engg approach to design of tapered dielectric rod horn ant

7/27/2019 Engg Approach to Design of Tapered Dielectric Rod Horn Ant

1/9

U.D.C. 621.396.677.75

EngineeringApproach to theDesign of TaperedDielectric-rod andHorn AntennasJ. R. JAMES,B.Sc, Ph.D., A.F.I.M.A., C.Eng., M.I.E.R.E.*

SUMMARYThe taper profile of optimized dielectric-rod andhorn antennas is synthesized as a series of non-interacting planar radiating apertures. The method issemi-empirical, straightforward to apply, enablesthe dielectric-rod antenna to be satisfactorilyoptimized and provides a means of evaluating andoptimizing a dielectric-horn antenna with variablewa ll thickness. The optim um profiles are takenas those which smoothly transform the surface-wavepower from the launcher to the radiatingaperture. The optimization of the dielectric-rodantenna considerably improves the radiation patternwhile computations supported by measurementsconfirm earlier reports that a dielectric-horn canhave a higher gain than a metal horn of similardimension but side-lobe level is seen to be animportant issue. Wide flare-angle horns give idealE-plane patterns at the expense of a high side-lobelevel in the H-plane; for small flare angles the dielectrichorn gives similar patterns to the tapered rodantenna and thus preserves rotational symmetry.Calculations throughout are restricted to cylindricalgeometry but other geometries and variations on thedielectric -horn principles are described. Usefulengineering design data have been compiledfor both the dielectric-horn and rod antennas andcurves are given which determine near-optimumparameters for gains up to about 20 dB whichis seen to be a practical operating limit for thesesurface wave devices. A unified impression ofdielectric antennas emerges with the importantconclusion that, when optimized, dielectric-rod andhorn antennas are in fact competitive wth smallmetal horns for some a pplications ; furthermore thedielectric-horn antenna, used singly or in arrays, is anideal device for producing a low side-lobe levelin the E-plane.

* Department of Electrical and Electronic Eng ineering, RoyalMilitary College of Science, Shrivenham, Swindon, Wiltshire.

1. IntroductionThe dielectric-rod antenna (Fig. l(a)) has beenextensively studied in recent years 1"5 and has proved tobe a useful device in certain specialized applications;the dielectric-tube antenna has also received someattention1 '6 and is found to be similar in performanceto the uniform dielectric-rod antenna. The dielectric-hornantenna also shown in Fig. l(a) is perhaps the least knownof all dielectric antennas and little has been establishedabout it apart from the early re po rts 1 ' 7th at its gain appearsto exceed that of a metal horn of similar dimensions andthis could be a significant property. Dielectric antennasbring about some saving in weight, have good sealing andcorrosion properties and are now attracting renewedinterest in view of the ease with which dielectrics can bemanufactured with present-day techniques. The theo-retical solution of these antennas is difficult because theradiation does not emanate from a simple plane apertureas in the metallic horn case; many isolated aspects of theproblem have been successfully analysed26 but until theoverall radiating system can be treated as a whole theoptimization of these antennas rests largely with theexperimenter. In this paper an engineering approach tothis problem is adopted whereby theoretical techniquesare supplemented by experimental information. Opti-mized versions of these antennas are then comparedwith one another and also with the conventional metalhorn.

Perspex

0-4-Xc

(a) Dielectric-horn antenna with uniform wall thickness (Kiely 1)and dielectric-rod antenna with linear taper.Metalwaveguide

Launcheriaperture ' S tep iaperture i apertureStep 2 Step 1 i Rod endaperture j aperture(b) Approximating the dielectric-rod profile by a series of steps,each one assumed to produce a planar aperture of infinite extent.

Fig. 1.The investigation concentrates on tapered dielectric-rod and horn antennas and as little is known about thelatter the paper is concerned with their evaluation. Thedielectric-horn antenna is certainly no less involved adevice than the tapered dielectric-rod antenna for which

no accurate design data exists; the difficulty lies inThe Radio and Electronic Engineer, Vol. 42, No. 6, June 1972 251


2/9

J. R. JAMES

calculating the radiation from the tapered structure to asufficient accuracy and much the same problem occurs fordielectric-horn antenna s. We describe a semi-empiricalapproach which assists in the selection of an optimumtaper profile for the dielectric-rod antenna and usefuldesign curves are obtained as a consequence. Thedielectric-horn antenna may be calculated in a similarmanner by synthesizing its profile by a system of shortconcentric dielectric tubes; the predominant radiationcharacteristics are, however, exhibited by theradiation from the end aperture alone and we evaluatethe dielectric-horn on this simpler basis. Suitableapplications for the antenna are pointed out and theessential design parameters identified. Calculationsare confined to cylindrical geometry for simplicity and thetreatment of other cross-sectional shapes follows in asimilar manner but is computationally involved.2. Semi-empirical Approach to Tapered

Dielectric-rod AntennasThe diameter of the dielectric-rod is tapered alongthe antenna length to obtain a large efficient launchingaperture in conjunction with a large radiating apertureat the rod end.4 Unfortunately radiation leaks outalong the taper and it is shown below that this radiationdoes not contribute usefully to the radiation pattern.The main design problem is therefore to select a taperprofile which radiates least power and no exact means ofcalculation appears to exist. The writer has previouslyoutlined9 the following step-synthesis procedurewhereby a step discontinuity on the rod is regarded as aradiating planar aperture of infinite extent in the planeof the discontinuity; with this approximation a tapereddielectric-rod antenna whose profile is assumed to be aseries of m small non-interacting steps, may be treated asa system of m radiating planar apertures (Fig. l(b)).The radiated E-field is given by the sum of theradiation from each aperture, thusE = - ' 4n R ;= o

(n x H,)] e "- B l da (1)where the far-field vector Kirchhoff integral is employedwith the notation of Silver12 and (E i? H,-) is the aperturefield of the ith aperture. The aperture f ield is taken as theunperturbed surface wave on each rod segment and forthe diameters and dielectric constant of interest onlythe dominant HE1X wave exists (see, for instance, Fig.8.6 on page 72 of ref. 13). K t for i = 1, 2, 3, . . . , m, isa factor which defines the proportion of incident powerradiated at each step. The launcher radiation correspond-ing to / = 0 is obtained approximately from a chopped-surface-wave-distribution as described in ref. 4. Theradiation lost at each step can be estimated by theLorentz reciprocity theorem 13 but in the present author'sexperience the resulting radiation pattern predictionsare still too inaccurate to be useful; this is attributedto the fact that reflected waves at each step are neglectedas is the interstep radiation coupling.9 It was thoughtthat a better approximation would result by taking theaperture field as the difference between the unperturbedfield s on either side of the step but po or results were again

obtained. Eventually experiments indicated that theleakage of radiation along a smooth monotonicallydecreasing taper was to a large extent governed by therate of change of power flow ratios along the taper andthe following empirical formula>RAD

7i (2)wherewas found to give some approximate functional descrip-tion of the radiating process. P* AD is the power lost atthe rth step and P\ and P? are respectively the surface wavepower flowing in the interior and exterior regions of theith segment, / = 1, 2, 3, . . . , m; W is an empiricalconstant which depends on rod geometry and dielectricconstant and is to be determined by curve fitting thecomputed to the measured radiation patterns. From aphysical standpoint fFmay be interpreted as a parameterembodying the complex effects which arise when thesmall steps are spaced closer together.

- 2 0

- 3 0

\

1

\w4-8 X

Metal guide0 -3 3 \ o

2 %Power radiatedat step 1%

End aperture radiation 78-3o'Launcher rad iat ion 12-8


3/9

TAPERED DIELECTRIC-ROD AND HORN ANTENNAS

o


4/9

J. R. JAMES

2.2. Design Curves fo r Tapered Dielectric-rod AntennasThe curves in Fig. 4 are based on the radiating proper-ties established abo ve and it is assumed that th e o ptim um

taper profile is one for which the ratio yi+1 /yi inequatio n (2) is consta nt from step to step. With thiscriterion the taper profile is immediately defined oncethe terminal rod diameters are stated. The curvesembrace a useful range of antenna performances up tothe above mentioned gain threshold of 20 dB below wh ichwe have been able to optimize the tapers. The launche rradius is taken as 0-4/l0 and we assume that the rod willfill the latter in which case the launcher radiation isknown to be about 10% of the incident metal guidepower; 4 the curves are based on this figure and theremaining power is assumed to emanate from the endapertu re. The side-lobe can be improved by launchingfrom a larger aperture or alternatively utilizing a ringsource launch er.13 There are many other techniques forincreasing the launching efficiency such as placing aflange or choke 1 7 around the metal waveguide aperturebut all improvements necessitate additional metalliccomponents, increase the space occupied by the launcherand in some cases narrow the operating bandwidth.Practical system constraints will determine whether ornot the launcher efficiency is to be optimized beyond thatof the simple metal cup arrangement considered here(Fig. l(a)); antenna gain and beamwidth will not varyappreciably with more efficient launching but the per-formance figures for side-lobe level in Fig. 4 are to beregarded as a conservative assessment of what can beachieved.

To demonstrate the use of Fig. 4 suppose that ananten na gain of 15 dB is required. A rod terminal radiusof O-223AO, 1st side-lobe level of - 7 - 3 dB , beamwid th24 and rod length of 4-4A0 are given by the lower setof curves. To obtain the taper profile mark the powerratio curve at points x and y that correspond to thelauncher and terminal radii of 0-4A0 an d O-223vlo. In te r-mediate radii along the taper are then defined by pro-jecting the line xy on to the abscissa via the powerratio curve (solid line in uppe r graph). To dem onstratethis further th e line xy already draw n in Fig. 4 correspond sto the terminal radii of the antenn a in Fig. 3 and th e poi ntz when projected to q via p predicts an optimum radiusof 0-21A0 at a distance of Xo along the taper as opposedto the value of 0-224i0 obtained by curve fitting. This isin good agreemen t considering the dielectric constan t andlauncher radius do not correspond exactly to the con-ditions of Fig. 4; similar agreement is obtained forother points along this taper.Since the dielectric-rod antenna possesses radiationpatterns with rotational symmetry,4 the 1st side-lobelevel quoted may be taken with good approximation asthe value for either the E or H plan e. It is believed tha tthe curves are accurate enough for most design purposesand certainly much more informative than existing databased on a simplified sin x/x calculation.2 A completeantenna design must embody the bandwidth and matchingaspects of the launcher feed but our experiments haveshown 27 that the radiation patterns are not sensitive toconditions within the launcher provided the latter is not

heavily over-moded; for the launcher radius consideredhere it is easy to maintain the H n mode and the feeddesign can be pursued independently.3. The Dielectric-horn Antenna

Early references to the dielectric-horn antenna areKiely1 (1953) and Prochazka7 (1959). These experimentswere confined to horns with uniform wall thickness andcalculations based on the assumed simplified radiationmechanism gave some agreement with measurements7bu t were generally insufficiently repres entati ve; as far as isknown no attempt has previously been made to investi-gate the effect of ho rn wall thickn ess. The presentimproved knowledge of dielectric-rod and also dielectrictube a n te nna s 6 ' 1 7 now enables the dielectric-hornantenna with or without varying wall thickness to beevaluated in the following simplified way: the horn isregarded as a system of short concentric n on-overlappingdielectric tubes with various external and internal radiib an d a respectively, which approximate to the actualhorn dimensions in a similar fashion to the synthesis ofthe stepped dielectric-rod profile (Fig. l(b)). Experim entsshow th at the ho rn could b e justifiably represented inthis way and furthermore the radiating system closelyparallels the situation for the tapered rod antenna in thatlow side-lobe patterns corresponded to negligibleradiation from the ho rn conical surface. From anevaluation standpoint it is therefore sufficient to consideronly the radiation from the launcher and horn mouthand the subsequent computations carry this simplifica-

Fig. 5. Values of /? for HEn mode on dielectric tube. e r = 2-2.254 The Radio and Electronic Engineer, Vol. 42, No. 6


5/9


tion. The hor n mou th is taken as a radiating dielectrictube and since a detailed study of the dielectric-tubeantenna has already been described elsewhere 1 ' 6 ' 1 6we emphasize only those details that are relevant to thehorn situation.3.1. Surface-wave Modes on a Dielectric Tube

One physical interpretation of the tapered dielectric-rod antenna is that the launcher field is progressivelytransformed along the taper until the surface-wavewavelength approximately matches that of free space.This view may be extended to the d ielectric-horn antenn aand to illustrate this further we have computed a rangeof wavelengths for the dominant hybrid mode ondielectric tubes whose dimensions are relevant to thehorn situation. In Fig. 5 values of the phase constant /?normalized by k, are plotted as a function of tube radiusand thickness for sr = 2-2 since most of our experimentsuse paraffin wax . /?/ = /L0M where Xo an d Xare the free-space and surface-wave wavelengthsrespectively. Taki ng the simplified step ped -tub e mo delfor the horn we see that the horn wall thicknessmust taper to a very small value at its mouth if theterminal radius is to be large whereas at the launcher endthe wall needs to be very thick for good launchin g. Inthe designs described below we take c{ = alb) 0 at thelauncher in which case the launchin g efficiency from ametal cup as in Fig. l(a) should be at least as good asfor the dielectric rod case. It was conjectured th at theefficiency would be higher since some rays emanatingfrom the launcher would be diffracted forward by a wideflare angle dielectric horn and evidence of this effect isapparent in the experimental results below.A second point of interest is whether it is reasonablyvalid to assume that only the H E ^ mod e need be con -sidered on each elemental tube in this simplified model.To examine this, the cut-off dimensions for the nextfour higher modes on a dielectric-tube are plotted as afunction of tube radius and thickness in Fig. 6; alsosuperimposed in this figure is a horn profile which istypical of those used in the subsequent experiments. Wesee that only the symmetric modes and in particularthe H o i, can exist on any of the elemental tubes in additio nto the H E n mode but with careful launching from theH l t mode in the dielectric-filled waveguide thesesymmetric modes should not be excited. For the casesconsidered below there is no evidence that symmetricmodes are being excited and the assumption that onlythe H E ^ mode need be considered is justified.3.2. Theoretical Evaluation of the Dielectric-hornAntenna

The radiation calculation is carried out in a similarmanner to that for the dielectric-rod antenna but themanipulations are necessarily more extensive due to theadditional boundary condition; the salient details aresummarized as follows. The launcher radiation is calcu-lated from the unperturbed H n mode in the latter,as was done in equation (1), and the launchingefficiency is again based on a chopped-surface-wavedistribution. 4 The horn mouth is regarded as a dielectric

tube aperture for the purposes of calculating theradiation pattern; the relevant ft is given in Fig. 5 andthe unperturbed HE 1X mode is taken as the aperturefield. Using equation (1) bu t with m = 0 and 1 only,since we are assuming that an optimum non -radiatin gtaper profile exists, the radiation pattern resolves intofunctions of Lommel integrals as in the rod case; 4these equations are given in detail elsewhere. 17 As theantenna has a pencil-type beam, the gain can be readilycalculated from E- and H-plane radiation patterns. 1 8

0 -9

0 -3

Fig. 6. Cut-off conditions of the next four modes above the HEmode. er = 2-26, profile of typical dielectric ho rn.

Since the dielectric-horn antenna would appear to be anatural replacement for metal horns we will use the latterfor the purpo ses of com parison. The radiatio n pattern sof the metal horn have been obtained by applying thevector Kirchhoff integral12 to the unperturbed H nmode in the horn mouth; other modes have beenneglected but the results agree well with measurementsand geometric optics calculations1 9 in the main loberegion which is of interest here.

Tn Fig. 7 the gain of an optimized dielectric-hornantenna is plotted as a function of the radius and wallthickness of the horn mouth; also given is the gain of ametal horn of identical mouth radius b. If a horn taperprofile can be selected which is optimum, i.e. radiationis emitted solely from the launcher and horn mouth,then these curves show that the dielectric ho rn can alwaysbe arranged to have a higher gain than a metal horn ofidentical mouth dimensions by tapering the wall thick-ness to a very small value. This is in agreemen t with th eearlier observations.1June 1972 255


6/9

J. Ft. JAMES

3 0

^= o> ^ ? ^

~~"~" 0-9 9

- _ . 0 - 9 7

, 0 - 9 5 ^ ^ ^ " ^

_ ab10

Fig. 7. Antenna power gain computed using the vector Kirchhoffintegral. cylindrical metal horn aperture radius b; di-electric horn, er = 2-2, launcher radius = 0-4Ao, launcher loss =1 0 %Unfortunately antenna gain is not the only systemparameter of interest and the computations of the firstside-lobe level (Fig. 8) show that the dielectric-horn issuperior to the metal horn in the E -plane but considerablyinferior in the H-plane; the high side-lobe level in the H-plane is not due to interference with the launcher sinceradiation from the latter is deliberately omitted in thesecurves. The explanation is that the E-field is concentratedin the dielectric walls, thus the field distribution tends toresemble two vertical line sources and creates aninterference pattern in the H-plan e; however this fielddistribution appears to be ideal as regards the E-plane.The inference from Figs. 7 and 8 is that a superior antennato the metal horn can be obtained if the dielectric hornterminal radius is large and the wall very thin but this

Fig. 8. First side-lobe level relative to main-lobe amplitudecomputed using the vector Kirchhoff formula. metal hornaperture radius b, dielectric horn aperture, e r = 2-2. Insetsketch shows concentration of E field in the dielectric tube w alls.

depends on whether a non-radiating taper profile can befound.3.3. Experimental Substantiation of Dielectric-horn

Antenna DesignAfter much experimenting it was found that the situa-tion showed little departure from the dielectric rod casediscussed above: that is, horn profiles that producednegligible radiation along their length could be foundfor antenna gains below abo ut 20 dB while gains inexcess of this fig ure are difficult to achieve. Similarly,radiation emanating from the taper increased the overallside-lobe level and did not contribute usefully to theradiation pattern s. This immediately constrains the wallthickness in Fig. 7 to values of c < 0-93 that producevery high H-plane side-lobe levels in Fig. 8 although theE-plane pattern continues to have a low side-lobe level.The H-plane side-lobe level can be improved by reducingthe dielectric horn terminal diameter until the horn haszero flare angle.

H-PLANE

- 2 0

- 3 0

\\\\\ /1/i

v\\ \\ \\\\\\\

.. AV 3-3X o E-PLANE$ = 90

3 0 9 0

Fig. 9. Radiation patterns of w ide flare angle dielectric h orn.er = 2-2 (paraffin wax), Ao = 3-2 cm, c = 0-95; measured,computed assuming zero taper radiation loss and 12-8%launcher loss.Several dielectric horns have been tested 23 which bearout these theoretical predictions and it is sufficient hereto show three cases. In Fig. 9 the horn was too short for

the given terminal wall diameter and thickness; it isconsidered that appreciable radiation occurs along thehorn taper and this is responsible for the disagreementbetween the measured and computed patterns. Ofparticular interest here is the high side-lobe level in theH-plane as opposed to the low level in the E-plane andthe fact that the contribution from the taper spoils theradiation characteristics in both planes. On reducingthe flare angle and thickening the horn walls (Fig. 10),less radiation is lost along the horn taper and themeasured results tend more to the computed values.Finally, Fig. 11 (a) shows the patterns of a zero flareangle dielectric horn; the radiation along the horn taperhas been considerably reduced and there is a low H-plane side-lobe level. The E-plane pattern was verysimilar thus exhibiting rotational symmetry of thebeam. In fact this zero flare angle horn gives similarresults to the linearly-tapered dielectric-rod antenna256 The Radio and Electronic Engineer. Vol. 42. No. 6


7/9


- 2 0

a> - 1 0

- 2 0

- 3 0

\\yVvV

H-PLANE(6 = 0

\V\\\\y

Jor = 0.(b) Measured radiation patterns of cylindrical metal horn.Measured gain = 18-5 dB ; E-plane, H-plane.

to a conventional cylindrical metal horn and comparisonof Fig. ll(b) with Fig. 3 emphasizes the competitiveperformance of these dielectric antennas which can bebrought about by optimizing the various taper profiles.If however a narrow beam with low side-lobe level isrequired in the E-plane only, the dielectric horn iswithout doubt more suitable than either the dielectric-rod or cylindrical metal horn antenna. From theexperiments shown here it appears that a dielectricantenna will generally be 50 % longer than a metal hornof similar performance; the physical diameter will ofcourse be much less. Similar results hold for dielectricantennas and metal horns in rectangular geometry.

Fig. 12. Fraction of surface wave power within horn boundarybased on step model approxim ation. Launcher and horn mouthradii correspond to point x and y; c values on line xy as a functionof radius b define the near optimum profile, er = 2-2.3.4 Optimization of the Dielectric-horn Taper

In principle the dielectric horn could be flared out ina variety of ways but here we have confined our studyto flares with linear exterior walls in which case onlythe interior wall profile can be optimized to minimizethe taper radiation losses. Tests have shown that as forthe dielectric rod case, the fraction of surface wave powerwithin the horn antenna exterior boundary wall as afunction of radius b gives some indication as to how tochoose the inside wall profile; these curves are given inFig. 12, and define the inner wall profile as follows:the values of b and c that correspond to the horndimensions at the launcher and mouth are marked onthe curves at x and y; c values on the line xy thendefine the inner wall radius since b varies linearly alongthe horn length. Wefind however that the profile differslittle from a straight line as the sketch in Fig. 12 showsand since a linear wall flare is easier to construct it hasbeen adopted in most of our designs; the radiation alongJune 1972 257


8/9

J. R. JAMES

- 1 0 \\

3 0

0-7\,

Fig. 13. E-plane radiation patterns of long wide flare angle horn.sr = 2-2 (paraffin wax), Xo = 3-2cm, c = 0-96; measured,computed with gain 24 dB assuming no rad iation from ordissipative losses in the taper and 12-8% launcher loss. = 90.the horn taper is then controlled by the terminal wallthickness and horn length. To design a zero flare anglehorn profile the line xy becomes parallel to the ordinateaxis and the c values vary in a smoother monotonicfashion giving a concave profile as in the rod case.

An example of the type of E-plane pattern that can beobtained from a long horn with linear walls is shown inFig. 13. The radiatio n along the length is not negligiblebut is reduced to well below the launcher radiation leveland computations agree well with measurements; furtheroptimization of this horn was not attempted because ofmachining difficulties. The possibility of using an array ofsuch horns suggests itself but grating lobe constraintsneed to be considered. It was anticipated that a wideflare-ang le horn would collect some of the launcherradiation and for the wide flare angles used here someweak evidence exists in this respect since the computedside-lobe levels away from the main beam are generallyhigher than those actually measured. This is particularlyso for the E-plane patterns of Figs. 9 and 10.

(a) rectangular horn antenna

(b) wedge antenna with converging and diverging taper profiles

(c) double wedge-shaped antennaFig. 14. Various types of rectangular dielectric anten na suitablefor rectangular waveguide launchers.

Another possibility is to maintain a constant value of calong the flare so that the internal and external wallsdiverge in a radial manner. It is evident from the curvesabove that it is not possible to choose c to satisfy bothgood radiation pattern and efficient launching con-ditions. To develop this principle further it is necessaryto increase the size of the launcher and length of theantenna; a practical realization of this idea is the infra-red detection antenna found in moths which is tens ofwavelengths long and has a relatively large diameterlauncher.4. Variations on the Dielectric Horn

Rectangular waveguide is more common than circularand it is possible to develop the above antennas forrectangular launchers. In fact a large variety of surfacewave antennas may be developed using rectangular,elliptical, triangular and other cross-sectional shapesof rods and tubes. Unfortunately the trapped surfacewaves cannot be expressed in analytic form and recoursemust be made to point-matching calculations.20 Aperturefields are then obtained as a series and radiation patterns

- 9 0

Fig. 15. Measured H-plane radiation patterns of dielectric filledoverm oded cylindrical waveguide, e, = 2-2 (paraffin wax ),Ao = 3 cm ; dielectric-filled waveguide apertu re, withdielectric filling extended to form a short horn as sketched. = 0.can only be obtained by numerical integration. Someexamples of variations on the cylindrical dielectric hornare shown in Fig. 14 and we have tested them experi-mentally: these antennas are interesting but show nomarked advantages over conventional forms althoughthey may have specialized applications.Much attention has been focused on the use ofdielectric inserts in the mouth of the metal horns in orderto reduce side-lobe level. For instance a recently reporteddielectric wedge-shaped antenna 2 1 '28 is very similar tothe zero flare-angle horn (Fig. 11 (a)). The insertionof dielectric strips into metal horn mouths 22 condensesthe E-field in a similar way to the dielectric tube (Fig. 8).Another interesting effect is shown in Fig. 15 whereby ashort thick dielectric horn protrudes from an over-moded cylindrical waveguide and enables a broad beampattern to be obtained from what was previously a multi-lobed radiation pattern. One explanation25 of this modefiltering action is th at radiation coupling between thelauncher and the horn adjusts itself to minimum for acertain set of modes in the metal launcher and a givengeometry of the dielectric insert; another explanation isthat the dielectric extension forms a resonant open cavity

258 The Radio and Electronic Engineer, Vol. 42, No 6


9/9


for these dimensions. Man y other dielectric insertconfigurations are possible but their mechanism, as inthe above case, is generally involved and much relianceis necessarily placed on actual measurements.5. Discussion and Conclusions

The optimization of the taper profile of dielectricantennas is seen to present considerable theoreticaldifficulties; the approach developed here is to supplementtheoretical considerations based on a simplified modelwith actual measured data and a high degree of optimiza-tion has been found possible. A conclusion of practicalimportance is that the optimization of the taper profileis well worthwhile and places both the dielectric-rod andzero flare-angle dielectric-horn antennas on a com-petitive footing with conventional metal horns ofmoderate gain. Design curves have been presented forcommonly occurring values of dielectric constant butdetails of the general computer programs are availablefrom the auth or. It is not envisaged tha t these dielectricdevices will generally be alternatives to metal hornsbecause robustness and power handling requirementsobviously favour the metal antenn a. However thereare many applications where the low cost and weight,ease of manufacture and good corrosion and sealingproperties associated with dielectric antennas will bean advantage; one such application is in large arrays.

Evaluation of the dielectric-horn antenna has shownthat it could be particularly useful when directionalproperties in one plane only are required as is so oftenthe case. An array of these devices is an in terestingpossibility.When the flare angle is reduced to zero the dielectric-horn has similar patterns to the tapered rod antennaand exhibits rotation al symmetry. The bandwid th of thedielectric-horn antenna is comparable to that of thetapered dielectric-rod antenn a and is typically 1 GH z atX band, but this is more often than not constrainedby the bandwidth of the feed system. An alternativelauncher design has not been considered since themethod used here seems to be widely applicable inpractice to both horn and rod antennas.An interesting fact has emerged that the dielectric-rod and horn antennas have similar actions despite theirwidely differing geometries; in essence the principle isto transform the launcher aperture into a larger onehaving an acceptable aperture distributio n. Thetransformation may also be carried out by any one of avariety of open dielectric structures which are suitablytapered along their length. Fro m a practical poin t ofview. there seems little to choo se betw een th e p er-formance of these various forms of surface-wave antennaand the final choice rests with the overall system con-straints ; in particular the gain limitation experienced forthe dielectric-rod and horn antennas seems to be ageneral feature.

6. AcknowledgmentsThanks are due to M r. T. N. L. G allett and M r. E.

Sigee for assistance with computations and measure-ments respectively, and to Professor M. H. N. Potok forhis continued interest and advice.

7. References1. Kiely, D. G ., 'Dielectric Aerials' (Methuen Mo nograp h,London, 1953).2. Zucker, F. J., 'Surface and leaky-wave antennas', in Jasik, H.(Ed.), 'Antenna Engineering Han dbook ', Chap. 16. (McGraw-Hill, New York, 1961.)3. Brown, J. and Spector, J. O., 'The radiating properties of end-fire aerials', Proc. Instn Elect. Engrs, 104B, pp. 27-34, 1957.4. James, J. R., 'Theoretical investigation of cylindrical dielectric-rod antennas', Proc. Instn Elect. Engrs, 114, pp. 309-19, 1967.5. Bach Andersen, J., 'Metallic and Dielectric Antennas'(Polyteknisk Forlag, Denmark, 1970).6. Gallett, I. N. L ., 'Radiation patterns of tubu lar dielectricstructures'. European Microwave Conference, London, Sep-tember 1969.7. Prochazka , M., 'Die dielektrische h ornan tenna ', Hoch-frequenztechnik u. Electroak., 68, pp. 93-103, 1959.8. James, J. R., 'Studies of cylindrical dielectric-rod antennas'Tech. Note RT41, Royal Military College of Science, June 1968.9. James, J. R., 'Aperture coupling in cylindrical dielectric-rodantennas', Electronics Letters, 4, pp. 39-41, 1968.10. Marcuse, D., 'Radiation losses of the dominant mode inround dielectric waveguides', BellSyst. Tech. J.,49, pp. 1665-93,1970.11. James, J. R., 'Some further properties of cylindrical dielectric-rod antennas', Tech. Note RT37, Royal Military College ofScience, July 1967.12. Silver, S., 'Microwave Antenna Theory and Design', p. 161(McGraw-Hill, New York, 1949).13. Barlow, H. M. and Brown, J., 'Radio Surface Waves', (Claren-don Press, Oxford, 1962).14. Snyder, A. W., 'Radiation losses due to variations of radius ondielectric or optical fibres', I.E.E.E. Trans, on MicrowaveTheory and Techniques, MTT-18, No. 9, pp. 608-15,September 1970.15. Barlow, H. M. and Brown, J. loc. cit., p. 164.16. Kharadly, M. M. Z. and Lewis, J. E., 'Properties of dielectric-tube waveguides', Proc. I.E.E., 116, No. 2, pp. 214-24,February 1969.

17. Potok, M. H. N., James, J. R. and Gallett, I. N. L., 'Dielectricantennas' 2nd Annual Research Report ELS/PR/No. 18,R.M.C.S. February 1970.18. Silver, S., loc. cit., p. 581.19. Ham id, M. A. K., 'Diffraction by a conical horn', I.E.E.E.Trans, on Antennas and Propagation, AP-16, No. 5, pp. 520-8,September 1968.20. James, J. R. and Gallett, I. N. L., 'Point-matched solutionsfor propagating modes on arbitrarily shaped dielectric-rods',The Radio and Electronic Engineer, 42 , No. 3, pp. 103-13,March 1972.21. Boulanger, R. J. and Ham id, M . A. K., 'A new type of dielectric-loaded waveguide antenna', Microwave J., December 1970,pp. 67-8.22. Hamid, M. A. K. et ah, 'Radiation characteristics of dielectric-loaded horn antennas', Electronics Letters, 6, pp. 20- 1, 1970.23. James, J. R., Gallett, I. N. L. and Sigee, E., 'New ideas andresults on dielectric horns', Proceedings of the EuropeanMicrowave Conference, Stockholm, August 1971.24. James, J. R., 'Leaky waves on a dielectric rod', ElectronicsLetters, 5, No . 11, pp. 252-4, May 1969.25. James, J. R., 'Aperture coupling in surface wave antennas',Tech. Note RT33, Royal Military College of Science,February 1967.26. Collin, R. E. and Zucker, F. J., 'Antenna Theory', Parts I andII . Inter-University Electronics Series (McGraw-Hill, NewYork, 1969).27. Potok, M. H. N., James, J. R. and Gallett, I. N. L.,'Dielectric antennas' 1st Annual Research Report ELS/PR/N o. 16, R.M.C.S., November 1968.28. Hamid, M. A. K. et ah, 'A dielectric loaded circular wave-guide antenna', I.E.E.E. Trans., AP-20, 1, pp. 96-97January 1972.Manus cript first received by the Institution on 29th February 1972and in final form on 27th March 1972. {Paper No. 1451/CC130).

The Institution of Electronic and Radio Engineers, 1972June 1972 25 9

engg approach to design of tapered dielectric rod horn ant

Documents