permittivity
•, withe-itottimeconvention,
thefield
consists of the incident and scattered as
Eipi(p,
z) = eikz/p (1)
Ep•(p,z) = -e -ikz/p_
2fo
• • •(•)R'(•:p)
sin(•
d (2)
Ez•(p, z) =- •(•)R(Kp) cos(•z)d• (3)
H4•i(p,z) = e-ikz/(T•p)
+ itoe- -•(•)R' (Kp)cos(•z)d• (4)
whereK = X/k2 - •2, k = 6oX//x•- 2,r/X,R(Kp)
= Jo(•cp)No(•ca) - No(•Cp)Jo(•ca), and R'( ) =
dR( )/d( ). Jo( ) is the zeroth-order Bessel
function,and No( ) is the zeroth-orderNeumann
function.
Herert - X/We,and•i(•)R(Kb) is the
Fourier cosine transform of Ezi(b, z) given by
= fo z) cos(z)az.
In region (H) of permittivitye, the field is repre-
2. Field Representation sented as a summation of the discrete modes:
Considerthe DFES antennaconsistingof a coaxial
line whose outer conductor is circumferentiallyre-
moved and replacedby a dielectricwith permittivity Epli(p,z)
= - E --R•)(•:mp)
am sin (amZ) (5)
m=0Km
• = •0 •r (see Figure 1). Assumethat the top plane
(z = 0) inside the coaxial line is shorted. Let a
transverseelectromagnetic(TEM) wave be incident
from below insidethe coaxialline. In region (I) of EzlI(P,
z) = E R0(•Cmp)
cos(amZ) (6)
m=0
Copyright1997 by the American GeophysicalUnion.
Paper number97RS02203. H6•( p,z) = itoe• -- R b(KmQ) cos (a mZ)
Km
(7)
0048-6604/97/97RS-02203511.00 m=0
2149
wherego(trmp) : PmJo(trmp)+ qmNo(trmp), Pm -d<z<0
andqm areunknowncoefficients
the boundaryconditions,
to be determinedby
t(m -- V'k2 -am2, and
Ezz(b,z)={•
zII(b'z)
otherwise (11)
a m = m rr/d. Applyingthe Fouriercosinetransformto (11) gives
In region(III) of permittivity•0, the radiatedfield
is
• Ro(•cmb)
•1(•)- •m=O
•(-• mm(•)
(12)
EplII(p,
z)= 2yr• -[Eiii(•')Ho(1)'(•p)e
-i•zd•' (8)
where
EzIII(P,
z)= • I /*_Em(O
mo(1)(
-i•z
fp)e d• (9)
Fm(•)
--
i•'(- 1)m[e-i•d __ei•d]
2(•2- a2m) (13)
z)=•iweo
Hq,iii(p, fo•1• •iii(OHo (10) TheH6 fieldcontinuity
(•)'(,•p)e
-i• d•' at p - b gives,
for -d < z < 0,
i
2 Hq,i(b,z) + Hq,i(b
, z) = Hq,ii(b,z) (14)
where
f = X/k0 - •2, k0_ to•/-•e0, andH0(•)( )
is the zeroth-order Hankel function of the first kind.
Multiplying(14) bycos(asZ
) (wheres = 0, 1, 2, ...)
and integratingwith respectto z from -d to 0, we
obtain
3. Matching Boundary Conditions
In orderto determinethe unknowncoefficients
Pm
and qm, we enforcethe boundaryconditionson the • Jo(Kmb)Ims
q-0.5 Jl (•Cmb)Sms
Pm
field continuities. rn=0 Krn
2151
amd •rOtm d
+ E No(•cmb)Ims
+ 0.5
-- Nl (gmb)•Jmsqm + E No(•:mC)Jmk
+ 0.5
-- Nl (•:mC)amk qm
m=0 m=0
2i = 0 (22)
-- (15)
kb where
where•ms is the Kroneckerdelta,a m = 2 (m - 0),
1 (m - 1, 2,...), and
=•1f •• H(01)'(•c)
Jmk •S0(1)(•c) (23)
Gm(•)G•(•) d•
Ims
=- 2f0
••-/•-(-•3
R'(gb) Fs(OFm(•)d; (16) where the [Link]
evaluationof Jmk is summarizedin the appendix.
Performingthe residuecalculus,
we representIms in From (15) to (22) we obtaina matrixequationforP m
terms of a fast convergingseries
and qm:
dR ' ( •(b )
F
Ims
=2•(g(•(b• øt
m•Jms
l•=am xItxIt
•3 2 •4
(24)
• i•(_l)m+s(1 - e2i;d) where P and Q are column vectors consistingof
= b(• 2- m)• -as) elementsPm and qm, respectively,and the •1, •2,
•3, •4, and F elementsare
ik(- 1)m+s(1
-- e2it:a) amd
(•7)
qtl,ms + 0.5 --Jl(•Cmb)ams
-- Jo(gmb)Ims (25)
2b
ln(2)(k2-
a2m)(k2-
as
2) amd
where•n isgivenbyR(Kb)lg=•n-- 0. = No(•:mb)Ims+ 0.5--Nl(•(mb)ams
qt2,ms (26)
Similarly,E z continuityat p = c gives
-d<z<0 •rOtm d
Eziu(c,z)={•
zII(c'z) otherwise (18) q13,mk q-0.5--
-- Jo(•(mC)Jmk J1(•:m½)amk (27)
Applyingthe Fouriertransformto (18) gives •rOtm d
qt4,mk + 0.5--
= No(•:mC)Jmk Nl (•:mC)amk (28)
• Ro(gmC)
(19)
•III(•) =E H-•],-•;•Sm(•)
m=0 (- 1)S[eikd
-- e-ikd]
where
Vs
= b(k2
- a•2) (29)
-i •[ 1 - (- 1)me-i•d] Note that J l( ) and N l( ) are the first-order
Besseland Neumann functions,[Link] the
Gm(•)
= •2
--a m 2 (20)
low-frequencyregime (d << X), the lowestmode
corresponding to rn = 0 becomesdominant,and
continuity
at p = c gives
Pm - qm = 0 (m >- 1). Hence a dominant-mode
H4,u(c,z) = H4•iii(c, Z) (21) solutionis given as
Multiplying(21) by cos (akz) and integratingwith
respectto z from -d to 0, (30)
,00)
E Jo(•(m½)Jmk
m=0
q-0.5-- •(m
erøtmdJ1(•(m½)amk Pm
,00)
