Analysis of Rect Horn Ant via Uniform Asymptotic Theory

download Analysis of Rect Horn Ant via Uniform Asymptotic Theory

of 10

Transcript of Analysis of Rect Horn Ant via Uniform Asymptotic Theory

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    1/10

    AND PROPAGATION, V O L . AP-30, NO. 2, MARCH 1982Analysis of Rectangular Horn Antennas via Uniform Asymptotic Theory

    RONALD C . ME N E N D E Z , MEh lBER , LEE , A N D SH U N G - W U LEE, FELLOW, IEEE

    Abstract-Approximate analyticxpressionsorhear-fieldpattern ndmodal eflection coeffkient ofhehree-ntenna ave een eveloped. Based upon a

    modelf thewo-dimensionalorn-waveguideapproachoffersgreatergenerality than previously

    nd accuracy. Theasymptotic heory CAT),which semployed hroughout,

    accuracy in the solution. It is demonstrated that, in most cases, theof the horn is reducible to the pattern of a simple slit

    an absorbing creen lluminated by anarrayofpatterned ineThemeasured patternsofseveralhorns are convincingly

    ered by the prediction of the analytical model. The same model iso calculatehemodaleflectionoefficient of the horn-

    and, again, comparison with measured results is

    I . I N T R O D U C T I O NY R A M I D A L andsectoralhornantennashavebeenana-lyzed by a variety of app roach es [ l ] -[ 61 wh ich use rayMost a re ybr id pproacheswhi c h oup l e ay

    iding a comp le te ray p ic turet he ho rn a n t e nna . The goa ls o f h i s pa pe r a re to d e m o n -e ha t suc h a p i c t u re ma y be fo rmul a t e d , a nd t o ob t a i ndom inant far- field pa t te rn and the mod a l re f lec t ion coef-

    of th e horn-waveguide anten na s t ruc ture . Fur the rm ore ,a comple te ray p ic ture will be show n to be more genera lThis paper is organized as fol lows. After a concise summarythe re levant d i f f rac t ion theory in Sec t ion 11, a s t a t e me n t o fproblem i s presented n Sec t ion 111. I t is fo llowed by th eesults of th e far-fie ld adiat ionpattern Sectioncompari sonsof hepredicted patterns wi thme a s -

    IV-D). A discussion of the meri t s of th i sh e la tive to previouswo rk is given in Section IV-E.heappl ica t ion of th i smult iple magem o d el t o o b t a i n h eeflect ionoeffic ients resentedn ec t ion V-A,owed by a compari son of e xper imen ta l resul t s and predic -in Sec t ion V-B.T h r o u g h o u th isw o r k ,he n i fo rm sympt o t i ch e o r y) is e mpl oye d t o de sc r ibe hediffract ionprocesses n-The nota t ion and te rminology will be those employ ed[ 8 ] andma nyof he esul t spresentedheredrawheavi ly

    Severa l onvent ions used throughou this apero l l ow.( - i w t ) and i s suppressed. 2 ) The d i f -problem i s lways educed to wod i m e n s i o n s n oManuscript eceivedMarch18 ,1980; revised Ju ne 2 5 , 1981 .Th i s\vas supported by he National Science Foundation under GrantR . C. Menendttz is with Bell Labora tories , Whippany Road, Whip-S. W . Lee is with t h e Elec t romagnet ics Labora tory . Department of

    NG 77-20820 .NJ 07981 .

    niversity of Ill inois, Urbana, IL 61801 .

    24 1

    y var ia t io n) . 3) Both he t ransverse magne t ic (TM)case (non-ze ro f ie ld components H y , E,, E,) an d the transverse electric(TE) case (E, , , H ,, H z ) are rea ted s imul taneou s ly , wi th hehe l p o f t he t wo symbo l s a nd r such tha tfo rh l : u = H y and T = + If o rE : u =E, , and r = - l .

    11. S U MM A R Y O F G T D A N D U A TF or onve n i e nc eof e fe rence nd nt roduc t ionofno t a -

    t ions ,hee levant or t ions fhe eomet r ica lhe o ry fd i f f r ac t i on (GTD) a nd t he UAT a re summa r i z e d i n t h i s s e c t ion[71 - [141 .

    A . Geometr ica l Theory ofDif frac t ion [9]As show n in F ig . 1 , a wedg e i s i l lumina ted by the inc identfield u' f rom a l ine source a t 7 = Ti. The prob l e m is t o d e t e r -mine the asym ptot ic so lu t ion of the to ta l f ie ld u r m u p t o t h eo r d e ro f k - ' I 2 inc luded re la tive t o u'). According to t h eG T D , u t is given byG T D : ur(?) = ug(?) + ud(?) + O(k-l), k + m. (1)

    Here ug i s the geom et r ica l opt ics f ie ld , comp osed of he in-cident field, u z and the reflected fie ld u r . Corresponding to t h efac t tha t u z s prod uced by the source a t ?iF i g . l ) , u r ma y beident i fied wi th the f ie ld radia ted f rom an image source a t T r .The presence of th e wedge casts a shadow with respect to uiand one w i th respec t ' to ur in the geom et r ica l opt ics sense . Toexpress this , fact mathematical ly, le t us i n t roduc e t wo sha dowindica tors E' and 8 , uch tha t '

    + 1, if i i s in he geom et r ica l shadow region&zr(r) = of u2rr

    -1, if ? is in the geo metrical l i t region of u'".(2)

    The nma y ewrit ten s~gc)O ( - E ~ ) ~ ~ ( ? ) + e(-Er)ur(?j (3 )

    whe re 8 is a uni t -s tep func t ion: B ( x ) = 1 if x > 0 an d e ( x ) = 0if x < 0. The se c ond t e rm ud in (1) is the (Keller ' s) diffractedfield, given b yud(?) =g(kr){X,,,(Jr')u'(;= 01 + X , ~ ( W ) U ' ( ; = 0)). (4)

    He re g ( k r ) is a cyl indrical w ave factor,

    0018-926X/82/0300-0241$00.75 0 1 9 8 2 IEEE

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    2/10

    2 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. AP-30, NO. 2 , MARCH 1982

    / 0 '

    POINT T

    1 . Point source at f i illuminating wedge of exterior angle wn.OBi and OBr are the shadow boundaries.

    is the diffraction coeff icient of a wedg e of exterior angle( = [ F ( @)- (ti)] u ' ( 7 )+ [F(E') - )I u ' (? ) .

    (9 ) is defined by,-in14 -

    F(.$)=- i t2 d t , for real t .6For a la rge argument ,F(E) has an asympto t ic expansion

    ~ . ' ( t )e(-- [) + : ( t ) + NE- ), I t I+ 03here

    Th e d e to u r p a r ame te r l l ( l r )n (9) measu res the phase differ-ence be twee n the inc iden t ( ref lec ted) f ie ld and the d i f f rac tedfield, as defined in [ 1 3 , eq . (5.5)] . For the specia l case whenthe source is a l ine source (Fig . l ) , ti>'' ssume s imple forms,namely ,p(;) = 2 , ' Iv'k(a + b - "') I ( 1 3 )

    where & e r ) i s the d i rec t d is tance f rom ?@') t o 7, while (a -I-b ) is that via the edge.hus, is th ehasei f ferencebetween ui n d ud a t ?. The to ta l f ie ld ut given in (8) is validfo r a l u', i n c lu d in g p o in t s o n th e sh ad o w b o u n d a r i e s o r n ea rthe edge . In fac t , u t in (8) satisf ies the edge condition at ?= 0.

    The fo l lowing observat ions can be m ade when com par ingthe GTD so lu t ion (1) and the UAT so lu t ion8).1) If I i f " I + 00, ( 1 1 ) may b e u sed n (9) and uG + $.Thus , th e UAT so lu t ion reduces to tha t of t h e GTD.2) If I [ 1 1 r I + 0, the GT D fa i ls because u t in (1 ) becomesinfiniteanddiscontinuous. ncontrast , heU A Tsolu-tion in (8) remains valid.

    In other words, the GTD agrees asymptotically to t h e o r d e r ofk - l 2 with the UAT as long as I tJ>"are large enough so t h a tthe asympto t ic expansion (1 1) may be used in (9) . I t is usefulto in t r o d u ce two t r an s i t i o n r eg io n sT' and T'. When the obser -vation point 7 falls inside p>',he GTD solution differs, signif-ican t ly rom heU A Tsolution. When u' isoutside p s r , h eGTD olu t ion greeswi th heUAT o lu t ionwithin pre-scribed tolerance. The transit ion regions are defined byTi": I $i*r(Q I < M ( 1 4 )

    where M isasuitably argenumber . See [7] and [ 8 ] f o r amore com prehensive discussion.) -

    111. STATEMENT OF THE PR OB LEMConsiderectangularwaveguidendyramidalornoriented such that the z-axis is in the ax ia l d i rec t ion , and the

    normals o he waveguide walls a re o r ien ted o ie a long hex or y axes, Fig. 2 . We approximate th is fu l l th reed imensionh o r n wav eg u id e s tr u c tu r e b y wo wo d imen s io n a l s tr u c tu r e swhichare heprojections of theh o r nwav eg u id eo n to h ex - z or y - z plane . The pat tern of the hree-d imensionalhorn in the x - z plane is approximated by that of the two-d imensional s t ruc ture pro jec ted on to the x - z plane. Thus, apyramidal o rnspproximated by twowo-dimensionalh o r n s (of possibly different flare angles), while a sectoral hornis modeledbya wo-d imensionalhorn noneplaneanda(9 ) parallel-plate waveguide in t h e o th e r . In addi t ion to projectingthe physica l s t ruc ture of the horn on to two p lan es , the inci -dentmodemust tselfbe eso lved on to hese wo p r in c ip a lplanes. In effect , this corresponds to ignoring he f ield varia-t ion in the y-direction when considering the two-dimensional(10) horn whic h is the pro jec t ion n he x - z p lane . The approx-imatio n of the f ields of three-dimen sional structures by thoseof two-dim ensional structures is common in ray analyses, buti t does imi t he analysis here to the two pr in c ip le p lanes of

    Th e r e f o re , n wh a t f o l lo ws , we an a ly ze a two d imen s io n a lhorn-waveguidestructurewith he ntention of apply ing he' resu lts to th ree-d imensional s t ruc tures . Our goal here i s to p re-(1 2 ) sen t a generalcohesive aypictureof his tructureas well

    (1 1) th eh o r n .

    as to learn the shor tcomings of such an approach .

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    3/10

    MENENDEZ AND L E E : ANALYSIS OF KECTANGULAR HORN ANTENNAS 243

    Fig. 2. Three-dimensional ectangular hor n and i t s two two-dimen-sional horn projections.

    Fig. 3 . Geometry of two-dimensional horn a n te n n a . (a ) Horn-wave-guide geometry. (b ) Cascaded aper tures of horn-waveguide s t ruc-t u r e .We assum e ingle rbitrarilypolarizedguidedmode spropaga t ingwi thin hewaveguideand s nc identupon hehorn . The geom et ry of he genera l wo-dimensiona lproblemis sketched nFig.3(a).Thecompletehorn-waveguide struc-

    ture is specified by three parameters: guide half-width-a , flarelength-L,and lare angle-@. Th eguide i s compo sed of tw oparal le luns taggeredperfect lyconduc t inghalf-planes C1 andCz , separated by awidth 2a. The single propagatingmodeinc ident upon the aper ture f rom z = -O O i s of the formTM: H, .(x , z) = 2 co s & (X + a ) ) ex p ($,a)

    n = 0, I I 2, ... ( 1 5)

    T E : E I s n ( x , ) = 2i sin (l,(x +a ) ) e x p [ifl,.)n = 1 : 2 , 3, ... ( 1 6 )

    where n i s the mod e number , 4, is the transverse wavenumber( E , l ,= n7r/2a), and p,, is he ongitudinalwavenumber p,, =+ ( k - - E 1 z 2 ) 1 / 2 . T h e TM expression ( 1 5) a lsoc on t a i ns heTE M m o d e ( n = 0). The inc ident f ie ld may be readi ly decom-pose d i n t o i t s t wo p l a ne wa ve c omp one n t s a nd , wi t h t he c on -ve n t i ons on u and 7 descr ibed previous ly , both pola r iza t ionsof the incident fie ld may be wri t ten as

    ui = u+i + u- i ( 1 7 4whe re

    u+ = e x p i (x+ a ) +& z)and

    u- = 7 e x p ( - t n ( X + a ) + &z). (1c )Clearly, u+ i (u - i ) represents a p lane wave propaga t ing a t angle@,(-@,) with respec t to the -axis ,

    @,l = sin- (n7r/2kn). (18 )Not ice ha t u p i epresen ts he nc iden t f ie ld for he wedgeW1 whi le orresponds to he f ie ld e f lec ted rom W1 . orth esecondwe dge W, he ro l e s of u+ and u - i are eversed(Fig. 3(b)).As show n nFig .3(b) , hehorn -wa ve gu i des t ruc turec on-s i st s of two aper tur es , A an d A 2 , in se r ies. Aper ture A con-sistsof thega pbe t we e n W , and W 2 while aperture A2 isformed offlareedges El a nd E 2 . This pa i r of aper tures m-media te ly sugges ts a number of ques t ions regarding the ana l -ysis of such a structure by ray techniques.

    1 ) What will be the role of the pr imary inc ident f ie ld ( themod a l rays w i th in the guide) n the f la re region (be tween41 a n d A 2 ) n d n h e a r f ie ld ( b e l J o n d A 2 ) ? 1 9 a )

    2) Can the in te rac tions be tween the throa t wedges and theflare edges be correct ly accounted for in l ight of the dif-ficul ty in cascading U A T solut ions?1b)

    3) What will be t he role of the diffra cted sour ces near thethroa t in thea rieldbe yond A 2 ) ? (1c )

    Theabovequest ionsar eanswered n he ol lowingsections.In answering those ques t ions , we d raw upon two conc lus ionsderived in [8 I . We now summarize the two conc lus ions .F i rs t ,anarbi t ra ryaper ture l lumina tedbyaplane wavegeneratesa ar ieldwhichcontains n o d i rec t geomet r ica l)ray on tribution. notherword s, he perture ar ie ld iscomple te ly descr ibed by he Ke l le r d i ff rac ted rays emana t ingfrom the edges . Fur the rmo re , for an aper ture which i s of theorder of a typica l open-ended waveguide ( l ike aper ture A , inFig. 2 , 2 k a < IO) , th eun i fo rme xpre s s i ons o r hepa t t e rn[ 171and he ar-fie ld Keller pat ternconvergewithina ewguide widths of the aper tu re . S ince aper ture A is i l luminatedby two planewaves, u+ and u- I , the d i rec t cont r ibut ion ofthese plane waves in the far fie ld is absent . Except very near

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    4/10

    IEEE TRANSACTIONS ONat, the f ield in the f lare region is,wellymodeled by th e

    u+(u-) inc iden t uponW 2 (W I ). Since hese ields re ay ields, hey re

    A 2 u n d e r h e o n -of the UAT.Another conclusion der ived in [SI conce rns the use of theheransmissionhrough an absorbing(a screen which produces no ref lec ted f ie ld for an n-t ray f ield, i .e ., ref lectioncoefficient 7 = 0). Referr ingFig. 4, consider a l ine source having a radiation pattern ( 0 )7 behind two absorb ing screens. (The z igzag l inesFig. 4 are meant t o represen tabsorbingscreens.)Let the7 = ( Y , 6 ) b e ex -ed b y, as Y + =,

    U ( Y , 6 )g ( k r ) P ( 6 ; ? ) + O ( k - 3 / 2 ) . ( 2 0 )[81 leads to a f ina l so lu t ion for

    e far-field patte rn P(e ,? ) given by

    * e x p [ i k r cos (e + e ) ] p ( e )

    S represen ts the s li t ha l f -wid th ; he de tour parametersre= sin*1,2/2)> (22)

    , r l ,2, and Ql2 a r e sh o wn in Fig . 4. The incident f ield isof the am e ormas (20). We use he xpressionin (21) la ter .I V . FAR - FI ELD PATTER N

    . Ana ly t ic a l Resul t sTh e r aypicture of thed if f rac t ion processesoccurring ne horn waveguidestructuremaybeshown to be ormallycal to ha t o f much implerproblem: slit in n

    creen l luminatedby n rrayofpatterned ineP ( 0 ) of the horn is repre-y a uperposi t ion of te rmso f h e o r m o f211,9

    ~ ( 0 ) p ( e ; + o(k - ( 2 3 )n= 1

    ere, in m ost cases the sl it s the ape rture A 2 ( th e sl i t half-S = a + L s in@ ), and the se t o f line sources loca ted a t, ~ represen ts hed if f rac ted ources twed g es W l an d W 2E l W1 and2 W 2 . Thesemultiply eflected aysmay be convenientlyof images [31 . Conceptual ly , the

    bouncing to and f ro in the la re eg ionmay be equiva-described as the result of a set of image sources in free. The interaction of these image sources with the secondA 2 provid es the far-f ield pa ttern.

    I

    ANTENNAS AND PROPAGATION, VOL. AP-30, NO. 2 , MARCH 1982I

    2

    Fig. 4 . Singlepatternedine ource at r illuminating a slit in anabsorbingscreen.Hornmay be treatedassuch a slit lluminatedby several such sources.Consid er the horn show n in Fig. 5 . Fo r th e ang le 0, larelength L , and gu ide wid th a chosen here , there are ef fec t ive lyfou r sources: the tw o wedges Wl ,2 and the i r images W3,4. Ifwe egard the aysd i f f rac ted rom W1 ,2 asarising from apat terned ine source a t W 1 2 , i t s no t d i f f icu l t to see tha tsource W3 is imply the imageof W 2 in theupper la rewhile W4 is the imageof W 1 in the ower la re . The se t o fsources lies on a circle of radius b(b= a/s in 9) cen te r ed ab o u tth e p o in t a t wh ich th e u p p e r an d lo wer f l a r e su r f aces wo u ld

    in tersec t . The sources are separa ted f rom hei r ne ighbors byangle 20. The set of im age source s is alwa ys f inite in n u m b e rbut can be large if 0 s small.Aside from ach sources individual ontr ibu tion in t h ed i rec tion of the observat ion po in t , o ther fac tors are a lso im-p o r t an t .Th e ay s r o me a c h m a g eso u r ce o l lo w a u n iq u ep a th o h eo b se r v a t io np o in t .T h ep a th en g thd i f f e r en cesare crucial . The r icochetted paths from flare to f lare are pre-cisely acco unted orby he elativeposi t ions of the mag esources . naddi t ion , ac torsaccount ing or henumber ofbounces, he po lar iza t ion , and he d i f ference in t h e n c id en texci ta t ions of W l and W 2 must en c lud ed to ma in ta incor rec tphase e la t ionsamong he mage ource ie lds .Th ecomple te set of mages adiating in free spaceautomat ica l lysa tisfies the oundary ondi t ion longh elare surfaces.Rela t ive to the kO-ord er inc iden t mod e in the waveguide, thefield produced by the wedge sources is of k- /*-ord er . How-ever , s ince none of the kO-ord er f ie ld penet ra tes aper ture A Ito the ar ie ld , hek- /* -order ie ld of the imagesourceswill be the dom inant te rm in the far f ie ld of the th roa t aper -tu re .The ie ldofan mag esourcemust , n u rn ,penet ra teaper ture A , ; the geometr ica l op t ics contr ibu t ion of the t rans-mit ted f ie ld s o f p r imary mpor tance to the far f ie ld of thehorn . The cascading of the throat-diffracted f ields with aper-tu re A , ra ises the impor tan t quest ion of wheth er the th roat -diffracted f ields may be represented as an optical f ield (or sumof optical f ields) in order that the UAT may be applied at 42 .If the f lare angle 0 and f lare length L are bo th la rge enoughthat edges El and E 2 arecompletelyoutside he ransit ion

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    5/10

    D L E E : ANALYSIS OF R E C T A N G U L A RO R NN T E N N A S 245

    5. Modeling of mul t iple images as point sources i l luminating a s l i tin anabsorbingscreen.Source W 1 adiates hrough sli t form ed ofedges E l an d E,, (a) an d (b). Source W4 (image of K 1 in lowerflare) radiates through sl i t formed of edge E 2 and wedge W 2 . (Simi-larly for W 2 an d W 3 . )withhewaveguidemode ,here i s n o

    mul t i p l ea re of he proper form to cascadewithedgesHowever, hor ns do not n general meet hiscri terion.E l ,, re bur ied n he rans i t ion regions a t tached t oW1 ,2 . Fortun a te ly , the guide width i s typica l ly qui tel compared to he lar e e n g t h L , andherans i t ion

    ons over lap to suc h an ex tent ne a rEl ,2 tha t the d i f f rac tedare a good approximat ion to the uni form solut ion . Tha tl thoughedge El may be so n e a r o B I Z small detour)di f f rac tedraycont r ibut ionfrom W l o E , isexces-E l is simultaneously near B z r and the d i f f rac ted cont r i -W2. ill largely com pensa te for he s ingula rbe-B 1 . ( In the inf in i te ly fa r f ie ld , the compensa t ionPerfectcompensa t ionfor hesingulari ty is n o tcted in the near field. As a result , he edge diffractedrising fro m E l , 2 , due o l lumina t ion by the Ke lle rW l 2 , may be som ewhat larger than would be derivedrigorous cascading of UA T solut ions. Since the diffract ionth e image source fie ld by E l ,2 is of k- -order (relative to

    e ko guide mode) , the e ffec t of the e rrors men t ioned aboveminimal n th e mainbeamwhere hek-12-orderof t he t h roa t sou rc e s i s domi na n t . Howe ve r , fo r 0 S 9,

    ed rays are dominant . Discrepancy in this region is notsing. Ou r goal s t o obtain he dominan t fie ld behaviori n t he u ll pa t t e rn , i s k - l I2 - o rde r .Large Flare Horns (@>30)Returning to Fig. 5 , we cons id er the in te rac t ion of sourceand i ts image W4 with aper ture A 2 . It is c lear that sourc el adiatesdirect ly hrough he perture ormedof dgesE 2 , Fig. 5(a ) . Other d i f f rac ted rays n te rsec t he oppos i te

    the upper f la re again bu t pass o ut the aper ture wi th-

    out ur the r n te rac t ion .Obvious ly , hedi rec t n te rac t ionofthe ays i f f rac ted rom W 1 wi t h pe r t u re A2 (F i g . ( a ) )can be re in te rpre ted as source W 1 in f ree space i l lumina t ingth e s l it shown in Fig. 5(b). The rays reflected from the low erf la re (F ig . 5(c )) may be cons idered to a r i se f rom image sourceW 4 which i s t ransmi t t ing through an aper ture be tween W 2 ndE, (F ig .5(d)) . n hecomple te aypic ture ,sources W 1 n dW 2 adia te hrough aper ture E l E 2 whilesources W3 a nd W4radia te hrough aper tures E2W2 and El W 1 , respectively. Nosingle source aperture nteract ion sat isfies he bound ary con-dit io ns alon g the flare . However, the aggregate does satisfy th eboundary ondi t ion . For example , ons ide rheayromW 1 o E2 shown in F ig . 5(b) . S ince the sc reens an absorber,on l y he nc i de n tpor t ion of theU A Tsolution isnecessaryt o obta in the d i ff rac ted f ie ld of edge E,. However, th e diffrac-t ion of the ray from W4 o E2 (Fig. 5(d ) ) by he a bso rb i ngscreen wil l combinewi t h hea bove osa t i s fy hebounda rycondi t iona long ine W 2 E 2 . (The nte rac t ionof W1 with W 2does not involve an absorbing screen.) The ray diffracted fromW 2 a s no subse que n t i ma ge a nd , a s suc h , t he c ompl e t e UATwedge iffractionxpressionmust e sedor this finalin te rac t ion . This fac t is depic ted by the nonabsorbing w edgeinserted in Fig. 5(d).C. Narr ow Flare Horns ( Q

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    6/10

    246 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, V O L . AP-30, NO. 2 , MARCH 1982

    w. ./31

    .

    31

    - - -- - -- -& . ...

    (dFig. 6. Modeling of small flare angle horn withmany image sources.Sources W , and W4 rad ia te th rough slit E 1 E 2 in absorbing screen,(a) a n d (b). Source W 5 radiates through the narrow aspect slitshow n in (c). (Sources 2 , 3 , and 6 ar eanalogous to 1 , 4, and 5 ,respectively.)cons ider ing he s impl ic i ty of he mode l . The predic ted curvefo l l owsheme a su re desu lt very closelyver th e range0 < 9 < Q. F o r I9 >@,he predic t ion shi f t s upw ard -2 d Bf rom hemeasured esul t ,butc on ti nue s t o fo l l ow he obes t ruc ture . A possible explana t ionfor h i sbehavio r is that nthe region 0 > @ .the d i rec t rays f rom the four image sourcesn o longer exis t and only he d i ff rac ted rays f rom E l and E 2are significant . As discussed earlier, waves incident up on edgesE l and E2 f rom the throa t a re incorr ec t individua lly a s E l , 2arewi th in he rans i t ion egionsof he hroa tdiffract ion.Howe ve r , t o hee x t e n t h a t E l are in the ar ie ld of th et h roa t ( r S a ) , the singulari t ies should izearly cance l . The re -sul t ing rays diffracted from E l ,2 are typica l ly too la r ge , w hichwou l d ac c oun t fo r he +2 dB shi f t . This hypothes i s i s borneou t by the nex t case in which Q a n d a are unchanged but Li s inc reased. T he shi f t obse rved s smal le r a s expec ted .

    Case 2: Fo r th e s eco nd case, the relevant parameters areQ = 17.5' ~ = f l

    ka = 3 R = 0.k l , = 145 .85 ('5)

    The obe s t ruc tu re of the pa t te rn i s now even more nvolved(compare F ig . 7 and F ig . 8) . Again th e agree me nt in the mainbeam region (I9 < 9) i s exce l lent , but a sh i f t (smalle r than thefirst case) is again observed fo r 0 >9. Some disagreement wi ththe lobe s t ruc ture occurs near 0 = 60". No explana t ion for th i sbehavior is available,butotheranalyses [ 1 1 , [SI have alsofailed to recover this po rt ion of t he pa t t e rn .

    s8 , - ALCULATEDMEASURED(-5 B I !- . . _.m0 9

    W

    250 50.0 780 1w.oe ( d 4

    Fig. 7 . For hornandexcitation of (24) compositepattern of horn(in dB) is compared with m easured values fo r 0 < 90". (F or clar i tymeasured values are shifted down by 5 dB.)

    C A L W U T E DMEASURED( - 5 dB )

    S G 50.0 750 loooe ( d e g )Fig. 8. For horn an d excitation of (25) composite pattern of horn (indB) is compared w i t h measured values for 0 < 90" . (For clar itymeasured values are shi f ted do\vn by 5 dB.)Case 3: The th i rd case (F ig . 9 ) i s qui te d i f fe rent from t h e

    p re c e d i ng t wo . The ho rn a nd mode pa ra me t e r s a re@ = 0 . 6 " . r = + l

    kQ = 3 n = 0.k L = 18.94 (26)

    For such a na rrow horn , the re a re s ix image sources . S ince kLis so. small , the ape rtu re effect ively emoves the geomet r ica lopt ics cont r ibut ion of the throa t sources in the fa r f ie ld . Onem i g h te xpe c t hepa t t e rn os h o wan overal l shift ince heedge diffracted rays are dominant and E l , 2 are typical ly largeby a co ns tant ac tor . Reca l l heshi ft in Fig. 7 fo r 0 > 6.)However,since the c ons t a n t fa c t o ra pp l i e s o a l l angles, theshi f t i s not observed in the dB plot . Accord ingly , the pred ic tedresults ol low hemeasuredvalues nd ecover thepa t t e rnlobe structur e very well. (It is l ikely tha t the abso lute value ofth epredicted result is large by some cons tant value.) A pos-sible way to circum vent this difficul ty is t o pe r fo rm a r i go rouscascadingof the uni form hroa t f ie lds wi th he aper ture A 2using the appro ach of Lee and Boersma [ 1 6 1 .

    Case 4: The p re v i ous fou r e xa mpl e s ha ve l l involved a TE Mgui de mode . The a pp roa c h t a ke n i n [ 1 1 , d o m i n a n t h o r n m o d eanalysis, is suff ic ient t o desc r ibe the f ie ld in the throa t regionfor the first three cases in the region I9 < @ (pa r t icula r ly t rue

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    7/10

    L E E : A N A L Y S I S OF R E C T A N G U L A R H O R N A N T E N N A S 247

    25.0 50.0 1 8 08 deg)

    I

    9 . For horn and exci ta t ion of ( 2 6 ) compos i t e pa t t e r n o f hor n (indB) is comparedwithmeasuredvalues or 0 < 90 . (For clari tymeasured values are shiftedby 5 dB.)3) . For larger $ horns , the throa t pa t te rn can devia te8 < Q. Higher orderalsoproduce hroa tpat ternswhichareconsiderably

    T h e h o r n in case 4 is both large Q and the inc ident mode i se fi rst TM mode. T he param eters employed are l isted below.Q = 2 5 O ~ = + 1

    ka = 10 n = + I .

    mo de angle Q1 , fo r t h i scase is roughly 9.Reca ll tha t the TM, mode i s ant i symmetr ic in x . As a re -the far fie ld of the thr oa t is zero in the direct ion 8 = 0.e overal l pat tern (Fig. 10) is c learly anisotropic in the region< Q a nd , in fact , fol lo ws he far-field pat tern of the fou rRecall that he Kellerpat terns of W l 2 are ingular a t= Q1 but hat hes e singulari tiescancel n the farf ie ld toce a finite result 181. For th i s cance l la t ion o occur ne f a r fie ld of the sl i t , the geometrical optics factors of the

    direct contributions of sources W1 and W 2 ,be dentical orsources W l nd W2. or a TEM (TMo)(G o = 0), the sources an d the singularities of the fieldssymmetr ic wi th respec t to t he z-axis. The modi fying func-(28) are utom atical ly dentical t B = 0 ( f rom hehorn) . For t he TEM mode , he re fo re , healwayscancel n he ar ield of the lit.Suchon does not occur fo r h igher order mod e exc i ta t ion .hemodeangle Qn # 0 fo r n # 0. As aresult , hepair ofies which must cancel in the dire ctio n Q, = 8 is n osymmetr icwith espect t o the z-axis. Themodi fying( 2 8 ) i s not , n genera l , he same for source W , and

    W 2 in thedi rec t ion 8 = I f,however , we choose tomi ts of (28) when Itl 2 z I>M and should the transi t ionof edges E l ,2 not overlap over some range of angles,modi fying ac tors (28) areuni formly1over ha tofangles. If theshadowboundarysingularities of W 1nd W 2 ieentirelywithin hislit egion, hesingularities

    08(deg)

    Fig. 10. F orhornandexci ta t ion of (27) compos i t epat ternofhornexhibits a significant off-axis peak. In cont ras t , pat tern of horn de-scribed in (30) is a l so plot ted above and exhibi t s an on-axis peakand low sidelobe levels. See text for d iscussion.againcancel n he ar ield of thesl i t .T hecondi t ionsde-scribedabove orwhich he ingularities of ahigherordermode are manageable may be expressed as

    (Q- sin- (M/&KC>) >9, PI + 0. (29)F or hehornandexci ta t ion l isted n ( 2 7 ) , the hadowboundarydiscont inui t ie sdo ie n he l i t egion sa t i s fy(29 ) ) , and hesingularities do c omb i ne o y i el da ini te ar

    f ie ld for the horn a t = $1 .Case 5: F o rc ompl e t e ne s s , he horn n he p re c e d i ngex -ample sevaluatedagain or the first-orderTEmodeexcita-t ion . The horn and mode paramete rs a re

    Q = 2 5 O ~ = - 1ka = 10 n = f l .k L = 1 5 0 (30)

    The singular behavior of W 1 2 a t B = Q1 emains in the indivi-dual pat terns, but since the horn and mode described in (30)satisfy (29) , t he t o t a l ho rn pa t t e rn p rove s t o be un i fo rm (F i g .10). Again a strong geom etrical optics contribution from hethroa t sources i s present in the main beam. The ho rn , exc i tedby the istT Em o d e ,p roduc e s hee x t re me l ysmoo t h a r -fie ldpa t tern how n.Themain eatures of thepa t t e rn - t heext remely m oothmain eam lendingnto he raduallyincreasing lobestructureatwideangles-are nmarkedcon-tra st o he TM patterns bservedn th e first our ases.Nonetheless, he TE patterns shown n [2] exhibit he sameoveralleatures. A possible xplanation of the extremelysmooth pa t te rn ( in compari son to t h a t for th e TM case)ex -hibi ted by the TE case wo uld be the absence of significant dif-fracted fields from E , 2 . Since the boun dary condi t ion a longthe flare surfaces requires zero fie ld, the nulls of the com positethroa t a t te rn o in t o E , , 2 . Accordingly ,he iffractedfields ( k - -order) are negligible, a n d essential ly the geometri-cal beam from the throa t domina tes in the forw ard d i rec t ion .E. Discussion of Alte rnate Approaches

    The comprehensive analysis of the horn radiator providedby Yu et ~ l .1 ] did not employ the mult iple image model of

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    8/10

    248 IEEE T R A N S A C T I O N S O N A N T E N N A SN DR O P A G A T I O N , VOL. AP-30,O. 2 , MARCH 1982t he f i e l d s in t he f l a re r e g i on o f t he ho rn . In s t e a d , t he y c hoset o c onc e n t ra t e upon t he E-p l ane o f na r row se c t o ra l ho rns fo rwh ich the fie lds in the flare region can be well-represented byt h e d o m i n a n t m o d eof a corner reflector.T hedominantmode na lys i s epresents he ie ld n heflare region as arising from a single sour ce si tuated at the apexof a comer re f lec tor . The source of approach [ 1 ] is an so-t ropic radia tor which l lumina tes aper ture A , to p r o l i d e t h efar-fie ld p at ter n. T he val idi ty of this approximation rests uponthecollect io n of "throat ources"being ufficiently closelyclustered to merge in to a single l ine source. This merg ing willa lsodependupon hewaveguidemodeand t spolarizat ion.Th e degree t o which this condit ion is sat isfied in several prac-tical cases is illustrated in [ 11 . The domi na n t mode a pp roa c hcan recover he pa t te rn of appropr ia te horns and exc i ta t ionsqu ite convincingly. How ever, n he case of wide flare horn s,h igher order m ode exc i ta t ions , or the o the r pola r iza t ion , thedomi na n t mode a pp rox i ma t ion ca n bre a k down .At he cos t of re lat ively i t tle ncrease n complexity, hemul t ip le mage approach can ikewise eva lua te he pa t te rn ofthese narrow flare horns, and offers ncreased general i ty. Forexample , he pa t te rns n F igs . 7-9 requi re e i the r four or s ixsources .Tha t s , ( 2 1 ) mus tb eeva lua tede i the r ouro rs ixt imes . In re turnfor h i sext racomplexi ty , h i sapproach re -moves theassumpt ion ha t hefie lds n he flarecanbeap-p rox i ma t e d by t he domi na n t ho rn mode . The mul t i p le i ma gemode loffersgreater lexibilitywi th espect to he nc id entm o d e ( n , 7). The image ource osi t ions nd eomet r ica lrelat ionsi.e., athengths)emain nchanged.Onlyheindividual p at ter ns of the set of image sources need change t oaccommodate the new mod e rays inc ident upon W 1 2 . ( S o m eres t r ic t ions on the inc ident mode must be appl ied when in te r-ac t ion wi th the aper ture A 2 i s in t rod uced. ) Unl ike the domi-nant mode ana lys i s , the mul t ip le image mode l depends s t ronglyupon a l l threehornparamete rs : a, L , 9. Inpart icular, hen u m b e r o fmages an ecome ui tearge or smalllareangles. Acco rdingly, this mod el is bet ter sui ted for large flareho rns .

    V. R E F L E C T I O N C O E F F I C I E N TThe modal fie ld propagating within he guide undergoes are f lec t ion upon encounte r ing aper ture A l . This throa t re flec -t ion is a unctio n of the horn-waveguide unctionand nde-pe nde n t of L. The f ie ld t ransm i t ted in to the f la re region i s inturn re f lec ted a t aper ture A , . This reflected fie ld couples t oth eeflectedwaveguidemodes .hesew oontributions

    provide thedomi na n tbehavior of the o ta l e f lec ted ie ld .I t is possible to describe hese nteract ionsus ing he am emultiple-imagemodel of the horn-w aveguide unction as wasemp loyed n he far-fie ld pat tern alculat ion.Comple tede-tails are presented in [ 71.A . Formulat ion

    In thecalcula t ion of the eflect ion a t he hroa ta nd hecouplingof he ie lds eflected rom he lare dge to heguide, we make use of he ray-to-mode conversion factor pu tforward by Lee [ 141 for parallel-plate waveguides. Th e ampli-t ude CT",'' of the re f lec t~ed mode m excited by a unit amp li-t ude r a y (mode n ) nc ident upon one edge of the gu ide is given

    byCTn2J= [ X,J9k ' ) + 7x , v ( \ k y ) ]

    [f(-n + @ n ) f ( ~ @m)l[ 4 ~ , a co s @,I - (3 1)

    9' 8 = -(n -@,,, t n) f o r w = 2. ( 32 )He re, he first bracket is the famil iar Keller diffract ion coef-f ic ient l inking the inc ident mode ray and the re f lec ted mo deray.The econd racke t onta inshemodi f ica t ion ohediffract ionoeffic ientdescribedarl ier)risingromheproximi tyof hesecondhalf-plane whichmakesup heguide .The final bracket is termed the ray-to-m ode conversion factor.Within this factor, the term E,= takes the value 2 f o r m = 0 ,and1,otherwise .The ac tor ka represents hegu i deha l f -width nd Qr n is themod e angle ssoc ia tedwi thmode m.Stric t lypeaking,heay- to-modeonvers ionactorw asderived fo r the pa ra lle l-p late guide as was the mo di f ic a t ion tothe iffract ion oeffic ient . orhe aral le l-platetructure(w = 2 ) , ( 3 1 ) i s dent ical to he exac t so lu t ion obta ine d bythe Wiener-Hopf techn ique.T hegeneral izat ion of (31 ) o he h roa t e f l e c t ion o f ahorn i s obta ined by the subs t i tu t ion X + x , Lee has shown[ 141 ha t fo r the flanged waveguide (LC = 1.5) the above sub-s t i tu t ion agrees to order k- ' l2 with he results obtained byothers. In addit ion, Jul l [61 has obta ined the fo l lowing throa treflect ion coeffic ient r r y matching the dominant guide andhorn mode s a t t he t h roa t

    H , (* )(kb)-jH,(Z)(kb)H l ( 2 ) ( k h )+ jHO(*)(kh)rT = ( 3 3 )

    where b = a /s in @.Since he coordina te surfaces over whicht he ma t c h i ng oc c u rs on l y c o i nc i de fo r Q, = 0 , (3 3) i s val id fornarrow la reho rns .Fur the rmo re , 33) s e s t r ic ted o TEM-to-TEM re f lect ion . For n = m = 0 and in the l imi t q5+ O(w+l ) , (31) and (33) provide the ame ef lec t ion oeff ic ient(rT= CToo) .Accord ingly, we a ccep t the generalized form of(31) as the e f lec t ionc oup l i ng rom nc i de n tm o d e n t o r e-f lec ted m ode m for arbi trary flare angle @.In a dd i t i on o he h roa t e f l e c t i on , he e t u rn rom hef la ree dge sente rs n to he o ta l e f lec ted ie ld .T hemul t i -bounce ray fie lds originat ing at wedges W1 , are representedas arising from the same series of ima ge sources as employ edin the far-field patterncalculat ion.Here , he i rs tbracket isthe familiar Keller diffract ion coeffic ient l inking the incidentmo de ray and he eflectedmode ay .The econdbracketc on t a i ns hemod i f i c a t i on o hed i f f r a c t i onc oe f f i c i e n t de -scribed earl ier) arising from the proxim ity of the seco nd half-p lane . S ince the horn i s symmetr ic , i t i s suff ic ient t o cons iderthe re turn produced by jus t one edge as the magni tudes of there turn of the two edges a re ident ica l . Each of the N sourceslaunches a ray to he edge n ques t io n . The f ie ldarriving atth ee dge romsource p fo ra n nc i de n tm o d e n i s denotedZ P " . The f ie lddiffractsas if from he edge of an absorbingscreen. T he edge-diffracted ieldmay ollowan y of N pa t hsback to ntersectwedges W1 , 2 andc oup le t o he e f l e c t e dmodes. Each of these N pa ths corresp ond s to the s tra ight -l ine

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    9/10

    A N D LEE: ANALYSIS OF R E C T A N G U LA R H O R N A N T E N N A S 2490 5 I I I

    - ALCULATED0 0 MEaSUREDk 020-

    wuz2 0 1 5 -w

    I I I7 8 9 10 I

    X ( C dFig. 11. Magni tude of reflection coefficient of horn described in ( 36)versus free-space wavelength of excitat ion.f rom t he e dge t o one o f t he N image sources . The cou-of the f ie ld f rom source p ba c k t o sou rc e q is given bye Keller diffraction coeffic ient a t th e ed ge a nd s denoted byThe ray propagat ing toward source q will cou ple to th e

    d mode a c c o rd i ng o (31 ) wi t h an e f f i c ie nc y de no t e dC P . In al l , there are N. t e rms c on t r ibu t i ng t o t he r e t u rn

    hr N( 3 4 )

    ,= I p = lz t o m o d emgiven by

    ern, n = cTm, + cFn l , ( 3 5)The above resul t i s for the twodim ensiona l horn . To incor-the threedim ensiona l na ture of the phys ica l horn , theof the mod e in the second principal plane (at rightes to he f i rs t ) mu st be inc luded.T heprojec t ionof herays n h i s p lane i susedas the nc ident f ie ld and he

    For asec tora lhorn , hemode l ingc ompose d of a wo-dimensiona lhornandaparal le lplateThecomplete eflected fieldof the hree-dimen sionall horn is givenby an expression of t h e f o r m ( 3 5 ) a n d

    umerical resultsThe preceding analys i s has been mplemented for a rangef re que nc y romne a rc u t o f f o heonse t of the econdmode in the o l lowing ec tora lho rn -wa ve gu i de

    q5 = 16.5, 2a = 3.404cmL = 36.16m , 2c = 7.214 c m .3 6)

    e first hreevariables havebeen definedearl ier; c i s heto t he p l a ne o f a and. The i nc i de n t mode i s TEl o . The p ro j e c t i ons o f t h i s mode

    to th e tw o principal planes of he physical horn prov ide aM inc ident f ie ld for he wodim ensiona l horn and a TE,de for the paral le l-plate guide.

    In F ig . 11 , the abso lute va lue of the re f lec t ion coeff ic ienti sp l o t t e dversus he reespace wave lengthof the nc identf ie ld . The re f lec t ion vanes f rom roug hly 20-5 pe rcent and theoverall discrepancybe tween hecalculatedandobserved e-flect ion is 2-3 percent (da ta f rom Jul l [ 6 ). In pa r t icula r , theoveral l pat tern of reflect ion peaks and val leys is bro ug ht outqu ite well. We observe th at he fi t is bet te ra t he h i gh fre -quency end of the graph as expec ted . In addi t ion , the ca lcu-la ted resul t s a re typica l ly too la rge as was the case in the fa r-f ie ld pa t te rn . In the la t te r case , the d i sc repan cy was asc ribedto the Ke l le rdiffracted ays rom he hroatwedges .Thesesamedifficult ies houldexplain he l ightoverest imation inthe ca lcula ted resul t .Theagreementobtainedby hemult iple-imagemode loft h e h o r n in the far-fie ld pat tern predict ion and the reflect ioncoeffic ient calculat ion speaks well for th e val idity of the raypic ture of the horn . The disc repanc ies encounte redn t h e U A Tanalysisapparently temnot us t rom hemode lbut alsofrom the difficul ty cascading the fie lds of the f i rs t aper ture Awith the second aper ture A 2 . How e ve r , f rom t he unde r s t a nd -in g of the single aperture [ 8 ] , i ts associated transi t ion regionsand dominant f ie ld behavior, i t i s possib le to produ ce mean ing-fu l r e su l t s fo r t he s e r i e s o f a pe r t u re s i n t he ho rn de sp i t e t heinherent cascading difficul ty.

    REFERENCESJ . S . Yu : R .C .Rudduck, nd L . Peters , J r . , Comprehensiveanalysis for E-plane of horn antennas by edge diffraction theory,I E E E TrUJlS. Antennas Pr op ag at. , vol. AP-1 4, pp. 138-149, 1966.C . A. IMentzer, Analysis nddesign of high-beam fficiencyaperture ntennas.Ph.D. issertation.TheOhioStateUniv.,Columbus,1974.Y . Ohba. On the radiation pattern of a cnrner reflector finite inwidth, IEEE Trans.AntennasPropaga t . , vol.AP-12,pp. 127-132,1963.IM. A . K . Hamid,Near-fieldransmissionetweenornantennas ,Dept .E lec .Eng. ,Univ.Toronto,Toronto,Canada,Res . Rep. no. 43, 1966.P . M. Russo, R. C. Rudduck. and L. Peters, J r . , A method forcomputing -planeatternf horn antenna s,E EE Trans.Antennas Propagar. , vol. AP-1 3, pp. 219-244, 3965.E. V . Jull. Reflection from the aperture of a long -plane sectoralhor n , IEEE Trans . Antennas Propa go t . , vol . A P -20 . pp . 6 2 4 8 ,1972.R.C .Menendez nd S. W . Lee. Uniform symptotic theoryapplied to apertu re diffraction , Dept. Elec. Eng., Lniv. Illinois,Urbana, Tech. Rep. EM76-9, Aug. 1976.

  • 7/27/2019 Analysis of Rect Horn Ant via Uniform Asymptotic Theory

    10/10

    IEEE T R A N S A C T I O N S O N A N T E N N A S A N D P R O P A G A T I O N , VOL. AP-30, NO. 2 , MA R C H 1982

    R. C . Menendez and S . W . Lee, On the role of the geom etricalopticield in apertureiffraction, IEEE Trans.ntennasP ro p a g a t . , vol . AP-25. no. 5 , pp. 688-695 , Sept . 1977.J . B.Keller.Geom etrical heory of diffraction. J. O p r . SOC.Atner. . vol. 5 2 , pp. 116-130, 1962.R. M. Lewis and J . Boersm a. Uniform asymp totic theory of edgediffraction, J . Math. P tzys . , vol. 10, pp. 2291-2305, 1969.D. S. Ahluwal ia ,R . M . Lewis,nd J . Boersma,Uniformasymptotic theory of diffraction by a p lane screen, S f A M J . Appl .M a t h . , vol. 16, pp. 783-807, 196 8.J . Boersma and P . H . hl. Kersten, Uniform asymptotic theory ofelectromagnetic diffraction by a plane screen, Dept. Math.. Tech.UniversityofEindhoven,Netherlands, inDutch)Tech.Rep..1967.S. W. Lee and G. A. Deschamps. A uniform asymptotic theory ofelectromagneticdiffraction by a urvedwedge. l E E E Trans.Antennas Propaga t . , vo l . AP-24, pp. 25-34, 1976.S . W . Lee, Ray theory of diffraction by open-ended waveguides.I . Field in waveguides, J . M a th . P h y s . , vol. I I , pp. 2830-2850.1970.Applications, J . M a t h . P h p s . , vol . 13, pp. 656-664, 1972.S. W .Leeand J . Boersma,Ray-opticalanalysis of fieldsonshadow boundaries of two parallel plates,J. M a th . P h y s . , vol. 16,1 7 4 6 1 7 6 4 , 1 97 5.R . C . Menendez and S . W . Lee,Near ield of th eopen-endedparallel plate waveguide, Wave Morion , pp. 239-243, Jan. 1979.

    - Ray heory of diffraction by open-ended waveguides. 11.

    Ronald C . Menendez (S75-M77) received theB.S.degree nphysics romWashington Uni-versi ty,S t.Louis ,MO , in 1971,and he M.S .and Ph.D. degrees in electrical engineering fromth eUniv ersity of Illinois,Urbana, in 1973an d1976, respectively.Following earlong ostdoctoralppoint-ment with he Coordinated Science Laboratory,research nterestshave ncludedmagnetic levi-tation of high-speedgroundvehicles, ray tech-niques, lectrostat ic ffects urroun ding EHV power ines, ightning-induced t rans ient mi t igat ion, and inves t igat ions into hu man physiologicalef fects of electrical st imuli .Dr . Menendez i s a member of Tau Beta P i , S igma Xi , and Phi KappaPhi .

    %9 6 . heoinedel lelephoneaboratories.is

    Shung-Wu Lee (S63-M664M73-F81),orhotographndbiographyleaseeeag e17fhe May 1980ssue of thisTRANSACTIONS.