US2227372A - Tunable efficient resonant circuit and use thereof - Google Patents

Description

Dec. 31, 1940. D. L. WEBSTER EIAL 2,227,372

TUNABLE EFFICIENT RESONANT CIRCUIT AND USE THEREOF Filed July 21, 1938 2 Sheets-Sheet 1 INVENTORS DAVID L. WEBSTER WILL/3Z7 HAN EN A TORNE 1940- i D. L. WEBSTER ETAL ,372

TUNABLE EFFICIENT RESONANT CIRCUIT AND USE THEREOF Filed July 21, 1938 2 Sheets-Sheet 2 INVENTOR8 DA v/o'L. lA/EBJTER IL LIA/1 HA/ViEN .8)

J ATTOR Patented Dec. 31, 1940 UNITED STATES PATENT OFFICE TUNABLE EFFICIENT RESONANT CIRCUIT AND USE THEREOF tion or California Application July 21, 1938, Serial No. 220,414

8 Claims.

This invention relates, generally, to resonant circuits and the invention has reference, more particularly, to a novel tunable high efficiency resonant circuit characterized by an electromag- 5 netic field contained within a hollow conducting member, the said circuit being of the general type disclosed in copending application Serial No. 92,787 filed July 27, 1936 of William W. Hansen patented Feb. 20, 1940, No. 2,190,712.

10 One object of the present invention is to provide a high efiiciency resonant circuit of the above type having novel means for tuning or changing the resonant frequency of the same as desired.

Another object of the present invention lies in the provision of a novel circuit of the above character that is particularly adapted in use for transferring energy to or from electron streams.

Still another object of the present invention is to provide a novel resonant circuit of the above character wherein a hollow conducting member is provided with re-entrant end portions of smaller area than the ends of said member, whereby any required acceleration of electrons traversing the space between said re-entrant end portions can be accomplished with less RI loss in said conducting member than if said conducting member had fiat ends.

Other objects and advantages will become apparent from the specification, taken in connection with the accompanying drawings wherein the invention is embodied in concrete form.

In the drawings, Figure 1 is a central sectional schematic view illustrating the principles of the present invention.

5 Figure 2 is a central sectional view illustrating a form of re-entrant end resonant circuit showing the conducting member provided with an external hollow fiange providing a space for receiving energy from electron streams.

40 Figure 3 illustrates a modified detail of construction.

Figure 4 is a central sectional view of a practical form of reentrant end resonant circuit employed for accelerating or decelerating electrons.

. 5 Figure 5 illustrates the use of a sphere with reentrant cones, and Fig. 6 illustrates another modified form of the invention.

Similar characters of reference are used in all of the above figures to indicate corresponding 50 parts.

Referring now to Figure l of the drawings the reference numeral I designates a hollow cylindrical conducting member as of copper, the same having fiat ends 2 and 3 provided with central reentrant portions or electrodes 4 and 5. These reentrant portions or electrodes are illustrated by the arrows as adjustable as to their depth of projection into the interior of member l. Member I is adapted to contain oscillations set up in any suitable manner such as described in the before 10 mentioned application and as will further appear, standing electromagnetic waves being established within the member, having their magnetic field component substantially as represented by the dots of Figure 1 and their electric field component substantially as represented by the lines extending between the ends 2 and 3 and the opposed faces 6 and I of the reentrant end portions or electrodes 4 and 5 in this figure. There is no radiation from this closed member in spite of the fact that the physical dimensions may be of the order of the wave length produced in free space by oscillations of the same frequency. 1

Assuming that the total height of the member I is 2b and that the extent to which each of the electrodes 4 and 5 projects into the interior of the member I is h, then for a given total charge on the ends of member I, by increasing h continuously from zero, as by pushing the members 4 and 5 inwardly'of member I, it will be evident that the charges 'on the ends 2 and 3 must 'fiow more and more onto the electrode surfaces 6 and l, for

otherwise there would exist an electric field with a component tangential to the sides of the electrodes which would effect the fiow of the charges to surfaces 6 and I. To explain the changes in frequency produced by changes in the positions of these ends it should be noted that the currents at all points on said member oscillate substantially in the same phase. 40 1 but less drastic reduction of B elsewhere.

one of these forms must aiTect it equally in the other.

If the ends are made slightly more reentrant, as by increasing h in Figure 1, but the total charge of either sign is made to attain the same peak value as before, the change in electric energy can be predicted by expressing this energy as 1 f .E dl) where E is the strength of the electric field in Heaviside-Lorentz units and do an element of volume. While E is changed somewhat at any point within said conducting member, the greatest changes are in the regions covered by the advances of said reentrant ends, where E is cut from an exceptionally large value to zero. Consequently the electric energy is reduced.

The magnetic field in said regions, on the other hand, is exceptionally weak. Consequently the magnetic energy is not reduced so much by the reduction of the magnetic field strength B to zero in this region. Therefore, it must be reduced by a more general This must be accomplished by a general reduction of the current density and'this in turn, under the above postulate of constant maximum total charge of either sign, can be accomplished only by a reduction in the frequency. Consequently, if the ends are made more reentrant, other dimensions being constant, the frequency is reduced.

This change is somewhat analogous to the change in an ordinary resonant circuit, consisting of a condenser and a coil, when the plates of the condenser are brought nearer together making its dielectric thinner and increasing its capacitance. In the extreme case where said reentrant portions approach one another infinitely closely, so that h b, they can indeed be considered as approximating condenser plates, so that the frequency does then approach zero like In all other cases, however, as in shapes like those of Figure 1, it would be misleading to speak of capacitance between said ends or even between said reentrant portions thereof. This is because the electric field is not of a static type, its lines and distribution of strength depart widely from those of an electrostatic field between electrodes similar to said ends of said member, and the concept of potential difference between the plates of a condenser fails to apply to the ends of said member. [his failure toapply consists partly in the fact that said potential difference in a condenser is the same for all points on one plate and all points on the other, whereas there is no corresponding quantity with the same value for all pairs of points in the present invention.

A further failure of the concepts of capacitance and potential difference here appears when one attempts to remedy this first failure by introducing the concept of distributed capacitance in the manner successful with ordinary radio circuits. This failure arises from a breakdown in the ordinary definition of potential difference in the present case. In an ordinary circuit the potential diflerence between one point A and another point B can be defined accurately enough as the work done by the electric field on a unit quantity of electricity taken from A to B. This is a useful concept because,for static fields, and with some limitations also with alternating fields of relatively low frequency, the work done in moving a charge from one conductor of a certain potential to another conductor ofanother potential is independent of the path followed and the time taken in making the transfer, and the concept of potential is given a definite meaning by the existence of this constant amount ofwork. Unfortunately, in dealing with the subject matter of this application, this simple equivalence of work no longer holds true, and therefore the concept of.potential loses its ordinary meaning. Hence, in order to avoid considerable confusion and misconception it is necessary to abandon the concepts of the electrical engineer which are built around the concept of potential, and consider the more fundamental concepts of physicists.

The necessity for this change of concepts will be evident from the following: In Figure 2 the hollow conducting resonator shown, when excited, will have an alternating electric field along its axis. This field is the actual observable phenomenon and from it the existence of a potential on the surfaces 2, 3, and 6, 1 might be inferred. But if an attempt is made to measure this potential by transporting a charge from A to B, for example, one value will be obtained if the charge is transported from A to B along the straight dash line of Figure 2, whereas a smaller answer will be obtained if we transport the charge by an arcuate path, shown by a dash-dot line, and we will get an exceedingly small answer if we transport the charge along the inner surface of the hollow conductor. In fact in this latter case, the computed potential will be merely where P is the power dissipated in the hollow resonator, and I is the circulating current. This value of the computed potential will usually be less than a thousandth part of the potential computed from the work done on a charge moving along the axis of member i from surface 6 to surface I.

This is not to deny the existence, or utility in other ways, of a potential function with diiferent values at different parts of member I. There is indeed a scalar potential V that satisfies the equation where A is the corresponding vector potential, DtA its partial derivative with respect to time t, c the speed of light and wherein grad is defined as in Equation 2, page 14, of Abraham and Becker's book entitled Classical Electricity and Magnetism, published 1932 by Blackie and Son, Ltd., of London, England. In this case, however, it must be noted that this equation alone is not suflicient to define either V or A, but must be supplemented by some other equation, and that said supplementary equation can be chosen somewhat arbltrarily. The two most commonly used supplementary equations, namely div A= and divA+%D,V=O

where div is defined on page 18, Equation 9, of Abraham and Becker. These equations lead to definitions of V as potential integrals of the charge density, instantaneous and retarded, respectively. As applied to a condenser in an ordinary circuit, these functions are nearly identical,

so that either of them can be used to define the diflerence or potential between the condenser plates; and E, at any point within the condenser, is given accurately enough by the term -grad V ii alone. Likewise as applied to the inductive part of an ordinary circuit, the induced electromotive.

are of coordinate importance. Indeed the electric field E is zero outside the inner surface of member I only because these terms are equal in magnitude but opposite in direction at all outside points, whereas near the middle of member i they are nearly in the same direction. Qualitatively speaking, if one associates the term grad V with condensers and electrostatic fields, and the term with induced electromotive forces, the electric field of the present invention is a sum of fields of the two kinds in somewhat arbitrary proportions depending on the arbitrary defining equaion.

In summary, therefore, the ordinary concepts 40 oi potential differences and capacitances are inadequate for the analysis of the present invention; but the reasoning stated above in terms of magnetic and electric energy values is valid, and the change in the ends in Figure 1 to make 45 them more reentrant does therefore reduce the frequency of free or resonant oscillations.

Useful variations in frequency can be obtained by adjusting the value of h by arranging the electrodes or projections l and 5 as.structural 50 members that can be moved in and out of the member I as shown in Figure 1, or as in Figure 3 wherein the sides of the electrodes 4 and 5' are shown as of the sylphon or bellows type adapted to be varied as to length by actuation of rods 8 s5 and 9. The structure of Figure 1 is adapted to be contained within an evacuated chamber (not shown) whereas the oscillatory circuit member I of Figure 3 is adapted to be evacuated without the necessity ofusing an enclosing chamber.

The above approximate calculation shows how, by starting from a cylindrical resonator whose resonant wavelength, for example, can be computed exactly, one can predict in which sense certain distortions will alter the wavelength.

Moreover, the above calculation can be done in greater detail to find approximate numerical values, the approximation being good when the final shape does not departgreatly from a cylinder with flat. ends.

conclusions on reentrant ends is a quantitative calculation of frequencies for resonators of two shapes which lend themselves readily to such calculations and bear some similarity to the flatend cases treated above. Two such shapes are:

(l) The simple sphere described in application Serial No. 92,787, along with the fiat-ended cylinder described there, and (2) a sphere modified and shown in Figure 5 as SI, having two reentrant conicalend portions II and 31. These end portions are supposed to come very close together and in fact the calculation to be reported assumes them to be separated by only an infinitesimal mount.

Assuming this shape, it can easily'be shown that the electric field is like that qualitatively shown by the curved lines in Figure 5 and that the resonant wavelength of the ground state is just 4 times the radius of the sphere. This result does not depend on the angles 0! the cones. For comparison we may note that the wavelength of a cylinder with fiat ends is 2.61 times the radius of the cylinder, thus proving that the acceleration of the two reentrant conical end portions made a suitable reduction in the frequency of oscillation in a manner similar to that described above in connection with Figures 1 to 4.

The shape illustrated in Figure 5 is often a rather close approximation to shapes desirable in certain practical applications. For example, use has been found for the device illustrated in Figure 6, having substantially the form of a cube a with two opposite corners turned inside out to provide reentrant end portions 3. and 40.. It has been shown experimentally that the wavelength of this device could be predicted with good accuracy from the above knowledge relative to shapes like Figure 5.

Parenthetically, it should be noted that the lines of electric field shown in the several figures will be distorted in rather complex ways by any conducting loops that may be inserted, such as it in Figures 2 and 4. Such distortions are localized near the loops and have not been shown in the drawings.

In describing the driving mechanism of Figure 2, use is made of the same method of analysis, so that in place of the idea of potential of the walls of the conducting chamber-ll we now deal only with the electric and magnetic fields set up within the chamber as a boundary of the field. The energy stored in the resonator is in the form of electromagnetic energy which is bounded by the shielding and/or reflecting eflect of the metal walls. The skin depth of the current flowing in the walls is the thickness of metal that is just sufllcient for the induced currents set up by the waves within the chamber to reduce the field beyond this depth due to the waves to of its value in the space inside. (e being the base of natural or Napierian logarithms.) It is clear that a metal wall of thickness many times the skin depth will reduce the field to a negligible amount. Any wire totally outside the chamber is therefore not subject to the oscillatory field at all; and in the case of battery H, for example, it makes no difference where the contact of the battery lead to the outside of member H is made so long as the capacity between plate i5 and member II is large for by-passing radio fre- 70 An alternative method of proving the above quency currents. Whenever a wire runs inside member Ii, on the contrary, as in the case of loop IE, it is subject to the oscillatory electric field, and must have oscillatory currents induced in it. The voltages driving such currents can be predicted from the nature oithis electric field,

with proper allowance for the distortion of the field by any oscillatory currents on the wire.

In Figure 2 the hollow cylindrical evacuated member ll shown, is provided with a hollow external annular flange ill. The flange In, like the electrodes or reentrant projections 4 and 5 of Figure 1. changes the frequency of the circuit. by decreasing the frequency with increasing dimensions of the flange.

In addition to changing the frequency of the member II. the projections 4" and 5" of this member and the flange ill have the advantage of presenting traversable intervals in the contained held that are relatively short. This is a distinct i5 advantage in certain applications of the circuit requiring the transfer of energy into or out of the same by use of electron beams. In Serial No. 92.787 three methods of energy transfer in a can shaped hollow member are shown, namely: an

inductive coupling loop, a capacitive coupling plate or grid, andan electronic coupling beam. When an electron beam is used for coupling, the best results are ordinarily obtained when the electrons of the beam traverse the field of the hollow member in less than one-half period of oscillation. This condition is readily fulfilled when using the structure of the present inventron: partly because the reduction of frequency by the addition of either the internal projections 4" and 5" or the external flange i0 makes any given time of transit through the field a smaller traction of a cycle; and partly because the dislance through which an electron must pass, either between the internal projections, from surface 6 to surface 1, or across the flange, from grid l2 to anode surface I5. is less than the distance from surface 2 to surface 3.

In Figure 2 the flange ill on the exterior of the member H is used for coupling the circuit to a beam of low velocity electrons. An emitting filament i3 is arranged next to the grid I 2 in a suitable evacuated space I 4, and connected to the outside of member it through a tube IS. A plate I 5 insulated from member II is con- 5 nected to battery I! showing its other terminal grounded on member II. The grid I2 is placed in the wall of flange l0 adjacent to filament l3 and connected to filament l3 through the tubular coupling loop which may be as wide as is needed to carry the required high-frequency current or may consist of several similar loops in parallel, linking the same lines of force of the magnetic field. There may be several units, each consisting of elements similar to l2, l3, l4, I5, I6, and i1, spaced around flange it), either equally or unequally spaced.

In operation, electrons are drawn from the filament l3 during the alternate half-cycles of oscillation within member II when the filament I3 is negative with respect to grid l2. They enter the space between grid l2 and plate l5, where the electric field is a resultant of two superposed components, one the electrostatic field due to the charges imposed by battery I! on grid l2 and plate i5, and the other a part of the alternating, or dynamic, electric field due to the oscillations within member ll. During the half of the cycle in which the electrons enter the flange interior this dynamic component opposes their motion,

while the static component helps it. Consequently, if the static component is stronger than the dynamic, it will drive the electrons across the flange, forcing them against the dynamic field and thereby feeding energy to the oscillations.

Plate l5 may be a plate, as shown in Figure 2,

or it may be perforated to form a grid. It should be understood that the important consideration is transit of the electrons through the Id. between grid l2 and plate II and that it es no fundamental diflerence in the operation of the system whether the electrons are stopped by the plate II orpermitted to on through as they will if plate fl is perforated, and an evacuated space is provided beyond. Accordingly, it will be evident that if there were no resistance or other loss of power from the oscillations they would increase in strength until their dynamic electric field was strong enough to prevent the transit of electrons during a part of the half cycle, either by actually reversing their motion or by merely delaying them too much, or both. Since there always is resistance, and in the practical uses of this invention there are other deductions of power from the oscillations, as in the use of the electric field between surfaces 8 and I for accelerating other electrons as described later, the oscillations are often weaker than those just de-* scribed; but the oscillations may still attain that strength if the electron current taken from battery i1 is great enough to allow battery I! to supply the requisite power for all losses, both by resistance and by electron impact, and for all uses of power drawn from the oscillations. The use of much more current, or of a battery of electromotive force much greater than that needed to secure the transit of the electrons from grid i2 to surface l5 during the time of a strong opposing dynamic electric field, causes an excessive waste of power in electron impacts on surface l5, and therefore is usually inadvisable.

It is evident that there aremany other methods of sending electrons through flange ill from grid l2 to plate I5 in greater numbers during the time when the dynamic component of the electric field in said flange opposes their motion than during the time when said field assists their motion. All such methods are essentially equivalent to the method described herein, in that they result in an equivalent transfer of power to the oscillations in member II, and Figure 2 shall be interpreted as illustrative of such methods and not in a limiting sense.

If it is desired to produce an electric field of any given strength along the axis of member II, as indicated by the lines extending between surfaces 6 and I, the dimensions of the external flange Ill can be chosen so as to cause the field between grid i2 and surface l5 to have such strength that the electro-motive force of a battery chosen as directed above has a convenient I value. These observations follow from the general theory of this type of circuit-as set forth in Serial No. 92,787. Thus, it is possible to produce a high field between the top and bottom of member II when using a comparatively lowvoltage electron source at the faces l2 and ii of the flange It]. This arrangement also permits the coupling of low voltage energy sources adaptable to small dimensions for energy transfer into resonant circuits of this type wherein large dimensions are desirable to get the desired high energy transfer to electrons.

The adaptation of low-voltage energy transfer to the production of high-speed electrons is accomplished as shown in Figure 4. In this figure the hollow resonant member I 8 is shown with internal projections or reentrant ends i9 and 20. The electrode faces 2| and 22 are perforated to form a pair of spaced grids. An electron emitter 23, such as a heated active oxide-coated surface is provided and connected with a battery 24 for accelerating the emitted electrons toward the grids 2| and ,22. A control grid 25 is' placed between the electron emitter 22 and the first resonant circuit grid 2|. Control grid 2! is connected with a biasing. battery 2| through a phasing network comprising a variable inductance 22 and a variable capacitance 2!, battery 22 being con nected to the electron emitter 22. Inductance 22 is coupled to a second inductance II which is connected to a coupling loop ll in the field of the resonant circuit within member II. The use of low velocity electrons directed'across the interior of flange It is also employed in this flgure for establishing and maintaining an oscillating electro-magnetic field within the hollow circuit member II. If desired, loop I! may be connected directly to an alternating current converter (not shown) of any desired form, which converter may be arranged to act either as a driver or as a load.

In the operation of the system, a beam of electrons is projected through the grids 2| and 22 from the emitter 23. Electromagnetic oscillations are produced in member it by electrons emitted from filament II. An alternating potential is impressed on grid 25 by the coaction of loop 3| and the connections thereto. The phasing network 28-29 is adjusted so that grid 25 passes electrons in the maximum amount at any desired part of the alternating .cycie. For example, if the electrons are controlled so they enter at the maximum rate during the part of the cycle when the field between grid 22 and grid 2| is in the direction to oppose the electrons, the motion of the electrons between grids 2| and 22 will be retarded and the electrons will leave grid 22 with velocities less than the velocities with which they enter grid 2| and they will consequently deliver energy to the field of member l8. Conversely, if the electrons pass grid 2| at the maximum rate when the field between this grid and grid 22 is in a direction to assist the motion of the electrons, they will be accelerated in transit between grids 2| and 22 and energy will be transferred from the electromagnetic field of member id to the electron beam. The transfer of energy both as to direction and rate can be controlled by adjustment of the phasing network. Consequently in the event a converter is used attached to loop IS the transfer of energy between the field and the converter can similarly be controlled.

Owing to the use of the reentrant ends l9 and 20, the electric field strength between grids 2| and 22 is increased while the flight distance of electrons in passing through the resonant circuit field is shortened so that it is easily possible to pass electrons between these grids in one half cycle as indicated by the curve 33 in Figure 4, or in less than one half cycle.

The present invention is especially well adapted for the production of high-voltage X-rays after the manner described in copending application Serial No. 92,787, by its use to accelerate electrons to extremely high velocities, so that they may subsequently produce X-rays. In this use and in many other uses the requirements to be met are chiefly the production of an electric field of at least a given strength all along a line of a given length, as along the axis of member i8 from grid 2| to grid 22. In such a case these grids may be taken as the first fixed points of the design. Then to prolong the half cycle well beyond the time of transit of the electrons from grid 2| to grid 22 it is desirable to increase the diameter of member ll. However, if member I! had flat ends. spaced at the same distance as grids 2| and 22, the charges on these ends would be much greater than those of the grids alone. This fact makes it desirable to move most of the end surfaces farther apart, as in member II. The reduction in the frequency following this change helps also to reduce the oscillatory current and its ohmic power loss. The best dimen sions in practice depend on many factors, such as the speeds of the electrons entering member l8 at grid 2|, the desired axial field strength, distance and frequency, the power carried by the electron stream, the space requirements of other devices associated with this one, and many other factors in the various uses of this device.

In the claims the expression shunt impedance will be understood from the following: Suppose that it requires P watts to maintain a root mean square voltage V across a certain resonant circuit operating at its resonant frequency. Then is a quantity of the nature of a resistance which is often called the shunt impedance.

As many changes could be made in the above construction and many apparently widely different embodiments of this invention could be made without departing from the scope thereof, it is intended that all matter contained in the above description or shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.

What is claimed is: I

l. A resonant circuit comprising a hollow conducting member arranged to contain a standing electromagnetic field, said member having opposed hollow reentrant portions for determining the frequency of oscillation of said field, said reentrant portions serving to establish a relatively intense electric field therebetween, said reentrant portions being apertured, and means including a thermionic emitter for passing electrons through said field by way of said apertured reentrant portions for changing the velocity of the v electrons.

2. A high efiiciency resonant circuit comprising enclosing conducting walls bounding an electromagnetic field, said walls having a regular geometrical form and defining a substantially closed conducting body, an exterior hollow flange projecting outwardly therefrom, the electric vector of said field being relatively weak within said flange and relatively strong at'the center of said body removed from said flange, and means for projecting electrons across the space within said flange in a direction substantially perpendicular to the opposite walls thereof for setting up and maintaining the electromagnetic field within said conducting body.

3. A resonant circuit comprising, an evacuated container of substantially cylindrical form provided with an external outwardly projecting hollow annular flange having spaced opposed walls and an outer closing wall, and means for setting up standing electromagnetic oscillations within said container, said means comprising a thermionic emitter and means for applying electron accelerating voltage between said emitter and a portion of said flange for driving electrons in a direction at right angles to the opposed walls of said flange and across the space provided within said hollow flange, in which space the efiective 1'5 electric field strength is considerably less than the maximum electric field strength existing within said container.

4. A resonant circuit comprising, an evacuated container of substantially cylindrical torm provided with an external hollow outwardly projecting annular flange, and means for setting up standing electromagnetic oscillations within said container, said means comprising cathode and anode means for drivingelectrons across the space provided within said hollow fiange, in which space the effective field strength is considerably less than the maximum field strength existing within said container said container being provided with opposed reentrant end portions whereby the strength of electric field between said end portions is greatly enhanced.

5. .A resonant circuit comprising, an evacuated container of substantially cylindrical form provided with an external hollow annular flange, and

means for setting up standing electromagnetic oscillations within said container, said means serving to drive electrons across the space provided within said hollow flange, in which space the effective field strength is considerably less than the maximum field strength existing within said container, said container being provided with opposed reentrant end portions of relatively low 8. An electromagnetic oscillator comprising a hollow resonant circuit member having an external hollow flange, means for passing electrons through the space within said fiange ior setting up and maintaining an electromagnetic field within said member, a source of electrons, means for forming the electrons into a beam and pro- ,lecting them through the said electromagnetic field, an element providing a controlling field for said electron beam, and means for energizing said controlling element from the energy of the said electromagnetic field in proper phase relationship ior varying the velocities of the electrons of said beam.

7. A resonant circuit comprising a hollow conducting member arranged to contain a standing electromagnetic field, means for establishing and maintaining said field within said member, said member having an apertured reentrant portion, and means including a thermionic emitter tor passing electrons through said field by way of said apertured reentrant portion for changing the velocity of the electrons.

8. A resonant circuit comprising a hollow conducting member arranged to contain a standing electromagnetic field, means, including a thermionic emitter, for driving electrons through a por-' tion 01 the space within said member for setting up and maintaining said electromagnetic field, the electrons passing through that portion of said space where the electric field vector is relatively weak, said member having an apertured reentrant portion at a point thereof where the electric field vector is relatively strong, and means including a second thermionic emitter for passing electrons through said field by way of said apertured reentrant portion for changing the velocity of the electrons.

DAVID L. WEBSTER.

WILLIAM W. HANSEN.

Images