US2424002A - High-frequency electronic tube - Google Patents

Description

15, 1947- I D. H. SLOAN 2,424,002

- HIGH FREQUENCY ELECTRONIC TUBE- Filed Nov. 4, 1940 l0 Sheets-Sheet 1 (Nmvmlz, DA wo H. SLOAN.

ATTORNEYS.

July 15, 1947.

D. H. SLOAN HIGH FREQUENCY ELECTRONIC TUBE Filed Nov. 4, 1940 10 Sheets-Sheet 2 INVENTOR, DAV/D H. SLOA/V ATTORNEXS.

July 15,1947,

D. H. SLOAN HIGH FREQUENCY ELECTRONIC TUBE 10 Sheets-Sheet 3 Filed Nov. 4, 1940 INVENTOR, DAV/D H. SLOAN.

Mv/W

ATTORNEYS.

July 15,1947. D. HVSLOAN 2,424,002

HIGH FREQUENCY ELECTRONIC TUBE Filed Nov. 4, 1940 10 Sheets-Sheet 4 I n5 INVENTOR,

//7 DAV/0 H. SLOAN.

' M 2 H3 9 I ATTORNEY;

July 15, 1947. D. H. S LOAN HIGH FREQUENCY ELECTRQNIC TUBE Filed Nov. 4, 1940 10 Sheets-Sheet 5 INVENTOR, 0A v/0 H. SLOAN.

ATTORNEYS.

July 15, 1947. SLOAN f 7 2,424,002

HIGH FREQUENCY ELECTRONIC TUBE Filed Nov. 4, 1940 10 Sheets-Sheet 6 INVENTOR, 0A V/D H. 51. 0AM

A 7" TORNEVS.

July 15, 1947. D, H SLOA 2,424,002

HIGH FREQUENCY ELECTRONIC TUBE Filed Nov. 4, 1940 10 Sheets- Sheet 7 INVENTOR, DAV/D H. SLOAN.

ATTOR/VEVS.

- July 15,1947. D. H. SLOAN HIGH FREQUENCY ELECTRONIC TUBE Filed Nov. 4, 1940 10 Sheets-Sheet 8 uvvswron, DAV/0 H. SLOAN. g M M ATTORNEYS.

l0 Sheets-Sheet 9 Filed Nov. 4, 1940 2 Q a v i \h DAV/D H. SLOAN. 4 ,M- JW ATTORNEYS.

R m a M a m w 1 @N M a n 1 w a M 6 7 5 w/ m 3 8 6 3 Q i. wmfl a m w July 15,1947. D. H. SLOAN HIGH FREQUENCY ELECTRONIC TUBE Filed Nov. 4, 1940 10 Sheets-Sheet 10 INVENTOR DAV/D H. SLOA/V.

BY Q

A TTORNE VS Patented July 15, 1947 HIGH-FREQUENCY ELECTRONIC TUBE David H. Sloan, Berkeley, Calif., assignor to Research Corporation, New York, N. Y., a corporation of New York Application November 4, 1940, Serial No. 364,284

21 Claims.

1 This invention relates to electronic tubes, and particularly to tubes adapted for the production and modulation of ultra-high frequency oscillations, i. e., oscillations of frequencies of the orderof 1,000 megacycles.

The progress of electronic and radio development since the inception of the art has been marked by two steady advances. One of these advances has been toward higher power, the other toward higher frequencies. The latter line of ad- Vancement has been, to a certain extent at least, incompatible with the first, since with increasing frequency the effect of interelectrode capacity has become greater and more troublesome. Nevertheless, up to the last few years, the difficulties have been met by a steady evolutionary process consisting in large degree of refinement in detail, which has enabled the vacuum tube art to keep pace with the increasingly rigid demands of the manufacturers and operators of transmitting and receiving apparatus.

The attempt within recent years to carry the useful spectrum into the range of wavelengths in the range of a meter and less has involved difficulties of a new order of magnitude. For one thing, the frequencies involved are so high that the transit time of an electron stream across the interelectrode spaces of the tubes becomes an appreciable fraction of a cycle. For another, even with connecting leads reduced to minimum lengths, their inductance has been sufiicient so that the capacities required in tuning them to the desired frequencies are small in comparison with the interelectrode capacities in conventional tube structures and these capacities have therefore become not merely a nuisance, limiting the efficiency of operation, but frequently an absolute bar to such operation; so much so, in fact, that it has been only with tubes of very small size and consequent small power output that operation has been obtainable at all.

There therefore exists at the present time a need for a tube which will meet the severe requirements of producing large power outputs by generation or amplification at extremely high frequencies. These requirements are first, a cathode-grid structure which will effectively modulate an electron stream without the application of excessive control voltages; second, a cathodegrid structure whose capacity and inductance relationships are so proportioned that they may be tuned to the high operating frequencies desired; third, a structure lending itself to circuits of low relative radio-frequency,resistance of high impedance, so that excessive energy will not be required to swing them through the necessary range of control voltages; fourth, fixed relationship between the various elements, irrespective of temperature or ordinary shock, so that the frequency to which the device as a whole is tuned will not be affected by relative changes of position; fifth, minimum undesired or incidental radiation from the various elements of the tube and its auxiliaries; sixth, a minimum of insulating material subjected to high-frequency fields. To these may be added the secondary requirements of demountability for replacement of filaments, facility in water cooling and avoidance of hot spots, and ease of tuning.

From the conventional approach these requirements are incompatible to a large degree. A high degree of control requires close spacing of cathode and grid, which leads to high interelectrode capacity. Rigid structure ordinarily means massive structure, which again leads to high interelectrode capacity. Water cooling systems tend to form effective antennae, leading to large stray power radiation. The broad purpose of my invention is therefore to reconcile these and other apparent incompatibles.

Pursuant to this general purpose, among the objects of this invention are: To provide a tube which is capable of producing many kilowatts of power at extremely high frequencies; to produce a high frequency generator of great frequency stability; to produce a high frequency amplifier and oscillator tube of relatively high efficiency, and particularly to produce such a tube wherein the losses due to undesired radiation from the tube itself are reduced to negligible proportions; to produce a high frequency oscillator and amplifier which may be tuned to operate at any desired frequency throughout a reasonably wide range; to provide a high frequency oscillator and amplifier which may be constructed with the high degree of accuracy required to meet the close tolerances demanded by the frequency of operation and to maintain those tolerances under the changes of temperature produced by such operation; to provide an electronic tube of the character described which may be fully fluid cooled and wherein the cooling system does not introduce material parasitic radiation of radio-frequency power; to provide a high power oscillator and amplifier .tube which is readily demountable for replacement of filaments; to provide means of density-modulating an electron stream at ultra-high frequencies, in order to produce extremely short bursts of electron emission occurring at the peaks of the oscillation and of substantially uniform velocity, whereby the conversion of energy into high frequency power occurs at high efficiency; to provide a type of structure for high frequency electronic tubes which is of great flexibility, and which will, because of such flexibility, permit the construction of tubes exactly adapted to a large variety of powers and services; to provide a novel and effective method of tuning apparatus of the character described;

and to providea type of electrode support for high frequency electronic devices which is massive and rugged, and which, at the same time, does not introduce interelectrode capacities which either severely limit the frequencies upon which the device is operative or the power which may be developed at such frequencies.

My invention possesses numerous other objects and features of advantage, some of which, together with the foregoing, will be set forth in the following description of specific apparatus embodying and utilizing my novel method. It is therefore to be understood that my method is applicable to other apparatus, and that I do not limit myself, in any. way, to the apparatus of the present application, as I may adopt various other apparatus embodiments, utilizing the method, within the scope of the appended claims.

The tube of my invention involves two basic concepts. The first of these comprises forming electrode supports of sturdy coaxial metallic cylinders which constitute a radio-frequency transmission line of at least one and preferably a plurality of quarter wave-length electrical links, with impedance irregularities at or near certain of the quarter wave points,-the electrodes themselves forming a portion of these transmission lines as considered electrically. Means are preferably provided for varying the position of the impedance irregularities to provide exact tuning, but this is not essential since, as will hereafter be shown, by a proper combination of the characteristic impedances of the quarter-wave sections'and their terminating impedances, it is possible to make the radio-frequency impedance-of the supports as viewed from the electrodes themselves extremely high, so that the overall efiect'i's almost as though the electrode capacities together with the inductances required to resonate them were supported freely in the space within an unbroken metallic shield.- Thislatter feature is secured by providing multiple coaxial line sections forming branch paths of greatly different impedance, certain of these paths acting as by-passes of negligible impedance at. points where it is necessary that some circulating currents should flow although D.-C. insulation must be maintained while at the'same time maintaining the high impedance desired in other paths which would otherwise lead to radiation By placing these. bypass sections at current nodes, the FR losses therein may be made too small to need consideration.

The second fundamental concept comprises mounting on the ends'of such supports, preferably in biaxially symmetrical configuration, one or a plurality of cathode-grid combinations which act, as before stated, as the termini of the transmission lines formed by the supports; mounting the grid opposed to an anode or other acceleratin the immediate neighborhood of the grid; and

mounting the cathode in the region of sharpest curvature. One of the best ways of obtaining such a structure is to form the grid of pairs of cylindrical surfaces whose axes are parallel to the plane of the anode, and to make the cathode as a filament or strip having a flat or slightly concave face lying between the cylindrical surfaces of the grid. This results in the lines of force from acceleratoror anode normally terminating in the grid structure,'none of them reaching the cathode, from which emission is therefore normally suppressed. A few volts relative change in potential of the cathode, as referred to either accelerator or control grid, results in some of the lines of force from the accelerator terminating in the cathode surface, which is accordingly subjected to an extremely powerful field causin very large emission to the anode. The result is What may be termed an explosive type of emission, giving electron bursts of high density for very short periods at the peaks of the cycles. It will be evident that this structure results in a relatively high capacity as between cathode and grid, but the tuned transmission line support enables this capacity to be effectively resonated with an inductance as small as may be desired and still form a sharply tuned, high Q circuit whose high resonant input impedance may appear as resistive, capacitive or inductive as the conditions of operation may require.

. Referring to the drawings:

Fig. 1 is a longitudinal section through a highfrequency oscillator tube embodying my invention, the particular tube illustrated employing a radial arrangement of filaments and grids.

Fig. 2 is a transverse section of the tube of Fig. 1,' showing the multiple coaxial grid-filament supports, and Water connections for COOllIlg the filament mounting, the plane of section being on the line 2-2 of Fig. 1.

Fig. 3 is a transverse section through the anode structure of the tube, the plane of section being 7 indicated by the line 33 of Fig. 1.

Fig. 4 is an enlarged detailed View illustrating water-cooling connections from the exterior of the tube to the filament mounting.

Fig. 5 is a schematic sectional View through filaments, control grid, accelerating grid, boundary grid and anode of the tube roughly indicating the field effective at cutoff.

Fig. 6 is a sectional view through the grid support line, showing the radio frequency by-pass between accelerating and boundary grids and the tuning mechanism for isolating the control grid and water cooling the same.

Fig. 7 is an enlarged detail showing the method of insulating certain of the supporting rings upon which the coaxial electrode elements are carried.

Fig. 8 is' a section taken at right angles to the View of Fig. 1, and showing the anode-supporting, cooling, and tuning system;

Fig. 9 is a perspective view of the accelerator grid. 7

Fig. 10 is an elevation of the control-grid structure.

Fig. 11 is a sectional View, taken on the plane between the filament and grid structures, and showing in detail the filament support.

Fig. 12 is a fragmentary axial section taken on the line |2--|2 of Fig. 11.

Fig. 13 is an elevation of the active face of the anode.

Fig. 14 is a fragmentary section of the anode, the plane of section being indicated by the line i4|4 in the preceding figure.

Fig. 15 is an elevation of the boundary grid.

Fig. 16 is a sectional view taken on the line Iii-J6 of Fig; 15, and showing a portion of the anode'in elevation,

Fig; 17 is an elevation of one of thefilaments.

Fig.,18 shows a modified form of coaxial line structure for grid-filament supportin atube generally similar to Fig. l, but adapted for use either as an amplifier oran oscillator with inductive feed-back.

Figs. 19, 20 and 21 are-sectional views through the tube of Fig. 18, taken on the lines numbered in accordance with the figures.

Fig. 22 isalongitudinal section through a tube built in accordance with this invention but-where in a cylindrical, rather than a radial filamentgrid and. anode arrangement is used.

Figs. 23, 24 and 25 are transverse sections through the tube of Fig, 22, taken on the lines indicated by the respective numerals.

Figs, 26 and 27 are detailed views indicating the tuning'mechanism for the tube of Fig. 22.

Fig. 28 is a longitudinal section on a larger scale through the filament support of the tube of Fig; 22.

Fig. ZQ'iS a transverse section through the supporting columns of the tube of Fig. 22, showing the construction of the centering mechanism, the plane of section being indicated by the line 29-29 of Fig. 28.

Fig. 30 is a sectional view taken on the line 3ll30 of Fig. 22.

Fig. 31 is a fragmentary section. taken on the line 3l3l of the preceding figure, showing the passage of the cooling pipe-past the anode and between the two sections of the filament sup" port.

Fig. 32 is an impedance diagram for an openended, half wavelength section of transmission line.

In the ensuing specification the invention will first be described in its various aspects as applied to an oscillator tube of moderatepower (i. e., approximately 10 kw. peak output at 20 to 40 centimeters wavelength). Following this the-re will be described two modifications illustrating respectively the application of the principles of my invention to a similar tube adapted for amplification or for generation of oscillations by inductive feed-back, and to a somewhat higher output device showing the principles asapplied to a tube constructed with cylindrical rather than radial arrangement of electrodes.

The tube shown in longitudinal section in Fig. 1 is of the demountable, constantly pumped type, as, in fact, are all of those herein described although the principles involved are not limited for use in such tubes. From the structural point of View the tube comprises a series of flanges connected by sections of tubing and held together in compression. From a practical point of view it is advantageous to have the flanges pierced for and held by circumferential bolts to hold the parts firmly in position when the tube is not under vacuum, but when in use the external air pressure tends to hold the entire device together, and the tube has actually been operated without the retaining bolts, these have therefore been omitted in the drawings since they add a further complexity of detail to an already comfits an internal recess or counterbore in an annular grid flange 3, clamping a, boundary grid 4 between the housing tube 2 and the flange 3. A rabbet on the outer periphery of the flange 3 receives one end of a main support cylinder 5, whose other end terminates in another annular flange 1. All of the parts thus far mentioned are of metal, andI have found it convenient to make the flanges of steel, and the tube 5 also ofseamless steel tubing, while the cylinder 2 may be either chromium or copper plated steel or solid copper, with copper preferred, since it forms a portion of a resonating circuit. Carrying on from the flange l is a glass or Pyrex cylinder 9 which abuts a terminal flange ill.

As has been mentioned already, the device as a Whole is fully demountable. The endsof the tubes contact the flanges with smooth machine fits. The joints thus formed are sealed by applying thereto ordinary wide elastic bands as indicated by the reference characters H, these bands being smeared before application with a small amount of vacuum line stop-cock grease.

It may be pointed out at this time that all of the structure thus far described with the exception of the terminal flange II] is at D.-C. ground potential, and as will later be shown in detail that the entire exterior structure is substantially at radio-frequency ground. This means that the insulating section formed by the cylinder 3 is not subjected to R.-F. fields. It also renders easy the support of the device by any desired external means. Part of such support may be the connection to the pump, which is by a pipe l2 of relatively large interior diameter, welded or otherwise secured into the bottom of the anode housing 2. This pipe is not shown in Fig. l, but is clearly visible at the bottom of Fig. 8.

The various elements which contribute to the electronic'action of the tube are mounted within the envelope thus formed on columnar supports each of which has transmission line characteristics designed to meet its particular function. These elements are shown in schematic arrangement in Fig. 5, and comprise an anode 13, a boundary grid t, an accelerating grid IS, a control grid Hi, and a filamentary cathode ll.

Fig. 5 is drawn to a greatly enlarged scale and shows a fragmentary section of the elements cooperating with a single filamentary cathode. In the tube here shown six such cathodes are used and the portions shown of the other elements are repeated for each cathode. One advantage of the type of structure here shown is that the ability of the tube to supply power output varies almost directly as the number of filaments used, and that the changes required to add additional filaments are relatively minor. Tubes have been designed conforming substantially to the structure here shown. with as high as twenty-four filaments, each with its attendant grid-anode structure, but since each of these assemblies is merely the duplicate of the others as far as performance is concerned it is sufiicient for the present to consider one only.

Considering, therefore, the portion of the elements shown in Fig. 5, the anode 13, preferably made of high conductivity oxygen-free copper, is operated at the maximum potential of the system, say 10 to 50 thousand volts positive. It is provided with a V-shaped groove 2% with its axis parallel to the axis of the filament.

d, which is also preferably made of oxygen-free copper. This is provided with an aperture surf- Next, pro-- ceeding toward the filament, is the boundary grid rounded by acollar 2| in accurate alinement with the groove 28 in the anode. Next in line is the accelerator grid l5, with an aperture 22 which is somewhat narrower than the opening in the boundary grid, and Which is operated at a potential above the cathode of from 5 to 20 thousand volts. All potentials mentioned are illustrative and relative only, since the actual valuesnsed will depend upon the size, power output and operating frequency of the device. Furthermore, modifications in design are possible whereby the functions of certain of the grids are combined, other electrodes are operated at ground potential, etc. Such modifications will be considered later; the purpose here is to show the application of the principles of my invention to the present tube.

The most important portion of the combination is the arrangement of the cathode-grid structure. The important features here are first, that the control-grid elements comprise parallel cylindrical surfaces, curved as they are presente to the filament. In the present case they are rods or wires, but they could be cylindrical surfaces formed as the edges of a slot in a flat plate without affecting their performance. Between these surfaces, and slightly back of the plane of their centers of curvature, lies the filament, which has a flat or preferably a slightly hollow ground face presented to the anode. It is convenient to perate the filament at ground potential (disregarding for the moment the slight voltage drop along the filament) and, for the powers here considered, to operate the grid IE at 200 to 500 volts negative.

It will be seen that at the orders of voltages given th major fields are from the accelerator grid l5 to the control grid. As is well known, the lines of force constituting such a field terminate at right angles to the surfaces of the field-defining electrodes. It follows that in the region adjacent the cathode the lines of force emerge from the grid wires in the general direction of the cathode and then curve very sharply toward the anode in a direction nearly at right angles to their direction of emergence. There is also a fairly strong field between the control grid and the oathode itself, which is superimposed locally upon the field between the control grid and accelerator grid, and is directed toward, instead of away from the cathode. As a result of the interaction of these two fields none of'the lines of force from the accelerator-grid normally terminate upon the surface of the cathode. Emission has therefore no tendency to leave the latter, since the space adjacent it is nearly neutral, with such weak fieldas exists therein directed toward the cathode.

As is the case with any grid-controlled tube operated through cut-01f, when th grid swings positive some of the lines of force from the accelerator-grid which formerly terminated'on the control grid now terminate on the cathode, and as the cycle progresses the cathode-control grid field weakens or even reverses, permitting emission toward the anode, and a space charge builds up in the region immediately in front of the cathode face which has the usual effect of limiting emis-. sion. The distinguishing feature here is that the region where the field is weak enough to permit such space charge effect is very shallow, so that even'withthe low velocities imparted to them by such relatively weak field the electrons can and do traverse it in a reasonably small fraction of a cycle.

The biasing potential between cathode and grid is so adjusted that emission can occur only for an instant at the cycle peaks, and cut-ofi may occur even before the first electrons emitted have traversed the space charge region. Furthermore, while in this region there is a maximum difference of velocity as between electrons, both by reason of differing initial velocities of emission and, more importannby reason of differing acceleration due both to phase of emission and field strength at various parts of the cathode surface.

The important point is that because the region is so shallow all of the electrons emitted do get through it before the cycle has advanced too far, and, having traversed it, fall into the region of high potential gradient where acceleration toward the anode is very great; space charge effect is of no further moment, and they receive so large a portion of their final velocity that their differences of velocity in the initial region are immaterial.

It should be realized that while space-charge effect prevents some emission and decreases the acceleration of electrons emitted, it will not drive those which have been emitted back to the oathode nor prevent their reaching the anode. It follows that the space-charge region may be considered as a reservoir for emitted electrons. With conventional grid-cathode structures it is relatively deep, so that, at the frequencies and powers at which this tube is intended to operate, transit therethrough occupies a major portion of the cycle, and with the varying velocities obtaining while in this region the electrons straggle through to reach the anode in such varying phases that the density modulation of the stream is almost if not entirely lost.

With the arrangement of my invention, however, the space-charge region is so shallow that even the stragglers among the emitted electrons traverse it in less than a quarter cycle and instead of the density modulation of the electrons being lost they reach the anode in bursts of such power and suddenness and with such close velocity grouping that I have termed cathode-grid combinations of this type explosive. The object of the design is to make the electron reservoir constituted by the space-charge region as shallow as possible, and in practice the ideal can be so far realized as to permit density modulation of electron streams at frequencies in the range of 1,500 megacycles, Where in the past it ha been necessary to use velocity modulation, involving larger and more complicated structures, to get reasonably effective results, even in smaller sizes and at much lower powers than those here contemplated.

When the tube here-shown is used as an oscillator in the manner now to be described, the various potentials are so arranged and proportioned that the transittime of the burst of electrons is substantially one-half cycle. The anode i 3 is in a timed circuit, as is also the grid Hi. The condition of oscillation then is that the potentials of the anode and the grid swing in the for this arrangement is to spread the area of impact and so increase the size and decrease the intensity of maximum local heating, while increasing the cross-sectional area of thermal conductivity by which cooling occurs, and also to insure that secondary electrons are not projected into regions of high field intensity which would laccelerate them so that they, in turn, would cause serious'heating effects.

It has already been stated that the transit time of the electron burst is one-half cycle of oscillation, and it follows that immediately after the electron burst has occurred the anode has started to swing negative. The electrons accordingly reach their maximum velocity-at or about the plane of the accelerator grid-ideally, just as they pass the eifectiveplane of the boundary grid 4. As the anode continues to swing negative they encounter a decelerating "field, eitherin'an absolute sense or, if still being accelerated by the D.-C. field, from the anode at least in comparison to the acceleration of the D.-C. field alone. In passing through this decelerating field the electrons are delivering energy to the anode circuit, and they are traveling at minimum relative velocity when they enter the slot 20. This slot acts in some degree as a Faraday space, and the electrons suffer little change in velocity or energy as they pass through it. Therefore their work is done and their transit time effectively over as soon as they have entered it.

Sln'cethe acceleration of the entire burst of electrons takes place with substantial uniformity they retain their close grouping at the time of impact, and since the impact occurs when the electrons constituting the burst have suffered maximum deceleration there is a minimum of energy wasted as heat and the oscillator consequently operates atrelatively high efliciency.

For operation int-he manner described the desiderata are that the control grid l6 and filament I! should be effectively isolated from each other both as regards D.-C. and radio-frequency potentials, and should have an effective capacity sufficiently low so that it may be tuned to the desired operating frequency, or, in other terms, it must be capable of being connected in circuit with an inductance sufliciently small to tune to that frequency. The accelerator grid l must be insulated from the other elements to maintain its D;-,C. voltage, but should be effectively grounded as regards radio-frequency potentials. The

boundary grid 4 should also be grounded to radiofrequencyan'd for convenience in operation and safetys sake should preferably also be grounded as regards D.'-C. potential, as it is electrically continuous, with the envelope. The anode should be free as regards both A.C.'and D.-C. potentials. The various mounting and auxiliary means next to be described are designed to meet the desiderata as fully as possible. In this description terms such as above or below are used to indicate position as shown in Fig; 1. They have no other significanca'as the device may be operated in any position;

, Starting at the bottom of Fig.1, with the flange Ill, '9. high conductivity column or pipe 25 extends a major portion of the length of the entire device to the'plane of the flange 3 and the boundary grid 4. This column is brazed or otherwise permanently secured into the flange In so as to be accurately concentric with the remainder of the tube structure and, of course, to be vacuum tight. At its upper'end it is threaded to receive a gridsup-portri'ng {21, which'is clamped between locking nuts 29 and 30, and an additional locking :10 screw 3| (Fig. 10) is also provided for further security. The pairs of parallel grid wires l6 project from the ring 2'! parallel to its radii, six pairs of grid wires being provided in the present design, the pairs being equidistantly spaced around the periphery of the ring.

Two sliders are mounted on the column 25. The upper slider 28 comprises a short section of tubing 32 surfaced to a sliding fit on the column 25 and shouldered at each end to receive discs '33 and 34 between which a short section of tubing 35 is clamped. The column 25 is provided in this region with a, longitudinal slot for the passage of a screw 31which engages a piece of tubing 39' sliding within the column. The tubing 39 terminates in an annular block 40, and an adjusting rod 4| is threaded into one side of the block and passes to the exterior of the tube through a gland box 42 and a Wilson seal 43.

It is apparent that the position of the slider may small for the rod 50 which it is desired to seal.

The seal is lubricated with a small quantity of Vacuum stop-cock grease. Such a seal is vacuum tight under conditions where other known types of packing would leak badly, ince the differential pressure on the washer serves to make it hug the central rod more tightly and it remains tight whether the rod be subjected to rotary or sliding motion in either direction.

Returning to the general tube structure, the second and lower slider 5! is essentially similar in construction to that just described, except that it actuating'rod 52 is mounted externally ofthe column 25 through the gland box 42 and Wilson seal 53.

The slider 5| makes a close sliding fit within a cylindrical conductor 54 mounted in the flange l0 accurately'concentric with the column '25 and maintained in this concentric relation both by the slider 5| itself and by an auxiliary-diaphragm or spacer 55. The tubing 54 does not extend the full length of the central column, but terminates a distance above the flange i which is somewhere in the neighborhood of one-eighth of a wavelength at the mean frequency for which the'tub'e is designed.

Accurately coaxial with the column 25 and its surrounding conductor 5! is a third conducting cylinder '51, mounted on the flange 'l and extending below it for approximately one-eighth Wavelength, [so that the two conductors 54and 51 overlap by. a distance approximately equal to 'one quarter wavelength of the average operating frequency of the'device, wavelengths in this sense being-used to mean the wavelength of the frequency transmitted along the two tubes as a coaxial transmission line. There is no metallic contact between the two conductors 54 and'fil, and they are separated by vacuum so that dielectric loss does not occur in the space between them.

Column 51 is brazed or otherwise secured in the flange 1, is made of highly conducting material {preferably oxygen-free copper) and is preferably provided witha cooling system comprising a water pipe 60 coiled around and soldered 11 to the external surface of the column. Theends of this .pipe are brought out through the flange 1 at the right of Fig. 1.

The upper end of the column 51 carries an intermediate ring El which supports indirectly one end of each of the filaments IT. The other ends of these filaments are carried by a group (here six) of pipes 62 mounted in the annular inter-space between the column 51 and the outer shell 5. The lower ends of the pipes 62 are mounted in a ring 63 which is bolted to and insulated from the flange as is shown in Figs. 2 and 7. V The ring '63 is counterbored at three equidistant point to receive insulating beads 64 of porcelain, lava, or other refractory insulating material which beads space the rings slightly from the flange '1. A cap screw 65 passes in turn through a clamping cap 61, a second bead 69, the ring 63 and the bead 64 to clamp the ring firmly to the flange. It should be noted that the potentials which this arrangement must withstand are of lowfrequency and are only those across the filament, i. e., the insulation need only be of sufiicient value to withstand a few volts (three at 60 cycles in the instant structure) and the insulating material is not subject to dielectric heating from radio-frequency fields.

The actual filament mounting can best be seen in Figs. 11 and 12. Each of the tubes 62 carries an inwardly projecting L-shaped lug 10, and the inner ends of the lug are provided with slots H for receiving the downturned ends of the staple-shaped filaments H, th ends being clamped into place by set-screws 12. The inner ends of the filaments are clamped in an annular groove 13, formed in an inner mounting ring 14 which is supported on column 51 by the intermediate ring 6| before mentioned. The actual clamping of the filament ends is accomplished by pairs of set-screws 15 bearing on small blocks TI.

The pipes 62 are surrounded by open-ended cylindrical conductors 19, which terminate at the level of the upper end of the lug and extend down over the pipe 62 for approximately one-quarter wavelength and are supported by the ring 6|.

- Within the conductors 19 are inner tubular conductors 80 of substantially the same length, open at their upper ends and mounted by their lower ends on the pipes 62 by means of conductive blocks 8|. The concentric tubes 19 and 80 are both'open at the filament end, being notched to clearthe lugs 10 and also being provided with alined holes to permit tightening of the setscrews 12. It will thus be seen that the only connection between the innercolumn 51 with its supporting rings BI and 14 and the group of filament support tubes 62 is the filaments themselves.

These are shown in Fig. 17, and as will be seen are relatively short and rigid. They are preferably of pure tungsten and have a considerable degree of strength. It will further be seen that the support afforded their outer ends by the tubes 62 and lugs 10 is light and of small inertia and that the tubes 62 have relatively large resiliency, The filaments therefore are very unlikely to be ruptured by shock on the device as a whole, and there is ample flexibility to take up their expansion.

Each filament is preferably formed of round tungsten wire, one surface of which is ground flat or slightly concave. The diameter here used is 50 mils. The grinding is preferably performed in a jig which deforms the Wire slightly in the longitudinal direction, so that the ends" of the,fila=- ment are ground a few thousandthsof an inch thinner thanjs the central portion. This grinding forms the flat emitting surface of the filament, and if done with a relatively small wheel whose axis is maintained parallel to the length of the filament, it gives the slight hollow grinding which has already been stated to be advantageous. Theeffect of thinning the two ends, adja cent the point where the filament is clamped, is to give a greater current density at these points, with a consequent greater-liberation of heat which compensates for the heat conduction to the clamping means andrcsults in substantially constant temperature; and substantially constant emission over the entire effective length of the filament. Being of pure tungsten, the filament retains a degree of resiliency even at its emitting temperature, and this, together with its resilient support, prevents buckling or change of plane of the emitting surface when in operation and keeps the electrical constants of the device fixed under such minor variations in operating temperature and expansion, and inequality in these factors as between the several filaments, as inevitably occur in practice.

Cooling for the support of the inner ends of the filaments is accomplished by conduction through the support rings 14 and'BI to the column 51 and thence through the cooling coil 60.

Cooling for the outer supports is by circulatory 7 system within thesupport pipes 62 themselves. A small water pipe enters theside of each of the support pipes 62, and extends axially within it to a point adjacent the lug 10. so that water entering this pipe will be squirted against the inner end of the lug. From there it returns through the pipe 62 externally of the pipe 90 to the bottom of pipe 62, where the end 90 of the next pipe is connected to carry the water to the next filament support, circulation thereby occurring through each of the support pipes 62 in I succession.

The supply for this circulatory system is through a fitting designated by the general reference character 9|, comprising coaxial pipes 92 and 93 which connect respectively to the two ends of the system. The outer of these pipes is permanently secured to the support ring 63 (see'Fig. 4). The fitting 9| passes through the flange l and is insulated therefrom by insulating bushings 94 of steatite or other refractory between which is a compressed rubber washer 95 forming a vacuum-tight seal. A connecting lug 91 for connecting one filament supply lead is mounted upon the fitting 9|,and the ring 63, and-the circulatory system comprising pipes 90 and 62 all constitute the conductingsystemfor supplying the filament current. The return circuit is through the column 51 and the flange 1, to which a second connecting lug (not shown) is attached.

.There are two otherfeatures comprised within the filament-grid'structure and their supporting systems. The first of these is a sliding plug 99 mounted in the end of the inner support column 25, and adjustable as to position by means of an operating rod I00, and anoffset extension rod IUI passing through a Wilson seal H12 in the gland box 42. The second is a cooling pipe I03 which extends'substantially the full lengthof the inner column .25 and is soldered thereto adjacent its 7 upper end for better heat transfer,

We are now in a position to consider the electrical characteristics of the filament-grid structure in view of the desiderataabove set forth, and

it is believed appropriate to do this at this point, since the same principles are involved in the supports for the remaining elements of the device and the explanation of all will be simplified if these principles are in mind. The necessary separation of the elements as regards D.-C. or low frequency potentials have already been accounted for. There is no metallic connection between the grid-support column 25 and the filament-support system comprising the column and the support pipes 62. Remaining to beaccounted for is the impedance relationships between the grid and filament members, and this is dependent upon the impedance of the coaxial transmission line formed by the inner and outer columns 25 and 51 and the coaxial conductors associated therewith.

The impedance characteristics of transmission lines whose lengths are of the same order of magnitude as the wavelengths of electrical oscillations transmitted thereby are now well known, but they are restated here for convenience in the explanations that follow. Most of them can be derived from the impedance diagram of a half-wave line open at the output end, as shown in Fig. 32, which indicates such a line diagrammatically, and shows the approximate curve of relative impedance looking into any portion of the line from the right, resistance of the conductors themselves being assumed to approach zero. Extremely short sections show a high capacitive reactance, which falls to the characteristic impedance of of the line at the /8 wavelength point, and to zero at the quarter-wave point, i. e., a quarter- Wave open-ended line acts as a dead short. From this point on the apparent reactance is inductive, rising again to the value at the point and approaching infinity at /zt.

Thesame diagram maybe taken as representing the impedance of a shorted-end line if the origin be taken at the nodal or quarter-wave point, which appears as a short when looking into the line. For short sections the rea'ctance is small and inductive, it rises to at the /8k point and approaches infinity at line is the geometric mean between its input and closing impedances. The short-circuit and opencircuit conditions are, of course, merely special cases of this general relation.

The lines comprising the element supports of the tube of my invention may be considered from a number of aspects, all depending on the general relationships above set forth, but in the treatment here adopted they are generally considered as divided into sections of quarter-wave length, or thereabout, as this is believed to lead to the simplest explanations.

We are interested in the impedance of the gridfilament support line as viewed from the grid end, but this impedance, is dependent upon the terminating or output impedances of the various sections, and therefore, in order to determine what the grid-end impedance will be, we must consider the various elements, section by section, starting from the outermost or lower end of the tube as shown in Fig. 1.

From this aspect the first section of the structure is the section including the adjusting rods 52, 4i, etc., the flange I0, and the section of the tubular conductor 54 illustrated as below the end of the column 51. Electrically this portion of the structure is a single conductor, and viewed from its upper end constitutes an end-fed antenna. It is preferable that its length be of the order of one-half wavelength at the operating frequency of the device, in which case its effective impedance Z2 will be in the neighborhood of 1,000 ohms. If its length be reduced to onequarter wavelength its effective input impedance will likewise be reduced to the neighborhood of from 50 to ohms, the quarter wavelength condition being the least desirable in practice. This antenna is considered as being fed by the coaxial transmission line comprising the tubular conductor 54 as the inner element and the column 51 as the outer element. With the spacing shown such a transmission line will have a characteristic or surge impedance 20, of about 10 ohms, and as has already been stated the length of this section of transmission line is approximately where is the wavelength at the frequency of operation. If we consider the quarter-wave condition to be fulfilled exactly the impedance looking into the coaxial line from the grid end will be If the antenna section of the system have an impedance of the order of 1,000 ohms, the characteristic impedance of the line being 10 ohms, the input impedance of the line will therefore be iii of an ohm. This low impedance therefore becomes the closing impedance of the section of line immediately preceding it. From one point of view it acts as a radio-frequency by-pass between the inner conductor 54 and the outer conductor 51, so that viewed from the input end, "at radio-frequencies the cylinder 54 and outer column 51 appear as a single conductor, and form, in connection with inner column 25, a single radio-frequency transmission line considered as fed from the grid-filament end through a slight impedance irregularity where the inner cylinder 54 terminates. Its effect from another point of view will be considered later.

Even if the conditions as to impedance of antenna and length of the coaxial line constituting the column 51 and cylinder 54 are not exactly met the result will be substantially the same. The antenna impedance can easily be kept above 100 ohms, making the impedance looking into the quarter wavelength line 1 ohm. If the length of the linesection is not exactly one-quarter wave, but is still materially greater than one-eighth wavelength, the input impedance will still be low in comparison with the characteristic impedance of the line, and although more power will escape than if optimum conditions are met the amount of power wasted by such undesired radiation will be very small.

The section of the inner line comprising the cylinder 54 and column 25 terminates in the slider 5|, which, as it is of large area and makes good contact with both conductors, may be considered as of zero impedance. This section may be tuned to exactly one-quarter wave by moving the slider. Due to the spacing between the two conductors the characteristic impedance of this section of transmission line is high, and if the resistance of the line were zero the input impedance would be infinite. Actually it may always be made to exceed 100,000 ohms and under optimum conditions may reach ten times this value. This section therefore forms a tuned radio-frequency choke of extremely high impedance interposed between the filament-grid structure and the outside world, and the impedance involved is so high that practically all energy reaching it is reflected back to its source.

What actually happens can be expressed more nearly in the terms of low frequency power line transmission if we think of the antenna as a load which is fed by a line terminating immediately above the top of the column 54. Current fed to this line from the central column 25 must proceed down the column, across the slider, and back to the top of the conductor 54, since owing to skin effect none will flow transversely through the wall of the conductor. In so flowing the current meets an enormous impedance-say 100,000 ohms. From there the line continues down the outside of conductor 54 to the antenna and back within the column 5'! to the terminus above the top of conductor 54. This latter length of line, including the load imposed by the antenna, has an impedance of, say, 1 ohm, and since the voltage available at the termini of the line will divide itself across this low impedance and the high impedance line section in series therewith in the proportions of the magnitudes of those impedances, and since the current flowing at the input points of the respective sections is the same, it follows that the energy delivered to the respective impedances will also be in proportion to their magnitudes, and only fi o of that delivered to the line will be transmitted to the antenna to be radiated thereby-stil1 less if optimum conditions are met.

From still a slightly difierent aspect, the small and largely resistive impedance ofiered by the outer line is at a current node. We therefore have a very small current flowing, and therefore the value of PR is vanishingly small, the R in this case being the apparent input impedance of the outer line and PR (practically) the energy radiated.

From whatever aspect the matter be considered the result is the same: The sections of the transmission line above the current node terminate in what is equivalent to an open circuit, just as would a low frequency line connected across an ordinarily good insulator. There is some consumption of power, which can be neglected in further consideration (as in the case of the insulator), and the succeeding sections can be treated as if they terminated at this point in an infinite impedance. It should be noted, however, that at the frequencies we are considering substitution of an insulator for the line sections would drop the impedance to a finite value and introduce large losses through radiation and dielectric phenomena.

The design problem to be met, therefore, is the design of a structure which, when terminated by an impedance approaching infinity, will have the properties of an anti-resonant circuit as viewed from cathode and grid. This structure is provided by two additional quarter-wave sections of the same line.

The first of these sections extends to include the upper slider 28, and its design is such that its electrical length may be changed in opposite sense to its physical length; i. e., such that it may be fitted in beneath the section above it even when the length of the upper section increases with decreased frequency of operation or vice versa.

This effect is obtained by means of the irregularity introduced by the low-impedance line section constituted by the slider 28. From the top of the high impedance section already described to the slider is a length of relatively high impedance line of less than n wavelength which therefore appears as a capacity variable from zero to some small value as the slider is moved to change its length from zero toward To this is connected the relatively great capacity of the slider portion of the line about 4 in length, but presenting an effective capacity many times as great as and apparently in parallel with that of the lower section, so that moving the.

slider to change the length of the section below it changes the apparent capacity as viewed from above relatively little. Therefore a very short length of the high impedance line above the slider is all that is necessary to tune this effective capacity to resonance, thus completing the quarter-wave section of line and bringing the node or quarter-wave point of the composite section a small distance above the upper slider face. The distance between the slider and the node will vary with frequency, of course, but only slightly with the position of the slider.

It has already been pointed out that the node is effectively equivalent to a short-circuit, and hence, since by moving the slider we may move the position of the node, by so doing we may tune the uppermost section of the line including the filament and grid. We have made that portion of the line below the slider and above the impedance loop relatively ineffective in tuning, so that we have an elastic or extensible quarter-wave section of line.

The final or grid-filament section may thus be resonated or otherwise tuned to give optimum operating conditions. In the case of Fig. 1, where capacity feed-back between anode l3 and grid cap 99 is used, the desired tuning of this section must provide a capacitive reactance. This is obtained by making the grid-filament section slightly longer than one-half wavelength or, in other terms, tuning it to a slightly lower frequency than that of the desired oscillation, so that as viewed at grid and filament it presents a small anti-resonant capacitive reactance. Under these circumstances the filament-grid system appears as a capacity in series with the capacity between the grid structure and the anode, and this latter capacity is adjustable by varying the position of the cap 99. When, therefore, the potential of the anode swings, the grid will assume a potential with respect to the filament (and ground) which is intermediate between cathode and anode potential, and which bears the proportion to the total potential between anode and filament that the effective series capacity between anode-grid and grid-filament bears to the apparent capacity between filament and grid. In other words, the arrangement is essentially a capacitive voltage divider which swings the grid potential in the same sense that the anode potential swings, and in fixed and .predetermined proportion thereto. Since the criterion for oscillation of the device is that the grid and anode should swing in the same sense and in step, the result is a highly effective capacity feed-back which is under control either by varying the actual capacity coupling with the cap 99 or by varying the efiective resonant input capacity of the grid-filament circuit by varying the position of the slider 28.

By the use of the two sliders the device is thus given its very considerable tuning range. The lower slider brings the current node to the point where the antenna is fed; the upper slider 28 moves the nodal point immediately above it and thus tunes the filament-grid section. The actual point of importance is that by adjusting the position of the sliders the effective resonant impedance of the filament-grid combination may be made to assume any value which may be desired, since the node above the slider 28 may be moved near enough to the rather large lumped cathode-grid capacity to embrace between the node and that capacity the exact small line inductance required for tuning it. In actual practice the efiective impedance will be made capacitive and small in comparison with the physical grid-cathode capacity, but it might, if desired, equally well be made inductive or resistive. Furthermore, since the effective resistances in the circuit are extremely low, and the losses are also small even though the circulating currents may be large.

A system of transmission lines, chokes and bypasses similar to that used in the filament-grid circuit is employed across the filament to prevent transmission of energy to D.-C. insulation and to prevent filament damage by R.-F. currents. The actual ground point on the filament circuit is the flange I on the outer casing 5 of the device. This, however, is unimportant and the effect of the transmission line arrangement may be considered as though the ground point were at the inner end of the filament. This may be considered as terminus of a quarter wavelength coaxial transmission line comprising the tubular conductors I9 and 8B, which is open at its lower end, terminating in high impedance. The transmission line impedance is again low, being of the order of, say, 5 ohms, and the line therefore forms a negligible series impedance as before, acting as a by-pass to the inner conductor. This, again, is a quarter wavelength line with the pipe 62 as its inner conductor, terminating in a dead short, and therefore ofiering very high impedance. As the potential imposed across this impedance is merely that which can build up across the short filament, amounting to a few volts at most, the escape of power through the filament support may be neglected, and the high impedance effectively in series with the filament prevents circulation of R.-F. currents which might otherwise cause hot spots and burn-outs.

We are now ready to consider the mounting of the remaining elements, i. e., the accelerator grid, boundary grid, and anode, which elements are shown in elevation in Figs. 9, 15 and 13 respectively. The accelerator grid is mounted from a side tube I85, which is welded to project through the wall of the housing 5 immediately below the flange 3. This side tube carries at its outer end a flange I01 which is surfaced to receive the tubular glass insulator I09, and the latter, in turn, carries a terminal flange Hi]. This structure may best be seen in the enlarged detail view of Fig. 6. As in the case of the main tube envelope, the tie-bolts which hold the structure together are omitted, but it will be understood that it is assembled in the same fashion as is the main envelope, with ground surfaces reenforced by greased rubber bands or gaskets III which form the seals. Two tubular conductors are fixed to and project inwardly from flange I I B. The inner conductor I I2 is spaced from the outer conductor H3 and is held accurately concentric therewith by an annular spacer H4. The accelerator grid I5 is supported from the inner member by a tubular bracket II5, the end of which fits within the conductor H2 and is rigidly secured thereto. A cooling pipe II'I, bent into a ring to surround the accelerator grid, has its ends brought down parallel to the support bracket H5 and enters the inner conductor on either side thereof, the ends of the pipe passing into the interconductor space distally of the spacer I I4 and emerging through the fiange HE]. A tuning slider H9, which nearly fills the space between the inner and outer conductors and does not make actual contact therebetween, is operated by means of a hook I28 whose end projects through a longitudinal slot in the conductor H2. A control rod I.2I is threaded to the end of the hook and emerges through a Wilson seal I22.

The supporting bracket H5 and cooling tubes II'I are carried up to the interspace between the control grid and the boundary grid through an angular fitting or shield I25, which passes through a notch I21 cut in one side of the filament support ring 6|. This construction is shown in Figs. 1, 6 and 11, each of these figures showing sections of the shield. The shield is electrically continuous with a pan I29 overlying-and contacting the filament support ring I4 and slotted immediately above the filaments, which forms an additional shield or barrier to separate completely the anode and control-grid sections of the tube except at the points where intercommunication is necessary or desired. The shield and pan therefore form one terminus, and the accelerator grid and cooling pipe II'I form the other terminus of the radio-frequency transmission line comprising the side tube I and the tubular conductors I I2 and I I3.

From what has gone before it is believed that the operation of this arrangement will be readily apparent. Again we have an antenna system comprising the control rods I2I and coolin tubes I ll, plus the projecting end of the conductor I I3, which is fed by and offers a relatively high impedance to a quarter wavelength transmission line of low impedance formed by the side tube I at and the conductor H3, and there is accordingly a radio-frequency by-pass between the grounded outer case 5 and tube Hi5 of the conductor H3. Within this there is another series section of transmission line comprising the conductors H2 and H3 and terminating in a short formed by the spacer II4. This inner line is tuned to a quarter wavelength by means of the slider I20, which acts as a loading capacity and increases greatly the electrical length of the line. In practice this slider is moved back to a point from which the line appears as a very large inductance at the operating wavelength. The proper point is that at which the remaining inductance and capacitance of the line, considered from the grid end, make it just a quarter wavelength from the inner end to the shorting spacer I I4, forming a' very high impedance at the shield where the grid I and cooling tube III are supported, and preventing any appreciable power being transmitted past this point to be radiated. The capacity of the grid I5 to the boundary grid 4 is large, and that to the control grid I5 is small; there is little coupling tending to swing the accelerator grid I5, and it consequently tends to maintain very nearly zero R.-F. potential.

As has already been described and as shown in detail in Fig. 16 the boundary grid 4 is firmly clamped between the flange 3 and the anode housing 2, and is therefore physically and definitely at the ground potential of the housing. The boundary grid and the anode face I3 again form the termini of a resonant line, comprising the housing 2 as its outer conductor and a cylindrical anode body, designated by the general reference character I30, within the housing. This resonant line is one-half wavelength long, and may be considered as terminating between the inner face of the flange I and the end I3I of the anode body. This will be recognized as an openended half wave line, and therefore of extremely high impedance when viewed from either end.

The construction and method of support of the anode body are best shown in Fig. 8. The support is from the midor quarter wavelength point of the anode, i. e., at a potential node, so that there is little tendency for power to escape from the support structure. Such tendency as there is for power to leak from the support point is suppressed by either or both of two methods. First, and preferable in the cases where the tube may be predesigned to operate at a fixed wavelength, is a movable plate I32 mounted on the sliding rod 50 of the Wilson seal first described, and making contact with the flange I by means of a sprin skirt I33. This may be adjusted to bring the node of the resonant line accurately at the point of support. This method of preventing direct radiation from the anode was adopted in the first of these devices constructed. It was quickly found, however, that the plate I32 was more useful as a tuning device, and therefore the principle of transmission-line-choke support was again employed to prevent power escape. In the construction then adopted and here shown a side tube I40 of relatively large diameter is welded at substantially the midpoint of the anode housing 2. The side tube carries a metallic flange I4I, with a glass insulator tube I42 fitted against it and in turn carrying a terminal flange I43. Through this terminal flange passes a pipe I44 which projects through a pass hole I45 in the side of the anode housing and on the end of which the anode body is attached. The action here will be described following the mechanical description of the anode, as the expedients adopted are predicated upon the necessities of the mechanical structure.

From the electrical point of View the anode body is a simple cylinder with closed ends. Its

complexity, as shown in Fig. 8, is due primarily.

to the provision for circulating cooling water within it, and to the provision of what-may be termed a rough tuning device.

Owing to the necessity for providing cooling the body itself must be water-tight, and accordingly it is constructed of a flared cylinder I41, to the flared end of which the anode face I3 is hard-soldered. The other end of the cylinder is closed by a threaded disc I49.

The supporting pipe I44 enters the flared cylinder I4'I through an aperture in the side thereof. The end of the pipe is threaded into a boss I48 on an inner bafile cylinder I 50, which boss is soldered to the inner wall of the cylinder I41. The boss I48 extends internally to form a cylindrical chamber I5I, which connects by a side pipe I52 through the end I53 of the baffle cylinder, so that water introduced through the pipe I44 is discharged directly against the active face I3 of the anode, and thence is forced around the exterior of the baffle cylinder to reenter its open end. It can then return within the cylinder to enter the open end of a return pipe I54, which is mounted concentrically within the pipe I44 by means of a perforated cap I55 which fits over the end of the pipe I44, its lower end passing out through the discharge chamber I5I. The cap compresses a rubber gasket I46, sealing the joint between the pipe I44 and the anode body to make it water and vacuum tight.

The upper end of the pipe I54 is centered in the pipe I44 by means of a metal bellows I51 which is sealed to both pipes and permits differential expansion between the two. Water is introduced into the pipe I44 through a side pipe I59, and its course can be traced by the arrows in the drawings through the outer pipe, the perforations in the cap I55, the side pipe I52, and thence around the bafile cylinder I50 and back through the central pipe I54,

The action of the mounting follows the principles already set forth, although the application is somewhat diiferent. A disc I50 is connected to the flange I 4| both electrically and mechanically, and carries a cylinder NH. The pipe I44 and the cylinders I40 and IBI form a transmission line one full wavelength long. Electrically this might equally well be a half wavelength line, but additional space is needed for the insulating cylinder I 42, which must withstand the full D.-C. anode potential of 20,000 volts or more. The length of this section is measured from the anode and its housing, and the impedance at its outer end is very high, so that looking into it from the anode the impedance is also very high.

This high impedance i connected in shunt across the line formed by the anode body I30 and anode housing 2 very near the nodal point, where the impedance of the latter line is low, and accordingly a very small portion of the current flowing at this point will take the high impedance path to the outer world.

In other terms, the full wave line is connected so near the node of the main anode oscillator circuit that only a few volts are effective across its termini, and therefore very small currents will tend to flow therein, representing a power loss of where V is the small input voltage and Z the large input impedance. Moreover, since the line is one Wavelength long, only the small voltage V will be

Images