FIELD OF THE INVENTION : The present invention relates to driving nonlinear loads such as high intensity discharge lamps, plasma tubes, gas lasers, x-ray units, spark gaps and similar.

The present invention discloses new analytically optimized linear torus inductors with low external component need performing high voltage generation, current limiting, and power limiting.

The present invention discloses new analytically derived circuit constraints that obtain a triple root network operation with predictable response that allows small capacitors external to the torus inductor and an optimal near loss-less operation of the solid state switching devices.

OBJECTS AND ADVANTAGES : An object of the present invention is to obtain hitherto unrealized performance of high frequency power inductors for extreme applications such as HID lamps, plasma power supplies, gas laser drivers and x-ray power supplies.

An object of the present invention is to add a new dimension to the effort that has hitherto been made to reduce both electromagnetic radiation and acoustic noise emanating from high frequency electronic power controllers and drivers.

An advantage of the present invention is that the efficiency and linearity of the driver inductor is increased by using magnetically active cores of low permeability material in a new optimized geometry, compared to the conventional high permeability cubic cores with air-gap or high permeability torus cores in the prior art.

An advantage of the present invention is selecting a torus radius ratio based on new and novel analytical optimisation-such as"k = b/a = 1/7"for circular cross-section, to obtain minimal wire resistance for the completed torus inductor and minimal EM pollution outside the component.

An advantage of the present invention is the inherent use of low permeability materials for the torus core-which can now be casted with prewound solenoidal coils-in effect eliminating the complex winding process in expensive machines.

An advantage of the present invention is that by obtaining new stable triple root circuit-operation, energy storage in capacitors is minimized and near absent compared with prior art electronic drivers with bulky power factor correction capacitors, which have the same or worse shortcomings as the capacitors for passive 50/60Hz dampers/drivers.

A further advantage to the present invention is that it addresses the ecological demanding of low EM pollution and high efficiency and low weight.

BACKGROUND AND PRIOR ART FOR THE INVENTION : A search of the background art directed to the subject matter of the present invention was conducted by the"Patent-og Varemaerkestyrelsen"in Denmark- on the gth of July 2004, and disclosed the following five patents or patent application : US5389857, US5294868, GB2356081, DE3632272 and EP0187493.

Thorough evaluation of the sited documents has and will show that these documents do not address the novel features taught in the present invention- specifically the new analytical optimal construction procedure taught in the present invention which can be applied to a wide range driver/damper/ballast applications for nonlinear high voltage and high current loads.

Passive dampers/drivers for nonlinear plasma loads-such as HID lamps and X- ray lamps in the prior art can be illustrated as in Fig. 3.1 require bulky, complex and expensive components, often with limited lifetime. Passive drivers still monopolize the industry in the higher wattage applications and it is only relatively recently that active or electronic drivers have appeared for the lower power members of the HID lamp family and gas discharge tubes, but they suffer from acoustic-, conducted-and radiated noise together with heat dissipation making poor use of energy and polluting the environment in a broad sense.

Prior art electronic or active drivers are more or less based on the Switching Power Supply topology illustrated in Fig. 3.2 that was developed due to demand in the communication and computer industry many years ago and do not address how to optimize the torus geometry together with critical material selection

methods of the present invention. In the prior art, high frequency ferrite inductor with inherent or intended air-gaps as seen in Fig. 3.3 in effect, replaces the laminated iron-nickel core of the older 50/60Hz passive damper/driver.

High frequency power inductors in prior art electronic drivers use rather high permeability materials for the core to help confine the magnetic field inside the core structure. High permeability materials in power inductors saturate easily, and therefore air-gaps, large or small, are introduced into the magnetic path. The presence of the air-gap promotes energy gradients in its vicinity and considerable acoustic energy is radiated which makes prior art power inductors noisy besides being warm due to Eddy-, hysteresis-and magnetostriction losses.

In FIG. 3.2, we illustrate a schematic view of a typical power controller for prior art electronic drivers for gas discharge tubes, sectioned into five main building blocks, from the mains input to the gas discharge unit or HID lamp. This is a basic switching power supply with AC output although the topology can differ in details, especially the inverter topology from half-bridge to full-bridge, push-pull, fly-back and fore-ward.

BRIEF DESCRIPTION OF THE INVENTION : The present invention discloses new analytical solution to the nonlinear algebraic equation encountered in both geometry, and material selection for a thermally and electromagnetically optimized torus inductor with very high efficiency.

Further, the present invention discloses new analytical constraints that govern optimal component selection and optimal operating conditions in solid-state electronic AC-AC power controllers for nonlinear high voltage and high current loads. More specifically, the present invention teaches a new method to design high performance, high voltage, high current and high frequency torus inductors, wound on a core made from magnetically active material with low or moderate permeability, very low electrical conductivity and very linear susceptibility to electric and magnetic fields. This creates a unique driver/inductor component with very linear properties, very high short circuit current capability and very high open circuit voltage capacity necessary to ignite discharge lamps and tolerate the short circuit currents thereafter necessary to successfully drive HID lamps and plasma devices.

The present invention also discloses method to design a torus inductor with very small induced current losses and very small nonlinear hysteresis losses and very small dielectric losses and very small acoustic losses that can be achieved by a large variety of materials. Candidate materials are low permeability nickel-zinc ferrites, almost ferrite-less ceramics, single domain ferromagnetic particles in dielectric matrix, nano-crystalline low permeability alloys, thin film ferrite polymers, rare earth doped ceramic or solids, to the extreme case of an empty air-core. Selecting precisely the proper permeability value for the core is instrumental in obtaining the simultaneous solution to the nonlinear algebraic equations governing the geometry, current and voltage constraints.

Fig. 4.1 depictures a minimal implementation of the present invention for AC- mains supplied electronic driver comprised of torus inductors, capacitors and a switching device. Fig. 4.2 illustrates, not to scale, minimal energy storage waveform feeding that track the input envelope. Fig. 4.3 illustrates a controller governing the ON-time and OFF-time of switching devices individually through a separate feedback-loop with the OFF time related to the state of the input voltage, but the ON time is related to the state of the output current. By combining the two, efficient start, ignition and warm-up, is secured, together with a versatile power limiting function.

DETAILED DESCRIPTION OF THE ANALYTICALLY OPTIMISED TORUS INDUCTOR : Up until the present invention, inductor geometry design was guided by complex tables, graphs, procedures and expensive software, and it can be fairly said, that no two designers use the same method with trial and error ruling the game. This situation is rooted in the rather large number of adjustable parameters in geometry and material selection of both the core and the wire. To make matters worse, the algebraic equations involved are both non-linear and not uniquely solvable.

The present invention, discloses a relatively small, near linear power inductor, wound on a magnetically active torus core, which can store and release magnetic energy with minimal external effects. The near linear performance allows induction of very high voltages, such as 5000 volts or higher, together with a

high short circuit current capability of tens of amperes, thus allowing up to 50kW virtual power to be delivered from a relatively small volume.

A torus inductor according to the present invention is wound on a core of low or moderate permeability material to promote low or near absent losses. An empty core, also known as the air-core, has the same permeability as vacuum,"ß0 = 1.2566 plH/m". At the other extreme high density ferrites have been constructed with permeability up to or over 100 000 times that, or "µ = 0.12566 H/m".

Driver inductor design is governed by the desired power handling capacity, the short circuit current desired, the open circuit voltage and the desired frequency of operation. Four main classes of driver torus geometries are considered in the present invention and are illustrated in Fig. 7.1. Two material constants, the Magnetic Permeability of the core "µ = µ0(1+#m)", and the Electric Conductivity of the wire"6 = eneTe/nie"are need in this analysis, and the Electric Permittivity of the core material"s = so (l+Xe)" is advantageous for upper frequency limit determination.

The maximum current (i), and the maximum voltage (u), are both fixed by the application. The time of conduction (t) or (toN) is rather flexible and can be decreased for more power handling or increased for less. Four quantities; the Resistance (R) of the wire, the Magnetic Field Intensity (H) inside the core, the Magnetic Flux Density (B) inside the core, and the Inductance (L) of the complete torus characterizes the design. To be explicit with reference to Fig 7. 1, let us use a wire of diameter (d) and wind it on a torus with core radius (b) and magnetic radius (a): Here"N = 27v (a-b)/d"is the number of tight windings on the torus cores = 2bN"is the total length of the wire used,"Sc = 7Eb 2,, is the cross-sectional area of the torus, and"/c = 21can is the magnetic length, which is also the circumference of the core centre. As a hint to obtain our final result, multiply (L) and (H) to get "LH = slik2N3/47rn, and from this equation, solve for (N3) and insert into the equation for the resistance (R):

Both equations for (R) and (a) have a minimum value at = 1/3", and it so happens that the majority of industry standard manufactured torus cores yield closely to this value as seen in Fig. 7.3. This is the obvious choice if either (a) or (R) is fixed by the design requirements and should be considered as prior art.

The present invention goes further and reveals that it is possible to eliminate (N3), (N2) & (Nt) from all three equations for (R), (L) & (H), and the result is a single expression as a function of the single dimensionless variable = b/a" : These now disclosed expressions are the key to the optimal design where both the electrical resistance of the wire, and the electromagnetic fringing outside the core are minimized. In the equation on the right, the current has been scaled down to"io = i/4"and thus eliminating the factor"210 = 1024"from the left equation. This is done to stress the simplicity and regularity of our expression, and to simplify appearance further, we use the conductance of the wire"G = 1/R".

A certain task fixes"k = b/a"by a polynomial of the 7th degree which can be solved graphically or with iterative methods. By equating the 15t derivative of the polynomial to zero, a unique maximum is found at"k = 1/7"while other values give two solutions, the FAT Torus with"k > 1/7"and the THIN Torus with"k < 1/7".

By selecting"k=1/7"for a circular cross-section torus, the smallest possible wire resistance is obtained and the smallest possible magnetic"fringing"outside the core of the inductor component is obtained at the expense of larger torus diameter which is now closer to a thin ring than to a fat torus.

Fig. 7.2 plots the torus polynomial for a circular core and circular diameter wire, as a function of the dimensionless radius ratio"k=b/a".

A necessary condition for optimal design is to be able to select (H) and (, u) as to bring the polynomial expression"k (1-k) 6n down to, or below"66/77 = 46 656/ 823 543 = 0. 05665277..." as seen in Fig. 7.2.

After selecting the value for"k=b/a"from the graph in Fig. 7.2, it is possible to calculate the resulting optimized (N), (d), (a) and (b) values : A closer look at the expression for the major radius (a) reveals the torus volume "V = 2#2 k2 a3" and the energy equation"L i2 = g H2 V"which expresses the balance between the current induced energy"Ej = L i2/2"and the magnetic stored energy "Em = µ H2 V/2".

To show an example of applying the new design equation, assume that we need an inductor with inductance"L = 200 µH" with current ramping to"i = 6 A". Further, let the resistance be required to be less than or equal to"R = 8 mQ"or let the conductance be required to be greater than or equal to"G = 125 S". To keep the magnetic flux density below 0.21 T, we select a core material with permeability "µ = 83 µ0" where "µ0 = 4n 10-7 H/m"is the permeability of vacuum.

This selection also keeps the magnetic field intensity below"Hmax = 2000 A/m".

For the winding we use a good copper wire with electrical conductivity": s = 56 MS/m"and we calculate : The value 0.0549 is below the maximum "k(1-k)6 = 0.05665", therefore a solution exists. To be specific take the radius ratio value to be"k = 0. 16", just above the minimum value 1/7 and calculate the resulting construction parameters (N), (d), (a) & (b):

These values satisfy the design goals for the complete inductor, and keeps"B < µHmax". The outer diameter is 75.3 mm, the inner diameter is 54.5 mm and the height is 10.4 mm. The Voltage-time Integral is"Li =Judt = 1200 Vlisn. With a conduction time of"t = 3.7 lisn we can apply 325 Volts rendering a virtual power of P = 1950 Watts for the 5500 mm3 torus core.

Other torus cross-sections of interest are illustrated in Fig. 7.1, that is square, rectangular and elliptical. The rectangular cross-section and the elliptical cross- section can be used to obtain smaller diameter, as the area is now larger. This can be verified by applying the same method for the circular torus by a skilled person-after the method has been disclosed.

The thermal power generated from losses inside the core, should be larger, or equal, to the thermal power generated inside the coil wire, to keep thermal management simple.

By using optimized torus geometry for a near linear and near lossless inductor core, the magnetic flux is automatically confined inside the component. The confinement increases with the cores permeability and can be increased further by demanding the wire to cover the whole surface of the torus, by varying its thickness and breath along the wire. Such inductors can cope with high short- circuit current and high open circuit voltage such as will occur in HID electronic drivers and plasma power supplies.

MATERIAL SELECTION FOR ELECTRIZATION AND MAGNETIZATION : In the present invention, the electrization (P) and the magnetization (M) represent the response from atoms in materials and structures and can have a variety of characteristics based on polarity, orientation, interaction and history or memory. All materials possess both dielectricnP = Xde Dn and diamagneticnM =

Xdm H poiarization where (xe) is a positive electric dielectric susceptibility tensor and (xdm) is a negative diamagnetic susceptibility tensor.

Permanent electrization"P = nd p due to oriented local dipolar strength (p) of density (nd), and permanent magnetization M = nµ µ" due to spinning and orbiting electrons with magnetic dipole moment (l) of density (nµ) is possessed by many materials and molecular structures. If an effective atomic or molecular mass (mA), in kilograms can be defined, a mass density (A) can be correlated to the electric monopole charge density"Pe = Z (e/mA). CA"or the electric dipole density"nd = Zd (l/mA). #A" or the magnetic dipole density"np = J (1/mA). CA" where (Z) is the integer or fraction of free electrons in each atom and (Zd) is a dipolar distortion in each atom and (J) is the integer or fractional angular quantum number for each atom. Good conductors have"Z > 1"and"Zd = on, while good insulators have"Z # 0" and "Zd > 0".

In the present invention, we calculate the dielectric susceptibility ( de) and diamagnetic susceptibility (xdm) for materials possessing a homogenous and isotropic electron density (ne), electrical conductivity (a), and electron relaxation time (#e) as follows : Here (me) is the electron mass in kilograms and (re) is the classical radius of an electron, (f) is the dimensionless oscillator strength of a bound electron and (r,) is the radius of an orbiting electron number (i). The dielectric susceptibility is positive but the diamagnetic susceptibility is negative, and both are slowly temperature dependant through the electron density (ne). The diamagnetic susceptibility is rooted in the orbital motion of electrons as is apparent from Eq.

6.6 The present invention acknowledges that when permanent electric or magnetic dipolar moments are perfectly randomly oriented, the effective electrization or magnetization is macroscopically equal to zero. When external fields are applied, such random dipoles will orient themselves to an extent limited by the local temperature of the thermally agitated atomic system. We can calculate the paraelectric susceptibility (#pe) and paramagnetic susceptibility (xpm) for materials possessing electric dipole moment (po) with dipole separation (do) as follows :

Here (kB) is Boltzmann's constant, (EB) is the Bohr energy for the orbiting electron and (h-bar) is Dirac's constant. Both the paraelectric and paramagnetic susceptibility is positive and inversely dependant on the temperature through the thermal energy factor (kBT), which at room temperature is about 0.026 eV. At sufficiently low temperatures, long-range effects can enter the picture with accumulative result. Paraelectric materials become ferroelectric and paramagnetic materials become ferromagnetic materials below the so-called Curie temperature.

The present invention gives a new insight into the electric and magnetic susceptibility of matter in state of solid, liquid, gas and plasma. The electron plasma frequency (cep) and the conductive skin-depth (8) occurs repeatedly when discussing electric conductors, inductor core materials, dielectrics in capacitors and the plasma unit or HID lamp itself The electron plasma frequency (Oep) is used to define a corresponding electron plasma radius"rep = C/ (sep = 1/47vrene". We the go further and define a quantum moment of inertia"Ie = (h-bar) 2/mec2"for an electron which is equal to 1. 358 10' "Kg Kg m2 and corresponds to the Compton wavelength. Armed with these definitions we can write clear and transparent equations for the basic electric and magnetic susceptibilities as:

The paraelectric susceptibility (,pPe) is proportional to the energy of a permanent electron dipole (e do) oscillating at the plasma frequency (cep) divided by the thermal energy (kBT). The paramagnetic susceptibility (Xpm) S proportional to the mechanical energy of a permanent spinning electron possessing a 2nd moment (Ie) divided by the thermal energy (kBT) and is therefore bound to break down at low temperatures, and below the Curie temperature, phase transition indeed do occur, and the material can become ferromagnetic. Conversely, ferromagnetic materials phase transit to paramagnetic when heated above the Curie temperature, indicating long-range interactions and correlations between the atoms and electrons in materials that can lead to spontaneous order and accumulated memory and history effects.

LONG RANGE INTERACTION AND MEMORY EFFECTS IN MATERIALS: At low enough temperature, dipole-dipole interaction and spin-spin interactions can give rise to ferroelectric and ferromagnetic susceptibility and both effects are of concern for capacitors and inductors in the present invention. Long-range electrostatic dipole-dipole interactions are needed to establish ferroelectric order and the same can be said for spin-spin interactions establishing ferromagnetic order. This introduces the more exotic ferroelectric and ferromagnetic materials.

Ferroelectric and ferromagnetic materials do not return perfectly to the initial state after removal of an external electric or magnetic field, but leaves behind a remenance displacement field (Dr) or a remanence flux density (Br) respectively.

To completely remove the electrization or magnetization from the material, we must apply the electric coercive field (Ec) or the magnetic coercive field (Hc) respectively.

The deep theoretical base above for electric and magnetic materials in an external electric and magnetic field allow first-principles relativistic calculations of non-linear electric and magnetic susceptibilities. It is possible for materials to be

simultaneously ferroelectric and ferromagnetic and thus both possess electric and magnetic dipole moments. Complex materials can have simultaneously ferroelectric, dielectric, piezoelectric and pyroelectric properties and indirect coupling between ferroelectric and ferromagnetic phases can result in amplified response via a local magnetostrictive effect.

STRUCTURAL AND ACOUSTIC EFFECTS IN MATERIALS: The present invention disclosed a method to minimize radiated and conducted acoustic noise from the components and assemblies of a preferred embodiment of the invention. All materials deform under electric and magnetic fields.

Electrostrictive and magnetostrictive effects can generate considerable noise and must be respected when tailoring the electric and magnetic susceptibilities, especially ferroelectric and ferromagnetic materials in capacitors and inductors as discussed in Sec. 6.5. To extend the electric and magnetic fields introduced in Sec. 6.1 in Eq. 6.1 for structural deformation dependence, we need the stress vector (X) measured in Pascal and the dimensionless strain vector (x): <BR> <BR> <BR> <BR> <BR> # = dX## + # + 1/c#G0## + GB##<BR> <BR> <BR> <BR> # = sE## + dX## (6. SO) <BR> <BR> <BR> H=mx X-M+c Go B+GE E<BR> <BR> <BR> <BR> <BR> X=sB*X+mX'B The elastic compliance tensors (SE) and (SB) relates the mechanical stress (X) to the strain (x). The piezoelectric tensor (dx) is measured in Coulombs per Newton and the analogous poezomagnetic tensor (mx) is measured in Coulomb-metres per Newton-seconds, which is in fact mobility or inverse Tesla. Ferroelectric and ferromagnetic often coexist with structural modulation effects. Ferroelectric and magnetic orders in ferro-electromagnets with secondary indirect interactions can result in a magneto-dielectric effect where the dielectric and magnetic properties can be modified by the onset of a magnetic and a dielectric transition on the application of an external magnetic and electric field.

Ferroelectric and ferromagnetic crystals can possess mechanical resonances and excitation quantified as polariton that can display temporal wave motion.

DETAILED DESCRIPTION OF LOW LOSS MAGNETIC MATERIAL SELECTION : When a torus driver inductor core is constructed with low permeability materials possessing small residual magnetization, we can use approximate linear relationship"M = Xm H"between the Magnetic Induction (M) and the Magnetic Field (H) with the susceptibility tensor (xm) as a proportionality factor. The maximum magnetic field intensity (Hmax) and the corresponding magnetic permeability" i (t+% m)"determines the maximum magnetic flux density (Bmax) the core will experience.

In recent years, there has been a renewed interest in high frequency dynamic properties of finite size magnetic structures, and in a series of new publications, high-frequency magnetization dynamics in magnetic particles have been the subject of much experimental and theoretical attention.

The high frequency characteristics of single-domain ferromagnetic particles can be analysed in terms of relaxation time, anisotropy energy barrier and thermal fluctuations. Deriving an equation for the probability density of orientations for the random electric and magnetic dipoles is the starting point for a calculation of the relaxation rate. This equation can be solved in the limiting cases of high and low energy barrier and shows that the time dependence of the magnetization is dominated by a single exponential, with well defined relaxation rate.

It is now possible to analyze whole ferrite-particle systems, similarly to semiconductor quantum wells, and compute the dielectric constant and the magnetic permeability from statistical properties and local fluctuations. These bulk quantities describe the physical properties over large length and time scales and will normally have well defined values.

The bulk properties of random media can be favourable to obtain linear, isotropic and homogenous magnetic materials to use in capacitor dielectrics and inductor cores with low losses. The most general approximations for the evolution of a dynamic magnetic system in the presence of magnetic field fluctuations and alternative electromagnetic fields are based on quasi-static processes that are characterized by specimen size, which can be much smaller than the free-space electromagnetic wavelength. The energy states of a spin system coupled by exchange interactions are wavelike, and spin waves appear as intrinsic

excitations in magnetic materials. The energy of a spin wave is quantized, and the unit of energy of a spin wave is called magnon. On the corpuscular language, an oscillating process in a magnetically ordered body is a collection of magnons.

Magneto-static oscillations in ferrite samples can have the wavelength much smaller than the electromagnetic wavelength and, at the same time, much larger than the exchange-interaction spin wavelength. This intermediate position between a pure electromagnetic and a spin-wave process, reveals very special geometrical effects. The unique properties of artificial atomic structures can be employed to make low permeability magnetic material for torus cores with very low losses.

The dynamic losses due to irreversible processes in magnetic core materials can be classified into four different but interactive phenomena: a) Eddy current losses from induced currents b) Hysteresis losses from magnetic memory effects c) Domain-wall-movement losses d) Magnetostrictive and acoustic losses e) Electrostrictive and dielectric losses To attack Eddy losses, the core can be sectioned into smaller and smaller domains. Two-dimensional domains are used in 50/60Hz laminated ferromagnetic cores and new ferrite cores such as FINEMET are constructed from thin films or ribbons. One-dimensional domains can be promoted by suspending electrically isolated magnetic particles of extremely small size, such 5-50 nanometres in a nonmagnetic and dielectric matrix. Modern nano-crystalline alloys and thin film ferrite polymers make a good candidate.

To attack losses from domain-wall movement, single-domain magnetic particles such as nano-crystals can be employed and will reduce Eddy-losses at the same time.

To attack hysteresis-losses, small randomly oriented magnetic particles or quantum well magnetic atoms can help to give very thin B-H loops with small coercive field intensity (Hc) and small remanence flux density (Br).

To attack Magnetostriction, small domain particles in elastic matrix can be used and are promoted in the present invention.

The following magnetic materials are all candidates for synthesising low-loss magnetic torus cores: a) Low permeability and low conductivity ferrites such Philips 4E1 and 4D1. b) Low permeability and low conductivity ferrites such Siemens T38 and U17. c) Diluted near ferrite-less ceramic as in a) & b). d) Nano-crystalline composites such as FINEMET e) Nano-crystalline magnetic particles in a dielectric matrix. f) Low permeability paramagnetic core of very low electrical conductivity. g) Super-paramagnetic Chromium FINEMET-type alloys. h) HITPERM nano-crystalline alloys.

Using the core materials above as a starting point, we can synthesize the desired magnetic permeability demanded by the optimized torus constraints. For lowest losses we aim at the relatively low permeability cores from about 10-80, and with electric conductivity from 0.1-1 mS/m.

Thin films have been made that show soft magnetic properties and are isotropic with no preferred easy-axis orientation. The effect of nitrogen reactive sputtering on the magnetic properties show increased linearity.

In Fig. 7.4 it is seen that by perfectly randomizing the orientation of the magnetic particles, linear performance is enhanced and the permeability is reduced. It is indicated in Fig. 7.5, that very low hysteresis losses can be achieved by keeping the magnetic particles below the critical"domain formation size". Both strategies also guarantees that magnetostriction is almost absent.

DETAILED DESCRIPTION OF TRIPLE ROOT CIRCUIT OPERATION : In the present invention a new circuit theory allows significantly higher power to be controlled with smaller components utilizing dynamic effects to increase efficiency and lower radiation losses.

A preferred embodiment uses two-torus topology to implement a mains connected electronic driver that doubly isolates the short circuit current in HID lamps or plasma device from the AC or DC mains supply. In the following subsections, we define the optimal operating conditions and auxiliary capacitor component selection required for the optimized torus driver operation of the present invention.

A network that allows generation of high voltages to ignite HID lamps, and performs the current limiting function after ignition, is drawn up in Fig 8.1. The "G"subscript refers to the High Frequency Voltage Generator, the"B"subscript refers to the Driver Function and the"L"subscript refers to the Load. The HID lamp will have two extreme states: An open circuit with"RL = =00"when the lamp is off before ignition, and an almost short circuit with RL = 0 just after ignition.

The two states described have different"resonant"frequency. For the open circuit case, the effective capacity is Ceff = CB CL/ (CI3 + CL) but for the short circuit case only Ce is in effect. The voltage transfer function (UL/UG) can be simplified in two simple steps as shown in the following equations: To accomplish this reduction we have defined the following 8 quantities: Driver frequency (BB = 2N fB = 1 (LBCB) Normalized frequency s = #B z Load frequency = 2n fL = 1/ (RLCv)" Capacitor ratio p = (CB/CL) Characteristic impedance ZB = H (LB/CB) Load factor p = (RL/ZB) Source factor s = (RG/ZB) Load-driver-ratio # = #L/#B = ß/#

The cubic denominator can be transformed into its normal form "w3 - 3 b1 w + 2 bo"where"w = z + (+e)/3"and (bi) and (bo) are the new coefficients. To simplify the procedure let us assume that (RG) is so small that (E) can be ignored in relation to both ( ?,) and (p). Then the cubic polynomial in the denominator can be shown to have a triple zero.

Fig. 8. 3 & 8.4 plot the 3rd order polynomial coefficients (bo) and (b1) respectively, as a function of (p) with (p) as a parameter for different capacitor ratio. A triple root (zo) appears when both (bo) and (b1) are zero. Then"0 = 8"and therefore lip = 8/#27" and "# = A127". The component values that give the triple root are"CB = 8 CL" and "LB = 27 CL (RL)2 / 8".

Both the individual sign and the relative magnitude of the reduced coefficients (bo) and (bl) govern the frequency and time domain solutions of the HID matching network presented. Written in terms of the parameter"x = (3p/p) 2", the discriminating function, which boarders are traced out in FIG. 8.5, can be written as: The discriminating function defines three zones in the (b0-b1) plane with different network behaviour. A triple real root at the origin gives a critically damped 3rd order system, double and single real root, or distinct real roots give a flexible damper/driver and complex conjugate roots give oscillations for high Q values.

The object of the present invention is to operate at low Q values and well into the real root zones and use small inductors and capacitors under stiff analogue or digital control.

The efficiency of the electronic driver is increased compared to the conventional high Q energy storage method in the prior art, because the triple root operation allows smaller inductors for the driver function.

According to a preferred embodiment of the invention a low frequency AC electrical power is fed to the first torus inductor controlled by a switch preferably operating in the vicinity of the triple-root of the network, then drained and applied to a 2nd torus DC-blocking driver, also preferably operating near the triple-root before being applied to the HID lamp or plasma tube constituting the load.

A network that can isolate short circuit condition present in loads such as HID lamps from the mains input voltage is drawn schematically in Fig 8.2. The"G" subscript refers to the Low Frequency Mains Input Voltage and the"L"subscript refers to the Effective Load consisting of the driver torus and the HID lamp or plasma load. Let (Z) be the impedance looking into the generator capacitor (CG) so that (RG) does not enter into the expression: Here (M) is the mutual inductance when the two coils are wound on the same core. This expression can be made more intelligible by introducing two angular velocity parameters :"03M = 27C fM = 1/# (LGCL)" and "#R = 2W fR = 1/ (RLCu)", a capacitor ratio = (CL/CG)"and a load factor"p = (RL/ZM)" where "ZM = 71 (LG/CL)"is the characteristic impedance of the LGCL combination.

Further simplifications can be made by the change of variable : "s = com z"and introducing the ratio = #R/#M = ß/#":

The cubic polynomial in the denominator can be transformed into its normal form "w3-3 b1 w + 2 bo"where"w = z + (#+#)/3" and (b1) and (bo) are new coefficients. To simplify the task, let us assume that (RG) is so small that (s) can be ignored in relation to both (X) and (p). Then the new coefficients (bo) and (bl) are now functions of (#) and (p), or (p) and (ß) : A triple root (zo) is obtained when both (bo) and (b1) are zero. To find this triple root, we solve the equation'o = i"which has the solutions = 1/8"and therefore"p = 4/#27" and "# = #27". The component values that give the triple root are"CG = 8 CL"and"LG = 27 CL (RL) 2/ 8".

Both the individual sign and the relative magnitude of the reduced coefficients (bo) and (b1) govern the frequency and time domain solutions of the network presented. Written in terms of the parameter x = (X/3) = (ß/3p), the discriminating function can be written as: This equation is similar to the driver inductor equation, but differ in numerical coefficients. The quietness and efficiency of the electronic driver is increased compared to the conventional high Q band-stop filters in the prior art, because the triple root operation allows smaller inductors for the filtering function.

DETAILED DESCRIPTION OF THE TRIPLE ROOT CONTROLLER : The object is to generate bipolar waveform as see Fig. 4.2, and feed it to the output circuit functioning as a final current limiter and as a loss-less damper/driver for nonlinear and variable loads, such as HID lamps, plasma devices, LASER power supplies and x-ray machines.

It has been found that the efficiency of the switching element is strongly improved by the dynamic operation of the switch and that the efficiency may be optimised by selection of ON-times and OFF-times adapted for the state of a) the current in the inductors, b) state of the input voltage, whether it is a DC mains or a slowly varying AC mains such as 50Hz or 60Hz, and c) the state of the plasma in the HID lamp.

Many variants of the full-bridge, forward, half-bridge, push-pull, fly-back, SEPIC and Cuk-converter topologies were built around a driver torus and mains isolation torus according to the present invention. Fig. 10.2 illustrates the basic aspects of some of these circuits.

For AC/AC operation, the physical switching device performing the bi-directional switching may preferably be constructed from semiconductor devices, such as bipolar transistors, field effect transistors or isolated gate bipolar transistors, avalanche rectifier etc.

A control circuit is adapted for dynamically regulating the switching frequency, so that it always matches a situation wherein the efficiency of the lamp driver is optimal simultaneously matching the state of the plasma with the dynamic state of the resonant inductor-capacitor-switch.

A control circuit adapted for dynamically regulating the ON-time and OFF-time of the switching devices, so that it always matches a situation wherein the efficiency of the power driver is optimal, may control the switch, either by analogue or digital control.

With reference to Fig. 4.3 and Fig. 10.4, the ON-times and OFF-times are individually controllable through a separate feedback-loop with the OFF time related to the input voltage but the ON time related to the output current.

Preferably the ON-times and OFF-times reference values are adjustable, e. g. stepwise adjustable in predefined steps for specific conditions, e. g. according to pre-specified loads that can be variable.

Fig. 10. 1 disclosed a possible embodiment of the present invention for interfacing to 115V/60Hz mains. The topology used is a Push-Pull driver with a second torus performing as a transformer to drive the first torus acting as the driver.

A light emitting diode that illuminates a light dependant resistor that modulates the OFF-time in a one-shot RC timer samples the input voltage. The same is done for the current sampler regulating the ON-time By a dynamical switching with independent ON and OFF timers, controlled by separate feedback-loop, the performance of the lamp driver is greatly simplified without sacrificing performance as compared to the conventional power-control methods. The switching period or the frequency of operation is only a statistical quantity in this invention, and conventional feedback control methods such as phase locking loops can't be used.

Preferably the heat from the switching devices shall be the dominant external dissipation in the power controller.

COUNTERACTING ACOUSTIC INSTABILITY IN PLASMA DRIVERS : By the discovery of new subtle effects in plasma devices, the present invention adds a new dimension to the effort that has hitherto been made to reduce electromagnetic radiation and noise from high power electronic driver controllers.

The present invention disclose an electronic driver which converts electric energy into a current limited high voltage potential intended to ionise and excite vaporised molecules of electro-optically active atoms. The minimal features of plasma devices or HID lamps are illustrated in Fig 9.1.

A gas discharge or plasma tube will have many modes of reonances which equivalent circuit is depictured in Fig 9. 2. There is an electron gyrofrequency (eB/mec), ion gyrofrequency (ZeB/mic), electron plasma frequency (ne e2/s0 me) ln, ion plasma frequency (nj e2/80 m) t/2, electron trapping rate (eKE/me) t/2, ion trapping rate (ZeKE/mj) t/2, electron collision rate, ion collision rate, plasma container cavity resonance, ionic charge transport resonance, acoustic resonance and finally thermal resonance.

An important aspect of the present invention is based on a specific method of utilising the dissimilar relaxation times for light electrons and heavy ionic charge carriers in a plasma chamber for light generation. By a dynamic operation, the internally generated Joule heat is reduced and the photon generation is

increased. It is thus possible to increase the efficiency through the dynamic operation.

When constructing the electronic driver, attention is paid to the special relationship between the electronic switching frequency and the plasma characteristics as well as acoustic characteristics and geometric design parameters of the plasma container or bulb. This takes into account the fact that many modes and harmonics play a role in the electro-photo-acoustic-thermal interaction in the plasma material, as indicated in Fig. 9.2.

The use of a high frequency Ionic Transit Time sensing enables a high efficiency operation of a gas discharge lamp of varying geometries for the optimal generation of photons. This applies to a variety of materials and devices including Metal Halide Lamps, High Pressure Sodium lamps, Low-pressure Sodium lamps and other plasma devices.

The radius (R) at time (t) of a strong spherical blast wave resulting from the explosive release of energy (E) in a plasma medium with uniform density (p) is "R = (Et2/p) 1/5". To supress acoustic resonance, the"Ionic Transit Time"must be recognised. If the lamp is short circuited before the"Ionic Transit Time"has evolved, a loss-less damping can be effected. Thus power can be increased significantly before instability turns on.

Depending on molecular composition of the plasma in the lamp that has been selected, and depending upon the geometry of the plasma container or bulb, the selection of ON and OFF times which determine the average switching frequency, may be guided by the range of acoustic resonance harmonics of the plasma container, the ion and electron plasma resonance frequencies and the cyclotron frequency as a result of the induced magnetic field due to the conducting ionic current itself.

It has been found that by exploiting the different relaxation times of the electrons on one hand and the ions on the other hand, the efficiency of the lamp can be enhanced. The electrons and the ions in the plasma move with different velocity and different species can accumulate at alternative anode and cathode terminal when high frequency electric fields switch polarity.

The chemical composition of the plasma material could be selected from a group comprising halides, inert gas, metallic vapour, and oxide thereof. As an example the plasma material could be made from Xenon gas or from a mixture comprising Sodium vapour.

Preferably the plasma container, see Fig. 9. 1, has an ionic path length in the size of 1-5 cm. such as in the size of 3 cm. The plasma container may comprise a large number of resonance modes depending on plasma container geometry and material selection, such as container thickness and density of gas.

The load resistance of the HID or plasma device can be substantially equal to zero so that short-circuit is created when the lamp ignites and the next seconds thereafter.

As an example at one of the acoustic resonance frequencies at around 230kHz in a 250W metal halide lamp using potential differences of about 100V between the plasma tube terminals with 3 cm path lengths, an appreciable efficiency increase has been measured.

If as in the dynamic case electrons decelerate against an electric field, their energy will be transferred to the plasma load by polarisation and magnetisation currents but not by the conventional charge carrier current of ions and electrons alone.

DESCRIPTION OF DRAWNGS : The various details discussed in the present invention are described in the following numbered drawings: FIG. 3.1 displays a schematic view of passive driver in the prior art for driving gas discharge lamps, here shown without an optional igniter.

FIG. 3.2 displays a schematic view of electronic driver in the prior art for driving gas discharge lamps, here shown with a half-bridge topology but without the igniter.

FIG. 3.3 illustrates some cubical ferrite cores with inherent air-gap used in prior art electronic drivers and power controllers.

FIG. 4. 1 shows a preferred embodiment of the Triple root damper/driver with two torus inductors L & LB used in the present invention to control ignition, warm-up and power limiting function for a HID lamp.

FIG. 4. 2 shows a Bipolar Chopped Sine waveform to drive a HID lamp with minimal energy storage in capacitors where fh, pp/fin=19 is used for clarity only.

FIG. 4.3 shows a preferred embodiment of the Triple root damper/driver with two torus with dual feedback ON OFF time controllers in a dynamic configuration FIG. 7.1 draws the four principal torus cross-sections evaluated in this invention.

FIG. 7.2 plots the Optimal Circular Torus 7th Order Polynomial as a function of the radius ratio"k = b/a".

FIG. 7.3 plots the industry standard Torus radius ratio as a function of size.

FIG. 7.4 displays the effect of the relative magnetic orientation of small magnetic particles on the B-H hysteresis plots.

FIG. 7.5 displays the effect of magnetic particle size on the thickness of hysteresis loops, e. g small particle size give thinner hysteresis loops, and also the effect of temperature on hysteresis-loop area.

FIG. 8. 1 shows a network that allows generation of high voltages to ignite HID lamps, and performs the current limiting function after ignition.

FIG. 8.2 shows a network that can isolate short circuit condition present in loads such as HID lamps from the mains input voltage.

FIG. 8. 3 shows capacitor selection curves based on the 3rd order polynomial coefficient bo, to obtain triple-root operation to obtain lossless damping.

FIG. 8.4 shows capacitor selection curves based on the 3rd order polynomial coefficient bl, to obtain triple-root operation to obtain lossless damping.

FIG. 8.5 traces the discriminating borders for a 3rd order polynomial showing polynomial coefficient bl, as a function of polynomial coefficient bo.

FIG. 9.1 shows a HID lamp or a transparent plasma container with ionisable vapour and electric terminals.

FIG. 9.2 shows the various energy processes involved in the plasma container represented by equivalent electromagnetic-, electric charge-, thermal-and acoustic circuitry.

FIG. 9.3 shows the Voltage-Amperes characteristics for a 250W plasma device: A) Ignition.

B) Current surging.

C) Warmup.

D) Increasing lamp resistance.

E) Reignition for old lamp devices.

FIG. 9.4 shows the Volt-Current characteristic curves during warm-up of a 250Watt High Pressure Sodium HID lamp corresponding to the curve B-C in Fig.

9.3.

FIG. 9.5 shows the Power-Current characteristic curve during warm-up of a 250Watt High Pressure Sodium HID lamp corresponding to the curve B-C in Fig.

9.3.

FIG. 9.6 shows the Resistance-Current characteristic curve during warm-up of a 250Watt High Pressure Sodium HID lamp corresponding to the curve B-C in Fig.

9.3.

FIG. 10. 1 shows a Push-Pull electronic driver with two torus built to verify and test the present invention with dual feedback ON OFF time controllers in a dynamic configuration for ignition, warm-up and power limiting.

FIG. 10. 2 shows circuit topologies tested with the methods of the present invention : b) Fly-back electronic power controller topology. c) Buck-boost electronic power controller topology d) CLASS E and Boost electronic power controller topology

e) SEPIC electronic power controller topology f) Buck electronic power controller topology g) CUK electronic power controller topology FIG. 10.3 shows circuit topologies implemented with semiconductors for testing the methods of the present invention: a) Fly-back electronic power controller topology. b) Push-Pull power controller topology. c) Half-bridge electronic power controller topology. d) Full-bridge electronic power controller topology.

FIG. 10.4 shows a preferred embodiment of the Triple root damper/driver with bipolar switches and split output inductor and inherent or external capacitors.

FIG. 10.5 shows an ON-time OFF-time controller with dual RC time feedback mechanism for dynamic control of ignition, warm-up and power limiting.