US3010083A - Nonreciprocal microwave devices - Google Patents

Description

4 NO 2 1 J. A. WEISS 3,010,083

NONRECIPROCAL MICROWAVE DEVICES uvvavrok x y Y J A WEISS A TTQRNEV Nov. 21, 1961 J. A. WEISS NONRECIPROCAL MICROWAVE DEVICES 5 Sheets-Sheet 2 Filed Aug. 3, 1959 FIG. 2

w RS 05 F, n: \E. .M U M N EW 3% x V 5 a W G H Y B a a 9 2 ,W 5 n: EX N w G 0 F a a F E W J4 II t I w w m x 9 y E x x Q 9 x 5 E E 5 4 3 m. g 4 M B C G E 4 4 H m m m ATTORNEY E9 (Ni) Nov. 21, 1961 J. A. WEISS NONRECIPROCAL MICROWAVE DEVICES 5 Sheets-Sheet 5 Filed Aug. 3, 1959 FORWARD D/RE C 7'/0N FIG. 6

REVERSE D/PEC T/ON //v l/ENTOR J. A. WEISS af/zimu A T TODAIE Nov. 21, 1961 Filed Aug. 3, 1959 J. A. WEISS 3,010,083

NONRECIPROCAL MICROWAVE DEVICES 5 Sheets-Sheet 4 lNl/EA/ TOR J A. WEISS BY A TZ'ORNE V Nov. 21, 1961 J. A. WEISS 3,010,083

NONRECIPROCAL MICROWAVE DEVICES Filed Aug. 3, 1959 5 Sheets-Sheet 5 FIG. 9

INVENTOR .1 A. WEISS A 7' TORNE V United States Patent Ofitice 3,010,083 Patented Nov. 21, 1961 3,010,083 NONRECIPRGCAL MICROWAVE DEVICES Jerald A. Weiss, Summit, N.'J., assignor to Bell Telephone Laboratories, Incorporated, New York, N.Y., a corporation of New York Filed Aug. 3, 1959, Ser. No. 831,416 14 Claims. (Cl. 333-9) This invention relates to electromagnetic wave transmission systems, and, in particular, to multibranch .circuits having nonreciprocal transmission properties for use in said systems.

Many of the known microwave devices whose operation is founded upon gyromagnetic phenomenon reveal practical operating limitations when attempts are made to utilize them at the higher microwave frequencies. For example, resonance and field-displacement isolators tend to require too high a magnetic biasing field since, for these devices, the intensity of the biasing field is proportional to the operating frequency. Isolators relying upon rotational effects are subject to variations in attenuation produced by frequency or temperature changes. Similarly, other types of microwave components which rely upon any of the above-mentioned efiects are likewise adversely affected.

It is, therefore, the general object of this invention to produce microwave devices capable of operating at high frequencies with low biasing fields and which are inherently broad-band.

In the copending application of A. G. Fox, Serial No. 501,241, filed April 14, 1955, there is described a transmission device for producing nonreciprocal rotation of the plane of polarization of linearly polarized electromagnetic wave energy. The device comprises a tapered pencil of gyrom gnetic material axially located in a conductively-bounded guide whose cross-sectional dimensions vary from cross section to cross section along the longitudinal axis of the guide. The device therein described is of particular interest in applied at either end emerges at the mid-point of the structure as a circularly polarized wave. Furthermore, the device operates with relatively low biasing fields and is inherently broad-band. These unusual properties of a half section of the device are utilized to produce nonreciprocal polarization rotation similar to the Faraday effect.

It has since been discovered, however, that other polarizations may be produced at the half-way point in a tapered structure similar to the Fox structure, which give rise to an entirely new class of nonreciprocal microwave components. These alternate polarizations result from the special design of the taper. In particular, it has been discovered that a linear polarization may be produced whose direction of polarization is inclined at a forty-five degree angle from that of the incident wave energy and Whose direction of inclination is a function of the direction of propagation. By coupling to the structure in the region of linear polarization, nonreciprocal directional effects are produced. In particular, couplers placed in the region of the inclined polarizations may be utilized to produce isolator action and three and four port circulators.

These and other objects and advantages, the nature of the present invention, and its various features, will appear more fully upon consideration-of the various illustrative embodiments now to be described in detail in connection with the accompanying drawings, in which:

FIG. 1 is a perspective view of the tetrahedral tapered waveguide;

FIGS. 1A to 1C show cross-sections of FIG. 1;

that a linearly polarized wave FIG. 2 shows, by way of illustration, the change in polarization of a linearly polarized wave produced by the gyromagnetic material;

FIG. 3 is a perspective view of the butt joint section, a modified form of the tetrahedral tapered section;

FIG. 4 shows, by way of illustration, the nature of the wave energy in various regions of the butt'joint section of FIG. 3;

FIGS. 4A to 4E show, by way of illustration, the polarization of the wave energy in the various regions of the butt joint section shown in FIG. 4;

FIG. 5 is a perspective view of van isolator using the tetrahedral tapered section;

FIG. 6 is a perspective view of the tetrahedral section connection as a t-hree'port circulator;

FIG; 7 is a perspective view of two tetrahedral sections connected and utilized as a four port circulator;

FIG. 8 is a perspective view of a tetrahedral section connected as a four port circulator;

FIG. 9 is a perspective view of a butt joint junction using coaxial cables; and

' FIG. 10 shows, by way of illustration, the field configuration at the butt joint junction of FIG. 9.

Basic to the several embodiments of the invention is the so-called tetrahedral waveguide section 11 shown'in FIG. 1 interposed between the two crossed rectangular waveguides 12 and 13. Waveguide 11 tapers smoothly and gradually along its length from a rectangular transverse cross section at guide 12, having its wide dimension extending horizontally, to the rectangular cross section of guide 13, which has its wide dimension extendingvertically. The name tetrahedral derives from the fact that by connecting the sides of the two crossed guides 12 and 13, in the-manner shown, the resulting tapered section 11 is in the form of a doubly truncated tetrahedron. I Y

Guides 12 and 13 are proportioned to have a .wide in-. tern-al dimension of just less than one wavelength of the wave energy to be supported therein, and a narrow dimension substantially one-half of the wide dimension. So

- proportioned, these guides are supportive of the dominant ing magnetic field component.

mode of Wave propagation in which the electric lines of force extend in a direction parallel to the narrow walls, but are cut off for wave energy whose electric field is parallel to the wide walls.

In the particular embodiment of the tetrahedral waveguide shown in FIG. 1, guides 11, 12 and 13 are colinearly aligned along a common longitudinal axis, with guide 13 rotated about said axis ninety guide 11, section A-A, tangular cross section with its wide dimension horizontal, which tapers into the symmetrical cross section B-B, shown in FIG. 1B. In particular, the cross section of section BB is square. From section B'-B, guide 11 tapers to the rectangular cross section CC shown in FIG. 1C, which has its wide dimensions directed vertically.

Extending along the axis in the region of the tapered section 11 is a cylindrical rod of gyromagnetic material 14, suitably supported at the ends by means of the dielectric supports 15 and 16. The term gyromagnetic material is employed here in its accepted sense as designating the class of magnetic polarizable materials having unpaired spin systems involving portions of the atoms thereof that are capable of being aligned by an external magnetic polarizing field and which exhibit a significant precessional motion at a frequency within the rangecontemplated by the invention under the combined influence of said polarizing field and an orthogonally directed vary- This precessional motion is characterized as having an angular momentum and a degrees with respect to guide 12. Taking sectional views of the tetrahedral waveshown in FIG. 1A, has a recmagnetic moment. Typical 'of such materials are ionized gases, paramagnetic materials and ferromagnetic material-s, the latter including the spinels such as magnesium aluminum ferrite, aluminum zinc ferrite and the garnetlike materials such as yttrium iron garnet.

Rod 14 is biased by a steady polarizing magnetic field of a strength to be described. As illustrated in FIG. 1, this field is applied longitudinally, i.e., in the direction of propagation of wave energy in guide 11, and may be externally supplied in any of the usual ways well known in the art or rod 14 may be permanently magnetized if desired.

Before proceeding further with a detailed examination of the operation of waveguide components utilizing the tetrahedral waveguide, it is preferable that there be a clear understanding of the operation of the tetrahedral waveguide itself. Consider a dominant mode wave applied to guide 12, which for the purposes of explanation is located in the x---yz coordinate system shown in FIG. 1. The electric field vectors are all polarized in the x direction as represented by the vector E Because the guide is less than half a wavelength in the x direction, as indicated in FIG. 1A, guide '12 is incapable of supporting wave energy polarized in the y direction. In the absence of any coupling mechanism between the x and y directions (such as a magnetized gyromagnetic rod), the wave energy would tend to propagate along the tapered guide 11 in the z direction until the y dimension of the guide decreased to less than the cut-off dimension, at which point the energy would be reflected back toward guide 12. In the presence of the polarized element 14, however, energy is coupled from the applied wave, which is polarized in the x direction, to a wave polarized in the y direction. The coupling produced by element 14 can be explained by the recognition that the gyromagnetic material of element 14 contains unpaired atomic spins which tend to line up with the appliedfield H These spins have an associated magneticmoment which can be made to precess about the line of the biasing magnetic field, keeping an essentially constant moment in the direction of the applied biasing field and at the same time providing a moment component which may rotate in a plane normal to the field direction. Thus, when a reciprocating high-frequency magnetic field of electromagnetic wave energy is impressed upon the moment, the moment will commence to precess in the preferred angular sense. The effect of this precession is to produce a reciprocating field at right angles in space to the applied field and displaced in time from the applied field.

Thus, when the wave E is applied to the tetrahedral section 11 in which rod 14 is located, there is a tendency for a component to be induced at right angle to E This is shown in FIG. 2,. where E represents the impressed electric field and E the induced electric field. Because of the space and phase differences between E and B the resultant field will, in general, assume an elliptical polarization, represented by the ellipse 21. While the effect of the presence of the gyromagnetic rod is to induce a component B this component will be unable to propagate independently in the tetrahedral guide until the electrical dimension of the guide in the x direction is greater than approximately one-half a wavelength. This will, in general, occur somedistance along the guide 11 away from section AA. It should be noted that while wave energy polarized in the y direction is incapable of propagating over an interval of the tetrahedral section, it is nevertheless preferred that the gyromagnetic material extend into the region wherein E is cut oif. This enables the spurious local radio frequency fringing fields generated by the discontinuity introduced by the end of rod 14 to decline to negligible amplitudes in the region of the wave path wherein both polarized waves can be supported.

As the wave energy propagates in the z direction there is a constant transfer of energy from the E polarization to the B polarization, until ultimately substantially all the energy leaves by way of guide 13 polarized in the y direction as shown by vector B in FIG. 1. Of particular interest, however, is the nature of the wave configuration in the region midway along guide 11 since it is the nature of the field pattern in this region which determines the uses to which the tetrahedral waveguide may be put.

The tetrahedral taper, and the various forms thereof, may be characterized, in general, as comprising:

(i) A transition between two transmission lines which are of the same kind, capable of supporting propagation in only a single mode, which possesses a unique polarization, and which are so oriented that the directions of polarization of the propagating modes in the two wave paths are mutually orthogonal;

'(ii) The transition possesses the symmetry of the geometrical point group D as defined by G. Herzberg in Infrared & Raman Spectra, at page 8;

(iii) The gyromagnetic medium (such as ferrite, magnetic garnet, parametric, or plasma) is magnetized by a steady magnetic field in the z direction (direction of propagation) In the case of ferrite or garnet it is usually convenient to have the material in the form of a rod symmetrically located along the path axis, but this is only one of many possible embodiments. For example, in the case of a plasma the medium may completely fill the tapered guide.

For the purposes of analysis a special or degenerate form of the tetrahedral waveguide is considered. As shown in FIG; 3, this comprises two abutting rectangular waveguides 31 and 32 coaxially aligned along a common longitudinal axis and rotated ninety degrees with respect to each other so as to be cross-polarized. The dimensions of the guides are proportioned to operate in the dominant mode in which the electric lines of force extend in a direction perpendicular to the wide guide walls. An element of gyromagnet-ic material 33 is symmetrically disposed at the junction of the two guides and, extends into each of them so as to interact with the radiation fields in both. The configuration shown in FIG. 3 is a special form of the tetrahedral structure of FIG. 1 in which the longitudinal dimension of the tapered section 11 is reduced to zero.

The solution of the scattering problem for the type of junction shown in FIG. 3 is given using a simplified model in which:

(1) The input and output guides are represented by uniform anisotropic media;

(2) The transition between the two mutually orthogonal polarizations takes place at the plane z=0 (at the butt joint);

(3) The radiation is in the form of plane waves;

(4) At the joint z=0, the radio frequency electric and magnetic fields are continuous;

(S) The presence of the gyromagnetic material is represented by assigning to the media a magnetic permeability in the form of Polders tensor for longitudinal magnetization. (See 011 the Theory of Ferromagnetic Resonance, by D. Polder, Philosophical Magazine, Volume 40, 1949, page 99.)

An analysis of the tetrahedral waveguide starts with Maxwells equations: 7

l a curl E at (1) 1 e5 curl H; a (2) wherein E is the electric field vector of the electromagnetic wave; 'H is the magnetic field vector of the electromagnetic wave; c is the velocity of light in free space;

E is the magnetic induction, and

E is the electric displacement.

Substituting ,uH for B To represent the gyromagnetic effect produced by rod 14- magnetized in the z direction, the components ,u and x of the Polder permeability tensor are substituted for F, i.e.,

p ix p.

0 0 [L Similarly, the property of a waveguide which possesses diiferent propagation characteristics for waves polarized in the x and y directions, respectively, is indicated by assigning a phenomenological tensor dielectric constant for; In diagonal form 60 (See A Phenomenological Theory of the Reggia-Spencer Phase Shifter by J. A. Weiss, Proceedings Institute of Radio Engineers, June 1959, page 1132.)

The two significant components of e are e and e For a rectangular guide which supports propagation in the dominant mode only, e is, in general, not equal to e With reference to the x-y-z coordinate system shown in FIG. 3, e for guide 31 is positive whereas e is negative. For guide 32, on the other hand, e is negative and is positive. Making the indicated substitutions for the tensor permeability and tensor permittivity and assuming that for plane-wave propagation in the :2 directions at angular frequency w, the field equations may be solved. The solution yields the propagation constants and states of polarization of the two normal modes, indicated by 1+ and which are given in terms of the magnetic parameters n and K, the dielectric coeificients e and e and the angular frequency as e= /2 (e -Fe and a= /2 G e It should be noted that the effective values of 6 and e themselves may be frequency-dependent, depending upon the type of Waveguide used. Also, in this simplified model, the effective values of ,u. and :c involve the shape and size of the ferrite or other gyromagnetic element as well as its intrinsic properties.

Substituting for e and 5 Equation 3 may be rewritten where realizable, however, only those waves that are will be considered in the discussion that fellows:

\ For the purposes of explanation, the butt joint configuration of FIG. 3 is schematically represented in FIG. 4. To further facilitate the explanation, the wave path is divided into four discrete regions, I, II, III and IV, each of which has distinctly different propagating characteristics. Region I of guide 31, for example, is a simple rectangular Waveguide, whose permeability and permittivity are real and constant. Region II in guide 3 1 and Region III in guide 32 are regions of uniform cross section wherein the permeability and permittivity are functions of the gyromagnetic material as well as of the waveguides. Region IV in guide 32 is comparable to region I, except that the direction of polarization of the wave energy supportable therein is rotated ninety degrees. Let us now examine the effects produced by the butt junction and the gyromagnetic material upon an incident wave, E of unit amplitude and zero phase angle, polarized in the x direction. This wave is indicated in FIG. 4 as being applied to guide 31 and is shown as a propagating wave in region I, the region of guide 31 without gyromagnetic material. Upon entering region II of guide 31 containing the gyromagnetic element 33, the propagation constant of the wave and its polarization undergo a change. The propagation constant for a propagating wave in this region is given as 7+ I\ b l"\ F' i xii yl] (5) where e in guide 31 is greater than zero and 6 is less than zero.

In general, since n 6 1 [e ][e y+=ii 3, a positive real number.

The propagation constant root of 7 i.e., 'y .=ij3, wave of constant amplitude direction, designated P}. It can be shown (see June 1959 Proceedings Institute of Radio Engineers, cited above) that this wave is elliptically polarized with its major axis in the x direction, as indicated in FIG. 4A. Since guide 31 cannot support polarization in the y direction, there is no continuous transfer'of energy from the x direction of polarization to the y direction of polarization, notwithstanding the presence of the gyromagnetic material tending to couple between the two polarizations. The P wave, instead, propagates along guide 31, maintaining a substantially constant amplitude and elliptical eccentricity. Upon reaching the junction of guides 31 and 32, (z=0), however, the incident Wave P is scattered into four waves: P a propagating reflected wave; N a decaying, nonpropagating wave in guide 31; N a decaying nonpropagating wave in guide 32; and P a propagating wave in guide 32.

Considering the waves in guide 3-1, the reflected wave P is of the same polarization and wavelength as the incident wave, the latter being given by the positive square root of 'y From Equation 5,

This wave, designated P propagates in the minus z direction, and, as shown in FIG. 4B, is elliptically polarized with its major axis in the x direction.

There is, in addition, 31, nonpropagating because it is elliptically polarized with its major axis along the y, or cut-oil? direction. The propagation constant of this wave, designated NJ, is obtained'by evaluating Equation 5 for the positive value of 'y which is a positive real number, that is,'

where 5 is given by the negative square indicates the presence of a propagating in the plus 2 "Y a positive real propagation constant represents a decaya nonpropagating wavein guide,

elliptically polarized, as shown, but with its major axis in the y direction.

Similar evaluations of 'y in guide 32 may be made, with the condition that e is less than zero, and 6 is greater than zero, and in guide 3-2 equals e in guide 31, and e in guide 32 equals e in guide 31. Such evaluations yield two waves of consequence. One is a decaying wave N} of elliptical polarization whose major axis, as shown in FIG. 4D, is in the x, or cut-oif direction, whereas the second is a propagating wave P3 of elliptical polarization whose major axis is along the y direction, as shown in FIG. 4B. These are represented in region III of FIG. 4 by the curves N3 and P3 The output wave is represented, in region IV, by E of amplitude E and phase 90 +6. The 90 degrees represents a time phase shift occurring at the junction :0, whereas 0 is the time phase shift over the regions II and III.

It is evident from the above that the solution of Maxwells equations for the butt joint structure of FIG. 3 comprises waves of two principal modes. In particular, as has been shown, there are four principal scattered waves, as represented in FIG. 4, which include the P wave, a propagating reflected wave in guide 31 the Ni} wave, an evanescent wave in guide 3-1;

the

P} wave, a propagating transmitted wave in guide 32 and the N wave, an evanescent wave in guide 32.

The amplitudes of these several waves at the butt junction (i.e., 2:0 plane) are given in terms of a, 4:,

and w as follows:

The scattering coetlicients are seen to possess qualitatively the same properties as are observed in the physical embodiments of the device, namely:

gyromagnetic medium is unma'gnetized junction. The inclcorresponds to the cross-sectional dimensions of the rectangular guide), the components of the permeability tensor and the frequency, there is full transmission of the incident wave and no reflection;

(5) Under other conditions involving the same parameters there is full reflection and no transmission except to the extent that in this case there is, in general, an evanescent, nonpropagating mode present at the output side of the junction. When the applied field is such that condition 4 occurs at some frequency, condition 5 will be found to occur at a lower frequency and condition 2 at a still lower frequency. The upper limit of the frequency range at which condition 4 occurs will be at the point at which the input guide becomes capable of supporting more than a single mode and the properties of the junction become drastically difierent (for example, the Faraday rotation effect makes its appearance and the evanescent waves disappear);

(6) The phase of Waves propagating through the junction in opposite directions differ by i.e., the device is a gyrator;

(7) In general, the state of polarization at the plane 2:0 is elliptical. It may, under given conditions, be circ'ular. Of particular importance, however, is the situation wherein the polarization is linear and at an angle of 45 to the principal axis of the anisotropic medium. Either circular and linear polarization can be made to occur together with condition 4, i.e., with full transmission.

Let us now consider under what conditions full transmission occurs with no reflection. This means that the refiected wave, 1; is zero which, from Equation 6 implies one or'more of the following:

b=is 10 c=is (11) The last of these need not be considered since this implies that 8:0, i.e., that the wave path is either circular or square with equal propagation for the E and the B polarizations. In accordance with the invention, however, the waveguides are not symmetric, and 6 is not equal to zero. For either of the other two conditions, full transmission occurs without further restrictions on the wave path cross-sectional dimensions. However, for each of these two cases transmision occurs in a different manner. For example, referring to Equations 7 and 9, it is noted that for s -z'c, both N and N3 vanish, whereas for s: ib they, in general, do not.

From the amplitudes of the scattered waves given by Equations 6 through 9, the values for E and E at the 2:0 plane can be obtained. They are given by For either b=is or c=is, the condition for no reflection, the amplitudes of E and E are equal and the electric field at 2:0 is elliptically polarized with its major axis inclined at an angle of forty-five degrees with respect to the directions of polarization in the guides. In particular,if c=is, the electric field is circularly polarized, which is the condition of operation contemplated by A. G. Fox in his copending application cited above. If, on the other hand, in accordance with the invention, b=is, and in addition 1i y s-ic E and 13,, are both equal and in time phase. As a consequence of this particular adjustment of the wave paths, the electric field at the 1:0 plane is linearly polarized at an angle of'forty-five degrees with respect to the directions of polarization of the wave energy in each of the waveguides. Furthermore, the polarization angle is a function of the direction of propagation, there being a ninety degree change in the direction of polarization at the junction as the direction of propagation is reversed. This is the condition required for the isolator and circulator applications contemplated by the present invention.

The explanation of the operation of the various embodiments of the invention using the tetrahedral taper of FIG. 1 that is to be given hereinafter is based upon the analysis made of the special form of the tetrahedral taper shown in FIG. 3. The butt joint configuration was analyzed rather than the tetrahedral taper for simplicity in that the tetrahedral taper, being a transmission line of constantly changing cross-sectional dimensions, is considerably more difficult to analyze mathematically than the butt joint junction which comprises two uniform crossed guides, especially when the tapered section is further loaded by means of a gyromagnetic substance. Nevertheless, the basic principle of operation is observed, in most respects, to be substantially the same in both devices. The dilferences are essentially difierences in detail and degree, which dilferences would determine which of the structures might be used in some particular application. This can best be illustrated by considering two applications in which one or the other of the two devices are used to best advantage.

For example, while the butt joint structure of FIG. 3 has been used essentially as a mathematical model for the purpose of analysis, it makes an excellent magnetically controlled reactive switch. As a switch, it possesses a very high insertion loss in the reflecting (nonmagnetized) state, low loss in the transmitting (magnetized) state (which is, in principle, lower than that attainable in any of the currently known ferrite-waveguide devices), high switching speed, broad bandwidth, and little sensitivity to variations in applied field and magnetic density of the gyromagnetic medium. In addition to its use as a switch, it may be used as a reversible gyrator, whose direction of phase shift is a function of the direction of polarization of the gyromagnetic element.

As a switch, the non-transmitting state is that state in which the coupling effect of the gyromagnetic material is nullified by demagnetizing it. In this non-cupling state the extent to which spurious transmission is suppressed is determined entirely by the degree of mechanical perfection of the junction. The simplest and most precise structure for this purpose is the butt joint in which all surfaces are either mutually parallel or perpendicular. It should also be noted that for this application the state of polarization at the joint is of little importance in its performance as a switch. On the other hand, in a resistance vane isolator, of the type to be described in greater detail hereinafter, the state of polarization of the wave energy is of great importance since it determines the forward-to-reverse attenuation ratio. Equally important, however, is the extent over which the desired polarization is obtainable. Obviously, the desired polarization must persist over a sufiiciently long longitudinal region of the device to permit the adequate absorption of the wave energy in the reverse direction by the resistive vane. -In this application, the butt joint junction would not be suitable since the polarization of interest exists only at the plane of the junction. On the other hand, in the tapered section, because the changes in the propagation conditions experienced by the wave energy take place more gradually, the consequences of these changes are not as abruptly distributed in space as they are in the butt joint section. Therefore, the tapered section of FIG. 1 would be more suitable for use as an isolator, and, in general, for all ap plications in which a particular state of polarization is to be maintained over an interval.

It should be again emphasized that the principle of operation is not affected by the gradual transition of the 10 waveguide dimensions in the taper since a wave which is not transformed from the initial state of polarization in the input guide to the preferred state of polarization in the output guide through the agency of the gyromagnetic medium ultimately must reach the condition of cutoff in essentially the same manner in the tapered guide as in the butt joint junction.

In designing a tapered section, however, certain restrictions must be observed. At the butt joint of FIG. 3, both the x and y polarizations are momentarily cut off at the junction. It has been similarly found that in the tapered section .11, there may be a finite longitudinal interval wherein both polarizations are cut off. If this region is made too long, however, there Will be substantially no transmission through the section. On the other hand, the overall length of the taper, and particularly the length of the taper wherein both orthogonally polarized waves can propagate, should not be greater than half a wavelength for the highest frequency to be transmitted therethrough. 'If made larger than half a wavelength, Faraday rotation elfects will be observed which will interfere with the operation of the tetrahedral section as contemplated by the invention. The gyromagnetic material should extend over an interval coextensive with that occupied by the evanescent waves NJ and N In addition, it should be noted that not only the metallic waveguide walls but also the gyromagnetic substance contribute to the effective tapering of the transmission line characteristics. This effect is especially important in those cases where the rod diameter is so large as to cause appreciable distortion of the distribution of radio frequency electromagnetic energy over the cross section of the wave path (dielectric waveguide effect). Hence, the achievement of the proper field distribution necessary to satisfy the conditions specified by Equations 10 and 15 may require that the gyromagnetic element and biasing field be tapered as well as the waveguide. In some applications, changes in the dimensions of the wav guide over the critical region (near the region of linear polarization) may be second order and can be neglected. However, for those applications in which even the slightest degree of ellipticity in the polarization is objectionable, particular attention must be given to the contribution to the effective tapering by the gyromagnetic material. In the latter class of devices it will be necessary to appropriately proportion the distribution of gyromagnetic material over the critical region.

Having defined the conditions necessary to produce linear polarization in the central region of the tetrahedral waveguide, it is now proposed to utilize the tetrahedral tapered Waveguide as a nonreciprocal waveguide component. In FIG. 5 there is shown a perspective View of the tetrahedral tapered section of the present invention connected and utilized to produce nonreciprocal transmission effects. The device of FIG. 5, in particular, is an attenuator utilizing the ninety degree difference in the direction of polarization occurring in the tetrahedral section to produce nonreciprocal operation.

Referring more specifically to FIG. 5, the isolator, in accordance with the invention, comprises the two coaxially aligned dominant mode rectangular waveguides 51 and 52 separated by the tetrahedral section 53. Waveguide 52 is rotated ninety degrees about the common axis with respect to guide 51 so that the directions of wave polarization in the two paths are mutually orthogonal.

Extending along the axis over an interval of the tapered section 53 is .a cylindrical rod of gyromagnetic material 54. Rod 54 is longitudinally biased by the steady biasing field H. Field H may be externally supplied in any of the usual ways well known in the art, or rod 54 may be permanently magnetized.

Longitudinally disposed along rod 54 are the resistive vanes 55 and 56 which extend substantially equal distances from the center of guide 53 toward guides 51 and 52. The vanes are located in a plane inclined at an angle of fomtydive degrees to the directions of polarization in guides 51 and 52, so as to absorb and dissipate waves having their plane of polarization parallel to the plane of vanes 55 and 56, but to pass substantially unaitected waves having their plane of polarization perpendicular to the plane of the vanes.

Designating, for the purposes of explanation, the direction from a to b, the forward direction, and from b to a, the reverse direction, the operation of the isolator of FIG. may be explained by first considering a wave traveling from a to b. This wave is horizontally polarized in guide 52. With the tetrahedral section adjusted to satisfy the conditions specified by Equations and 15, the wave will be linearly polarized in the region of the resistive vanes but at an angle of forty-five degrees relative to its initial direction of polarization, and at an angle of ninety degrees with respect to the plane of the resistive vanes. So oriented, the wave will pass through the tapered section unafi'ected and leave by way of guide 51. In the reverse direction from b to a, however, the plane of polarization of the wave will be rotated into coincidence with the plane of the lossy vanes. Thus, energy propagating through the tapered section in the reverse direction Will be substantially attenuated by virtue of the presence of the lossy vanes.

Other, non-lossy selective coupling means may be utilized in conjunction with the tetrahedral taper to produce other nonreciprocal effects. For example, in FIG. 6, there is shown a three port circulator comprising the waveguides 61, 62, the tapered section 63 and the biased gyromagnetic element 64, arranged in a manner similar to that shown in FIG. 5. Instead of the lossy vanes, however, there is a'probe 69 extending into the tapered guide. Probe 69 is centrally located along the tapered section (in the region wherein the wave polarization is substantially linear), and is oriented to be parallel to said polarization for waves propagating through the taper for one direction of propagation, but to be normal to the direction of polarization for waves propagating in the reverse direction.

Energy intercepted by the probe 69 is coupled to the third waveguide 65 by way of coaxial lead 67. Guide 65 is shown as terminated at one end by shorting plane 70. Coupling to guide 65 is therefore made in a manner to induce wave propagation away from end 70 in accordance with techniques well known in the art.

In operation, the plane of polarization of wave energy entering section 63 from guide 62 is rotated into coincidence with the probe 69. Energy so polarized is thereby coupled to guide 65 and hence out through port b, with none reaching guide 61. Wave energy introduced into guide 65, on the other hand, is coupled into section 63 by way of probe 69 and propagates in the direction of guide 61, with none reaching guide 62.. Finally, energy introduced from guide 61 reaches the coupling region polarized normal to the preferred direction of coupling to probe 69. Consequently, substantially none of the energy is coupled to guide 65, and all is propagated through to guide 62. Thus, the sequence of propagation is a b, b c, and c n.

FIG. 7 shows a four port circulator using two tetrahedral tapers 71 and 72, coupled by means of the coaxial connection 73 and the associated electrostatic probes 74 and 75. As before, the probes are inclined at an angle of forty-five degrees to correspond to the direction of polarization of Wave energy propagating in one direction through the respective tapered sections and to be normal to the direction of polarization for propagation in the reverse direction. So arranged, transmission takes place from a 11, b a, ca, and da in typical circulator fashion. v

FIG. 8, alternately shows a four-port circulator using a single tetrahedral tapered section 81. In this'embodiment the two branch guides 82 and 83 are directly coupled to the tetrahedral by means of the apertures 84 and 85, respectively. The apertures are situated in adjacent corners of the tapered section 81 in the region of linear polarization. For one direction of propagation, energy is coupled through aperture 84 to guide 82, whereas for propagation through the taper in the reverse direction, energy is coupled through aperture 85 to guide 83. The shift in coupling, as explained above, is due to the ninety degree change in the direction of polarization of the wave energy resulting from the change in the direction of propagation through the tapered section.

It should be noted that while the tapered sections shown in the several illustrative embodiments of the invention comprise sections of conduct-ively bounded waveguide, the taper may be composed of finline or other types of waveguiding structures without in any way detracting from the effectiveness of .its operation. In addition, the taper need not be linear but may vary in any prescribed manner in accordance withthe requirements of the particular application. Similarly, the use of coaxial connections between the several components comprising the three and four-port circulators of FIGS. 6 and 7 was merely illustrative. Other types of suitable coupling means, well known in the art, may be used in their place equally as well.

An illustration of the use of other types of waveguiding structures to practice the invention is given in FIG. 9 where the equivalent of the butt joint structure of FIG. 3 is shown using coaxial cable, In particular, there are shown two abutting sections of cable and 91 whose longitudinal axes are transversely displaced a given distance with respect to each other. The coaxial cable 90 comprises the inner cylindrical conductor 92 separated from a coaxially disposed outer cylindrical conductor 93 by a suitable dielectric material 94. Similarly, cable 91 comprises an inner cylindrical conductor 95 and an outer coaxial cylindrical conductor 96 separated by the dielectrio material 97.

Located in the overlapping regions A and B, common to both coaxial cables, and extending a distance into each, are the two gyromagnetic elements 98 and 99. Each of these elements is longitudinally biased by means of a steady magnetic field of amplitude H with the direction of the biasing field in element 98 reversed with respect to the direction of the biasing field in element 99.

FIG. 10 shows in greater detail the arrangement of the two cables and, in particular, the magnetic field distribution in the common regions A and B bounded by the inner conductors 92 and 95 and the outer conductors 93 and 96. In these two regions the radio frequency magnetic fields (which, in general, consist of closed loops of magnetic fiux surrounding the inner conductor of each coaxial cable) intersect. Illustrative of these intersecting fields are the intersecting magnetic field components, i and E and i and 1 shown in the two common regions, A and B, respectively. For a particular transverse displacement d of the cable axes, the intersecting radio frequency fields can be made to be orthogonal over the regions A and B, thereby precluding any coupling of wave energy between the two abutting sections of cable. If, now, however, the two magnetized elements 98 and 99 of gyrornagnetic material are inserted into the common regions A and B of the butt junction, coupling between the two orthogonal field configurations is made possible in a manner analogous to that obtained in the butt joint junction of FIG. 3. Thus, the junction of coaxial lines is in every essential respect equivalent to the waveguide version described above and shown in FIG. 3 and as such is suitable for use as a switch or gyrator. It should be noted that its ability to function as a switch or gyrator does not depend upon the establishment of a unique mode of polarization of the wave energy at the 13 junction. In addition, this structure has the advantage that by operating in the TEM mode it possesses no cutoff and can therefore serve as a practical embodiment of the butt joint switch and gyrator at frequencies below those conveniently utilized in waveguide structures.

However, it is equally important to note that by complying with the conditions set forth in Equations and 15, linear polarization of the wave energy may be obtained at the junction, and utilized to produce nonreciprocal transmission effects, as was explained hereinbefore.

Of particular interest in all of these devices is the steady biasing field required to operate the several embodiments of the invention. As was pointed out earlier, resonance and field displacement gyromagnetic microwave components require biasing fields which are proportional to the operating frequencies, whereas devices relying on Faraday rotational efiects are sensitive to variations in frequency and temperature. The tetrahedral junction, on the other hand, operates satisfactorily over a range of biasing fields, which range is substantially independent of the operating frequency and ambient temperature and includes biasing fields below the saturation level of the gyromagnetic material. This property of the tetrahedral junction greatly facilitates the magnetic biasing problem. In particular, it permits the principles of the invention to be applied to devices operating at very high frequencies (e.g., the millimeter wave range) without requiring extensive and awkward magnetizing facilities. Furthermore, at any frequency, it makes possible the design of conveniently small, lightweight units. In applications which involve the changing of the applied magnetic field, it permits such changes to be performed at high speeds because of the small inductance of the magnetizing solenoid and absence of a ferromagnetic yoke.

In all cases it is understood that the above-described arrangements are illustrative of a small number of the many possible specific embodiments which can represent applications of the principles of the invention. Numerous and varied other arrangements can readily be devised in accordance with these principles by those skilled in the art without departing from the spirit and scope of the invention.

What is claimed is:

'1. In an electromagnetic wave transmission path adapted for propagating wave energy in opposite longitudinal directions, a first region of said path supportive of wave energy polarized in a first direction, a second region of said path supportive of wave energy polarized in a second direction orthogonal to said first direction, means for coupling said wave energy from said first region to said second region comprising an element of longitudinally biased gyromagnetic material, said gyromagnetic element and said first and said second transmission regions proportioned to induce a linearly polarized wave in an inter- -val of said path between said first and said second regions whose direction of polarization differs by ninety degrees for opposite longitudinal directions of propagation along said path.

2. In an electromagnetic wave transmission system, a first wave path supportive of wave energy polarized in a first direction, a second wave path supportive of wave energy polarized in a second direction orthogonal to said first direction, means for coupling wave energy from said first path to said second path comprising an element of gyromagnetic material magnetically biased in the direction of wave propagation, said paths and said biased gyromagnetic element proportioned to induce a linearly polarized wave in a region between said paths whose direction of polarization is inclined at an angle of fortyfive degrees with respect to the directions of polarization of said wave energy in said first and second paths.

3. The combination according to claim 2 wherein said tioned such that where e= /2(e +e and 5=%(e s f0! which e and e are the transverse components of the effective tensor dielectric constant of the wave paths and u and K are the components .of the Polder permeability tensor.

4. In an electromagnetic wave transmission system, a conductively bounded waveguiding structure that varies smoothly and continuously from a transverse cross section of a given shape at a first longitudinal location to a transverse cross section of a different shape at a second longitudinal location, an elongated member of magnetically polarizable material capable of exhibiting gyroT magnetic properties at the operating frequency of said system extending longitudinally within said structure between said locations, a magnetizing field applied longitudinally to said member, coupling means located between said first and said second locations for coupling to wave energy propagating through said guide in one direction to a substantially different degree than to wave energy propagating through said guide in the reverse direction, and means for varying the amplitude and direction of said magnetizing field.

5. The combination according to claim 4 wherein the cross-sectional dimensions of said waveguiding structure and the parameters of said gyromagnetic member are adjusted to produce a linearly polarized wave in a region of said guide substantially midway between said first and said second locations.

6. The combination according to claim 5 wherein said coupling means comprises a vane of lossy material lying and in a plane parallel to the direction of said linear polarization for waves propagating through said guide in the reverse direction.

7. The combination according to claim 5 wherein said coupling means comprises an electrostatic probe extending in a direction parallel to the direction of said linear polarization for wave energy propagating in said one direction.

8. A four-port circulator comprising a pair of electromagnetically coupled tapered waveguide sections each supportive of wave energy having a first sense of polarization at one end thereof and a second sense of polarization at the other end thereof orthogonal to said first sense, means for coupling wave energy from said first sense of polarization to said second sense of polarization in each of said sections comprising a magnetically biased element of gyromagnetic material, said sections and said elements proportioned to induce linearly polarized wave energy in a region within said section between said one end and said other end inclined at an angle of forty-five degrees to said first sense of polarization and said second sense of polarization, and means for coupling between said sections located in said regions of induced lineraly polarized wave energy.

9. An electromagnetic wave transmission path adapted for propagating wave energy in respectively opposite longitudinal directions whose cross-sectional dimensions vary continuouslyfrom a first transverse cross section supportive of wave energy polarized in a first sense, to a second transverse cross section supportive of wave energy polarized in a second sense orthogonal to said first sense,

said linearly polarized Wave in said region in which said second sense of said linearly polarized wave in said'region in which said direction of polarization is different for the opposite direction of propagation.

10. In an electromagnetic wave system first and second sections of coaxial cable each comprising an inner conductor surrounded by a coaxially disposed outer conductor, said sections having the cofacing ends thereof abutting upon each other with the longitudinal axis of said first section transversely displaced with respect to the longitudinal axis of said second section defining a pair of regions common to both said cables bounded by said inner conductors and said outer conductors, at least one element of gyronragnetic material disposed Within one of said regions and means for magnetically biasing said element.

11. In an electromagnetic wave system first and second sections of coaxial cable each comprising an inner conductor surrounded by a coaxially disposed outer conductor proportioned to support wave energy in the TEM mode, said wave energy as supported in each'of said sections having circular magnetic field components coaxially distributed over the interval between said inner and said outer conductor, said sections having the cotacing ends thereof abutting upon each other with the longitudinal axis of said first section transversely displaced with respect to the longitudinal axis of said second section defining first and second overlapping regions of magnetic 'field hav- 16 ing mutually orthogonal field components, an element of gyromagnetic material disposed within at least one of said regions and means for longitudinally biasing said element.

12, The combination. according to claim 11 wherein a gyromagnetic element is disposed within each of said regions and wherein said gyrornagnetic elements are biased in mutually opposite directions.

13. In combination first and second sections of rectangular waveguide coaxially disposed along a common longitudinal axis,vthe cofacing ends of said waveguides abutting upon each other to form a butt joint, said second guide being rotated ninety degrees about said axis with respect to said first guide, an element of gyromagnetic material extending from said first guide through said joint into said second guide, a source of magnetic field for biasing said element in a direction parallel to said axis and means for varying the amplitude and sense of said biasing field.

14. The combination according to claim 13 wherein each of said Waveguides is supportive of electromagnetic wave energy in only one sense of polarization.

References Cited in the file of this patent UNITED STATES PATENTS

Images