter of the specified operating device must operate correctly, which requires at least one signal return path. As already explained above, it is sufficient to report only the presence of a margin on this signal return path, which represents a second essential property of the specified control method and the operating device provided for it. For this purpose, the post-regulator has a second feedback line 51, which is essentially connected to the negative output 61 of the post-regulator and thus carries a differential voltage with respect to the output-side circuit ground 21 that corresponds to the margin SR.
[0100] The voltage on feedback line 51 is fed into a detection circuit 7, into which the aforementioned limit G is permanently designed. The detection circuit 7 compares the level on feedback line 51 relative to the output-side circuit ground 21 with the limit G and connects the transmission line 17 to ground 21 with a low impedance whenever the level 51 is above the limit G, i.e., whenever the margin is present and, according to the definition of G, is also large enough.
[0101] The second optocoupler OK2 always sends a signal back to the control circuit 9, which is received in two ways. A digital output DIGI is connected to the auxiliary voltage 3V3, limited only by the series resistor 29, as soon as the first photon passes through the optocoupler. A high-impedance pull-down resistor 39 is connected in parallel to this, which in turn is connected in parallel to a low-pass filter consisting of a filter resistor 49 and a filter capacitor 59. Its analog output ANA delivers a voltage proportional to the time-averaged value of the analog measurement voltage M explained above.
[0102] Purely numerical reference symbols such as 41 or 51 can, both above and throughout the following, describe voltages, their associated voltage-time profiles, or their associated electrical connections, paths, lines, or nodes. Furthermore, the same reference symbols also describe the functions of the voltages or lines, so the combination of only the function and the associated reference symbol can occur, such as "feedback voltage 51." Since it is clear from the context which aspect is meant in each case, splitting the reference symbols according to these aspects makes no sense. Instead, the usual structure in electronics based on subnetworks is followed.Accordingly, the same identifier and / or symbol (especially a ground symbol) can appear multiple times in one and the same figure, thus adequately describing the corresponding subnetwork without showing the direct electrical connections of the subnetwork. Furthermore, all described electronic components are designated purely numerically. However, if the values of the designated components or electrical quantities are required in formulas, they are preceded by the qualifying letters commonly used in electrical engineering, such as R45 for the value of resistance 45 in ohms or U22 for the value of the voltage on line 22 in volts.
[0103] A first switch-mode converter 10, which forms the input stage and is indicated as a flyback converter, supplies power to the entire operating device. This topology is often used when the input stage formed by it is required to perform the following three additional tasks in addition to supplying power:
[0104] • A power factor correction as required for all general lighting purposes, i.e. the shaping of an input current or mains current coming from an AC mains supply into a waveform proportional to the mains voltage of the same mains supply, ideally into a sine wave, and the minimization of the phase shift between mains current and mains voltage,
[0105] • galvanic isolation of the output side of the operating device from its input side, and
[0106] • the generation of an intermediate circuit voltage VZ that is lower than the peak value of the mains voltage. Its most important component is an isolation transformer 20, which not only temporarily stores the inductive energy required for every switch-mode converter, but also transmits the required power across the isolation barrier 99, thereby ensuring the required isolation voltage of several thousand volts. Because at least two of its windings are DC-loaded, the core of the isolation transformer 20 requires an air gap. One of its windings recharges the intermediate circuit capacitor 11 via a rectifying diode 30 and the positive pole 31. The return for this recharging occurs via the output-side circuit ground 21, to which the same winding that is connected to the anode of the rectifying diode 30 is also connected.The other winding of the isolation transformer 20 is connected to an input-side terminating capacitor 40 and to the working electrode of an actively switched power transistor or inverter switch 80, shown only symbolically. Its reference electrode is coupled to the input-side circuit ground GND with at least a low resistance. In order for the flyback converter to deliver a positive output voltage, the two windings of the isolation transformer 20 must be oriented in opposite directions, as shown by the dots. The actively switched power transistor 80 is controlled via a control line 90, often also called the gate or base line. A pulsed voltage is applied to this line, which corresponds in time to a similarly pulsed output voltage 110 from the control circuit 9. This voltage controls the first switched converter 10 and is amplified by a control amplifier 100.The recharging of the terminating capacitor 40 is performed via a mains rectifier 50, which is preceded by a radio interference filter 60. The entire specified circuit arrangement is supplied with electrical energy via a mains input 70, which is connected to the radio interference filter 60 and is configured for connection to an AC mains supply.
[0107] FIG. 8 shows a simple embodiment of the specified operating device. As shown, the load 8 consists of a series circuit of several light-emitting diodes, which are preferably intended for general lighting technology. The post-regulator 1 is designed as a simple linear regulator, comprising a current-sensing resistor 71 and, connected in series with it, a linearly amplifying transistor 81, which, as shown, is advantageously designed as a MOS field-effect transistor. This type of transistor is controlled by a variable voltage at its gate 91, which can come directly from the output 22 of the PI controller without imposing a static load on this output. Advantageously, a matching resistor 15 is also connected between the output 22 and the gate line 91.The current measuring resistor 71 is connected in series between the output-side circuit ground 21 and the source of the linear amplifying transistor 81, and the node therebetween is connected to the first feedback line 41. Thus, a voltage corresponding to the output current l_LED is present on this line relative to ground 21. The drain terminal of transistor 81 is connected to the negative terminal 61 of the load so that this linear amplifying transistor can conduct the output current l_LED. The second feedback line 51 is also connected to the node therebetween, so that a voltage is present there relative to ground 21 that corresponds to the difference between the intermediate circuit voltage VZ and the output voltage and thus precisely to the margin SR to be monitored.This eliminates the need for direct measurement of the intermediate circuit voltage VZ, which would otherwise require a third transmission path to the control circuit 9 located on the input side, as can be seen in the circuit diagram. Since the third transmission path is also eliminated, this design is particularly cost-effective.
[0108] The detection circuit 7 essentially consists of an inverting NPN bipolar transistor 27, although any other transistor type can be used in its respective inverting circuit at the same location. A softening resistor 37 is located between its emitter and the output-side circuit ground 21, across which the measuring voltage M drops during normal operation of the specified operating device. The base-emitter forward voltage of the bipolar transistor 27 forms the limit G, whereby the relationship "M = SR - G if SR > G" explained above is a solution realized with real components. If SR is smaller than a forward voltage, the inverting transistor 27 switches off completely; otherwise, it conducts according to the term M / R 37a current that also flows through the transmitting line 17, which is directly connected to the collector of transistor 27. The transmitting diode of the second optocoupler OK2 is thus flowed through by a current according to M / R37, which is why the term "softening resistor" 37 makes sense. The larger the value of R 37 is selected, the less current flows through the transmitting diode of the second optocoupler OK2. Since the output auxiliary voltage S6V7 of this optocoupler, which also supplies the operational amplifier 12 in controller 2, is always selected to be greater than SR, the voltage difference is absorbed by the collector of transistor 27. A base series resistor 47 connects the second feedback line 51 to the base of transistor 27.
[0109] In the simplest case shown here, the voltage shaper 4 consists only of a Zener diode 34, which ensures a required uniform voltage level for the high level of signal 24, which in turn must be lower than the secondary-side auxiliary voltage S6V7, and a pull-down resistor 44 connected in parallel to the Zener diode 34, which ensures a zero voltage for a low level of signal 24. For temperature compensation, a flux diode 64 can be connected in series with the Zener diode 34, and a series resistor 54 can be connected upstream of both to limit the current in the receiving transistor of the first optocoupler OK1. Thus, the signal 24 is generated either at the anode of the flux diode 64 or at the cathode of the Zener diode 34. The signal 24 contains its information only in its pulse-pause ratio and is passed on to the two-stage low-pass filter 3.All of this, especially the generation of uniform voltage levels for signal 24, requires the connection of the Zener diode 34, the pull-down resistor 44 and the low-pass filter 3 to the output circuit ground 21.
[0110] FIG 9 further develops the previously described embodiment. In particular, the detection circuit 7 is implemented in more detail. The inverting bipolar transistor 27, including the base series resistor 47, now represents only one output stage, which terminates a differential amplifier, a subtracting transistor 57, an adding transistor 67, and a current source resistor 77. The latter is supplied by the same output auxiliary voltage S6V7 as the transmitting diode of the second optocoupler OK2. A terminating resistor 87 is connected in series between the collector of the subtracting transistor 57 and the output circuit ground 21 and is thus parallel to the already known series circuit comprising the base series resistor 47, the base-emitter path of the inverting transistor 27, and the softening resistor 37.This is because the base series resistor 47 is connected directly to the output of the differential amplifier, i.e., to the collector of the subtracting transistor 57. If a current mirror were placed opposite the differential amplifier, as is usual with comparators or operational amplifiers, the output of such an overall arrangement would be at the collector of the adding transistor. A protective resistor 97 is connected between the transmit line 17 and the collector of the inverting transistor 27, and a trimming resistor 107 connected to the auxiliary voltage S6V7 helps adjust the operating point for transistor 27.
[0111] The purpose of this further development is, firstly, to map the limit G more sharply and, in particular, to make it lower than a base-emitter forward voltage, and secondly, to make the voltage signal for the margin SR independent of the current output current l_LED. To achieve the latter, this voltage across the current measuring resistor 71 must be subtracted from the voltage on the second feedback line 51, which, as already described above, corresponds to the sum of the operating voltage of the linearly amplifying transistor 81, i.e., the actual margin, and the voltage across the current measuring resistor 71 as a measurement signal on the first feedback line 41 and is thus output current-dependent. The previously described differential amplifier 57+67 in the detection circuit 7 is responsible for this.For its operation, line 41 is also routed out at another point on post-regulator 1 and coupled to the base of subtracting transistor 57 via a subtracting series resistor 127. The second feedback line 51 is coupled to the base of adding transistor 67 via an adding series resistor 137. The subtraction of the current measurement voltage on line 41 is an "inherent advantage" and can also be omitted by connecting the left end of subtracting series resistor 127 to output-side circuit ground 21 (not shown) instead of to current measurement resistor 71. This is because the voltage divider consisting of a reference resistor 117 and subtracting series resistor 127 primarily defines the limit G according to the formula G = S6V7 * RI. 27 / (R117+R127) , according to which, with an appropriately low-ohm subtracting series resistor 127, very low limits G can be represented, even lower than a diode forward voltage, which is particularly advantageous for linear regulators as post-regulators 1. Towards the end of the description of Figure 5, it is stated that such a limit can be 0.5 V. The desired sharpness of the limit detection is taken over by the differential amplifier in the following way: If the margin SR in the form of the voltage on the second feedback line 51 becomes greater than the limit G, the emitters of both differential amplifier transistors 57 and 67 follow this upward movement, which causes the transistor 57 to turn on. This also turns on the inverting transistor 27 and sends a current of the magnitude S6V7 / (R97+R37) through the second optocoupler OK2 if the softening resistor 37 is very low-ohm or zero.If the margin SR becomes lower than the limit G, both emitters follow this downward movement again, which leads to the switching off of transistor 57 and thus also of transistor 27, thus ending transmission. Exceeding the limit G by the margin SR is detected much more sharply here and reported back to the control circuit 9 than in the circuit shown in the previous figure. A softening of this detection, which, as already described, can also be useful here, is achieved by a higher-ohm softening resistor 37 and current source resistor 77, as well as by a lower-ohm terminating resistor 87. The advantage of the free adjustability of the limit G by resistors 117 and 127 remains.
[0112] In the event of a short circuit between the output terminals 31 and 61 of post-regulator 1, regulator 2 is unable to regulate the gate voltage 91 quickly enough. At the same time, a voltage corresponding to a high short-circuit current drops across the current-sense resistor 71, which can be greater than 1 V. This is because the intermediate circuit capacitor 11 is ultimately discharged via the current-sense resistor 71—only slowed by the impedance of the short circuit itself. A circuit consisting of a series resistor 111 and a protection transistor 101 then intervenes because the voltage across the parallel-connected sense resistor 71 becomes greater than a base-emitter forward voltage. The protection transistor 101 thus becomes low-impedance and pulls the gate voltage 91 against the output-side circuit ground 21, causing the amplifying transistor 81 to become high-impedance particularly quickly, thus protecting the post-regulator 1 from destruction.To avoid overloading the output 22 of regulator 2, the matching resistor 15 must no longer be bridged, unlike in the previous figure. A second matching resistor 105, parallel to the working path of the protection transistor 101, forms a voltage divider with resistor 15 in front of the gate 91, which allows the regulator output voltage 22 to be slightly attenuated during normal operation.
[0113] FIG 10 leads to the clocked post-regulators 1', which, as already derived, basically comprise a buck converter, since this is the only sensible converter topology at this point that behaves like a linear regulator over the entire operating range and, in particular, can statically connect its input to its output without any margin and without clocking. The advantage of such clocked post-regulators is their significantly lower losses compared to their linear counterparts, and this is also significantly less dependent on their margin. Nevertheless, it also makes a lot of sense to keep the margin small here. Because of the particularly small spread between the input and output voltage of such buck converters, they can be operated particularly advantageously even in continuous conduction mode (CCM), since the particularly lossy diode conduction phases are then particularly short.Conversely, with otherwise identical ambient parameters, the clock frequencies here are higher than in so-called critical conduction mode (CRM) or transient conduction mode (TCM) and much higher than in the even lower-frequency discontinuous conduction mode (DCM). Because the voltages across a buck converter coil or "buck choke" 121 are significantly lower here and, above all, reliably consistently lower at all possible operating points – their positive amplitude corresponds to the margin – than in corresponding circuits with a fixed intermediate circuit voltage designed for the highest possible output voltage, the buck converter coil 121 can be selected to be up to five times smaller than in those circuits. An EF16 or EF20 design can be replaced here with a "yarn core coil" 121.The linearly amplifying transistor 81 above becomes the clocked power transistor 81', advantageously of the same type, and swaps the order with the current measuring resistor 71. The measuring resistor is now connected in series with the drain terminal of the power transistor 81' and thus directly to the negative terminal 61 of the post-regulator 1'. If a clocked power transistor 81' and a buck converter coil 121 are present, a freewheeling diode 131 is also required. To improve its output, the entire clocked topology is terminated by the smoothing capacitor 141, which directly connects the two output terminals 31 and 61. Because the intermediate circuit capacitor 11 must terminate the input of each buck converter according to converter theory, the intermediate circuit capacitor 11 is more important in the post-regulator 1' than in the linear regulator 1 above.It is particularly advantageous if the intermediate circuit voltage VZ is applied to the input of the buck converter, which is generated and maintained by a first clocked converter as an isolating and power factor correcting input stage. Gate line 91' carries an amplified and clocked control voltage from gate driver 101'. This increases a control signal 56 with levels below 7 V at its input to a level of 12 V, which is why it is supplied by a higher secondary auxiliary voltage S12V. This higher auxiliary voltage advantageously also supplies the transmitting diode of the second optocoupler OK2, the reason for which will be explained in more detail later.
[0114] Because the existing output auxiliary voltage S12V is too high for the regulator 2, a second, much lower output auxiliary voltage S3V is generated to supply the amplifier 12 contained therein in order to reduce its losses.
[0115] Continuously operated buck converters with a small spread, i.e. a small difference between the input and output voltage, are particularly advantageously controlled via a so-called hysteresis band, whereby the necessary pulse pattern for the control signal 56 is generated directly. This requires a pulse generator 5, which contains a comparator 35 as its central and only active part, which, like the amplifier 12, can be designed as an operational amplifier. The pulse generator 5 is also particularly advantageously supplied by the smaller auxiliary voltage S3V. A double component with two identical amplifiers can then be used for the regulator and the pulse generator simultaneously. The measuring voltage across the current measuring resistor 71 is now superimposed with a gradient generated by the buck converter coil 121, which can result in an intersection with the regulator output voltage 22.If the comparator with an output voltage of 125 is converted into a Schmitt trigger by feeding the output voltage via a hysteresis resistor 45 to the positive input of the same comparator, to which the regulator output 22 is also connected as a reference via the already known matching resistor 15, the necessary hysteresis band is created. Its height corresponds to the hysteresis of the Schmitt trigger, which is defined, among other things, by the ratio between the hysteresis resistor 45 and the matching resistor 15. The higher the hysteresis, the lower the clock frequency of the continuously operated buck converter 1' will be. With a symmetrical supply, i.e. with, for example, -3 V instead of circuit ground 21 at the negative supply terminal of the comparator 35, the hysteresis is always symmetrical to a reference voltage, which here is given by the regulator output voltage 22.Above this reference voltage, there is an input voltage 41 at the negative input of comparator 35, which can also be called the measurement input, above which the comparator's output voltage 125 tilts approximately to its negative supply voltage. This so-called tilt voltage is referred to as V_down in the following. Precisely because of this tilt to an edge of the supply, unlike the linear operational amplifier 12, the value of V_down also depends on the level of the auxiliary voltage supplying comparator 35—here, the 3 V of the second and smaller output-side auxiliary voltage S3V. The exact value of the tilt voltage V_down, depending on all necessary parameters, is given by the equation according to relevant literature.
[0116] V_down - (R45 U22 +R15 * 3 V) / (R45 + R15). As with the linear operational amplifier 12, the resistance before the positive feedback, i.e., the matching resistor 15, plays a crucial role here, and the denominator of the equation, here as well, is the sum of both resistors involved in feedback (negative or positive feedback). A so-called flip-up voltage V_up, which is present at the positive input of the comparator output 125 when the level is low, and which must be undershot by the input voltage 41 so that the comparator output 125 flips back to the positive supply voltage S3V, would result from the same equation with a symmetrical supply of the comparator 35, if the plus sign in the numerator is replaced by a minus sign. Due to the lack of the negative part of the auxiliary voltage S3V, a monopolar supply is present, and the calculation is simplified to
[0117] V_up = R45 U22 / (R45 + R15) and the hysteresis is asymmetrical relative to the regulator output voltage 22 acting as the reference voltage. The height of the hysteresis is given by
[0118] V down - V_up = R15 * 3 V / (R45 + R15) to a value that, apart from the constant "3 V" = S3V, essentially depends only on the ratio of R15 / R45 of the matching resistor to the hysteresis resistor, especially when the hysteresis is small compared to the auxiliary voltage. In this case, the hysteresis resistor 45 must be chosen with a very high resistance. From the same reasoning, it follows that the pull-up voltage V_up is only just below the value U22 of the regulator output voltage 22, because the large value of R45 in the penultimate equation is dominant.
[0119] A comparator designed in this way is suitable as a hysteresis band controller if the same signal 41, which is output as an actual value to the feedback resistor 52 of the controller 2, is also used as input voltage 41 for the comparator 35 of the pulse generator 5, if necessary adjusted by an intermediate resistor 25. The second adjustment resistor 105, which occupies a different position compared to the previous figure, can possibly reduce the level of the reference voltage here as well.Because the current through the buck converter coil 121, represented by the voltage across the current measuring resistor 71 as the actual value, is sawtooth-shaped and thus periodically alternates higher or lower than the desired smooth actual value l_LED at the clock frequency of the post-regulator, which in turn should correspond to the setpoint 63 for controller 2, the operation of the buck converter included in the post-regulator T is perfectly simulated by the design of pulse generator 5, making it ideal for hysteresis band control. The correspondence between l_LED and the setpoint 63 is achieved by using the regulator output voltage 22 as a reference for the comparator 35, which has been converted into a Schmitt trigger, in pulse generator 5.Not only the above asymmetry of its hysteresis, but also any static effects of extreme duty cycles in the control signal 56 for the gate driver 101' with the clock frequency of the buck converter are compensated by the integrating capacitor 32 of the PI controller 2. Since this clock frequency is ultimately defined by when the measured current through the buck converter coil 121 leaves the hysteresis, the expression fl. ~ Rl5 * 3 V / ((R45 + Ri5) L121) ultimately results as a reference for the clock frequency fr of the buck converter in the post-regulator 1'.
[0120] By swapping the order between power transistor 81' and measuring resistor 71 in post-regulator 1', a new reference potential 51' is created for regulator 2 and, due to the high signal interconnection, also for pulse generator 5, low-pass filter 3 and voltage shaper 4. The end of current measuring resistor 71 connected to the buck converter coil forms this new reference potential 51', and instead of the second feedback signal 51, the input voltage 41 for comparator 35 in pulse generator 5 is now created at the same point, which voltage is also present on the first feedback line 41. This is because the upper end of measuring resistor 71 is connected precisely to this line.In order for the entire circuit to be able to start at all, a starting resistor 151 is provided, which extends from the lower end of the measuring resistor 71 to the circuit ground 21 and therefore separates all potentials on the output side in the first milliseconds after the operating device is started.
[0121] The new reference potential 51' lies above the output-side circuit ground 21 by the margin SR, or more precisely, reduced by the voltage across the current measuring resistor 71, which is not a problem, because the first optocoupler OK1 connected to the voltage shaper 4 can isolate a slightly different voltage than the second optocoupler OK2, which is controlled by the detection circuit 7, which is still referenced to the output-side circuit ground 21. Only the output 125 of the comparator 35 requires a level-shift stage in order to be able to switch the lower-lying gate driver 101' on and, in particular, off. A pnp level-shift transistor 65 performs this task if its emitter is connected to the comparator output 125 and, via a pickup resistor 75 between its base and the new reference potential 51', pulls the collector 95 upwards when the voltage 125 assumes a positive level.This also causes the control signal 56 at the input of gate driver 101' to become positive, transmitted by a series resistor 16, which can also be bridged. A quenching resistor 85 facilitates the re-switching off of level-shift transistor 65, causing its collector 95 to become high-impedance. A turn-off resistor 46 then pulls the control signal 56, together with the collector, against circuit ground 21, thereby reliably turning off gate driver 101* again. In its off state during normal operation, the level-shift transistor 65 must therefore only be able to block the margin SR, and in the event of an output short circuit, the intermediate circuit voltage VZ for a short time. The turn-off resistor 46 can also be connected directly to the collector 95 of the level-shift transistor 65 (not shown).
[0122] The detection circuit 7 is identical to that shown in the second-to-last figure, including all external signal connections. Only a Zener diode 147 is now connected between the emitter of the inverting transistor 27 and the softening resistor 37, with its anode connected to the resistor and its cathode to the transistor. This increases the built-in limit G, which now corresponds to the sum of the diode forward voltage of the base-emitter diode of transistor 27 and the Zener voltage of diode 147. As already mentioned, for clocked post-regulators 1', the margin SR no longer needs to be minimal; rather, a somewhat larger margin is permissible and sensible.Because of the voltage across Zener diode 147, the supply voltage for inverting transistor 27 on transmit line 17 must also be higher. Therefore, the transmit diode of the second optocoupler is supplied by the new auxiliary output voltage S12V, which is particularly advantageous at 12 V and is already used for gate driver 101*. Due to the change in order just explained, the feedback line is called 41 here and no longer 51 as in the second-to-last figure, but is connected to the negative output terminal 61 in the same way as there.
[0123] The voltage shaper 4 is significantly improved here and is supplied by a third new auxiliary voltage S5V on the output side, which is advantageously 5 V or slightly more. It even includes a cascade of the second new auxiliary voltage S3V, because this is generated from the third auxiliary voltage S5V in a voltage limiter comprising a decoupling resistor 74, a controllable Zener diode 84 (e.g., a TL431), and a voltage divider consisting of resistors 94 and 104, at a particularly advantageous level of 3 V. This voltage is adjusted via the divider consisting of resistors 84 and 104, with its tap typically being 2.5 V above the new reference potential 51'. The voltage limiter replaces the Zener diode 34 of both previous figures, in conjunction with a pulse transistor 114.Its collector is connected to the second new auxiliary voltage S3V, and its emitter is identical to the input signal 24 for the following low-pass filter 3. Because its base voltage can be higher than its collector voltage, the pulse transistor is saturated as long as it is intended to transmit a pulse. As a result, it develops only a very small and, above all, very constant forward voltage. The difference between the base and collector voltages is absorbed by the already known resistor 54, and the equally known pull-down resistor 44 ensures that the pulse transistor 114 is switched off and that the input signal 24 is reliably zero between pulses. Unlike shown, this pull-down resistor can also extend directly from the base of the pulse transistor 114 to the new reference potential 51'.
[0124] However, the circuit shown in Figure 10 encounters problems when its output current l_LED is required to be particularly small, although the asymmetry of the hysteresis of the Schmitt trigger 35 in the pulse generator, derived above, actually helps in this regard. This is because its input 41 only needs to be slightly below the reference 22 to cause the comparator 35 to flip up again. However, if the reference 22 with voltage value U22 is barely greater than zero due to the tiny required output current, the resolution for the flip-up voltage is completely lost according to the equation V_up = R45 U22 / (R45 + R15) derived above. In other words, the negative supply voltage for the comparator 35 is missing.
[0125] FIG 11 solves this problem by, first, removing the comparator 35's positive feedback 45 and thus its Schmitt trigger characteristic. Its hysteresis is virtually infinitely small, and when inputting equal potentials 41 and 22, i.e., when the output side—which is the entire purpose—is statically steady, the comparator would switch its output 125 between S3V and zero at an almost infinitely high frequency. In the event of a control deviation, however, its output 125 would be statically high or low, which in turn is unsuitable for directly clocking the buck converter 1' ultimately connected to it. Therefore, second, a delay circuit 6 is inserted between the level-shift stage following output 125, with its collector 95, and the gate driver 101*.In addition to the already known off-resistor 46, it includes at least one delay capacitor 36, a delay resistor 16, and a discharge diode 26, which discharges the delay capacitor 36 via the off-resistor 46 significantly faster than it is recharged by a positive voltage at collector 95 via the delay resistor 16. This delays all off commands somewhat, but all on commands significantly more. Both delay times are fixed. Therefore, this is a constant off-time control instead of the hysteresis band control from the previous figure.As a result, the system comprising pulse generator 5, delay circuit 6, and clocked post-regulator 1' appears to always generate an output current l_LED that is too small. This is because, at a suitable current level, the power transistor 81' is switched off, causing a drop in the current level through the buck converter coil 121. Only after the constant off time has elapsed is it switched on again, at which time the current in the buck converter coil 121 is significantly lower. Due to the sawtooth shape of this current, the output current l_LED corresponds to the time-averaged value of the current through the buck converter coil and is thus initially smaller than the setpoint. The PI controller 2 notices this, however, and increases its output voltage 22 accordingly. The controller output 22 is therefore exactly the amount above the setpoint 63 by which the above system generates "too little" current. This "above" contains the solution to the problem described at the beginning for this figure.This is because all measurement and other signals in controller 2 and pulse generator 5 are still above zero, even if the output current l_LED should already be almost zero.
[0126] The average value control from the previous figure has been transformed into peak current control in Figure 11, or more precisely, control by switching off upon peak current detection, because the buck converter 1' can also be operated continuously. Current peaks can be significantly higher than the corresponding average value. This effect is exploited here and leads to significantly improved system characteristics for another reason as well. The current drop in the buck converter coil is defined by the output voltage between terminals 31 and 61 and is significantly steeper than the current rise at the same point. This is because this current rise is determined by the margin SR, which is always smaller, in many cases even smaller by a factor of the output voltage.Assuming constant levels of the maxima and minima in the buck converter coil current, this results in a turn-off time period for the power transistor 81' that is shorter or even significantly shorter than its turn-on time period, which two time periods follow one another alternately due to clocking. It is not only more advantageous to fix a shorter time period in a circuit, here in the delay circuit 6, but also the one that results from a fixed value, here from the output voltage, which is firmly impressed by a load 8, which preferably consists of at least one series connection of light-emitting diodes. On the other hand, the turn-on time period now controlled by the control is not only the one based on the constant, but ultimately controlled margin SR, but above all the one that directly determines the peak current in the buck converter coil, from which, after smoothing, the required output current l_LED is generated.
[0127] In contrast to all previous circuit diagrams, the input of the detection circuit is not connected directly to the SR margin via lines 51 or 41, but rather to the new reference potential 51'. This is known to be below the SR margin by the voltage across the current measuring resistor 71. The difference between them is only 0.1 V for an output current l_LED = 1 A, over which a sawtooth-shaped ripple with a maximum amplitude of 0.025 V is superimposed, caused by the buck converter's clocking. This ripple is negligible at first glance.
[0128] At second glance, this even becomes a feature. Detection circuits are designed to detect things that are either not there or already there. Because they always have repercussions on the things to be detected, whenever the thing to be detected is barely distinguishable from its absence, linear and quasi-stable intermediate stages can develop that run counter to sharp detection. If a ripple is superimposed on the thing to be detected—here, a voltage—distinguishability is increased, and the detection circuit can report a new "thing detected" state in a discrete-time, synchronous manner to the frequency of this ripple.
[0129] The generalization of this detail means that all the circuit blocks shown in Figures 7 to 12 can be interchanged, although not in their general configuration as shown in Figure 7. In particular, a further developed detection circuit 7 from Figure 9 can also be used in Figures 10 to 12, or an improved voltage shaper 4 from Figure 10 can also be used in Figures 8 and 9.
[0130] FIG. 12 offers another solution, which also works for loads other than LED loads, for the problem in FIG. 10, which is very similar to the present figure. However, the wiring of the measuring resistor 71 in the clocked post-regulator 1' is different from the two previous figures. Here in FIG. 12, it no longer measures the buck converter coil current, but only directly the output current l_LED, because the filter capacitor 141 is connected "below" the measuring resistor, so to speak. As a result, the new reference potential 51 for circuit blocks 2, 3, 4, and 5 migrates to the lower end of the filter capacitor 141, which is now located exactly by the margin SR above the output-side circuit ground 21. In return, the potential of the negative load terminal 61 is now applied unfiltered to the first feedback line 41 for the controller 2 and at the same time applies to the negative measuring input 41 for the comparator 35.
[0131] However, the voltage at this input 41 lacks the ripple resulting from the buck converter coil. In order for the pulse generator 5 to still generate a clock signal from the comparison of the two relatively smooth voltages 22 and 41, it firstly requires positive feedback as in Figure 10 so that the comparator 35 can actually tip somewhere. At a high output level, a constant current flows through the hysteresis resistor 45 from the output 125 toward the reference 22, and at zero potential of the output 125, a reversely constant current flows from the reference to the output along the same path. Secondly, the positive feedback must also be frequency-determining, which is why a clock capacitor 55 is inserted in series with the hysteresis resistor 45 in Figure 12.The positive feedback currents described above generate exponentially ramped voltages at the reference input of comparator 35 due to the new serial clock capacitor 55. This creates intersections with the input voltage 41 at the measurement input, which directly lead to the generation of a clock signal at the output 125 of comparator 35. Due to the clock capacitor 55, negative voltages can arise at the reference input even with a monopolar supply, which can still be correctly evaluated by comparator 35. This avoids the loss of resolution described in Figure 10 for very small reference voltages 22.
[0132] FIG 13 helps with the hitherto unanswered question of which frequencies the two optocouplers OK1 and OK2 are operated at. The feedback on the margin SR is either purely digital, namely whether the margin is currently larger or smaller than a given limit G, or a mixed digital-analog signal, whereby in the time periods when the limit G is exceeded, the extent M of the excess is also fed back. Since the margin SR is almost always superimposed on mains hum, this also defines the frequency for the feedback by OK2, i.e. a feedback frequency, since it is almost always this hum that causes the limit to be exceeded. The specified control method corresponds to that of a two-point controller, which here, however, does not act on a primarily integrating system as is usual, but on a primarily proportional or linear system.To do this, the curve between two points, where a lower point signals "range SR < limit G" and an upper point "range SR > limit G," i.e., the curve of the DIGI signal, must first be low-pass filtered. Strictly speaking, the mean value of the DIGI signal must be determined, which, in relative and thus preferred terms, corresponds to determining its duty cycle. All of these filtering, determinations, or determinations are handled by control circuit 9, located on the input side of the operating device.
[0133] An exception to the feedback frequency is a very low output current l_LED. Then, the mains hum almost completely disappears from the headroom SR, which now lies, so to speak, permanently at the limit G and has also become very soft due to the lack of power input and output, and other artifacts in its voltage can cause its limit G to be exceeded. Although the feedback frequency can be almost arbitrary at very low output currents l_LED, the basic principle of the control process is nevertheless maintained, especially in its purely digital form, which is why this is particularly advantageous. As long as any pulse pattern is fed back via the second optocoupler OK2, the headroom adjustment via the power factor correcting input stage operates as described, only with almost no power. If a sustained high level is fed back, the input stage switches off completely for this duration.Otherwise, the feedback frequency for OK2 is defined by twice the frequency of the available AC mains supply and is typically 100 Hz. The frequency for the setpoint transmission via the first optocoupler OK1, i.e. a setpoint frequency, can either be precisely synchronized to the feedback frequency, although this becomes extremely complicated for very small output currents l_LED. Or it can be selected to be constant and maximally asynchronous, for which the prime number ratio 5 : 7 is particularly suitable. The setpoint frequency can then be 140 Hz, but can also be lower than the feedback frequency. The difference to the feedback frequency should be large enough to avoid beats. However, the physical properties of an optocoupler must also be taken into account, particularly its dynamic and thermal properties.
[0134] The upper trio of graphs shows a purely digital optocoupler transmission at low frequency. At the top, a rectangular input signal OK_IN is shown, and in the middle, a resulting analog output signal OK_COLD is shown in solid lines for a cold optocoupler, or OK_HOT, which is shown in dashed lines and is generated by a warm optocoupler. Third, a reference level is shown in dash-dotted lines, with which the analog output signals are compared to arrive at the digital output signal OK_OUT in the third graph. A comparison of the third and first graphs shows that every optocoupler naturally transmits with a slight delay, but also that the transmitted pulses become shorter the warmer the optocoupler becomes. If a mark-to-space ratio contains the signal's information, as in purely digital transmission, the observed pulse shortening is a clear corruption of information.
[0135] The lower trio of graphs shows the same situation at five times the frequency. While the pulse shortening with a warm optocoupler is absolutely the same per pulse, it distorts the signal five times more because the pulses themselves are significantly shorter. This effect is even visible to the naked eye. It follows that the setpoint frequency should be chosen as low as possible. The feedback frequency of typically 100 Hz already represents something of an ideal point, which is why the setpoint frequency should not be much higher, for example, at 140 Hz. A further advantage of such low feedback frequencies is the particularly high signal resolution in the control circuit 9 on the input side, which operates with a constant and, above all, limited clock frequency.Since in purely digital transmission the value of the transmitted signal must be obtained from the intervals between its edges, the input-side resolution is higher the further apart the edges are in time, i.e. the lower the setpoint frequency is.
[0136] When controlling the headroom SR, the inherent slowness due to such low setpoint frequencies is irrelevant, since the input-side controller (not described further for the sake of general familiarity) for a first clocked converter intended to operate as a power factor correcting input stage must be set even slower. Otherwise, the power factor correction will not work. However, the lower limit for the setpoint frequency, at which the setpoint for the output current l_LED is transmitted via the first optocoupler OK1, is determined by low-pass filter 3. This filter should filter as well as possible without becoming arbitrarily sluggish, because it should also be able to transmit setpoint changes, which are particularly common in LiFi. Therefore, this filter 3 is designed as a two-stage low-pass filter, as shown unchanged in all Figures 7 to 12.This is because any residual ripples in the setpoint signal 63 on the output side of the specified operating device are passed on to the output current l_LED almost undamped by the very fast and direct control of post-regulator 1 or 1' and must not exceed small percentages of the mean value of the same current. The same percentages also apply to the ripple of the setpoint signal 63, thus limiting the setpoint frequency downwards.
[0137] All of this results in the optimum transmission of the setpoint for the output current l_LED, for example, at 140 Hz or 145 Hz. The setpoint frequency preferably ranges from 30 Hz to 1 kHz, and is particularly advantageous from 60 Hz to 170 Hz.
[0138] The following describes the fine adjustment of the headroom voltage depending on the current internal condition of the device. This fine adjustment is used in all of the above-mentioned embodiments. The current internal condition of the device depends primarily on the thermal situation within the device, which in turn depends on the output current and the output voltage. However, the current internal condition of the device can also depend on the number of operating hours and the associated aging of critical components (e.g., electrolytic capacitors).
[0139] If the power loss in the post-regulator is below the critical limit, e.g. due to an output current of less than 90% of the maximum output current, or if the temperature, which is measured by the microcontroller using an NTC resistor, at the critical components of the post-regulator (this is usually the transistor connected as a linear regulator when the post-regulator is designed as a linear regulator, or the step-down converter coil or buck choke when the post-regulator is designed as a clocked step-down converter) is below the critical limit of 110% to 120°C, the microcontroller increases the margin voltage by means of the intermediate circuit voltage in the fine-tuning range up to an upper value, so that the power loss in the circuit arrangement and especially in the post-regulator is increased up to a predetermined limit in order to minimize the ripple of the output current, so that the following dimensioning rule applies to the ripple of the output current:
[0140] The amplitude modulation of the output current = (maximum output current - minimum output current) / (maximum output current + minimum output current) should be less than 3%.
[0141] If the power loss in the post-regulator is above the critical limit, e.g. due to an output current greater than 90% of the maximum output current, or if the temperature, which is measured by the microcontroller using an NTC resistor, at the critical components of the post-regulator (this is usually the transistor connected as a linear regulator when the post-regulator is designed as a linear regulator 1 or the buck converter coil when the post-regulator is designed as a clocked buck converter 1') is above the critical limit of 110°C to 120°C, the microcontroller reduces the margin voltage by means of the intermediate circuit voltage in the fine-tuning range to a lower value, so that the power loss in the circuit arrangement and especially in the post-regulator is reduced in order to increase the ripple of the output current.so that the following dimensioning rule applies to the ripple of the output current: The amplitude modulation of the output current = (maximum value output current - minimum value output current) / (maximum value output current + minimum value output current) should be greater than 5% and less than 10%.
[0142] If the number of operating hours is above the critical limit of 40,000 hours and the power loss in the post-regulator is above the critical limit, e.g.by an output current greater than 90% of the maximum output current, or if the temperature, which is measured by the microcontroller with an NTC resistor, at the critical components of the post-regulator (this is usually the transistor connected as a linear regulator when the post-regulator is designed as a linear regulator or the buck converter choke when the post-regulator is designed as a clocked buck converter) is above the critical limit of 110% to 120°C, the microcontroller reduces the headroom voltage by means of the intermediate circuit voltage in the fine-tuning range to a lower value, so that the power loss in the circuit arrangement and especially in the post-regulator is reduced in order to increase the ripple of the output current, so that the following dimensioning rule applies to the ripple of the output current:.
[0143] The amplitude modulation of the output current = (maximum output current - minimum output current) / (maximum output current + minimum output current) should be greater than 10% and less than 15%.
[0144] LIST OF REFERENCE SYMBOLS (CONTAINS ONLY PURE NUMBERS) linear post-regulator ' clocked post-regulator 1 intermediate circuit capacitor 1 output-side circuit ground 1 positive pole of an intermediate circuit voltage 1 first feedback line 1 second feedback line 1 ' new reference potential on a feedback line or new
[0145] Feedback line 1 Negative pole of an output voltage or negative connection for a connectable load 1 Current measuring resistor 1 Linear amplifying transistor T Switched power transistor 1 Gate or gate line or gate voltage of the linear amplifying
[0146] Transistor 1 ' same for the clocked power transistor 01 Protection transistor 0T Gate driver 11 Base series resistor 21 Buck converter coil 31 Freewheeling diode 41 Filter capacitor 51 Starting resistor 61 Positive pole of the output voltage or positive connection for a connectable load PI controller 2 Operational amplifier 2 Controller output or controller output voltage or reference voltage 2 Integrating capacitor 2 Negative feedback resistor 2 Feedback resistor 2 First small inductance 2 Second small inductance 2 Possible coupling of the small inductances by a common ferrite bead Two-stage low-pass filter 3 Upper voltage divider resistor 3 Lower voltage divider resistor 3 Filter capacitor 3 Intermediate resistor 3 Smoothing capacitor 3 Setpoint voltage shaper
[0147] Reception line
[0148] Output signal of the voltage shaper or input signal for the
[0149] Low-pass filter
[0150] Zener diode
[0151] pull-down resistor
[0152] Series resistance
[0153] Flux diode
[0154] Decoupling resistor controllable Zener diode (e.g. TL431) lower voltage divider resistor upper voltage divider resistor
[0155] Pulse transistor
[0156] Pulse generator
[0157] Adaptation resistance
[0158] Input resistance
[0159] Comparator or second operational amplifier
[0160] Hysteresis resistance
[0161] Frequency capacitor
[0162] Level-shift transistor
[0163] Driver resistance
[0164] Extinguishing resistance
[0165] Emitter of the level shift transistor or its line second matching resistor
[0166] Output of the comparator
[0167] Delay circuit
[0168] Series resistance or delay resistance
[0169] discharge diode
[0170] Delay capacitor
[0171] pull-down resistor
[0172] Detection circuit
[0173] Transmission line inverting transistor
[0174] Softening resistance
[0175] Base resistor subtracting transistor of a differential amplifier adding transistor of the differential amplifier
[0176] Current source resistance
[0177] Terminating resistor
[0178] Protective resistor
[0179] Trim resistor
[0180] Reference resistor subtracting series resistor adding series resistor electrical load control circuit on the input side of an operating device
[0181] Series resistor for the transmitting diode of a first optocoupler
[0182] Series resistor for the receiving transistor of a second optocoupler
[0183] Parallel resistance
[0184] Low-pass resistor
[0185] Low-pass capacitor
[0186] Isolation barrier of the first switch-mode converter
[0187] Isolation transformer
[0188] Rectifying diode
[0189] Terminating capacitor
[0190] Mains rectifier
[0191] Radio interference filter
[0192] Mains input actively switched power transistor or inverter switch
[0193] (symbolically represented)
[0194] Control line of the inverter switch
[0195] Control amplifier for the inverter switch pulsed output voltage from the control circuit 9
Claims
PATENT CLAIMS Circuit arrangement comprising: - an input (70) for inputting a network power, and - an output with a positive output terminal (161 ) and a negative output terminal (61 ) for operating a load (8) with an output current (l_LED) and an associated output voltage (V_LED), - a first clocked converter (10) with a SELV transistor (99) between its primary and secondary sides to generate a converter output current in order to produce an intermediate circuit voltage (ICV) with a ripple on the secondary side and to regulate to an adjustable mean value of the intermediate circuit voltage, - wherein the circuit arrangement is set up to control the first clocked converter in such a way that the mean value of the intermediate circuit voltage (VZ) can follow the output voltage (V_LED) with a distance corresponding to the clearance voltage (SR), - a post-regulator (1 , T) connected downstream of the first clocked converter for controlling the ripple on the intermediate circuit voltage (VZ) generated by the first clocked converter, - a first optocoupler (OK1 ) for transmitting information from the primary side to the secondary side via a current setpoint of the output current (l_LED), -a second optocoupler (OK2) for transmitting information about a clearance voltage (SR) from the secondary side to the primary side to enable control of the intermediate circuit voltage (VZ) by the first clocked converter (10), wherein the clearance voltage (SR) corresponds to the potential difference between the negative output terminal (61 ) and a circuit ground (21 ), - exactly one control circuit (9) on the primary side for controlling the circuit arrangement, - wherein the control circuit (9) is designed to regulate the clearance voltage (SR) to the smallest possible value in order to minimize power loss in the circuit arrangement and especially during operation to set the post-regulator (1, T) to a small value and simultaneously keep the ripple of the output current (l_LED) as low as possible. Circuit arrangement according to claim 1, characterized in that the control circuit (9) regulates the value of the clearance voltage (SR) and thereby fine-tunes it depending on the output current (l_LED) and / or the output voltage (V_LED) and / or the temperature of the circuit components and / or the number of operating hours, such that the clearance voltage (SR) is increased in the fine-tuning range up to an upper value, so that the power dissipation in the circuit arrangement and especially in the post-regulator (1) is increased up to a predetermined limit in order to minimize the ripple of the output current (l_LED).Circuit arrangement according to claim 1, characterized in that the control circuit (9) regulates the value of the clearance voltage (SR) and thereby fine-tunes it depending on the output current (l_LED) and / or the output voltage (V_LED) and / or the temperature of the circuit components and / or the number of operating hours in such a way that the clearance voltage (SR) is reduced to a lower value in the fine-tuning range, so that the ripple is increased to a predetermined limit in order to reduce the power loss in the circuit arrangement and especially in the post-regulator (1) to an even smaller value.Circuit arrangement according to claims 1 to 3 characterized in that the control circuit (9) is configured to operate the first optocoupler (OK1) purely digitally and to operate the second optocoupler (OK2) either in a mixed mode, wherein the second optocoupler (OK2) transmits a periodically recurring analog signal (M) with gaps between the sections of the analog signal (M) which are greater than zero, or the control circuit (9) is configured to operate the second optocoupler (OK2) purely digitally, wherein the second optocoupler (OK2) transmits a digital signal (DIGI). which has a low level corresponding to the gaps in the analog signal (M) in time, and otherwise has a high level. Circuit arrangement according to claims 1 to 4, characterized in that the first clocked converter additionally functions as a power factor corrector. Circuit arrangement according to any one of claims 1 to 5, characterized in that a microcontroller is part of the control circuit (9). Circuit arrangement according to any one of the preceding claims 1 to 6, characterized in that the post-regulator is a linear regulator. Circuit arrangement according to any one of the preceding claims 1 to 6, characterized in that the post-regulator (T) is a second clocked converter in the form of a buck converter. Circuit arrangement according to claim 8, characterized in that the buck converter is controlled by means of hysteresis band control.Circuit arrangement according to claim 8, characterized in that the buck converter is controlled by constant-time switch-off control after peak current detection. Circuit arrangement according to any one of the preceding claims, characterized in that the post-regulator (1, T) is configured to control the output current (l_LED) to a setpoint without transmitting any actual value of the output current (l_LED) back to the control circuit (9). Circuit arrangement according to any one of the preceding claims, characterized in that the control circuit (9) is configured to transmit the setpoint purely digitally to the post-regulator (1, T) via the first optocoupler (OK1). Circuit arrangement according to any one of the preceding claims, characterized in that the control circuit (9) is configured to transmit the setpoint as the duty cycle of a pulse-width modulation via the first optocoupler (OK1) to the post-regulator (1, T), wherein the associated frequency is a setpoint frequency and is 30 Hz to 1 kHz, advantageously 60 Hz to 170 Hz, and particularly advantageously 140 Hz or 145 Hz. Circuit arrangement according to one of the preceding claims, characterized in that it is configured to measure the clearance voltage (SR) in the post-regulator (1, 1') and to evaluate the measurement result by a detection circuit (7), wherein the detection circuit (7) outputs a signal (M) to the second optocoupler (OK2) as long as the clearance voltage (SR) exceeds a limit (G), wherein the signal (M) includes the amount by which the limit (G) is exceeded by the clearance voltage (SR), and wherein the limit (G) is stored in the detection circuit (7). Control method for a circuit arrangement according to one of the preceding claims, characterized by the following steps: - Receiving the signal (M) from the control circuit (9), - Rules of the first clocked converter to an intermediate circuit voltage (VZ) according to the value of the signal (M) - Transmission of the setpoint of the output current (l_LED) via the first optocoupler to the post-regulator (1 ), - Control of the ripple of the intermediate circuit voltage (VZ) and control of the output current (l_LED) of the circuit arrangement by the post-regulator (1 ), - Transmission of the signal (M) from the detection circuit (7) to the control circuit (9). Control method according to claim 15, characterized in that the minimum of the margin voltage is 0.0V to 0.5V and particularly advantageously 0.0V to 0.2V, and that the margin voltage range is 1V to 2.5V and particularly advantageously 1.3V to 1.8V when the post-regulator (1, 1') is a linear regulator. Control method according to claim 15, characterized in that the minimum of the margin voltage is 0.0V to 10V and particularly advantageously 0.0V to 2.5V, and that the margin voltage range is 2V to 12V. and particularly advantageously 5V...8V when the post-regulator (1, T) is a buck converter. Control method according to one of claims 15 to 17, characterized in that the control circuit (9) measures a temperature of the circuit arrangement and reduces the DC link voltage (VZ) when the measured temperature exceeds a predetermined temperature.
