Вы находитесь на странице: 1из 7

436

IEEE TRANSACTIONS ON ANTENNAS AKD PROPAQATION, VOL.

AP-20,

NO.

4, JULY 1972

condition. Also, there is a noticeable improvement i n the pattern structure a t scan angles near en&e. An isotropic element array is used for comparison with the data of Lo et al. [SI and the Hansen-Woodyard condition; however, the opt,imization of a practical array should include mutual coupling effects Khich may introduce differences from the theoretical results illustrated in this paper. Phase optimization is useful for electronically scanned arrays when varying the phase with fixed amplitude. For the case of gain optimization, the improvement is most pronounced near endfire scan.

ACKKOWLEDGMEKT
The authors wish to thank Dr. S. A I . Sanzgiri for his helpful discussions.

REFEREKCES
[I] A. I. Uzkov,. A; approach to the problem of optimum directive antenna deslgn, Dokl. A M . Nauk SSSR, vol. 53, pp. 35-38, June 1946. [2] E. N. Gilbert and S. P. Morgan, Optimum design of direct.ive

antenna arrays subject to random variations, Bell Syst. Tech. J., vol. 34, pp. 637-663, May 1955. [3] A. Bloch, R. G. Medhurst, and S. D. Pool, A new approach to the design of superdiiective aerial &Grays, Pric; Znst. Elec. Eng., vol. 100, pp. 303-314, Sept. 1953. [4] M. Uzsoky and L. Solyrnar, Theory of superdirective linear arrays, A d a Phys., vol. 6, pp. 1.85-204, 1956. 1 5 1 E. I. Krupitsky, On the maxlmumdirectivity of antennas consist.1ng of &Crete radiatom, Sov. Phys. Dokl., vol. 7, no. 3, pp. 257-259, Sept. 1962. [SI Y . T. L o , S. W. Lee, and Q. H. Lee,Optimization of directivity and signal-to-nolse ratlo of an arbitrary antenna array, Proc. ZEEE, vol. 54, pp. 1033-1045, Aug. 1966. [i] J. K. Butler and H. Unz, Beam efficiency and gain optimization o f antenna arrays with nonuniform spacings, Radio S a . . , . vol. 2, no. 7, pp. 711-720, July 1967. [ti] 8. M . Sanzgiri and J. I ( . But.ler, Constrained optimization of the performanceindices of arbitrary array antennas, IEEE Trans. Antmnas Propagat., vol. AP-19, pp. 493-498, July 197 1. [9] D. K. Cheng and P. D. Raymond, Jr., Optimisation of array directiv1t.y by phase adjustments, Electron. Lett., vol. 7, no. 18. DD. 562-554. &Dt. 1971. [ 101 T.&.& Saaty and J. Bram, Nonlinear YaUlmatics. New York: McGraw-Hill, 1964, pp. 53-174. Ill] R. L. Fox, Optimiration MeUlods for Engineering Design. Reading, Mass.: Addison-Wesley, 1971, pp. 51-58. 1121 W. Hansen and J. R. Woodvard. A new Drincidein . . W. directional antenna design, P T O C . ~ I Rvo.. E , 26, pb. 332-345, Mar. 1938. 113) R. E. Collin and F. J. Zucker, Antenna Theory, pt. I. New York: McGraw-Hill, 1969, pp. 156-157.

Cross Coupling in a Five Horn Monopulse Tracking System


WILLIAM M . BRIDGE

Abstracf-A model is developed which explains the observed effects of secondary pattern crosstalk in a five horn circularly polarized monopulse tracking system. The model predicts unstable operation and loss of track for s a c i e n t l y depolarizing targets as well as polarization dependent uncertainties in the angular location of off-axis targets. Thecross coupling is characterizedby two complex coupling coeEcients between adjacent horns in the monoThis characterization is assumed to be valid pulsefeedsystem. irrespective of whether the observedsecondarypattern effects are the result of mutual couplings within the feed structure itself or caused by cross-polarized backscatter from the paraboloidal reflector. A ridge and nonridge polarization terminology is adopted which is physically related to the fact that the error horns in a five horn circularly polarized system are heavily ridge loaded for onelinear polarization and not the other.

I. INTRODUCTION

NU

1;IEROUS possibilities exist for cross coupling in a followfive horn monopulse tracking system. For tmhe ing discussion, cross coupling w i l l be defined as the spuManuscript received December 2, 1971; revised February 17, 1972. The author iswith theMITRE Corporation, Bedford, Mas.01730.

rious coupling of energy from one error channel t o another. It includes t,he effects of mutual coupling between adjacent horns, finite isolation between the orthogonal signals of each dual polarized horn, and the inherentcross-polarized backscat.ter from a parabolic reflector. I n a circularly polarized five horn system, the desired error signals are usua1l;v obtained bytakingthe difference between the horizontal and vertical components of the circularly polarized input signal andthen combining t.hese H and V difference signals in a 3 dB quadrature hybrid toproduce the desired error signal output as shown in Figs. 1 and 2. Consequently the feed basically receives H and V signals and, in this paper, any cross coupling which exists d l be considered as between the H and V components of the signal. The previous definition of cross coupling is too general to be very useful in characterizing a particular network. Isolation is more definit,ive and refers to measurements a t specific ports of .t.he network (notapertures),The insertion loss between aaimut,h ridge (AR) and elevation nonridge (EN) ports of Fig. 1 would be defined as ridgeto-nonridge isolation. Crosstalk, on the other hand, refers

BRIDGE: FIVE HORN YONOPULSE TRACKING SYSTEM

437

5=$= -$
I
1 1

Fig. 1. Ridge-to-ridge crosstalk, pedest.al -42> target AZ.

Fig. 2. Nonridge-to-nonridge crosstalk, pedestal AZ

> target

AZ.

to pattern-t,ype measurements associated with the reception of signals at. the apertures. Crosst.alk may be termed as primary or secondary depending on whether or not, a sec0nda.q reflector is included. The following discussions willbe limited to the results obtained from secondary pat.tern crosstalk measurements performed on an 84 dish with an F / D rat.io of 0.375 and mounted with conventional azimuth and elevation axes.

C. Ridge-to-Ridge Crosstalk
Thisterm refers to crosst.alk between t,he vertically polarized azimuth channel a,nd the horizontally polarized elevation channel. Forexample, this crosstalk could be measured by noting t,he spurious signal present in the ER channel when the antenna is rotated and illuminated as described in Section I-A.

-4.Cross-Polarized Crosstalk
Cross-polarized crosstalk refers to asingle axis (azimuth or elevation) and is a measure of the relative response of the H and V channels to a cross-polarized input signal. For example, azimuth cross-polarized crosstalk could be measured by observingthe level of spurious signal present at port AN (Fig. 1 or 2) when theant,enna is rot.at,ed about a vert.ica1axis while it is illuminated by a vertically polarized source located in the farfield.

D . LVmridge-to-Konl.idge Crosstalk
This t,erm refers to crosstalk between the horizontally polarized azimut,h channeland the vertically polarized elevation channel. For example, t.his crosstalk couldbe measured by noting the spurious signal in the E N channel when the antenna is rotated as i n Section 1-8,while it is illuminated by a horizont,a,llypolarized source located in the far field. It should be pointed out that both ridge-to-ridge crosstalk and nonridge-t,o-nonridge crosstalk could also be obt.ained by rotating t.he antenna about a. horizonta.1 axis while illuminating it, with an appropriately polarized farfield source. I n a welldesignedfeed, the cross-polarized crosstalk should bea t least 20 dB less than theresponse to correctly polarizedsignals. I n addition, it can bereasoned from symmetry arguments t,hat ridge-t.o-nonridge crosstalk should be independent of angular offset in the principal

B. Ridge-to-Sonridge Crosstalk
This term refers to crosstalk betxeen azimuth and elevation channels of t,he same polarization. For example, ridge-to-nonridge crosstalk could be measured by observing the level of spurious signal present, in the EN channel m-hen the antenna is rotated and illuminated as described in Section I-A.

438

IEEE TRLLNSACTIONS ON ANTENNAS AND PROPAGATION, JULY

1972

RIDGE-TORIDGE
1 2

2 3 4 5 6 7 8

VERT E L RIDGE-NOW -RIDGE


9

0 0 7

TIME

IN SEC. (2 M w / S )

Fig. 3. Vertical azimut,hscan.

output porta.nd 180"when directed away from the output port. 5) The direction of the cross-coupled vector is determined by extending the vector representing the signal in the principal axis about a 90' arc of a circle into the horn which receives the cross-coupled energy. 6) The feed and its associated components are assumed to be rotationallysymmetric and ideal with the exception t.hat the error horns need not have thesame gain for ridge and nonridge excitation.

111. DERIVATION OF CROSSTALK EQUATIONS


a.xis and that, the coupled signals should cancel in the output port of the difference hybrid. This is verified by the data shown in Fig. 3, which includes ridge-to-ridge crosst.alk as well, The abnormally high secondary pattern crosstalk of this antenna was substant.iated by primary pa.tternand isolation measurements, and may not be representative of other five horn systems. The measurement, of cross-polarized crosstalk and ridge-to-nonridge crossta.lk requires t,he removal of the 3-dB quadrature hybrids to isolate the H and V channels. Ron-ever, having once established that t,hese crosstalk levels are negligible, the remaining crosst.alk can be evaluat,ed without, removal . of these hybrids. The following general expressions for crosstalk are derived from the assumption that any arbitrary input signal canbe expressed in terms of leftrhand and right-hand circular components, and thateach of these can be further divided into H and V components. An LC signal of magnitude ( A ) is always assumed present. Without loss in genera.lity a "viewing time" may be chosen when the instantaneous orientation of this vector is vertical. The right-circular component is then assumed to haveany magnitude ( K A ) and an inst,antaneous orientation angle f relative to the LC component, L f = L LC - L RC. The vertical and horizontal components of the left- and right-circular input signals can be expressed as follows:

11.

MODEL

DEVELOPMENT

LC (vertical) A

AL O

The observed effects of crosstalk can be predicted quite accurately with the a.id of the model shonn in Figs. 1 and 2. This model assumes a ridge-to-ridge coupling coefficient (0L 7 ) and a nonridge-t,o-nonridge coupling coefficient (cxL6) between adjacent horns. Before these c0efficient.s can be evaluated, it is necessary to determine the phase constants of the system. I n general, the only phase informat,ion available is the IF phase difference between the sumchannel andeither of the twoerrorchannels (AAZ - Z) and (AEL - 2 ) . The RF phaseconstants . +AR and +EN can be determined by noting the change in difference phase for vertical and horizontal input signals from a boresight source a t fixed angular offsets. The IF phasing of the system BEL, ez, and eAz can be determined within a constant by noting the asymptotic phase differences during azimuth and elevation scans. The crosstalk coupling coefficients 0 L y and a L 6 ca.n be obtained from principal axis scans of a linearly polarized boresight source a.s shown in Figs. 1 and 2. The following arbitrary conventions were selected in establishing the modelshonm in Figs. 1 and 2. 1) An LC received signal corresponds to the clock-wise rotation of electric vector. 2) A + A Z offset (pedestal A Z > target A Z ) causes larger signa.1 in the right horn of the diagram and produces a 90" (AAZ/Z) phase. 3) 4+EL offset (pedestal EL > target EL) causes larger signal inthebottomhorn of the diagram and produces a 90" (AEL/Z) phase. 4) The inst.antaneous phase of a signal is assumed to be zero when t.he electric vector is directed toward the

LC (horizontal) B
RC (vertical) A'

=jA =

KA L f

RC (horizontal) B'

- j K A L \k.

The center reference horn is assumed tohaveperfect circularity so that it does not respond to the RC component of the input signal. This assumption is supported by thefactthat t.he measured RC cancellation from spherical t.argets wason the order of -30 to -35 dB. The reference signal Z of Fig. 1 can now be expressed as
n

Fora positive azimuth offset (pedestal A Z > target AZ) , the azimuth error signal (AAZ) can be expressed

A A Z (RC) = - j

(" 2 ") K A

L f [ 1 - GL$AR] Leaz
(2)

where R and L represent the unequal voltages induced in right and left horns. The factor G has been introduced to account for the difference in secondary error pattern gain between ridge and nonridge excitation. Using (1) the normalized azimuth

BRIDGE: FIVE HORN MONOPULSE TRACKING SYSTEM

439

t,hen error

becomes
+

v 5 C Z (LC) = - ( R - L) [ z 2
= z (RC)

*
G L '$-4R] L 2
(OAZ

and RC component.s wit.h an instant.aneous Dhme angle = 0". Equat,ions ( 3 ) , (4)and (5), (6) the; reduce to:
=

- 6,)

for +AZ offset (vertical signal, K

1,

Oo)

AAZ

- v% - (R - L)
2

Fora positive azimuth offset the spurious normalized elevation error signal can be expressed as: AEL

for +EL offset (vert,ical signal, K

1,

0")

- (LC)

l 6
2

AEL
I: =

fi
d

- (R - L)

j y( B - T ) ~ L [ + E N

+ (eEL

e,)]

(9)

AEL v 5 (RC) = - ( R - L )

The gain factor G, which i s principally t.he result, of differentilluminationfunctions for ridge and nonridge polarizations, does not appear in the previous expressions because it is assumed that the crosstalkisprincipally caused by mutual coupling between adjacent error horns. I n any event, the effect of differential gain could be included in the ma.gnitude of a and 6. The analogous expressions for a positive elevation offset (pedestal EL > target EL) are

This condit.ion is particularly useful for evaluating the coupling coefficients L Y L 6 and P L y as shown in Figs. 1 and 2. A similar situation exists for horizont,al signals which may also be used for evaluating the couplingcoefficients. The LC only case is represented by the first expression of each set. It is intereding to note that in general the magnitude of the spurious crosstalk relative to the principal error signal is less for LC signaLs than for the linear H or V signal which maximizes the crosstalk. This can be seen from the following LC ratios for AZ and EL offsets: for +AZ offset ( L C )

-(RC) = j -

AEL

fi
2

( B - T)
For G = 1 and ' $ a ~ and EN negligibly small, the only way the LC spurious crosstalk could be as severe as the worst. of the H or 1 ' crosstalks would be for the ridge and nonridge coupling coefficients to be equal and oppositely phased (a = a,y = 6 T). If both azimuth and elevation offsets are assumed, ea.ch error signal is a combination of a true error signal and a spurious crosstalksignal. The magnit.ude of t,he spurious crosstalk signal dependson the phase constants of the syst.em as well as the polarization cha.racteristics of the input signal, and its sign depends on t.he direction of t,he offset in the otheraxis. Thus for certain offset and polarization conditions, the tracking system could become unstable and drive in the opposite direction. This stability condition d l be examined only for the case of negligible small '$AR a.nd +EX and G = 1 (equal ridge a.nd nonridge gains).Underthese condit,ions t,he normalized azimuth

-( L C )

AAZ

-j-

v 5
2

( B - T)

- (RC) = - j -

AAZ

d2
2

(B- T)

(6)

A few special cases are w0rt.h examining in detail. A pure vertical signal ca.n be considered as cont,aining equal LC

440

IEEE TRBNSACTIONS ON ANTENNAS AND PROPAGATION, JULY

1972

and elevation error signals become

4) Ridge-to-ridge crosstalk referenced tothe

6 dB principal error signal. 5) Nonridge-to-nonridge crosstalk referenced 15.5 dB to the principa.1error signal. 6) Error charnel H-to-V ratio (G), ridge 2.5 d B polarizat.ion low. 7) Circular polarization crosstalk (sphere) 12 dB.
The derived model characteristics are as follows: Excess azimuth ridge phase = 8". Excess elevation nonridge phase = 5". Azimuth IF phasing = 97". Reference IF phasing = 20". Elevation IF phasing = 0". Ridge-to-ridge voltage coupling coefficient = 4. Ridge-to-ridge coupling phase = 180'. Sonridge-to-nonridge to voltage coupling coe6cient = Sonridge-to-nonridge coupling phase = 255". Secondary error pattern voltage gain difference for ridge and nonridge excitation = 1.

The normalized driving signals are obtained by multiplying each error signal by t.he sin function of the phase angle between it. and t.he reference signal. Thus

AAZ I @ -

sine=-(R-L)+-(B-T)[~sin6-Osiny I: 2 2
+Kasin (6

I @

A.

+e) + KPsin (y + JE)]


I @
2

(15)

The model --as first evaluated for pure LC returns from a sphere. Equations( 3 ) and (4) indicate that theLC crosstalk level should be 12 dB, and this is precisely the level of crosstalk observed. The theoretical crosstalk has also been evaluated for various linear polarization conditions, using the derived phase constants of the system. The results are plotted along -n-it,h the experimental data in Fig. 4 and the agreement is very close. The experimental data a cos 6 p cos y = tan-' (17) was obtained by rotating a linearly polarized horn located asins+Psiny' on a. boresight tower. The effect of cross coupling on the angular location of This condition is sat.isfied for 2 values of differing by off-axis targets is graphicallydepicted in Fig. 5. The 180"; one resu1t.s in a maximum value of crosstalk while apparent, target position for fixed 3 n1ra.d offsets in azimuth the other results% either minimum crosstalk or a maxiand elevation is plot,ted as a function of orientation for a mum vaIue of opposite sense crosstalk. The sense of the linearly polarized return (I LC I = I RC I). The boresight crosstalk term is determined by the direction of the offtracking 1ocat.ion was first established for the existing sets, and forcombined unit offsets in the "left-up" or "domn-right." direction the tracking system may become phasing. Then, approximately 3 mrad offsets were taken unstable for sufficiently large K. For t.he system to actu- from this particular designat,ion point and data were recordedwhile t.he boresight linearly polarized horn was ally drive off, itappears necessa.ry that both AZ and EL rotated. The data clea.rly indicate the uncertaint,y in offerror signals musthave the incorrect sign. This means axis target location for target. returns having equal LC the magnitude of t.he smaller crosstalk maxima must be and RC magnitudes. greater than t,he true unit error signal. The effect of off-axis target locat,ion for ot,her polarization ratios has been investigated theoretically using the EXPERIhlEhTALVERIFICATION OF A:fODEL model previously developed. For these analyses a ridgeThe model thus derived has been very successful in to-nonridge gain difference of 2.5 dB and ideal phasing predicting the performance of theantenna system for were assumed, various polarization conditions. The following list gives the measured secondary char- ( G ~ R= 0 = GEN), (eAz - e,) = go", (BEL - e,) = 0". acteristics of this particular five horn antenna system. The results are shown in Fig. 6 for four polarizations 1) Linear polarizationnull depths referenced 40 dB (LC, RC = +LC,RC = LC, RC = 2LC), where the cross to t.he sum pattern (sphere). coupling is depicted on t.he basis of normalized error volt2 ) Cross-polarized crosstalk referenced to the 23 dB ages for unit.offsets. The shape of the t,heoretical RC = LC principal error signal. contours is similar to the measured contours shown in 3) Ridge-t,o-nonridge crosstalk referenced to 30 dB Fig. 5. The elliptical nature of the contours is caused by the principal error signal. the gain difference between t.he ridge and nonridge po1a.r-sine
2:
=

AEL

I6
2

-((B

T ) +-(R-L)[-a.sin6+psiny

m.

BRIDGE: FIVE HORN MONOPULSE TRACKING SYSTEN

441

vA A Z

I " ' ! x
180

NORMALIZED-EL OFFSET PHASE = 270"


=TffiORETICAL AMP. Q =THEORETICAL WASE -=MEASURED DATA * EL CHANNELGAIN'1.2 rb,HIGH 4 ZZY
x

nma

4 9

ORIENTATION OF LINEAR INFUT SIGNAL

V E R T .

se I D I u .

ncmu.

IW2W

1"

167.5

*EL

Fig. 7. Composite crosstalk for A2 and EL offset, K = 4.

Fig. 4. Elevation cross coupling for +3 mrad AZ offset.


t EL OFFSET IN

MRAD

- AZ OFFSET
I ; 4

NORMALIZED

-1:z -9 -$

NORMALIZED + A2 OFFSET

2250
-AZ

OFFSET
2 3

+ AZ OFFSET IN MRAD
4

NORMALIZED -EL OFFSET

-\
' 3

v
135'
-21
Y

Fig. 8. Theoretical elevat.ion crosstalk for BEL K = 3.2.

-30") 0,

+ 30",

-EL OFFSET IN MRAO

Fig. .5.

Crosstalk for off axis targets n4t.h I LC

I RC I.

PHASE = 90 NORMALIZED +EL OFFSET

where the magnitude of the spurious crosstalk is equal to the principal error signal. The det.ailed sha.pe of the measured contours in Fig. 5 is not symmetrical and is not in exact agreement, with the theoretical results of Fig. 6. However, at the time of these dat,a, t.he reference cha.nne1 pha.sing prior to themicrowave circular combining hybrid had not been adjusted for optimum reference channel circularity. This misalignment (approximately 15") combinedwith any IF phase misalignment (BAZ,BZ,OEL)could account for much of the observed differences. The bottom contour of Fig. 5 is somewhat unusual and this is attributed to different, ground reflections for H a.nd V polarizations at boresight elevation angles less than 28 m a d .

V. COMPUTER RESULTS
The model equations presented in the previous section ha.ve been programmed on a digital computer so that the effect. of perturbing t.he various parameters can be investigated.Thisprogram ha.s been used toinvestigatethe theoretica.1 stability of t,he antenna system for idea.1phasing and a unit offset in the "down-right" direction. The results for G = 2, K = 4 (RC signal 12 dB greater than LC signal) is shown in Fig. 7. Fig. 7 is a composite plot for a unit offset, in both azimuth and elevation and clearly indica.tes a region of instability for beheen -80" and - 110". The program has also been used to show the effect of IF phasing for a unit azimuth offset with G = E, K = 3.2, (RC 10 dB > L C ) . This dat.a is shown in Fig. 8 and indi-

NORMALIZED- ELOFFSET PHASE'ZW

Fig. 6. Theoretical crosstalk for ellipt,ical polarizations.

izations. If this gain difference could be reduced to zero or compensated by the introduction of pads (2.5 dB) in the nonridge error lines, t.he contours mouldcollapse to straight lines. It should be pointed out t.hat t,he center (RC = 0) as well as the size and shape of the contours is critically dependent on the IF phasing. The dashed lines at 45"/135O are included in Fig. 6 to indicate the point

442

IEEE TRANSACTIONS ON H T E N N A S AND PROPAGATION, VOL. AP-20, NO.

4, m Y 1972

cates Dhe variability in crosstalk with IF phasing. Perhaps the most significant effect is the marked difference in apparent target elevation for the same incoming polarization angle i P . This points up the necessity for fairly st.ringent phase stability if the effect of crosstalk is to be removed by computat,ion. The cross-coupling effect of varying thedifferential elevation phase 4 E N was negligible and this parameter as well as ~ A R should be optimized for best error channel circularity rat,her tha.n minimizationof crosstalk.

RC returns was measured. I n order to accomplish this the absoluteinsertionphase of each receiver channelmust
eit-herbe set to a prescribed value or measured as part of premission calibrations. The most severecrosstalk in the measured antenna syst,em was the ridge-to-ridge variety. It can account for much of the observed 2.5-dB loss in secondary pattern i f the effects of reradiated ridge polarimtion difference gain energyare considered. The analysisis notparticularly helpful in the design of an improved feed structure but does point up the need for extensive secondary patiern measurements to characterize the final device. Future modifications and new feed designs should attempt to equalize the ridge and nonridge secondary pattern difference gains, by reducing the crosstalk levels and possibly by the insertion of fixed pads in the proper channels. A scheme of intentionally introducing additional cross coupling of the proper magnitude and phase to cancel the undesired cross coupling has also been suggested. ACKNO~EDGMENT The author would like to thank Dr. J. Ruze of M.I.T. Lincoln Laboratory, for valuable criticism and discussions, and J. Pearlman, who programmed the equations.

VI. CONCLUSIONS
A model has been developed for the secondary patt.ern crosstalk of a five horn monopulse tracking system. The model w a s adequate for predicting the system performance for arbit,rary incoming polarizat,ions. The most severe effect of crosstalk is to cause angle tracking to become unst.able for sufficiently large cross polarized returns. The .crosstalk also resultsin serious pola.rizationdependent uncertainties in the angular location of off-axis targets. This latt.er effect iscriticallydependent on the system IF phasing but could be removed by computer processing provided the phase difference between incoming LC and

The Behavior o f Electromagnetic Fields at Edges


JOSEF MEIXlUER

Abstract-The behavior of an electromagnetic fieldin the neighborhood of the common edge of angular dielectric or conducting regions is determined from the condition that the energy density must be integrableoverany ~ t domain e (theso-callededge condition). Two cases are treated in detail. 1) A region consisting of a conducting wedge and two different dielectric wedges with a common edge. 2) A regionconsisting of two differentdielectric wedges with a common edge. It is also shown that near such edges, electrostatic and magnetostatic fields will exhibit the same behavior as the electromagneticfield.

The order of this singularity is, however, subject to the so-called edge condition [l], [a]. The edge condition states that the electromagnetic energy density must be integrable over anyfinite domain even if this domain contains singularities of t.he electromagnetic field. In other words, t.he electromagnetic energy in any finite domain must be finite. In the case of a perfectly conducting surface n-ith an edge, one concludes from this condition that near the edge the singular components of the electric and magnetic field vectors are of the order p-lI2, where p is the I. INTRODUCTIOK distance from the edge, while the component,s of the electric and magnetic field strengths parallel to the edge are N THE SOLUTION of diffraction problems, it is ah-ays finite. found that at sharp edges of the diffracting obst.acle In this paper we generalize the preceding result to find the electromagnetic field vect.ors may become infinite. t.hebehavior of the field vectorsnear the edges of dielect.ric and perfectly conducting bodies. We restrict ourManuscript. received January 21, 1971 ; revised January 2 4 , 1972. selves to the neighborhood of points for which there is a T h i s work was supported by A i r Force Cambridge Research Cent.er well-defined tangent along the edge. We may then conunder Contract, AF-19(122)-42. The aut.hor was with Courant Instituteof Mathematical Science, sider the edge as locally straight. Hence it suftices to conNew YorkUniversity, New York, N. Y., on leave from the Institute sider a space filled with wedges of homogeneous materia1 of Theoretical Physics, Rhine-W&falian Institute of Technology, wititha common straight edge. We shall treat in det,ail Aachen, Germany.

Вам также может понравиться