US3035201A - Cold cathode switching devices - Google Patents

Description

y 1952 T. M. JACKSON ETAL 3,035,201

cow CATHODE swncnmc DEVICES Filed April 50, 1958 w 4 Sheets-Sheet 1 3 I I I I BC 30 I I I I Inventor TMJackson EALSeII Attorney y 15, 1962 T. M. JACKSON ETAL 3,035,201

cow CATHODE SWITCHING DEVICES Filed April 50, 1958 4 Sheets-Sheet 2 bsagfs gaggg Attorney y 1962 T. M. JACKSON ETAL 3,035,201

COLD CATHODE SWITCHING DEVICES 4 Sheets-Sheet 3 Filed April 30. 1958 Inventor 5e. By M Tlfiulackson EA 1 ll 5711215324214 ol I! Attorney y 1962 T. M. JACKSON ETAL 3,035,201

COLD CATHODE SWITCHING DEVICES Filed April 30, 1958 4 SheetsSheet 4 .L 11 1; h L

Inventor: tnJackson EAR 5c" Attomy United States Patent Ofiice 3,635,201 Patented May 15, 1962 3,035,201 COLD CATHODE SWITCHING DEVICES Thomas Meirion Jackson and Eric Andreas Frederik Sell,

London, England, assignors to International Standard Electric Corporation, New York, N.Y.

Filed Apr. 30, 1958, Scr. No. 732,091 Claims priority, application Great Britain May 17, 1957 7 Claims. (Cl. 313-497) The present invention relates to cold cathode gas-filled tubes and circuits using such tubes for high speed electric switching arrangements.

In recent years much work has been done in the telephone and telegraph switching field with multi-electrode cold cathode tubes using the principles of ionisation coupling within the gaseous atmosphere of the tube and flow transfer from one to another of an array of ionisation-coupled cathodes. The application of the general switching techniques of telecommunications to the computer field leads to demands for higher and higher switching speeds; but our present knowledge of the physics of the glow transfer processes indicates that there is a basic limitation in the speed of some of these which prohibits the use of glow spread or glow transfer devices at pulse repetition frequencies exceeding some tens of kilocycles per second. Attention has therefore been directed to switching circuit arrangements employing separate discharge tubes with external coupling between them. Thus, in US. Patent 2,631,261 glow discharge trigger tubes are disclosed which have been operated at pulse repetition frequencies in the neighbourhood of one hundred kilocycles per second. The limitations on higher speeds are the formative delay time (of the trigger gap in the tubes just mentioned) and the deionisation time. Both these quantities require careful definition, which will be gone into later, but for the moment it is suflicient to say that deionisation time is determined by applying to a gap across which glow discharge is being maintained an extinguishing pulse so as to reduce the anode-cathode voltage; with a given pulse amplitude the minimum width of the extinguishing pulse which ensures that the glow discharge does not restart after passage of the pulse is a measure of the deionisation time. Similarly, to fire an extinguished gap into glow discharge a pulse is applied to increase the anode-cathode voltage; for a firing pulse of given amlitude the minimum pulse Width which will ensure that glow discharge may be maintained after passage of the pulse is a measure of the formative delay time of the gap.

It is relevant to add here that short deionisation times are favoured by concentric or coaxial geometries with the cathode surrounding the anode, but for minimum formative delay times it has been concluded that the anode should surround the cathode. In the said US. patent, the influence of gap voltages on the speed of transfer from trigger to main gap was considered and, as a result, three principal embodiments were illustrated, one employing an outer cathode and one an outer main anode in respective coaxial constructions, while the third used a planar electrode construction, giving a compromise between the conflicting requirements of short deionisation time and low transfer voltage. The factors govering formative delay time in a diode have some points in common with those governing transfer voltage in a three electrode trigger tube. In the present invention we are primarily concerned with the physics of diode gaps and not with trigger tubes, but from what has just been said, it might appear that for high speed operation, a planar geometry would present the best compromise between conflicting requirements of a short formative delay time and a short deionisation time.

Our investigations into formative relay and deionisation times in diode cold cathode tubes have red us to the discovery that, contrary to what has been suggested above, minimum formative delay times, as well as short deionisation times, are obtained with an electrode geometry in which the cathode effectively surrounds the anode in either a coaxial or hemispherical arrangement. This is only true, however, subject to this important proviso viz; a substantial amount of ionisation, in one important class of embodiments of the invention in evidence as a large pre-breakdown current, must be present in the gap before the application of a firing pulse. Such large pre-breakdown ionisation cannot be maintained in a gap having planar electrodes; it would cause the gap to break down spontaneously into glow discharge.

The implication of our discovery and the basis of the present invention is this: since a large part of the formative delay time is normally occupied in pre-breakdown ion production, if there exists some means for copious ion production in the gap before the application of a firing pulse and if the gap geometry permits the existence of this ionisation without causing the gap to break down spontaneously into glow discharge in the presence of a steady anode-cathode applied potential exceeding the glow discharge maintaining voltage, then a significant fraction of the normal formative delay time is eliminated.

The present invention, therefore, provides a switching device comprising a cold cathode gas-filled electric glow discharge tube having a discharge gap formed between a cathode surrounding its anode so that the lines of electric force diverge from the anode to the cathode, the device further comprising means for maintaining in the said gap a large pro-breakdown ionisation level, for example by means of corona discharge in the gap, such that on the application of a firing pulse of 60 volts overvoltage the formative delay time as herein defined is less than one microsecond.

By large pre-breakdown ionisation level we mean large in comparison with that which could be tolerated in a gap between planar electrodes having similar breakdown and maintaining voltages and passing the same glow discharge current. By way of example, a planar diode gap having a breakdown voltage of 400 volts and passing l0 miliiamperes of glow discharge current at 200 volts maintaining potential may pass a pre-breakdown current of very much less than one microampere, whereas a coaxial diode constructed in accordance with the principles of the present invention for operation at the same voltages and glow discharge current can support a pre-breakdown current of ramp.

We have further found that the large pre-breakdown ionisation may be very efiectively provided by means of corona discharge, the anode radius being made sufliciently small for the standing electric field intensity in its neighbourhood to exceed the dielectric strength of the gasfilling. The nature of the gas-filling is, of course, also a matter of some importance, but the principles governing the choice of gas mixtures are commonly known by those skilled in the art and need not be discussed at present. We have also found, however, that given an appropriate gas mixture, the deionisation time is largely a matter of anode-cathode separation and varies linearly with that separation.

On the basis of the above findings the invention pro vides a diode tube with an outer cathode and inner anode so that the lines of electric force diverge from the anode to the cathode, having a glow discharge deionisation time, as hereinafter to be more precisely defined, of less than two microseconds and in which a continuous corona discharge may be maintained without breakdown into glow discharge during the application of an anode-cathode potential exceeding the glow discharge maintaining voltage but less than the glow discharge breakdown potential.

With cold cathode glow discharge tubes using corona self-priming in accordance with the invention we have been able to achieve deionisation times less than 1 ,usec. and still shorter formative delay times.

We are not aware of any previous glow discharge tubes whose construction allows the attainment of both the short formative delay times and short deionisation times of the present invention. Since, however, as has already been emphasised, the present invention is based on our discovery, so far as formative delay time is concerned, of the importance of large pre-breakdovm ionisation in a glow discharge gap, and because of our finding that such ionisation may very conveniently be provided by corona discharge at the anode of the glow discharge gap, we have investigated the properties of some existing corona stabiliser tubes when operated under circuit conditions which allow, at will, either corona or glow discharge to be maintained as alternative states of operation. These corona stabiliser tubes are designated for voltage stabilisation and depend for their operation upon the wide variation of corona current, measured in microamperes, accompanied by corresponding small changes in anodecathode voltage. They are not intended at all for operation in the glow discharge regime, though some types are capable of passing useful glow discharge currents. Some corona stabiliser tubes we have examined under these conditions of operation have been found to give short formative delay times, but the anode-cathode separation is invariably too great to meet our deionisation time requirement. Nevertheless, we have been able to operate an existing corona stabiliser tube as a glow discharge tube using the principles of the present invention in a special pulse operated circuit at a pulse repetition speed of about 300 kc./s., the speed being limited by the 2.3 ,usec. deionsation time of the tube. This speed, though well below that we can attain with tubes specially constructed for the purpose, still marks a notable improvement on what we believe hitherto to have been the fastest operation of glow discharge tubes.

, Although, as has been discussed, one mode of operation of the invention is to use corona current for selfpriming a diode glow discharge and so providing the large pre-b'reakdown ionisation, it is characteristic of corona discharge that a large change of corona current is accompanied by a very small change of voltage across the gap in fact this is what makes corona discharge useful in a stabiliser tube. The transition between corona and glow discharge, with its much lower maintaining voltage, is very abrupt in such tubes. It follows that special circuit arrangements must be provided to cater for manufacturing variations from tube to tube and to allow for adequate tolerances on component and supply voltage values if it be desired to use corona self-priming in carrying out the invention. Accordingly we provide a high impedance circuit in series with the cold cathode diode, the current-dependent voltage drop of the circuit absorbing supply voltage variations and allowing for component and tube tolerances, ideally only corona current passing through this series circuit, and we further provide a low impedance by-pass circuit, allowing passage of the much larger glow discharge current, access to the by-pass circuit being controlled by a gate, normally a crystal diode, which gate is actuated by the pulses applied for switching the glow discharge on or E.

Use of corona discharge as a self-priming mechanism in a glow discharge tube is not the only way in which the principles of the invention may be applied. Indeed, due to factors such as the limited'backward resistance of commonly available crystal diodes, it may be more convenient to provide the large pre-breakdown ionisation by other means. In particular an auxiliary glow discharge in the field of the main gap may be utilised, the main gap geometry, as before, providing a short main gap deionisation time and permitting the large amount of ionisation from the auxiliary discharge to be maintained without spontaneous breakdown of the main gap into glow discharge. The auxiliary glow discharge then furnishes what we may call an artificial corona in the main gap, but its maintenance may be made largely independent of the main gap potentials.

For this artificial corona mode of operation the invention provides a particularly favourable type of tube: construction which is a modification of the tube first specified above. In these modified tubes, in place of a puncti form or filamentary anode used to facilitate corona dis-- charge, the single anode is replaced by a pair of elec.-- trodes, providing between them a gap for the passage of an auxiliary glow discharge, thus forming electrode: means for setting up the artificial corona in the tube.- The outermost electrode of the. resulting triode remains. the cathode of what is now the main gap, and one electrode of the additional pair becomes the main gap anode, the main gap geometry providing the necessary short deionising time and also permitting a large number of' ions, compared to that permissible with planar electrode geometries, to be present in the gap without causing it to break down into glow discharge in spite of the application of a steady anode-cathode voltage exceeding the: maintaining voltage. The formative delay time of the main gap is thus affected in analogous manner to that of the tubes with corona priming previously discussed, but, the tube can be used in a simpler circuit in which component and supply voltage limits may have quite wide tolerances.

In a coaxial electrode arrangement of a triode tube according to the invention a hollow cylindrical main cathode surrounds a cylindrical main anode, which may be split into two hollow cylinders with a gap between the two halves, and a fine wire auxiliary cathode is strung along the axis of the anode. The priming discharge occurs between the auxiliary cathode wire and the edges of the two anode halves so that positive ions are continually present in the gap between the two anode halves opposite the cathode.

in a spherical arrangement according to the invention the main cathode is formed as a dome covering the tips of a pair of fine wires which are connected as anode and auxiliary cathode respectively.

Both the above electrode arrangements have been used in tubes which have been operated at pulse repetition rates of one megacycle per second.

The invention will now be described in fuller detail with reference to the accompanying drawings in which:

FIG. 1 is a circuit diagram of a known arrangement using gas-filled diodes as relay tubes;

FIG. 2 shows the general form of the curve relating deionisation time of a glow discharge gap with anodecathode voltage existing during an extinguishing pulse;

FIG. 3 shows curves relating to the potential variation across a discharge gap under various conditions of electrode geometry;

FIG. 4 shows typical curves relating applied voltage and pre-breakdown and glow discharge currents in a gas-filled tube;

FIG. 5 illustrates the construction of a corona-primed diode according to the invention;

FIG. 6 is a modification according to the invention of the circuit arrangement of FIG. 1 for use with tubes to be described with reference to FIG. 5;

FIG. 7 shows the construction of an artificial corona primed tube according tothe invention;

FIG. 8 shows an embodiment of the invention alternative to that of FIG. 7; and

FIG. 9 shows formative delay time and deionisation time curves for the tubes of FIGS. 7 and 8.

In order to illustrate the use of diode cold cathode tubes in a glow transfer switching circuit, the arrangement disclosed in British Patent 727,415, with a modification of the split pulse drive system is reproduced in FIG. 1 of the accompanying drawings. Part of a counting chain is there shown comprising an even number of glow discharge diode tubes T T T These tubes are inserted in series with respective rectifiers W to W,,; each poled to present a low impedance path to the discharge current through its own tube, and cathode load resistors R to R between one or other of a pair of positive busbars la and 1b and an earth line 2. The odd numbered tubes have their anodes connected to busbar 1a and the even numbered tubes are fed from busbar 1b. Respective capacitors C to C connect the cathodes of tubes T to T to the junction of the rectifier and resistor in the cathode circuit of the immediately preceding tube. Capacitor C joins the cathode of T to the junction of W and R The busbars 1a and 1b are fed from the respective outputs of a bistable device BC which is supplied with pulses 3. The output BCl feeding busbar 1a supplies in synchronism with pulses 3 derived rectangular pulses 3a whose potential excursion is from a value V sufiicient to maintain discharge in any one of the discharge tubes T, but supplying less than the breakdown voltage to these tubes, to a value V below the maintaining potential of the tubes. The output BC provides similar pulses 3b to busbar 111 but in antiphase with the pulses 3a, so that when the voltage of busbar 1a changes from V to V that of busbar 1b falls from V to V and vice versa.

The operation of the circuit is as follows. Assume that tube T is firedi.e. it is the only tube which is passing a glow discharge. The discharge current, passing in the low resistance direction through W sets up a voltage across the load resistor R The cathodes of the remaining tubes are all held at earth potential. The busbar 1b is at voltage V and busbar 1a at V On the arrival of the next pulse 3 the anode of T is taken down to V so that the discharge is extinguished, and, simultaneously, the anodes of T and the other odd numbered tubes are raised to the voltage V This increase in anode-cathode voltage is insufficient to fire tubes T to T,, but the cessation of current through R carries the cathode of T negative by an amount sufficient to cause this tube to fire. Until it fires, the discharge circuit of C includes the high back resistance of rectifier W so that the cathode of T is held negative. On the arrival of the next pulse 3, the discharge in T is extinguished and T is fired. Thus the discharge is stepped from tube to tube along the counting chain by each fresh pulse 3. Since the anode of a tube which has just been extinguished is held at V below the maintaining potential, for the whole of the inter-pulse interval, the limit to pulse repetition speed is reached when the time interval between the arrival of pulses 3 equals the deionisation time of the tubes T, which is normally the longest time lag of the tubes and circuit.

We must now consider precisely how deionisation time is to be defined and the factors which govern it, keeping in mind the general manner of use of the discharge tube, of which the arrangement of FIG. 1 is typical but by no means exclusive. From the point of view of the tube user, it is required that a steady glow discharge be maintained for a time which may be indefinitely long, that this discharge be extinguished by means such as a pulse, and that the discharge remain extinguished after passage of the said pulse until re-fired by another pulse. A steady voltage must then normally be applied to the tube capable of maintaining the required discharge current but not so great as to cause breakdown without augmentation from a firing pulse or the 'like. The glow discharge must be extinguished by temporary reduction of the said steady voltage and the tube must not refire when the steady voltage is restored. It is evident that the steady supply voltage must be reduced below the maintaining voltage of the tube for a minimum time, otherwise the tube will refire. This minimum time will evidently be a function of the amount of ionisation in the tubei.e. the value of the discharge current which has been maintained; and possibly the duration of the discharge may be a factor to be taken into account. The minimum time for permanent extinction is 'found to depend markedly on the amplitude of the extinguishing pulse.

In FIG. 2 there is plotted a typical curve relating minimum extinguishing pulse duration T, for a given glow discharge current, with the pulse amplitude, the abscissa being plotted, in arbitrary units, in terms of V the anodecathode voltage obtaining during a rectangular extinguishing pulse. V must obviously be less than V the gap maintaining voltage, so that the curve is asymptotic with respect to V V If V is reduced too much, the minimum pulse duration rises to a maximum at V =0, corresponding to simple removal of the anode-cathode volt age. If V were made negative, the curve to the left of the T axis would be found to be approximately a mirror image of that in the first quadrant. The phenomena causing these results, although well known, merit attention for the understanding of the present invention and will be discussed below. We are, however, now in a position to define, for the purposes of the present invention, the quantity dynamic deionisation time. (Unless otherwise stated, where reference in this specification is made to deionisation time, it is to be understood that dynamic deionisation time as here defined, is implied.) Evidently, for the given initial conditions of glow discharge current in the gap to which FIG. 2 relates, there is an absolute minimum value of T and a corresponding optimum non-zero value of V We therefore define the dynamic deionisation time of a tube as the minimum duration of a rectangular extinguishing pulse of optimum amplitude applied to reduce the voltage maintained across a gap which has been passing a given glow discharge current for a stated minimum time, the gap voltage 11'sing after passage of the pulse to a valuethe reapplied voltage-which is a given amount above the maintaining voltage of the gap. In the present specification, in order to limit ourselves to tubes and circuits which we consider to be of use in our high speed switching applications, we specify, unless it is otherwise explicitly stated, that the deionisation times be in terms of glow discharge currents of not less than 10 milliamperes and of greater than 10 seconds prior duration and that the gap voltage be allowed to rise after the extinguishing pulse and to remain before the gap is next fired at least 50 volts above the gap maintaining voltage.

In order to appreciate the way in which the electrode geometry afiects the deionisation time, attention is first directed to FIG. 3 in which the dotted curve a depicts the variation of potential V across a glowing discharge gap, the abscissa being in terms of r, the distance from the cathode, the positions of the cathode and anode being indicated at K and A respectively. The cathode glow is situated 'at r and between K and r the potential rises sharply and uniformally to a value but little less than the maintaining voltage V Beyond r the potential falls slightly and then rises up to the potential of the anode. For V of the order of volts, the potentials at r and at A may typically difiFer by about 10 volts. Electrons are emitted from the cathode by photon excitation due to the light of the cathode glow at r,; and by positive ion bombardment of the cathode. The emitted electrons are accelerated by the relatively intense field between K and r and at r they have acquired sufficient energy to excite the atoms of the gas-filling. Those electrons which have sufiicient energy to penetrate the space charge of the cathode glow and the subsequent potential trough proceed to the anode or ionise further atoms whose freed electrons join them on this journey. Between r and A there is a plasma region in which equal numbers of electrons and of positive ions are proceeding in opposite directions. The ionisation products, ions and electrons, are continually being removed from the anode-cathode gap by loss to the electrodes themselves, by recombination to form neutral atoms and, in some cases, by lateral diffusion out of the field of the gap. Stability of the glow discharge is established when the rate of removal of ionisation products is just balanced by the rate of formation of new ions and electrons. If, now, the exciting potential be removed from the gap electrodes, the electric field collapses and the ionisation products are left to recombine; unless they be carried there by their existing velocities when the field collapses, or by later random movements, ionisation products are no longer removed by absorption at the gap electrodes. Recombination is a slow process; in fact in a typical gas mixture considerable ionisation may still remain in a discharge gap some seconds after removal of exciting voltage. This accounts for the long deionisation times near the origin of the curve of FIG. 2. If, however, the exciting potential be reduced to only just below V the field strength in the gap is still sufi'icient to provide ion multiplication in the gas by collision processes and by secondary emission from the cathode due to ion bombardment; the rate of removal of ionisation products is only slightly greater than their rate of formation by these secondary processes. Deionisation times are therefore long, as at the right of the curve of FIG. 2. It follows, then, that to achieve short deionisation times, ionisation products should be swept out of the gap by electric fields, the reduction by recombination being largely ignored, but the field strengths should not be too high. Furthermore because of their so much lower mobility, it is the positive ions which have to be considered rather than the electrons.

Summing up the conclusions to be drawn from the above statements, it follows that the criterion for short deionisation times is the speed with which positive ions can be taken to the cathode, without, however, accelerating them to the extent that further activation of the gap is provided by collisions or by the ejection of electrons from the cathode surface due to the ion bombardment. Here choice of the general geometry of the tube can be of assistance. In FIG. 3, in addition to the curve a already discussed, three other curves, B, C, and D have been drawn. They all show the potential variation from cathode to anode in the absence of space charge modification, b-ut curve C, a straight line, relates to the potential between a plane cathode and a plane anode while the other two are for the cases of concentric cylindrical geometry. B is for the case where the outer electrode is positivei.e. the anode surrounds the cathodewhile curve D is for the cathode surrounding the anode. Curves similar to B and C would obtain for spherical geometries. It is seen at once that the field strength in the neighbourhood of the cathode is least in case D, where the cathode surrounds the anode, whereas with the anode surrounding the cathode, curve B, the field strength is greatest at the cathode and one would expect that, in this case, deionisation would be largely oifset by the gain of energy of particles traversing this field. It should be understood that the speed of a charged particle in a gas is given by the product of the field strength and the mobility of the particle: mobility takes into account collision processes and is a function of the gas pressure; the enengy a particle attains is not, therefore, as it would be in free flight, proportional to the potential through which it has fallen, but is, rather, at any time, proportional to the field strength in its immediate neighbourhood.

Experiment confirms that a concentric geometry with central cathode gives long deionisation times. For the concentric geometry with external cathode, however, matters are complicated by the fact that the major part of the anode-cathode potential difference is located between the cathode and the cathode glow, close to the cathode. While space charge persists to any appreciable extent, therefore, the potential distribution both for the external cathode case and that of a planar geometry is nearer to curve a than D or C, respectively. We find, in fact, that with respect to deionisation times there is little to choose between the planar and the outer cathode cases. We do find, however, that, in both cases, the

eionisation time is a linear function of the anode-cathode spacing d and of the reciprocal of the gap maintaining voltage V That this is reasonable can be seen by considering that in practice the optimum voltage across a gap during an extinguishing pulse is not very much below V,,,, and that the mobility of an ion is inversely proportional to the gas pressure p while the breakdown voltage V proportional to the product p d, will be chosen in the light of circuit considerations. Thus pd can be taken as constant and, then, the mobility is seen to be a constant times the gap length. From our empirical relationship between deionisation time, gap length and maintaining voltage, it is a fairly straightforward procedure to deter mine the gap length required for a given V and an acceptable deionisation time and then to choose the pressure, for any given gas mixture, which will give the chosen striking voltage V In practice as d is made smaller and p increased (to keep V constant), the current density, which is inversely proportional to p becomes very high, resulting in con traction of the area of glow and demanding a correspondingly small cathode area. Reduction of the cathode area magnifies edge efiects and, therefore, a compromise has to be made so as to achieve usable values of V discharge current and output voltage obtainable across a load resistor. Factors controlling the gas filling are positive ion mobility and dielectric strength, the latter also governing the tube breakdown voltage. Hydrogen best satisfies requirements except for the fact that its undiluted use results in a maintaining potential of the order of 300 volts, which entails correspondingly higher supply voltages. In order to work with more moderate voltages we use gas mixtures including hydrogen, the proportion of hydrogen being chosen in accordance with user requirements on operating conditions.

Attention can now be directed to the question of formative processes and delay time. In the range of gap lengths and gas pressures with which we are concerned we have found that the generalised Townsend theory of the gas discharge holds and is adequate for the understanding and utilisation of the present invention. As the invention involves the phenomena of discharge in the low current regimes before breakdown into glow discharge occurs, these prebreakdown processes will be discussed in some detail.

in the Townsend theory of discharge between a cathode and an anode in an ionisable gas it is assumed that each electron creates new electrons by collision processes with atoms of the gas during each centimetre of its path in the direction of the field, considered uniform, between anode and cathode. If the gap length be x, there is then obtained for the current i flowing in the external circuit between anode and cathode an expression of the form Here i corresponds to the theoretical case of a gap vanishingly small, so that no gaseous collisions occur. In practice i is the photo-electric current emitted by the cathode as produced in the gas due to its irradiation by light, 'y-radiation and so on. Equation .1 therefore posstulates that no current can flow without some form of irradiation and must cease if the source of radiation is removed. in the present specification we shall assume that such a source is always available. It is also mentioned here that it is common knowledge that breakdown of a gap cannot occur in the complete absence of light or of some ionisation in the gap; where random e ents, such as passage of cosmic rays, are the only source of this initial ionisation, there will be a corresponding random delay in firing the gap, known as the statistical delay. As we are assuming in the present specification that there is always present some source such as a priming discharge evoking the emission of the current i it follows that the statistical delay of firing is in all cases eliminated, other than, perhaps, when a circuit is first switched on, and this delay requires no further mention.

If the electric field in the discharge gap is sufiicient, besides the primary process of ionic multiplication by electron-atom collisions, there are numerous secondary processess, including ionisation by positive ion collisions, the more important process of secondary emission by positive ion bombardment of the cathode, and increased photo-emission from the cathode by photons emitted from excited atoms. Inclusion of these several secondary processes leads to the standard Townsend equation In practice e l, so that Equation 2 indicates an indefinite increase of currenti.e. breakdown into glow discharge-when the secondary multiplication processes have grown to the extent that 'ye" =l. For other than uniform fields similar, but more complicated, expressions for the Townsend equation apply; for our present purposes Equation 2 is adequate.

In FIG. 4 we have drawn graphs representing prebreakdown and glow discharge currents plotted on a logarithmic scale of current I with a linear scale of ordinates V in respect of two gaps of different geometry, both passing the same normal glow discharge current and having the same maintaining and striking potentials. The graph ABCDE is for a cylindrical geometry with external cathode and A'B'D'E is for a planar electrode geometry.

Consider first the planar case. At A, with no applied voltage, there is a very small background current. Over the greater part of the region A'B' the curve is linear, corresponding to the growth of current due to the primary ionisation effects of Equation 1. The slope of the part AB' varies considerably, as is to be expected, with the conditions of illumination of the cathode. Beyond B the secondary processes associated with the second Townsend coefficient 7 become predominant and, when the discharge current is about l.5 l0 amp. the gap breaks down, at a static striking voltage of 400, into glow discharge. The glow discharge region extends from about 0.1 milliarnpere to some tens of milliamperes. If an attempt be made to restrict the glow discharge current and raise the voltage, so as to trace backwards the characteristic curve towards D, it is usually found that the unavoidable interelectrode and Wiring capacitances set up relaxation oscillations and a steady discharge cannot be maintained; the curve in this region has therefore been drawn dotted.

For the cylindrical geometry the characteristic curve is very different. The region of primary ionic multiplication AB is much the same as A'B, but at B, in addition to those secondary processes already mentioned, corona discharge may occur. Reference to curve D of FIG. 3 shows that the field strength near the anode may become very great: it will be governed very largely by the anode diameter. With a fine wire anode, as there was in the tube to which FIG. 4 relates, the field strength becomes so great that, as in the Geiger-Muller counter, passage of radiation through the gas can give rise to relatively large bursts of current. In this intermittent corona regime between B and C the voltage remains practically constant While the current may change from less than 10- amp. to some microamperes. At C the curve rises again and over a range of some 50 volts the current increases to about 0.1 milliarnpere. This is the regime of DC. corona, where the electric field exceeds the dielectric strength of the gas and ionisation is set up by the field excitation of the atoms; the current may be maintained steady up to the point D. At D breakdown to glow discharge occurs, the negative slope portion of the characteristic being shown dotted as in the planar case to indicate that steady discharge cannot normally be maintained over the indicated range. In the present invention we are particularly interested in the region CD of the cylindrical characteristic. As mentioned above, in this region a steady discharge may be maintained and the characteristic corona glow is normally visible around the anode Wire; we have not encountered evidence of relaxation oscillations or other forms of intermittent discharge in this region. It is particularly to be noted that with the cylindrical geometry a pre-breakdown current greater than by a factor of 10- than in the case of a planar geometry may be tolerated; furthermore such a relatively large pre-breakdown current may be obtained by a steady corona discharge.

In the above description of the voltage-current characteristics of FIG. 4, we have tacitly traced the rise of pre-breakdown current with slow increase of applied voltage. In the present invention, however, except, perhaps, when first switching-on the circuit, we are not primarily concerned with this state of affairs. Rather the glow discharge tube has applied to it, in the absence of glow-discharge, a steady potential, greater than the maintaining voltage but less than the static striking potential. Firing of the tube into glow discharge is to be initiated by the application of a pulse momentarily increasing the anode-cathode voltage beyond the static striking voltage. It can be stated quite generally that the time lag between application of the firing pulse and breakdown into glow discharge is shorter the'greater the overvoltage-the excess of applied voltage above the static striking potential.

Remembering that at all times deionisation processes occur simultaneously with formative processes, we can interpret the condition we :1 as that in which the rate of ionisation formative processes in a gap exceeds the rate of loss of ionisation products. In other Words the criterion for breakdown into glow discharge is the onset of an unstable state in which the secondary processes summed up in the coefiicient 7 becomes cumulative. The time which elapses between the application of a firing pulse and the onset of this unstable state is the formative delay time, as strictly defined, of the gap at the overvoltage used. The formative delay time as thus defined is a quantity which is not very easy to measure and, in practice, a somewhat less exact definition is more convenient and will be adopted for purposes of the present specification.

Reference once more to curve a of FIG. 3 reminds us that glow discharge cannot be maintained until the necessary space charge and consequent redistribution of potentials has been set up in the gap. Time must obviously be taken for this space charge build-up even when the breakdown condition e" =1 has has been achieved. Some quite significant space charge build-up and consequent potential redistribution will have been produced by the large pre-breakdown ionisation in applications of the present invention, but the remaining space charge required for stable maintenance of glow discharge must be furnished by current flow from the external circuit; the space charge build-up time is thus influenced by the external wiring and circuit connections, that is by conditions not altogether within the control of the tube manufacturer. Nevertheless, in circuit arrangements such as those described above with reference to FIG. 1, the space charge build-up time has been found to be in the region of milli-microseconds; the error in neglecting this time in terval is therefore small and, in any case, will give an overestimate of the formative delay time if it be included therewith.

Once breakdown has occurred and the space charge within the gap has been built up to an adequate extent, the gap is said to be fired; that is to say glow discharge may now be maintained provided the anode-cathode potential across the gap does not fall below the glow maintaining potential. We therefore define, for purposes of the present specification, the formative delay time of a glow discharge gap as the time interval between the application of a given over-voltage to the gap electrodes and the gap being fired, the word fired being used in the sense above explained. For practical measurement we consider a gap which has been left without glow discharge for an indefinite time, longer than the dynamic deionisation time, with an applied voltage between the maintaining and striking voltages of the gap, and measure the minimum width of a pulse raising the applied voltage to a given overvoltage, glow-discharge being maintained after passage of the pulse.

in the present specification consideration is limited to tubes whose formative delay times, as above defined and measured, are less than 1 sec. for overvoltages less that 60 volts.

There is one further factor, which, though it does not alter our basic definition of formative delay time must nevertheless be considered when any attempt is made to measure it. As will be discussed below, we find that with tubes capable of passing steady corona current, if they are not maintained, in the intervals between firing pulses, in the region CD of FIG. 4, formative delay times are long. In arrangements in accordance with the principles of the present invention, this corona maintenance condition, or equivalent priming arrangements providing a relatively large pre-breakdown ionisation level in the tube, are automatically provided.

What has just been said emphasises our discovery that one of the main factors affecting the formative delay time is the amount of ionisation already present in the gap. Other factors are: (a) the rate of ion formation in the gas by electron and positive ion collision processes; (b) the rate at which photons arrive at the cathode; and (c) the rate of increase of positive ion bombardment of the cathode. The properties of the gas and the gas pressure, both governing the mobility of the ions, is obviously important, but the electrode geometry, determining the field distribution and permissible pre-breakdown current, exercises a predominating influence when comparisons are made between different tubes having similar gas-fillings.

Considering the three geometries corresponding to curves B, C and D of FIG. 3, in the case of the central cathode, curve B, the field distribution in the absence of space charge distortion is the nearest to that eventually obtaining in the glow discharge condition, curve a. The field at the cathode is highest, and hence we should expect factors a and above to be most favoured by this geometry. In the case of a trigger tube, where the space charge of the main gap is built up, so far as possible, at the trigger gap, it is found, as discussed in our aforementioned specification No. 686,317 (Hough-Iackson 122), ease of transfer of glow from trigger gap to main gap is favoured by an internal cathode, external anode, main gap geometry. It might well be assumed therefore, that a similar geometry would yield the lowest formative delay time. This is the case, but only in the absence of appreciable ionisation in the gap before the application of a firing pulse. It will be apparent that, because in the central cathode case the electric field is greatest at the cathode, a relative small number of positive ions in the gap can be very efiective in cathode bombardment and consequent breakdown of the gap. This is bourne out by the curves of FIG. 4 which show how very much less pre-breakdown current, and hence ionisation, occurs in the planar gap than in the case of the external cathode. For an internal cathode the pre-breakdown current would be even smaller.

From our experiments it appears that of the three factors a, b and c mentioned above, if the gap geometry is such that a large pre-breakdown current may exist, i.e. if a concentric geometry with external cathode is adopted, the factor a is the most important in that it is possible to arrange that a large number of the ions which would otherwise have to be formed by collision process can be pre-existant and the time required for their formation eliminated. Obviously the more ionisation products there are available in a gap before application of a firing pulse, the greater will be the rate of formation of fresh ionisation products when the pulse is applied. We are not aware, however, that this aspect of pre-breakdown ionisation level has previously been recognised. The realisation of its importance has enabled us to operate tubes with formative delay times ten times shorter than has hitherto been possible.

An embodiment of a corona primed diode glow discharge tube according to the invention is illustrated, very much enlarged, in FIG. 5 in which a standard subminiature glass envelope 4 encloses a cylindrical cathode 5 which is supported by rods 6 from the glass press 7 and coaxially surrounds an anode rod 8. One of the rods 6 and the anode rod 8 are shown taken through the glass press. The critical parameters of the tube and consequential dimensions which together characterise the construction as an embodiment of the invention are as follows:

Corona breakdown 360 volts max. voltage.

And the essential dimensions:

Cathode internal diameter Cathode length Anode diameter From what has been explained above, it will be understood that, broadly speaking, the gas filling and pressure are chosen to give the required breakdown and maintaining voltages, the anode-cathode spacing determines the deionisation time, the anode diameter, the corona breakdown voltage and the cathode area is chosen to provide the required glow discharge current. The formative delay time is a function of the pre-breakdown current, determined largely by the properties of the gas and the potential gradient at the anode. Since few of the parameters and dimensions are altogether independent of one another, some compromise in design characteristics will normally be involved in making tubes to a specification, as has previously been mentioned.

For the tube whose dimensions and basic characteristics have been quoted, the corona current at 380 volts is about 50 amp. and for glow extinguishing purposes the anode-cathode voltage should be reduced to about volts.

On account of the fact that the anode-cathode voltage of a tube next to fire is held at the extinguishing value until the arrival of the firing pulse, the circuit of FIG. 1 is not, as it stands, suitable for the operation of tubes of the type just described with reference to PEG. 1. To obtain the short formative delay times of the FIG. 5 type of tubes, corona current should flow immediately prior to application of a firing pulse. The arrangement of FIG. 1 could be modified by the substitution of a single train or" rectangular extinguishing pulses, such as the train 3a, applied to all the tubes in common, in place of the complementary pulse trains 3a and 3b, but, even so, the circuit would not provide much margin against variation of supply voltages, component values or tube characteristics. These dilficulties are overcome by the circuit of FIG. 6.

In the arrangement of FIG. 6 the cathode circuits are the same as in FIG. 1 and are identified by the same reference symbols; the split pulse supply system is also similar, in that complementary pulse trains 3a and 3b are fed to the busbars 1a and 1b respectively. Instead, however, of the anodes being connected directly to their respective 1 mm. 2 mm. 0.076 mm. (0.003 inch).

busbars, an isolating diode W' W' W is inserted in series with each anode, so that glow discharge current for tube T must flow from 1a through W' for tube T glow discharge current flows from 1b through W and so on. In addition, each anode is connected through a high resistance, provided by resistors R; to R respectively, to a source of corona maintaining supply labelled 430 v. For use with the tubes of FIG. 5, the amplitude excursion of the pulses 3a and 3b is from 335 volts to 160 volts, the cathode resistors R to R are each of value 13,500 ohms, so as to provide, with a glow discharge current of ma. and 335 volts reapplied voltage, the specified 70 volts over- Voltage for firing. R to R are each 1 megohm, allowing corona current of 50 microamperes to flow through each quiescent tube so that the anodes of these tubes are held at 380 volts, thus normally blocking the anode recti fiers W' to W',,. A value of 20 pf. is satisfactory for the capacitors C to C,,.

The operation of the circuit of FIG. 6 is as follows. Assume that glow discharge is being maintained in T The major part of the normal discharge current of 10 ma. is supplied from the busbar 1b and flows through diode W so that the anode of T is maintained at 335 volts and about 95 microamperes flows through R',,. The voltage across R is the difierence between the anode voltage and the maintaining voltagei.e. 135 volts. All the other tubes, T to T are passing corona discharge, the diodes W to W all being blocked as explained above, and their cathodes are substantially at earth potential. When the next pulse 3 arrives the voltage on busbar 1b drops to 160 volts and the diode W becomes blocked, thus inserting the high impedance of R into the discharge path; glow discharge cannot be maintained but is replaced by corona discharge. (It may be mentioned that the course of the anode voltage and of deionisation processes inside the tube under these circuit conditions has not yet been fully investigated; for minimum deionisation time the anode voltage should fall to 160 volts, but it is doubtful that in fact it falls so far before rising again towards the 430 volt potential through R,,; the deionisation time has, however, been measured and is found to be some 10% longer than in the ideal case.)

The sudden fall of current through R,, charges (I so that the cathode of T falls towards 135 volts with respect to earth potential, W presenting a high impedance discharge path for the capacitor C At the same time the voltage on busbar 1a rises to 335 volts. Due to the high impedance of R the anode of T follows the cathode potential with increase of corona current until the diode W' becomes unblocked and glow discharge can now flow.

In the arrangement of FIG. 6, if the several supply voltages are derived from a common source, with the values of components given a supply voltage variation of more than 20% could be tolerated when dealing with pulses 3 applied at a pulse repetition rate of 1.1 megacycles per second.

It will be appreciated that in the circuit of FIG. 6 the tube anodes are normally each fed from a 'high impedance source, efiectively passing only corona current, and that glow discharge current is fed to each tube from a low impedance source through a gate, the anode crystal diode, which gate is controlled by the firing and extinguishing pulses.

it has been assumed above that the crystal diodes W' to W have infinite backward resistance. In practice, however, it is difiicult to obtain diodes whose backward resistance is sufficiently high not to reduce appreciably the anode voltage for the corona current during the low voltage excursions of the pulses 3a, 312. It will be seen that if the anode diode has a backward resistance comparable with that of the anode resistor, appreciable current will flow through the high impedance anode circuit during these intervals; this increased current through the anode resistors will further drop the anode voltage so that the anode cathode potential may fall below the corona striking potential of 360 volts, with consequent loss of priming and long formative delay times. Although there are now silicon diodes which meet the severe requirements of the diodes in the anode circuits, their supply is, at present, not very plentiful. It has previously been mentioned, however, that corona self-priming is not the only way in which the invention can be realised and that a separate glow discharge, functioning as an artificial corona discharge, may be utilised. Such an arrangement, though involving a somewhat more complicated tube construction than the simple diode of FIG. 5, yet allows the use of simpler circuits; in particular the circuit of FIG. la has been used with ample voltage and component value tolerances when incorporating the tubes now to be described, both tube types having been operated at pulse repetition speeds in excess of one megaclcle per second, the tube with spherical geometry giving the higher speed.

One embodiment of an artificial corona tube according to the invention is illustrated in FIG. 7. Here a cylindrical cathode 9 coaxially surrounds a tubular anode 10 which is formed by a Pair of axially separated tubular members 11 and 12 of which only the ends protrude. into the enclosure of the cathode. A central wire 13 provides an auxiliary cathode and continuous glow discharges are maintained between the wire and the anode members 11 and 12 during operation of the tube.

The cathode 9 is bonded at each end to a respective one of a pair of shallow metal cups 14. Each cup 14 serves to locate a ceramic sleeve 15 coaxially with and at the respective ends of the cathode. The anode members 11 and 12 are each received in a central aperture in the base of a respective deep metal cup 16 which fits, into the open end of the respective sleeve 15.

Each cup 16 has an outwardly extending rim which seats on the end of the corresponding sleeve 15 and the anode members 11 and 12 are bonded to the cups as indicated at 17. The ends of the anode members 11 and 12 protruding into the cathode enclosure are drawn down slightly so as to provide internal seatings for respective ceramic sleeves 18 fitting inside the anode members with their other ends protruding beyond the rims of the cups 16. The auxiliary cathode wire 13 is threaded through the sleeves 18 and an eyelet 19 is passed over and welded to the wire 13 at each end of the electrode assembly, the eyelets bearing against the respective sleeves 18 so as to clamp the assembly together and to keep the Wire 13 taut within the assembly. This electrode assembly is housed within a conventional miniature radio valve envelope 20 having a glass button base 21 carrying leads 22. The assembly is supported on the leads 22 by means of wire connections thereto. The auxiliary cathode wire 13 is secured to one of the leads 22, a wire 23 sheathed in an insulating sleeve 24 joins the cathode 9 to another lead 22, and further wires 25 and 26 join the cups 17, and hence the anode members 11 and 12, respectively, to two further leads 22. The wires 25 and 26 are shrouded by insulating sleeves 27 and 28 respectively. After processing the assembled tube, the envelope is filled with the chosen gas mixture, apertures 29 in the bases of the cups 16 aflording access to the interior of the electrode structure.

The essential dimensions and the gas filling of an embodiment of the invention as illustrated in FIG. 7 are:

Gas filling 20% hydrogen, 1% argon,

79% neon. Gas pressure 200 mm. mercury, at 20 C. Cathode internal diameter 0.116 inch. Cathode length 0.060 inch. Anode outer diameter 0.056 inch. Electrode material High purity nickeh The diameter of the auxiliary cathode wire 13 is not critical and may be anywhere between 0.005 inch and 0.050 inch, the smaller dimension favouring a more stable glow discharge. The cathode outer diameter is, of course, unimportant electrically, 0.215 inch being a convenient size.

FIG. 7, besides illustrating the construction of this embodiment of the invention, also shows diagrammatic circuit connections to the several electrodes of the tube. If it be envisaged that the circuit of FIG. 1 use tubes described with reference to FIG. 7, then the circuit connections shown in FIG. 7 correspond with those for tube T of FIG. 1, the corresponding components and leads being given the same reference numerals in the two figures. The only circuit difference from FIG. 1 in FIG. 7 is the provision of the auxiliary cathode 13 connected through a current limiting resistor R to a terminal which is maintained at l50 volts with respect to the earth line 2. Discussion of the working of this auxiliary discharge circuit, the artificial corona, and the resulting characteristics of the tube of FIG. 7 will be given after description of an alternative spherical electrode embodiment illustrated in FIG. 8.

In the embodiment of FIG. 8 the cathode proper is the hemispherical dome 31 drawn out in the base of a cupshaped member 32 of high purity nickel. The anode 33 and auxiliary cathode 34 are provided by the respective tips of a pair of wires 35 and 36, which are symmetrically positioned underneath the dome 31. These wires are sealed through a glass bead 37 which is fused to one side of a ceramic disc 38, apertured to receive the wires and to locate their ends, which fits inside the member 32. The disc 38 is held in member 32 by means of a metal sleeve 39 bonded to the inner surface of the skirt of member 32. The electrode assembly is mounted by means of the wires 35 and 36 together with wires 40 and 41, both joined to the outer surface of cathode member 32, in a standard sub-miniature radio valve envelope 42, the wires being sealed in the pinch 43. Access to the interior of the electrode assembly is provided by means of an aperture 44 in the dome 31, in conjunction with a split spacing collar 45. As in the case of FIG. 7, circuit connections are shown for connection as tube T in the circuit of FIG. 1, together with the provision of a -l volt connection to the auxiliary cathode from a terminal 30 through current limiting resistor R With the same gas filling and gas pressure as in the case of the embodiment of FIG. 7, the essential dimensions for the tube of FIG. 8, are

Inches Internal diameter of cathode hemisphere 0.040

Diameter of anode and auxiliary cathode wires 0.005 Length of tips of wires 33, 34 available as discharge electrodes 0.015

The overall diameter of the electrode structure is about 0.280 inch and the length of the skirt portion of the cathode member 32 is 0.140 inch. The gap between the tips of the anode and auxiliary cathode wires is 0.005 inch.

For both the tubes described above with reference to FIGS. 7 and 8 the static breakdown voltage for glow discharge is just over 400 volts, the glow discharge maintaining potential being about 200 volts. Both tubes pass a normal glow discharge current of 10 milliamperes.

In operation in the circuit of FIG. 1 the resistors R, should each be 220,000 ohms; the priming current from the auxiliary cathode then has the values 0.5 ma. when the anode is at 160 volts and 1.3 ma. when the anode rises to 335 volts. Talc'ng into account the stray capacitances of the electrodes and wiring, it is found that these values give approximately the same eifects on the formative delay and deionisation times of the main gap as does a constant priming current of 1 ma.

As has been explained in connection with the definitions of formative delay time and deionisation time, the formative delay time is a function of the overvoltage applied to fire a gap and the deionisation time is a function of the voltage re-applied to a gap following an extinguishing pulse. Both the tubes with cylindrical and spherical geometries described above with reference to FIGS. 7 and 8 respectively and having the dimensions quoted have very similar deionisation characteristics, and curve A of FIG. 9 is a graph of the relationship between re-applied voltage (the value being indicated to the right of the graph) and deionisation time for either tube. The relation between overvoltage, values being given on the left of the graph, and formative delay time, is given by the curve B in the case of the cylindrical geometry of FIG. 7 and by curve C for the spherical geometry of FIG. 8. All three curves A, B and C relate to the condition in which a priming, or artificial corona, current of one milliampere is flowing across the gap-or gaps, in the case of FIG. 7between anode and auxiliary cathode. It will be seen that, for a given overvoltage, the spherical geometry gives much shorter formative delay times than does the cylindrical arrangement. Still shorter formative delay times, and deionisation times, can be obtained by the use of a pure hydrogen filling for the tube, at the expense, as has been mentioned previously, of higher supply voltages. The curve D of FIG. 9 shows the variation of formative delay time with overvoltage for a tube having the same dimensions as described above with reference to FIG. 8, but filled with hydrogen, a priming current of one milliampere, as for the other curves, passing through the auxiliary cathode-anode gap.

By the use of the artificial corona priming, large prebreakdow-n ionisation can be tolerated and short deionisation times obtained in these tubes because of the geometrical and scaling factors discussed previously. Formative delay and deionisation times are then reduced to a fraction of those obtaining in conventional types of trigger tube and ample margins are available for circuit and voltage tolerances.

summarising briefly the foregoing description, it can be stated that, arising from our investigations into prebreakdown ionisation levels and discharge currents in glow discharge tubes, the invention provides a diode form of gas tube with hitherto unrealised short deionisation and formative delay times, a high speed operating circuit for use with these tubes, but also of advantage with some forms of known corona voltage stabiliser tubes when used as glow discharge tubes, and, avoiding some of the difficulties associated with this new circuit arrangement, a modified form of what is, in its essentials, a diode relay tube primed 'by an auxiliary discharge so as to have characteristics similar to and even improved upon the corona self-primed diode of the invention.

While the principles of the invention have been described above in connection with specific embodiments, and particular modifications thereof, it is to be clearly understood that this description is made only by way of example and not as a limitation on the scope of the invention.

What we claim is:

1. A gas discharge tube comprising a cup-shaped main cathode member formed with a dome projecting from the middle of the base of the cathode member, the interior of the dome providing the cathode discharge surface, a ceramic disc held within the cup-shaped cathode member, an anode member and an auxiliary cathode member, said members being coaxial with said cathode and projecting through respective apertures in said ceramic disc,

the ends of said members projecting within the dome.

2. A gas discharge tube according to claim 1 in which the said anode member and cathode members are sealed through a glass button secured to the said ceramic disc on the side remote from the cathode dome.

3. A gas discharge tube according to claim 1 in which the tube is filled with a mixture of 20% hydrogen, 1% argon and 79% neon.

4. A gas discharge tube comprising a hollow cylindrical cathode, a tubular anode member smaller in diameter than said cathode and coaxial therewith, one end of Said anode member providing the discharge surface, said end projecting into one end of said cathode, and an auxiliary cathode located coaxially of and extending through said anode member and said cathode.

5. A gas discharge tube as in claim 4, wherein "a second tubular anode member having a diameter substantially equal to the diameter of said first mentioned anode member and spaced therefrom with one end of said second anode member projecting into the other end of said cathode and being coaxial with said cathode and electrically connected to said first mentioned anode member.

6. A discharge tube according to claim 4 in which a pair of ceramic sleeves, coaxial with the anode member, abut against the respective ends of the cathode, a pair of metal cups are received one in each of the said sleeves, and the said anode members are each secured in a central aperture in the base of a respective one of the said cups.

7. A discharge tube according to claim 6 in which a further pair of ceramic sleeves is held one in each of 18 the tubular anode members, the said auxiliary cathode is threaded through these further sleeves, and a pair of eyelets bearing against the ends of the said further sleeves are secured to the auxiliary cathode to maintain it in tension and to clamp together the assembly of electrodes and ceramic sleeves.

References Cited in the file of this patent UNITED STATES PATENTS 1,722,588 Metcalf July 30, 1929 2,098,301 Mendenhall Nov. 9, 1937 2,331,398 Ingram Oct. 12, 1943 2,433,755 Haine et a1 Dec. 30, 1947 2,457,891 Henninger et a1 J an. 4, 1949 2,471,263 Depew May 24, 1949 2,492,295 Knochel Dec. 27, 1949 2,646,534 Manley July 21, 1953 2,724,789 Overbeck Nov. 22, 1955

Images