Power supplies are necessary to supply an electrical device with electrical power at a stabi- lized voltage which is different to the line voltage, e. g. 110 V or 230 V. A basic element of a power supply is a transformer transforming an input voltage to an output voltage. In common transformers, the output voltage is a fixed multiple of the input voltage with a factor given by the relation of the number of turns of a primary winding and the number of turns of a secon- dary winding which are magnetically coupled. Also, the voltage decreases for a increasing load at the secondary winding as the current is increased, too, and the secondary windings inherently comprise a resistor.

For applications in the field of welding, usual power supplies show the effect that the current through the light arc strongly depends on the distance of the electrode form the material to be welded. Especially for welding method with consuming electrodes, this requires for a very sensitive power control to avoid irregularities in the weld seam. Also, the weight and thus the need for core material of the transformers is high and should be reduced in view of portable welding devices.

Undesired influences of load and input voltage on the output voltage are usually compensated by an electronic controlling device measuring the output voltage and controlling the input voltage. If a decrease of output voltage is measured, the input voltage is increased or vice versa.

Further, high frequency components are produced, if the input voltage is provided by a chop- per power supply. Typical Power supply applications therefore have the need for wave filter- ing in order to remove high frequency components.

Document WO 97/29494 describes an transformer with multiple taps at the secondary wind- ings which are selectively connected on order to control the output voltage.

Another possibility to transform the AC output voltage by an rectifier to DC voltage and to control the DC voltage by means of Z-Diodes or similar devices in order to achieve a stabi- lized output voltage.

These techniques require a huge amount of controlling and stabilizing elements.

It is desirable to have a power supply with a reduced number of elements and preferable a transformer providing an output voltage that is widely independent from the load and the sup- ply voltage level. Further it is desirable to have a transformer capable of wave forming and provides a current limitation The invention therefore relates to an transformer according to claim 1 fulfilling the above mentioned object.

The invention relates to a transformers having a stabilized output voltage, whereby the stabi- lizing effect is inherently achieved by the electric and magnetic structure. The ouput voltage is stabilized by controlling the magnetic connection between at least a primary (input) winding and the secondary (output) windings.

At least primary winding, and at least a first secondary winding and a second secondary winding are each wound around one of three legs respectively and are magnetically connected by an upper with an air gap and by a lower yoke. A resonant circuit formed by the secondary windings and a capacitor provides a high saturation of the magnetic coupling between the secondary windings. The resonance is limited by the air gap which also controls the saturation as well as by a load current which is provided by the secondary windings to a connected load.

If the behavior is mainly characterised be the influence of the resonance circuit, the inductiv- ity related to the secondary windings forming a part of the resonance circuit appers to be

highly increased. For varying conditions, i. e. varying load or varying input voltage at the at least one primary winding, the total flux in the secondary windings and therefore the output voltage at the secondary windings is maintained constant by controlling the saturation of the magnetic connection which is mainly depending on the resonance circuit for low loads and depending on the load for high loads. In other words, the magnetic connection between the primary winding and the secondary windings and thus the power transfer between the primary winding used as input winding and the secondary windings used as output windings is influ- enced by the load as the additional flux produced by load and the flux produced by the reso- nance circuit control the level of saturation in the magnetic connection. Also, the attenuation of the resonance increases in parallel with the load. This controlling behavior can be influ- enced by an air gap which has an direct influence on the magnetic connection.

In one embodiment, one primary winding is carried by a first leg, while at least two secondary windings are carried by two legs, respectively. The secondary windings are connected with a capacity thus forming a resonant circuit. The legs are magnetically connected at their respec- tive ends by means of two yokes, whereby at least one of the yokes comprises an air gap be- tween the respective legs carrying the secondary windings. With this embodiment, the stabi- lizing effect can be achieved with a minimum of magnetic legs.

With this embodiment, the stabilizing effect can be achieved with a minimum of magnetic legs. In a first range of load, the voltage is constant and therefore any influences of the input voltage or the load are compensated. In a second range with a higher load, the voltage rapidly decreases and thus provides for current limitation. This range of operation can be used for welding applications as the current is constant for a wide range and thus compensates for changes caused by varying distances between electrode and welding surface. Therefore the transformer provides a very stable arc without any additional control means and thus the weld seam is more regular in comparison with results achieved with common welding power sup- plies. Additionally, the amount of core material can be reasonably reduced in comparison to common welding devices assuming the same welding power.

Further, a wave forming effect is provided by the transformer according to the invention re- ducing high frequency distortions, e. g. , if a chopped supply voltage is used. Also, the speed control of DC motors can be effected more easily without the need for costly rectifying de- vices.

A second embodiment comprises an additional leg and an additional primary winding. The two primary windings are located on two outer legs and the two secondary windings are lo- cated on the inner legs. The two primary windings are electrically connected and the secon- dary windings are electrically connected. This embodiment provides a symmetrical structure and provides constant output voltage at the primary windings over a wide range of input volt- ages supplied at the secondary windings.

Preferably, the secondary windings provide intermediate taps, providing a possibility to select the desired output voltage by choosing the number of turns of the secondary windings on both sides of the intermediate taps. Alternatively, the transformer comprises third and fourth sec- ondary windings, whereby the first and the second windings form the resonant circuit together with a capacitor and the third and the fourth windings provide the output voltage. In this way, the two functions of the secondary windings can be separated.

Further, the secondary windings supplying the output voltage can be connected in series. In this way, a higher total output voltage can be achieved. Also, the secondary windings can be connected in series, whereby the inductivities of the respective secondary windings add to- gether. Thus, a lower capacity value can be used for a resonance with the desired frequency, i. e. the supply frequency, e. g. 50 or 60 Hz. For an embodiment of the invenntion being sup- plied by a chopper supply and therefore with a high frequency voltage, the values of the ca- pacitor as well as the values of the inductivities provided by the secondary windings used for the resonance circuit can be reasonably decreased in comparison to applications with 50/60 Hz.

The magnetic connection of the legs preferably is provided by two yokes, one for each side of the legs. Theses yoke can comprise an air gap or a magnetic material causing a different qual- ity of magnetic connection between the legs carrying the secondary legs and the magnetic connection between the other respective legs. In this way, the level of saturation and the trans- former characteristic caused by the saturation can be controlled easily. Also, the transformer can be manufactured in a standard process.

The winding direction of all windings can be the same. This also simplifies the manufacture process of the transformers.

Preferably, two primary windings are connected in series. This reduces the number of neces- sary turns for a given output voltage with a given output voltage at the secondary windings.

Advantageously, the magnetic connection between the legs carrying the windings which form a part of at least one resonance circuit provide a magnetic leakage flux and/a ferromagnetic loss which both can be used to control the attenuation of the resonance circuit and therewith its behavior and influence on the stabilizing characteristics of the transformer.

Preferably all primary windings and all secondary windings have the same number of turns, respectively. This provides a complete symmetrical structure and allows for a simplified pro- duction process.

In order to influence the magnetic connection and the saturation in the magnetic material of the transformer, the cross the legs carrying the secondary windings can differ from sections of the legs carrying the primary windings. This also allows for a control the saturation and therewith the characteristics of the transformer.

An embodiment comprising an additional primary winding would allow for a more flexible connection of the transformer with an electrical power supply network.

Short Description of the Figures Fig. la shows a first embodiment of the transformer according to the invention having three legs comprising a primary winding and two secondary windings.

Fig. lb shows a second embodiment of the transformer according to the invention having four legs comprising two primary and two secondary windings Fig. 2 shows a third embodiment of the transformer according to the invention in which two secondary windings are used for a resonance circuit, each comprising an intermediate tap.

Fig. 3 shows a forth embodiment of the transformer according to the invention in which two secondary windings form a resonance circuit and another two windings are used as output winding.

Fig. 4 shows the behavior of the output voltage depending on a varying load for three trans- formers according to the invention having different air gap widths.

Fig. 5 shows the wave filtering and wave forming characteristics of one embodiment of the invention.

Fig. 6 shows the output voltage in dependency of a varying input voltage for different load situations.

Fig. 7 shown the output current characteristics dependent on a variable load resistance Detailed description of the Figures Figure la shows a circuit diagram of the first embodiment of the stabilized transformer ac- cording to the invention. The transformer comprises three legs 1,2, 3 which are parallel to each other and magnetically connected by two yokes 9,9'. This structure can be made of an usual EI-core. Two secondary windings 5,6 are electrically connected with a condenser 7 thus forming a resonance circuit having two inductivities which are represented by the secon- dary windings, each located on a different leg. The magnetic connection between the legs carrying the secondary windings 5,6 comprises an air gap 8. A primary winding 4 is carried by the first leg 1 while the secondary windings 5,6 are carried by the legs 2 and, respectively.

One of the legs carrying the secondary windings 5,6 is arranged as center leg of the EI-core.

Figure lb shows a circuit diagram of the second embodiment of the stabilized transformer according to the invention. The transformer comprises four legs 10,12, 14,16 which are par- allel to each other and are magnetically connected. The legs 10-16 have the form of a rod are of the same length. Each leg has two ends, an upper end and a lower end, all upper ends are magnetically connected by an upper yoke 50 perpendicular to the legs and all lower ends are connected by a lower yoke 52 which is also perpendicular to the legs 10-16 and thus parallel to the upper yoke 50. Regarding the legs 10-16, it can be distinguished between two outer legs 10,12 and two inner legs 14,16.

A first primary winding 20 with a first number of turns np is wound around a first outer leg 10 and a second primary winding 22 having a second number of turns np'is wound around the second outer leg 12. Both primary windings 20,22 are electrically connected in series and an input voltage Up can supplied to the primary windings 20,22. Driven by this input voltage Up, the first primary winding 20 generates a magnetic flux in the first outer leg 20 which is anti- parallel to the magnetic flux produced by the second primary winding 22 in the second pri- mary leg 12. It would also be possible to connect both primary windings anti-parallel, which would lead to the same effect. Alternatively, it would be possible to connect two primary windings in parallel, whereby the primary windings comprise winding directions, which are opposite to each other.

A first 30 and a second 32 secondary winding are wound around both inner legs 14,16, re- spectively, having the same winding direction. The secondary windings 30,32 are connected in parallel to each other, on one side by a direct electrical connection and on the other side of the secondary windings by a capacitor C, 40. The value of the capacitor C is selected in order to form a resonance circuit with the first and second secondary windings at the frequency of the output voltage, for example the rated (supply) frequency 50 Hz.

When the input voltage Up is supplied to the first and second primary windings (20,22), which are serially connected, each of them introduced a magnetic flux in the first (10) and the second (12) outer leg, respectively. Due to the magnetic connection provided by the upper (50) and the lower yoke (52) and the electrical connection, the fluxes produced by the input voltage are added in the outer legs (10, 12). Also, the flux induced by current floating through the secondary windings (30,32), is added in the inner legs (14,16), when a load is applied to the secondary windings.

Between the first and the second inner leg (14,16), the upper yoke (50) comprises a section having an air gap (60). At this section, the magnetic guidance differs from magnetic guidance of the remaining yoke. Magnetic guidance denotes the effective permeability and the therefore the ability of the magnetic connection, i. e. the yokes, to concentrate the magnetic flux in the material. This ability is strictly depending on the effective permeability of the yoke being de- pendent on the magnetic permeability, the geometry of the magnetic connection and by ef- fects causing a leakage field, e. g. air gaps.

The difference of magnetic guidance and therefore the effective permeability can also be caused by a yoke section having a wider or narrower cross section. Especially a tapering re- gion of the yoke causes a higher flux density and therefore, the material guiding the magnetic flux is at an higher saturation level than the remaining yoke material. Also, a section com- prising material having a lower permeability or magnetic material comprising a lower level of maximum saturation would cause the desired dependency of the magnetic connection on the flux located in the yoke.

When input voltage is applied to the two primary windings (20,22) a electromagnetic force EMF is induced in the first (30) and the second secondary winding (32) which are adjacent to the primary windings (20,22).

Some current is produced in the first (30) and in the second (32) secondary winding and in the capacitor C, 40, by the EMF and therefore an additional self-EMF is induced in the two sec- ondary windings 30, 32. This self-EMF is constructively added to the EMF induced by the current floating through the primary windings 20,22 and thus the voltage provided at con- denser C, 40 is additionally increased. If the voltage provided at condenser 40 is increased, the current applied to the secondary windings (30,32) is increased, leading to a resonance rise of the additional self-EMF. The limitation of the self-EMF depends on the saturation charac- teristics of the magnetic material of the yokes (50,52) (and the legs) which are not linear to the inducing magnetic field. In this way, the voltage Us provided by the secondary windings 30, 32 is not linear to the input voltage Up which is provided at the primary windings 20,22, and further the relation of the input and output voltages is characterised by the saturation pro- cess in which the resonance rise of the resonance circuit compensates for varying loads at the secondary (i. e. output) windings and for varying input voltages applied to the primary wind- ings.

In the embodiment shown in Fig. lb, the number of windings in the first primary winding np is equal to the number of windings of the second primary winding np. The number of wind- ings ns of the first secondary winding is equal to the number of windings of the second secon- dary winding In this example, np = nu is assumed for sake of convenience.

The first 30 and the second secondary winding 32 represent two inductivities which are con- nected in series. Therefore, their inductivity is added and connected in parallel to capacitor C,

40. The inductivity depends on the geometry of the secondary windings 30,32, the respective legs 14,16 as well as on the numbers of turns ns and Us ; further, the inductivity depends on other effective permeability which varies with the degree of saturation of the magnetic mate- rial of the legs 14,16. Also, the material of the yokes 50,52 is saturated for high magnetic fluxes. The combined inductivity of the first and second secondary windings 30, 32 and the value of capacitor C, 40, is chosen to be in resonance for the desired frequency of the output voltage Us, for example 50 Hz. Assuming the input voltage Up comprising considerable fre- quency portions nearby or at the resonance frequency, the secondary windings will be excited to a resonance at this frequency. The amount of energy is exchanged between the capacitor 40 and inductivity of the secondary windings 30,32 is resonance rised and limited by the maxi- mum magnetic flux at which a complete magnetic saturation of the magnetic material at the yoke section having the highest flux density is reached. Further, the resonance is attenuated by the load occurring at the output of the secondary windings 30,32 and by losses due to magnetic leakage.

Assuming an input voltage Up applied at the primary windings 20,22 and the load occurring at the second windings 30,32 shall be neglectable at the beginning. In this case, that magnetic connection, represented by the upper and lower yoke, is saturated by the nearly unattenuated resonance circuit formed by the secondary windings 30,32 and the capacity 40.

The resonance frequency of the resonance circuit should be at or nearby the frequency of the input voltage supplied at the primary windings and therefore the value of the capacity 40 (or the inductivity of the secondary windings) should be selected to fulfill this condition. The output voltage Us at the secondary windings 30,32 is limited by the maximum degree of satu- ration of the yokes 50,52, especially of the section 60 between both inner legs 30,32. With an increased load, the resonant circuit is attenuated by the load and therefore has a decreased influence on the saturation. The saturation is increasingly influenced by the high flux being introduced by the (load) current floating through the secondary windings 30,32 (and the load). Both effects compensate for each other and consequently a de-or increasing load is balanced by the in-or decreasing influence of the resonance circuit resulting in a nearly con- stant output voltage Us at the secondary windings 30, 32 for a varying load.

The behavior curve of the output voltage Up depending on a varied load (and therefore de- pending on the output current) is influenced by additional air gaps 60,260, 260'in one 50 or

in both yokes 50,52, taking into account the magnetic characteristics of the material as well as the ferromagnetic loss.

In Fig. 2, a third embodiments of the transformer according to the invention is shown.

It also comprises four legs 110-116 connected with two yokes 150,152 and comprises two primary windings 120,122 which are serially connected and arranged symmetrically at the outer legs (110,112). Further, two secondary windings also arranged symmetrically at the inner legs being connected in series by a capacitor 140 thus forming an resonance circuit.

Between the inner legs 114,116, one of the yokes 150are interrupted by an air gap 160.

In contrast to the embodiment shown in Fig. l, the secondary windings 130, 132 each com- prise an intermediate tap 170 at which the output voltage Us is supplied. The resonance circuit is formed by the complete secondary windings 130,132 and the capacity 140, whereas the output voltage Us is provided by only a part of the secondary windings 130, 132, respectively.

The relationship between the windings comprised by this part of the respective secondary windings and the number of turns of the total secondary windings 130,132 gives a possibility to determine the relation between the input voltage Up and the constant output voltage Us.

Also, the influence of the resonance circuit on the balancing between attenuation of the reso- nance circuit and saturation of the yokes 150,152 can be influenced on this way.

In Fig. 3, a fourth embodiment of the transformer according to the invention is illustrated.

This embodiment has a similar four-legged structure as the embodiments of Figs. 1 and 2, but the secondary windings 230,232, 234,236 have been divided into two functional parts which are not electrically connected. The first functional part is the resonance circuit formed by the capacitor 240 and first and second secondary windings 230,232. The second functional part of the secondary windings is comprised by third and fourth secondary windings, 234 and 236, both symmetrically arranged at two inner legs 214,216. These third and fourth secondary windings 234 and 236, are electrically connected and provide an output voltage Us. With this structure, the output voltage can be selected absolutely independent from the characteristics of the resonance circuit, which is formed by the first and second secondary windings 230,232 and capacity 240. The number of turns of all windings has been chosen to be symmetrical, i. e. np = np', ns = ns'and no = no'. The number of turn for at least one of the windings can also be asymetrical, depending on the application and desired behavior characteristics. Between the

inner legs 214,216, the yokes 250,252 each comprise an air gap 260,260'providing the de- sired influence on saturation characteristic and on attenuation of the resonance circuit.

Fig. 4 is a chart illustrating the influence of the width of an air gap. The dependencies be- tween secondary output voltage Us, which is provided by the secondary windings 30,32, 130, 132,230, 232 of the transformer according to the invention and the current Is floating through a load connected with the secondary windings is shown. For a first range of output current Is, which is denoted as (A) the voltage remains nearly constant at the voltage value Uc. In a sec- ond range (B), the output voltage Us rapidly breaks down for a small increase of output cur- rent Is. This behavior can be used for short circuit protection or for overload situations as well as for a constant current supply with a working point in range (B), e. g. for welding machines.

The behavior of the transformer regarding the output voltage Us and output current Is in range (B) can be interpreted as the behavior of a constant current source. Especially with regard to curve (2), the current remains nearly constant at a value Ic while the voltage varies over a wide range between Us and 0. The transformer according to the invention operating at this working range or having a working point at this steep decrease of output voltage U2 is espe- cially accomodated for welding applications. With a constant current Ic provided at the output of the transformer, the electric arc is maintained stable and thus less sparks are produced and the welding result is very homogeneous. In general, for power supply applications using an electric arc at varying loads, this characteristic of the transformer according to the invention provides very stable output characteristics. The characteristics of this second (shut down or constant current source) area (B) can be selected by amending the width of the air gap 60, 160,260, 260'whereby curve (1) shows the behavior for a large air gap, e. g. 3 mm, curve (2) shows the constant current characteristics for a narrow air gap, e. g. 2 mrn. and curve (3) shows the characteristics for a very small air gap, e. g. 0.5 mm. It is to be noted that the core structure of curve (1) introduces a higher amount of magnetic leakage flux than the core structures of curve (2) or (3) for a total flux comprised by the respective yokes. Therefore, the core structure of curve (1) has a higher influence on the attenuation of the resonance circuit as the magnetic leakage flux forms a major part of the attenuation effects on the resonance cir- cuit. Also, the higher leakage flux for reasonable loads influences the magnetic connection between respectively opposed primary and secondary windings, 20,36, rsp. 22 and 34 and therefore has an impact on the stabilized output voltage for high load situations.

In the case of neglectable load connected with the secondary windings 30,32, the stabilized output voltage at the secondary windings is defined by the maximum saturation of the mag- netic material and the magnetic leakage flux while an increasing load attenuates the resonance circuit keeping the voltage constant by the decreasing balancing effect between the resonance circuit and the saturation (i. e. the magnetic connection) of the yoke section between the two inner legs 14,16.

As shown in Fig. 5, the usage of a resonance circuit also enhances the frequency stability of the transformer combined with a wave forming feature as the resonance circuit has the char- acteristics of a band pass filter. The wave forming feature allows a sinus curve at the output of the transformer, which is nearly independent from a symmetrical input wave form or the load at the secondary windings. Any symmetrical waveform supplied at the input Uin, (i. e. the primary windings), is formed and as result, the output voltage Uout of the transformer com- prises a sinusoidal form.

Fig. 6 shows the output voltage characteristics of the transformer according to the invention.

The curves show the dependency of the voltage U2 at the secondary windings on the voltage supplied at the primary windings 20,22, which is denoted as Ul. When the load connected with the secondary windings 31,32 is neglectable and the input voltage supplied to the pri- mary windings 20,22 reaches a certain value, which is denoted as (b), the voltage at the sec- ondary windings reaches a limit value, denoted as (d), which is directly related to the satura- tion limit of the magnetic circuit. If the supply voltage at the primary windings increases above (b), an electromagnetic force is induced in one of the secondary winding 30 adjacent to the respective primary winding 20 according to the increased value of the voltage at the pri- mary windings 20,22. This electromagnetic force is equal to the resultant vector of the two electromagnetic forces induced in the concerned secondary winding 30 by the adjacent pri- mary winding 20 and the distant primary winding 22. For the other secondary winding 32, this process is provided in a symmetrical manner in the leg 16 around which the other secon- dary winding 32 is wound, induced by the flux of the respective adjacent primary winding 22 and by the flux of the respective distant winding 20. It is to be noted that the flux induced in an inner leg (e. g. 14) by the respective adjacent primary winding (e. g. 22) is always higher or at least different than the flux induced by the respective distant winding (e. g. 20) due to the air gap 60 between the inner legs 14,16. As described above, any means capable of decreas- ing the magnetic guidance can be used in the yoke sections between the two inner legs 14,16.

Consequently, a compensation between saturation produced by the load current at the secon- dary windings and saturation produced by the resonance circuit is established. Therefore the output voltage is stabilized at a nearly constant level (c).. (d), while the input voltage at the primary windings can vary on a wide range (b).. (a).

Due to the reasonable current in the resonance circuit formed by the secondary windings 30, 32 and C, 40, the output voltage at the secondary windings (30,32) is provided at a constant value as the current in the resonance circuit having a phase difference of nearly + 90° to the phase of the current through the load has a higher level than the load current at a small load.

Reducing the input voltage U1 would lead to less magnetic flux and thus to a lower level of saturation regarding the magnetic connection 50,52, 60. A lower level of saturation causes a better connection at the yoke section 60 between the two inner legs 14, 16 as well as a higher current oscillating in the resonant circuit as the resonant circuit is less attenuated. Due to this, the level of saturation is increased until the increased attenuation by saturation is in balance with the lower flux introduced by the primary windings 20,22 due to the lower input voltage.

This balancing effect leads to a stable output voltage, which is in wide ranges independent from the input voltage.

Also shown in Fig. 6 is the dependency between the input voltage and the output voltage for different load situations. Curve 1 shows the dependency between Ul and U2 for a neglectable load, Curve 2 shows the dependency between U1 and U2 for a reasonable load and curve 3 shows the dependency for a high load which is higher the loads referring to curves 1 and 2.

As described above, a certain input voltage Ul = b, corresponding to the output voltage U2 = d, a balanced situation is reached and for increased voltages, the output voltage remains nearly constant. It is to be noted that for a wide range b.. a of input voltage Ul, the input volt- age U2 is varied over a small range d.. c, only. The range in which the output voltage U2 is nearly constant, starts at a minimum input voltage Ul that depends on the load connected to the input (i. e. respective secondary windings) of the transformer. For a high load (curve 3), this minimum input voltage is higher than the minimum voltage for a low (curve 2) or ne- glectable (curve 1) load connected to the secondary winding.

Fig. 7 illustrates the behavior of the transformer according to the invention for varying loads showing a diagram of the input current I1 which is applied at the primary winding (s) and the output current I2 produced at the secondary windings for a varying load R.

For high loads, equivalent to R in the range of 0.. Rr, a stable current Is is provided at the output. This corresponds to the"overload"case shown in Fig. 4, range (B). For loads with a resistance larger than Rr, the current is continuously decreasing. Even if not shon in this dia- gramm, the corresponding output voltage at the primary winding (s) is constant. The range for R > Rr is equivalent to the range (A) in Fig. 4. For a better understanding, of the characteris- tics of the transformer, Fig. 7 should be regarded in connection with Fig. 4, whereby Fig. 4 shows the output voltage characteristics and Fig. 7 shows the output current characteristics for a varying load (i. e. for a varying output current).

The core of the transformer can comprise laminated, sintered or die casted magnetic material which saturates for high magnetic fluxes. Materials having different saturation characteristics can be provided for different parts of the transformer, i. e. a material 1 for the inner legs, a material 1'for the outer legs, a material 2 for the magnetic connection between both inner legs, referring to an embodiment with four legs. Magnetic materials such as Permalloy or other ferrite materials can be used and may be selected depending on the applied frequency range. The transformer according to the invention can also be used for high frequency appli- cations, e. g. with a chopper power supply as the wave forming characteristics of the trans- former assure a sinusoidal output wave form. In the case of high frequency input and output voltages, the core material used for legs and yokes can be selected accordingly. The core sec- tion of the core is preferably constant and can be squared, round or of similar form. The air gap can be formed by a yoke section with a reduced length leaving out a gap and thus influ- encing the magnetic connection. Also, ferromagnetic materials or a yoke section with a re- duced cross section can be used instead of or in combination with the air gap in order to achieve the desired magnetic properties.

In the shown embodiments, only one capacitor is used. It is also possible to use more than one capacitor in combination with one or more secondary windings. The capacitor (s) and the sec- ondary winding (s) form a resonance circuit which can be done by a parallel or serial connec- tion or an appropriate combination thereof. Having more than one capacity and thus more than one resonant circuit, the respective resonant frequencies can differ from each other while the resonant circuits can oscillate at least partly independently.

Also, magnetic materials comprising a magnetic hysteresis can be used for attenuating the resonance circuit, thereby transforming parts of energy oscillating between capacitor and in- ductivity into magnetic loss and therefore into heat. Further, the copper resistance of the windings can lead to a substantial loss of oscillating energy in the resonance circuit and has

also to be taken into account when the attenuations are regarded, which have an influence on the resonance circuit.

The primary and/or the secondary windings can comprise a number of intermediate taps to simplify an adjustment of the transformer according to varying voltage or load conditions and according to different applications.