Inductive Cooking Point Heating System

The invention relates to an inductive cooker heating system for cooking vessels or the like.

Induction heating systems have the advantage of very low-inertia heat generation directly in the cooking vessel, namely in the base of the cooking pot. The actual cooking appliance remains largely cold. The disadvantage is the relatively high construction expenditure and the control problems. As electronic compounds are required for the necessary high frequency production and the control thereof and as the dissipated heat in the electronics and the induction coil there is greater heating of the induction generating means, it has been necessary to place the conversion and control electronics separately from the cooker heating system. Thus, installation in normal cookers or hobs was impeded and therefore induction cooker heating systems were generally installed in special equipment.

The object of the invention is to create in a hot point heating system a low-loss, easily controllable or regulatable hot point heating system operating with limited self-heating.

The invention proposes a oscillation package control for the induction generating means.

Unlike in the case of conventional phase angle control, oscillation packet control according to the invention does not have a d.c. voltage component and therefore as no repercussions on the mains. There is essentially no harmonic formation, so that interference suppression is simplified. This is helped by the fact that the high frequency inverter is constructed in freely oscillating manner, i.e. its frequency changes in current and damping-dependent manner. This is unusual because the resonant circuit is controlled by a control clock frequency of a phase locked loop (PLL), but which takes over the resonant circuit frequency in power operation. The electronic switches also switch in the zero passage.

As a result of the control principle of the full half-wave control the inverter always operates at the maximum efficiency point. It therefore has in particular a high cooking-on (partial power) efficiency. Therefore the power components can be correspondingly smaller and be constituted by commercially available components. The control principle is optimized with respect to the repercussions on the mains and the influence on pacemakers is minimized.

Through the choice of the half-wave control in the basic pattern (partial intervals), which cover few mains periods (e.g. six half-waves) and the variation, repetition or combination of said partial intervals within an overall time interval of a few seconds (between 1 and 10 and preferably approximately 2 s), there is a considerable variation range of the power setting possibilities. The setting range can also be made non-linear over the setting path (toggle position), so that it can be ergonomically optimized. Smaller powers can be more precisely set.

Limited switching losses result from the freely oscillating rectifier. Therefore no overdimensioning of the electronic power switches or circuit breakers are necessary (IGBT's). There is no constant frequency and instead frequency modulation occurs due to saturation effects.

Fundamentally temperature monitoring is not required in an induction heating system, because the heat is only formed outside the latter, namely in the cooking vessel. However, from the latter heat can be transferred to the plate and therefore inadmissibly overheat the glass ceramic plate. It is difficult to sense said plate using conventional means. Thus, according to the invention, a novel optical measuring device is used for measuring the plate temperature. It contains an infrared sensor, e.g. a silicon photodiode, which carries out a temperature measurement utilizing Planck's radiation law. With increasing glass ceramic plate temperature there is also a rise in the maximum of the frequency of the irradiated photons (Wien's displacement law). As from a given temperature the energy of the irradiated photons corresponds to the spectral sensitivity of the sensor, so that an evaluatable signal is obtained, which is used for switching off or reducing the power of the heating system.

As such an overheating of the glass ceramic plate can only occur if the heating system is incorrectly used, e.g. by depositing an empty pot, the temperature limiting means must fulfil a barrier function, i.e. the cooking point must remain switched off when the temperature limiting circuit responds until it is manually disconnected and then reconnected again. This can easily be brought about by the control electronics, e.g. a microcomputer.

These and other features of the invention can be gathered from the claims, description and drawings, the individual features being realizable in an embodiment of the invention and in other fields, either singly or in the form of subcombinations, and can represent advantageous, independently protectable constructions for which protection is hereby claimed.

Embodiments of the invention are described in greater detail hereinafter relative to the drawings, wherein show:

FIG. 1 A plan view of an inductive cooker heating system component.

FIG. 2 A diagrammatic longitudinal section through the component.

FIG. 3 A cross-section.

FIG. 4 A block circuit diagram of the control and power supply of two induction coils.

FIG. 5 A part detailed diagram for the operation of an induction coil. FIGS. 6 & 7 Diagrammatic representations of a shield.

FIGS. 8a to d "Current over time" representations of different bask pulse patterns.

FIG. 9 A table representation of the individual power stages of basic pulse patterns.

FIG. 10 An explanatory diagram of a current/time pattern.

FIGS. 11a to 11b The current/time pattern and the associated on-periods of a pot detection testing cycle.

FIG. 12 A cross-section through a strand from which the induction coil is formed.

FIGS. 1 to 3 show a component 11 for two induction cooker 10. It is provided for placing under a plate 12, e.g. a glass ceramic plate. The component forms a compact, relatively flat, easily handlable constructional unit which, with the exception of the power supply and a setting and regulating member 27 with knob 26, which can also incorporate a power control device, contains all the elements necessary for operation. The component can e.g. be pressed from below against the plate 12 by not shown spring elements. Through this arrangement and the inclusion of all the essential components the induction heating system can also be placed instead of and in addition to conventional radiant cooking points in a glass ceramic cooking zone.

In a sheet metal tray 23 the component contains a cooling body 15, preferably a shaped aluminium part with a surface substantially closed at the top and cooling ribs 18 on the bottom, which form cooling channels 19 between them and run roughly along an axis 9 connecting the two heating systems 10. On the top the cooling body has recesses 29 in which are located induction generating means 14 and which are in each case associated with a heating system 10. On the underside of the cooling body is provided a mounting plate 16, which is e.g. screwed to the outer cooling ribs, so that the cooling channels 19 and further larger areas 28 serving as cooling channels on the underside of the cooling body 15 are enclosed. Electronic power control elements 21, preferably in heat conducting connection with the cooling body 15 are located therein. The mounting plate also carries electronic components, but mainly the elements used for control purposes and therefore working with relatively small currents and limited heating. Everything fits into a sheet metal tray. However, the mounting plate could itself form the lower cover. In the vicinity of a short marginal side 24 of the elongated, rectangular component 11 ventilation openings 25 are provided through which a fan 37 arranged in a recess of the cooling body 15 sucks air or blows it out after flowing through the cooling channels 19, 28. It is also possible to have a fan arranged centrally on the cooling body with an air outlet to two or more sides. Therefore the power control elements and the control electronics are directly cooled by the cooling air flow and the power control elements also give off their heat by conduction to the air-cooled cooling body.

The induction generating or producing means 14 comprise an induction coil 30 in the form of a flat, disk-like or circular plate, magnetic yoke means 31 positioned below it and a thermal insulation 32 on the side facing the plate and in the vicinity of which can be provided a shield 33.

The induction coil 30 contains strands 38 wound in helical and/or spiral manner and which are constituted by single conductors (cf. FIG. 12). The strands 38 are formed from several, preferably five to nine and in the present case seven elements 40, which are twisted together and in turn contain between five and nine and in the present case seven twisted together single wires. The individual conductors are electrically insulated against one another in conventional manner, e.g. by a heat-resistant varnish coating. The copper single conductors 39 have a diameter d between 0.1 and 0.4 mm, preferably 0.2 mm. This value applies to the presently preferred frequency of the current supplied to the induction coil of between 20 and 30 kHz, preferably approximately 25 kHz. For other frequencies it is possible to determine a basic value D of the single conductor diameter according to the following formula: ##EQU1## in which D is determined in meters. The electrical conductivity k of the single conductor material is given in. A/V*m, its permeability .mu. in V*s/A*m and the frequency f in 1/s. The preferred wire thickness d is between a 1/4 and 3/4 of the basic value D calculated according to this formula. It has surprisingly been found that with such shall single conductor diameters the power dissipation in the induction coil 30 can be significantly reduced.

On the basis of all existing findings and which have also been proved by theoretical calculations, the coil losses should decrease on reducing the diameter d to a value the same as the basic value D according to the above formula, but should then scarcely undergo any reduction. The theoretical findings considered to be proven up to now are based on the skin effect of a single conductor and determine for the aforementioned diameter an optimum quantity, because then there is a uniform flow through the total diameter despite the current displacement towards the surface. The basic value D corresponds to the penetration depth of the current in a conductor surface and due to the circular wire shape there is a simultaneous penetration from all sides and therefore a uniform current coverage over the cross-section. This theoretically based consideration has been surprisingly disproved by tests. It would in fact be preferable to have a diameter below 0.2 mm, i.e. smaller than half the basic value D, but the diameter reduction is limited by the mechanical working possibilities.

Tests have shown that the losses by eddy currents and ohmic losses in the single conductors due to the induction produced by the coil in the case of the wire thicknesses used up to now (basic value D of 0.4 mm at a frequency of 25 kHz) were 70 to 100 W, whereas they are halved in the case of a coil having the same power and a wire diameter d of 0.2 mm and are only roughly 40 W. Therefore the coil heating is much lower and, apart from not inconsiderable energy savings, it would be possible to eliminate otherwise occurring problems connected with coil insulation and heat dissipation from the coil.

The magnetic yoke means 31 formed from ferrite segments is also placed below the coil in the form of a flat, circular layer with a central opening 35. The magnetic field formed on the underside of the induction coil is closed with limited magnetic resistance, but high electrical resistant, so that also there the eddy current losses remain low. No significant induction field is formed on the underside of the induction generating means 14. The magnetic yoke means 31 also form a heat conducting bridge between the induction coil 30 and the cooling body on which they engage, so that the coil loss heat is immediately dissipated into the cooling body.

The thermal insulation 32 is in the form of a plate with a central opening 35 between the latter and the glass ceramic plate 12 and which covers the induction coil 30. It is made from a very good heat protecting and preferably also electrically insulating material, e.g. a pyrogenic silica aerogel, which is compressed or moulded into a plate.

It would appear to be unusual to shield the actual heating element, namely the induction coil, in thermal manner with respect to the heat-absorbing cooking vessel. Even if account is taken of the fact that the energy transmission takes place by induction and not by heat transfer, it would be thought that at least for the dissipation of the loss heat into the induction coil a very good heat closure to the load, i.e. the cooking vessel 13 would be advantageous. However, it has been found that the induction coil, particularly in the case of the aforementioned low-loss coil construction, generates so little heat that through a heat bridge to the load heat is removed from rather than supplied to the latter. As a result of the heat protection the induction coil is kept at a lowest temperature level, which is advantageous for coil design and insulation. There is also an efficiency improvement because the heat of the cooking vessel 13 is not carried off downwards through the glass ceramic plate. The thermal insulation 32 advantageously simultaneously forms an electrical insulation against the glass ceramic plate 12, which becomes electrically conductive at elevated temperatures.

In the vicinity of the central opening 35, which passes through the insulation 32, the induction coil 30 and the yoke means 31, is provided an optical sensor 36, which senses the radiation from the glass ceramic plate. Therefore indirectly the cooking vessel temperature which could become harmful to the glass ceramic plate by means of a contact-free measurement, which would be difficult to perform in the magnetic field of an induction cooking point. Therefore it is a question of a measurement of the cause of the thermal hazard to the glass ceramic plate, because the latter is only heated by the cooking vessel. The glass ceramic largely transmits the radiation and cannot therefore be measured in contact-free manner. However, in the case of other plate materials the latter could constitute the radiation source.

The optical sensor is an infrared detector, whose spectral sensitivity is in the infrared range. With increasing cooking vessel temperature there is a rise in the maximum of the frequency of the irradiated photons according to Wien's displacement law. As from a predetermined temperature the energy of the irradiated photons corresponds to the spectral sensitivity of the IR detector, so that an evaluatable signal is formed, which is then used for disconnecting or reducing the power of the induction heating system. For this purpose the optical sensors 36 of each induction cooking point act by means of comparators 41 on a microcomputer 42 (FIG. 4), one being provided in each case for the control and regulation of an induction cooking point. It is adjustable by means of the setting member with the knob 26 to a specific temperature or power stage. The optical sensors 36 can be silicon diodes.

Alternatively precision resistors could be applied to the plate, e.g. between the latter and the insulation in the coil area, if said resistors are not or are only slightly influenced by the magnetic field and any influencing can be compensated on a circuitry basis or in the measuring program.

The shield 33 is provided between the induction coil 30 and the glass ceramic plate 12. It can be located on or is advantageously embedded in the top or bottom of the thermal insulation 32. The shield e.g. comprises a wire or strip structure shown in FIGS. 4 and 6 and which is constructed in low eddy current manner. This means that the thickness of the individual structural elements 45 (wires, strips, etc.) is smaller than the current penetration depth at the frequency used and also the structures are not electrically closed. Thus, in FIG. 6 there is an open ring conductor 46 with inwardly projecting branches 45, which are of varying length, so that the entire surface is uniformly covered. The ring 46 is connected to an earthing device 34, e.g. by connection to the earthed sheet metal tray 23 of the component 11 (FIG. 1).

Without any significant losses occuring, as a result of the said shield, the electrical field formed around the induction coil is shielded in the upwards direction and consequently so is the stray electrical radiation. In addition, the discharge currents from the cooking vessel can be reduced. The shield could also be formed by an earthed resistance material layer. It is important that the material is not magnetic and for avoiding eddy current losses has a relatively high electrical resistance compared with metallic conductors.

FIG. 4 is a block circuit diagram and FIG. 5 a more detailed view relative to the power supply, regulation and control of the induction coils 30. FIG. 4 shows that the alternating current from the power supply 22 is supplied across a radio suppression means 50 and rectification means 51 to a common intermediate circuit 52, from where the supply takes place for the two inverters 53, which could also be referred to as high frequency generators, for each induction coil 30. The intermediate circuit and inverters are controlled by a control means 54, which in turn receives signals from the microcomputers (MC) 42.

FIG. 5 shows the circuit of an induction coil 30 in greater detail, in which the control, inverters 53 and induction coil 30 of a second cooking point, which is also connected to the intermediate circuit 52, are not shown so as not to over burden the representation. Reference should be made to FIG. 5 for circuit details.

Each induction coil 30 is located in a resonant circuit with a half-bridge circuit, i.e. there are two branches 55, 56, in each of which there is a capacitor 57, 58 and an electronic switch 60, 61. They can be IGBT components, i.e. electronic semiconductor components incorporating several transistor functions and which are controlled by the control means 62 and can switch extremely rapidly. A free-wheeling diode 63, 64 and a resistor 65, 66 is in each case connected in parallel to said power switches 60, 61. These elements form the inverters 53 constructed as a resonant circuit, upstream of which is connected the intermediate circuit 52 and the rectifying means 51. A rectifier bridge produces a pulsating d.c. voltage, i.e. in which by rectifying the mains alternating current sinusoidal half-waves of in each case the same polarity are combined. The outputs of the rectifier bridge 51 are applied to the two branches 55, 56. In the intermediate circuit there is a common capacitor 67 between the two branches and a resistor 68 switched by an electronic switch 69, which can be a MOS-FET, which in conjunction with the resistor ensures that there are no clicks when switching on the inverter and it discharges the intermediate circuit.

In the control or driving path to the switches 60, 61 is in each case provided a driving unit 80, which contains an isolation between the low voltage part 54 and the power side, e.g. by optical couplers. Moreover, it supplies the switches with the control energy. The latter is supplied by means of supply units 81, which are located in the branches of the resistors 65, 66 and which in each case contain a Zener diode 82, a diode 83 and a capacitor 84. The Zener diode limits the voltage to the control voltage necessary for the switches 60, 61 and the diode and capacitor serve as a rectifying means. This leads to a simple "mains device" for the switch driving energy, which obtains its energy from the resistor branch, i.e. from an energy source which is in any case provided. Therefore the resistors produce less loss energy and in spite of this the other conditions are not impaired, e.g. the current value at 70.

The represented resonant circuit in symmetrical circuitry could be replaced by one having asymmetrical circuit, in which in place of the two resonant circuit capacitors 57, 58 only one is provided. The resonant circuit only then takes energy from the mains half-side. However, this simpler circuitry could be advantageous in cases where precise radio suppression values do not have to be respected.

At a tapping point 70 between the induction coil 30 and the capacitors 57, 58 of the resonant circuit is connected a switching control 71 for the inverter 53, which contains a sample and hold element 72, a limit value memory 73, a comparator 74 and an on-off memory 75. This switching control is provided in order to immediately disconnect the induction heating system if no power decrease occurs, e.g. if the cooking vessel 13 is removed from the cooking point and is only to be switched on again when a cooking vessel is present. For this purpose, in relatively short time intervals, a check is made to detect such a presence and this takes place by measurement of the damping of the induction coil 30.

The switching on of the resonant circuit takes place in the zero passage of the mains voltage in accordance with a predetermined diagram, which is given by the microcomputer 42 and which will be explained hereinafter. The resonant circuit is controlled by means of the electronic power switches 60, 61, namely from the control 62. Prior to each half-wave of the generated high frequency voltage of approximately 25 kHz, in the zero passage there is a switching over between the said switches 60, 61. Thus, a completely freely oscillating inverter or inverted rectifier 53 is obtained, which has low switching losses. As will be explained, no phase angle control is used for power setting or regulating purposes. The frequency is not constant and can be adjusted in accordance with the saturation effects by frequency modulation. Therefore there is no need for the overdimensioning of the power switches 60, 61 and there is a limited harmonic generation.

Power setting takes place by means of an oscillation packet control. In normal operation the inverter is always switched on for a full mains half-wave. The basis for the power setting is the different power stages are determined by switch-on patterns, which consist of a combination of identical or different, intrinsically basic patterns of wave packets. Mains repercussions are minimized by the complete symmetry.

FIGS. 8 and 9 show an example of a pattern occupancy plan for such an oscillation packet control. A total time interval Z of 2.1 seconds is subdivided into 35 partial intervals T of in each case 60 milliseconds, i.e. six mains half-waves at a frequency of 50 Hz. There are in all four basic patterns of partial intervals T, shown in FIGS. 8a) to d) as "voltage over time" diagrams. FIG. 8a) shows a partial interval T with the designation * in which all six mains half-waves are present, i.e. it is a "full per" interval. FIG. 8b) shows a partial interval T with the designation X in which in all four mains half-waves are so distributed that in all there is a symmetrical distribution. Compared with the "full power" pattern according to FIG. 8a), the third and sixth mains half-waves are absent (in each case a positive and a negative half-wave), so that this partial interval X has a 2/3 capacity. FIG. 8c) contains only two mains half-waves, namely the first positive and the fourth negative, so that once again there is a symmetrical distribution. The partial interval T with the designation. Y consequently has a 1/3 power. FIG. 8d) shows the zero power, i.e. during this partial power interval 0 no power is provided.

FIG. 9 shows the occupancy plans using the 35 partial intervals T, which together form the time interval Z of 2.1 seconds. In exemplified manner there are different power stages, e.g. corresponding to the toggle position of the knob 44 and with which are associated the different combinations of basic patterns in accordance with FIG. 8, in each case arranged in series. The following power release percentages reveal that in this way the power characteristic in the case of a power-controlled induction heating system can be adapted at random to the practical requirements. Thus, e.g. the power in the lower setting stages can be regulated much more finely than in the upper stages, which is in accordance with practical requirements. As each basic pattern Y according to FIG. 8c) only corresponds to less than 1% power within the time Z, the power can be adapted on a percentage basis. It is also possible to obtain completely irregular or non-constant paths, if this is appropriate. Nevertheless switching in the voltage zero passage is ensured. FIG. 8 shows positive and negative mains half-waves, as occur upstream of the rectifying means, to demonstrate the freedom from repercussions on the mains. In the resonant circuit there are mains half-waves in the form of rectified alternating current.

In the time interval Z, which is 2.1 seconds in the illustrated example, but can be of random length and subdivided into random partial intervals T, the basic patterns are randomly mixed controlled by the microcomputer and in this way produce a mains-side, d.c.-free control or regulation in relatively short pulses, but in each case containing a complete mains half-wave. The setting by means of the setting elements 43, as shown in FIG. 9, can be purely power-dependent, but there can also be influences on the part of temperature sensors or the like on the microcomputer, so that a control loop is obtained.

The start of the resonant circuit for producing the high frequency supplying the induction coil 30 commences in the zero passage of the mains voltage and amplitude and frequency in the resonant circuit change with the rise and fall of the current and voltage over the individual mains half-waves. Thus, at the start of each half-wave the frequency is higher and decreases in the vicinity of its maximum, because the inverter freely oscillates. Moreover, the frequency not only changes with current, but also with the pot material, because e.g. the inductance is not constant due to magnetic saturation in the pot bottom. If the inductance of the overall arrangement is lower, a higher frequency is obtained. This arrangement also has advantages in connection with radio suppression, because broad-band interference sources can be more easily suppressed. In addition, less harmonics are produced, because no phase gating is required.

The pot detection shown in FIG. 5 and which also protects the environment against excessive induction fields and provides a self-protection of the inverter, functions as follows. If with the heating system switched on the cooking vessel is removed therefrom, there is a pronounced rise in the current in the resonant circuit, because the damping decreases. The current in the inverter is tapped at 70 and detected by the sample and hold element 72. If it exceeds the limit stored in the limit value memory 73, then the inverter is disconnected by means of the control 62, in that the power switches 60, 61 are closed or not opened and this can also take place within a mains half-wave. The energy then present in the resonant circuit is returned via the free-wheeling diodes 63, 64 into the intermediate circuit 52. Therefore the disconnection takes place as a function of the current in the resonant circuit in an extremely rapid and loss-free manner.

Despite the switched on cooking point no power is released until a suitable cooking vessel is again placed thereon. This on-check takes place at the start of each time interval Z (e.g. 2.1 seconds). The testing process takes place as follows. In the control 62 there is a phase locked loop or PLL supplying the control clock frequency for the power switches 60, 61. During the operation of the resonant circuit it sets itself to the frequency of the main resonant circuit end alternately switches over the power switches 60, 61. Under no-load conditions, i.e. during the testing phase, the phase locked loop on excitation by the microcomputer and closure of one of the two switches 60 or 61 releases a semi-oscillation. Previously, by means of the resistors, 65, 66, the tapping point 70 was charged to a specific voltage and therefore a certain energy was present in the resonant circuit. On switching on one of the power switches current flows for a high frequency halfwave. The sample and hold element, e.g. a peak value detector which also contains a current converter in order to convert the currents flowing into measurement currents, measures the current during this preoscillation and stores the result end corresponds to the value i.sub.max in FIG. 10. In the resonant circuit the amplitude decays in accordance with the energy consumption through damping in accordance with a specific function (corresponding to an e-function). If this decay takes place too slowly, the damping is too low and power switch-on conditions do not exist. This is e.g. shown in FIG. 10, where a decaying oscillation is shown and the limit values G1, G2, G3 and G4 give values which could be stored in the limit value memory 73. If they are exceeded, this means "no adequate damping" and a signal "no switch-on" is given to the microcomputer.

Thus, the pot detection operates according to the damping measurement principle, testing only taking place with half the inverter, so that the power resonant circuit does not start and for this purpose it would be necessary to have an alternate switching on of the two power switches 60, 61. In the circuit embodiment according to FIGS. 4 and 5 the testing process takes place in such a way that from the first oscillation on switching on one of the power transistors 60 or 61 the current value is measured for a very short time E of e.g. 20 microseconds (roughly a half-oscillation in the idler frequency), is established by the sample and hold element and from this in the limit value memory 73 the following limit values, e.g. G1 to G5 are derived. Under the control of the microcomputer the phase locked loop PLL then introduces intervals P of the same order of magnitude and then switches on the power transistor again. From the current drop in the next oscillation (cf. FIG. 11a), by comparison with the limit values by means of the comparator 74, it is possible to establish whether the current exceeds these limits (here G2 and G3). The result of this check is buffer stored in the memory 75.

There is then a second switching on when the limits G4 and G5 are used for the comparison. This second measurement takes place for safety reasons, so as to avoid falsification by a pronounced frequency swing, e.g. in the case of an aluminium or copper article in place of a cooking vessel. If this measurement also reveals no exceeding of the limit values, then the damping is adequate and there is a power switch-on of the resonant circuit by the control 62. As the entire measurement only takes microseconds, the energy decayed in the resonant circuit, because it could not be replaced in this time by the high-ohmic voltage dividers 65, 66 connected in parallel to the power switches 60, 61. Up to the next test cycle at the start of the next time interval Z (after 2.1 seconds), the resonant circuit is supplied by said voltage dividers with the corresponding teat voltage and a new test can begin, if any exceeding of the limit values is established and therefore "too little damping" is detected and the resonant circuit was not switched in power operation.

Testing can take place with a very low testing current, e.g. 1/10 of the rated current in the case of power operation. Since also as a result of the very short on times of e.g. 20 microseconds within the test cycle of 2 seconds the resonant circuit is only in testing operation for approximately 1/100000 of the total time, the total power release during testing is an insignificant fraction of the total power of the cooking point and can be ignored from the energy standpoint and also with respect to the influencing of the environment. It is approximately 1 to mW in the case of a 2,000 W cooking point.

As a result of this pot detection by means of checking the possible power decrease (damping), there is a very reliable, quick-acting and test energy-low measurement. In place of current measurement in the resonant circuit, it is e.g. possible to use a voltage measurement on the resonant circuit capacitor, in order, by measuring the decay of the voltage amplitude, to carry out a comparison with the limit values determined on the basis of the initial measurement. Testing only takes place with half the inverter, so that the power resonant circuit does not start during the testing phase. If in the two successive measurements (second and third switching on of the PLL) both the values stored in the memory are found to be adequate for damping (limit values not exceeded), in the control 72 and accompanied by the timing of the phase locked loop PLLL, the resonant circuit is put into operation with full power by the alternate switching on of the power switches 60, 61. The power release then takes place in accordance with the power diagram explained relative to FIGS. 8 and 9 until either the cooking point is switched off by means of the setting element 43, or by removing the pot the self-protection comes into effect and the power is disconnected, so that it once again passes into the testing phase.