Transformer circuit.

The invention relates to a transformer circuit according to the preamble of claim 1.

Such transformer circuits are used, with the aid of at least one actuating unit, which can be brought into different switching states, to change the amplitude of an AC supply voltage output by a voltage source, if necessary, before it is applied to a consumer as AC load voltage.

Such a transformer circuit is known for example from DE-OS 25 00 065. This circuit comprises a single actuating unit with a transformer, the primary winding of which is fed by the supply voltage supplied by the voltage source. Several taps are provided on the secondary winding, which can be optionally connected to the lines leading to the load by means of automatically controllable switches. This ensures that the same AC voltage amplitude is always supplied to the load even when the circuit is connected to voltage sources which emit different AC voltages with respect to the amplitude.

However, this known arrangement has a number of disadvantages. The entire power supplied to the load must be passed on via the magnetic field of the transformer. The dimensioning of the transformer must therefore be adapted to this total output and there are correspondingly high losses. If the load consumes very high powers, the transformer must therefore be designed very large and moreover cooled, which leads to considerable manufacturing and operating costs. In addition, the known circuit is not suitable for switching frequently and quickly in order to keep the load voltage at least approximately constant despite corresponding changes in the amplitude of the supply voltage. If one were to operate the known arrangement in this way, considerable problems would also arise from the fact that the entire power supplied to the load flows through the changeover switches. On the one hand, these switches would have to be operated under load and, on the other hand, special measures would have to be taken to prevent the power supply from being interrupted during the switchover.

In contrast, the invention has for its object to provide a transformer circuit of the type mentioned, with the help of which the amplitude of the supply voltage emitted by the voltage source can be changed in a simple and quick manner and with extremely low energy losses.

To achieve this object, the invention provides the features set out in claim 1.

These measures are based on the consideration that in many applications the required change in the amplitude of the supply voltage emitted by the voltage source is only a comparatively small percentage, for example of + 25% of the amplitude. Therefore, according to the invention, the main part of the power of the load is supplied via the first winding of the transformer by galvanic means, the inductance of this first winding being very low due to the low number of turns of this winding and the low frequencies with which high powers are delivered to loads Voltage drop generated with correspondingly small losses, which also occur in an increased manner in the conventional transformer circuits.

According to the invention, the at least one control unit of the transformer circuit can be applied by applying a control voltage U_s be brought to the further winding in at least one switching state in which a voltage ΔU in the first winding of the transformer₁ is induced, which, depending on the sense of the turn of the further winding, adds or subtracts from the input winding with respect to the first winding, so that for the output voltage U_A applies:<maths id="math0001" num=""><img file="EP0169488A2_D0001.tif" /></maths>

Here is the relative size of ΔU₁ with respect to the control voltage U_s by the turns ratio w₁/ w_w given the first winding of the transformer for further winding:<maths id="math0002" num=""><img file="EP0169488A2_D0002.tif" /></maths>

The turns ratio w_1/w_w is much smaller than 1 here and is preferably in the range from 1: 7 to 1: 200. In addition, the current that flows through the further winding in the first switching state must be matched to the nominal load current that flows through the first winding of the transformer so that, for a given turns ratio, the fluxes of both windings are approximately the same in amount and have such an angular displacement that the magnetic flux that results in the transformer core, to the desired induced additive or subtractive voltage drop ΔU ₁ leads to the first winding of the transformer. It can be seen that under these conditions the induced voltage drop Δ U₁ is largely independent of the load current, so that even if the load current fluctuates in relation to its nominal value, a constant difference between the input and output voltage of the actuating unit can be maintained.

A major advantage of this arrangement is that the magnetic coupling of the transformer is only the small part of the power that is required for the induced change in amplitude. This significantly reduces the energy losses caused by the inductive energy transfer from one transformer winding to the other. The transformer can thus be dimensioned correspondingly smaller and the effort required for cooling the transformer can be reduced. The switches, which can be used to apply the control voltage to the further winding of the transformer, also result in only a small part of the total power, so that the switches are subjected to much less stress even with frequent switching operations. moreover, even with very large loads, semiconductor switches, e.g. Triacs or switches constructed from V-MOS transistors can be used, which enable switching to be carried out considerably faster than the mechanical switches used in such cases according to the prior art. A complete interruption of the energy supply to the load during switching cannot occur in principle, since the galvanic connection between the load and the voltage source is constantly maintained via the first winding of the transformer.

For a universal usability of such an actuating unit, it is expedient to ensure that even in the periods in which the actuating unit is not in the first switching state, the magnetization of the transformer core is not essentially caused by the flooding of the first winding alone. This can be done, for example, by an auxiliary winding in the periods in which the further winding is not. is at a control voltage, with the help of switches, for example is short-circuited. By appropriately dimensioning the number of turns and the current which then flows through the auxiliary winding, the through-flow of this auxiliary winding can be adjusted so that no significant induced voltage drop occurs on the first winding.

In the periods in which the auxiliary winding is short-circuited, the output voltage of the actuating unit is approximately equal to the input voltage. However, this equality can only be achieved approximately and the equipment required for this is comparatively large.

However, the at least one control unit of the transformer circuit can preferably be brought into different switching states by applying different control voltages to one or more further windings, as will be explained below for various embodiments:<ul id="ul0001" list-style="none"><li>A) control unit with a single further winding to which two control voltages can be applied.</li></ul>

One designates the switching state in which such an actuating unit occurs when the first control voltage U is applied_S1 is, as the first switching state, which is described by the above equations (1) and (2), it results when the second control voltage U is applied_S2 to the further winding under the same conditions as above, a second switching state in which a defined second voltage drop AU2, which is largely independent of the load current, is induced on the first winding. In this case, the output voltage U applies_A:<maths id="math0003" num=""><img file="EP0169488A2_D0003.tif" /></maths>

ΔU depends on this₂ from the control voltage U_S2 also according to equation (2) above.

The input voltages U are preferably used as control voltages_E and the output voltage U_A the control unit use, to which the further winding is galvanically connected with the help of the switches, taking into account the sense of the winding, in such a way that the induced voltage ΔU₁ added to the input voltage and the other induced voltage ΔU₂ from the input voltage U_E subtracted.

The following therefore applies to the output voltage U_A in the first switching state<maths id="math0004" num=""><img file="EP0169488A2_D0004.tif" /></maths>and in the second switching state<maths id="math0005" num=""><img file="EP0169488A2_D0005.tif" /></maths>

However, these two inducible voltages ΔU₁ and ΔU₂ cannot be chosen independently. Rather, they are according to the equations<maths id="math0006" num=""><img file="EP0169488A2_D0006.tif" /></maths>and<maths id="math0007" num=""><img file="EP0169488A2_D0007.tif" /></maths>linked together if w₁ is the number of turns of the first winding and w is the number of turns of the further winding of the transformer.

So that an unchanged transmission of the amplitude of the input voltage of the control unit to the output connections of the control unit is also possible, the control unit can be brought into a third switching state in which no voltage is induced in the first winding of the transformer. So that the first winding does not develop a throttle effect with a correspondingly high voltage drop in this third switching state, care must be taken that the magnetization of the transformer core is not essentially caused by the flooding of the first winding alone. This can be done in various ways, as will be explained in more detail below. It is essential that in this third switching state only an extremely low voltage drops across the first winding of the transformer, so that with a good approximation the output voltage of the actuating unit is equal to the input voltage:<maths id="math0008" num=""><img file="EP0169488A2_D0008.tif" /></maths>

Because of the small voltage drop across the first winding, only a small voltage is induced in the further winding, so that the short-circuit current flowing in the circuit of the further winding remains small and causes very little power loss.

A first possibility for realizing the third switching state is to provide a switch with the aid of which the further winding can be short-circuited, whereby it is disconnected from all control voltages at the same time.

In order not to overload the transformer, it must be ensured that the short-circuit switch is only closed when the switches used to apply the control voltages are open. It must also be ensured that the switches used to apply the one control voltage are only closed when the switches which are used to apply the other control voltage are open, and vice versa.

In order to make a simultaneous closing of these switches impossible, the switching state of each switch is monitored with the aid of an associated sensor unit and a closing command for a previously open switch is suppressed by a blocking circuit if the output signal of the sensor unit of the other switches indicates that one of these other switches is still closed.

It is desirable that when switching from one switching state to the other, the output voltage U_A the control unit as quickly as possible and as "smoothly" as possible, ie without strong fluctuations in the absolute amplitude of the output AC voltage going up or down from its old to the new amplitude value. However, this cannot be optimally achieved in the embodiment in which the third switching state is produced by short-circuiting the further winding, since certain switching criteria must be observed for the closing and opening of the switches, which make it impossible to derive from an amplitude value Switching output voltage to another so quickly that after less than a full oscillation period of the load AC voltage, the amplitude value is reached in a stable manner.

It is therefore preferably provided to produce the third switching state by electrically connecting the further winding of the transformer to the first winding in parallel in such a way that a short-circuited transformer is obtained with two windings wound antiparallel to its core and connected to the same voltage. The currents that flow in the two antiparallel windings each try to build up a magnetic field in the core of the transformer; however, these fields face each other and essentially cancel each other out. The leakage inductance and the ohmic resistance of the first winding through which the load current flows are very small. The voltage drop occurring at it is therefore very small and the above equation (8) applies with a good approximation. The current flowing through the further winding is correspondingly small, since the further winding has a significantly higher impedance than the first winding of the transformer. As a result, the load current practically flows exclusively through this first winding.

In principle, four switches are sufficient for a transformer which has only a single additional winding in order to be able to bring the relevant actuating unit into the three different switching states mentioned.

If no further measures are taken, care must also be taken in this case to ensure that the input voltage and / or the output voltage of the actuating unit is not short-circuited by simultaneous closing of corresponding switches, as a result of which an inadmissibly high short-circuit current would flow. However, this would mean that here again certain switching criteria would have to be observed for the opening and closing of the switches, which would delay reaching the new amplitude value when changing from one switching state to another.

To avoid this, the use of one or more current limiting circuits is provided in particularly preferred embodiments of the actuating unit according to the invention.

In the case of a transformer which only comprises a single further winding, the third and fourth switches, ie the two switches with which the two ends of the further winding can be connected to the connecting connecting conductor of the actuating unit, for example themselves each be designed as a current limiting circuit in such a way that they do not let any current through at all in the open state and the current flowing through them in the closed state only oppose a very small, constant resistance as long as this current remains below a predetermined limit value, prevent the current from rising above this limit.

The transition from the first to the second switching state or from the second to the first switching state then simply takes place in such a way that the two switches opened in the previous switching state are also closed, which corresponds to a transition to the third switching state, and only then do the switches opened, which must be open in the new switching state. Because of their current limiting properties, the third and fourth switches prevent impermissibly high short-circuit currents from flowing in the third switching state.

Another possibility for a transformer with a single further winding is that the third and fourth switches, ie the two switches with which the two ends of the further winding can be connected to the connection connecting conductor of the actuating unit, are not directly connected to this connection. Lead the connecting conductor. Instead, the third and fourth switches are directly electrically connected to one another by a further conductor and a circuit arrangement is provided between this further conductor and the connecting connecting conductor, which on the one hand connects the two conductors in an electrically conductive manner and on the other hand the flow of an impermissibly large current of one of these two conductors prevented on the other. In the simplest case, this circuit arrangement can be a switch which is always opened when the actuating unit is to be brought into its third switching state, in which an impermissibly high short-circuit current would otherwise flow via this switch. However, such switches can only be opened at very specific times, so that the optimum switching speed cannot yet be achieved with them.

Instead of this, an automatically operating current limiting circuit is preferably used as the circuit arrangement, which opposes the current flowing through it with only a very small, constant resistance, as long as this current is less than a predetermined limit value. However, if the current approaches this limit too much, the current limiting circuit steadily increases its resistance so that the current cannot exceed the specified limit. In contrast to a simple switch, which suddenly limits the current flowing through it to zero when opened, this continuous limiting process has the advantage that no voltage peaks occur in the output voltage of the actuating unit. The limit value is chosen so that it is only slightly greater than the current that must flow through the further winding in the first or second switching state and also through the current limiting circuit lying in series with the further winding in these two switching states.

With this arrangement, since the further conductor, which connects the third and fourth switches to one another, would short-circuit the further winding if the third and fourth switches are closed at the same time, the transition from the first to the second switching state is preferably carried out here in such a way that first the second switch is closed, which connects the second end of the further winding to the output end of the first winding. Since in the first switching state the first switch is closed, which connects the first end of the further winding to the input-side end of the first winding, and since this first switch initially remains closed, the two windings are temporarily electrically parallel to one another and the actuating unit is located in the third switching state. The current limiting circuit prevents an inadmissibly high short-circuit current from flowing through the closed second switch and the fourth switch, which is also still closed and which connects the second end of the further winding to the further conductor and thus also to the connecting connecting conductor. The switching process is then continued in such a way that the fourth switch is opened and then the third switch is closed, which connects the first end of the further winding to the further conductor. Even with this switch position, the actuating unit is in the third switching state, since the first and the second switch are still closed. An impermissibly high short-circuit current could now flow through the first and third switches, but this is prevented again by the current limiting circuit. Finally, the first switch is then opened so that the actuating unit changes to the second switching state.

The same applies to switching from the second to the first switching state.

If the actuating unit is not to be kept in the third switching state temporarily, but for a longer period of time, the current limiting circuit can advantageously be designed in such a way that it can be switched to at least one second current limiting value, which is substantially lower than the first current limiting value, preferably equal to zero. In this way, the further winding lying parallel to the first winding of the transformer is practically completely independent of the input voltage U._E disconnected and there is no longer any short-circuit current to the connecting connecting conductor.

An automatically operating current limiting circuit has the advantage over a switch, in addition to the already mentioned avoidance of switching peaks, in that it prevents the current flowing through it from exceeding the predetermined limit value without any delay.

According to a particularly preferred embodiment, it is provided that the limit value to which the current limiting circuit limits the current flowing through it can not only be switched back and forth between two values ​​but can be changed continuously in a predetermined range. This makes it possible on the one hand to limit the short-circuit current flowing in the third switching state to an uncritical value and on the other hand to limit the currents which second switching state flow through the relevant further winding, if necessary to control or regulate.

If triacs are used as switches, which are known to be closed at any point in time but can only be opened when the current flowing through them is zero, no special additional criteria with regard to the switching point need to be taken into account in the switching operations described above.

For the various embodiments of actuating units according to the invention with a transformer with a single further winding, the following time sequences result when switching over:

If an actuating unit comprises a transformer with a single further winding and four switches, of which the first and second are designed as triac and the third and fourth as current limiting circuits, switching to from the first (second) to the second (first) switching state can be carried out up to open switch at the start of switching, ie the second (first) and third (fourth) switches are closed immediately and without any delay, as a result of which the actuating unit changes to the third switching state. To get from this into the second (first) switching state, the first (second) and fourth (third) switch must be opened. Since it is assumed here that the first (second) switch is a triac, this is only possible if the short-circuit current flowing through it and the further winding has a zero crossing. This leads to a time delay, which in the worst case can be half a period of the alternating current. This applies in the same way if the actuating unit not only briefly passes through the third switching state during the transition from the first to the second or from the second to the first switching state, but has been in the third switching state for a long time and is brought into the first or second switching state by the latter should.

The following effect also occurs with all of these transitions whenever the third switching state is left: After opening the switch, the first or second switching state by the further winding then lying at its control voltage, a current which is driven by a completely different voltage source than the short-circuit current, namely in the first switching state by the input voltage of the actuating unit and in the second switching state by the output voltage of the actuating unit; this current is phase-shifted depending on the load current against the short-circuit current flowing before the switch opens, ie as a rule, these two currents are not in phase. Thus, when using triacs during the transition from the third to the second or first switching state in the further winding, which is then connected to the control voltage, there is a strong change in the current flowing through this further winding, which is reflected in the output voltage of the actuating unit by a voltage spike on the first half-wave following the opening of the relevant switch. Only the second following half-wave then has the exact new amplitude value and no longer has any overshoots or voltage peaks. In all of these switching processes, in connection with the above-mentioned waiting time until the next zero crossing of the short-circuit current occurs, a total switching time may result which is too long for certain applications.

The situation is even more unfavorable for an actuating unit in which the transformer has a single additional winding and in which all four switches are designed as triacs. As already described above, when switching from the first (second) switching state to the second (first) switching state, the two switches that were open until the start of the switching process, namely the second (first) and the third (fourth) switch, must not be closed at the same time. Rather, only the second (first) switch may be closed here initially; then the fourth (third) switch must be opened, which is only possible when using triacs at the next zero crossing of the current flowing through this switch. The third (fourth) switch can then be closed with a certain safety margin and only then is it possible to open the first (second) switch, for which a zero current crossing must be waited for again. When switching from the first to the second switching state or vice versa, a waiting time of two half-periods can result in the worst case. If the actuator is held in the third switching state for a long time, the third and fourth switches can be opened. If a transition to the first (or second) switching state is then to take place, the fourth (third) switch must first be closed, which can happen at any time; then the second (first) switch is opened, for which a zero current crossing must again be waited for.

Since the short-circuit current and the current which flows through the further winding in the new switching state are usually phase-shifted with respect to one another in these cases, the voltage peak described above again occurs on the first half-wave of the output voltage, which follows the last switching step of the entire switching process . This results in total changeover times which are as long in the transition from the third to the first or second switching state as in the first embodiment of an actuating unit according to the invention, and which are even longer in the transition from the first to the second or from the second to the first switching state .

If one wants to make the switching processes even faster in these embodiments, the invention provides instead of using triacs electronic switches which can not only be closed at any time but also opened again. For this purpose, for example, V-MOS transistors are available, of which two each have to be connected in series with their source-drain paths with opposite polarity in order to set up an AC voltage switch. With these switches, there are no waiting times until the next zero current crossing. In addition, a switching criterion that is independent of the zero crossing of the short-circuit current can be used for the opening processes, each of which leads from the third switching state to the first or second switching state, which leads to the smallest possible change in the current in the further winding lying on its control voltage after the switching process . If, for example, the point in time at which the current which flows through the further winding connected to its control voltage after the switching process has its zero crossing is used as the switching time, it can be achieved that already at the first half-wave that follows this switching process Output voltage of the actuator without voltage peaks or dips has exactly the new amplitude value.

_Da in the intermediate period in which the actuating unit is in the third switching state, a different current flows through the further winding than if the further winding in the new switching state is connected to its control voltage, according to the invention the time interval of the zero crossing of the last-mentioned current is from Zero crossing of the input AC voltage measured and stored at an earlier point in time, in which the actuating unit is in the relevant switching state. With the help of this stored value, the above-mentioned favorable switching time can then be determined on the basis of a zero crossing of the input AC voltage.

Thus, the times that elapse between the initiation of a switching process and the point in time at which the output voltage has reached its new amplitude value in a stable manner, that is to say without impressed voltage peaks or voltage dips, can thus be shortened considerably. If the actuating unit is in the first or second switching state and a switchover to the second or first switching state is necessary, then in the embodiments equipped with V-MOS transistors as switches, the first half of the change occurring in the output voltage can occur at any time immediately and carry out the second half of this change within a half period of the AC voltage to be switched.

Such a change or influencing of the output voltage in two very quick successive steps is extremely advantageous, because in spite of the great speed with which the new state is reached, this leaves the system with enough time to switch from one switching state to the other without switching peaks and overshoots switch.

A fourth switching state can be produced for an actuating unit, the transformer of which has only a single further winding, in that the switches of the actuating unit are actuated in such a way that the circuit of the further winding has a high resistance value, which, even after step-down transformation, on the side of the first Winding provides a high resistance value. In this switching state, the entire magnetization of the transformer core is caused by the flooding of the first winding. A voltage drop, which is dependent on the size of this flow and thus on the size of the load current, then occurs at the first winding. This throttling effect of the first winding in the fourth switching state can be used to limit the power supplied to the load to a safe level when a short circuit occurs at the load.

B) Actuator with two more windings

According to a second embodiment, however, the transformer can also have two further windings, the flooding and winding conditions for the first winding meet the same conditions as those specified above for the further winding. In this case, the actuating unit is brought into the first switching state in that a control voltage is only applied to the first further winding; In contrast, the control unit is brought into the second switching state in that a control voltage is only applied to the second further winding.

The number of turns, the control voltages and the winding direction of the two further windings with respect to the first winding are preferably selected such that the amplitudes of the two inducible voltages ΔU₁ and ΔU₂ are approximately the same size, but the two inducible voltages have opposite signs to the input voltage U_E are stampable. In this case, the two equations (4) and (5) above again apply to the first and second switching states.

The control voltages can indeed be generated in various ways and applied to the further windings. However, in the first switching state, the first further winding is preferably galvanized directly with the input voltage U using the switches_E connected to the control unit, while in the second switching state the second further winding is directly galvanically connected to the output voltage U_A the actuator is connected so that an autotransformer arrangement is obtained in both switching states. One of the two further windings is used exclusively as an adding winding and the other is used exclusively as a subtracting winding. So there is also only an additive induced voltage + ΔU₁ and a subtractive induced voltage -AU2 are available. However, these two voltages are not necessarily linked to one another via the above equations (6) and (7), since a separate number of turns w for each of the two further windings_w1 or w_w2 can be chosen. The equations apply here to the impressable induced voltages:<maths id="math0009" num=""><img file="EP0169488A2_D0009.tif" /></maths>and<maths id="math0010" num=""><img file="EP0169488A2_D0010.tif" /></maths>If you choose w_w1 and w_w2 for example so that<maths id="math0011" num=""><img file="EP0169488A2_D0011.tif" /></maths>applies, the input voltage U_E exactly symmetrical output voltages U_A+ and U_A- achieve. Alternatively, if desired, the asymmetry between + ΔU shown in equations (9) and (10) can also be used₁ and -ΔU₂ be reinforced.

In addition, this embodiment allows one end of each of the two further windings to be firmly connected and only the other end to be either connected to the input or output voltage in an electrically conductive manner or separated from it using a switch. So fewer switches are needed.

As already mentioned, such a transformer circuit is particularly advantageous if the voltages to be induced + ΔU₁ and - ΔU₂ only a comparatively small percentage of the input voltage U_E turn off. The relationship w₁/ w_wb^between₁/^w_w1 and w1 / ww2 are therefore generally less than 1 and are preferably in a range from 1: 3 to 1: 200.

This second embodiment can also be constructed in different variants, which enable the production of a third switching state, in which the output voltage of the actuating unit is practically the same as the input voltage, in different ways.

A first possibility is that switches are provided, by means of which the two further windings can each be short-circuited. Here too, special measures must be taken to avoid overloading the transformer, which ensure that the switch or switches used to apply a control voltage are closed only for one of the two further windings. For the preferred embodiment with two further windings, one of which is hardwired as an adding winding and the other hardwired as a subtracting winding, this means that the two switches are not operated in an overlapping manner. It must also be prevented that a control voltage is applied to one or both further windings while the associated short-circuit switch is closed.

To make simultaneous closing of the switches in question impossible, the switching state of each switch is also monitored here with the aid of an associated sensor unit and a closing command for a previously open switch is suppressed by a blocking circuit if the output signal of the sensor unit of the other switches indicates that one this other switch is still closed.

In order for energy losses and switching peaks to be as small as possible when switching from one switching state to the other, it is necessary with this variant to open or switch the switches at certain phase angles or in certain phase angle ranges of the magnetic flux which penetrates the first winding of the transformer . close. This phase angle or Phase angle ranges are chosen so that this magnetic flux changes little due to the opening or closing process.

However, this leads to switching criteria which delay the switching from one switching state to another in such a way that the new amplitude value of the output voltage cannot be reached stably within an oscillation period of the load alternating voltage.

Therefore, in a second variant of a transformer which has two further windings, each with one of its two ends with the front or, as seen from the voltage source, rear end of the first winding are provided, that in order to achieve the third switching state, a series circuit consisting of these two further windings is connected in parallel with the first winding; these two further windings lying in series with one another can be regarded as a single winding with a continuous winding direction.

A short-circuited transformer is again obtained with two windings wound antiparallel on the core and connected to the same voltage. The currents in these antiparallel windings try to build up opposing magnetic fields in the core of the transformer, which essentially cancel each other out. The above equation (8) applies again. The current flowing through the two further windings lying in series with one another is very small, since these further windings have a significantly higher impedance than the first winding. The load current therefore flows almost exclusively through the first winding.

In principle , three switches are sufficient for such a transformer , which has two further windings of the type specified above, in order to be able to bring the relevant actuating unit into the three different switching states mentioned.

If no further measures are taken, care must also be taken here that the input voltage of the actuating unit is not applied to the other windings connected in parallel with the first winding, in which an impermissibly high short-circuit current would then flow, by simultaneously closing the switches. However, this would mean that here again certain switching criteria would have to be observed for the opening and closing of the switches, which would delay reaching the new amplitude value when changing from one switching state to another.

To avoid this, the use of a current limiting circuit is preferably provided here.

Preferably, the two switches, with which the two free ends of the two further windings can be connected to the connection connecting conductor, are likewise directly electrically conductively connected to one another by a further conductor, and there is a circuit arrangement between the further conductor and the connecting connecting conductor Provided above under A) type, which is preferably again formed as a current limiting circuit.

Here, too, the previously open switch is first closed during the transition from the first to the second switching state or from the second to the first switching state, as a result of which the actuating unit temporarily changes to the third switching state; the current limiting circuit in turn prevents the flow of an impermissibly high short-circuit current. A short time later, the switch which was closed in the previous switching state is opened, as a result of which the control unit changes to the new switching state.

The statements made above under A) about the preferred configuration and control of the current limiting circuit apply here in a corresponding manner.

In the present embodiment, the following time sequences result for switching from one switching state to another:

In the worst case, when using triacs as switches, a half-period must also be waited until the corresponding switch can be opened when changing from the third to the first or second switching state. It is again irrelevant whether the actuating unit has been in the third switching state for a long time or whether it briefly runs through when switching from the first to the second or from the second to the first switching state.

Here too, when switching to the first or second switching state, the current which flows through the further winding after the switching process has ended, which is due to its control voltage in the new switching state, is phase-shifted with respect to the short-circuit current previously flowing through this winding, so that the same disturbing voltage peak results as in the embodiments described above.

To carry out the switching processes more quickly, it is also possible here to use electronic switches instead of triacs, which, for example, each consist of a series connection of two V-MOS transistors and can be closed and opened at any time.

Once again, the waiting times until the next current zero crossing no longer apply and the switching criterion described above, which is independent of the zero crossing of the short-circuit current, can be used if a switch has to be opened to transition from the third switching state to the first or second switching state.

Again, a point in time is used as the switching time at which the current which flows through the further winding connected to its control voltage after the switching process has its zero crossing. Since a different current flows through the two further windings in the third switching state than when the corresponding further winding is connected to its control voltage in the new switching state, the time interval between the zero crossing of the last-mentioned current and the zero crossing of the input AC voltage is also in a previous one Period measured and the measured value saved. With the help of this stored value, the above-mentioned favorable switching time can then be determined again.

Thus, the time periods required for the switching processes can also be made extremely short. If the actuating unit is in the first or in the second switching state and a switchover to the second or first switching state is required, the first half of the change occurring in the output voltage can be any in the embodiments equipped with V-MOS transistors as switches Immediately carry out the instant and the second half of this change within a half period of the AC voltage to be switched.

Such a change or influencing of the output voltage in two very quick successive steps is extremely advantageous, because in spite of the great speed with which the new state is reached, this leaves the system with enough time to switch from one switching state to the other without switching peaks and overshoots switch.

Such a change, which takes place in two steps, is however not possible with a single actuating unit if it is already in the third switching state and is to be brought out of this into the first or second switching state. Although it only changes the output voltage by half the maximum possible change, this half must be managed in a single step.

If, for a given input voltage of the control unit, more than three different output voltages are to be optionally available in succession at the output connections, the transformer can indeed have several further windings, each of which can have different numbers of turns. These number of turns can be within the range from 1: 3 to 1: 200 mentioned above, but should only differ from one another to such an extent that if the associated voltage is applied to the further winding with the smallest number of turns, none in the other further windings excessive voltages are induced. A corresponding number of switches can be provided, with the aid of which each of these windings is connected to a control voltage or can be separated from this. It is also possible to apply a control voltage to only one or to two or more of the further windings at the same time.

C) Actuator with another winding to which more than two control voltages can be applied.

A preferred possibility, according to the invention, of optionally providing more than three different output voltages in succession at the output of a single actuating unit is, however, alternatively connecting one of a plurality of control voltages U to the at least one further winding with the aid of switches_S1, ..., display US2q^u- place that differ at least partially in their amplitude. Q is any integer greater than 1.

To generate these control voltages ^U_S1, ..., U_S2q an alternating voltage source is preferably used which has a plurality of taps between which different tap voltages U_X1, ..., U_XP are constantly available and can be tapped. p is also an integer greater than 1 and preferably less than q. With the help of switches, these tap voltages can be applied either individually or in groups as control voltages to the further winding of the transformer.

It is an essential aspect of the invention to provide a transformer circuit which has a predeterminable change range + ΔU_Max enables a digital change in the voltage applied to a load and thus also in the power delivered to the load. In special cases, the change range can only be positive or only negative; ie only the additive or only the subtractive impression of induced voltages AU on the input or supply voltage may be required. In the following, however, the general case of a change range symmetrical to change zero (input voltage equals output voltage) is ± ΔU_Max explained.

With a digital change in the output voltage in this range ⁺ ΔU_Max it is understood that there is a small, noticeable voltage change + ΔU on both the positive and the negative side_min or - ΔU_min there and that in the positive part of the change range q positive stresses - ΔU₂ (ν = 1, ..., q) and in the negative part of the change range q negative impressible voltages - ΔU (ν = 1, ..., q) are available, whereby the following applies:<maths id="math0012" num=""><img file="EP0169488A2_D0012.tif" /></maths>and<maths id="math0013" num=""><img file="EP0169488A2_D0013.tif" /></maths>

This means that in the positive as well as in the negative part of the change range any voltage that can be impressed + ΔU₂ an integer multiple of the associated smallest impressable voltage + Δ_Umin is and that₂ can take any integer between 1 and q. The greatest inducible voltage in each direction is also the limit of the range of change:<maths id="math0014" num=""><img file="EP0169488A2_D0014.tif" /></maths>

It can be seen that the range of change is selected both by choosing the smallest change ± ΔU_min and thus the step size, as well as by choosing the number q of cuts. However, increasing the step size leads to a reduction in the accuracy with which the load voltage U is used, for example, when using the transformer circuit according to the invention as a control device_L can be kept constant at a predetermined value. On the other hand, an increase in q means an increase in the technical outlay. So when determining the quantities q and ± ΔU_min an optimization tailored to the respective application.

The amplitudes are preferably of + Δ U_min and - Δ U_min at least approximately the same size, so that the following also applies at least approximately to the other inducible voltages:<maths id="math0015" num=""><img file="EP0169488A2_D0015.tif" /></maths>

The control voltages U to be applied to the further winding are also corresponding_Sν structured digitally according to the invention, ie there is a smallest control voltage U_Sminwhich are used to impress the smallest induced voltage ΔU_min leads, and the remaining control voltages are integer multiples of this smallest control voltage:<maths id="math0016" num=""><img file="EP0169488A2_D0016.tif" /></maths>

whereν again runs through all values ​​from 1 to q. Around the symmetrical change range specified above⁺ ΔU_Max To be able to cover with 2q steps, only q control voltages U_S2 can be provided since with the help of the switches each voltage tapped from the AC voltage source can be applied to the further winding in two different ways so that in one of the two cases the winding sense of the further winding with respect to the first winding of the transformer "the winding sense in the other Trap is just opposite. As a result, the induced voltage ΔU is additively impressed in one case and subtractively in the other case on the input voltage of the actuating unit.

Here, too, there is again the possibility of short-circuiting the further winding so that the output voltage of the actuating unit is equal to the input voltage, or to interrupt the circuit of the further winding in order to limit the load current due to the resulting choke effect of the first winding.

To generate the q control voltages U_S2 According to the invention, it is not necessary to provide q + 1 taps on the AC voltage source in such a way that one of the smallest control voltage U_Smin corresponding tap voltage U_Xmin falls off.

Rather, the amplitudes of the tap voltages are graded according to a suitable code so that with a minimal number of taps (and thus also a minimal number of switches) all required control voltages U_Sν add up by combining several tap voltages, as long as they do not correspond directly to one of the voltages available between two adjacent taps. So that the smallest control voltage U_Smin is available, at least one pair of adjacent taps must be provided. _hen be between whom an ab^Griffs^tension U_Xmin U_Smin falls off. Between the other pairs of adjacent taps, tap voltages can then be at least partially provided, the integer multiples of the smallest tap voltage U that are to be determined according to the above-mentioned code and different from 1_Xmin are. The cheapest code here is the pure binary code, in which each tap voltage occurs only once and the tap voltages 1 in sequence between successive tap pairs. U_Xmin, 2nd U_Xmin, 4th U_Xmin, 8th . U_Xmin etc. fall off.

The use of this code, however, presupposes that tap pairs that are necessary for the additive composition of a control voltage U just required_Sν not required, can be short-circuited without further notice.

In the case of an alternating voltage source preferred according to the invention, which consists of an additional transformer arrangement with a winding to which an alternating voltage is applied and which is divided into a plurality of winding sections between which the taps are used _Agrasp the tap voltage^Gen U_X1, ..., U_XP are brought out, the above-mentioned condition for the use of a pure binary code does not exist. For this reason, a code is preferably used here which allows each control voltage required to be tapped from a group of taps which follow one another in as far as they cannot be tapped directly from a single tapping pair. In general, this means that at least the smallest tap^panswer U_{Xmin '} in some cases, however, some of the integer multiples must be able to be tapped several times. For example, for the generation of eight control voltages 1 .U_{Smin '}2nd U_{Smin '} ..., 8th U_Smin four winding sections may be provided on the winding of the additional transformer arrangement, the number of turns of which are selected such that the tapping voltages are in turn at the taps 1 · U_Xmin, 2 · U_Xmin, 4 · U_Xmin, 1 . U_Xmin fall off, with U_Xmin = U_Smin is. _Mto see that the control voltages 1. U_Smin. 2nd U_Smin and 4th U_Smin can be tapped directly on the first or second or third winding section (counted from the left in the above row) while the control voltage ^{3 .} U_Smin via a combination of the first and second winding sections, the control voltage 5. U_Smin via a combination of the third and fourth winding sections, the control voltage 6. U_Smin via a combination of the second and third winding section, the control voltage 7. U_Smin via a combination of the first, second and third winding section and the control voltage 8. U_Smin can be tapped over the combination of all four winding sections. The code just given as an example, however, is not the only possible one with this number of required control voltages and four available winding sections. For example, all eight control voltages can also be tapped if the integer multiples of the smallest tapping voltage correspond to code 1,3,2,2.

The number of turns of the winding sections is preferably selected such that the tap voltage 1 at the section lying at one of the two ends of the row of winding sections. UXmin and at the section that lies at the opposite end, the tap voltage 1. U_Xmin are directly accessible, as is the case with the first of the two examples above.

It is essential that the code is always chosen so that with a minimum number of winding sections or taps all the necessary control voltages U_S2 be available. In addition, if possible, the maximum alternating voltage that can be tapped across the combination of all winding sections should be equal to or at least not significantly greater than the maximum required control voltage U._Smax be.

The additional transformer arrangement preferably consists of only a single winding which is subdivided into the various sections and at the extreme ends of which a corresponding AC voltage is applied. For example, the input or output voltage of the actuating unit itself can be used for this purpose.

In order to be able to use a transformer circuit, which consists of a single actuating unit, on the further winding of which different control voltages can be applied with the aid of switches in the manner just described, as voltage constants and / or voltage regulators, the invention further provides that the voltage applied to the load Voltage U_L is measured using a sensor arrangement that a comparator the output signal of the ^{Sensor arrangement} with a reference value U _ref compares, which represents the target value S of the load voltage, and that a switch control is provided which controls the switches based on the difference signal, which is emitted by the comparator arrangement, such that the voltage changes Δ U. induced in the first winding of the transformer_ν any fluctuations in the load voltage U_L counteract and compensate for these fluctuations.

If it is desired to be able to provide more than three different load voltages in succession for a given supply voltage Uv, it is advantageous as an alternative to the single-stage arrangement just described to provide a transformer circuit in which two or more stages, each of which can consist of one or more control units, are connected in series with one another in such a way that the supply voltage U_v as input voltage U_E is present, the output voltage U_A this first stage as input voltage U_E is applied to the second stage, etc. and that the output voltage of the last stage of the load as the load voltage U_L is fed. Seen from the voltage source, the first windings of the transformers of all stages are in series with each other and with the load.

The stages connected in series with one another can each consist of a single actuating unit which is designed with one or more, in particular two further windings, and according to one of the embodiments described above, at least into the equations (4), (5) and ( 8) defined three different switching states can be brought.

As an alternative to this, the stages of such a transformer circuit can also each consist of two actuating units connected in series, which are combined to form a pair of actuating units.

This should be understood to mean the following: These are two actuating units, which also have two further windings, one of which is used as an additive and the other as a subtracting winding. The two transformers are dimensioned in such a way that each of the two actuating units, both in an adding and in a subtracting manner, is able to effect approximately half of the total voltage change that is to be applied by the actuating pair. For example the actuator unit pair its input voltage U_EP by + ΔU_P can change, each of the two actuating units can change the input voltage of + ΔUp / 2 supplied to them alone. If each of the two actuating units is in its first switching state, this is referred to as the first switching state combination of the actuating unit pair and the output voltage of the actuating unit pair applies<maths id="math0017" num=""><img file="EP0169488A2_D0017.tif" /></maths>if U_EP is the input voltage of the actuator pair.

If each of the two actuating units is in its second switching state, this is referred to as the second switching state combination of the actuating unit pair, and it applies<maths id="math0018" num=""><img file="EP0169488A2_D0018.tif" /></maths>

Furthermore, the turn ratios of the two transformers are matched to one another in such a way that the effects of the two actuating units compensate one another when the actuating unit pair is in a third switching state combination; In this third switching state combination, for example, the first actuating unit, which is closer to the supply voltage source, is in the first and the second actuating unit is in the second switching state. It then applies to the output voltage of the pair of actuators<maths id="math0019" num=""><img file="EP0169488A2_D0019.tif" /></maths>

It is a great advantage that the losses that occur in the control units are extremely low in all three switching state combinations. In particular, the unchanged transmission of the input voltage to the output of the pair of actuating units in the third switching state combination is practically loss-free.

Compared to an actuating unit, which alone can assume the three switching states corresponding to equations (4), (5) and (8), such an actuating unit pair has the advantage that only half of those for each actuating unit The relevant voltage or power change must be applied. Although two transformers are required, they can also be dimensioned considerably smaller and lighter in accordance with half the power. This is particularly advantageous for the manufacture, transport and spare parts management of transformer circuits for high performance.

In general, the fourth switching state combination remains unused for a pair of actuating units, in which the first actuating unit is in the second switching state and the second actuating unit is in the first switching state. The information just given can be summarized in the following Table 1:<tables id="tabl0001" num="0001"><img file="EP0169488A2_D0020.tif" /></tables>

Basically, it is not necessary here that each of the two actuating units of the pair of actuating units can be brought into the third switching state by itself.

Preferably, however, in the case of a pair of actuating units, each of the two actuating units is designed such that one further winding or both further windings can be connected in parallel with the first winding, that is to say each of the two actuating units can be brought into the third switching state on its own; you can see the above-mentioned current limiting circuit or Current limiting circuits before, so with the help of V-MOS transistor switches an extremely fast, in several steps switching from each switching state combination of the actuating unit pair to any other switching state combination can be carried out.

For example, if the pair of actuators from the second switching state combination (U_AP2 = U_EP- Δ U_P) in the third switching state combination (U_AP3 = U_EP) can be done with such a pair of actuating units without delay by closing the switch in both actuating units, by closing the actuating unit itself being brought into its third switching state, as described above. This results in a further switching state combination, which with regard to the output voltage U_AP of the actuator pair with the third switching state combination described above is equivalent. So UAP3 '= U also applies here_EP. This change in the output voltage by ΔU_P can take place at any time and the output voltage goes from the old voltage value U practically without delay_AP2 to the new voltage value U_AP3' over.

The same applies to a transition of the actuator pair from the first switching state combination (U_AP1 = U_EP + Δ ^U_P) in the further switching status combination.

However, it is useful to pair the actuators when the output voltage U_AP longer than the input voltage U_EP should remain to switch from this further switching state combination to the third switching state combination described above. This takes place at the most favorable times in that the first actuating unit is brought into its first switching state and the second actuating unit into its second switching state by opening the corresponding switch. The output voltage of the pair of actuators goes from^U_AP3' ^{= U}_EP on ^U_AP3^{= U}_EP about, so practically does not change.

The third switching state combination has the advantage over the other switching state combination that, if necessary, a transition to the first or the second switching state combination can take place in two equally large change steps, of which the first can be carried out without any delay, that the second or the first actuating unit is brought into its third switching state by closing the switch in question. As a result, the output voltage of the pair of actuators immediately goes from U._Ap3 = U_E on U_E + ΔU_P/ 2 or U_E -ΔU_P/ 2 over. At the next favorable time, which occurs at the latest within the next half cycle of the AC voltage, the second or the first actuating unit is then brought from the third into the first or the second switching state, as a result of which the actuating unit pair changes into the first or second switching state combination , in the U_Ap1- U_E⁺ ΔU_p/ 2 + ΔU_P/2 ^{or U}_AP2⁼ UE - ΔU_P/ 2 - ΔU_P/ 2 applies.

The transition from the first to the second or from the second to the first switching state combination likewise takes place in two steps, of which the first can be carried out immediately and the second at the latest within the next half cycle of the AC voltage. In this case, the first step consists in bringing both actuating units into their third switching state simultaneously by closing the corresponding switches; in the second step, the two actuating units are each converted into their second or their first switching state by opening the corresponding switches.

If a transformer circuit consisting of one or more such pairs of actuating units (which can then cause different voltage changes) is used as a voltage regulator or voltage constant, it can also be used to meet the extremely high requirements with regard to switching speed and switching accuracy, such as those used in the Power supply of data processing systems are provided.

In order to be able to cover a larger range of output voltage values ​​in small voltage steps, it is advantageous to have several stages which either consist of individual actuating units, each of which can be brought into the third switching state, or consist of the pairs of actuating units described above

(whereby both types can be mixed in an arrangement), to be connected in series and the voltage differences ± ΔU₁, ..., ± ΔU_nthat can generate n such stages to choose different from each other. It is particularly advantageous if the percentage values ​​that result when each of these voltage differences is divided by the supply voltage divided by 100 are in relation to one another in integer powers of three. This applies to the smallest voltage difference that can be generated by one of the stages⁺ ΔU_min:<maths id="math0020" num=""><img file="EP0169488A2_D0021.tif" /></maths> so the voltage differences of the other stages are chosen so that they are approximately equal to + 3A%, ± 9A% etc. of the supply voltage U_v are.

For example, if three stages are connected in series in a transformer circuit, and for each stage the three switching states or Switching state combinations are used, so twenty-seven combinations of switching states are possible for the entire transformer circuit, one of which allows the supply voltage emitted by the voltage source to reach the load with an almost unchanged amplitude, while thirteen combinations reduce the amplitude of the supply voltage approximately by integer multiples of A. % increase and thirteen combinations decrease this amplitude approximately by integer multiples of A%. This is shown in more detail in Table 2.

In the left column of this table, the consecutive number n of the respective combination of switching states is shown, whereby the superscript "+" or "-" indicates whether it is a combination that leads to an enlargement ("+" ) leads to the amplitude of the supply voltage or a combination that lowers the supply voltage ("-").<tables id="tabl0002" num="0002"><img file="EP0169488A2_D0022.tif" /></tables> In the middle column, a "+" means that one or both actuating units of a pair are in the first switching state in the relevant stage, so that the amplitude of the supply voltage is increased by 9A%, 3A% or A% while a "-" means a corresponding reduction and "O" symbolizes the third switching state of an individual actuating unit or the switching state combination 3 (see table 1) of the relevant actuating unit pair, in which or in which the amplitude of the input AC voltage is passed on unchanged. The right column shows the total changes in amplitude that can be achieved by the respective combination of the switching states of all stages. Only rounded values ​​are given, which do not take into account that the input voltage of the stages closer to the load can change depending on the switching state of the preceding stages.

It can be seen that the amplitude change takes place with the aid of such a transformer circuit according to the invention in discrete steps, the step size from one switching state combination to the next always being approximately equal to A% of the respective supply voltage.

If a stage is made up of two actuating units that form a pair, as an alternative to the arrangement just explained, only two switching state combinations can be used for each actuating unit pair, for example the switching state combination O in which the output voltage is equal to the input voltage , and the combination "-", in which the output voltage by n._A% is lower than the input voltage, where n takes a different integer value for each pair of actuators. For this use case, it is possible to construct the pairs of actuators in such a way that they can only assume these two switching state combinations. This can be done in such a way that, for example the front actuating unit of each pair has a hard-wired, non-switchable further winding that permanently induces, for example, a negatively impressed voltage - (n / 2) · A%, while the second actuating unit has an adding and a subtracting further winding, which is alternatively so can be switched so that they either have a voltage of + (n / 2) .A% or of - (n / 2)_'Induce A%, which in conjunction with the induced voltage - (n / 2) · A% of the front actuator results in either a change in voltage O or - nA%. Correspondingly, actuator unit pairs can also be provided, which can only assume the two switching state combinations O and + n · A%.

In all these cases, the change in the output voltage of the entire transformer circuit with respect to the input voltage does not take place according to the ternary code shown in Table 2, but according to a binary code. To cover the same voltage change range, more actuator pairs are required than with the ternary code; However, there are applications in which the input voltage is only to be changed in one direction based on an overall change O and / or the voltage change range is not large. The advantage of a purely binary control can then outweigh the increased need for control units.

Regardless of how many stages are connected in series and whether a binary or a ternary or other code is used, it is a salient feature of a transformer circuit according to the invention constructed in this way that it allows a gradual or digital influencing even of very large powers. In contrast to analog systems, it has an extraordinarily high regulating or control speed. The accuracy achieved in each case essentially depends only on the number of actuating units or stages used.

However, the typical and preferred application of a transformer circuit according to the invention consisting of two, three or more stages does not consist in the fact that nine, twenty-seven or more output voltages should be able to be generated one after the other starting from a fixed supply voltage originating from a voltage source.

Rather, the use of such a transformer circuit as a voltage constant and / or voltage regulator is provided in a particularly preferred application. This means that the setpoint S_L for the voltage supplied to the load, either the nominal value of the supply voltage Uv emitted by the voltage source or another voltage value can be selected. However, such a different setpoint must lie within the change range of the transformer circuit according to the invention. If it is very close to the limit of this change range, the load voltage U is regulated_L only possible in the case of deviations from the setpoint S in one direction. However, this is completely sufficient in cases where deviations in the other direction do not occur.

In the following, the application as a symmetrical voltage regulator is explained in more detail, with the aid of which it is prevented that the amplitude of the load voltage supplied to a load of a predetermined setpoint value S_Ldeviates by more than + δ%, which is equal to the nominal value of the supply voltage U_V which can fluctuate in a much larger range, for example by a maximum of ± Δ% of the nominal value.

For this purpose, a circuit arrangement according to the invention comprises, in addition to a transformer circuit with a corresponding number of stages, a sensor arrangement that measures the amplitude of the supply voltage and / or the amplitude of the load voltage, a comparator arrangement that compares the sensor signal or signals with one or more reference values ​​and, in the event of deviations, corresponding difference signals generated, as well as a switch control, which compares these difference signals, for example, with a permanently programmed table of difference signal values. From this comparison, the switch controller determines the combination n or n of switching states (see Table 2) that is required to compensate for the deviation of the supply voltage from the nominal value that has occurred, so that the load voltage is within the predetermined range S_Lf% remains.

It is now assumed that the amplitude of the supply voltage is initially the nominal value and thus also the target value S_L corresponds, but then deviates from this nominal value to an increasing degree, for example: In this case, the switch control must change from the existing switching state combination n = O (see Table 2), at which the load voltage U_L equal to the supply voltage U_v is in time for the switching state combination n = 1^-, with further increase to the combination n = 2^- etc. pass over. As a result, a corresponding integer multiple of A% is subtracted from the supply voltage and thus the load voltage in the desired range S_LHeld δ%.

With a continuously increasing positive deviation, the transition from the nth combination to (n + 1) takes place^--th combination in each case at a certain switching threshold SW_n- / (n + 1)^-, ie a fixed amplitude value of the supply voltage. If the positive deviation steadily decreases again, the transition from the (n + 1) th combination to the n th combination of switching states takes place at the same switching threshold in the opposite direction. It is advantageous to separate the last two switching thresholds from one another by means of a small voltage difference. The "hysteresis" achieved in this way prevents an excessive switching cycle in cases in which the supply voltage U_V has a value for a long time that is equal to a switching threshold and fluctuates slightly around this value.

The same applies to negative deviations of the amplitude of the supply voltage from the nominal value only that the switching thresholds with SW_n+_{/ (n + 1)}+ are called, because in this case, with increasing deviation downwards from the additive impression of n times the minimum amplitude change A% to the additive impression of (n + 1) times of A%, the desired constancy of the amplitude has to be used to achieve the load voltage.

With each transition from a combination of switching states to an adjacent combination, the amplitude of the load voltage changes abruptly by approximately A%. The switching thresholds are preferably set so that when the amplitude of the supply voltage passes the value of the switching threshold in question without a sudden change, the amplitude values ​​U_Lvor and U_Lnach lie symmetrically to the setpoint. Here U_Lvor the amplitude of the load voltage before _Uswitching and U_Lnach the amplitude of the load voltage after the switching process. The following should therefore apply with the best possible approximation:<maths id="math0021" num=""><img file="EP0169488A2_D0023.tif" /></maths>still applies | U_LVOR-U_Lnach| U_Lnach| = A · S. 100. The percentage value A is constant, but is not at the setpoint S._Lbut related to the amplitude of the input voltage of the respective stage. Hence the size of U_Lvor and U_Lnach depends on which combination of switching states a transition to an adjacent combination takes place.

The above equation (14) can be maintained in any case by suitable selection of the switching thresholds SW. It is also possible according to the invention to ensure that U_Lvor and U_Lnach within the desired control accuracy S_L+ δ% predetermined amplitude range, with the percentage on the target value S_L= 100% related.

When determining the value of A, it should be taken into account that on the one hand A should be as large as possible so that as few actuators as possible are required to cover a given fluctuation range Δ, but on the other hand A should not be chosen too large, because otherwise the desired control accuracy J is not can be observed. According to the invention, A is preferably chosen so that it is between 1.6 ∫ and 1.8 ƒ.

It should be pointed out again here that the switching thresholds can be used regardless of whether the circuit arrangement works as a voltage constant or as a voltage regulator, ie whether the load voltage U_L on a setpoint S_L is maintained, which is equal to the nominal value of the supply voltage emitted by the voltage source or to a nominal value which differs from this nominal value.

The use of these switching thresholds is also independent of whether the supply voltage or the load voltage is measured with the sensor arrangement. In the first case, the difference between the above switching thresholds and the setpoint S_Ldirectly contained in the table used by the switch control, with which the difference signal supplied by the comparator is compared. In the second case, the switch control must be based on the approximation of the amplitude of the load voltage to one of the values ​​U_Lvor and U_Lnach and / or the knowledge of the currently valid combination of switching states determine which switching threshold the supply voltage is currently approaching and which switchover must therefore be carried out.

Another possibility is that the sensor arrangement measures the amplitude of the alternating voltages in front of and behind the transformer circuit. The changes both in the supply voltage Uv and in the load voltage U are then_L detected and evaluated in such a way that the switches of the actuating units are controlled in such a way that the amplitude of the voltage supplied to the load is as constant as possible.

A transformer circuit according to the invention can advantageously be used in multiphase systems with or without a neutral conductor. In the first case, at least one control unit is provided for each phase, the first winding of which lies in the respective phase conductor in such a way that the load current flowing on this phase conductor flows through it, while the connecting connecting conductor of each control unit is connected to the neutral conductor of the multiphase system connected is.

If the multiphase system does not have a neutral conductor leading from the voltage source to consumption, the first windings of the actuating units which are provided for a specific phase are switched back into the phase conductor and all the connecting connecting conductors are connected to one another, thereby creating an artificial zero -Conductor is formed, which can be at any potential.

Finally, in the case of a multi-phase system without a neutral conductor, the actuating units provided for the different phases can be arranged in a daisy-chained circuit.

The invention is described below using exemplary embodiments with reference to the drawing; in this shows:<ul id="ul0002" list-style="none"><li>1 is a transformer circuit, in which a control unit is arranged between the voltage source and the load, which according to a first embodiment has a transformer with a single, short-circuitable further winding,</li><li>2 shows a transformer circuit in which a control unit is arranged between the voltage source and the load, which according to a second embodiment has a transformer with two further short-circuitable windings,</li><li>3 shows a detail from FIG. 2, which shows the details of a sensor unit,</li><li>4 two actuating units connected in series, which form a pair of actuating units and each of which, according to a third embodiment, has a transformer with two further windings which cannot be short-circuited,</li><li>5 shows a transformer circuit constructed as a single-phase voltage constant with three stages connected in series,</li><li>6 shows a further embodiment of a voltage constant for a 3-phase system,</li><li>7 shows an embodiment in which a single actuating unit can be brought into a multiplicity of switching states and is used as a voltage regulator,</li><li>8 shows a further embodiment of an actuating unit for a transformer circuit according to the invention, in which the transformer has only a single further winding which can be connected in parallel with the first winding,</li><li>9 shows an embodiment of an actuating unit for a transformer circuit according to the invention, in which the transformer comprises two further windings which can be connected in parallel with one another in series with the first winding,</li><li>Fig. 10 shows the structure of a current limiting circuit, as used in the actuating units shown in Figs. 8 and 9, and</li><li>Fig. 11 is a diagram for explaining the selection of the cheapest switching times when changing from one switching state to another.</li></ul>

Fig. 1 shows an AC voltage source 1, the supply voltage U_v outputs the input terminals 2, 3 of an actuator 4 as input voltage U_E is fed. An output voltage U appears at the output connections 5, 6 of the actuating unit 4_Athat a load 7 as a load voltage U_L is fed.

With the help of the actuator 4 according to the invention, the amplitude of the output voltage U_A versus the amplitude of the input voltage U_E changeable. For this purpose, the control unit 4 comprises a transformer 8, the first winding 9 of which is connected between the input connection 2 and the output connection 5, while the input connection 3 is connected directly to the output connection 6 by means of the connection connecting conductor 10. In this way, seen from the voltage source 1, the first winding 9 is connected in series with the load 7.

The transformer 8 has a further winding 11, which is magnetically coupled to the first winding 9 via the iron core 12 of the transformer 8. With the two ends 13, 14 of the further winding 11 two pairs of switches 15, 16 and a short-circuit switch 17 are connected.

With the help of the two pairs of switches 15, 16 and the short-circuit switch 17, the actuating unit 4 can be brought into four different switching states. In the first switching state, in which the pair of switches 15 is closed and the switches 16, 17 are open, the input voltage U is applied to the further winding 11_E created. The winding direction of the windings 9, 11 defined by the points 19, 20 is chosen so that the voltage ΔU₁, which is induced in this first switching state by the further winding 11 in the first winding 9, to the input voltage U_E added. The voltage is thus obtained between the output connections 5, 6 of the control unit<maths id="math0022" num=""><img file="EP0169488A2_D0024.tif" /></maths>

As already mentioned, the value, ie the absolute absolute value of the induced voltage ΔU₁ by the turns ratio w_1/w_w the first winding 9 to the further winding 11 according to the equation ΔU₁ = w_I.U_E/^w_w fixed.

In the second switching state, which is shown in FIG. 1, the switches 15 and 17 are open and the pair of switches 16 is closed, as a result of which the output voltage U_A the actuator 4 is placed. At the same time, the winding direction of the further winding 11 is reversed compared to the first switching state. This subtracts the voltage ΔU₂, which is induced in this switching state in the first winding 9 of the transformer 8, from the input voltage U_He so that you get at exit 5, 6:<maths id="math0023" num=""><img file="EP0169488A2_D0025.tif" /></maths>In this case, Δ applies to the induced voltage^U₂ = W₁U_E/ (W_W⁺ w₁). Since the turn ratio of the first winding 9 to the further winding 11 is typically less than 1: 7 according to the invention, the voltage ΔU induced in the second switching state is therefore₂ always slightly smaller than the voltage ΔU induced in the first switching state₁. However, in practice the increase in the output voltage U which can be achieved with the circuit arrangement according to FIG. 1 in the first switching state_A compared to the input voltage U_E with very good accuracy equal to that achievable in the second switching state

Voltage reduction can be set because the first winding 9 represents a complex resistor for the load current flowing from the voltage source 1 to the load 7. Since the first winding 9 generally comprises very few turns, this resistance is low, but it does lead to a certain voltage drop, which is independent of the switching state of the actuating unit 4. The value of U_A is in both switching states slightly below the values ​​that result from the above simplified equations. The output voltages that can be achieved in the two switching states are therefore symmetrical to the input voltage with good accuracy:<maths id="math0024" num=""><img file="EP0169488A2_D0026.tif" /></maths>

In a third switching state of the actuating unit 4, the two switch pairs 15, 16 are open and the short-circuit switch 17 is closed. The circuit 11 thus short-circuited further has a very low resistance, which, due to the fact that the turns ratio w 1 / ww is substantially smaller than 1, appears to be appropriately transformed down on the side of the first winding 9. As a result, the first winding 9 in this switching state represents an extremely small resistor for the load current, to which practically no voltage drops, so that the following applies with very good approximation:<maths id="math0025" num=""><img file="EP0169488A2_D0027.tif" /></maths>

or<maths id="math0026" num=""><img file="EP0169488A2_D0028.tif" /></maths>

Because of the extremely low voltage drop across the first winding 9, only a low voltage is induced in the further winding 11, so that despite the short circuit, only a relatively small short-circuit current flows through the further winding 11. The losses occurring here can be kept far less than 1% of the power delivered to the load 7.

Since the losses occurring in the control unit 4 are well below 1% of the load power even in the first two switching states, such a transformer circuit forms an extremely advantageous possibility, starting from a given input voltage U._E digitally three different output voltages U_A to provide.

In a fourth switching state, all switches 15, 16, 17 are open. The circuit of the further winding 11 then has an almost infinitely high resistance value which, even after step-down transformation on the side of the first winding 9, provides a high resistance value, so that a voltage drop dependent on the magnitude of the load current occurs at the first winding. This throttling effect of the first winding 9 in the fourth switching state can be used to limit the power supplied to the load to a safe level at least until a short circuit occurs at the load until further switch-off measures have been taken.

The two switch pairs 15 and 16 and the short-circuit switch are actuated by a switch control 23, which controls the switches 15, 16 and 17, which can be formed, for example, by triacs, in the required manner via the lines 25, 26 and 27. It is ensured that the switches 15, 16 and 17 are never closed at the same time and, on the other hand, the periods in which the switch is made from one switching state to another are kept as short as possible. In the event of a transition from the first or second switching state to the third or vice versa, the switch pairs 15 or 16 must be opened shortly before the time or closed shortly after the time in which the short-circuit switch 17 is closed or opened. In the case of a transition from the first to the second or from the second to the first switching state, an almost simultaneous closing and opening of the switch is not favorable, as will be explained in more detail below. Rather, a short time interval is maintained between the opening of the previously closed switch pair and the closing of the previously open switch pair.

In order to avoid that there is a drop in the output voltage U in these short switching intervals due to the above-described choke effect of the first winding 9_A 1, the transformer 8 has its own short-circuit winding 28 which can be short-circuited with the aid of a switch 29 which is parallel to it. This switch 29 is controlled by the switch controller 23 via a line 30 and is only closed for those periods during which the two switch pairs 15, 16 are temporarily open simultaneously when switching from one switching state to the other.

FIG. 2 shows a transformer circuit with an actuating unit 34, the structure of which differs from that of the actuating unit 4. In principle, however, the function of the actuator 34 is the same as that of the actuator 4.

The actuating unit 34 in turn comprises a transformer 8, the first winding 9 of which is connected between the input terminal 2 and the output terminal 5, while the other input terminal 3 is directly electrically connected to the other output terminal 6 via the terminal connecting conductor 10.

In contrast to the actuating unit 4 from FIG. 1 In the present case, the transformer 8 has two further windings 35, 36, one of which, as an additional winding 35, is permanently connected at one end to the end of the first winding 9, which is directly connected to the input terminal 2 in a galvanically conductive manner while the other end of the adding winding 35 can be connected to or disconnected from the connection connecting conductor 10 by means of a switch 37. The other of the two further windings is fixed as a subtracting further winding 36 with one end and is directly galvanically conductively connected to the end of the first winding 9, which is directly galvanically conductively connected to the output terminal 5 of the actuating unit 34, while the other end of the subtracting another winding 36 can be connected to or disconnected from the connecting connecting conductor 10 with the aid of a switch 38. The sense of winding of the three windings 9, 35 and 36, which are magnetically coupled to one another via the core 12, is identified by points 19, 20 and 21. It is chosen so that the voltage Δ U₁, which is induced by the adding winding 35 in the first winding 9 when the switch 37 is closed, to the input voltage U_E added, and that the voltage 0 U2, which is induced by the subtracting winding 36 in the first winding 9 when the switch 38 is closed, from the input voltage U_E subtracted.

A short-circuit switch 31, 32 is arranged parallel to each of the two further windings 35, 36, which short-circuits the associated further winding 35 or 36 in the closed state. The two short-circuit switches 31, 32 are controlled via a line 33 so that they are always open or closed at the same time. The switches 31, 32, 37 and 38 are controlled so that either only the switch 37 or only the switch 38 or only the switches 31, 32 are closed. The actuating unit 34 can thus be brought into the same three switching states as were described above for the actuating unit 4. Likewise, the actuating unit 34 can be brought into a corresponding fourth switching state by opening all the switches 31, 32, 37 and 38, which is not considered a "normal"_Be-drive state is used, but can be used to limit the load short-circuit current in the event of a load short circuit.

Basically, it would be sufficient to provide only one short-circuit switch 31 or 32 and to close it in order to produce the third switching state. However, if both other windings 35, 36 are short-circuited, only half the short-circuit current flows in each of the windings 35, 36, which enables a smaller dimensioning._If this outweighs the disadvantage of a second short-circuit switch is an optimization question that has to be decided in the specific individual case.

In any case, one switch is less required in the embodiment according to FIG. 2 than in the embodiment in FIG. 1, whereby the disadvantage of a second further winding is largely compensated for.

In addition, the embodiment shown in FIG. 2 offers the possibility of Δ U₁ within certain limits of Δ U₂ to be chosen independently, so that here the two output voltages U_A1- ^U_E + ΔU₁ and ^U_A2 = UE - ΔU₂ no longer necessarily symmetrical to the input voltage U_E must lie.

In the present exemplary embodiment, too, a switch controller 23 is provided, which outputs the control signals for the switches 37, 38 and 31, 32 via the lines 25, 26, 27. However, the lines 25, 26, 27 are not connected directly to the switches 37, 38, 31, 32, but rather are each connected to an input of an AND gate 39, 40, 41, the other inputs of which are controlled by sensor units 42. Each of the sensor units 42 has two input connections, with the aid of which it queries the voltage drop across the associated switch 37, 38 or 31. The purpose of these sensor units 42 and the AND gates 39, 40, 41 is to ensure that each of the two switches 37, 38 or the two switches 31, 32 can only be closed by a corresponding signal from the switch control 23 if the other switches have been opened beforehand.

If, for example, as shown in FIG. 2, the switch 37 of the actuating unit 34 is closed, then no voltage drops across this switch 37. Therefore, the associated sensor unit 42 generates a logic 0 signal at its output, which blocks the AND gates 40, 41 and prevents a closing signal from switch controller 23 from reaching switches 38 and 31, 32. These switches can therefore only be closed when the switch 37 has been opened, which is indicated by the sensor unit 42 in that it supplies the AND gates 40, 41 with a logic 1. Conversely, the same also applies, of course, to the interrogation of the closed state of the switches 38 and 31, 32 by the associated sensor units 42 and a corresponding blocking or release of the AND gate 39.

If triacs are used as switches 37, 38, 31, 32, these can of course not be controlled directly by the AND gates 39, 40, 41, but there is one between the output of these AND gates and the gate electrode of the triacs of the usual triac drive circuits is provided, which is omitted in Fig. 2 for the sake of clarity. The sensor circuits 42 are described in more detail below with reference to FIG. 3.

If the switching unit 34 shown in FIG. 2 is to be switched from the first switching state to the second or from the second switching state to the first, the previously closed switch 37 or 38 must be opened and the previously opened switch 38 or 37 must be closed a short time later become. The output voltage U_Athe setting unit 34 as quickly as possible and without the occurrence of additional voltage peaks or voltage dips from the old to the new amplitude value. In order to achieve this, it is expedient to open the previously closed switch 37 or 38 when the current flowing through the associated winding 35 or 36 has a zero crossing . If a triac is used as switch 37 or 38, this results in the switch opening at the correct time, ie automatically at zero crossing of the current by preventing re-ignition in the other direction after self-extinguishing of the triac at zero current crossing. A previously opened switch 38 or 37 is preferably closed at those phase angles of the magnetic flux passing through the winding 9 at which the change in this magnetic flux caused by the closing of the switch 38 or 37 is as small as possible. The phase angle of the magnetic flux, at which this criterion is met, depends on the load current, so that it cannot be given an exact value, but only a range. For the switch 37, this area is in the vicinity of the zero crossing of the magnetic flux, while for the switch 38 it is in the vicinity of the maximum of the absolute value of this magnetic flux.

To determine the most favorable closing times for the switches 37 and 38, the transformer 8 has a fourth winding, which serves as a sensor winding 43. When switches 37 and 38 are open, a voltage is induced in this sensor winding which has a constant phase shift with respect to the magnetic flux in the winding 9 which is independent of the load. This phase shift is constantly equal to 90 °, so that the switch 37 must always be closed in the area of ​​the absolute maximum of this voltage and the switch 38 in the area of ​​a zero crossing of this voltage. The information required for this is supplied to the switch controller 23 from the winding 43 via the lines 44.

An example of the sensor units 42 shown only schematically in FIG. 2 is explained below with reference to FIG. 3. 3 shows only the two connecting lines to the actuating unit 34, which supply the voltage dropping at the associated switch, for example at the switch 37, from above, and the line which, at the bottom, supplies the control signal for the two AND gates of the other switches, for the AND gates 40, 41 of the switches 38 and 31, 32.

The AC voltage dropping across the switch 37 in the open state is rectified with the aid of a rectifier 46, the DC voltage outputs of which are connected to one another via a resistor 47 and a photodiode 48 of an optocoupler 49. A phototransistor 50 of the optocoupler 49 is connected on the one hand via a resistor 51 to a supply voltage V and on the other hand directly to ground. The voltage which can be tapped off from the ground between the collector of the phototransistor 50 and the resistor 51 is fed via a line 52 to an inverter 53, the output of which is connected to the output line leading to the AND gates 40, 41, which the closing signals can come, release or block from the switch control 23 via the lines 26, 27.

If the switch 37 is open, the rectifier 46 generates a direct voltage from the alternating voltage then dropping at the switch 37, which causes the diode 48 of the optocoupler 49 to light up. The "low" signal then emitted by the phototransistor 50 is inverted by the inverter 53 into a "high" signal, which the AND gates 40, 41 enable.

On the other hand, if the switch 37 is closed, no AC voltage drops across it and the rectifier 46 does not generate any DC voltage. Thus, the diode 48 of the optocoupler does not light up and the phototransistor 50 emits a "high" signal, which is inverted by the inverter 53 into a "low" signal for blocking the AND gates 40, 41.

4 shows two actuating units 54, 54 ', which have an identical structure, which differs from the structure of the actuating unit 34 shown in FIG. 2 only in that the two short-circuit switches 31, 32 are omitted. As a result of this, the AND gate 41 from FIG. 2, which controls these two switches 31, 32, and the one of the three sensor units 42, which queries the switching state of the switches 31, 32, are also omitted. The two remaining AND gates 39, 40 accordingly only require two instead of three signal inputs. Otherwise, the basic structure of the actuating units 54, 54 'is the same as that of the actuating unit 34 and the corresponding parts are provided with the same reference numerals.

The two actuating units 54, 54 'are connected to one another in series, ie the output voltage U appearing at the output connections 5, 6 of the actuating unit 54_A is the input terminals 2 ', 3' of the actuator 54 'directly as the input voltage U_{E '} fed. _Da moreover, the input connections 2, 3 of the control unit 54 the supply voltage U emitted by the voltage source 1_v is supplied as the input voltage and the output voltage output at the output connections 5 ', 6' of the actuating unit 54 'as the load voltage U_L is applied to the load 7, the two first windings 9, 9 'of the two transformers 8, 8' seen from the voltage source 1 with the load 7 in series.

The fact that the two actuating units 54, 54 'have no short-circuit switches means that each of them can only be brought into three of the four switching states defined above. If the fourth switching state, in which the switches 37, 38, 37 ', 38' are all open, is left aside only in the event of a load short circuit, then only operating switching states remain for each of the two actuating units 54, 54 ' the first two switching states, in which they can be brought independently of each other.

This results in a total of four different switching state combinations for the transformer circuit shown in FIG. 4.

Because of the absence of the third switching state, each of the two actuating units 54, 54 'can input voltage U_E or U_E, only with a changed amplitude, ie either with an additive or a subtractive voltage change + ΔU₁ or - ΔU₂ or + ΔU₁'or - ΔU₂' pass on. Since the turns ratios of the further windings 35, 36 and 35 ', 36' to the respectively associated first winding 9, 9 'can in principle be determined independently of one another, a total of four different load voltages U can be given for a given supply voltage Uv_L produce.

Preferably, however, these turns ratios to form a pair of actuators are set so that the percentage increase in the output voltage U_AP of the pair compared to the input voltage U_EP of the pair, which results when the switches 37, 37 'are closed, is equal to the percentage reduction in the output voltage U_AP compared to the input voltage U_EP is, which results with closed switches 38, 38 ', and that the output voltage U_AP is, with great accuracy, equal to the input voltage UEP when the switches 37 and 38 'are closed, that is to say the front actuating unit 54, ie closer to the voltage source 1, is in the first switching state and the rear actuating unit 54' is in the second switching state. With this special switching state combination, the effects of the two actuating units 54 and 54 'mutually cancel each other out, so that the AC voltage emitted by the voltage source 1 is applied to the load 7 practically unchanged. It is of particular importance that this unchanged transmission of the input voltage of the actuator pair to the output takes place almost without loss, so that even if one connects several such actuator pairs in series, an efficiency of more than 99% can be achieved.

The pair of actuating units thus has four combinations of switching states, three of which correspond to the three switching states of the individual actuating units 4 and 34 described above:<maths id="math0027" num=""><img file="EP0169488A2_D0029.tif" /></maths><maths id="math0028" num=""><img file="EP0169488A2_D0030.tif" /></maths><maths id="math0029" num=""><img file="EP0169488A2_D0031.tif" /></maths>

The fourth switching state combination remains unused. With regard to the switching options, the function of such a pair of actuating units 54, 54 'is practically the same as the function of an individual actuating unit 4 or 34.

A pair of actuating units offers the advantage, however, that given the size of the voltage to be applied and thus the power to be switched, each of the two actuating units only has to manage half of this switching power and can therefore be dimensioned correspondingly smaller. Although one more transformer is required, the two transformers 8, 8 'of the pair of actuating units 54, 54' together are only slightly larger and heavier than the one transformer 8 of an actuating unit 4 and 34, respectively, with the same switching capacity. A single actuating unit 54 or 54 'is in any case considerably smaller and lighter than an actuating unit 4 or 34, ie this results in smaller and lighter sub-units, which brings considerable structural advantages to arrangements in which a large number of such actuating units or actuating unit pairs are connected in series. Transport is also much easier if you can break down such a system into several smaller and lighter sub-units. Two smaller units also have the advantage that they lead to smaller losses than a single unit with the same switching capacity.

To achieve particularly high switching speeds, such a pair of actuating units can also be constructed from two actuating units 174, 174 ', as will be described below with reference to FIG. 9.

5 shows a transformer circuit which is used as a single-phase voltage constant for the voltage U supplied to the load 7_L serves. It is assumed that a setpoint S. For the amplitude of the AC voltage supplied to the load 7_L is specified, which is set to 100% below, and from which the voltage actually applied to the load 7 may deviate by a maximum of + δ%. Furthermore, it is assumed that the supply voltage supplied by the AC voltage source 1_nowng _Uv in amplitude by + A% of the nominal value U_Vnenn can deviate. In principle, the setpoint S_L the load voltage U_L equal to the nominal value U_Vnenn the supply voltage U_v or be different from this nominal value. It is a particular advantage of the transformer circuit according to the invention that it enables the load voltage U_L also to a setpoint S_Lto regulate, for example, at or near the limit of the intended control range. However, this is only useful if deviations in the supply voltage can only occur in one direction. If the supply voltage is e.g. generated with the help of an inverter from a battery arrangement, this requirement is met without further ado, since the battery DC voltage and thus also the amplitude of the AC voltage generated therefrom cannot increase with prolonged operation with progressive discharge of the battery arrangement.

In the following, however, the first case (U_Vnenn = S_L) considered and assumed that Δ &gt;&gt; δ, so that a regulation of the amplitude of the supply voltage U_V to the setpoint S_L is required.

For this purpose, a transformer circuit according to the invention is provided between the voltage source 1 and the load 7, which consists of three stages 55, 56, 57 connected in series with one another, each of which is operated either by an actuating unit 4, 34, 144 or 174 according to FIG. 1 , 2, 8 or 9 or can be formed by a pair of actuating units 54, 54 'according to FIG. 4 or by a pair of actuating units which is constructed from two actuating units 174, 174' according to FIG. 9. The stages 55, 56, 57 are controlled with the aid of a switch controller 23 which is connected to each stage 55, 56, 57 via a pair of lines 61, 62. Depending on whether the stages 55, 56, 57 are formed by an actuating unit 4, an actuating unit 34, an actuating unit 144, an actuating unit 174, an actuating unit pair 54, 54 'or an actuating unit pair 174, 174', these line pairs symbolize , the lines 25, 26, 27 and 30 (see FIG. 1), the lines 25, 26, 27 and 44 (see FIG. Fig. 2), lines 158, 159, 160, 161 and 30 (see Fig. 8), lines 163, 164, 165 and 30 (see Fig. 9), lines 26, 27, 44, 26 ', 27' and 44 '(see FIG. 4) or twice the lines 163, 164, 165 and 30 (see FIG. 9).

The switch control 23 issues the switching commands to the switches of the stages 55, 56, 57 via the lines 61 and receives the information generated by the sensor windings 43 about the phase position of the magnetic flux in the first windings 9 of the transformers 8 and thus via the lines 62 the favorable closing times and periods for the switches. Furthermore, a first comparator 63 is provided, which has a reference voltage U at one of its two inputs_ref1 receives the setpoint S_L for the load voltage U_L represents. The other of its two inputs is supplied with the output signal of a first sensor 64 which detects the load voltage U_L measures. Via the line 65, the comparator 63 sends a difference signal to the switch control 23, which indicates whether and to what extent the load voltage U_L from the setpoint S_L deviates. Before this deviation from the permissible range + 6_% runs out, the switch controller 23 changes the switching states of stages 55, 56, 57, which thereupon the supply voltage U_V impress a new amplitude change and thus the load voltage U_L keep within the permissible control range ± δ%. In addition, a second comparator 66 is provided, which has a nominal value U_Vnenn the Ve^r- supply voltage Uv corresponding reference voltage U_ref2 with the output signal of a second sensor 67 which compares this supply voltage U_v measures. The differential signal emitted by the second comparator 66 is likewise fed via line 68 to the switch control 23, which can thus operate not only in the control mode but also in the control mode or in a combination of both. This has the advantage that in the event of a short circuit on the load side, ie at U_L = O, the switch control from the fact that Uv is still different from zero, can recognize the fault and does not try to load voltage U_L regulate up; instead, it can bring the control units of all stages into the fourth switching state defined above, in which the first windings 9 of all transformers 8 exert a strong choke effect and thus limit the load short-circuit current. The switch controller 23 preferably comprises a microprocessor for processing the information coming in via the lines 62, 68 and 65 and for converting this information into corresponding switching commands.

As already mentioned, the stages 55, 56, 57 are constructed in such a way that each stage increases the input voltage supplied to it in a first switching state or in a first switching state combination by a predetermined percentage, in a second switching state or in a second switching state -Combination reduced by approximately the same percentage rate and passed on unchanged in a third switching state or in a third switching state combination. In this sense, switching state combinations are also referred to in simplified form as first, second and third switching states whenever there is no explicit mention of actuator pairs.

The predetermined percentages by which the individual stages can change the input voltage supplied in each case differ from stage to stage and are preferably approximately in relation to one another in the form of integer powers of three. In the exemplary embodiment shown in FIG. 5, the last stage 57, which is closest to the load 7, can change the input voltage supplied to it, for example, by + A% or pass it on almost unchanged. The middle stage 56 can change the input voltage supplied to it by approx. + 3A% or pass it on almost unchanged, and the foremost stage 55 closest to the voltage source 1 can change the input voltage supplied to it by approx.<tables id="tabl0003" num="0003"><img file="EP0169488A2_D0032.tif" /></tables> For an exemplary embodiment in which each stage 55, 56, 57 is formed by a pair of actuating units 54, 54 'or 174, 174', this is shown in more detail in Table 3 for the case where + A%_%w + 1% is selected so that there is a possible amplitude change of approx. + 3% of the input voltage supplied to this pair for the actuating unit pair of the middle stage 56 and a possible amplitude change of approx. + for the actuating unit pair of the foremost stage 55 9% results.

As can be seen from Table 3, in order to achieve a change in the respective input voltage of a pair of actuating units that is as symmetrical as possible, the percentage voltage changes that can be achieved by the individual further windings are selected at least partially differently.

It applies to the pair of the foremost stage 55 that the adding winding of the actuating unit 54 or 174 can bring about a change of + 4.5%, while the subtracting winding can bring about a change of - 4.9%, and the adding one or subtracting winding of the actuating unit 54 'or 174' can impress a change of + 4.4% or - 4.2% on the input voltage of this rear actuating unit 54 'or 174' of stage 55.

These percentages are selected by a corresponding choice of the turn ratios so that the total changes shown in Table 3 on the right result for the three switching states used in the first stage, which are + 9.1%, -8.9% and + 0.1% . These three values ​​are selected at 0.1% higher than the target + 9%, - 9% and 0%. This compensates for the voltage drop that results when the load current flows due to the remaining throttling effect on the relevant first turns of these two actuating units 54, 54 'and 174, 174'.

The same applies to level 56, with values ​​that are approximately 0.02% to 0.03% higher, as can be seen in Table 3 without any problems.

In Table 4, similar to Table 2 on the left, the twenty-seven switching state combinations are listed again, which can be achieved with a transformer circuit comprising three actuator unit pairs according to FIG. 5, if only three switching state combinations are used for each actuator unit pair. In addition, Table 4 shows for each actuator 54, 54 'of the three actuator pairs whether the adding or subtracting winding is connected to the associated input or output voltage. A "1" means that the relevant further winding is connected to the associated voltage, while an "O" indicates that the winding can be opened by opening the relevant switch 37, 37 'or 38, 38' from the connecting connecting conductor 10 ( see Fig. 4) separately and therefore not connected to the input or output voltage. The number combination 1001 for a pair of actuating units thus means that in the front actuating unit, ie closer to the voltage source 1, the additive winding is switched on and the subtracting winding is switched off, while in the rear actuating unit arranged closer to the load 7, the additive winding switched off and the subtracting winding is switched on. A pair of actuators identified in this way is therefore in the third switching state combination defined above, in which the effects of the front and rear actuating units virtually cancel each other out, so that the input voltage appears at the output of the pair of actuating units with an almost unchanged amplitude.<tables id="tabl0004" num="0004"><img file="EP0169488A2_D0033.tif" /></tables> With the combination n = O, all three pairs of actuating units are in the state just described and one can see from the rightmost column of Table 4 that the ratio of load voltage U_L to the supply voltage U_v in this case 1.0014 is practically equal to 1.

In contrast, there are, for example, n = 13⁺ all three pairs of actuators in a state in which the adding winding is switched on in both actuators (first switching state combination identified by 1010). The right column shows that the load voltage U_L 13.52% greater than the supply voltage U_v is.

The switch controller 23 selects this combination when the supply voltage U_V has dropped sharply compared to the setpoint.

If one assumes that the deviation% of the load voltage from the nominal value S, which is set here equal to 100%, may amount to a maximum of + 0.5%, then the supply voltage U_v drop to 87.65% of this setpoint because the transformer circuit according to the invention reduces this decreased supply voltage U by 13.52% (based on U_V = 100%) can raise; the resulting value for the load voltage of<maths id="math0030" num=""><img file="EP0169488A2_D0034.tif" /></maths>lies at the lower limit of 99.5% (based on the setpoint) and thus within the permissible range. For the switching state combination n = 13^- applies accordingly that here the supply voltage U_V may have risen to 114.84% of the setpoint without the load voltage<maths id="math0031" num=""><img file="EP0169488A2_D0035.tif" /></maths>the upper limit exceeds 100.5% of the permissible range.

The same can be achieved for all other switching state combinations n. It is preferably always at such values ​​of the supply voltage U_v from one switching state combination to the next, in which the amplitude of the load voltage U_Lvor before switching and the amplitude of the load voltage U_Lnach after switching to setpoint S_Lare approximately symmetrical [see equation (14) _7 above. From the above values ​​it follows that in this exemplary embodiment fluctuations in the supply voltage U_v from + Δ = + 14.84% (based on the setpoint S = 100%) to -Δ = - 13.35% (also based on S = 100%) can be compensated so that the load voltage U_L fluctuates only within a range of S + 0.5%.

The corresponding — but to a limited extent — also applies if the stages 55, 56, 57 are formed by a pair of actuating units comprising two actuating units 174, 174 'according to FIG. 9 or by individual actuating units 4 or 34 or 144.

If a larger fluctuation range ± Δ% is to be detected, either the minimum amplitude change A must be increased, which is at the expense of the control accuracy δ, or the number of stages must be increased. It can be useful to add a level whose change range is not equal to the next integer power of three from A, here not equal to + 27A, but is only an integer multiple less than 27 of A, which is so large that if all four steps in the same direction, ie all additive or all subtractive, the required fluctuation range + ä can just be covered.

FIG. 6 shows a modification of the circuit arrangement according to the invention, as can be used to control the voltage output by a three-phase network.

As can be seen from FIG. 6, a transformer circuit 75, 76, 77 according to the invention is provided for each of the three phase conductors R, S and T, which is constructed in each case in the same way as the transformer circuit in FIG. 5 each of these three transformer circuits 75, 76, 77 from three stages 55, 56, 57 connected in series, each of which here consists of a pair of actuating units 54, 54 'and 174, 174' and can assume four different switching states. The AC voltage on each of the three phase conductors R, S and T can thus be subjected to change amounts that are in a ratio of 1: 3: 9, or the AC input voltage can be passed on unchanged or the load current can be throttled.

In order to be able to bring the stages of the transformer circuits 75, 76, 77 into the three different switching states as required, each of the transformer circuits 75, 76, 77 is not only with its associated phase conductor R, S or T, but also with the zero -N conductor connected. A three-phase network 80 is used here as the voltage source.

The voltage amplitudes supplied by the network 80 on the individual phase conductors R, S, T are continuously measured with the aid of a sensor arrangement 81, which supplies the three measurement signals to a comparator arrangement 82. There, the measurement signals with a common reference value U_ref compared. Alternatively, a separate reference value can also be specified for each phase conductor R, S and T.

The comparator 82 generates a separate difference signal for each of the three phase conductors R, S, T, which is fed to a switch controller 83. This controls via the line groups 85, 86, 87, the switches of the stages 55, 56, 57 in each of the transformer circuits 75, 76, 77 in the manner as has been explained in detail above. Of course, here too, each control unit is connected to the switch control 83 via several lines, as shown in FIGS. 1, 2, 4, 8 and 9. For the sake of simplicity, however, these lines have only been shown in FIG. 6 as a single bidirectional line.

A phase conductor R forms the output of each transformer circuit 75, 76, 77_K, P_K or T_{K '} the letter "K" indicating that an AC voltage with a constant amplitude is available on these phase conductors. These voltages can either be applied together to a single load that requires a three-phase current, or different loads, each of which only has to be operated with a 1-phase alternating current.

Alternatively, the sensor arrangement 81 can also be designed in a multiphase system in such a way that it detects the phase conductors R_K, P_K, T_K of the AC voltage or loads supplied.

With greater demands on the control accuracy or with even larger control ranges, more than three stages can also be provided for the transformer circuits 75, 76, 77. Analogously to FIG. 6, the circuit arrangement according to the invention can also be used in multi-phase systems which comprise less or more than three phases.

FIG. 7 shows a further embodiment of a transformer circuit according to the invention, which comprises only a single actuating unit 94. As in the exemplary embodiment shown in FIG. 1, the input connections 2, 3 of the actuating unit 94 are also used as the input voltage U here_E a supply voltage U_v supplied, which comes from a voltage source 1. An output voltage U appears at the output connections 5, 6_Athat a load 7 as a load voltage U_L is fed. Furthermore, the actuating unit 94 comprises a transformer 8, the first winding 9 of which is connected between the input terminal 2 and the output terminal 5, while the other input terminal 3 is directly electrically connected to the second output terminal 6 by means of the connecting connecting conductor 10. The transformer 8 also has a further winding 11 which is magnetically coupled to the first winding 9 via the iron core 12 of the transformer 8.

In contrast to the exemplary embodiment in FIG. 1, the actuating unit 94 of the present exemplary embodiment can be brought not only into four but into thirty-four different switching states, so that it is possible to make a total of thirty-two different amplitude differences between the input voltage U_E and the output voltage U_A to generate the one control unit 94, the input voltage U_E To provide unchanged at the output connections 5, 6 or to throttle the load current in the event of a short circuit on the load.

This great variation possibility allows the transformer circuit shown in FIG. 7 to be used as a voltage regulator and / or voltage constant similar to the transformer circuits in FIGS. 5 and 6.

In Fig. 7 the use case is shown as a voltage regulator, in which in turn a sensor arrangement 64 via lines 95, 96, the output voltage U_A the actuator 94, which is equal to the load voltage U_L is fed. The sensor 64 transmits a measurement signal to a comparator 63, which uses this measurement signal with a reference voltage U_ref compares the setpoint S_L the load voltage U_L corresponds. The comparator 63 gives the difference between the measurement signal and the reference voltage U via line 65_ref representing the differential signal to a switch controller 23, which controls a switch group 98 consisting of fourteen switches via lines 97 in order to bring the actuating unit 94 into the different switching states, as will be explained in more detail below.

In order to be able to bring the actuating unit 94 into thirty-two further switching states in addition to the two switching states, in which the further winding 11 is either short-circuited or completely open, thirty-two control voltages U_S1 to U_S32 which are generated according to the invention with the aid of a single AC voltage source 100.

The AC voltage source 100 is formed by an additional transformer arrangement 101, which in the present case consists of six winding sections 104 to 109 which are electrically connected in series and are magnetically coupled to one another via a common transformer core 111.

One end of the series circuit consisting of the winding sections 104 to 109 is directly electrically connected to the one pole of the AC voltage source 1, to which the input connection 3 of the actuating unit 94 is also connected, which is connected via the connection^Gs-conductor 10 is directly electrically connected to the output terminal 6 of the actuator 94. The other end of the series circuit consisting of the winding sections 104 to 109 is connected via a line 114 to the second output connection 5 of the actuating unit 94. Thus, the output voltage U is connected to the series connection of the winding sections 104 to 109_A the actuator 94.

The series connection of the winding sections 104 to 109 has seven taps 121 to 127, of which the taps 121 and 127 are connected to the two outer ends of the series connection, while the taps 122 to 126 are each led out between two adjacent winding sections.

Each of the taps 121 to 127 is connected to a pair of on / off switches from the switch group 98. In the closed state, one switch of each pair of switches connects the associated tap to a line 129 which is connected to the lower end of the further winding 11 in FIG. 7. In the closed state, the other switch of each pair connects the associated tap to a line 130 which is connected to the other end of the further winding 11. As already mentioned, all switches of the switch group 98 are controlled by the switch control 23 via the lines 97 in such a way that the control voltage U just required is always present on the further winding 11_S1 to U_S32 is present, or that the two switches of any pair are closed at the same time to short-circuit the further winding 11, or that all switches 98 are open to throttle the load current.

For a symmetrical regulation of the load voltage U_L around the setpoint S_L the actuator can be brought into thirty-two different switching states, sixteen of which are used to additively impress the respectively induced voltages ΔU₁ to ΔU₃₁ and sixteen for the negative impression of the respectively induced voltage ΔU₂ up to 4U32 are provided. The amplitude of each positively impressed voltage is equal to the amplitude of a corresponding negatively impressed voltage.

Since the sign of the stamping results from the winding direction with which the further winding 11 is connected to a control voltage, there are therefore only sixteen control voltages U._S1 to U_S16 required with different amplitudes, since the further winding 11 can be placed with the help of the switches 98 with two different directions of the winding sense on the different taps 121 to 127.

In order to be able to tap the sixteen different control voltage amplitudes, the number of turns of the winding sections 104 to 109 are matched to one another according to a code which is optimized so that on the one hand the smallest possible number of winding sections 104 to 109 and thus also taps 121 to 127 and switches 98 is required, and that, on the other hand, the maximum required control voltage U_Smax between the most distant taps 121 and 127 can be tapped.

According to this optimized code, the winding section 109 has a number of turns such that when the series connection of all winding sections 104 to 109 the output voltage U_A of the actuating unit 94, a tap voltage 1 of this winding section 109. U_Xmin is tapped, the smallest required control voltage U_Smin corresponds.

By simultaneously closing the upper switch of the switch pair 132 in FIG. 7 and the lower switch of the switch pair 131, the smallest control voltage U required can be applied to the further winding 11_Smin are applied so that the voltage ΔU induced thereby in the first winding 9 of the transformer 8_min is subtractively impressed on the input voltage UE. If instead the lower switch of the pair 132 in FIG. 7 and the upper switch of the pair 131 are closed simultaneously, the same smallest control voltage U is applied to the further winding 11_Smin on, but the sense of winding is the other _Wickung 11 inverted compared to the previous case, so that now the induced voltage .DELTA.U_min is impressively impressed on the input voltage U. The same also applies to the control voltages that can be tapped between any other taps 121 to 127.

According to the optimized code, the number of turns of the other winding sections 104 to 108 are selected such that the following tap voltages are available between adjacent taps 121 to 126:<tables id="tabl0005" num="0005"><img file="EP0169488A2_D0036.tif" /></tables>

Together with the tension 1. U_Xmin on tap pair 126, 127 this gives the possibility of all control voltage amplitudes of 1. U_Smin until 16. U_Smin either directly at taps or between taps that are further apart, as shown in the following table 6:<tables id="tabl0006" num="0006"><img file="EP0169488A2_D0037.tif" /></tables>

It can be seen that the optimized code is also distinguished here by the fact that on the one winding section 109 located at the end of the series connection, 1 times the minimum tap voltage U_Xmin and at the winding section 104 at the other end, twice U_Xmin is tapped.

In order to obtain the same conditions as in the embodiment shown in FIG. 1, it can be provided in the embodiment shown in FIG. 7 that the line 114 is not firmly connected to the line 95 at point 140. Instead, two push-pull switches can be arranged here, with the help of which the end of the line 114 which is remote from the series connection of the windings 104 to 109 via corresponding lines either with the line leading from the output connection 5 to the load 7 or with the line from the voltage source 1 to the input terminal 2 leading line can be connected. These switches are then also controlled by the switch controller 23 in order to connect either the input voltage U to the series connection of the windings 104 to 109_E or the output voltage U_Athe actuator 94 to create. The former preferably takes place when a corresponding control voltage U is applied to the further winding 11_S1'...., U_S31 a voltage ΔU in the first winding 9₁, ..., ΔU₃₁ should be induced, which relates to the input voltage U_E additive imprints. To the output voltage U_A line 114, on the other hand, is preferably connected when a voltage Δ U in first winding 9₂, ... ΔU₃₂ should be induced, which relates to the input voltage U_E subtractive.

8 again shows a single actuating unit 144, which is constructed similarly to the actuating unit 4 from FIG. 1 and is connected in the same way to change the amplitude of an AC voltage between an AC voltage source 1 and a load 7. Circuit parts in Fig. 8, which are present in the same way in Fig. 1, again have the same reference numerals. In particular, also in FIG. 8th the transformer only has a single further winding 11 which is magnetically coupled to the first winding 9 via the iron core 12 of the transformer 8. Two switches 150, 152 and 151, 153 are connected to the two ends 13, 14 of the further winding 11.

If the switch 150 is closed, it connects the end 13 of the further winding 11 to the input terminal 2, to which the one end of the first winding 9 is also connected. If the switch 151 is closed, it connects the other end 14 of the further winding 11 to the output terminal 5, to which the other end of the first winding 9 is connected.

If the switch 152 is closed, it connects the end 13 of the further winding 11 to a line 155, with which the switch 153 also connects the other end 14 of the further winding 11 in the closed state. Between the line 155 and the connecting connecting conductor 10, a circuit arrangement 157 is provided, which can be a simple controllable off / on switch, but is preferably formed by a current limiting circuit, as will be explained in more detail below with reference to FIG. 10 .

With the help of the switches 150 to 153, the actuating unit 144 can be brought into four different switching states. In the first switching state, in which the switches 150 and 153 are closed, the input voltage U is applied to the further winding 11 and the current limiting circuit 157 lying in series therewith_E created. Since the limit value to which the current limiting circuit 157 limits the current flowing through it is chosen to be greater than the current which flows through the further winding 11 in this first switching state, the voltage drop across the current limiting circuit 157 is very small and it is practically the whole Input voltage U_E to the further winding 11 as a control voltage. The winding direction of the windings 9, 11 defined by the points 19, 20 is selected so that the voltage Δ U₁, which is induced in this first switching state by the further winding 11 in the first winding 9, to the input voltage U_E added. The voltage is thus obtained between the output connections 5, 6 of the control unit<maths id="math0032" num=""><img file="EP0169488A2_D0038.tif" /></maths>

The absolute amplitude of the induced voltage ΔU₁ is the turn ratio w_1/w_w the first winding 9 to the further winding 11 according to the equation AU1 = w₁U_E/ w_w fixed.

In the second switching state, which is shown in FIG. 8, the switches 150 and 153 are open and the switches 151 and 152 are closed, as a result of which the output voltage U is applied to the further winding 11 and the current limiting circuit 157 which is in series with it again_A the actuator 144 is placed. Since the current flowing through the further winding 11 in this second switching state is approximately equal to the current flowing through the further winding 11 in the first switching state, this current is also below that

Limit value of the current limiting circuit 157, so that its resistance is very small even in this second switching state and practically the entire output voltage U_A bears against the further winding 11. The winding direction of the further winding 11 is reversed compared to the first switching state. This subtracts the voltage ΔU₂, which is induced in this second switching state in the first winding 9 of the transformer 8, from the input voltage U_., so that you get at exit 5, 6:<maths id="math0033" num=""><img file="EP0169488A2_D0039.tif" /></maths>

In this case, ΔU applies to the induced voltage₂= w₁U_E/ (w_w+ w₁). It is therefore the voltage ΔU induced in the second switching state₂ somewhat smaller than the voltage ΔU induced in the first switching state₁.

In a third switching state of the actuating unit 144, at least the two switches 150 and 151 are closed, so that the further winding 11 with antiparallel winding direction to the first winding 9 and electrically parallel to this first winding 9 is at the same voltage as this. In this switching state, the transformer 8 is therefore short-circuited and the currents flowing in the two antiparallel windings 9, 11 attempt to build up a magnetic field in each case; however, these fields face each other and almost cancel each other out.

The leakage inductance of the first winding 9 can be kept so low that the first winding 9 only opposes the load current flowing through it in this switching state with its very small ohmic resistance, as a result of which the voltage drop occurring on the first winding 9 is very small. This means that applies in this third switching state<maths id="math0034" num=""><img file="EP0169488A2_D0040.tif" /></maths>

Only the slight voltage drop across the first winding 9 is available as the driving voltage for the short-circuit current flowing through the further winding 11, so that the short-circuit current through the further winding 11 also remains very low. Since the impedance of the further winding 11 is considerably greater than that of the first winding 9, the load current flows practically exclusively through the first winding 9.

If it is always ensured that the switches 152, 153 are both open when the switches 150, 151 are closed, the current limiting circuit 157 can be dispensed with, ie the conductor 155 can be connected directly to the connecting connecting conductor 10 in a galvanically conductive manner. However, this has the consequence that when switching over, for example, from that shown in FIG. 8th shown second switching state in the first switching state, the switches 151, 152 must first be opened, and that only when these switches are open with certainty, the switches 150, 153 can be closed. If, in the absence of a current limiting circuit 157, all four switches 150 to 153 are closed simultaneously, both the input voltage U_E as well as the output voltage U_A short-circuited, which would lead to impermissibly high short-circuit currents and to an undesirable breakdown of these voltages.

Without a current limiting circuit 157, the previously closed switches would first have to be opened when changing from one switching state to the other, which, if triacs were used as the switch, would only be possible when the current flowing through them was zero, and those for the new switch state to be closed switches are closed, for which specific times would have to be waited again, in which the lowest possible switching peaks in the output voltage U_A surrender. Overall, this leads to the fact that at the earliest after one and a half to two periods of the AC output voltage U_Athe new amplitude value is stable available.

To accelerate the switching processes, it is therefore advantageous to provide the current limiting circuit 157. In the case of a switchover process by which the actuating unit is to be switched from the second switching state shown in FIG. 8 to the first switching state, for example, the actuating unit 144 is first brought into the third switching state, which is done by closing the first switch 150. A short time later, the third switch 152 is then opened and the fourth switch 153 is then closed. The actuating unit remains in the third switching state since the first switch 150 and the second switch 151 are closed during this time. Short-circuiting of the two windings 9 and 11 by the further conductor 155 is avoided in that the two switches 152 and 153 are not closed at the same time. During the entire time that the actuating unit 144 is in the third switching state, the current limiting circuit 157 prevents the flow of an impermissibly large short-circuit current from terminal 5 or from terminal 2 to the connecting connecting conductor 10 via the simultaneously closed switches 151, 153 or switches 150, 152 closed at the same time. As the last step of the switching process, the switch 151 is then opened, as a result of which the actuating unit changes from the third switching state to the first switching state.

The same also applies to a switching process that leads from the first to the second switching state.

In the switching processes just described, the actuating unit 144 also briefly passes through the third switching state whenever it is intended to change from the first to the second or from the second to the first switching state. If the actuating unit 144 is to be kept in the third switching state for a longer period of time, the switches 152 and / or 153 are opened so that the input connection 2 or no more currents can flow from the output connection 5 to the connection connecting conductor 10 and thus the power loss is reduced even further.

In a fourth switching state, all four switches 150 to 153 are open, so that the circuit of the further winding 11 has a high resistance value, which provides a high resistance value even after transformation down on the side of the first winding 9. A voltage drop that depends on the size of the load current thus occurs at the first winding. This throttling effect of the first winding 9 in the fourth switching state can be used to limit the power supplied to the load to a safe level at least until a short circuit occurs at the load until further switch-off measures have been taken.

The switches 150 to 153 are actuated by a switch control 23, which controls the switches via lines 158, 159, 160 and 161. The switch controller 23 can obtain the information necessary for this from a comparator (not shown in FIG. 8) which compares the load voltage U_L and / or the supply voltage U_v compared with setpoints and, in the event of deviations, emits corresponding difference signals, as described in detail above. Furthermore, the transformer 8 of the actuating unit 144 comprises a short-circuit winding 28 which can be short-circuited with the aid of a switch 29 which is parallel to it. This switch 29 is also controlled by the switch control 23 via a line 30. According to the invention, this only takes place when switches 150 to 153 or Certain disturbances occur in the current limiting circuit 157, as will be explained in more detail below.

As an alternative to the embodiment just described, the current limiting circuit 157 in the actuating unit 144 can be omitted without the delays in the switching process mentioned above having to occur. This is achieved in that the two switches 152, 153, which are then directly connected again to the connecting connecting conductor 10, are each designed as a current limiting circuit, the limit value of which can be switched back and forth between the value zero and a value other than zero . If such a current limiting circuit is switched to the limit value zero, this corresponds to the open state of a switch. If, on the other hand, it is switched to the limit value other than zero, it only opposes the current flowing through it with a very small, constant resistance, as long as this current remains significantly below the limit value. This limit value is selected so that it is greater than the current which must flow through the further winding 11 and the relevant switch 153 or 152 in the first or in the second switching state.

A circuit arrangement which has the properties just described is explained in more detail below with reference to FIG. 10.

In the case just described, the switchover from the first to the second switching state or from the second to the first switching state takes place in such a way that the two previously open switches are closed simultaneously and a short time later the two switches which are open in the new switching state are opened simultaneously have to. If switches 150 and 151 are implemented with the help of triacs, this opening process must be used to wait until the next zero crossing of the current which flows through the relevant switch 150 or 151 before opening.

In this embodiment too, the actuating unit can be brought into the fourth switching state by opening all four switches 150 to 153 simultaneously.

FIG. 9 shows a transformer circuit with an actuating unit 174, the structure of which differs from that of the actuating unit 144, but which in principle has the same functions.

The actuating unit 174 in turn comprises a transformer 8, the first winding of which is connected between the input connection 2 and the output connection 5, while the other input connection 3 is connected directly to the other output connection 6 via the connection connecting conductor 10.

Similar to the setting unit 34 in FIG. 2nd Here, the transformer 8 also has two further windings 35, 36, of which one end, as an additional winding 35, is firmly connected at one end to the end of the first winding 9, which is directly connected to the input terminal 2 in a galvanically conductive manner, while the other end of the adding winding 35 can be connected or disconnected from a line 185 by means of a switch 180, which in turn is connected to the connecting connecting conductor 10 via a current limiting circuit 157. The other of the two windings is fixed as a subtracting further winding 36 with one end and is directly galvanically conductively connected to the end of the first winding 9, which is directly galvanically conductively connected to the output terminal 5 of the actuating unit 174, while the other end of the further subtracting Winding 36 can be connected or disconnected from line 185 by means of a switch 181. The sense of winding of the three windings 9, 35 and 36, which are magnetically coupled to one another via the core 12, is identified by points 19, 20 and 21. It is chosen so that the voltage ΔU_1' which is induced by the further winding 35 in the first winding 9 with the switch 180 closed, to the input voltage U_E added (first switching state), and that the voltage .DELTA.U₂'Which is induced by the further winding 36 in the first winding 9 when the switch 181 is closed, by the input voltage U._E subtracted (second switching state). Here, too, the limit value of the current limiting circuit 157 is selected to be greater than the currents which flow through the adding winding 35 in the first switching state or through the subtracting winding 36 in the second switching state. Thus, in these two switching states, the resistance of the current limiting circuit 157 is practically negligible and the total input voltage U lies_E or the entire output voltage U_A on the adding winding 35 or on the subtracting winding 36.

In order to be able to bring this actuating unit shown in FIG. 9 into the third switching state, it is necessary to close the two switches 180 and 181 simultaneously, as a result of which the two further windings 35, 36 are connected in series with one another with the same winding sense and with an anti-parallel winding sense are connected in parallel to the first winding 9. Since in this switching state the two further windings 35, 36 can be regarded as a single winding, the same switching state is obtained as that described above as the third switching state of the actuating unit 144 from FIG. 8 and the input voltage U is also shown here_E passed on to the output of the control unit practically unchanged.

So that in this third switching state the input voltage U_E a current limiting circuit 157 is again provided between the conductor 185 and the connecting connecting conductor 10, which is again provided in principle by a controllable on / off, at the further winding 35 lying in the short circuit and drives an impermissibly high short circuit current from the input connection 2 to the input connection 3 Switch could be replaced. However, protection times would then have to be introduced and special verification circuits would have to be provided for switching from one switching state to the other, so that it is absolutely certain that switches 180 and 181 will be closed at the same time, as long as the switches connecting lines 185 and 10 are interconnected closed is. A current limiting circuit is therefore preferably used as the circuit arrangement 157, which automatically and without time delay prevents a further increase in the current flowing through it if this current threatens to exceed a predetermined limit value.

The actuating unit 174 can also be brought into a fourth switching state, as shown in FIG. 9. In this switching state, the two switches 180 and 181 are opened at the same time, as a result of which a strong throttling action of the first winding 9 occurs again, which can be used to limit the short-circuit current in the event of a load short circuit.

Switching from the first to the second or from the second to the first switching state also takes place here in such a way that the one of the two switches 180, 181 that was previously open is closed, and only then is the switch closed until then opened. The actuating unit 174 therefore also briefly passes through the third switching state with each transition from the first to the second or from the second to the first switching state.

So that when the third switching state is to be maintained for longer times, the power loss can be kept particularly small, it is provided in this embodiment that the current limiting circuit 157 is controlled via two lines 163 by the switch controller 23 so that its limit value is significantly smaller Value, preferably assumes the value zero. The current limiting circuit 157 then acts. an open switch and practically only the very small short-circuit current flows, which is driven by the small voltage drop across the first winding 9 in the two further windings 35, 36.

The switches 180, 181 are controlled by the switch control 23 via the lines 164, 165.

The transformer 8 of the actuating unit 174 also has a short-circuit winding 28 which can be short-circuited via a switch 29 which is controlled by the switch controller 23 via a line 30.

Because of the construction according to the invention, it is possible in certain malfunctions to keep the actuating unit 174 at least partially functional or at least to control it in such a way that it outputs its input voltage unchanged at the output connections 5, 6. If the control unit forms a link in a longer chain of control units, which are used overall as voltage constants, at least the other control units remain functional and the entire transformer circuit can maintain its control or regulating function, albeit to a limited extent. This is explained below for some typical malfunctions:<ul id="ul0003" list-style="none"><li>1. Short circuit in switch 180 or 181: Such a short circuit means that the switch in question can no longer be opened, that is to say if the control unit were not designed according to the invention, it would always remain in the first or second switching state. Assuming that, for example the switch 180 is permanently closed, the switch 181 can be closed and the current limiting circuit 157 can be switched to the smaller limit value due to the presence of the current limiting circuit 157 in all cases in which no additional impressions of the voltage induced in the winding 9 are desired. The control unit then changes to the third switching state and outputs the input voltage unchanged at the output. If the switch 181 is opened again and the current limiting circuit 157 is switched back to the larger limit value, the actuating unit returns to the first switching state. In spite of the disturbance, it can therefore still be switched back and forth between the first and the third switching state and the amplitude of the output voltage U_A change accordingly. In such a case, however, the second switching state can no longer be established. The same applies if a short circuit occurs in the switch 181, but the switch 180 remains functional. In this case, the actuating unit 174 can be switched back and forth between the second and third switching states, but can no longer assume the first switching state.</li><li>2nd Simultaneous short circuit in switches 180 and 181:<ul id="ul0004" list-style="none"><li>In this case, the current limiting circuit 157 is switched to the smaller limit value and the actuating unit remains permanently in the third switching state, in which the output voltage is equal to the input voltage. However, it can then no longer be brought into the first or second switching state.</li></ul></li><li>3rd Should a short circuit occur simultaneously in the two switches 180 and 181 and in the current limiting circuit 157, a very high short circuit current would initially flow from the connection 2 to the connection 3. In this case, a fuse 167 is connected in series with the current limiting circuit 157, which then blows and thus finally interrupts the connection between the lines 185 and 10. Because of the short circuit in the two switches 180 and 181, the actuating unit is then in the third switching state.</li><li>4th Line interruption in the current limiting circuit 157:<ul id="ul0005" list-style="none"><li>If the current limiting circuit 157 no longer allows current to flow due to a fault, then the switches 180 and 181 are permanently closed by the switch controller 23 and the actuating unit 174 is kept permanently in the third switching state which results in this way.</li></ul></li><li>5. Line interruption in one of the switches 180 and 181 If one of the two switches 180, 181 can no longer be closed, the strong throttling effect of the winding 9 formed above would occur whenever the other switch had to be opened. Because of the voltage drop that occurs at the choke 9 in this state, a voltage constant or voltage regulator in which an actuating unit shows this fault would practically no longer be able to perform its function. To prevent this, the short-circuit winding 28 is provided, the switch 29 is then closed. The actuating unit 174 is thus again in the third switching state; it can thus continue to be switched between the third switching state and one of the other two operating switching states.</li></ul>

The malfunctions just described can also occur in the actuating unit 144 shown in FIG. 8 and, due to their construction according to the invention, can be partially overcome in a manner similar to that just described. Of course, a fuse 167, which is connected in series with the current limiting circuit 157, can also be provided in the actuating unit 144 shown in FIG. 8.

A current limiting circuit 157 is shown in FIG. 10, as can be used in the actuating units 144, 174 in FIGS. 8 and 9.

This current limiting circuit has two current connections 187, 188, one of which is directly electrically connected to line 155 or line 185 and the other to the connecting connecting conductor 10. A series circuit is arranged between the two current connections 187, 188 and consists of the source-drain path of a first V-MOS transistor 190, two resistors 192, 193 and the source-drain path of a second V-MOS transistor 191 . In parallel to this series connection, two diodes 198, 199 are connected between the two current connections 187, 188, the forward directions of which are opposite to each other. The connection point 196 of the two diodes 198, 199 is electrically connected to the connection point 195 of the two resistors 192, 193.

Since each of the two transistors 190, 191 has a diode characteristic, that is to say can only develop its blocking effect in one direction, the two transistors 190, 191 are arranged such that their forward directions are parallel to the forward direction of the diodes 198 and 199 lying in the parallel branch and thus are facing each other. This current limiting circuit 157 can also be used to limit an alternating current in the required manner.

The diodes 198, 199 are selected such that the voltage drop occurring at them when the rated current flows is smaller than the corresponding voltage drop across the parallel V-MOS transistor 190 or 191. Since each diode 198 and 199 is not only the one parallel to it V-MOS transistor 190 or 191 but also its associated series resistor 192 or 193 bridged, the half-waves of the alternating current to be limited either flow via the diode 198 and further via the resistor 193 and the V-MOS transistor 191 or via the diode 199 and further via the resistor 192 and the V-MOS transistor 190 on the one hand, the alternating current in each half-wave through one of the two V-MOS transistors 190 or 191 be limited as necessary; on the other hand, it is avoided that the half-waves also have to flow through the second resistor and the second V-MOS transistor, which are only required for limiting the half-waves with the other sign. The power loss occurring in the current limiting circuit 157 can thus be kept particularly small.

The gate voltage for the two transistors 190, 191 supplied from the switch control 23 via the two lines 163 is applied between the connection point 195 of the two resistors 192, 193 and the two gate connections of the transistors 190, 191. As a result, the voltage that drops across resistors 192, 193 when a current flows between terminals 187 and 188 is subtracted from the gate voltage. The size of this gate voltage is selected so that the current flowing from one of the two connections 187, 188 to the respective other connection cannot exceed a predetermined limit value.

In the case described above that the current limiting circuit 157 is to be switched to a second, smaller limit value which is practically zero, the gate voltage supplied via the lines 163 is chosen to be so low that it is below the threshold voltage U_TH of V-MOS transistors 190, 191, which thus practically no longer allow current to flow through their source / drain path.

As already mentioned, triacs can be used as switches 150 to 153 or 180 and 181. However, this means that these switches can only be opened when the current flowing through them passes through a zero crossing. It has already been pointed out that, according to the invention, switches which are open until then are closed when changing from one switching state to the other. Then both the actuating unit 144 and the actuating unit 174 are each in their third switching state. The respective short-circuit current then flows through the switches 150, 151 or 152, 153 or 180, 181 and it can only be switched to the subsequent first or second switching state when this short-circuit current passes through a zero crossing.

If the switch in question is then opened, the further winding 11 or one of the two further windings 35, 36 is connected to its control voltage U._E or U_A, which as a rule tries to force the flow of a current which is phase-shifted against the short-circuit current flowing up to that point, that is to say at the point in time at which the respective switch is opened, has no zero crossing.

11 shows a diagram of the curve shape of an oscillation period of the input voltage U_E, The short-circuit current I flowing in the third switching state_K, of the current flowing in the first switching state 1₁ and the current 1 flowing in the second switching state₂ shown. The amplitude of the short-circuit current is I_K shown greatly enlarged for clarity.

It should be expressly pointed out that the three streams I_K, I₁ and I₂ cannot flow at the same time, since the actuating unit 144 or 174 can only ever be in one of the three switching states.

For the following considerations, it is now assumed that the actuating unit 144 or 174 is in the third switching state, of which the input voltage U during the first half period shown in FIG_E, ie between the times t₁ and t₄ to be switched to the first switching state. For this purpose, switch 151 must be opened in the embodiment according to FIG. 8 and switch 181 in the embodiment according to FIG. 9. Since these switches from the short-circuit current I_K if they are implemented with the help of triacs, they can only flow through at time t₄ be opened in which the short-circuit current I_K goes through a zero crossing. It can be seen from FIG. 11 that the current I₁, which should flow through the further winding 11 or the further winding 35 immediately following the opening of the switch, has a value which is equal to the zero crossing value of the short-circuit current I_{K '} which flowed through this further winding 11 or 35 before opening, is considerably different.

It is clear that the current flowing through the further winding 11 or 35 in the new switching state does not jump from zero to the actually required value I_S can increase. Instead, a compensation current I_G induced, whose value initially equal to -I_S and that exponentially decays over a longer period of time. Several oscillation periods of the input voltage U_E last until this compensation current I_G has completely disappeared.

The compensation current I_G is added to the current driven by the input voltage U through the further winding 11 or 35. Since the transformer 8 is dimensioned so that the current which normally flows through a further winding connected to its control voltage is just below the saturation limit, the transformer is replaced by this additive compensation current I_G driven to saturation. This has the consequence that a voltage drop occurs in the switching process just described, which means that the transition from the old to the new voltage amplitude does not run completely smoothly, but rather that voltage peaks are impressed on the first half-wave of the output voltage U following the switching process are.

In order to avoid this disruptive effect, the invention provides for switches 150 to 153 and 180, 181 to also be constructed with V-MOS transistors instead of triacs, two of which are each connected in series with opposite polarity. These transistors have the advantage that the switch they form can be opened regardless of the size of the current flowing through them. It no longer has to be the next zero crossing of the short-circuit current I_K be serviced, but the transition from the third to the first or second switching state can take place at a much more favorable time.

As can be seen from FIG. 11, the optimal switching times would be the times t₂ or t₃0 because in them the short-circuit current I_K, which flows in the other windings concerned before switching, is equal to the current that should flow in the respective further winding after the switching process.

Since these ideal times t₂ or t₃are very difficult to measure technically, they can be approximated by the times t₂'or t₃'are replaced, in which the current which flows through the further winding in the first or in the second switching state has a zero crossing. These replacement times t₂'or t₃'are from the ideal times t₂ or t₃ not too far away. Since, as already mentioned, the amplitude of I_K 11 is greatly exaggerated, that is when using the replacement times t₂'or t₃'The current change required is not particularly large either.

Since the time intervals τ₁ or τ₂that the replacement times t₂'or t₃'from the nearest zero crossing of the input voltage U_E have load-dependent, they can not be stored once and for all in the switch control 23. Instead, they are measured whenever the actuator 144 or 174 is in the first or second switching state, and the measured values ​​are stored. If the next switch from the third switching state to the first or second switching state is then to be made, the time which has passed since the zero crossing t₁ the input voltage U_E the switching time t has elapsed, on which the switching process is to follow₂'or the switching time t₃'can be easily specified.

These measures make it possible for the output voltage of the actuating unit to pass through the new amplitude value in a completely undisturbed manner at the next half-oscillation.

Should an actuator, which is equipped with V-MOS transistor switches and a current limiting circuit 157 and which impresses a voltage change + AU on its input voltage in the first switching state and causes a voltage change of -ΔU in the second switching state, from the first to the second switching state or reversed, the total voltage change 2ΔU that occurs can be carried out in two steps; The first step, in which the output voltage is changed by (ΔU), takes place immediately, ie at the same time as the changeover signal is generated. This is done by switching the control unit to the third switching state by closing one or more switches which were open until then. The second half the required change is then accomplished within a period of time which, in the worst case, is equal to half an oscillation period of the input voltage U_E is. Assuming that U_E has an oscillation frequency of 50 Hz, so the total change can be accomplished within a maximum of 10 ms. Then the output voltage U_A stable their new value.

The same also applies if an actuating unit is to be converted to the first or second switching state after it has been in the third switching state for a long time. Since only one or two switches have to be opened during such a transition, after the changeover signal has been generated it is only necessary to wait until the next favorable switching time t₂'or t₃'occurs. Since each of these points in time is available twice per alternating voltage period, in the worst case it must be waited for a period of time which corresponds to the length of a half period of the alternating voltage before switching can take place. Although the change in the output voltage takes place in a single step, the magnitude of this change is only half the size of the total change which is made during the transition from the first to the second or from the second to the first switching state.

A particularly fast and precise switchover occurs when two of the actuating units 174 described above are connected in series to form a pair of actuating units.