Spark-plug and plasma generation

The present invention relates generally to the generation of plasma in a gas, and more particularly to the plasma generation systems between two electrodes. Plasma generation is used in particular for the controlled ignition of internal combustion engines by the electrodes of a candle.

The ignition of gasoline internal combustion engines, consisting of initiating the combustion of an air-fuel mixture in a combustion chamber of said engine, is relatively well controlled in current engines. In spark-ignition engines with indirect injection, conventionally, a spark plug and an upstream electronic device make it possible to generate a spark capable of transmitting to the mixture sufficient energy for its combustion. The formation of this discharge requires high breakdown voltages (of the order of 30 kV per mm), so that the inter-electrode space of the candles is limited to about 1 mm, a relatively unfavorable distance to the initiation of combustion.

To meet the standards of depollution, the car manufacturers have developed spark ignition engines capable of operating with poor fuel mixtures, that is to say having an excess of air relative to the amount of fuel injected. These developments have been applied in particular to direct injection engines, in which the injection of fuel is directly into the combustion chamber.

Conventional ignition devices do not apply well on lean-burn and direct injection engines. Indeed, the ignition devices are then very difficult to develop. A flame front propagates correctly in a very poor mixture (richness lower than 0.3) but the initiation of the combustion generally requires wealth higher than 0.7, and preferably for richness close to stoichiometry. It is therefore essential to maintain a sufficiently high richness in the inter-electrode space.

The generation of stratified mixtures has therefore been developed. In contrast to a homogeneous mixture where the richness is globally the same in every respect, a stratified mixture presents a richness which decreases as one moves away from the candle. The stratification of the mixture in the combustion chamber is for example obtained by guiding the jet of fuel so that the jet meets the candle at the time of the production of the spark. The guidance of the jet is obtained in particular by aerodynamic phenomena, generated for example by a suitable shape of the piston.

Stratified mixtures pose several problems. It is difficult to coincide the moment of spark and the presence in the vicinity of the inter-electrode space of a mixture cloud with a richness close to 1, in a globally poor mixing environment. In addition, the mixture around the candle at the moment of the spark has significant inhomogeneities of richness, variable over time, which do not guarantee the initiation of combustion at the time of development of the spark. The size and duration of spark of conventional candles then imply a misfire rate incompatible with the current performance and pollution requirements. In addition, the jet of fuel often strikes directly the candle, resulting in fouling of the insulation of the candle. This fouling promotes leakage currents between the central electrode and the mass. The generation of sparks is affected because the spark is short-circuited by a low impedance carbon path that reduces the potential difference between the spark plug electrodes.

New spark plugs on the surface produce larger sparks to deal with the problem of spatio-temporal rendezvous. Thus, a higher mixing volume is ignited. The probability of initiation of combustion is then greatly increased in a direct injection engine spark ignition and stratified mixture. Such candles are described in particular in patent applications FR97-14799, FR99-09473 and FROO-13821. Such candles generate large sparks from reduced potential differences. Surface spark plugs have a dielectric separating the electrodes in the zone where the distance separating them is the lowest; the sparks formed between the electrodes are thus guided on the surface of the dielectric. These candles amplify the inter-electrode field on the surface of the dielectric. For this purpose, the elementary capacitances formed by the dielectric and an underlying electrode are gradually charged. Candles generate a spark propagating along the surface of the insulation in areas where the electric field in the air is strongest. A conventional engine ignition device, coupled with such spark plugs typically generates sparks having a length of 4 mm with breakdown voltages of between 5 and 25 kV. When the candle globally has a symmetry of revolution about its main axis, the discharge has a probability of appearance substantially identical anywhere around the insulation. On the contrary, conventional candles generate an electric arc occurring systematically in the same extremely small volume. This method of ignition by plasma generation still has disadvantages. It occurs in particular a passage to the arc following a single line. The initiation of combustion is not optimal.

There is therefore a need, which the invention aims to satisfy, for a plasma generation system between electrodes solving one or more of these disadvantages.

The invention thus relates to a plasma generation system, comprising:<ul id="ul0001" list-style="dash" compact="compact"><li>two electrodes;</li><li>a dielectric material separating the electrodes;</li><li>a voltage generator, connected to the electrodes and selectively applying an AC voltage having a frequency greater than 1 MHz and an amplitude between peaks greater than 5kV between the electrodes.</li></ul>

The invention also relates to a method for generating plasma between two electrodes separated by a dielectric material, the method comprising a step of applying an alternating voltage having a frequency greater than 1 MHz and an amplitude between peaks greater than 5 KV between the electrodes.

The term branched plasma used in the following refers to the simultaneous generation of at least several lines or ionization paths in a given volume, their branches being moreover omnidirectional.

While a volume plasma involves heating up the entire volume in which it is to be generated, the branched plasma only requires the heating in the path of the sparks formed. Thus, for a given volume, the energy required for a branched plasma is much lower than that required by a volume plasma.

The invention makes it possible to easily generate a volume plasma or a branched plasma. Such plasmas can in particular be used to initiate fuel combustion in a combustion engine. The plasma thus generated can also be used in other applications, such as exhaust gas pollution control, air conditioning system disinfection, air filter cleaning or surface treatment.

In a general manner, the following will be understood by high density, any molar density greater than 2.5 * 10^-3 mol / L. Combustion density will be called any molar density of gas greater than 5 * 10^-2 mol / L. A stream of positive ionization propagating from the anode will be referred to as a streamer.

Other features and advantages of the invention will become clear from reading the following description which is given by way of non-limiting example and with reference to the figures. These figures show:<ul id="ul0002" list-style="dash" compact="compact"><li>Figure 1, a diagram of operation of a spark plug surface spark;</li><li>Figure 2, the representation of applied fields and the spark generated between the electrodes of the candle during ignition initiation;</li><li>Figure 3, a diagram of the electrostatic field between the two electrodes of the candle during ignition initiation;</li><li>Figure 4, a schematic representation of the development of a streamer for a single voltage rise (local field and global field);</li><li>Figure 5 is a schematic representation of an embodiment of the plasma generating system according to the invention;</li><li>Figure 6, an electric model used for the design of the series resonator;</li><li>Figure 7, a variant in which the amplifier comprises a mid-point transformer;</li><li>Figure 8, another variant of the system wherein the amplifier comprises a power transistor control by a bipolar transistor;</li><li>Figure 9, timing of signals during the excitation of the resonator of Figure 7;</li><li>Figure 10, the various elements of the power supply of Figure 7 integrated on the same circuit;</li><li>Figure 11 is a schematic representation of a control loop included in the amplifier;</li><li>12, a variant of the system comprising a control loop and generating circuits of the first voltage oscillations;</li><li>Figure 13, another system variant comprising a control loop and generating circuits of the first voltage oscillations;</li><li>FIG. 14, an example of a transformer forming a current probe of the amplifier, produced on a printed circuit;</li><li>Figure 15, an embodiment of a parallel inductance on a printed circuit;</li><li>Figure 16, another embodiment of a parallel inductance on a printed circuit;</li><li>Figure 17, a variant of a system having a common power supply and amplifier for two resonators;</li><li>Figures 18 and 19, schematic sectional representations of an example of candle usable in the plasma generation system;</li><li>Figures 20 to 27, various configurations of candle heads adapted for radio frequency excitation.</li></ul>

The invention proposes to apply an AC voltage with a frequency greater than 1 MHz and an amplitude between peaks greater than 5 kV between two plasma generation electrodes, separated by an element made of dielectric material. This excitation will be designated by the sequence of radiofrequency excitation. A volume plasma or a branched plasma is thus generated.

FIG. 1 illustrates details of the structure of a surface spark spark plug for which the application of radio frequency excitation is particularly advantageous. We will first detail the operation of such a candle.

The surface effect candle 110 comprises a spark plug head for opening into the combustion chamber in the lower wall of the cylinder head of an engine. The spark plug comprises a low-voltage cylindrical electrode which serves as a metal base 103 intended to be screwed into a recess made in the cylinder head of the engine and opening inside the combustion chamber. The base 103 is intended to be electrically connected to ground.

The base 103 surrounds a cylindrical high voltage electrode 106 arranged in a central position. The electrode 106 is intended to be connected to a generator of a high ignition voltage. The electrode 106 is isolated from the base 103 via an insulating sleeve 100. The insulating sleeve consists of a material whose relative permittivity is greater than 3, for example a ceramic. The spark plug has a gap 105 between the dielectric 100 and one end of the electrode 103.

The electrode 106 and the insulating sleeve 100 protrude by a length 1 outside the base 103. This length 1 substantially corresponds to the length of the spark generated when a high voltage is applied between the electrodes 106 and 103.

The base or low-voltage electrode 103 comprises in one piece a body and a connecting piece supporting a flanged flange 101. The flange 101 has a beveled edge extending in the immediate vicinity of the surface of the insulator 100.

The dielectric 100 creates an electrostatic field amplification in the air in its vicinity. The spark generated between the beveled edge of the flange 101 of the base 103 and a free end 104 of the central electrode 106 is propagated on the surface of the insulator 100, where the electric field in the air is the strongest. .

The formation of a spark is initiated by tearing in the middle of a few electrons subjected to a large electric field. When applying a large voltage between the electrodes, electrons of the collar are accelerated by the generated electrostatic forces and hit molecules of the air. The end of the flange is the area that undergoes the most important electrostatic field, and is therefore the starting point of the first avalanche. The molecules of the air release an electron and an ionizing photon in their turn of other molecules of air. A chain reaction ionizes the air in the space 105 between the electrode 103 and the dielectric 100. The gas space 105 allows for prior ionization with a potential difference between the electrodes 103 and 106 relatively limited.

A conductive channel is thus created, as illustrated in FIG. 2. The discontinuous lines represent equipotentials of the electrostatic field when a high voltage is applied between the electrodes 103 and 106.

FIG. 3 represents an example of an electrostatic field amplitude between the end of the flange 101 and the end of the electrode 106, A designating the end of the flange, B denoting the end 104 of the electrode 106. . Once the air is ionized at the end of the flange, the ionization of the air creates a space charge with a potential close to that of the flange and therefore behaves as an extension of that -this. During the propagation of the avalanche front, the electric field is amplified upstream of the front and favors the creation of new avalanches. The phenomenon self-maintains along the sleeve 100, to create a conductive ionized channel to the end 104 of the central electrode.

In the candle of FIG. 1, the insulator is separated from the electrode 103 by an air space. This space is not essential for the operation of the candle but facilitates the manufacture of the candle with a flange having a very sharp angle near the surface of the insulation. It also reduces the influence of fouling phenomena.

The physical phenomenon implemented thanks to radiofrequency excitation has similarities with the propagation described above but makes it possible to considerably improve the effects thereof. FIG. 4 diagrammatically represents the electrostatic field during the departure of an avalanche. It can be noted that the propagation of the avalanche is limited by the local field due to the separation of atoms and their electrons. This local field limits in particular the propagation of the discharge over long lengths. The present invention proposes, among other things, an electrical excitation capable of inverting the polarity of the imposed global field before the electrons have been able to recombine with the atoms present in the medium. At each alternation of the polarity, the electrons are more and more accelerated in the opposite direction. A polarization wave is thus propagated in an oscillatory manner at the frequency of the excitation, recovering at each period the charges deposited in the previous period. Each alternation then produces a propagation of the wave greater than the preceding one; it is thus possible to obtain sparks of very long lengths with voltage amplitudes between the relatively limited electrodes. Radio frequency excitation also suppresses breakdown voltage variations between successive cycles.

For automotive ignition applications, those skilled in the art will utilize electrodes and insulation having materials and geometry adequate to initiate combustion in a mixture at a combustion density and to resist the plasma thus formed.

Plasma thus formed has many advantages in the context of automotive ignition: significant reduction in the rate of misfires in a stratified lean mixture system, reduction of wear of the electrodes and adaptation of the ignition initiation volume to density function. It is found that the excitation described is adapted to achieve ignition of a mixture having a density greater than 5 * 10^-2 mol / L. For this ignition application, the generator applies the excitation between 1.5 and 200 times per second, with an application duty ratio of between 10 and 1000, and preferably between 72 and 720.

The radiofrequency excitation described is also suitable for a plasma deposition application, in a gas having a density of between 10 and 10.^-2 mol / L and 5 * 10^-2 mol / L. The gas used in this application may typically be nitrogen.

The radiofrequency excitation is further adapted to an application for the depollution of a gas having a density of between 10 and^-2 mol / L and 5 * 10^-2 mol / L.

The radio frequency excitation is furthermore suitable for a lighting application using a gas having a molar density of between 0.2 mol / l and 1 mol / l.

An envisaged plasma generation system mainly comprises three functional subsystems:<ul id="ul0003" list-style="dash" compact="compact"><li>a generator capable of resonating an LC structure at a frequency greater than 1 MHz with a voltage across the capacitor greater than 5 kV, preferably greater than 6 kV.</li><li>a resonator connected at the output of the generator and having an overvoltage factor between 40 and 200 and having a resonance frequency greater than 1 MHz.</li><li>a candle head comprising two electrodes separated by an insulator, for generating a plasma during the application of the radiofrequency excitation.</li></ul>

FIG. 5 represents an embodiment of a plasma generation system 1 and its voltage generator 2. The voltage generator advantageously comprises:<ul id="ul0004" list-style="dash" compact="compact"><li>a low voltage supply 3 (generating a DC voltage of less than 1000 V);</li><li>a radio frequency amplifier 5, amplifying the DC voltage and generating an AC voltage at the frequency controlled by the switching control 4.</li></ul>

The alternating voltage of the amplifier 5 is applied to the resonator LC 6. The resonator LC 6 applies the alternating voltage according to the invention between the electrodes 103 and 106 of the candle head.

The voltage supplied by the power supply 3 is less than 1000V and the power supply preferably has a limited power. It can thus be provided that the energy applied between the electrodes is limited to 300mJ by ignition, for safety reasons. This also clamps the intensity in the voltage generator 2 and its power consumption. To generate DC voltages greater than 12 V in an automotive application, the power supply 3 may include a 12 Volt to Y Volt converter, where Y is the voltage supplied by the power supply to the amplifier. It is thus possible to generate the desired DC voltage level from a battery voltage. The stability of the DC voltage generated is not a priori a decisive criterion, it can be expected to use a switching power supply to power the amplifier, for its qualities of robustness and simplicity.

One can also consider, according to one variant, to apply across the amplifier a voltage of 42 V taken from the electrical circuit of the vehicle. This is the level of tension that will prevail in future standards for future motor vehicles. This variant, avoiding the voltage conversion by the power supply 3, substantially reduces the cost and the complexity of the voltage generator 2.

This voltage generator makes it possible to concentrate the highest voltages on the resonator 6. The amplifier 5 thus deals with much lower voltages than the voltages applied between the electrodes: it is therefore possible to use an amplifier 5 at a reasonable cost and having characteristics similar to those common to mass automobile production, the reliability of which is furthermore proven. In addition, such a voltage generator has a relatively small number of components. There is thus a voltage generation system having a reliability, volume, weight and ease of production of interest, especially for large series in an automotive application.

The amplifier 5 accumulates energy in the resonator 6 at each alternation of its voltage. A class E amplifier 5 will preferably be used as detailed in US Pat. No. 5,187,580. Such an amplifier makes it possible to maximize the overvoltage factor. Such an amplifier performs out-of-phase switching with respect to the amplifier described in US Pat. No. 3,919,656, which aims to perform switching with zero voltage and / or intensity and does not optimize the overvoltage factor of the amplifier. Those skilled in the art will of course associate a switching device adapted to the chosen amplifier, to support the requirements of voltage increases and to have an adequate switching speed.

The preferred class E amplifier comprises a parallel resonator 62. This parallel resonator 62 is preferably made on the same card as the amplifier 5 and its switching control 4. The parallel resonator 62 temporarily stores power supplied by the amplifier 5, and periodically supplies this energy to the series resonator 61. With the supply voltage values ​​specified elsewhere, an amplifier 5 having an overvoltage coefficient of the order of 3 will be used. This overvoltage coefficient corresponds to the ratio between the voltage supplied by the low voltage supply 3 and the amplitude between peaks of the voltage applied to the series resonator. The overvoltage coefficient of the associated series resonator 61 is then preferably between 40 and 200. The overvoltage coefficient of the series resonator is notably limited by its loss angle.

A preferred dimensioning of the inductive and capacitive elements of the series resonator 61 will be explained. Figure 6 illustrates an electric model of this resonator. Thus, the inductance series 65 has in series an inductance L and a resistor Rs taking into account the skin effect in the radiofrequency domain. The capacitor 119 has in parallel a capacitance C and a resistor Rp. The resistance Rp corresponds where appropriate to the dissipation in the ceramic of the candle. When the series resonator 61 is powered by a voltage at its resonance frequency f₀ (1 / (2π√ (L * C))), the amplitude across the capacitor C is amplified by the overvoltage coefficient Q defined by the following formula:<maths id="math0001" num=""><math display="block"><mrow><mtext mathvariant="italic">Q</mtext><mtext> = </mtext><mfrac><mrow><mtext>1</mtext></mrow><mrow><mfrac><mrow><msqrt><mfrac><mrow><mtext mathvariant="italic">The</mtext></mrow><mrow><mtext mathvariant="italic">C</mtext></mrow></mfrac></msqrt></mrow><mrow><mtext mathvariant="italic">Rs</mtext></mrow></mfrac><mtext> + </mtext><mfrac><mrow><mtext mathvariant="italic">rp</mtext></mrow><mrow><msqrt><mfrac><mrow><mtext mathvariant="italic">The</mtext></mrow><mrow><mtext mathvariant="italic">C</mtext></mrow></mfrac></msqrt></mrow></mfrac></mrow></mfrac></mrow></math><img file="EP1515594A2_D0001.tif" /></maths>

From the equation ω₀²= (2π * f₀)²= 1 / (L * C), we deduce that the following equation must be checked to obtain the maximum value of Q:<maths id="math0002" num=""><math display="block"><mrow><mtext>Rs = Rp * L / C</mtext></mrow></math><img file="EP1515594A2_D0002.tif" /></maths>

We will take into account the following conditions:<ul id="ul0005" list-style="dash" compact="compact"><li>f₀ is of the order of 5MHz;</li><li>the values ​​Rs and Rp are constant;</li><li>Rp is mainly induced by the fouling of the candle head and is worth on average 50 kQ;</li><li>Rs is approximately 10 Ω taking into account the skin effect.</li></ul>

We then deduct<maths id="math0003" num=""><math display="block"><mrow><mtext mathvariant="italic">C</mtext><mtext> ≈ </mtext><mfrac><mrow><mtext>1</mtext></mrow><mrow><mtext>ω</mtext><msqrt><mtext mathvariant="italic">Rs</mtext><mtext> * </mtext><mtext mathvariant="italic">rp</mtext></msqrt></mrow></mfrac><mtext> ≈45</mtext><mtext mathvariant="italic">pF</mtext></mrow></math><img file="EP1515594A2_D0003.tif" /></maths> and<maths id="math0004" num=""><math display="block"><mrow><mtext mathvariant="italic">The</mtext><mtext> ≈ </mtext><mfrac><mrow><msqrt><mtext mathvariant="italic">Rs</mtext><mtext> * </mtext><mtext mathvariant="italic">rp</mtext></msqrt></mrow><mrow><mtext>ω</mtext></mrow></mfrac><mtext> ≈ 22</mtext><mtext mathvariant="italic">uH</mtext></mrow></math><img file="EP1515594A2_D0004.tif" /></maths>

Another modeling also makes it possible to determine these characteristics. The resistance of the capacitance is modeled by the dielectric dissipation factor (tan (δ) = 1 / (Rp * C)) in the insulating material of the candle head, which is considered constant and solely dependent on the chosen material.

The overvoltage coefficient is then defined as follows:<maths id="math0005" num=""><img file="EP1515594A2_D0005.tif" /></maths>

The maximization of the overvoltage coefficient Q is then equivalent to the minimization of<maths id="math0006" num=""><math display="block"><mrow><msqrt><mfrac><mrow><mtext mathvariant="italic">The</mtext></mrow><mrow><mtext mathvariant="italic">C</mtext></mrow></mfrac></msqrt><mtext>.</mtext></mrow></math><img file="EP1515594A2_D0006.tif" /></maths> A high capacitance C and a reduced inductance L are then preferably selected.

These determination rules apply regardless of the type of series resonator used and therefore also apply to the coil-candle described later.

A compromise in the choice of values ​​is however necessary for the variant using a power MOS transistor as a switch, as described below. Indeed, the current flowing through the switch MOS then increases with the capacitance C. The value of the capacitance C must therefore be set as a function of the nominal current of the switch MOS.

Several variants of amplifiers 5 will now be described. In general, an amplifier having a power MOSFET transistor will preferably be used as a switch 51 controlling the commutations across the resonator 6. FIGS. 7 and 8 illustrate two embodiments of amplifiers 5 including M4 MOSFETs, such as switches 51. The amplitude and frequency constraints concerning the voltage to be generated between the electrodes can be solved with a power MOSFET having the following characteristics: an insulation greater than 500 V, a drain current capacity greater than 30 A, a time switching speed less than 20 ns (and preferably of the order of 10 ns when using a control loop) and a gate current capacity of up to 10A.

This MOSFET transistor will also preferably have an inductance of less than 7 nH on its connections between its active silicon surface and the printed circuit on which it is implanted. This avoids transients during high voltage peaks that would be detrimental to the fast switching of the transistor.

FIG. 7 represents a first embodiment of an amplifier 5 having such a switching control transistor M4. A midpoint transformer 56 is interposed between the control 4 and the power MOSFET M4. The M4 power MOSFET can thus be controlled very quickly with a symmetrical voltage able to block it effectively. Indeed, the application of a negative voltage on the gate of the MOSFET M4 transistor makes it possible to compensate for the overvoltages caused by the linkage inductance of M4 with the rest of the circuit. The blocking of the transistor is thus facilitated, especially since a negative voltage makes it possible to discharge the gate-drain capacitor particularly rapidly.

The amplifier 5 shown comprises two intermediate transistors M1 and M2 arranged to alternately feed the coils L1 and L2 of the primary of the midpoint transformer. A control circuit 57 applies respective control signals to the transistors M1 and M2. The control signals do not overlap temporally to avoid a short circuit in the primary. The control signals also advantageously have substantially equal activation times to limit the magnetizing current in the transformer 56. It is also possible to compensate an inequality of the activation times by a high value of the magnetising inductance of the transformer 56.

The timing diagram of FIG. 9 illustrates various signals during the excitation of the series resonator 61. The curve 91 represents the current flowing through the series resonator 61. The curve 92 illustrates the gate voltage of the MOSFET M4. Curve 93 illustrates the voltage at the input of the series resonator 61.

Advantageously integrates the amplifier 5 on the same printed circuit 8. It is thus possible to integrate the transformer 56, the transistors M1 to M4 and the control circuit 57 on the same printed circuit, according to the diagram shown in FIG. thus for a reduced cost a very compact amplifier 5. The leakage inductance of the transformer and the overvoltages at the terminals of the intermediate transistors M1 and M2 are also minimized.

The left part of Figure 10 shows several elements of the amplifier 5 and their connections. The central part of FIG. 10 represents the transistors M1 and M2 and their respective windings L11 and L12. The right part of FIG. 10 schematically represents the various elements integrated on the printed circuit board 8. The assembly formed by the transistors M1 to M4, the coils L11, L12 and L2, is preferably arranged on an edge of the printed circuit 8. The windings can thus be arranged in the air gap of a split torus 81.

FIG. 8 represents a second embodiment of an amplifier 5 having a MOSFET switching control transistor M4. The gates of the transistors M1 and M2 are linked. Transistors M1 and M2 thus switch simultaneously. The bipolar transistor M3 is therefore mounted as a follower. When M1 and M2 conduct, the bipolar transistor M3 is off, and therefore the MOSFET transistor M4 is also blocked. M1 and M2 intermediate transistors having the following characteristics are preferably used: a control voltage of 5V, a nominal current of 8A at this voltage, a resistor R_we less than 150 milliOhms and a response time of less than 20ns.

As shown in FIGS. 11 to 13, the amplifier 5 is advantageously servocontrolled to the charge current applied to the resonator 6. In practice, it is sought to slave a switch 51 controlling the commutations across the resonator 6. The amplifier 5 thus has a device 54 for measuring the current applied to the input of the resonator 6. The set point is applied to an input 58 of a comparator. The output signal of the comparator is applied to an amplification device 53 shown schematically. This optimizes the overvoltage factor of the amplifier 5 by driving the resonator 6 at its own frequency despite behavioral drifts of the components. This avoids the use of components whose cost and complexity are inappropriate for mass production. The slaving is for example carried out by re-injecting into the amplifier 5 a voltage proportional to the current flowing in the load. Phase correction can also be applied to the measured signal via the phase shift module 55.

In such a transformer 54, combined with a control loop, the parallel resistor R2 of the secondary of the transformer preferably fulfills two functions of the servocontrol: the feedback of a signal proportional to the current in the load, and the phase shift of the intensity crossing the load according to its resistance value.

It is advantageous to use a transformer 54 having a very low inductance value (for example between 10 and 20 nH) and whose windings support a current of the order of 10A. FIG. 14 thus shows an example of a transformer produced on a printed circuit, making it easier to obtain such characteristics. The left part of FIG. 14 independently represents the useful layers of the printed circuit. The right part of the figure represents these superimposed and assembled layers. The conductive element 151 forms the primary of a transformer, and is disposed on a first face of the substrate 152. This conductive element 151 is in the example made in substantially wire form. The conductive elements 153 and 154 form the secondary of the transformer. These conductive elements 153 and 154 are arranged on a second face of the substrate 152, opposite the conductive element 152. The elements 153 and 154 are electrically connected on the one hand along the dotted line, and on the other hand by the resistor 155. The resistor 155 can thus be used to measure the current flowing through the conductive element 151 and to form the phase shift module 55 described above.

Advantageously adapts the servocontrol of the switch 51 controlling the commutations at the terminals of the resonator 6 previously described to the embodiments having a power MOSFET switching control transistor as a switch. It is thus possible to switch the MOSFET transistor M4 at the optimum times.

In order for the servo structure to produce oscillations rapidly, despite an initial zero load current, several advantageous variants of the system are available.

The LC resonator 6 comprises a series resonator 61 and a parallel resonator 62. The series 61 resonator has a series 119 capacitance and a series 65 inductance. According to a first variant, the servocontrol structure comprises an astable oscillator 52 (for example a slot generator) for generating the first half-waves in the series capacitance 119 and stabilizing the oscillations under steady state conditions. It is expected that the frequency of the oscillator is close to the frequency of the excitation generated between the electrodes. The servo structure adds the current measurement signal and the signal of the astable oscillator 52 and thus enables the class E amplifier to perform the switching at the most favorable moments.

Moreover, the first slot generated by the oscillator 52 is approximately twice as short as the following: thus, the current in the series inductance 65 can be initialized to the value of this current under steady state conditions. The parallel resonator 62 comprises an inductor 621 and a capacitor 622 arranged in parallel. All pulses across inductance 621 and capacitance 622 are then equal. It is thus possible to avoid oversizing the switch 51 and exploit it optimally.

Figure 12 shows a second variant. The control signal applied to the switch 51 generates a voltage slot of short duration, that is to say of the order of 5 μs, initiating the first alternation in the resonator 6. The servo signal then controls the switch 51. The feedback loop of the servo structure has a high gain. Thus, the initial pulse making the servocontrol operational is sufficiently short, and the current flowing through the switch 51 remains reasonable. It is thus not necessary to over-size the switch 51 to start the servocontrol, in particular when the switch is formed of a power MOSFET transistor.

An advantageous combination of the parallel resonator 62 and the series resonator 61 optimizes the operation of the system when the natural frequency of the parallel resonator 62 is slightly greater than that of the series resonator 61. Thus, the voltage pulse generated by the closing of the switching transistor M4 has a duration less than the half-period of the series resonator 61. Thus, the pulse during the closing of the switch transistor M4 is anticipated by the internal inverse diode of the transistor M4 when the voltage of its drain returns to a zero value. It is then expected that the ratio between the respective characteristic impedances of the parallel resonator 62 and the series resonator 61 is less than 100 greater than 40. The lower value guarantees a good overvoltage coefficient. The upper value limits the currents in the transistor M4. A capacity of 1 nF and an inductance of 1 μH are typically used for the parallel resonator 62. The characteristic impedance of the parallel resonator 62 is then approximately 32 ohms.

Moreover, in the parallel resonator 62, it can be considered that the capacitances between the turns of the inductor 621 will be negligible with respect to the capacitance of the capacitor 622. It is therefore possible to realize the inductance 621 in the form of a superposition of substantially circular conductive tracks 623 made on the superimposed layers of a printed circuit. Examples of inductance structures 621 on a printed circuit are shown in FIGS. 15 and 16. The embodiments of these figures thus make it possible to produce an inductor 621 without a ferrite core. This reduces the cost and improves the performance of the inductor 621.

In Figures 15 and 16, the thick points represent connection pads of the different tracks. The vertical lines joining the connection pads represent conductive connections between the pads. The connected tracks thus form a coil. Advantageously, each track 623 is surrounded by a closed loop 625, in order to reduce the radiation of the inductance 621 formed by the tracks.

The diagram of Figure 15 shows a variant having an upper layer and a lower layer having no coil track. The upper layer and the lower layer each have a connection terminal 624 of the inductor 621.

The diagram of FIG. 16 represents a variant, in which the lower layer and the upper layer each have a coil track and a connection terminal. The curved lines 626 joining a connection pad to a connection terminal 624 represent an electrical connection reported on these printed circuit layers.

At the working frequencies of the resonator 6, the losses are significant. In order to limit these losses, the presence of magnetic material in the series resonator 61 is preferably minimized.

It should be noted in the foregoing that the mention of a series resonator does not necessarily imply that the resonator also includes a parallel resonator.

As illustrated in FIG. 17, it is possible to envisage a variant in which a common power supply and amplifier are used for two resonators 6 arranged in parallel. This variant reduces the weight, cost and overall complexity of the voltage generation system 1 for a spark ignition engine. Each resonator 6 corresponds to a respective combustion chamber 141 and 142, the two combustion chambers being in phase opposition. The amplifier 5 is controlled so that the ignition voltage is generated both during compression and during expansion for each combustion chamber. Indeed, the compression in a chamber 141 is synchronized with the trigger 142 in the other. During the generation of the voltage, the breakdown in the expansion chamber 142 is much faster than in the compression chamber 141. Indeed, the gas density in the expansion chamber is much lower than the density in the compression chamber. The equivalent discharge resistance of the expansion chamber 142 is thus much higher than that of the compression chamber. The candle present in the compression chamber then continues to rise in voltage until breakdown. The density of the gas in the expansion chamber is low enough not to disturb the coefficient of overvoltage in the compression chamber; the generation of the spark in the chamber in compression is thus undisturbed by the generation of the voltage in the other chamber.

FIG. 18 represents a sectional view of a spark plug advantageously integrating a series resonator 61. The spark plug 110 has a connection terminal 131, connected to a first end of an inductive winding 112. The second end of the inductive winding 112 is connected. at an inner end of the high voltage electrode 106. This end is also in contact with an insulating element 111 forming the capacitor.

The electrodes 103 and 106 are in this example separated by the dielectric material 100 for guiding sparks between these electrodes. The series resonator 61 integrated in the spark plug 110 comprises the inductive winding 112 and the insulating element 100 also forming the capacitor between the electrodes 103 and 106. The capacitor and the inductive coil 112 are arranged in series. The series capacitance of the series 61 resonator is formed of the capacitor and internal parasitic capacitances of the spark plug. This capacitor 119 is arranged in series with an inductor 65 to form the series resonator 61. As the length of the connection between the inductor and the capacitor is thus reduced, the parasitic capacitances in the spark plug are reduced. It is thus easy to obtain an overvoltage coefficient of the series resonator in the preferred range of 40 to 200 described above. The candle 110 is thus used to maintain the alternating voltage between the electrodes 103 and 106, in the desired frequency range.

The integrated series resonator in the candle preferably has a single coil 112, facilitating the manufacture of such a candle.

A large number of turns in the single winding 112 is necessary to obtain an inductance of the order of 50 μH (order of magnitude detailed later). However, a large number of turns generates parasitic capacitances. The single inductive winding 112 preferably has an axis (identified by the dashed line) and consists of a plurality of turns superimposed along its axis. It is thus understood that the projection of a turn is identical to the projection of all the turns along this axis. The parasitic capacitances are then limited by not superimposing turns radially.

The spark plug furthermore advantageously comprises a shield 132 connected to a mass and surrounding the inductive winding 112. The field lines are thus closed inside the shield 132. The shielding 132 thus reduces the parasitic electromagnetic emissions of the candle 110. The coil 112 can indeed generate intense electromagnetic fields with the radiofrequency excitation that it is intended to apply between the electrodes. These fields may notably disrupt embedded systems of a vehicle or exceed thresholds defined in emission standards. The shield 132 is preferably made of a non-ferrous material of high conductivity, such as copper. It is possible in particular to use a conductive loop as shielding 132.

For a shield 132 and a single coil 112 each having a generally cylindrical shape, the optimum ratio between their diameter is equal to the Euler number, or approximately 2.72, if it is desired to minimize the maximum electric field generated at the surface. turns. This avoids breakdown phenomena at the origin of energy dissipation in the candle. We will then preferably choose a ratio between their diameter of between 2.45 and 3.

The use of two coils 112 wound on one another and connected in parallel makes it possible to reduce the resistance of the winding formed. The skin effect, significantly increasing the resistance of the winding in the radio frequency range, is minimized by the winding one over the other of these two windings. If it is desired to minimize the length of the winding 112 for a predetermined inductance, the optimum ratio between the diameter of the shield 132 and the coil 112 is<maths id="math0007" num=""><math display="inline"><mrow><msqrt><mtext>2</mtext></msqrt></mrow></math><img file="EP1515594A2_D0007.tif" /></maths> by winding on one another two windings 112 connected in parallel by their ends. The two coils wound on one another have slightly different winding diameters and therefore slightly different inductances, which can disturb the operation of the candle in the radio frequency range. It has been determined that for the value<maths id="math0008" num=""><math display="inline"><mrow><msqrt><mtext>2</mtext></msqrt></mrow></math><img file="EP1515594A2_D0008.tif" /></maths> mentioned above, the difference of the inductances did not disturb the operation of the candle in the radiofrequency domain. In this case, a ratio of diameters between 1.35 and 1.5 will preferably be chosen.

The winding 112 and the shield 132 are preferably separated by an insulating sleeve 133 of a suitable dielectric material, in order to further reduce the risk of breakdown or corona discharge, causing energy dissipation. Of course, the lower the energy dissipation, the greater the amplitude of the voltage applied between the electrodes and the longer the life of the candle is high. The dielectric material may for example be one of the silicone resins sold under the references Elastosil M4601, Elastosil RTV-2 or Elastosil RT622 (the latter having a breakdown voltage of 25 kV / mm and a dielectric constant of 2.8). It can be provided that the outer surface of the sleeve 133 is metallized to form the shielding 132 above.

In general, it will be preferred to wind the winding 112 around a solid element 134 made of insulating and non-magnetic material. This further reduces the risk of breakdown and parasitic capacitances.

The set of dielectric materials is preferably strongly debulled, to further reduce the risk of breakdown. All the dielectric materials of the candle preferably have melting temperatures above 150 ° C.

In general, when the coil-candle comprises several insulating elements contiguous, there is a significant risk of creating air inclusions at the interface between these elements, especially when made of ceramic. However, for constructive reasons, it is envisaged that the coil-candle in most cases comprises several insulating elements contiguous. In particular, the connection between the insulation 134 of the coil and the insulator 111 of the candle head is also, for the same reasons of discharge, a very important source of dissipation. The aforementioned technique can, according to a new embodiment, be used in the ceramic to create equipotentials preventing the formation of electrical discharges.

Figure 19 shows a section of an insulating element 111 of candle head, also solving this problem. This insulating element 111 is intended to be associated with an insulating element 133 in the form of a silicone resin. This insulating element 111 has a non-circular section and is included in a circular part 136 belonging to the cathode 103. Thus, this element forms passages intended to let the silicone resin flow during its injection. The silicone resin can thus remove most of the air inclusions from the surface of the insulating elements.

The dielectric material used for the insulator 100 may for example be a ceramic based on alumina, aluminum nitride, aluminum oxide or silicon carbide.

At the working frequencies of the series resonator, the losses are significant. In order to limit these losses, those skilled in the art will limit as much as possible the presence of magnetic material in the series resonator.

According to a particularly advantageous variant illustrated in FIG. 18, the spark plug 110 furthermore has a current measurement winding 139 fulfilling the function of the module 54. This winding 139 comprises several turns surrounding the winding 112. The winding 139 is preferably arranged near the connector 131 and away from the candle head, in an area where the voltages are relatively low.

The spark plug of the invention can incorporate a certain number of other characteristics, such as the seat seal 130 of FIG. 18 disposed against a shoulder of the cathode 103, and intended to seal the breech at the level of the light. of the candle.

The candle head is the part of the candle that is placed in the gas in which the plasma must be formed. This candle head preferably comprises three elements: a central electrode 106, a ground electrode 103 and an insulator 100. The geometry of these elements is decisive for ensuring the formation of the volume plasma or the branched plasma at the desired location of the chamber, with the optimum properties, especially for ignition (large volume, optimum energy transfer to the gas , etc ...).

Figures 20 to 27 illustrate different configurations of candle heads, advantageously included in candles adapted to generate a plasma between their electrodes and adapted to be powered by a radiofrequency excitation.

Figure 20 shows a first group of variants of candle heads, which will be called candles with capacitive propagation. These geometries of candle heads have a cathode 103 partially covered by the insulator 100 in the axis of the candle. This geometry generates a capacitive propagation of the spark on the surface of the insulator 100.

Figure 20.I shows a candle head geometry known per se. In this figure, it can be seen that the cathode 103 protrudes axially beyond the insulator 100. It can also be seen that there is a direct path in the air between the anode 106 and the cathode 103. An electric arc can to form following this direct path.

The geometries of Figures 20.II and 20.III generate a better distribution of the plasma on the surface of the insulator 100. By lengthening the air path connecting the two electrodes, the probability of forming an arc is reduced. This creates multidirectional discharges between the electrodes. The plasma is distributed more evenly around the candle and the volume of gas affected is increased. The capacitive propagation effect between the electrodes is also reduced; the plasma can thus be generated at a greater distance from the surface of the insulator.

In the variant of FIG. 20.II, the cathode 103 is no longer projecting axially with respect to the insulator 100. The insulator 100, the cathode 103 and the anode 106 form substantially a flat surface, avoiding the formation of an electric arc between the anode 106 and the cathode 103.

In the variant of FIG. 20.III, the insulator 100 projects axially with respect to the ends of the electrodes 103 and 106. This further makes it possible to lengthen the air path between the electrodes 103 and 106. The protrusion of the insulator 100 forms a rounded boss.

The variant of Figure 21 proposes to reduce the capacitive effect. Thus, in the candle head, the cathode 103 does not extend radially under the insulator 100.

To lengthen the air path between the cathode and the anode, the cathode 103 of this variant is arranged axially recessed relative to the insulator. The central electrode or anode 106 is arranged flush with the insulation.

FIG. 22 proposes to make a cavity or a recess 120 in the insulator in order to amplify the depolarization phenomenon. The anode 106 also has an increasing section at its end, at the recess 120. Thus, in the recess, the final section of the anode 106 is greater than its intermediate section. Thus, a vacuum 121 is created axially between the end of the anode and the insulator 100, which locally amplifies the electric field.

In general, the variants to avoid the formation of a direct arc between the electrodes operate optimally in combination with the radiofrequency excitation. Radiofrequency excitation makes it possible to extend and curve the trajectory of the sparks.

FIGS. 23 to 25 show examples of tip-effect spark plugs characterized by a pointed anode portion projecting axially with respect to an axial end of the insulator and with respect to the cathode.

Fig. 23 shows a preferred embodiment of a peak effect candle head. The anode 106 consists of a core 1061 and a sheath 1062. The core 1061 is for example made of copper in order to promote the evacuation of heat along the anode 106. electrochemical erosion of the end of the anode. The sheath 1062 can be made of any suitable material, such as nickel.

Figure 24 shows several examples of peak effect candle heads. These candles thus have a ground electrode 103 recessed axially with respect to the insulator 100, in order to reduce the capacitive effect. The projecting end of the anode 106 also has a pointed shape.

Examples 24.II to 24.IV each have a cathode 103 forming an axial recess 122 near the insulator 100. This recess 122 has a rounded shape. This increases the ability of the candle to generate a branched spark. This reduces the probability that a plasma will only propagate on the surface of the insulation. The plasma thus tends to be distributed in a volume remote from the surface of the insulator 100.

Examples 24.III and 24.IV have an insulator 100 whose end has a rounded shape 123, in order to reduce its internal stresses. These constraints are related to the high levels of electric fields and temperature gradients near the end of the insulator 100.

The example of FIG. 24.IV comprises an anode 106 whose axial end 1063 has several points. A greater number of sparks are thus generated during the excitation and the erosion of the anode 106 is distributed over all the tips used.

The spark plugs may accidentally generate arcing between the anode 106 and the piston when the distance between the piston and the spark plug head is small. These arcs prematurely erode the tip of the anode 106 and prevent the formation of plasma volume or branched plasma. The candle head of Figure 25 thus provides a solution to this problem. The tip of the anode 106 is thus disposed in a counterbore 124 formed in the insulator 100. In order to reduce the electric field at the interface between the anode 106 and the insulator 100, a countersink and anode of cylindrical shapes and having diameters whose ratio is equal to the number of Euler are preferably provided. It is thus preferably provided that the ratio of their diameter is between 2.45 and 3. In FIG. 25, it can also be seen that the insulator 100 projects axially with respect to the tip of the anode 106. The insulator 100 also has an edge 125 projecting axially with respect to the cathode 103.

Figures 26 and 27 illustrate candle heads with dielectric barriers that will be designated later by blind candles. At the level of the candle head, the anode 106 is completely covered by the insulator 100. Such candles allow in particular to eliminate the formation of an electric arc between the anode and a piston, and eliminate erosion of the anode. The life of the spark plug is thus greatly increased, and can equal the life of a heat engine without requiring maintenance. Such candles only work because of the capacitive nature of the insulation 100.

The operation of a blind candle is made possible by the use of radiofrequency excitation. The application of radiofrequency excitation between the electrodes of a blind candle is also particularly advantageous. Electrode excitation forms space charges on the outer surface of the insulation. The insulator 100 then behaves as an anode and a volume plasma or a branched plasma is generated on its surface. Although the insulation has a relatively low charge, the radio frequency excitation generates a very large number of sparks on the surface of the insulation in a very short time. It can be provided in this variant that the insulator 100 forms the capacitor of the resonator. This reduces the energy dissipated in the candle.

According to a variant illustrated in FIG. 27, it is also conceivable to use a cylinder head of a heat engine as a ground electrode. The cost and complexity of the candle can be greatly reduced. In the blind candle of Figure 27, the cathode is constituted by the cylinder head.

It is also possible to integrate the candle in the cylinder head of the engine, still using the cylinder head as the cathode of the candle. The skilled person will then take any appropriate measure so that the life of the spark plug is at least equal to the life of the engine.

Moreover, although the heads of the candles shown have a symmetry of revolution about their axis, it is also possible to provide candle heads having other geometries, within the scope of the invention.