p669 Chap03

MICROSTRIP PATCH ANTENNAS
World Scientific Publishing Co. Pte. Ltd.

http://www.worldscibooks.com/engineering/p669.html
July 1, 2010 11:7 9in x 6in b937-ch03
CHAPTER 3
General Formulation of the Cavity Model
3.1 Introductory Remarks
There are two approaches to solving for the performance characteristics of
microstrip patch antennas. One is to devise a physical model based on a
number of simplifying assumptions; the other is to solve Maxwells equa-
tions subject to the boundary conditions.
Let us use the center-fed dipole antenna as an analogy. The simplify-
ing approach is to regard the dipole of length L as two segments of an
open ended parallel wire transmission line which are bended upward, as
illustrated in Figure 3.1
It is assumed that the current distribution on the bended segments (the
dipole) is essentially the same as that in the parallel open ended transmis-
sion line, namely, sinusoidal. The radiation and impedance characteristics
are then calculated based on the sinusoidal current distribution.
In the more rigorous approach, the currents on the dipole are regarded
as unknowns to be solved. By requiring the elds produced by this current
to satisfy the boundary condition on the conductor, at innity, and across
the feed points, an integral equation is obtained for the unknown currents.
The integral equation is then solved by some numerical method. Once the
currents are determined, the radiation and impedance characteristics are
then calculated.
Similarly, for the basic geometry of the microstrip patch antenna shown
in Figure 3.2, a simplied physical model, known as the cavity model, can
be developed to deduce the characteristics of the antenna. It is based on
a number of simplifying assumptions which are valid for thin substrates.
These assumptions enable the elds between the patch and the ground
plane to be determined analytically for a number of patch shapes. From
these, the various characteristics of the MPA can be calculated.
As in the dipole case, a more general and rigorous approach will regard
the currents on the antenna structures to be the unknowns to be solved.
39
July 1, 2010 11:7 9in x 6in b937-ch03
40 Microstrip Patch Antennas
Fig. 3.1 Development of a dipole antenna of length L from an open-circuited
two-wire transmission line.
Fig. 3.2 Basic geometry of a microstrip patch antenna.
By requiring the elds produced by the currents to satisfy the boundary and
excitation conditions, integral equations are obtained for the unknown cur-
rents, which are solved by some numerical method. This approach is known
as the full-wave method. It has the advantages of being able to handle thick
substrates and complicated structures, such as multi-layer structures con-
taining dielectric cover, several patches and parasitic elements. However, it
requires extensive computation time and provides little physical insight.
July 1, 2010 11:7 9in x 6in b937-ch03
General Formulation of the Cavity Model 41
In this chapter, we present the general formulation of the cavity model
for the case of a coaxially fed microstrip patch antenna. This formulation
will be used in chapters 46 to obtain the characteristics of the rectangular,
circular, annular-ring, and equitriangular patch antennas.
3.2 Introducing the Cavity Model
Consider the region between the patch and the ground plane in the basic
microstrip patch antenna geometry shown in Figure 3.2. This region resem-
bles an electromagnetic resonator or cavity, excited by a coaxial probe.
In the usual resonant cavity used in microwave circuits, the vertical walls
are also conducting walls and the structure is in the form of a conducting
box, as illustrated in Figure 3.3 for the case of a rectangular cavity.
The elds that can exist inside the conducting box are obtained by rst
solving Maxwells equations in the region inside the box and demanding
its solutions to satisfy the boundary conditions of the tangential electric
elds, E
t
= 0 on the top, bottom and on its sides. Similar to the vibrating
string problem reviewed in section 2.1, this leads to an innite number of
discrete solutions or eigenfunctions. Each solution describes a eld pattern
or mode, with its characteristic or resonant frequency. An innite series
of appropriately weighted modes or eigenfunctions are used to satisfy the
conditions imposed by the feed.
In the patch antenna case, the side walls are not enclosed by conducting
walls and the elds inside the cavity can leak out to space, leading to
radiation. To nd the radiation elds, we rst need to nd the elds inside
the cavity. To do this, we need to know what boundary condition to impose
on the side (vertical) walls.
The thickness of the substrate is denoted by the symbol t or h in the
literature. It will be shown in section 3.3 that, if the substrate is thin so that
t where is wavelength, the boundary condition on the side walls is
Fig. 3.3 Rectangular cavity with conducting walls.
July 1, 2010 11:7 9in x 6in b937-ch03
Fig. 3.4 Boundary conditions for the cavity in the region under the patch.
H
t
= 0. The boundary value problem to solve to obtain the eigenfunctions
or modes is illustrated in Figure 3.4 for the rectangular patch.
3.3 The Thin Substrate Assumption
Under the condition that the substrate thickness t is much less than wave-
length , there will be no variation of the elds with z. Since E
t
= 0 on the
perfectly conducting metal surfaces,
E = zE
z
(3.1)
Since

H

E, it follows that

H is transverse to z.
Consider the surface current density
J
s
. Referring to Figure 3.5, we can
write
J
s
= z

H|
z=0
(3.2)
z

J
s
= z ( z

H) = z( z

H) +

H( z z) =

H (3.3)
J
s
= (

J
s
) + n( n

J
s
) (3.4)
Fig. 3.5 Thin substrate assumption.
July 1, 2010 11:7 9in x 6in b937-ch03
where , n are unit vectors tangential and normal to the perimeter of the
patch, C.
Since there should be no component normal to the boundary at the
perimeter, n

J
s
|
c
= 0 for any point on C, we have:
H|
c
= z

J
s
|
c
= z J
s
|
c
= nJ
sc
|
c
(3.5)
Hence H has only a normal component at the boundary, or H
t
= 0 on
the boundary. Since there is no variation of the elds with z, it follows
that H
t
= 0 on the vertical sides of the cavity. A wall with zero tangential
magnetic eld is known as a magnetic wall. Thus the cavity is bounded by
magnetic walls on the sides, and by electric walls on the top and bottom.
3.4 Solution for Cavity Fields
Let

J be the current density of the feed, assumed to be directed along z.
Since t , we assume

J to be independent of z. Under these conditions,
the continuity equation reads

J =
J
x
x
+
J
y
y
+
J
z
z
= 0 = j = 0 (3.6)

E = 0
In the cavity (region between patch and conducting plane),

E) =

(

E)
2
E = j
0
(j
E +

J) (3.7)
E +
2
E = j
0
J (3.8)
2
E
z
+k
2
d
E
z
= j
0
J
z
(3.9)
where k
d
=
E
z
is obtained by rst solving the homogeneous equation subject to the
boundary conditions. The solution to the inhomogeneous equation is then
obtained by superposition.
Homogeneous equation:
(
2
+k
2
d
)E
z
= 0 (3.10)
Boundary condition:
H
t
= 0 on the side walls of the cavity.
From Maxwells equations, this translates into (
E
z
n
) = 0 on the side walls
of the cavity.
July 1, 2010 11:7 9in x 6in b937-ch03
Let the solutions (eigenfunctions) be
mn
and the eigenvalues of k
d
be
k
mn
. For example, for a rectangular patch, the eigenfunctions and eigen-
values are given by
E
z
= E
0
cos
mx
a
cos
ny
b
m, n = integers (3.11)
k
2
mn
=
_
m
a
_
2
+
_
n
b
_
2
(3.12)
Inhomogeneous equation:
2
E
z
+k
2
d
E
z
= j
0
J
z
(3.13)
To solve Eq. (3.13), we use the eigenfunction expansion technique described
in sections 2.1 and 2.2.
Let
E
z
=
n
A
mn
mn
(3.14)
where the coecients A
mn
are determined by requiring Eq. (3.14) to be the
solution to the inhomogeneous equation.
Substituting (3.14) into (3.13), we have
2
E
z
=
n
A
mn
mn
=
n
A
mn
k
2
mn
mn
= j
0
J
z
k
2
d
n
A
mn
mn
(3.15)
Multiplying the left and right hand sides of (3.15) by
rs
(
denotes complex
conjugate) and integrating over the domain of the patch we obtain
n
(k
2
d
k
2
mn
)A
mn
___

mn
rs
dv = j
0
___
J
z
rs
dv (3.16)
However, since the functions are independent of z, the volume integral in
(3.16) becomes the surface integral:
n
(k
2
d
k
2
mn
)A
mn
__
s
p
mn
rs
ds = j
0
__
s
p
J
z
rs
ds (3.17)
where integration is over the area of the patch s
p
.
July 1, 2010 11:7 9in x 6in b937-ch03
Assuming
mn
and
rs
are orthogonal over s
p
, which can be veried a
posterori,
A
rs
=
j
0
k
2
d
k
2
rs
J
z
rs
_
rs
rs
_
(3.18)
where J
z
rs
=
__
s
p
J
z
rs
ds,
rs
rs
=
__
s
p
rs
rs
ds
The solution to the inhomogeneous equation is therefore
E
z
= j
0
n
1
k
2
d
k
2
mn
mn
_
mn
mn
_
mn
(3.19)
The characteristic or resonant frequencies are determined from
k
2
d
= k
2
mn
(2f
mn
)
2
0
= k
2
mn
(3.20)
Equation (3.20) yields
f
mn
=
k
mn
2
(3.21)
If losses are not taken into account, k
d
is real and Eq. (3.19) shows that E
z
tends to innity at the resonant frequency. In the realistic case with losses,
k
d
is complex and Eq. (3.19) shows that, at the resonant frequencies, E
z
assumes large but nite values.
3.5 Radiation Field
To calculate the radiation eld, consider a closed surface S as shown in
Figure 3.6.
The top face of S lies just outside the patch and the bottom face lies
just outside the ground plane. The vertical face of S coincides with the
magnetic wall of the cavity. The elds exterior to S can be calculated from
the equivalent sources on S and their images; the latter is necessary to
Fig. 3.6 Microstrip antenna with a closed surface S.
July 1, 2010 11:7 9in x 6in b937-ch03
account for the ground plane, which is assumed to be innite in extent
for the purpose of analysis. Since the tangential electric elds on the top
and bottom faces, as well as the tangential magnetic eld on the vertical
surface, are zero, the only contribution to the equivalent sources are the
tangential electric eld E, on the vertical surface of the cavity. Together
with its image (section 2.2.8), the total equivalent magnetic current is
M = 2
E
s
n (3.22)
where n is the unit outward normal.
If the substrate thickness t is much less than the wavelength , its eect
on the radiation eld is small and

M can be assumed to radiate in free
space. From Eq. (2.47), the electric potential

F at a point r is given by:
F =

0
t
4
_

M(r
)
e
jk
0
|rr
|
|r r
|
d
(3.23)
where integration is over the perimeter of the patch.
The elds in the far-zone are given by (section 2.2.4),
_
_
_
H
r
0
H
jF
jF
_
_
_
E
r
0
E

0
H

0
H
where
o
=
_
o
3.6 Feed Modeling
In the development of the cavity model, the coaxial-line feed is represented
by a cylindrical band of electric current owing from the ground plane
to the patch. To simply matters, this is idealized by assuming that it is
equivalent to a uniform current of some eective width, centered on the
feed axis. For example, for a circular patch fed at a distance d from the
center, it is illustrated in Figure 3.7 and described by
J = zJ()( d)/d (3.24)

where is the Dirac delta function and
J() =
_
J w < < +w
0 elsewhere
July 1, 2010 11:7 9in x 6in b937-ch03
Fig. 3.7 Modeling of a coaxial feed by a current ribbon for a circular patch.
The eective annular width 2w is a parameter chosen so that good agree-
ment between the theoretical and experimental impedances is obtained.
Usually, the arc length 2wd is several times the physical dimension of the
inner conductor.
3.7 Losses in Cavity
The losses in the cavity under the patch comprise dielectric loss (P
d
), con-
ductor loss (P
c
), radiation loss (P
r
) and surface-wave loss (P
sw
).
Surface wave loss represents energy leaving the cavity in the form of
waves guided along the dielectric. For thin substrates, P
sw
can be neglected.
According to Wood [4] , it is required that t/
o
< 0.07 for
r
= 2.3 and
t/
o
< 0.023 for
r
= 9.8, if the antenna is to launch no more than 25% of
the total radiated power as surface waves. For a given thickness (in terms
of wavelength), the higher the dielectric constant, the more the surface
wave loss.
The dielectric loss (P
d
) and the conductor loss (P
c
) are calculated from
the electric eld under the cavity, while the radiation loss (P
r
) is calculated
from the far-zone electromagnetic eld. A good discussion of these can be
July 1, 2010 11:7 9in x 6in b937-ch03
found in Collin [5]. They are given by
P
d
=

2
___
|E
z
|
2
dv (3.25)
P
c
= 2
R
s
2
__
|
H|
2
ds (3.26)
P
r
=
1
2
_
2
0
_

0
|
E|
2
2
0
r
2
sin dd (3.27)
The quantity in (3.25) is the loss tangent of the dielectric and R
s
in
equation (3.26) is the surface resistivity of the conductors. (The loss tangent
is also denoted by tan ).
The radiation eciency is the ratio of radiated power to input power:
e% =
P
r
P
c
+P
d
+P
r
100% (3.28)
In calculating the losses, it is usual to make use of the resonance approx-
imation. This arises from the observation that, if the frequency is close to
the resonant frequency of a particular mode, the factor
1
k
2
d
k
2
mn
in equa-
tion (3.19) is very large and the contribution to E
z
, and hence to the
radiation eld E
z
, is due mainly to the resonant-mode term. At reso-
nance, the stored electric energy W
e
is equal to the stored magnetic energy
W
m
[5]:
W
e
=

4
___
|E
z
|
2
dv =
P
d
2
(3.29)
W
m
= W
e
(3.30)
And the total stored energy at resonance is
W
t
=

2
___
|E
z
|
2
dv =
P
d
(3.31)
In an ideal cavity with lossless walls, the loss tangent and the cavity quality
factor Q are related by
=
1
Q
(3.32)
To take into account the radiation loss and the copper loss, the concept of
eective loss tangent is dened as [2]
eff
=
1
Q
T
(3.33)
July 1, 2010 11:7 9in x 6in b937-ch03
where
Q
T
=
W
T
P
T
(3.34)
P
T
= P
d
+P
c
+P
r
(3.35)
In the formula for calculating the cavity eld E
z
, is to be replaced by
eff
. Since the calculation of the power losses depend on E
z
, the denition
of the eective loss tangent leads to a nonlinear equation for
eff
.
eff
can
be solved by an iterative procedure:
i+1
= [P
r
(
i
) +P
c
(
i
) +P
d
(
i
)]/2
0
W
e
(
i
) (3.36)
where the starting value is
o
= . In practice, the value of
eff
is found to
converge after one or two iterations.
3.8 Input Impedance
The input impedance at the feed of the antenna is given by
Z = R +jX =
V
I
=
E
av
t
I
(3.37)
where E
av
is the average value of the electric eld at the feed point and
I is the total current. For example, if the feed is modeled by Eq. (3.24),
we have
E
av
=
1
2w
_
+w
w
E
z
(d, )d (3.38)
and
I = J(2wd) (3.39)
Unlike the calculations of
eff
, it was found that non-resonant modes
must be included in the calculation of input impedance if good agreement
between theory and experiment was to be obtained. The appropriate equa-
tion for E
z
is therefore Eq. (3.19), which contains the factor
1
k
2
d
k
2
mn
. To
keep this term nite at resonance, the permittivity of the dielectric must
be considered complex. If only the dielectric loss is considered, we have
=
0
r
(1 j) (3.40)
k
2
d
=
2
0
= k
2
0
r
(1 j) (3.41)
However, Richard et al. [2] found that better agreement with experiment
was obtained if, instead of the loss tangent of the dielectric, the eective
July 1, 2010 11:7 9in x 6in b937-ch03
Fig. 3.8 Typical impedance characteristics around the resonant frequency of a
mode.
loss tangent is used. Thus, in calculating the input impedance, Eq. (3.19)
is modied to read
E
z
= j
0
m=0
n=0
1
k
2
eff
k
2
mn
J
z

mn
_
mn

mn
_
mn
(3.42)
where
k
2
eff
= k
2
0
r
(1 j
eff
) (3.43)
A typical impedance versus frequency curve is illustrated in Figure 3.8.
There is usually some reactance at the resonant frequency of a mode due
to the contributions from the non-resonant modes.
3.9 VSWR Bandwidth
The bandwidth of an antenna is the range of frequencies within which
the performance of the antenna, with respect to some characteristic, con-
forms to a specied standard. In the case of the microstrip patch antenna
which is basically a strongly resonant device, it is usually the variation of
July 1, 2010 11:7 9in x 6in b937-ch03
impedance, rather than pattern, which limits the standard of performance.
If the antenna impedance is matched to the transmission line at resonance,
the mismatch o resonance is related to the voltage standing wave ratio
(denoted by VSWR or SWR). The value of VSWR which can be toler-
ated then denes the bandwidth of the antenna. If this value is to be
less than S, the usable bandwidth of the antenna is related to the total
Q-factor by
B =
1
Q
S 1
S
(3.44)
To derive Eq. (3.44), consider the transmission line of characteristic
resistance R
o
terminated in the microstrip patch antenna, which has input
impedance Z
in
, as illustrated in Figure 3.9.
Consider the antenna as a resonant circuit. From elementary circuit
theory, the input impedance of the resonant circuit can be written in term
of the Q of the circuit and the frequency deviation from resonance, :
Z
in
= R
in
(1 +j2Q) (3.45)
where
=
f f
r
f
r
(3.46)
The voltage reection coecient and the standing wave ratio S are
given by
=
Z
in
R
0
Z
in
+R
0
(3.47)
S =
1 +||
1 ||
(3.48)
Fig. 3.9 Equivalent circuit of microstrip antenna connected to a transmission
line with characteristic resistance R
o
.
July 1, 2010 11:7 9in x 6in b937-ch03
Thus
S 1
S + 1
= || =
Z
in
R
0
Z
in
+R
0
(3.49)
Using Eq. (3.45), we have
S 1
S + 1
=
|1 +j2QR
0
/R
in
|
|1 +j2Q +R
0
/R
in
|
(3.50)
On letting T =
R
0
R
in
, we obtain
(1 T)
2
+ 4Q
2
2
(1 +T)
2
+ 4Q
2
2
=
_
S 1
S + 1
_
2
(3.51)
The solution for
2
is:
2
=
1
4Q
2
_
(TS 1)(S T)
S
_
(3.52)
Referring to Figure 3.10, let f
2
and f
1
respectively be the frequencies above
and below resonance at which the voltage standing wave ratio equals S.
The fractional bandwidth corresponding to this value of S is then
B =
f
2
f
1
f
r
=
2(f
2
f
r
)
f
r
= 2 (3.53)
Thus
B
2
4
=
1
4Q
2
(TS 1) (S T)
S
(3.54)
B =
1
Q
_
(TS 1) (S T)
S
(3.55)
Fig. 3.10 Illustration of VSWR against frequency plot.
July 1, 2010 11:7 9in x 6in b937-ch03
If the antenna is matched at the frequency f
r
, R
in
= R
o
and T = 1 and we
obtain Eq. (3.44):
B =
1
Q
S 1
S
In the literature, S is usually chosen to be 2 for dening the VSWR BW.
In industry, the more stringent standard S = 1.5 is used.
3.10 Qualitative Description of the Results Predicted
by the Cavity Model
In chapters 46, the equations presented in this chapter will be used to
obtain the specic results for a number of microstrip patch antennas. It is
instructive to describe here the qualitative features which are common to
MPAs. These features follow naturally from viewing the MPA as a leaky
cavity.
1. There are an innite number of resonant modes, each characterized by
a resonant frequency.
2. Because of fringing elds at the edge of the patch, the patch behaves as
if it has a slightly larger dimension. Semi-empirical factors are usually
introduced to obtain these eective dimensions. These factors vary from
patch to patch.
3. Each resonant mode has its own characteristic radiation pattern. The
lowest mode usually radiates strongest in the broadside direction. The
pattern of this mode is broad, with half-power beamwidths of the order
of 100
.
4. For coaxial-fed antennas, the input impedance is dependent on the feed
position. The variation of input resistance at resonance with feed position
essentially follows that of the cavity eld. For the lowest mode, it is
usually large when the feed is near the edge of the patch and decreases
as the feed moves inside the patch. Its magnitude can vary from tens to
hundreds of ohms. By choosing the feed position properly, an eective
match between the antenna and the transmission line can be obtained.
5. Since the cavity under the patch is basically a resonator, the total Q
and the impedance bandwidth are dependent on the thickness of the
substrate t and its permittivity
r
. For low values of
r
, the bandwidth
generally increases with increasing t and decreases with increasing
r
.
However, detailed analysis shows that the bandwidth and Q are compli-
cated functions of frequency, substrate thickness and permittivity.
July 1, 2010 11:7 9in x 6in b937-ch03
6. For thin substrates, the impedance bandwidth varies from less than one
to several percent.
3.11 Limitations of the Cavity Model Analysis
Let us recap the assumptions and limitations of the cavity model. The basic
assumption which renders the calculations of the cavity model relatively
simple is that the substrate thickness is assumed to be much smaller than
wavelength so that the electric eld has only a vertical (z) component which
does not vary with z. From this it follows that:
(1) The elds in the cavity are TM (transverse magnetic).
(2) The cavity is bounded by magnetic walls (H
t
= 0) on the sides.
(3) Surface wave excitation is negligible.
(4) The current in the coaxial probe is independent of z.
The coaxial probe is modeled by a current ribbon of a certain width,
which is a free parameter chosen to t the experimental data.
There are a number of limitations to the cavity model even if the thin
substrate condition is satised. The magnetic wall boundary condition leads
to resonant frequencies which do not agree well with experimental obser-
vations, and an ad hoc correction factor has to be introduced to account
for the eect of fringing elds. The width of the current ribbon used to
model the coaxial probe is another ad hoc parameter. The model cannot
handle designs involving parasitic elements, either on the same layer or on
another layer. It cannot analyze microstrip antennas with dielectric covers.
When the thickness of the substrate exceeds about 2% of the free space
wavelength, the cavity model results begin to become inaccurate, due to
the breakdown of (1)(4).
Despite the limitations described above, the cavity model has the advan-
tage of being simple and providing physical insight. The basic characteris-
tics and design information for rectangular, circular, annular, and triangular
patches can be obtained with relative ease, as we show in the next three
chapters.
3.12 Problems
1. (a) Assuming E
z
to be independent of z, solve
2
E
z
+ k
2
d
E
z
= 0 in a
rectangular cavity with electric walls (E
t
=0) on the top and bottom
and magnetic walls (H
t
=0) on the sides.
July 1, 2010 11:7 9in x 6in b937-ch03
(b) Obtain the formula for the characteristic (resonant) frequencies of
TM
mn
modes.
(c) Obtain the resonant frequency of the lowest mode for the case a > b.
x
z
a
b
o
y
2. If, in the rectangular cavity of problem 1, the side wall y = 0 is shorted
so that E
t
= 0. At y = 0 as well as on the top and bottom,
(a) nd the solution for E
z
;
(b) nd the formula for the resonant frequency of the lowest mode for
the case a > b.
References
[1] Y. T. Lo, D. Solomon and W. F. Richards, Theory and experiment
on microstrip antennas, IEEE Trans. Antennas Propagat., Vol. AP-27,
pp. 137145, 1979.
[2] W. F. Richards, Y. T. Lo and D. D. Harrison, An improved theory for
microstrip antennas and applications, IEEE Trans. Antennas Propagat.,
Vol. AP-29, pp. 3846, 1981.
[3] R. S. Elliott, Antenna Theory and Design, Prentice Hall, 1981, Sections 3.7
and 7.18.
[4] C. Wood, Analysis of microstrip circular patch antennas, IEE Proc.
Vol. 128H, pp. 6976, 1981.
[5] R. E. Collin, Foundations of Microwave Engineering, Second Edition,
McGraw-Hill, Inc. New York, 1992.
[6] K. F. Lee and J. S. Dahele, Characteristics of microstrip patch antennas and
some methods of improving frequency agility and bandwidth, in Handbook of
Microstrip Antennas, J. R. James and P. S. Hall (Editors), Peter Peregrinus,
Ltd., London, 1989.

p669 Chap03

Enviado por

Dados do documento

Descrição original:

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

p669 Chap03

Enviado por

Direitos autorais:

Formatos disponíveis

MICROSTRIP PATCH ANTENNAS

World Scientific Publishing Co. Pte. Ltd.

J = zJ()( d)/d (3.24)

Você também pode gostar