May Contributions to the Antenna World War I* · VanAtta andSilver: Antennas During World WarII The outstandinig development, however, was the production of cylindrically curvedphasefronts

PROCEEDINGS OF THE IRE

Contributions to the Antenna FieldDuring World War I*

L. C. VAN ATTAt, FELLOW, IRE, AND S. SILVER$, FELLOW, IRE

Summary-During World War II intensive efforts of radio engi-neers and physicists resulted in the invention of many new types ofantennas and in advances in fundamental antenna theory. Thepaper presents the results of this work in relation to the principalproblem areas that were recognized during that period: 1) sidelobesuppression; 2) beam-shaping techniques; 3) beam-scanning tech-niques; 4) broadbanding; 5) antenna siting. In each of these areasmajor advances were made, both in operating hardware and intheoretical understanding.

I. INTRODUCTION

VlvHE developments in the antenna field duringT World War II, the result of intensive and com-

bined efforts of radio engineers and physicists,were marked by the invention of many new types ofantennas, and also by advances in fundamental antennatheory. Indeed, the whole subject of microwave opticswas born and grew to maturity during the war years. Areview of the field leads us into many different direc-tions. In choosing the subjects for this short resume wehave arranged the material, including theoretical con-siderations, according to certain operational develop-ments. Because the achievements were so largely theresult of the combined efforts of many people in theUnited Kingdom, Canada, and the United States, wehave avoided ascribing any development to a particularperson, even though in a few instances this would havebeen possible.The review is divided into five sections: 1) sidelobe

suppression, which relates to developments based onthe concept of flat phase fronts; 2) beam-shaping tech-niques, which utilized new developments in microwaveoptics; 3) beam-scanning techniques; 4) broadbanding;5) antenna siting, a subject of great importance in realiz-ing the optimum performance of the systems and inmeasuring the characteristics of antennas.

II. PENCIL BEAMS AND SIDELOBE SUPPRESSIONThe first operational concept in the design of an-

tennas for radar applications was that of providing anarrow beam whereby targets could be located withhigh precision in azimuth and elevation. The pulse tech-nique of radar, of course, provides the range informa-tion, and the combination of beam and pulse techniquesyields the three coordinates of the target in space. Theoperational requirement, based on the searchlight con-

* Received by the IRE, June 30, 1961.t Hughes Research Labs., Malibu, Calif.I Dept. of Elec. Engrg. and the Space Sciences Lab., University

of California, Berkeley, Calif.

cept of optics, required not only that the main beambe narrow, but also that the sidelobe structure be keptlow to avoid the ambiguity of targets detected on thesidelobes. Two main approaches were taken to achievea narrow beam: the use of optical devices, lenses and re-flectors, having the property of transforming a family ofrays from a point source into a family of parallel rays;and the use of discrete arrays of elements based on theprinciple of interference.The directive antennas to produce a main beam cir-

cularly symmetric about the axis utilized a paraboloidof revolution with a primary feed at the focus. In orderto achieve maximum forward gain out of a given aper-ture the distribution in the field over the aperture shouldbe uniform in both amplitude and phase. However, thesidelobe structure is determined by the distribution ofamplitude within the uniform phase constraint, and amajor aspect of the theoretical work in this field was thestudy of the interrelation between the aperture distribu-tion and the entire radiation pattern of the system. Thisgreatly enlarged the understanding of diffraction theoryover what had been available prior to the war. It alsoled to a rediscovery of long forgotten monumentalcontributions made to the field of optics over a hundredyears ago.Much effort was devoted to the realization of a feed

system having, on the one hand, the characteristic of apoint source, that is, generating a spherical phase front,and, on the other hand, a power pattern which wouldproperly illuminate the paraboloid. When the coaxialline was the preferred form of transmission line, thenatural choice of primary radiator was the dipole and,in particular, the reflector-backed dipole. Dipole feedsdid only a fair job of illuminating the reflector sincethey produced unequal illumination in the orthogonalplanes, and hence asymmetry in the mainlobe. In addi-tion, their considerable backlobe radiation had del-eterious effects on gain and near-in sidelobe charac-teristics of the final pattern.As was indicated, it is necessary to trade sharpness of

the main beam for sidelobe reduction. The general re-sult established was that the aperture illuminationshould be tapered by some 15 db between the center ofthe aperture and the edge. The half-power width of themain beam is given by 6= kX/D, where X = free-spacewavelength, D =diameter of the aperture; designs weredeveloped for circular apertures which yield values of kin the range 1.2 <k < 1.5 with near-in sidelobes down 20to 25 db from the peak intensity of the mainlobe.

692 May

Van Atta and Silver: Antennas During World War II

The availability of higher power, the move towardhigher frequency and the toleration of heavier antennastructures, especially airborne, led to the use of rec-tangular uniconductor waveguides and compatible feedsystems. The development of these waveguides andhorns is certainly an outstanding feature of antennawork during World War II. Sectoral horns derived fromrectangular waveguides by flaring in one or the other ofthe two principal planes of the guide, composite sectoralhorns derived by flaring first in one plane and then in theorthogonal plane, and pyramidal horns derived byflaring simultaneously in both principal planes weredeveloped for various applications. By virtue of theflexibility available in the design of a horn feed, it waspossible to illuminate a reflector more effectively and toachieve improved performance with respect to backlobeand sidewise radiation. Three horn-feed types are shownin Fig. 1.A major development in reflector design that ac-

companied the use of horn feeds was that of directingthe axis of the horn pattern onto the reflector at anangle to the reflector axis. The reflector is then cut alongan equi-intensity contour, again at an edge illuminationsome 15 db below the peak value over the aperture.Since the phase center of the horn remains at the focus,the phase distribution over the aperture remains planeyielding a directive beam. The importance of the tech-nique is its flexibility in controlling sidelobe levels whileretaining the required mainlobe characteristics. Such anantenna is shown in Fig. 2.

Operational needs arose for beams having mainlobesnot circularly symmetric but of different beamwidthsin two orthogonal planes, one very narrow to retainhigh resolution in the plane of scan, and the other rela-tively large to give extended coverage in the orthogonaldirection. Such beams are known as fanned beams, andbasic diffraction theory shows that they can be obtainedby using rectangular or elliptical apertures with a cor-responding ratio of their principal dimensions. Theaperture illumination problem remains that of provid-ing a uniform phase and a tapered amplitude distribu-tion to control the sidelobe level; the horn feed solvedthis problem. Fig. 3 shows a paraboloid of revolutioncut into an elongated elliptical shape and fed by a flaredhorn.Another family of antennas designed to produce

fanned beams consists of a parabolic cylindrical reflectorbetween parallel plates illuminated by a sectoral hornfeed at the focus. The exit pupil of the system is then anarrow rectangular aperture. Such antennas, knownvariously as cheese or pillbox antennas, served as endsin themselves or as line sources for illuminating largerparabolic cylinders, as shown in Fig. 4.Microwave lenses were used in Germany prior to the

war and were investigated extensively in the UnitedStates during the latter years of the war. In most mili-tary applications, however, lenses were not competitivewith reflectors because of such factors as reduced gain,

higher sidelobes, frequency sensitivity, greater weightand unfavorable shape. One development of note, how-ever, must be mentioned in this survey. It was recog-nized that the dispersive property of a waveguide couldbe used in creating a medium having an effective indexof refraction less than unity. The waveguide or metal-plate lens thus came into being but did not find actualapplication during the war. Fig. 5 shows an early type ofwaveguide lens.

Linear arrays also received a great deal of attention.The designs which were built around the coaxial lineutilized dipole radiators, and there were many iingeniousconfigurations devised for beacon antennas and relatedoperational systems. Basically, however, these antennaswere adaptations of ideas and developments already inuse in radio-communications and direction-finder sys-tems before the war. The distinct war-period contribu-tion to the array system was the slotted waveguide ar-ray. The theory of slot radiators was developed exten-sively, and here it is appropriate to state that the basicwork was largely provided by groups in the UnitedKingdom and Canada.The excitation of a slot in the wall of a rectangular

waveguide can be controlled by its position on the walland the orientation of the axis of the slot with respectto the axis of the waveguide. It is this flexibility thatmakes possible the relatively easy control of excitationalong the array. Of the many developments that weremade in this field, the so-called Dolph-Tchebycheffarray deserves special recognition. In this type of arraythe amplitude distribution along the array is related tothe coefficients of a Tchebycheff polynomial. The re-sult is a beam having all sidelobes of equal amplitudeand therefore the narrowest beamwidth consistent witha prescribed sidelobe level. Sidelobe levels 30 db belowthe mainlobe peak were obtained, and in later develop-ments following the war even lower sidelobe levels wereachieved.

III. BEAM SHAPINGThe operational requirements that motivated the

design of fanned-beam antennas became rather quicklymore demanding with respect to the more efficientutilization of the available power in air-search systemsand more uniform ground illumination in airbornenavigational and bombing antennas. One solution to theproblem was obtained by the use of an extended feed inthe focal plane of a paraboloid reflector. Each elementof the feed produces a beam displaced from the axis byan angle proportional to the displacement of the elementfrom the focal point. The resultant of the overlappingbeams is a flared beam. High resolution in the trans-verse aspect is preserved until the coma aberration over-rides the collimating property of the reflector. Thetheory of aberrations was advanced markedly in thecourse of this developmental program. The antenna onthe left in Fig. 7 achieves a shaped beam by means of athree-horn distributed feed.

1962 693

PROCEEDINGS OF THE IRE

(a)

Fig. 3-Fanned-beam antenna.

(c)Fig. 1-Horn-feed types. (a) Electric-plane horn. (b)

Magnetic-plane horn. (c) Compound horn. Fig. 4-Pillbox antenna feeding a cylindrical reflector.

Fig. 5-Waveguide lens antenna.

Fig. 2-Antenna with asymmetrically cut reflector.

694 May

(b)

Van Atta and Silver: Antennas During World War II

The outstandinig development, however, was theproduction of cylindrically curved phase fronts over theaperture of the antenna. By developing a phase frontwhich is a section of a generalized cylinder, it is possibleto obtain a flared beam which, in the transverse aspect,retains sharpness over large flare angles. The comaeffect is eliminated since the rays are necessarily per-pendicular to the generator of the cylinder. Both theunderlying concepts and the realization of the systemwere major advances in microwave optics.The requisite phase front was realized in several ways.

One system utilized a line source feeding a cylindricalreflector whose cross section was designed to producethe dispersion of the ray system required for the flaringof the beam. The line source was aligned to be parallelto the generator of the reflecting cylinder. The sidelobelevel in the transverse planes is determined entirely bythe amplitude distribution along the line source. Bothpillboxes and arrays were used as line sources, and all ofthe techniques for controlling sidelobe levels derivedfrom flat phase fronts can be used in the design of thefeed. (Fig. 4 represents this type of antenna.)A second type of system utilized a point source feed

with a modified paraboloidal reflector. The earliestform was made up of a split paraboloid of revolutionwith one section displaced relative to the other, so thatthe resulting phase front had a large amount of third-degree phase error. Advances in theory, however, led toa new type of reflector whose curvature varied frompoint to point so as to fit the ray requirements com-pletely over the entire range of the beam. The com-plexity of the resulting surface was more than balancedby the simplicity inherent in the point source feed. Thistype of reflector, referred to as the doubly curved re-flector, was used increasingly toward the end of the war.Fig. 6 shows an operational antenna of this class.

IV. BEAM SCANNINGA radar system must be able to scan its directive

beam at a rate compatible with information rate calledfor by operational requirements. This must be accom-plished while preserving resolving power and an effectiveSNR. Among the outstanding contributions made to theantenna art during World War II, were the develop-ments in scanning techniques.The obvious method of scanning by moving the en-

tire antenna had soon to be superseded by other tech-niques as antennas became larger, and as higher scan-ning rates and complex scanning patterns becameneeded. In essence, the scanning problem is one ofchanging the orientation of the phase front at theaperture of the antenna. This must be done with mini-mum distortion of the phase front to preserve the struc-ture of the beam. The various rapid-scanning techniqueswhich were invented can be designated by the war-timecategories as optical scanning, phase-shift scanning, andfrequency-shift scanning.

In optical scanning only a part of the antenna is

Fig. 6-Shaped-beam antenna employing a modifiedparaboloidal reflector.

moved. For example, the feed of a parabolic reflectorcan be moved off axis in the focal plane to obtain alimited angle of scan. In particular, the feed may bedisplaced slightly and then spun about the axis toachieve a conical scan. The differential signal fromeither side of the cross-over point yields a high accuracyof pointing and a convenient tracking signal.The Robinson Roll antenna (right-hand antenna in

Fig. 7) was the most ambitious application of the dis-placed-feed technique. The reflector had a long focallength in one plane and a relatively short focal lengthin the other. The feed was located at the far focal point,but was confined in one plane between parallel plates tomeet the short-focus requirement. Moving the feedbetween the plates and parallel to the aperture thenprovided a scan in one plane. The final invention wasto fold the plates in such a way that the oscillatory feedmotion was replaced by a circular motion.One system which was developed to effect the same

purpose as conical scanning is essentially a data-process-ing technique. It uses a stationary feed system com-prised of four feeds clustered in a square about the axis.These, when fed all in-phase and in out-of-phase pairs,form a central or sum lobe and two split or differencelobes in orthogonal planes, respectively. By using thesum lobe in transmission and the difference lobes andthe sum lobe simultaneously on reception and compar-ing signals, one obtains a superior tracking antenna.

Three different examples of phase-shift scanningdeserve mention. The Navy Mark 8 fire-control antenna(Fig. 8) scanned in an azimuth sector the beam formedby a two-dimensional array of "polyrod" endfire radia-tors, by separately controlling the phase to each verticalbank of radiators. This control was accomplished byrotating impedance elements in the circularly polarizedfield of cylindrical waveguide sections. In the Fosterscanner a linear variation in phase was accomplishedby control of physical path length; a wave betweenparallel plates was conducted around a variable portionof the circumference of a cone. The Eagle antenna was a

1962 695

PROCEEDINGS OF THE IRE

Fig. 7-Shaped-beam antenna (on the left) and RobinsonRoll scanning antenna (on the right).

Fig. 8-Mark 8 "polyrod" scanning antenna.

linear array of dipole elements fed from rectangularwaveguide. A linear-phase variation was obtained inthis case by varying the major dimension of the wave-

guide and, hence, the optical path length betweenradiating elements.

Frequency-shift scanning makes use of the fact thatthe wavelength, and therefore the phase, along an ar-

ray of radiators fed from a waveguide depends upon thefrequency of the radiation. The scanning effect can beamplified by using a frequency-sensitive waveguide andby wrapping or folding the guide between radiators.This technique was explored during the war but did not

find extensive application until later because of lack offrequency flexibility in available radio-frequency power

sources.

V. BROADBANDING

In the early days microwave power sources, even

though designed for a spot frequency, came off the linewith a spread in frequency. Later an even greater spreadin frequency was required to reduce friendly jammingand to make enemy jamming more difficult. For thesereasons the RF system, including the antenna, had toperform satisfactorily over a band of frequencies.Studies of a large variety of systems led to a clearerunderstanding of the factors limiting impedance band-width and of the methods of combining mismatches to

cancel one another.

In the case of the antenna, satisfactory performancerequired that any of the essential antenna characteris-tics be maintained within certain limits over the speci-fied band. These characteristics included power gain,beam-width, sidelobe level and, in the conical scan pat-tern, the cross-over level. The characteristic of greatestconcern, however, in those days of the sensitive magne-tron was the impedance of the antenna.A constant impedance mismatch, even if large, could

be corrected by a transformer at the magnetron, but amismatch that varied widely and rapidly with fre-quency had to be eliminated or compensated at thepoint of origin. Frequency-sensitive mismatches resultedprimarily from the combined effect of a series of dis-continuities distributed along the RF line. Antenna mis-match was most serious since the antenna was farthestfrom the RF source.

Feeds for paraboloid reflectors were matched first in-dependently of the reflector. Dipole feeds from coaxiallines, with their associated chokes and directive dipolesor plates, were relatively frequency sensitive with manycritical dimensions requiring elaborate adjustments.Horn feeds from waveguides were basically bettermatched with fewer critical dimensions and were moresusceptible to calculation.The art of horn design reached an advanced state dur-

ing the war years. If the waveguide was flared in bothdimensions, the flare angles and their positions in theguide were chosen so that their discontinuities tended tocompensate each other and also that caused by themouth of the horn (see Fig. 1). Final correction was ac-complished by capacitive or inductive strips at or nearthe mouth of the horn, or better, in some cases, by thethickness and placement of a plastic cover over thehorn.When the feed was placed in the reflector, an addi-

tional discontinuity resulted from reflection back intothe feed. In some cases this discontinuity was correctedby a small plate placed a fraction of a wavelength infront of the vertex of the reflector. The size and place-ment of the vertex plate could be calculated from thegeometry and the feed pattern. The vertex plate sawlimited use, however, because it disturbed the apertureillumination and increased sidelobes. The ideal solutionwas found in the asymmetrically cut reflector (see Sec-tion II) which took the feed out, or almost out, of thereflected beam. In this case the feed was located at thefocal point but directed off-axis approximately towardthe center of the reflector area.The reflector mismatch was much more serious in

cylindrical reflectors, such as pillboxes fed by horns orlong cylindrical reflectors fed by linear arrays. For theseantennas it was essential that the feed be removed fromthe reflected beam. This was accomplished for the pill-box by the use of the folded pillbox or by the "hoghorn."In the folded pillbox the feed at one level was connectedto the linear aperture at another level by a parabolicbend. In the hoghorn the feed, still at the focus, illumi-

696 May

Van Atta and Silver: Antennas During World War II

nated half a pillbox with the aid of a guiding horn. Bothof these pillbox modifications, however, introduced addi-tional weight and a less satisfactory form factor. In thecase of a cylindrical reflector illuminated by a long lineararray, the reflected wave striking the array seriouslydisturbed both the impedance and the radiation char-acteristics of the array. Therefore the array had to bekept well out of the path of the reflected wave.The long linear array, in which only a small fraction

of the incident power reached the far end of the array,was usually terminated in a load. This so-called non-resonant array provided a relatively good impedancematch since the reflections from successive elementscould be made to curl up into a small resultant mis-match at the input to the array. Short arrays, operatedresonantly with a short-circuit termination, were morefrequency sensitive in their impedance. In both cases, ofcourse, well-matched or individually compensatedradiating elements greatly improved the over-all per-formance.

VI. ANTENNA SITING

An antenna tested under relatively "free" conditionswas found to perform quite differently under operatingconditions because of reflection of its radiation by thesurroundings. Reflection back into the antenna affectedits impedance match; forward reflection affected themainlobe or sidelobe characteristics of its radiation pat-tern. Toward the end of the war great strides were madein understanding and solving the complex problemscreated by the installation of many antennas on a singleship or aircraft.A shipborne antenna with a horizontal beam had its

elevation pattern sharpened and split into multiplelobes, and its gain increased by reflection of half of themainlobe by the sea surface. The over-all effect was notsimple since the reflectivity of the sea surface dependson wavelength and polarization of the radiation, angleof incidence, and sea state, and the combined patterndepends on height and orientation of the antenna.A more complex installation problem was created

aboard ship because many antennas were competingwith each other in the presence of stacks, masts andsuperstructure for some semblance of free-space condi-tions. This so-called antenna-system problem was par-ticularly severe for omnidirectional antennas. In thecase of HF communication antennas, intercoupling be-tween tranismitting and receiving antennas and thesharply lobed patterns of individual antennas wereknown only through operational experience, so that

maintaining satisfactory HF communications requiredblack art of a high order. Radars were frequently in-stalled in pairs with their antennas fore and aft or portand starboard in order to achieve complete azimuthalcoverage.

Military aircraft also presented a serious antenna-system problem. Antennas for navigation, short- andlong-range communication, and radar competed forfavorable sites in the face of increasingly severe aero-dynamic restrictions as aircraft speeds increased. Theradar antennas, aircraft intercept or bombing andnavigation, had the additional problem of operatingthrough a radome, a plastic housing necessary for aero-dynamic reasons.The radome problem first presented itself dramat-

ically when a naval aircraft search radar was found tobe blanked out in certain sectors. This was found to bedue to reflections from the radome back into the RFsystem pulling the frequency of the magnetron out of thepass band of the receiver. This impedance effect con-tinued to be the most serious consideration. Under cer-tain conditions, however, refractions and forward re-flections could distort the mainlobe and introduce ob-jectionable sidelobes.Radome wall designs included the thin wall (zero-.

thickness approximation), the sandwich wall (two thinskins spaced a quarter wavelength apart by plasticfoam), and the half-wave wall (a solid, high-densitywall, one-half wavelength thick). During the war muchattention was given to the sandwich wall, but shorterradar wavelengths, higher temperature requirements,and more severe microwave optical requirements even-tually favored the half-wave wall.Antenna siting problems were first encountered in

connection with measurements made in the course ofantenna development and test. With directive micro-wave antennas, measurements of the pattern and im-pedance were normally handled separately. For im-pedance measurements, satisfactory results could beobtained merely by pointing the antenna out through alarge open window. For pattern measurements, espe-cially since there was considerable interest in measuringlow-level sidelobes, it was necessary to have transmit-ting and receiving antennas facing each other across aconsiderable distance from elevated vantage points, sothat reflecting surfaces were well out of both beams. Asantenna apertures increased, the minimum patternrange also had to be increased according to the relationR min = 2D2/X, where D was the maximum dimension ofeither transmitting or receiving antenna.

1962 697