NO853872L - HOEY VOLTAGE CONDENSOR WITH HIGH ENERGY DENSITY. - Google Patents

Description

The invention relates to improvements in that object

which is described in the German patent application published as Publication No. 3 312 076.

This application describes a capacitor type with an energy density above 0.5 J/cm 3 and which can also reach 1.2 J/cm 3,

while the prior art has shown that an energy density of 1 J/cm 3 has never been achieved.

The range between 0.5 and 0.8 J/cm has already been achieved, but only with capacitors using a special dielectric (polyvinylidene fluoride, known under the trade name "K-film"). However, these "K-film" capacitors are

time barely able to achieve this level of energy density with the disadvantage of such high dielectric losses that they are completely excluded from all applications where the capacitor must work at a high repetition rate. The area of application for these "K-film" capacitors is then mainly limited to use with very low repetition rates, such as defibrillators. On the other hand, the price for this special dielectric

per Joule 3 or 4 times the price of a normal dielectric. The type of capacitor described in the above-mentioned German patent document, which does not have the "K-film" disadvantages,

is therefore the only one that can be considered for an energy density above 0.5 J/cm 3 in most cases other than defibrillator use.

This fact has brought the applicant to improve

this technology. The above-mentioned capacitor was precisely limited to an application area similar to or very close to the defibrillator application. The applicant now has appropriate

saw the application of this capacitor to areas other than the defibrillator area, where an energy density of from 0.5 to 1.2 J/cm 3 had not been achieved so far. Another factor to consider was that the energy densities of from 0.5 to 1.2

J/cm 3 which was achieved earlier, was only realized by

some prototypes, and the same values could not later be

accessed again. The present invention therefore also consists in establishing the principles that form the basis of these energy densities, in order to be able to reproduce them reliably and also to exceed them, as

thus achieving values higher than 2 J/cm 3 . The new areas of the invention are the areas of an elementary capacitor with high voltage and high energy, and which has various applications, such as energy storage, commutation and filtering applications. The capacitor is of the general type comprising two electrically conductive "armatures" or capacitor coatings which are separated by at least one dielectric layer. The new elementary capacitor according to the invention is characterized by the fact that each capacitor coating is connected to at least one layer of a first dielectric, that each coating is formed by a metal coating with a resistance per unit area between 2 and 30 ohms, being applied to a second dielectric

with a fibrous structure or construction that favors regeneration, so that the capacitor becomes regenerative (self-healing), that the capacitor is impregnated with a liquid dielectric, and that the thickness and nature of each dielectric as well as the liquid dielectric is chosen so as to ensure that when the rated voltage applied, the relationship between the electric field strength and the breakdown strength is essentially the same for all dielectrics.

Performances that can be achieved with these means can best

summarized by means of two provisions:

- the average electric field strength over the entire dielectric in the capacitor when it operates at the nominal voltage value: This field strength is from approx. 200 V/pm to more than 400 V/ym

- the volumetric energy density of the capacitor at

3

this same voltage: This is from 0.5 J/cm to more than

3

2 J/cm .

These two special characteristics are the most prominent for

a capacitor of the general type discussed above, and they are connected to each other.

The high electric field strength and the high energy density are achieved thanks to two main factors that are linked to each other to produce the given results, which results are as follows:

1. Regeneration capacity (self-healing).

2. Almost complete utilization of the breakdown strength for both dielectrics thanks to an equal ratio between the prevailing field strength and the breakdown strength in

both dielectrics.

Definitions

In this description, the term "rated voltage" means the maximum working voltage that the capacitor can handle. It corresponds to a working voltage for normal use, i.e. with a sufficient safety margin, but within applications where any limitations, apart from the voltage itself, are low or medium. These limitations are, for example, the repetition rate if the capacitor works on discharge, the temperature or other parameters. If any of these parameters are high, the capacitor will work in a non-rated mode (English: derating mode), with a voltage that is less than the rated voltage.

1. Regeneration ability

This regeneration ability consists in the ability to "cure" an insulation fault when it occurs, by burning out the metallization around the fault zone. It is also possible to use the full breakdown strength of the dielectric placed between the two capacitor coatings, up to a very high degree, due to the elimination of occurring errors.

This property consequently makes it possible to approach the isolation limit for the entire dielectric, whatever the isolation limit level may be. That is the role of the second dielectric and metallization properties

assume this regeneration ability. The regeneration process has in fact been known for many decades, but has never been used under the conditions of the present invention, and thus could never lead to the same results. In particular, this process has never been used at the same time up to such levels of voltage, energy and electric field strength as those permitted by the invention, due to the fact that at some of these different levels the process could not be controlled and led to capacitor failure.

This problem is well described in the French patent document No. 7908375 in the following terms: "It is known that during the checks at the end of the manufacture or during the use of such capacitors, the dielectric faults led to local discharges in the form of light arcs, oxidation or local volatilization of the metallized coating. These discharges are useful as they restore the normal insulation. This process is called self-healing.

Nevertheless, with capacitors of a certain capacity and voltages higher than a few thousand volts, the available energy can be at least locally destroyed. As the energy conserved in a capacitor is proportional to the square of the voltage, the above danger becomes greater, a fact which increases the danger of destruction of high-voltage capacitors."

The solution to this problem proposed according to the French patent No. 7908375 is to divide the metallized coating into several isolated parts, which is achieved with a laser beam, in order to reduce the energy that occurs during a self-cure.

The values of voltage, electric field strength and energy obtained in this way are 1500 V, 187 V/pm (1500 V: 8ym) and 112 J.

The solution proposed in the aforementioned patent, however, exhibits the following disadvantages: Complexity to remove metallization with a laser beam, and consequently loss of capacitance, and also loss of capacitance during the operating time when one or more metallized layers are separated.

The levels of voltage, electric field strength and energy achieved in this patent, on the other hand, are much lower than the values of the present invention, whose values are respectively 600 V to at least 8000 V, 200 V/ym and more than 400 V/ ym up to 500 J or more for an elementary capacitor. The given solution according to the present patent leads to much more interesting results even if simpler means are used which is a metallization with a surface resistance of between 2 and 30 ohms deposited on the second dielectric consisting of impregnated fiber structure.

This specificity, which is bound to other elements of the present invention, is, in other words, the basis for a perfect, skilled mastery with respect to the self-healing phenomenon, and this also for the extreme values of voltage, field strength or energy indicated above.

This championship can obviously be decided with the following statements: When a capacitor according to the invention is forced to regenerate due to a certain operating time, or due to the rated voltage being exceeded, the energy consumed or lost at the site of self-healing is always a very small part of the capacitor's total energy, e.g. 1%. For a charging process up to 5000 V, a capacitor according to the invention will therefore see a voltage drop of approx. 4980 V as a result of such self-healing. This small voltage drop and energy drop for such self-healing is due to the fact that the current flowing to the breakdown point is almost stopped by the relatively high surface resistance of the metallized layer. As the thickness of the metallized layer is very small, the mass of the metal to be burned will be small. A special feature of the capacitor according to the invention is that the voltage does not need to drop to stop the self-healing current. The current stops itself because of the surface resistance and because of the very low thickness of the layer.

A perfect demonstration of the regeneration property of a capacitor according to the invention is provided by the following experience: In a winding capacitor according to the invention, a needle is inserted to a depth of 1 mm normally on the axis of the winding, and is then removed. The capacitor is then charged. At the first charge, a certain crackling will occur (if the winding is not protected by oil), and afterwards the insulation will be completely restored. 2. Utilization of the two dielectrics with the same ratio This second feature of the invention is linked to the regeneration properties, because while the latter makes it possible to achieve the insulation limit, this second feature makes it possible to increase this limit up to values that have never been achieved previously.

The choice of the different parameters for both dielectrics, so that the ratio between the prevailing field strength in each dielectric and the resistance to breakdown discharge is essentially the same for all dielectrics, leads to operation at the same level with both dielectrics, e.g. 80%. With this method, the entire dielectric will be better used, and an average electric field strength is achieved at a level that has never been achieved before.

The obtained values of average electric field strength are from 200 V/m to more than 400 V/m. It is easy to show that the volumetric energy density d for a wound capacitor is:

whereEq = absolute dielectric constant of vacuum

e = relative dielectric constant

r

E = electric field strength in the dielectric

Since d is proportional to E 2 , it is clear that the present invention makes it possible to obtain an electric field strength that is more than twice the field strength corresponding to the current, known technique (300 to more than 400 V/um compared with 150 V/ym), and consequently more than four times higher density of energy stored in the capacitors. A preferred embodiment of the invention, in order to achieve that the two dielectric films work equally, consists in choosing the nature of these two dielectrics and also the liquid dielectric that impregnates them, so that the ratio between the resistance and the maximum electric field strength for each impregnated dielectric is practical spoke the same. The constant value of this ratio must also be confirmed at the high levels of electric field strength encompassed by the invention. It is obvious that the impregnation conditions (temperature, vacuum conditions, treatment duration) make it possible to influence these parameters. Not only is affecting the most favorable parts of electric fields in a composite dielectric not already known, but is inconsistent with commonly accepted electrical rules. When it comes to a composite dielectric, the electric field strength level in each of the two dielectrics depends on several dielectric parameters. Considering fig. la on the drawings, the two dielectrics 1 and 2 have the same surface, thicknesses e^

and q^ in relative dielectric constants e, and z^, and specific resistances (resistivities) r^and t^. If the capacitor is charged to a voltage value U, electric field strengths E^ and E^ appear in each of the two dielectrics, and the following voltages consequently appear in the two dielectrics in accordance with the relations:

For composite dielectrics, the known technique gives the relation:

This relation is only correct for a theoretical capacitor that has no leakage resistance. Such leakage resistances and R2 are shown in fig. lb in parallel with capacitors and C2 symbolizing the two dielectrics 1 and 2. In most practical applications, relation (1) is correct because R^ and R2 can be considered infinite. If the capacitors C-^and C2 are considered as if they. is connected in series and has no leakage resistors, it can be said that:

The hypothesis that and R^ are considered infinite is justified in virtually all applications of the present prior art, due to the fact that all known capacitors operate under a "low" electric field strength (below 200 V/ym). Under these conditions, the dielectric resistivity can be conventionally converted to ohm x m, but also to M ohm x ym. This value given in the literature for a field strength of only a few V/ym corresponds to a time constant of 50,000 seconds

(14 hours). As the charging time for a capacitor is always clearly less than this time period, the insulation resistance will not be sufficiently long to change the distribution of voltages as expressed in relations (1) and (2).

In accordance with the conditions according to the present invention, which more particularly consist in making the electric field values as close as possible to the maximum value allowed by the two dielectrics, i.e. their breakdown strength, the specific resistances or resistivities of the dielectrics are now only a part of those stated in the literature, and the insulation resistances R^and R 2 and also the time constants R^and R^C 2 thus become sufficiently low to give a voltage division according to the relation:

In the case that U, and U2 have reached their maximum values, we obtain: If the maximum value of the dielectric field strength for both dielectrics Ex, max and Ei0 insx (<gj>unfolding strength), their respective resistivities r^ and r2 and their thicknesses are introduced in equation (5), we get:

This relationship expresses a preferred choice of dielectric and also of the impregnation conditions according to the present invention.

Nevertheless, it is worth noting that relation (7) is entirely theoretical. In practice, it is actually advisable to use relations (3) or (4) which use the total insulation resistance for the given layer or layer thickness instead of the layer's resistivity, so that to establish relations (6) and (7) one has presented the hypotheses = r^e^ and R2= r2e2, although experience has shown that these hypotheses are not verified for films with thin dielectrics that are exposed to high field strengths. It has been observed that for such films the resistance is not a linear function of the thickness, which means that the resistivity is not constant. For example, the resistivity of polyester films with a thickness of 3 - 5 ym is much less compared to thicknesses of 8 - 12 ym (always exposed to high field strength). The choice of the dielectrics must thus satisfy relation (4) and must take into account the thickness of the given dielectric. In the following, an assembly of two dielectrics (fig. 1a) is considered, where one (1) is solid and made of a homogeneous plastic, the other (2) is porous and fibrous, and the entire assembly is impregnated with an insulating oil. If one studies these dielectrics in terms of resistivity, one can as a first approximation admit that the second dielectric 2 behaves like an oil reservoir and has a large part of the properties of oil, while the first dielectric 1, which is massive, mainly has the properties to its material. If such a massive dielectric is used alone or combined with others, and if it is impregnated with oil, according to common knowledge of the technology, one admits that this oil only serves to reinforce the dielectric's weak points and the breakdown strength in these points. One also expects a reduction of the "corona effects", especially at the edges.

When it comes to the resistivity of a solid dielectric, it has always been thought that this property belongs only to the material itself. However, within the results of the present invention, and particularly for the high field strengths that are assumed to be used, the first dielectric 1 has a resistivity that is not constant, but is firstly proportional to selected film thicknesses, and for a given thickness, secondly, this resistivity (or resistance if one considers this particular thickness) is a decreasing function of the dielectric field. Finally, this resistivity is no longer exclusively linked to the material, but is strongly dependent on the oil that impregnates the material.

This discovery has occurred by chance: a first capacitor according to the invention has been built which had an energy density of 1.3 J/cm 3 which was achieved completely reliably and reproducibly. This capacitor was impregnated with silicone.

As a second condenser, another one was built that was exactly identical to the previous one, but impregnated with castor oil. Only 75% of the previous voltage could be achieved, which meant 56% of the energy density. However, the breakthrough phase heat for castor oil is approximately the same as for silicone oil. When calculating the energy contained in each of the dielectrics, it was found that the energy contained in the first dielectric 1 represented 85% of the total energy. The energy deficit in the second capacitor must therefore come from the first dielectric 1, at least for a large part, as this capacitor had lost 44% of its total energy. The energy deficit in the second capacitor could therefore not only come from the second dielectric 2, neither for reasons of resistance nor resistivity, nor for reasons of the nature of the oil.

As the breakdown strength of the oil in the second capacitor was as good as for the first, the energy deficit for dielectric 1 could not have the breakdown strength as its origin, but only a modification of the resistance variation and a poor voltage distribution with regard to the breakdown strength, as is indicated by the invention. One therefore had to conclude that the type of oil has its significance in terms of the resistance and resistivity of dielectric 1, an opinion which, as far as is known, is not commonly accepted for solid dielectrics. When the evidence was presented, the following explanation was put forward: The probable reason is that the films, which are made of thin plastic, have a certain amount of microholes. The quantity of these holes per unit area is inversely proportional to the thickness. Above a certain thickness, the films are certainly "tight". These microholes are sufficiently small to avoid puncturing or penetration through the dielectric. They are firstly different from the film "craters" which are cavities of variable depth on the dielectric but which are larger and cause puncture at a certain applied voltage, and secondly they are different from microcavities which are enclosed in the dielectric and do not necessarily communicate. These microholes, which are more or less filled by the impregnation oil, are sufficiently numerous for a thin film, so that their presence in parallel and under high field strength leads to a current through the oil that fills the holes, as a result of ionization phenomena, impurity or residual moisture which is contained in the oil. This current leads to an actual drop in the film's insulation resistance. This phenomenon comes in addition to the polarization and ionization currents in the material itself and risks being mixed up with these. However, under the mentioned conditions or conditions, a current due to microholes is safe. The proof of this is that if one changes the type of oil and/or the related impregnation conditions and while measuring the leakage current (at high field strength) for the capacitors, such resistance variations were found that they could only have been produced by conduction through mechanical passages which was filled with oil in dielectric 1. If the plastic film had been dense, the oil would not have had such an impact.

If one considers the example with the second capacitor described above, the energy deficit is therefore a result of a resistance drop in R, (fig. la), so that the field in dielectric 1 has decreased sharply as a result of the influence of relation (3). The field in dielectric 2 increased—coincidentally—the higher and had reached its voltage limit much earlier compared to dielectric 1. The unusually strong drop of R^ in this second capacitor has been confirmed by means of autodischarge measurements. These measurements show the total leakage resistance (R^+ R2) for the capacitor, where R^ is largely included. One can also verify that the auto-discharge concerns a portion of energy that can only come from dielectric 1. The described experience shows that the energy density for the first capacitor is based on relations between resistors provided by the invention and which have been degraded by changing the impregnation agent on the other capacitor. However, it is possible that a castor oil of a different type or which has undergone a different treatment can satisfy the purpose of the invention. Oils of other types, mineral oils or organic oils, can also be considered.

A remarkable feature of the capacitors according to the invention is that the voltage distribution process for both dielectrics with regard to the insulation resistances (or leakage) only starts at a relatively high elec-

tric field strength, precisely when this distribution is useful.

When the insulation resistance drops for one or the other dielectric simultaneously and proportionally as the electric field approaches significantly the maximum field (e.g. 70~80%), it is not certain that this insulation resistance reduction occurs with the same ratio for each dielectric. In certain applications of the invention, the breakdown strength or electric field limit of each of the dielectrics is utilized by almost 80%. The fact that there are only a few self-healing punctures in this dielectric shows that the insulation resistances under these high electric field strengths are reduced in conditions that are in any case favorable for a good electric field distribution over both dielectrics. The insulation resistance at high electric field strengths already begins to decrease while the capacitor is being charged. When the electric field strength in one of the dielectrics approaches the breakdown strength, the insulation resistance of this dielectric decreases automatically, and its electric field and its voltage rise less quickly, while the field and voltage on the other dielectric will rise faster. One will understand this phenomenon if one considers the equivalent scheme described above (fig. 1b). During a charging process, the insulation resistance variation in relation to the field has the consequence that the ratio between the effective electric field strength and. the electric field boundary in each dielectric remains noticeably unchanged, thus guaranteeing the conditions which are the object of the invention. The more the electric field approaches the field boundary, the more the insulation smot stand decreases.

Of course, this automatic regulation only works within a certain area. If the field limit is reached for a given dielectric while the capacitor has only achieved a fraction of the expected maximum voltage, the capacitor will not be able to handle this voltage. The remedy in this case will consist in a better choice of the isolation resistors R1 and R2. The insulation resistance drop at high electric field strength has been confirmed using the following measuring arrangement: A capacitor has been realized in accordance with the invention consisting of a metallized paper layer and two polyester layers (for each capacitor coating), and of a liquid dielectric consisting of silicone oil.

This capacitor had a capacity of approx. 20 uF. An external resistance of 1 Gigaohm (Gohm) was connected between the contact terminals of this capacitor. The capacitor was charged with a direct voltage of 4280 volts, and the voltage reduction was recorded as a function of time. The discharge curve showed only small deviations compared to the theoretical discharge curve for a perfect capacitor (without leakage) discharging into a resistance of 1 Gohm. It was concluded that the internal resistance (R^+ R,,) in this capacitor at 4280 V is much higher than 1 Gohm, namely around 100 Gohm. Afterwards, the same capacitor was charged to 6000 V, which is its nominal voltage: already after 38 minutes the voltage had dropped to 5000 V. If one considers this voltage drop as the result of a constant insulation resistance in parallel with the resistance of 1 Gohm, but built in the capacitor (R^+ R2)/nar it a value of 1.55 Gohm. By analyzing the decline in the discharge curve in the immediate vicinity of where the charging ends, it was found that the decline is much lower compared to a natural exponential function, and an instantaneous time constant was then derived that was lower than 1 second at this particular point on the curve. This lower time constant already affects the end of the charging process and already favorably modifies the field strengths in the dielectrics before charging is achieved. From the beginning of the discharge curve, the time constant, which is then very low, starts to rise and after a time and/or less reaches the value it has at the end of the curve, which is a natural exponential function. This rise is so rapid that the capacitor very soon achieves a high insulation resistance. As soon as equilibrium is achieved between the voltages and U2i in accordance with the invention, the leakage becomes very small. The insulation resistance and time constant, which are greatly reduced at the beginning of the discharge curve, show a reduction of at least one of the two series-connected insulation resistances (Rj^ / ^2 ^ •

The strong reduction of the insulation resistance when approaching the field strength limit also has the following advantage for the life of the capacitor: If it is assumed that the described capacitor is charged for the first time to 6000 V,

must each capacitor element C^, C2 corresponding to each dielectric 1 and 2 (fig. 1b) store the same charge if the auto- or self-discharge as a result of the insulation resistances would not be present. The self-discharge, which is usually different for both dielectrics, causes one of the two elements C-^ or C2 after a time to have a charge which is less significant than the other. If the capacitor is now discharged to the outside, the elements will not be completely discharged as a result of the initially different charges before the discharge process: The element that had no internal charge loss will remain partially charged with the original polarity, while the other will be charged with the reverse polarity. The residual voltages on the capacitor elements and 0.^ are sufficiently low to consider the insulation resistance as infinite. The elements C^, C2 can then retain their charge for a long time, i.e. a few weeks or months. If you now charge the capacitor, the charging current must first completely discharge the capacitor element with the opposite polarity and then charge it with the polarity given by the charging current. If you now charge the capacitor with a given charge, this capacitor element will absorb a smaller charge compared to the other, and will therefore have a lower voltage than before, and will be less loaded during its subsequent recharges.

This effect, which is theoretical and cannot be measured, has been confirmed experimentally, not with the separate or individual dielectrics that form the capacitor (due to the fact that one cannot impose the field strengths on the individual dielectrics on a sufficient surface without permanent punctures), but with two complete capacitors (condensator windings) with different capacitance values. Two capacitors connected in series have behaved in the same way as the individual dielectrics in a single capacitor as described above. This of course only happens if you apply a voltage that is so high that one of the capacitors reaches its field limit (limit field strength) and has large losses or self-healing. After a discharge and a permanent short-circuit of the series connection, it has been measured that both capacitors remained charged with opposite polarities by approx. 10% of the final charge voltage. This effect, which has been observed on a real capacitor, turns out to be an advantage when connecting such capacitors in series because an irregular voltage distribution due to different capacitance values must be expected. A short-term difference between the voltages can also occur during the charging process.

Such errors due to poor voltage distribution do not affect capacitors according to the invention when connected in series, firstly because the capacitors are not destroyed thanks to the self-healing ability when they reach the limit of the mean field strength (or the field strength limit of one of the dielectrics ), and secondly because after the discharge you have a charge with the opposite polarity which protects one capacitor against overload during a further charging process.

The elementary capacitor according to the invention is thus particularly suitable for series connection of a large number of elements, and this constitutes a preferred application to which one will return later.

The principles described here have been verified automatically for elementary capacitors with voltages between 2 kV and 20 kV, where ethylene polyterephthalate is used as the first dielectric and metallized paper is used as the second dielectric. It has been observed that the achievable energy density for the entire voltage range was not constant. The given energy density for utilization conditions with medium severity had a value that was above 1 J/cm 3 for a voltage of approx. 5 - 8 kV (example: capacitor for 7 kV obtained with paper with a thickness of 7 nm with a metallization of 5 - 10 ohms per unit area and ethylene-polyterephthalate with a thickness of 14 ym

(2 layers of 7 ym), the whole impregnated with silicone oil; average electric field strength 333 V/ym; energy density 1.6 J/cm 3).

For elementary capacitors between approx. 5 kV and 2 kV or between approx. 8 kV and 20 kV, the achievable energy density gradually decreased from 1 J/cm 3 down to values lower than or equal to 0.5 J/cm 3. The reasons for this reduction were of a completely different nature and led to other improvements being made or made other arrangements.

Between 5 kV and 2 kV

The energy density reduction was due to a technological

limit for one of the dielectrics used.

In the following, the case of a capacitor with an assumed nominal voltage of 3 kV is considered. In order to obtain an energy density of 1 J/cm 3 , one must work with an electric field strength of 300 V/ym (a result of previous experiments). The thickness of the dielectric must therefore be 3000/300 = 10 ym. It is now intuitive that in order to keep the same mean field you must have almost the same thickness distribution when you use the same dielectrics. For the previous case with a capacitor of 7 kV, the thickness ratio between the paper and the total dielectric was 6/20, i.e. 30%. When applying this rule to the 3 kV capacitor in question, you had to have paper and ethylene-polyterephthalate of 3 ym and 7 ym respectively. However, paper thicknesses below 6 ym are not commercially available. One was thus forced to use the existing paper of 6 ym and to attach an ethylene-polyterephthalate of 4 ym to this. The unfavorable distribution of the dielectrics has consequently led to a capacitor which instead of 3 kV only holds 2 kV and has only achieved an energy density of 0.43 J/cm 3. This energy density was nevertheless acceptable compared to the known technique, but the applicant has developed paper with a thickness of less than 6 ym to make it possible to achieve the expected values.

A first technique for realizing paper with a thickness below 6 ym consists of taking the existing 6 ym paper and reducing its thickness by pressing or by calendering. The thickness that can thus be achieved is approx. 4 ym, which already allows an increase in the energy density of capacitors with a dielectric with a total thickness of 8 - 12 ym (paper ratio 1/3 to 1/2).

According to a preferred technique according to the invention, pressing is carried out by exposing the paper to a relative humidity of 20 - 40% and a temperature of 60 - 120° C. It would be advantageous to carry out the pressing using rollers between which existing paper passes on 6 - 12 ym as starting material. Instead of moistening the paper with water or more particularly with steam, it can also be moistened with a varnish that fixes the paper structure after pressing, or with a chemical product, and in particular a product that softens the fibers. You can also use several methods at the same time. Due to the non-negligible clearance between the roller bearings, it is difficult to set a very thin gap between the two rollers (e.g. 4 ym), as would be necessary to achieve the desired paper thickness. To solve this problem, the invention specifies the use of two cylindrical rollers which are in close contact, one of which is driven, and where the rollers rotate in opposite directions. The circumference of one of these rollers has a rectangular recess or depression with a width that is equal to the paper width and a depth that is less than the thickness of the paper used as starting material. This allows a close contact between the two rollers outside the recess, and the required thickness of the gap is ensured in the area of the recess. The pressing or rolling method must take a certain amount of time, so that the shape of the natural cellulose fibers that make up the paper can be modified by means of a form of creep or fly-thing.

A variant of the invention which is based on the same principle consists in wetting one paper side of ordinary condenser paper with a thickness of e.g. 6 ym with a solvent or a cellulose softener, and then pressing the paper. The amount of solvent used will be such that at least part of the capacitor paper's structure remains intact. Carbon disulphide (CS2) can be used as a plasticizer. According to the invention, another technique for obtaining this paper is based on destroying the membrane of natural cellulose fibers, so that thinner and denser cellulose fibrils are obtained than the constituent fibers. The pulp used for the production of normal paper already contains 10 - 20% of such cellulose fibrils. These fibrils can be extracted from the paper pulp, possibly diluted with water, by settling under the action of natural gravity or by centrifugation. Thicker cellulose fibers deposit more easily than thin fibrils.

If capacitor paper is produced from these fibrils as is done in a preferred embodiment, this paper can be produced with a thickness that is smaller than with fibers, due to the extremely thin, resulting coating, and this despite a sufficiently mutual, mechanical binding between the fibrils that looks like a coat. These flat fibrils also have the advantage of a 50% higher density and a correspondingly lower porosity, which results in a higher impact strength.

Cellulose fibers from coniferous wood have a diameter between 3 and 6 ym and a length of approx. 200 ym even though the corresponding fibrils have a diameter of approx. 0.2 - 0.4 ym and a length of approx. 3-4 etc. On the other hand, the so-called micro-fibrils, which are less interesting, have a much smaller diameter of approx. 0.06 ym - 0.08 ym. It will be possible to obtain much larger quantities of cellulose fibrils by treating the cellulose fibers using ultrasound. The preferred ultrasound frequency should be in the range between 400 kHz and 600 kHz. The membrane of the cellulose fibers can also be destroyed using chemical products to obtain the fibrils. This can be achieved with the help of a hypochlorite solution which, by releasing chlorine, destroys the membrane.

Regardless of the primary objective of the invention, which was to achieve a reduced paper thickness that allows an optimal division with respect to the associated dielectric (polyethylene terephthalate), this operation has other advantageous effects: The capacitor papers described above are much thinner than the plain paper. Considering that they are obtained from a greater thickness or smaller fibers, they have a higher density and consequently the advantage that they have smaller microcavities than ordinary paper, resulting in a higher impact strength than the impact strengths of the ordinary papers.

In accordance with the invention, the use of such papers makes the capacitor smaller, firstly because the reduced paper thickness results in a higher capacity, and secondly because the higher density and higher breakdown strength allow an increase in the capacitor voltage and consequently the accumulated energy compared to ordinary capacitors. Another advantage of this new paper is its ability for optimal metallization. Metallization on normal papers often has tips that penetrate deep into the paper porosity despite the varnish or lacquer coating. These tips cause punctures. As the new paper has less porosity, it will be less susceptible to this penetration. The metallization method has also been improved. The usual method consists in vaporizing metal, e.g. zinc, essentially at right angles to a paper tape. A new method for metallizing capacitor paper proposed according to the present invention foresees depositing the metal at an acute angle which is less than 75°, preferably with an angle between 10 and 45°, the optimum being an angle between 25 and 45° . It is preferred that the "path" of the metal particles has an orientation with a component in the direction of the movement of the paper. However, other path orientations of the metal particles are possible. The aforementioned embodiment has the advantage that the particles cannot penetrate deeply into any porosities in the paper web due to they hit the paper in an oblique direction. It is mainly on the aforementioned sections of paper that the metal is deposited. As a result, the paper's impact resistance is increased, and you will consequently have a capacitor that can be used for a higher voltage. The above-described metallization device for capacitor paper is characterized by a path for the metal particles that has a sharp angle of incidence in relation to the paper. This new paper with a thickness of 2.5 - 6 ym will, according to the invention, allow an advantageous realization of capacitors between 2 and 10 kV. With the smallest paper version or with special impregnation conditions can also be expected with versions between 600 V and 2 kV.

Between 8 kV and 20 kV

The gradual reduction in energy density observed on designs between 8 kV and 20 kV had another cause which is commonly referred to as edge or fringe effect. Some authors have shown that the edge of the metallic coating that makes up the capacitor coating, in this area of the dielectric forms a higher electric field strength than in the rest of the coating, which causes arcing or sparking at this location at a much lower voltage than in other areas. In the present case, these arc formations are not destructive as they are self-healing, but nevertheless they severely limit the nominal voltage of the capacitor. Various works have shown that the relationship between the edge field and the homogeneous dielectric field is a function of the square root of the dielectric thickness. This relationship has also been experimentally verified. This means that the capacitors according to the present invention in principle have an edge effect which occurs with a much higher voltage than for known capacitors, since for a given voltage you work with more than double the electric field with half the thickness. According to the invention, these capacitors allow, in principle, that their voltage can be raised to twice that of known capacitors before reaching the level where the edge effect becomes apparent.

As a result of this fact and in anticipation of the series connection of capacitor batteries, the optimum voltage for the present elementary capacitors has been set to a value of approx. 5 to 8 kV to take advantage of the maximum energy density for the entire battery. It is certain that the manufacture of elementary capacitors of 8 - 20 kV based on the principle according to the invention is not excluded,

either with an energy density that is lower than 1 J/crn^ or with a higher energy density, if a supplementary improvement is linked to the invention.

For the case where the second dielectric 2 is formed by several layers, such an improvement consists in particular in introducing a layer of the first dielectric 1, but not metallized, between two such layers. For example, one can have the following structure for each capacitor coating: Dielectric 2 (metallized) - dielectric 1 - dielectric 2 (not metallized) - dielectric 1. The electric field strength distribution as a function of the breakdown strengths will behave in the same way for a dielectric 2 as is metallized or not, and for any arrangement of the different layers.

It can also be noted that the principle of maximum utilization of a double dielectric according to the invention can also be applied in the case of a non-self-healing capacitor with thick capacitor coatings (aluminum layers) instead of metallized capacitor coatings. An advantageous application of the capacitor according to the invention is the formation of series batteries and series-parallel batteries. As indicated above, such elementary capacitors will have e.g. voltages of 5 - 8 kV if a maximum energy density is expected. The diameter of the elementary windings can vary between 10 and 100 mm and the height between 20 and 100 mm. A very valuable dimension is, for example, a winding with a diameter of 50 mm and a height of 80 mm. For a large part of the intended applications, however, the capacitor must emit very significant currents during the discharge process. However, according to some literature (for example the journal "Electronique de puissance" No. 1, supplement to No. 724 of "Electronique Actualité" dated September 9, 1983, page 69), capacitors with metallized capacitor coatings are in principle excluded for very high currents (500 A to 50,000 A for an elementary capacitor). In connection with the present invention, on the other hand, people have been surprised to see that in certain cases it was possible to emit repeated currents of approx. 10,000 A without damage, when using capacitors according to the invention. This is very surprising if you consider the high resistance per area unit (e.g. 7.5 ohms) for the capacitor coatings. To explain this fact, the following should be noted: A capacitor is formed by ribbons with an unwound length L and a width h (corresponding to the height of the winding). One can also consider the wound capacitor as an unwound which is composed of a number of L/h squares which are connected in series, but which are also connected in parallel at the level of the "side covering formation" (English: "scooping" ) (winding disc side metallization). The internal resistance is thus inversely proportional to this ratio L/h for a given film resistance. Since now the dielectric thickness in the capacitors according to the invention for a given voltage is very low compared to the known capacitors (a factor of less than 1/2), the length L will therefore be more than double for a given winding section, compared to known capacitors. This is the reason why the capacitor according to the invention has particularly high L/h ratios for the range of the intended voltages (500 - 10,000 V), allows peak currents of 5,000 - 100,000 A for an elementary capacitor and exhibits very low inductances. The highest L/h ratios, and consequently the highest currents, are obtained on capacitors with low height and with a large winding diameter (eg a diameter of about 70 mm and a height of 25 mm).

These high currents are even more surprising as the metallization used is very thin (eg 7.5 ohms per unit area). This result depends on the capacitor principle itself and on the capacitor's geometry. However, to further back up the current limit (or improve lifetime), one can reinforce the metallization edges in the "scooping" area, in the inactive part of the capacitor (the edge of the second capacitor coating) by a factor of 2 - 5 (e.g. a surface resistance of 1 - 3 ohms), because this is the area where the current density is most significant. Another advantage of this flat winding form is to build high-voltage capacitors where such elements are stacked up and connected in series inside an insulating tube.

The invention shall be described in more detail below in connection with preferred embodiments with reference to the drawings, where fig. 1 shows a diagram of a capacitor with a double dielectric, fig. 1b shows the equivalent diagram according to the invention of the capacitor in fig. 1a, fig. 2 shows an example of the film composition in a capacitor according to the invention, fig. 3a and 3b show the shape of windings according to the invention, fig. 4 shows a schematic view of a paper metallization device, fig. 5 shows a side view of an assembly for pressing capacitor paper, and fig. 6 shows a longitudinal section along the line VI - VI in fig. 5. Fig. la shows a schematic representation of a capacitor with a double dielectric. The two dielectrics 1 and 2 have respective thicknesses e^ and e2 and are contained between two capacitor coatings (English: armatures) 3 and 3' to charge the capacitor e.g. with respective polarities + and -, and to discharge this. The two dielectrics 1 and 2 have respective relative dielectric constants e, and £27 and when the capacitor is charged to its nominal voltage U, the respective electric field strengths are E^ and E2, and the resistivities in these fields are r^ and r^. The voltages that appear at both dielectric ends are U, and U2• Fig. 1b shows the equivalent diagram of the capacitor in fig. let. This equivalent kernel is by no means conventional. It is characteristic of the invention and indicates a preferred method for utilizing both dielectrics with the same ratio. This preferred method consists in the voltage distribution and across the contact terminals of the two capacitors C-^ and C~ which symbolize the two dielectrics 1 and 2, corresponds to the relation Fig. 2 shows an example of how the capacitor can be built in accordance with the invention. The figure shows two multi-layer or multi-layer plates A and B which are completely identical. Each composite layer comprises a paper 5 or 5' which is provided with a zinc metal layer 4 or 4', and two plastic foils 6 (6') and 7 (7<1>) of polyethylene terephthalate. The paper 5 (5') which is provided with the metal layer or metal layer 4 (4<1>), is a so-called metallized paper. The paper 5 (5') itself is a varnish-coated paper. The varnish coating consists of cellulose acetate, but cellulose acetobutyrate can also be used. The varnish coating provides a smooth surface which has a thickness of 0.5 um. The metal layer 4 does not extend to the right side edge of the paper 5, but it extends to and exceeds the left edge, while the metal layer 4' does not reach the left edge of the paper 5', but reaches to and exceeds the right edge. The resulting edges or margins on both sides amount to approx. 4 mm. The papers 5, 5' and the plastic foils 6 (6<1>) and 7 (7') have the same width. Those in fig. 2 right and left end surfaces 8 and 9 of the capacitor, which are formed by winding the composite layers A and B, are provided with sprayed-on metal coatings 8 and 9 of zinc which are in contact with the metal coatings 4 and 4', and to which connecting wires can be attached by soldering. For very high currents it is possible to tin these "side coatings" ("scooping"), or to apply other metals to the zinc coatings 8 and 9 instead of these coatings. The metal layer 4, 4' has a surface resistance of 7.5 ohms and a thickness of 15 mm. Underneath the zinc coating is a silver coating with a thickness of 0.2 - 0.5 nm. The dry paper has a specific insulation resistance of 10,000 Megohm x microfarad (measured at low field strength). Impregnated with silicone oil, it has a specific insulation resistance of approx. 15,000 Megohm x microfarad and a maximum field strength or breakdown strength of approx. 200 V/um. The specific insulation resistance of the polyester used is 50,000 Megohm x microfarad, and its breakdown strength is approx. 600 V/um. The paper layer 5 (5') and the two plastic layers 6 (6') and 7 (7<1>) each have a thickness of 7 ym. The width of the paper layer and the plastic layers visible in fig. 2, is 80 mm in the example shown, and the length that is not visible in the figure amounts to 100 m. As shown in fig. 3a and 3b, the layers are wound on an insulated mandrel 10, the layers being in close contact with each other and the finished capacitor is connected by means of the coatings 8 and 9 and impregnated with silicone oil, and the capacitor roll has a capacity of 22.5 uF. A self-adhesive polyester tape that has a thickness of approx. 70 ym and is wrapped around the condenser roll, preventing the roll from coming loose accidentally. During the unwinding of the capacitor, the foils are stretched with a force of 20 - 25 N.

The dielectric constant of the polyethylene terephthalate used is 3.2, and the dielectric constant of the paper impregnated with the liquid dielectric is 4.8. The double dielectric thus formed has a total average dielectric constant of 4.2.

This elementary capacitor has a nominal voltage of 7000 V and, considering its capacitance of 22.5 yF, an energy level of 551 J. Its volume is 344 cm and its energy density is 1.6 J/um. As soon as the capacitor is charged to 7000 V, the average electric field strength in the dielectric is equal to 333 V/um. The condenser has a L/h value of 1250

and withstand discharge currents of approx. 20,000 A. Samples of elementary capacitors are regularly held back from the production line and impregnated for testing. Using an oscilloscope and a printer, the capacitors' charging curves are measured up to the nominal voltage of 7000 V and further up to a test voltage of 7700 V. If self-curing effects occur, they should be low and disappear during the next charging procedures. Furthermore, the charging curves of the capacitors are recorded as described above, and as a third step, the printer strips are compared with reference printer strips for a reference elementary capacitor. One analyzes the discharge curve and calculates the time constant and the insulation resistances for several predetermined points on the curve, as well as the utilization percentage for each dielectric. As a normal case, you find, for example, 70 - 80% for each of the capacitors. These controls allow monitoring and possible correction of the dielectric quality, and in particular of the oil quality and the impregnation conditions. Before one knew the principles for such utilization of both dielectrics and the influence of the isolation resistances at high field strengths as described in the present description of the invention, it was in reality impossible to achieve a reproducibility and series progress

position of capacitors that achieved the expected field and energy density. Of course, a few unique prototypes that achieved such specifications have been achieved, but this has been purely by chance, and it has been impossible to achieve these results again until the time when the principles described here were discovered which provide the systematic mastery of the considered parameters. As an indication, the impregnation temperature is 100° C, the vacuum is 10 -2 to 10 -3 mm Hg, and the impregnation time is 24 - 48 hours. However, these conditions are constantly being adjusted, and may also differ from one model to another.

The capacitors for which the manufacturing processes have been described above are mostly intended to be built into boxes or containers individually, or to be connected as batteries. An example of a battery consists of two windings of the type described which are stacked in an aluminum container with a diameter of 75 mm and a height of 180 mm. The windings can be impregnated before or after the adaptation. A preferred technique consists of mounting the non-impregnated but pre-dried windings in the container and crimping the top to leave holes for the contact terminals. These holes will be used for the treatment. The windings inside the container are then dried again, but under vacuum for 48 hours, and then impregnated under vacuum for 24 hours. The container will be filled at the same time. At the end of this procedure and after breaking the vacuum, but without removing the capacitors from the bath, you can also close the holes by soldering in oil. The obtained capacitor has a capacity of 45 yF and a nominal voltage of 7000 V. It will be used, for example, as a discharge capacitor for flashing lamps for laser pumping. If the discharge frequency is lower than or equal to approx. 1 Hz, the capacitor will be able to work at its rated voltage or slightly below. If the repetition rate is higher, e.g. 20 - 30 Hz, the capacitor can be used in a "derating mode" at a voltage of 2000 - 3000 V. This same elementary capacitor can also be used to build batteries with very high voltage and/or very high energy, when connected in series or parallel. The voltages can reach several hundred kilovolts, and several megajoules. Certain models are suitable for building Marx generators.

A second example consists of an elementary capacitor that uses the same types and thicknesses of the dielectric, but instead of using a long cylinder as before (fig. 3a), a flat one is built (fig. 3b) by using films with a width of 25 mm. The winding height will then be 25 mm, the winding diameter 7 4 mm, the volume 107.5 cm 3 , the capacity 4.5 yF, the nominal voltage 7000 V, the energy 110 J, the energy density 1 J/cm 3 and the average field strength 333 V/ym. The length of the film is 100 m and the nominal L/h ratio is 4000.

This elementary capacitor allows currents of 500 - 100,000 A. The energy density of this embodiment is less, compared to the previous one, due to the fact that the edge proportion, which is subtracted twice from the width of the band (English: nappe), is more significant.

The great advantage of this model is both its significant L/h ratio, its ability to withstand high currents, and its low height. This elementary capacitor is quite rightly intended for the manufacture of batteries by stacking these flat cylinders in a tube with a diameter of 75 mm. In this way, you will get a capacitor for approx. 50 kV using eight stacked elements which give a container height of approx. 200 mm. Such a capacitor is suitable for gas lasers. It is also possible to manufacture models for 100 kV which have the manufacture of Marx generators as one application. However, since the energy density with a roll height of 25 mm is lower due to the edge proportions, some models of elementary capacitors with a height between 25 and 80 mm have also been planned.

A third design example consists of an elementary capacitor built in the same way as the first (fig. 2 and 3a), but with paper thicknesses of 6 ym (5 and 5') and with thicknesses of 5 ym for each of the layers 6, 6' and 7, 7'.

The winding diameter is 49 mm, the volume 150 cm^, the capacity

15 yF, the nominal voltage 5000 V, the energy 187 J, the energy density 1.25 J/cm 3 and the average field strength 312.5 V/ym. The film length is 53 m and the L/h ratio equal to 662. This elementary capacitor allows currents of 10,000 A. It can be mounted as a battery, as for the first model, in a container with a diameter of 50 mm and a height of 180 etc. As an example, the obtained capacitor of 30 yF is used

and 5000 V at its rated voltage for cardiology defibrillators. For other applications, such as certain fixed lasers or certain copiers, it will be capable of operating either at its rated voltage, or at a lower voltage in accordance with operating conditions. These elementary capacitors can also be used for batteries with very high voltages or energies.

A fourth capacitor model uses the new metallized paper with a thickness of 4 ym for the layers 4, 5, 4', 5', and a polyester film for the layers 6, 7, 6', 7<*>. The strips (English: nappes) with a width of 80 mn are wound on a two-part mandrel that has a diameter of 3 mm and which is removed after winding. The diameter of the winding is 20 mm, the volume 25 cm 3, the capacity 12 yF, the nominal voltage 3 kV and the energy 54 J. The average field strength is 375 V/ ym and the energy density is 2.15 J/cm 3. This is a capacitor for a implantable defibrillator. As the capacitor in such a device is not used very often and remains charged only for a very short time, the dielectric can be utilized beyond the limit used for industrial applications. Thus, the dielectric will be used at more than 90% of their breakdown strength. Under these conditions, this model can therefore also exceed the field strength and the indicated energy density.

A fifth and final example uses the same dielectric as for the fourth capacitor design, but is wound to a diameter of 74 mm. This design has a capacity of 172 yF and a nominal voltage of 2.5 kV. Its working voltage can vary between 2 and 3 kV according to the conditions of use, which corresponds to an energy density between 1 and 2.25 J/cm 3. The applications are different (lasers, defibrillators, etc.).

In this last example, you can also replace

the 4 ym paper with fibril paper of 2.5 ym and use an ethylene-polyterephthalate film of 2.5 - 4 ym. With film widths of 40 - 80 mm and a winding diameter of 15 - 20 mm, other models for implantable defibrillators with capacities of 10 - 30 yF and energy densities of 2 - 3 J/cm 3 can be built. In short, the same dielectric assemblies can be used with different field strengths and energy densities in accordance with the conditions of use and/or the expected lifetime: - 150 - 300 V/ym and 0.4 - 1.2 J/cm 3 for filtering, switching, discharge purposes where the requirements for medium working power or lifetime are high. These capacitor models can also be used below their nominal level (not rated value) for severe conditions; - 300 - 400 V/ym and 1.2 - 2 J/cm 3 for filtering, switching or discharge purposes where the operating conditions are less severe (external defibrillators or low repetition rate lasers); - 400 - 500 V/ym and 2 - 3 J/cm<3>for low working rate, for example implantable defibrillators, max 500 shocks and almost immediate discharge after the charging process.

Other materials can be used without changing the principles of the invention: - other metals or elements are possible for the metallization, such as aluminium, gold, silver, palladium or pure amorphous carbon; - other dielectrics are possible instead of dielectric 1, e.g. polypropylene, polycarbonate, polysulfone, polystyrene,

etc. ;

- other fiber structures are possible instead of dielectric

2; - other impregnation oils in accordance with the criteria according to the invention are possible.

Fig. 4, which shows a schematic representation of the paper metallization process, shows (not to scale) a vacuum chamber 100 in which is shown a mother or supply roll 102 which is filled with capacitor paper, and a receiving roll 104 on which the paper 106 will be wound. The condenser paper 106 passes along deflection rollers 108

and is in the upper part of the vacuum chamber 100 closed or shielded by means of a partition or blender 110, 112. A nozzle 101 connects the chamber to a vacuum pump. A melting vessel 116 is arranged in the chamber 100 in which zinc 118 is caused to evaporate. Electrodes 120 and 122, which are connected to a negative voltage and shield the supply roll 102 and the receiving roll 106, avoid a vapor deposit thereon. The path the vapors would have taken in the rolling directions if these screens were not present is illustrated by dashed lines. Another electrode 124, which has a positive potential, is located behind the paper 106 behind an opening 126

in the blender 110, 112. This opening 126 is situated in such a way that the metal vapors emitted from the melting vessel 116 and passing through the opening have an incidence angle of 45° towards the paper 106. If necessary, this angle can be changed.

In the example shown, the distance between the melting vessel 116, which is made of tantalum, and where the paper 106 is heated by the vapors behind the opening 126, is equal to 15 - 30 cm. The temperature of the metal is 400 - 800° C. The linear speed of the paper is 2 - 4 m/min. The potential difference between UQ

for the melting vessel and U+ for the electrode 124 is 400 - 800 V.

Electrodes 120 and 122 are at a negative potential of -100 V to 200 V relative to the vessel. The pressure in the chamber 100 is approx. 10<-2>- 10<-5>Torr. The metal deposit on the paper depends on the metal temperature, the angle of incidence, the paper speed, the voltages, the pressure in the chamber and the distance between the melting vessel and the paper.

Fig. 5 and 6 schematically show a device which has three metal rollers 200, 202, 204 with their axes in the same plane. The middle roller 200 is driven and transfers the turning movement by friction to the other rollers. The circumference of the middle roller 200 has a recess or recess 206 with a width of 85 mm and a depth of 4 ym. A paper tape or paper tap 108 of commercially available condenser paper with a thickness of 6 ym from a supply roll 210 passes in front of a steam nozzle 212 which moistens the web. The paper web passes further between the upper roller 202 to the middle roller 200, and further between this middle roller and the lower roller 204. The paper web passes exclusively in the area of the recess 206. The pressed paper is then wound up on a receiving roll 214.

A second nozzle 216 is arranged close to the middle roller 200

which can be used to evaporate water, varnish or other products, in accordance with the procedure that one wishes to use. The steam nozzle 212 impregnates the paper with a rela-

tive humidity of 25 - 35%. The rollers 200 and 204 are heated to a temperature between 80 and 110° C using infra-

red heating elements. The rollers 200 and 204 have the same diameter, which is 30 cm. The miter roller 200 is rotatable about an axle which is attached to a chassis 222, while the upper and lower rollers 202 respectively. 204 can be displaceable in the height direction and is pressed against the middle roller with a hydraulic piston system 224 which is fed by a hydraulic pump 226. Adjustment devices enable control of the pressure and thus allow an optimization of the force with which the rollers are pressed against each other. In the design example, this pressure is 1000 - 1500 dN/cm 2. The paper has a speed of 8 m/min - 10 m/min. After the pressing process, the paper has a thickness of 4 - 4.2 ym. After such a process, the paper has a density of approx. 1.4 - 1.5 g/cm . Paper made entirely of fibrils has the same density. The paper's impact resistance rises to 400 V/ym if it is impregnated with silicone oil. It can also be considered to increase the given surface resistance of 2 - 30 ohms above this value, in particular by using existing paper or new paper and/or the new metallization methods described above.