WO2023134898A1

WO2023134898A1 - Output switching stage with glow avoidance

Info

Publication number: WO2023134898A1
Application number: PCT/EP2022/081042
Authority: WO
Inventors: Felix Franck; Markus Heckmann
Original assignee: Osram Gmbh
Priority date: 2022-01-17
Filing date: 2022-11-08
Publication date: 2023-07-20
Also published as: DE102022200431A1

Abstract

The invention relates to a circuit arrangement for operating at least one LED, comprising an input for inputting an electrical system AC voltage, an inverse buck converter for setting a suitable operating current for the at least one LED, the inverse buck converter comprising a first and a second glow avoiding diode for avoiding capacitive displacement currents with respect to a protective ground in the switched-off state of the circuit arrangement. This measure advantageously ensures that glow effects are suppressed when the circuit arrangement is not operational and is still connected to the electrical system AC voltage, since the effects of parasitic capacitances that are responsible for said glow effects are prevented owing to the novel topology of the circuit arrangement.

Description

OUTPUT STAGE WITH GLOW PREVENTION

DESCRIPTION

technical field

A circuit topology for a step-down converter is specified, which is provided as a clocked output switching stage of an operating device that is set up in particular to supply light-emitting diodes in lighting technology. The output switching stage reliably prevents the LEDs from glowing when the operating device is switched off, even if the operating device is only switched off via its neutral conductor or via an external data interface. In the former, the phase of an AC mains supply is still connected to the input of the operating device, in the latter even both lines of the same, but the operating device clocks - entered via the data interface - not or not nearly. Without the additional glow prevention function, the LEDs cannot go completely dark in either case, even if they are supposed to.

background

Light-emitting diodes have an extremely high resistance due to their strongly non-linear characteristic curve at low power. However, this property does not prevent them from generating light from their comparatively tiny forward current, which, according to the flat branch of their nonlinear characteristic curve, is present up to relatively high forward voltages just below their nominal voltage forward current is correspondingly very dark, but still clearly visible in comparison to absolute darkness. This means that the smallest currents that are still emitted by an operating device that is intended for light-emitting diodes and is actually switched off can cause the light-emitting diodes connected to them to light up a little or just let them glow. From an electrical or control point of view, such an LED control gear works on the flat branch of a light-emitting diode characteristic curve in no-load operation, which is a difficult task, especially for clocked switching stages, which are generally used for low-loss energy conversion between the mains supply input and the output and which are actually supposed to work as a current source against a forward voltage reflected back from the output represents a solving task. Irrespective of this, even capacitive displacement currents, which are caused by the mains rectifier at the input of an LED control gear and reach the light-emitting diodes via parasitic capacitances, are sufficient to make them glow as long as at least one phase of an AC mains supply remains connected to the input of the LED control gear.

Large-area light arrangements with many individual light-emitting diodes are required for the illumination of offices or shops, which also have to be well cooled according to their number and their resulting electrical output. Therefore, the modules for the light-emitting diodes, the LED modules, are often built on so-called metal-core circuit boards, in which only a relatively thin layer provides insulation between the connections of the light-emitting diodes or LEDs to each other and to the metal core. The metal core is often - again for reasons of its cooling - connected directly to the lamp or even directly to the building and is therefore usually electrically at the potential of a protective earth. The construction just described also unintentionally includes very high parasitic capacitances between each of the LED connections and such a protective earth, to which the housing and the radio interference filter of an LED control gear can often also be connected. Such potential connections complete radio interference suppression, because they can be useful, for example, for shielding non-conducted radio interference. Because very often lighting constructions are originally intended for fluorescent lamps were to modernize. The mechanical and safety-related conditions often remain unchanged.

This can even extend into the basic design of LED drivers intended to replace the "ballasts" or fluorescent lamp drivers. In such classic control gear, there is often no galvanic isolation between the output to the lamp and the input for the AC mains supply. Such a structure usually improves the efficiency of an LED control gear. However, a DC voltage is expected at the output of the same, the ripple of which must be significantly smaller than that of the envelope of an AC voltage that can be emitted from fluorescent lamps so that these lamps light up evenly. Because in contrast to the often very sluggish classic lamps, which can even filter out an unsmoothed 100Hz mains hum like e.g. all incandescent lamps, the light emitting diodes are the fastest possible semiconductors and therefore the fastest light sources of all. It is not for nothing that infrared light-emitting diodes function as transmitters for almost all interior remote controls. In other words, every ripple in an output voltage of an LED control gear is made visible by the light-emitting diodes connected to it, which no classic lamp can do as long as it is supplied by a classic but electronic control gear that has appreciable alternating components in the output signal.

The input stage of a classic electronic control gear, which mostly or at least consists of a radio interference filter with a fuse, a rectifying circuit as a mains rectifier and a power factor corrector, is therefore adopted. This module is basically terminated by a large so-called intermediate circuit capacitor, which is connected in parallel to its DC voltage output poles. The rectification circuit is either a Graetz or a Delon bridge circuit. If the capacitive branch of the latter circuit is connected to the neutral conductor of an AC mains supply, the rare ideal case of quiescent potentials is present at both DC voltage output poles of the input stage, although the entire operating device under consideration with its AC mains supply is galvanically connected. In this case, the negative pole of the intermediate circuit capacitor is at a potential around at least the peak value of the mains AC voltage below the neutral conductor and the positive pole is at a potential around the same value above. At least as often, however, the capacitive branch of a Delon bridge can be connected - twisted, so to speak - to the phase of the mains AC voltage, so that all potentials of the input stage move with the voltage on this phase. This case is slightly weaker in all the rectifier circuits for the AC voltage mains supply, which consist of a Graetz bridge, as is always the case in the above. Irrespective of the distribution of the two bridge branches to phase and neutral conductor, the negative pole of the intermediate circuit capacitor is always drawn to negative voltages when the phase has a negative voltage compared to its neutral conductor. It is precisely this and the worsening caused by a Delon bridge that is connected the wrong way round that is the cause of the glowing of light-emitting diodes on an LED control gear that is actually switched off. This glowing is avoided by the circuit topology given here. At the same point, it also becomes clear why the galvanic disconnection of the neutral conductor with the phase still connected to an LED control gear cannot prevent glowing: The negative pole of the intermediate circuit capacitor coming from the connected LED module is connected with a very low resistance, e.g. to a protective earth. The connected phase periodically draws potentials below it, which means that the above-mentioned capacitive displacement currents, which cause the LEDs to glow, can continue to flow.

Because the intermediate circuit capacitor is supplied by an AC line voltage, which has been rectified in whatever way, and because it does not have an infinite capacity, its voltage, the so-called intermediate circuit voltage, is always superimposed on a mains ripple with double the supply mains frequency. This also applies if there is a power factor corrector between the rectifier circuit for the AC voltage mains supply and the intermediate circuit capacitor. This mains hum is almost never translated into a luminous flux hum by all classic lamps, but by light-emitting diodes it is. That is why it is used in almost every classic electronic control gear existing lamp inverters connected to the intermediate circuit capacitor are replaced by a linearly controlled or clocked DC-DC converter in order to form an LED control gear together with the almost unchanged input stage. The main task of this new DC/DC converter in LED control gear, which is often also referred to as a post-regulator or output switching stage, is to regulate the mains hum while at the same time maintaining the required LED current as precisely as possible. For this purpose, a special step-down converter is specified here as a clocked output switching stage.

While it was physically impossible to avoid a significant shift in the color locus when dimming fluorescent lamps or incandescent lamps, this avoidance is quite possible with light-emitting diodes as light sources. To do this, however, their current must be kept constant as long as they are lit. Instead of using a uniform current reduction, they can be dimmed alternatively or simultaneously using an output PWM, which in turn clocks the light-emitting diodes connected to an LED control gear. The clock frequency of this output PWM differs from the clock frequency of a clocked output switching stage and also usually from that of a power factor corrector. In particular, it, which is also referred to below as the dimming frequency, is lower than the other two clock frequencies. For such independent clocking of the LED as such, there is often a further electronic so-called dimmer switch or PWM switch in series with the negative pole of the output of an LED operating device. In EP-3-376-834-B1, a freewheeling diode is additionally connected from the negative to the positive pole of an output of an LED operating device that has been expanded by a dimmer switch, in order to block negative voltages across the switched-off light-emitting diodes caused by inductances.

A type of activity against the undesired glowing of actually switched off light-emitting diodes takes place on the LED module. DE-10-2016-119- 448-A1 proposes connecting a capacitor in parallel with each LED or at least each series connection of 2 LEDs, the capacitance of which is significantly greater than the parasitic capacitance between each LED connection and the board substrate is. As an alternative to the capacitors, high-impedance resistors can also be connected in parallel, which prevent glowing more reliably, but generate additional losses, which means that the LED module efficiency deteriorates. The cooling of the light-emitting diodes, which is already critical anyway, is additionally burdened by the waste heat of the resistors.

In contrast, DE-10-2015-106-268-A1 proposes expanding a clocked output switching stage with a diode between it and the intermediate circuit capacitor that supplies it. This additional diode is polarized in such a way that it allows a current flow from the intermediate circuit capacitor into the output switching stage, but no current flow from the positive pole of the output of the LED control gear back into the intermediate circuit capacitor. This capacitor also connects the negative pole of the entire LED control gear to this additional diode with low resistance, which is why it is assumed that this one additional diode is sufficient to avoid glowing.

Even in the case of an LED control gear with an output that is galvanically isolated from the mains supply input, light-emitting diodes that are actually switched off can glow. This is due to the y-capacitor, which is mandatory for radio interference suppression of the same control gear and which bridges the isolation barrier to compensate for parasitic capacitances of the isolating transformer. By default, as also shown in EP-3-376-834-B1 cited above, this y-capacitor is connected to the negative pole of the intermediate circuit voltage. DE-10-2019-207-182-A1, on the other hand, proposes connecting the negative pole of the output of an LED operating device via this y-capacitor directly to the pole of the mains supply input, to which the neutral conductor of the AC mains supply is to be connected. The y-capacitor is connected before the mains rectifier instead of more or less directly in parallel with the isolating transformer.

With LED control gear with a non-insulated output, it can make sense to use the dimmer switch in a glow avoidance. For this purpose, two flow diodes or functionally identical components such as synchronously controlled MOS field effect transistors are additionally required, as e.g of the LED control gear can be switched on in the same way. For effective avoidance of glowing, low-frequency alternating current must be blocked equally at both poles of the output of an appropriately equipped LED control gear. If this low-frequency alternating current has purely capacitive causes, as is the case here with glowing, it is sufficient to have just one in one of the supply lines to the LED module to block direction from it. This direction to be blocked is logically the opposite direction to the direction of the operating current desired for the light-emitting diodes from their control gear, and the supply line is the positive pole at the output of the control gear or the anode connection of the LED for reasons of accessibility for an otherwise necessary activation of an active circuit breaker that would otherwise be necessary module. For reliable glow avoidance, however, at least one of the supply lines to the LED module must also be able to be blocked for both current directions, which requires at least one actively controllable circuit breaker which - when switched on - carries the desired operating current and - when switched off - a possible one Glow current blocked in the same direction. This actively controllable circuit breaker can be formed by the dimmer switch. A glow current in the opposite direction, which can often flow through an inverse diode or through a base-collector diode of the dimmer switch, blocks a diode whose cathode is connected to the working electrode of the dimmer switch. Because this diode carries the operating current of the dimmer switch and thus the operating current required by the control gear for the LED module, it is a flow diode. In order to be able to control the dimmer switch easily and effectively, the at least one bidirectionally lockable supply line is in most cases the negative pole at the output of the control gear or the cathode connection of the LED module.

The series combination shown in DE-10-2015-202-370-A1 consisting of flux diode 15 and dimmer switch S1 can also be taken over by an IGBT as a dimmer switch with the same function, which then no longer requires an extra flux diode because it is integrated in it. A bipolar transistor used as a dimmer switch requires an extra flow diode, because its base-collector diode can act as an inverse diode together with a low-impedance base control. Task

It is the object of the invention to specify an LED operating device with simplified glow avoidance. This is because the LED operating devices described so far comprise at least four power switching stages in series, resulting in a disproportionate increase in the cost of materials, radio interference, size, price and losses. Starting from the AC mains supply, there is a radio interference filter, a rectification circuit as a mains rectifier, a power factor corrector that generates an intermediate circuit voltage, a post-regulator as an output switching stage, a dimmer switch and finally glow prevention. If the output is to be isolated from the mains supply input, either an appropriate further switching stage is placed between the power factor corrector and the post-regulator, or the power factor corrector is extended by a transformer, or the post-regulator is extended to an isolating output switching stage. Here, on the other hand, the output switching stage is to be combined with the dimmer switch and glow prevention. To minimize losses, the improved output switching stage is operated in a clocked manner. Since the intermediate circuit voltage can easily be set higher than the operating voltage or forward voltage of all LED modules that can be connected to the LED operating device under consideration, a step-down converter is preferably a suitable starting point for the improved output switching stage to be specified.

Presentation of the invention

The object is achieved according to the invention with a circuit arrangement for operating at least one LED, having an input for inputting an AC line voltage, a rectifying circuit for rectifying the AC line voltage into a pulsating DC voltage, a power factor correction circuit for setting the line power factor a value that satisfies local regulations, an inverse step-down converter for setting a suitable operating current for the at least one LED, the inverse step-down converter having a first and a second glow prevention diode to prevent capacitive displacement currents with respect to protective earth when the circuit arrangement is switched off, a first and a second output connection for operating a load, the first or the second glow prevention diode being connected in series with the first output connection, and the second or the first glow prevention diode being connected in series with a switching transistor of the inverse buck converter, with an input voltage of the inverse buck converter compared to its output voltage between the first and the second output port differs by more than 5% over more than 95% of all operating times. This measure advantageously ensures suppression of glow effects when the circuit arrangement is not running if it is still connected to the mains AC voltage, since the effects of parasitic capacitances, which are responsible for these glow effects, are suppressed due to the new topology of the circuit arrangement.

In a particularly preferred embodiment, the circuit arrangement has a rectifier circuit for rectifying the mains AC voltage into a pulsating DC voltage and a power factor correction circuit following the rectifier circuit for setting the mains power factor to a value that satisfies local regulations. This is an efficient method of ensuring compliance with current grid load regulations in many countries.

In another embodiment, the power factor correction circuit has a wide output voltage range in which the output voltage of the power factor correction circuit can become smaller and larger than the pulsating DC voltage. This ensures an advantageous circuit arrangement that can operate a wide variety of loads with a large output voltage range. In a preferred embodiment, both connections of a freewheeling diode of the inverse buck converter are directly connected to one of the glow prevention diodes in each case in order to decouple freewheeling phases of the buck converter from activities of the rectifier circuit. This measure has the advantage that the circuit arrangement can work more efficiently since the freewheeling phases can no longer be disturbed by rectification effects of the rectification circuit.

In a further preferred embodiment, the anode of the freewheeling diode is connected to the anode of the second corona avoidance diode in order to be able to switch a current in a step-down converter coil in the best possible way and at the same time reliably block it in the opposite direction. This measure also advantageously increases the converter efficiency and operational reliability.

In a further preferred embodiment, the cathode of the freewheeling diode is connected to the cathode of the first corona avoidance diode in order to shorten a freewheeling path and thus increase the efficiency of the circuit arrangement. As already described, the efficiency of the converter is advantageously further increased by optimizing the freewheeling path.

In a further preferred embodiment, a first resonance capacitor is connected in parallel with the buck converter coil in order to reduce the turn-off losses of a clocking power transistor of the inverse buck converter. This measure also advantageously increases the efficiency of the converter and thus of the entire circuit arrangement.

In a further preferred embodiment, the second glow prevention diode is connected in parallel with a second resonance capacitor in order to reduce the switch-on losses of the clocking power transistor. This measure also increases the efficiency of the converter and thus of the circuit arrangement, in that the turn-on losses of the switching transistor are advantageously reduced.

Further preferred embodiments can be found in the dependent claims and the disclosure as a whole, whereby the presentation does not always differentiate in detail between device and use aspects becomes; at least implicitly, the disclosure is to be read with regard to all claim categories.

Brief description of the drawings

Further advantages, features and details of the invention result from the following description of exemplary embodiments and from the drawings, in which identical or functionally identical elements are provided with identical reference symbols. show:

1a shows a basic circuit of a non-insulating LED operating device with an LED module connected to it, with a step-up converter as a power factor corrector and with a synchronously controlled step-down converter as an output switching stage (prior art),

1b shows the course of the currents in the three connecting lines, including a connection between the module heat sink and protective earth, between the LED control gear and module (prior art) and the potential course of the negative pole of the intermediate circuit voltage,

2a shows the basic circuit from FIG. 1a with an additional dimmer switch and two flow diodes as glow prevention between the LED operating device and the LED module (prior art),

2b shows the same curves as in FIG. 1b, only for the improvement of the circuit according to FIG. 2a (prior art),

3 shows an initial situation for glow avoidance with a step-down converter as a clocked output switching stage, dimmer switch and two flow diodes (prior art),

4 shows an improved clocked output switching stage as a step-down converter with integrated glow prevention,

5a shows some time curves from the circuit according to FIG. 4, 5b shows a zoomed section from the time curves according to FIG. 5a,

6 shows a further improved clocked output switching stage as a step-down converter with integrated glow prevention,

7 shows a zoomed section of the time curves in the circuit according to FIG. 6,

8a shows a clocked output switching stage with glow prevention for two outputs of an LED operating device with a SEPIC power factor corrector,

8b output switching stage according to FIG. 8a with shortened freewheeling paths,

8c output switching stage according to FIG. 8b with a common upper flow diode.

Preferred embodiment of the invention

Fig. 1a shows the situation of a non-isolating LED driver connected both to its LED module on the right side of the circuit diagram and to an AC mains voltage V1 on the left side. Using all switched off MOS field effect transistors M3, M1 and M2, it is demonstrated that this operating device has just been switched off via an external data interface (not shown), that all light-emitting diodes D1 to D30 should therefore be completely dark. One of the situations already mentioned at the outset is therefore present, namely with two AC voltage mains lines connected to the LED operating device, but the operating device is switched off due to appropriate control.

The two lines coming from the mains AC voltage V1 are routed via a very low-impedance radio interference filter (not shown) and connected to the two midpoints of a Grätz bridge circuit consisting of the four rectifier diodes D31 to D34. The Graetz bridge is the world's most common full-wave rectifier circuit for AC power supplies and is therefore always used here as a basis. Both mixed-polarity bridge arm midpoints together form their input, the neutral conductor N of the AC mains voltage V1 is connected to the one between D31 and D33, the cathode star from D31 and D32 forms the positive pole VJn of their output and the anode star from D33 and D34 forms the negative pole GND, the at the same time lies on a circuit ground. Between these two poles, a high-impedance resistor R1 ensures that an operating device that is switched off but not disconnected behaves similarly in a simulation as it does in reality. It simulates the low load on the rectifier circuit on the mains side, for example due to the standby consumption of a real operating device, which is intended to prevent the connected LEDs from glowing, and discharges all the capacitances in the vicinity of the rectifier circuit, including its terminating capacitor C1. The voltage VJn applied across it with respect to GND corresponds to the rectified mains AC voltage and is fed to a power factor corrector, for example in the form of a step-up converter. This consists of the components L2, M3 and D36, is terminated by an intermediate circuit capacitor C36 and does not need to be explained in more detail because it is widely known. Only the terminating capacitor C1 of the rectifier circuit at the input of the step-up converter, the high-frequency repercussions of this

The fact that the power factor corrector bypasses the rectifier during normal operation should be mentioned because it is not absolutely necessary according to converter theory, but it is essential for protecting the rectifier circuit on the mains side.

The intermediate circuit voltage is stored in C36 with its positive pole V_bus and its negative pole GND, which is also the negative pole of the rectifier circuit, the reference potential of the step-up converter transistor M3 and thus a circuit ground. In the following, all potentials are related to this circuit ground in order to specify voltages of the same name. Due to the step-up converter, the value of the intermediate circuit voltage V_bus is generally higher than the time value of the rectified mains AC voltage VJn, in most cases or periods even higher than the peak value of the AC mains supply, i.e. higher than the mains peak voltage. Only with very small capacitances of the intermediate circuit capacitor C36 or with a correspondingly high load on the same and in the temporal vicinity of the mains AC voltage zero crossings, it can occur that minimum values of the intermediate circuit voltage V_bus are below the maxima of the rectified mains AC voltage VJn occurring before and after. In between, the intermediate circuit voltage V_bus always reaches maximum values above the mains peak voltage in regular operation. The mains hum occurs between the minimum and maximum values, which basically has twice the mains frequency due to the full-wave rectification that has already taken place. Half-wave mains rectifiers are hardly ever used because of their poor and, in particular, one-sided utilization of the AC mains supply.

In order to minimize the influence of this mains ripple in the intermediate circuit voltage V_bus on the current flow of the connected LED module, an additional switching stage and/or a linear regulator and/or a filter must be installed between the LED module and the intermediate circuit capacitor C36. The circuit dimensioning and specification of its environment often ensures that the forward or operating voltage of an LED module to be connected, here essentially consisting of a series connection of light-emitting diodes D1 to D30, is lower than the intermediate circuit voltage V_bus in all positions and at all times of a mains voltage period . In these cases, a step-down converter is the most sensible topology to choose for a switching stage on the intermediate circuit capacitor C36 and upstream of the connected LED module in the direction of power flow, which is why only step-down converters are described below as output switching stages. The one here consists of the electronic switches or transistors M1, M2 and the buck converter coil or "buck choke" L1. Switch M2, the presence of which indicates a synchronously controlled step-down converter or "synchronous buck", can also be designed as a non-controllable freewheeling diode, which results in a normal step-down converter. Regardless of normal or synchronous control, this step-down converter, as a clocked output switching stage, regulates the output current for the light-emitting diodes connected to it to a desired value, with the mains ripple or another voltage difference representing the disturbance to be corrected. The input of each buck converter must be capacitively terminated, which is given here by the position of the intermediate circuit capacitor C36. This forms the termination of the output of the power factor-correcting step-up converter and the termination of the input of the step-down converter at the same time. In contrast, a storage capacitor C3 is not absolutely necessary at the output of the step-down converter and thus of the entire operating device. However, the storage capacitor C3 reduces the current ripple for the light-emitting diodes considerably, since it diverts at least part of the alternating component of the current in the buck converter coil L1, whose fundamental frequency corresponds to the clock frequency of the output switching stage, to the LED module. The combination of C3 with the lead resistances R_AN and R_KA, in which the differential resistances of all light-emitting diodes D1 to D30 are also subsumed, results in a first-order low-pass filter for the LED current. R_AN defines the outgoing line to the anode connection of the LED module at a positive output LED+ of the control gear coming from the upper end of the storage capacitor C3, and R_KA defines the return line from the cathode connection of the LED module to a negative output LED- of its control gear. This negative output - the circuit ground - is also connected to the lower ends of the storage capacitor C3 and the intermediate storage capacitor C36 and is at ground potential GND. The storage capacitor C3 is particularly important in the event of a sudden load shedding during otherwise normal operation with maximum output current or maximum output power. If it were missing, in this case there would only be an open output with the buck converter coil L1 in series with it. Because a current of approximately the same value as the previous output current is stored in it, theoretically infinitely high output voltages would arise at the output, and so quickly that no control circuit could take countermeasures in good time. C3 slows this voltage rise during load shedding to slopes that an output voltage control loop can still track and regulate. Therefore, although it is only optional from the point of view of the converter theory, the storage capacitor C3 is indispensable.

The middle line on the right-hand side of the circuit diagram of FIG. 1a is not intended or should be provided in the circuit layout, but results from the Mechanical construction of an LED module, especially if it is built on a metal core circuit board for cooling reasons. This also includes the fact that all capacitance symbols C5 to C35 leading to the middle line do not represent capacitors to be mounted discretely, but the parasitic capacitances of the LED connections and connection lines towards the metal core with a thin insulating layer as a dielectric. The heatsink line in the middle thus represents the metal core of a corresponding circuit board for the LED module under consideration or, as the name suggests, another mostly metallic heat sink. A metal core circuit board is the worst case and is always used as a basis for demonstrating the effectiveness of the measures specified here . For safety reasons, the metal core is often connected to a protective earth PE. The resistor R_PE in series sums up all "contact problems" on this path, but it is quite low-impedance because this connection must not be interruptible for safety reasons and must be designed for large currents in accordance with the relevant safety standards. All currents flowing between the LED module and its control gear are also shown. An anode current IAN flows in the connection R_AN, and a cathode current I _K A in the connection R_KA. Because it allows a y-shift of the control gear potential to be the cause in the first place, the most interesting current, a grounding current l _PE , flows in the connection R_PE between the heat sink or the metal core of the circuit board for the LED module or both, i.e. the LED Module cooling, and a protective earth connected to it. The counting arrows of all three currents are equally directed out of the LED module to the control gear, since the glow of the LED emanates from its module.

The lower graph of FIG. 1b shows the potential profile of the negative pole GND of the intermediate circuit voltage, ie the circuit ground, which causes all the glowing, in relation to the protective earth PE. The negative sinusoidal sections VT of the AC mains supply V1 can be clearly seen, which periodically pull down the entire circuit ground GND with the mains supply frequency whenever the phase of V1 has a potential that is becoming more negative than its neutral conductor. A phase with positive potential can be seen here in the almost horizontal GND curve sections for voltages of approximately zero, because during this positive mains oscillation the node GND is connected to the neutral conductor N via D33 from the previous figure, and because then the phase is actually decoupled from the unloaded mains-side rectification circuit. The connection via D33 takes place as long as the phase of the AC mains supply V1 shows increasing positive potential with respect to its neutral conductor, which is up to a point in time 1 at the maximum of the potential of the phase against. Therefore the curve for GND to the left of time 1 is continuously equal to the zero line for this lower graph, to the right of this D33 switches off. This is where a still very flat section of the curve begins, which due to a charge equalization between a parallel to the four switched-off rectifier diodes D31 to D34, as before and according to all other circuit diagram identifiers of the previous figure, effective capacitive voltage divider and the parallel connection of all parasitic capacitances C5 to C35 results. The flatness of this GND curve section to the right of point in time 1 reveals a dominance of this parallel circuit, whose capacitance is able to keep the ground potential GND approximately at its previous zero level, despite switching off the rectifier circuit on the mains side. Such flat GND curve sections suddenly transition into the negative sine sections at a point in time 2 . At such a point in time, the rectifier diode D34 switches on in each case because the potential of the phase of the AC voltage mains supply V1 then falls below the instantaneous potential of the circuit ground GND.

The terminating capacitor C1 takes this jump in the gradient in the ground potential GND with it to its upper end. Because this capacitor, due to its relatively low capacitance of e D34 diagonal upper rectifier diode D31 of the Graetz bridge is switched on, and the rectifier terminating capacitor C1 is charged to the time value of the magnitude of the mains AC voltage V1, which charging process corresponds to the time profile of the ground potential GND to the right of time point 2, which is reflected on the zero line. At this point, it is worth taking a look at the upper graph of the same figure 1b, which shows all currents flowing in the connections between the LED module and its operating device. The anode current or module current IAN flows in the positive pole of the device output, the module return current or cathode current I _K A in the associated negative pole, and most interestingly in the protective earth connection the earth current l _PE . As already mentioned, the counting arrows of all three currents point equally out of the LED module towards the control gear. As can be seen at first glance, the grounding current IRE to the right of time 2 corresponds to the negative sum of the module current IAN and the module return current I _K A through the actual connection lines connected to the positive and negative poles of the device output, which proves the purely capacitive character of all currents . This is because the grounding current I _PE can only flow in parallel across all the parasitic capacitances C5 to C35 from FIG. 1a, and thus the sum of both other currents as well. The vanishingly few "minus 2.94 nA" from the box in the bottom center of Figure 1b here for the time mean value of the earthing current I _PE say the same thing. Secondly, apart from an initial transient process, the shape of the curve of the module return current I _K A and the grounding current I _PE corresponds very precisely to a cosine which leads the negative sine section of the potential curve of the circuit ground GND causing these currents by essentially 90°.

If the circuit ground pulls down in relation to protective earth, as is the case after such a point in time 2, the grounding current I _PE flows capacitively leading into the heat sink and/or into the metal core of the circuit board for the LED module or into the LED module cooling because its values are negative are shown, and almost the same amount of current as module reverse current I _K A out of the module again. The remainder of the grounding current l _PE also flows out of the module, but as module current IAN from the positive pole, and charges or top-ups the storage capacitor C3 of the control gear to a voltage that corresponds to the voltage at which the light-emitting diodes composed of all the LEDs D1 to D30 non-linear characteristic begins to steepen. Both module currents in total flow via circuit ground GND forward to the rectifier circuit and from there via the lower rectifier diode D34, which according to FIG. 1a always conducts when the circuit ground is pulled down, into the AC power supply V1 in. Since the potential of their phase is negative right now and getting more negative, this is not a contradiction, it is causing all this behavior. The same current returns from the mains AC voltage V1 via the protective earth PE, which is ultimately connected to the neutral conductor N, and thus closes the circuit to the current I _PE , which flows into the LED module cooling system. Furthermore, after time 2, no current flows through the connection between PE and node N, which makes it clear that switching off the neutral conductor does not help against glowing, and that the phase of an AC mains supply connected to the operating device that is actually switched off alone is already causing the glowing of the LEDs connected to the control gear under consideration.

The crossing of all currents considered here at the value zero at a point in time 3 exactly when the driving voltage GND from the lower graph has its minimum, and their essentially point-symmetrical course until shortly thereafter, provide the third proof of their purely capacitive character . At a point in time 4 shortly after this current crossing, there is a kink in the current curves towards smaller values, together they form a kind of waistline, and at the same time the potential curve GND of the circuit ground in the lower graph changes from the negative sinusoidal section VT to a form that can be differentiated in a constantly differentiable manner decays exponentially and although it approaches the time axis, it does so more slowly than said sine segment VT. The latter indicates that at time 4 the lower rectifier diode D34 has switched off, which has pulled the potential of the circuit ground GND below that of the protective earth PE or kept it below this potential in the entire preceding time segment from the beginning of the potential drop of the circuit ground from time 2 to time 4 has.

After time 4 and during the peak in the current curves to l _PE , IAN, IKA or during the exponentially decaying time curve of GND, the energies stored in all parasitic capacitances and in the storage capacitor C3 are released by the lead resistances R_AN, R_KA and R_PE and by at least some of the light-emitting diodes D1 to D30, which have been glowing since time 2 and are still glowing, are used up. Both module currents IAN and IKA flow into the module in this time segment, and IAN quasi releases I _K A away. Because of this depletion, the negative potential of the circuit ground GND, which is essentially stored in the parasitic capacitances C5 to C35 between the LED module and its cooling, approaches the zero line of the lower graph. The currents I _K A and IAN flowing into the module in this section after point in time 4 keep the rectifier diode D31 conductive because they can only flow towards the module via the ground line. The terminating capacitor C1, which is charged at time 4 to the instantaneous value of the mains AC voltage V1 and subsequently to the mirrored value of GND, is helpful here, and which discharges as a result of the conduction of the two module currents. The ground potential GND is thus coupled to the instantaneous charge of the terminating capacitor C1 because its upper end is clamped to the neutral conductor N. With the increase in ground potential GND after point in time 4, the cathode potentials of all light-emitting diodes D1 to D30 also increase, because the parasitic capacitances C5 to C35 discharge in the same way as C1. This results in a grounding current l _PE out of the module. Because of this and because of the return via the protective earth to the mains AC voltage V1, the upper rectifier diode D31 remains switched on after time 4 throughout this consumption phase, regardless of the device's own consumption simulated by R1. At the same time, the negative sine wave section from the mains AC voltage, which is left at the continuously differentiable transition of the circuit ground potential profile GND at time 4, can be continued mentally as track VT. If this section of the negative sine half-wave reaches the zero line of the lower graph at a point in time 5, another kink occurs there in the GND potential curve of the circuit ground and at the same time there are even jumps in all the current curves shown above it.

At this point in time 5, the other of the lower rectifier diodes D33 cannot yet connect the circuit ground to the neutral conductor N because, among other things, the ground potential GND is still negative, being buffered by C1. Instead, D32 switches on, takes over the current previously flowing through D31 - which can be read from the fact that there are no crossovers between the module and ground currents at time 5 - and takes both ends of the terminating capacitor C1 along with it to potentials that become more positive in accordance with the V1 curve top end slightly stronger than its bottom end, which is at circuit ground. Therefore, the progression of GND between times 5 and 6 in Figure 1b here corresponds exactly to the potential of the lower end of C1 and thus quite exactly - shifted down by its charge and slightly falsified in the inclination due to its recharging - to that in this one Phase of zero rising profile of the AC line voltage V1. Because in this phase between times 5 and 6, the module current IAN that causes the glow flows mainly through the rectifier terminating capacitor C1, charges it again, continues to flow through the ground line and through the storage capacitor C3 to the positive device output LED+, with the storage capacitor C3 a little is discharged, which, after point in time 5, characterizes it as the source of this "bright glow phase" in addition to the mains AC voltage V1, which has just risen to positive values. Because of the mains-side connection between neutral conductor N and protective earth PE, it is possible for the grounding current IRE to flow out of the LED module cooling and through the input for the mains AC voltage V1 and through D32 back into the control gear. In this way, a low-impedance circuit is closed in the overall system under consideration; only the series connection of the capacitors C1 and C3 with the parallel connection of all parasitic capacitances C5 to C35 limits the module current IAN by means of the increase in the mains AC voltage V1. The module and earthing currents IAN and l _PE are correspondingly high and, due to the purely capacitive limitation, cosinusoidal, many or all of the light-emitting diodes D1 to D30 continue to glow, now even particularly "bright". As above, no current flows through the connection between PE and node N in this "bright glow phase", which underlines the minor importance of a connected neutral conductor when the glow occurs. This equalization process between times 5 and 6 ends when the ground potential has become zero due to the above entrainment effect by C1 and therefore the other lower rectifier diode D33, which is diagonal to the already conductive D32, switches on. This ends the change in potential of the circuit ground GND, which eliminates the cause of the glowing. After D33 is switched on, current can no longer flow through any of the parasitic capacitances C5 to C35. The fourth indicator that the grounding current I _PE is purely capacitive consists in the equal areas above and below the zero line of its time profile in the upper graph of FIG. 1b. The "40.8 pA" from the box in the lower center of the same figure for the effective value of the grounding current l _PE is nevertheless sufficient to let many or all of the connected light-emitting diodes D1 to D30 clearly glow.

The circuit diagram of the following _FIG . 2A corresponds to that already described in detail in FIG the output negative pole LED is expanded by two switching elements D _A G2 and M _AG explained later. The idea of switching other electronic switching elements into the connection R_PE between LED module cooling and protective earth to avoid glowing instead of all these should be rejected for safety reasons. Because this connection R_PE must be designed in such a way that it meets the requirements for a protective conductor specified in the relevant safety standards at all times (i.e. no switchable electronics are permitted).

M _AG corresponds to the dimmer switch already explained above, which is marked as switched off here like all other such electronic switches. Because the absolute darkness shown here corresponds to a dimming level of 0%, for which this dimmer switch has to be permanently switched off. Not shown is the inverse diode that is intrinsically present in each MOS field effect transistor chosen only as an example as an electronic dimmer switch M _AG and that forces the series connection of a lower or second flow or glow prevention diode D _AG2 . If this dimmer switch is to have an additional glow-preventing effect, as required here, a varistor and a SIDAC, for example, must be connected in parallel to protect against transient voltage peaks (not shown).

In FIG. 2B, for the circuit according to the previous figure, the same time curves are shown in the same scale, in the same direction and in the same time segment as in FIG. 1b. However, if the circuit of Figure 2a with the When compared with the time profiles of FIG. 1b, the effect of all three glow prevention components D _A GI, D _A G2 and M _AG can best be explained, and the origin of the time profiles in FIG. 2b can be derived from this.

At the beginning of the glow phase at point in time 2, the storage capacitor C3 is recharged according to the module current I _AN without glow avoidance, i.e. according to Figure 1b, which the upper flow diode D _AG I now prevents because this recharging current wants to flow out of the anode connection of the module and from D _AG I is blocked. At the same time, without avoiding glowing, a clear module reverse current I _KA occurs, which is now blocked by the switched-off dimmer switch M _AG because this current also wants to come out of the module. Both together have the effect that with glow prevention as shown in FIG. 2b here, no appreciable grounding current I _PE can flow from time 2. After the crossing of all _three current curves _at point in time 3, according to _FIG . In the absence of charging or recharging of the storage capacitor C3 after time 2 because of the module current I _AN blocked by D _AG I and in the absence of charging of all parasitic capacitances C5 to C35 because of the module reverse current I _KA blocked by M _AG , the glow phase after time 5 is now almost completely absent.

FIG. 2b shows that the section between times 5 and 6 runs in principle in the same way as explained above for FIGS. 1a and 1b. All variables I _PE , I _A N and GND involved here have the same sign here in FIG. 2b and above in the comparison figure 1b, and the gradient of GND is the same, so here, as above, corresponds fairly exactly to the simultaneous gradient of the AC mains voltage V1. According to Figure 2a, in the section between times 5 and 6, the rectifier diode D32 is conductive and finally leads the ground current I _PE coming from V1 through the rectifier terminating capacitor C1, charging through the ground line, discharging through the storage capacitor C3 and in the direction of flow through D _AG I Anode current l _AN into the LED module. The same current comes out again as earth current _lPE at the protective conductor connection PE of the LED module and flows back to the protective earth via the protective earth AC power supply V1. In this short period of time just described, all three corona avoidance components are ineffective because D _A GI cannot now block and because the grounding current I _PE flows past the other two, D _A G2 and M _AG .

A first difference to the above, by which the comparison figure 1b is always meant below, is the later position of the point in time 5 here in FIG. 2b. AC line voltage V1 is already positive when D32 turns on. At point in time 5, the terminating capacitor C1 is charged to a similar voltage as above, which can be read from the higher value of the ground potential GND at point in time 5 in FIG. 2b. This also causes its later position. Because in the previous section from time 4, at which D34 switches off, the ground potential GND is no longer firmly coupled to the charge of C1 as above, but can adjust itself freely after a capacitive voltage divider acting in parallel with the rectifier circuit on the mains side and thus tend towards zero more quickly. Because the current l _KA or l _AN flowing into the module at this time segment, which also keeps the rectifier diode D31 conductive in the entire section after time 4 and thus couples the ground potential to the terminating capacitor C1, is so small here that D31 more or less simultaneously with D34 turns off.

A second difference from the above is the significantly reduced values of the currents IRE and _IN after point in time 5. In this section, the topological current-limiting factors are the same elements of the circuit as already described, with and without glow prevention components, namely only capacitors and capacitances. The co-limiting capacitors C1 and C3 are the same as above, so that the earth and anode current can only be smaller here because fewer of the parasitic individual capacitances C5 to C35 are reversed charge at the same time. The fact that their initial charges are different here than above is due to the behavior of the improved circuit according to FIG. 2a in the sections after points in time 2 and 3 of FIG.

Only at the point in time 6, which can be read from the lower graph in FIG. 2b, of the complete charge matching of all parasitic capacitances C5 to C35 the chaining of the forward voltages of the “associated” light-emitting diodes D1 to D30, _current peaks occur in the upper graph of FIG. This is because only the number of currently active parasitic individual capacitances C5 to C35 can limit the grounding current I _PE after time 5, and if this is as high as at time 6, all of them are active. Glow cannot occur at any other point in time since all the capacitive equalizing currents according to FIG. 2b are significantly lower than the comparable ones according to FIG. 1b. Because the "minus 2.94 nA" from the box there in the lower center for the mean value over time of the earthing current I _PE has decreased significantly to "minus 665 pA" in Figure 2b here. And the glow-causing "40.8 pA" from the box in the lower center of comparison figure 1b for the effective value of the earthing current I _PE shrinks to just under 2 pA at the same point in figure 2b here, which is a reduction by a factor of a good 20 also makes a glowing of the light-emitting diodes D1 to D30 correspondingly darker, so that it is almost no longer perceptible.

A third difference from the above is the significantly greater drop in the ground potential GND between times 1 and 2, i.e. between the mains voltage maximum and the switch-on point of D34, at which the phase potential of the mains voltage V1 falls below the instantaneous ground potential GND. This clearly shows the decoupling of the ground potential GND from the parasitic capacitances C5 to C35 by the corona avoidance components D _A GI, D _A G2 and MAG, since the ground potential is now essentially only connected via a capacitive voltage divider parallel to the rectifier circuit consisting of D31 to D34 is defined. In the end, the switch-on time 2 of D34 is not only later than above, but above all at a significantly lower level, both of which weaken the kink in the GND curve. This alone reduces the jump height of a capacitive glow current if it were not already minimized anyway by the glow avoidance components D _A GI, D _A G2 and MAG. 3 shows the initial situation, a simplification of the prior art disclosed in DE-10-2015-202-370-A1, for the improved clocked output switching stage specified here with glow prevention integrated therein. The only functional difference from FIG. 2a is the replacement of the synchronously controlled lower buck converter switch M2 there by a freewheeling diode D43. Furthermore, all parasitic components are omitted here, since their effect has been sufficiently described above. The situation of a classic step-down converter in relation to circuit ground GND and the negative pole of its load, in this case the cathode connection LED- of the LED module consisting of D1 to D30, is easy to recognize. The two additional serial modules MAG and D _A G2 there do not significantly disturb the basic situation, especially because the low position of the storage capacitor C3, i.e. its connection to circuit ground GND, speaks the same language. This is, for example, "classic buck converter" or "high side switch buck". All of this leads to an approximately constant voltage difference between the intermediate circuit voltage V_bus and the output voltage Vc3, which has the mains ripple of twice the mains frequency as the dominant AC voltage component and must be positive at all times for a step-down converter to function properly at all. As far as radio interference suppression is concerned, both connections of the LED module can be treated in the same way with respect to protective earth.

Therefore, the same voltage difference can also occur between the negative pole of Vc3, ie the lower end of storage capacitor C3, and circuit ground GND. Then the upper connection LED+ to the load, in particular to the anode connection of an LED module, is connected almost directly to the positive pole of the intermediate circuit voltage V_bus, via which the power factor corrector supplies the rest of its control gear with energy, and also the upper end of the storage capacitor C3, which as a result also optically gets a high position. The bottom end of the same storage capacitor, which is literally “hanging in the air”, is attacked by the buck converter coil L1, which must be periodically connected to circuit ground for magnetization, which is done by the buck converter switch M1, and in between for demagnetization again via the freewheeling diode D43, which supplies the stored current can run free with V_bus. To do this, the two switching elements M1 and D43 must be in reverse order. Terms such as "inverse buck converter", "overhead buck converter", "inverse buck", "reverse buck" or "low side switch buck" have become common for this arrangement of a buck converter. The big advantage of this is that the step-down converter switch M1 works with respect to the circuit ground GND and can also be controlled from there. This change is the entry idea to the specified improved clocked output switching stage.

4 shows the result. The low-lying buck converter switch M1 combines two formerly different functions, that of the clock amplifier for the clocked output switching stage that regulates the mains ripple or other voltage differences, i.e. for the actual buck converter, and that of the dimmer switch. At the same time, it helps to avoid glowing. In its normal operation at high to medium dimming levels and with a continuous output current, for example from an LED module connected between the device output connections LED+ and LED-, consisting, for example, of the light-emitting diodes D6 to D30, the step-down converter switch M1 is operated in a continuous pulse width modulation with a clock frequency of dozens of kHz. If the output current is to have a PWM at a dimming frequency of many hundreds of Hz to a few kHz, i.e. be chopped, the mode of operation for M1 changes to a burst mode in which individual clocks or clock groups from the above continuous pulse width modulation are periodically missing. The dimming frequency, which is significantly lower than the clock frequency, results from the intervals between the first missing pulses. To avoid smoldering, which can be equated with a dimming level of 0%, the buck converter switch M1 is permanently switched off, which also results from an ever-increasing thinning out of the burst mode for dimming down into absolute darkness. Regardless of normal operation or chopped operation, the work of this output switching stage is characterized in that its input voltage V_bus deviates by at least 5% from its output voltage Vc3 over at least 95% of all its operating times and, in particular, is at least 5% higher than its output voltage, which is the forward voltage of the at least one light-emitting diode or the LED module, for example. Consisting of corresponds to the light-emitting diodes D6 to D30 which can be connected to this output switching stage.

Going beyond a known inverse buck converter, the improved clocked output switching stage has an upper glow prevention diode D _A GI, which blocks a return flow of current from the storage capacitor C3 at the output of the control gear into its intermediate circuit capacitor C36, which also stores it. Secondly, a lower glow prevention or flow diode D _A G2 is connected in series with the buck converter switch M1 in such a way that its cathode is connected to its working electrode. This results in the same configuration as in FIG. 2a or 3 in series with the negative pole LED- of the output, in particular with the cathode connection of an LED module, even increased by the series connection of the step-down converter coil L1. Because of both flow diodes D _AG I and D _AG 2, the storage capacitor C3, which not only terminates the clocked output switching stage directly, but also causes at least half of all glowing in the prior art, migrates into the glow avoidance area, i.e. as close as possible the LED module, at least consisting of D6 to D30, whose operating current he should keep free of as many alternating components as possible. Since the configuration of the corona avoidance components D _AG -i, D _AG2 and M1 has not changed compared to the actual load, the series circuit made up of many light-emitting diodes, e.g. D1 to D30, and since the storage capacitor C3, the source of some evil, is now of is decoupled from the cause of the problem, namely the y-shifts due to the protective earth PE, the improved clocked output switching stage contributes even more reliably to glow avoidance than the circuits from the prior art. Without the corona avoidance diodes D _AG I and D _AG2 , said y-shifts, which are equivalent to a periodic potential swing of the protective earth PE via the circuit ground GND, would occur by means of the parasitic capacitances, e.g. C5 to C32, and the light-emitting diodes D6 to D30 as pump diodes, both storing capacitors Charge C36 and C3 in such a way that smoldering can occur. An inverse diode (not shown) in the normally clocked power transistor M1 and the electrically conductive path via the ground line GND to the input stage, via the Graetz bridge as a rectifying circuit and via the AC mains supply back to the protective earth PE help here. Now is even Each individual freewheeling phase via D43 of the step-down converter working in a specified output switching stage is completely decoupled from the input stage of the operating device containing it by both corona avoidance diodes D _A GI and D _A G2, since the freewheeling diode D43 is located between the two corona avoidance diodes or flow diodes D _AG I and D _AG 2 . In accordance with FIGS. 1a and 2a, FIG. 4 here shows the output current as module return current or cathode current I _KA . A radio interference suppression capacitor C51 connects the negative output LED- of the control gear to protective earth PE, and a second radio interference suppression capacitor C52 connects circuit ground GND to protective earth PE. Both radio interference suppression capacitors often have a capacitance in the range of one nanofarad.

A small resonant capacitor C41 with a capacitance value of a few hundred picofarads, for example 220 pF, is connected in parallel with the buck converter coil L1. Its functions are explained in more detail below.

Fig. 5a shows some waveforms of important voltages or currents in a circuit according to the previous figure, the identifier of which is also referred to here. The top horizontal curve in its top graph describes the output voltage Vc3 of the improved clocked output switching stage, to which the storage capacitor C3 is charged in an approximately constant manner in parallel with its output. The wavy course directly below in the same graph describes the output current from the positive pole LED+ to the negative pole LED- of the output switching stage as a cathode current I _KA , which discharges the storage capacitor C3. The same capacitor is charged or recharged by the current l _i in the step-down converter coil L1, as shown in the bottom graph of FIG. 5a. The filtering effect of the storage capacitor C3 at the output of the improved output switching stage can be seen from the completely different shape of these two current curves l _KA and l _i . The middle graph illustrates the current I _Mi through the active buck switch M1 and the voltage Vm1 across it. Due to its clocking, periods of high and low values of both curves alternate. Whenever the buck converter coil current l _i of the bottom graph increases, the buck switch current Im1 also increases identically and is the Buck switch voltage Vm1 is approximately zero, and when the buck converter coil current decreases or undershoots, the buck switch current is zero and at the same time the buck switch voltage Vm1 corresponds to the input voltage V_bus of the output switching stage under consideration as long as the freewheeling diode D43 conducts. What happens when this freewheeling diode D43 also switches off in addition to the switched-off buck converter switch M1 is explained in more detail with reference to the following figure.

In FIG. 5b, a period is zoomed out of the preceding figure, ie shown in a time-extended form, so that more details can be seen. In particular, the step-down switch voltage Vm1 can be clearly seen in the center of its middle graph. This is where said small resonant capacitor C41 comes into play. Without this, the increase in the step-down switch voltage Vm1 would be virtually infinitely steep after a point in time t1 at which the step-down converter switch M1 has switched off and its current, the step-down switch current Im1, has consequently diminished. The resonant capacitor C41 with a capacitance of, for example, advantageously 220 pF flattens this voltage rise visibly, which is why the step-down converter switch M1 has enough time to reduce its current Im1 before a noticeable voltage Vm1 arises across it. As a result, the turn-off losses of the buck converter switch M1 are significantly reduced.

If the value of the step-down switch voltage reaches that of the current input voltage V_bus at a point in time t2, recognizable from the upper left corner in the associated time profile Vm1, the freewheeling diode D43 switches on and the step-down switch current Im1 jumps completely to zero. The current step that occurs in the step-down converter switch M1 corresponds to the Miller current, which also ends at the end of the step-down converter switch voltage change that caused it. The step-down converter coil current l _i curves downwards between the times t1 and t2, which is why its maximum lies between these points in time, that is to say shortly after the step-down converter switch M1 has been switched off. After time t2, the linear part of the demagnetization of the buck converter coil L1 begins, recognizable by the almost constantly falling course of its current l _i in the graph directly below, which current during this entire phase over the Freewheeling diode D43 flows. If this curve has crossed the current zero line, the freewheeling diode D43 switches off.

At this point in time t3, the voltage across the step-down converter coil L1 and across the resonance capacitor C41 connected in parallel to it is negative, as it was in the entire previous linear demagnetization phase, which is also what caused this phase. To relieve this situation, the freewheeling diode that has just been switched off allows a downward cosine oscillation of the step-down switch voltage Vm1, with the step-down converter coil current li_i continuously differentiable and sinusoidally overshooting into the negative at time t3. If this current reaches its zero line a second time after this half negative sine wave oscillation at a point in time t4, this time coming from below, the buck switch voltage Vm1 consequently reaches its minimum after half a cosine period, at which time the buck converter switch according to the usual valley detect mode M1 is to be switched on again. In the illustration here it is turned on a little too early, which - because it's definitely better than too late - is a common approximation of the optimum. The duration of this half period is essentially determined by the resonant capacitor C41 in cooperation with the step-down converter coil L1 and is longer the greater its capacitance. The longer the duration of this half period, ie the larger the resonant capacitor C41, the more precisely the point in time t4 of the above minimum can be determined, which indicates a second purpose of this capacitor C41. Without the resonance capacitor, this half period duration is very short, which makes it almost impossible for the time t4 of the next switch-on command to reach a minimum voltage. This switching on again at time t4 of the step-down converter switch minimum voltage minimizes the turn-on losses, and this even quadratically to the residual voltage Vdump to be switched. If the resonance capacitor C41 above enables a significant reduction in the turn-off losses, it also helps here with an even more significant reduction in the turn-on losses.

This Valley Detect mode, also known as Quasi-Square-Wave mode, CRM (CRitical conduction mode) or TCM (Transient conduction mode), It is known that if the voltage difference between the input and output of a step-down converter operated in this way is greater than half the instantaneous input voltage, the minimum of the step-down switch voltage at time t4 is at or even below its zero line due to said reverse oscillation. This meant that the step-down converter switch M1 could even be switched on completely without loss, which is clearly not the case according to FIG. 5b, visible from the residual voltage Vdump to be switched. Responsible for this is firstly the forward diode D _A G2 in series with the buck converter switch M1 and secondly the fact that the total voltage Vs1 is always considered across an entire switch, i.e. here the voltage across the series circuit of buck converter switch M1 and forward diode or lower glow prevention diode D _A G2 taken together as shown in FIG. 4. The deeper cause for all this is the parasitic parallel capacitances of the power semiconductors, here in particular the capacitance Cm1 of the step-down converter switch M1 and the capacitance Cdag2 of the lower corona avoidance diode D _AG2 . A diode path that is always present, for example that of the forward diode D _AG 2, leads to uneven charging of these parasitic parallel capacitances, in the example when the buck converter switch M1 is switched off to the fact that only the switch parallel capacitance Cm1 (not shown) is charged and, because the charging current required for this in Flow direction can flow through D _AG 2, the (also not shown) diode parallel capacitance Cdag2 not at all. The rising edge in the step-down switch voltage Vm1 from the middle graph in FIG. 5b and its high level on V_bus thus describe said sum voltage Vs1 and at the same time also the step-down switch voltage Vm1 as such, are therefore labeled Vm1=Vs1. Before the buck converter switch is switched on again, however, the inverse current required to discharge its parallel capacitance Cm1 for the purpose of residual voltage reduction is blocked by the forward diode D _AG2 , which is why the above half cosine ringback only describes the buck switch voltage Vm1 alone. The sum voltage Vs1, on the other hand, can oscillate significantly further downwards, as shown in dashed lines, because the only path of a current that can discharge Cm1 as shown must lead _AG2 via its parallel capacitance Cdag2 due to the reverse direction of D, which means a significantly larger change in the sum voltage Vs1 than driving force required. The difference between Vm1 and Vs1 is the

Voltage to which the glow prevention diode parallel capacitance Cdag2 charges, and thus corresponds to the blocking voltage of D _AG2 .

If the buck switch voltage reaches its minimum and the buck converter switch M1 is switched on as desired exactly then or shortly before the associated point in time t4, as shown in FIG. However, this peak is many times higher inside the switch M1, for example a power transistor, which is not visible or measurable from the outside. Because this internal current peak is the discharge of the parasitic parallel capacitance Cm1 from the residual voltage Vdump to zero (so-called "charge dump"), which does not take place via any external lines. At this point in time, however, the sum voltage Vs1 drawn in dashed lines is still negative, which is why the forward diode D _A G2 continues to block. Only when the time up to a point _in time t5 has elapsed with M1 switched on and the coil current l i being positive again is the parasitic diode parallel capacitance Cdag2 “charged” again to zero, so that D _AG 2 switch on and the step-down converter switch M1 takes over the step-down converter coil current l _i , which is then already positive can.

In Fig. 6, another small capacitor C4 is connected in parallel with D _AG 2 with a capacitance value similar to that of the resonant capacitor C41 in parallel with the buck converter coil L1, or particularly advantageously with a value of 100 picofarads, which is the only difference to Figure 4 is. Its capacitance should be in the range of that of the internal or parasitic parallel capacitance Cm1 of the step-down converter switch M1, or it can also be significantly larger than this. For a buck converter switch M1, which is designed, for example, as a MOS field effect transistor using silicon technology, the capacitance of this additional small capacitor C4 can particularly advantageously range from 22 pF to 330 pF.

Then, as FIG. 7 makes clear, the most important switching load can be achieved with clocking at a few tens of kHz up to the step-down converter switch M1 penetrate, namely the reduction of the forward voltage across the buck converter switch M1 to be switched on directly before the planned switch-on process to a residual voltage Vdump close to or ideally equal to zero, which is predetermined from the outside via the above-mentioned back oscillation and leads to the designation "zero voltage switching" or ZVS. The total voltage Vs1 shown here as a dashed line in FIG. 5b as a comparison figure has a similar profile to almost all other variables there, only the time resolution in FIG. 7 here is only half as high as above. However, because the total capacitance parallel to the forward diode D _A G2 is significantly greater than above and in particular greater than the parasitic capacitance Cm1 parallel to the active power transistor as buck converter switch M1, the buck switch voltage Vm1 follows this sum voltage Vs1 much better than above. This larger capacitance C4+Cdag2 can conduct much more of the current required to discharge the transistor parallel capacitance Cm1 than above, with the same total voltage curve Vs1 as the driving force. As a result, the residual voltage Vdump of the clocked power transistor in the function of a buck converter switch M1 is significantly reduced compared to above, namely almost to zero, and thus turn-on losses and radio interference emissions are also reduced, namely almost disappeared.

The principle behind this is a constructively controlled accumulation of the ringback voltage across the section of the internal circuit with the least capacitance acting in parallel. According to FIG. 6, the internal circuit includes the intermediate circuit capacitor C36, the upper glow prevention diode D _A GI, the freewheeling diode D43, the lower glow prevention diode or flow diode D _AG 2, the power MOS field effect transistor M1 as an actively clocked buck converter switch and the ground line GND. According to the above explanation for the most important switching load, the smallest capacitance must act parallel to M1 in order to be able to accumulate the most of the ringback voltage there. For this, C4 is already connected in parallel to D _AG 2. C41 and the storage capacitor C3 act in parallel with the freewheeling diode D43 at the output of the switching stage. Up to this point, however, no capacitance at all is connected in parallel with D _AG I and its parasitic capacitance is negligible. So according to the above principle, the most would change Accumulate ringback voltage where it is not useful and not across M1. Therefore, another small capacitor with a capacity in general of 22pF to 330pF or particularly advantageously with a capacity of 100 picofarads can also be connected in parallel to the upper corona avoidance diode D _A GI (not shown).

However, when the external circuitry on the right-hand side of Figure 6 is included, this becomes obsolete. Because the series connection of the capacitors C3, C51, C52 and C36 is already connected in parallel to the upper glow prevention diode via the detour of the ground line GND and forms a significantly larger capacitance than the 100 picofarads mentioned. In addition, the radio interference suppression capacitor C51 is increased by the parasitic capacitances C5 to C32 between the LED module cooling and all LED connections on the module circuit board, which is why the improved clocked output switching stage is specified at all.

A further novelty in FIG. 7 compared to FIG. 5b is a small piece of buck switch current immediately after the buck converter switch M1 has been switched on at time t4. This small piece of current flows through the forward diode parallel capacitor C4 to discharge it so that the forward diode D _A G2 can turn on again at time t5. The current slice has the same shape as the waveform of the buck inductor current l _i just below in the third graph, so it is an exact fraction of that current. The other part of it flows through the resonance capacitor C41, and the magnitude of the step in the buck switch current Im1 when the forward diode D _AG 2 turns on at time t5 quantifies the value of the current through the resonance capacitor C41.

Fig. 8a shows the basic circuit diagram of an entire LED operating device, with two of the specified clocked output switching stages being provided for two independent output channels, and with the power factor-correcting input stage of the operating device being formed by a SEPIC, as was already the case with a complete preceding operating device family of Applicant for classic lamps was given. The above statement "in order to form an LED control gear together with the almost unchanged input stage" also applies here. Mains supply input, radio interference filter, rectification circuit as a Grätz bridge and its terminating capacitor C1 are subsumed here on the left-hand side in the voltage source symbol V_in for a rectified mains AC voltage, which corresponds to a pulsating DC voltage, which is why it is also marked with + and -. The following input choke L2 together with the active PFC switch M3 represents the step-up part of the SEPIC, its rectifier diode D36 together with the storage coil L3 the step-down. In between is the SEPIC-typical internal capacitor C37, which is always charged to the instantaneous value of the rectified mains voltage V_in. The two chokes L2 and L3 can be coupled to one another, with the winding ends pointing towards the internal capacitor C37 providing the orientation. Furthermore, the storage coil L3 can be formed by an isolation transformer. A SEPIC input stage of this type has the additional advantage that, in addition to power factor correction, it can also generate intermediate circuit voltages that are lower than the mains peak voltage and at the same time have the correct polarity and correct ground reference. SEPIC input stages therefore have a wide output voltage range in which the output voltage (V_bus) can become smaller and larger than the pulsating DC voltage (V_in). This means that any intermediate circuit voltage V_bus can be set, which is slightly higher than the output voltage Vc3a or Vc3b, which corresponds to the forward voltage of the LED module "a" or "b" just connected to the control gear under consideration, so that the step-down converter in the specified output switching stage works properly at all and can regulate the mains hum, or which is a little more than twice as high as the output voltage Vc3a or Vc3b of the control gear under consideration, so that the same step-down converter not only works properly, but also the ZVS, which minimizes losses and radio interference adheres to Finally, mixed forms of this are even possible, up to real-time intermediate circuit voltage optimization with the aim of global loss minimization, including the input stage. Instead of the SEPIC, a simple step-up converter can be used as a power factor corrector (not shown), if it is ensured that the output voltage Vc3a, the corresponds to the forward voltage of module "a" consisting at least of the light-emitting diodes D2 to D5, and at the same time the output voltage Vc3b, which corresponds to the forward voltage of module "b" consisting at least of the light-emitting diodes D6 to D9, is greater than at least half of an intermediate circuit voltage V_bus , which a step-up converter can generate at the given mains voltage.

The two specified output switching stages are connected in parallel across the intermediate circuit capacitor C36, which terminates the input stage to its load and simultaneously terminates both output switching stages to their voltage source. Such circuits are required for so-called tunable white applications, in which the LED module has two parallel channels that are isolated from one another and are equipped with light-emitting diodes of different light colors, different color temperatures or different color locations. In this case, each channel is equipped uniformly, ie all LEDs D2 to D5, for example, as warm-white LEDs and all LEDs D6 to D9, for example, as cold-white LEDs. In this way, the color temperature of the light emitted by the module can be adjusted independently of a total brightness of the same light, which can be used, for example, to implement "human-centric lighting".

In contrast to FIGS. 4 and 6, each freewheeling diode D43a and D43b is connected here to the anode of the associated upper glow prevention diode D _A Gia and D _AG ib instead of to its cathode.

This is revised in Figure 8b. There, as in FIGS. 4 and 6, the freewheeling diodes D43a and D43b are again connected to the cathodes of their associated upper glow prevention diodes D _A Gia and D _AG ib. The freewheeling phases of both step-down converters are then opposed by one less diode forward voltage, which improves the efficiency of both output switching stages and thus the efficiency of a correspondingly constructed operating device. This arrangement, like that of FIG. 8a, is particularly advantageous when two separate LED modules with their respective light colors are connected to it, which can have different impedances to protective earth. Finally, FIG. 8c shows a further development of the arrangement from the previous figure, in which the two separate upper glow prevention diodes are merged into a common such D _A GI. This saves space and costs, but can facilitate so-called "glow crosstalk" between separate LED modules connected to it. However, if both color channels are wired to one and the same module, i.e. if this module has a common anode connection and only the two cathode connections are separate, "glow crosstalk" can no longer be limited by an operating device provided for this purpose, and is responsible for the supply and control of such Consequently, this arrangement can be used as a particularly advantageous combination module.

REFERENCE LIST

1 , 2, 3, 4, 5, 6 points in time within a period of a

mains AC voltage

C1 Terminating capacitor of a mains-side rectification circuit

C3 *) storage capacitor between the output terminals of the output switching stage

C4 ZVS capacitor in parallel with bottom flux diode

C5 to C35 Parasitic capacitances on an LED module

C36 intermediate circuit capacitor

C37 internal capacitor of a SEPIC as PFC

C41 resonant capacitor in parallel with buck converter coil

Cdag2 parasitic parallel capacitance of a flow diode (not shown in the drawing)

C51 RFI capacitor between a negative pole for the cathode contact of an LED module and protective earth

C52 radio interference suppression capacitor between protective earth and

circuit ground

Cm1 parasitic output parallel capacitance of a

power transistor (not shown in the drawing)

D1 to D30 LEDs

D31 to D34 diodes of the mains-side rectification circuit

D36 Diode that recharges the intermediate circuit capacitor C36 to terminate the PFC

D43 *) Freewheeling diode of the output switching stage

DAGI *) upper flow or glow prevention diode

DAG2 *) lower flow or glow prevention diode

GND line(s) or potential of a circuit ground

IAN Anode current coming out of an LED module

IKA cathode current coming out of an LED module

In current through the buck converter coil

Im1 current through the buck converter switch

IPE Earth current coming out of an LED module

L1 *) Buck converter coil

L2 boost coil

L3 storage coil for SEPIC

LED+ *) Positive output or connection or positive pole of

Output switching stage for the anode contact of an LED module

LED- *) Negative output or connection or negative pole of the output switching stage for the cathode contact of an LED module

M1 *) Buck converter switch

M2 synchronous switch instead of freewheeling diode D43

M3 boost converter or PFC switch

MAG dimmer switch t1, t2, t3, t4, t5 Points in time within a clock period of the output switching stage V1 Mains AC voltage

VT negative half wave of the AC mains voltage

V_bus intermediate circuit voltage to which C36 is charged

Vc3 *) Output voltage to which storage capacitor C3, C3a or C3b is charged

Vdump Residual voltage when the buck switch is turned on

M1

V_in Rectified mains AC voltage V1

Vm1 Blocking voltage of the buck converter switch M1 Vs1 Voltage across the series connection of DAG2 and M1

*) If there are several output switching stages, the indices "a" or "b" can be added to the corresponding reference symbols.

Claims

PATENT CLAIMS Circuit arrangement for operating at least one LED, having:

- an input for entering an AC line voltage (V1)

- An inverse step-down converter (L1, M1, D43, C3) for setting a suitable operating current for the at least one LED, the inverse step-down converter having a first (DAGI) and a second (DAG2) glow prevention diode to avoid capacitive displacement currents compared to a protective earth in the switched-off has the state of the circuit arrangement,

- a first (LED+) and a second (LED-) output terminal for operating a load (D1 ... D30), wherein the first or the second corona avoidance diode is connected in series to the first output terminal (LED+), and the second-or the first corona avoidance diode is connected in series to a switching transistor (M1) of the inverse buck converter (L1, M1, D43, C3), with an input voltage (V_bus) of the inverse buck converter (L1, M1, D43, C3) compared to its output voltage (Vc3) between the first and the second output port differs by more than 5% over more than 95% of all operating times. Circuit arrangement according to Claim 1, characterized in that it further comprises a rectification circuit (D31, D32, D33, D34) for rectifying the mains AC voltage (V1) into a pulsating DC voltage (V_in) and a power factor correction circuit (L2, M3, [C37 , L3,] D36) for setting the grid power factor to a value that satisfies local regulations. Circuit arrangement according to Claim 2, characterized in that the power factor correction circuit (L2, M3, C37, L3, D36) has a wide output voltage range in which the output voltage (V_bus) of the power factor correction circuit (L2, M3, C37, L3, D36) is smaller and larger than the pulsating DC voltage (V_in). Circuit arrangement according to one of Claims 1 to 3, characterized in that both connections of a freewheeling diode (D43) of the inverse buck converter (L1 , M1 , D43, C3) are directly connected to one of the glow prevention diodes (DAGI, DAG2) in order to avoid freewheeling phases of the To decouple buck converter activities of the rectifier circuit (D31, D32, D33, D34). Circuit arrangement according to Claim 4, characterized in that the anode of the freewheeling diode (D43) is connected to the anode of the second glow prevention diode (DAG2) in order to switch a current in a buck converter coil (L1) in the best possible way in one direction and at the same time reliably block it in the opposite direction can. Circuit arrangement according to one of the preceding claims 4 or 5, characterized in that the cathode of the freewheeling diode (D43) is connected to the cathode of the first corona avoidance diode (DAGI) in order to shorten a freewheeling path and thus increase the efficiency of the circuit arrangement. Circuit arrangement according to one of the preceding claims 5 or 6, characterized in that the step-down converter coil (L1) is connected in parallel with a first resonance capacitor (C41) in order to reduce the turn-off losses of a clocking power transistor (M1) of the inverse step-down converter. Circuit arrangement according to Claim 7, characterized in that a second resonance capacitor (C4) is connected in parallel with the second glow prevention diode (DAG2) in order to reduce the turn-on losses of the clocking power transistor (M1).