Introduction To Microstrip Antennas PDF

Introduction to Microstrip Antennas
David R. Jackson
Dept. of ECE
University of Houston
1
Contact Information
David R. Jackson
Dept. of ECE
N308 Engineering Building 1
University of Houston
Houston, TX 77204-4005
Phone: 713-743-4426
Fax: 713-743-4444
Email: djackson@uh.edu
2
Purpose of Short Course
 Provide an introduction to microstrip antennas.

 Provide a physical and mathematical basis for understanding how
microstrip antennas work.
 Provide a physical understanding of the basic physical properties of
microstrip antennas.
 Provide an overview of some of the recent advances and trends in
the area (but not an exhaustive survey – directed towards
understanding the fundamental principles).
3
Additional Resources
 Some basic references are provided at the end of these viewgraphs.

 You are welcome to visit a website that goes along with a course at the
University of Houston on microstrip antennas (PowerPoint viewgraphs
from the course may be found there, along with the viewgraphs from this
short course).
ECE 6345: Microstrip Antennas

http://www.egr.uh.edu/courses/ece/ece6345/web/welcome.html
Note: You are welcome to use anything that you find on this website,
as long as you please acknowledge the source.
4
Outline
 Overview of microstrip antennas
 Feeding methods
 Basic principles of operation
 General characteristics
 CAD Formulas
 Radiation pattern
 Input Impedance
 Circular polarization
 Circular patch
 Improving bandwidth
 Miniaturization
 Reducing surface waves and lateral radiation
5
Notation
c = speed of light in free space =c 2.99792458 ×108 [ m/s ]
λ0 = wavelength of free space λ0 = c / f
k0 = wavenumber of free space =k0 ω=

µ0ε 0 2π / λ0
k1 = k0 ε r
k1 = wavenumber of substrate
µ0
η0 = intrinsic impedance of free space η0 = ≈ 376.7303 [ Ω ]
ε0
η1 = intrinsic impedance of substrate η1 = η0 / ε r
ε r = relative permtitivity (dielectric constant) of substrate

1
c=
ε 0 µ0
ε reff = effective relative permtitivity
(accouting for fringing of flux lines at edges) µ=
0 4π × 10 7
[ H/m]
1
ε eff
= complex effective relative permtitivity ε 0 = 2 ≈ 8.854188 ×1012 [ F/m ]
rc µ0 c
(used in the cavity model to account for all losses)
6
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
7
Overview of Microstrip Antennas
Also called “patch antennas”
 One of the most useful antennas at microwave frequencies (f > 1 GHz).

 It usually consists of a metal “patch” on top of a grounded dielectric substrate.
 The patch may be in a variety of shapes, but rectangular and circular are the
most common.
Microstrip line feed Coax feed

8
Common Shapes
Rectangular Square Circular Annular ring
Elliptical Triangular
9
History
 Invented by Bob Munson in 1972 (but earlier work by Dechamps

goes back to1953).
 Became popular starting in the 1970s.
G. Deschamps and W. Sichak, “Microstrip Microwave Antennas,” Proc. of Third

Symp. on USAF Antenna Research and Development Program, October 18–22,
1953.
R. E. Munson, “Microstrip Phased Array Antennas,” Proc. of Twenty-Second Symp.

on USAF Antenna Research and Development Program, October 1972.
R. E. Munson, “Conformal Microstrip Antennas and Microstrip Phased Arrays,” IEEE

Trans. Antennas Propagat., vol. AP-22, no. 1 (January 1974): 74–78.
10
Advantages of Microstrip Antennas
 Low profile (can even be “conformal,” i.e. flexible to conform to a

surface).
 Easy to fabricate (use etching and photolithography).
 Easy to feed (coaxial cable, microstrip line, etc.).
 Easy to use in an array or incorporate with other microstrip circuit
elements.
 Patterns are somewhat hemispherical, with a moderate directivity
(about 6-8 dB is typical).
11
Disadvantages of Microstrip Antennas
 Low bandwidth (but can be improved by a variety of techniques). Bandwidths of

a few percent are typical. Bandwidth is roughly proportional to the substrate
thickness and inversely proportional to the substrate permittivity.
 Efficiency may be lower than with other antennas. Efficiency is limited by

conductor and dielectric losses*, and by surface-wave loss**.
 Only used at microwave frequencies and above (the substrate becomes too
large at lower frequencies).
 Cannot handle extremely large amounts of power (dielectric breakdown).
* Conductor and dielectric losses become more severe for thinner

substrates.
** Surface-wave losses become more severe for thicker substrates
(unless air or foam is used).
12
Applications
Applications include:
 Satellite communications
 Microwave communications
 Cell phone antennas
 GPS antennas
13
Microstrip
antenna
Filter
DC supply Micro-D
connector K-connector
LNA
PD
Fiber input with
collimating lens Diplexer
Microstrip Antenna Integrated into a System: HIC Antenna Base-Station for 28-43 GHz
(Photo courtesy of Dr. Rodney B. Waterhouse)
14
Arrays
Linear array (1-D corporate feed)
2×2 array
2-D 8X8 corporate-fed array 4 × 8 corporate-fed / series-fed array

15
Wraparound Array (conformal)
The substrate is so thin that it can be bent to “conform” to the surface.
16
Rectangular patch
W Js
x
L
h εr
Note: L is the resonant dimension (direction of current flow).

The width W is usually chosen to be larger than L (to get higher bandwidth).
However, usually W < 2L (to avoid problems with the (0,2) mode).
W = 1.5L is typical.
17
Circular Patch
a x
h εr
The location of the feed determines the direction of current flow and hence
the polarization of the radiated field.
18
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
19
Feeding Methods
Some of the more common methods for

feeding microstrip antennas are shown.
The feeding methods are illustrated for a rectangular patch,

but the principles apply for circular and other shapes as well.
20
Feeding Methods
z Coaxial Feed
εr h
x
Surface current
A feed along the centerline
is the most common
(minimizes higher-order
W modes and cross-pol).
Feed at (x0, y0) x

L
21
Feeding Methods
Coaxial Feed
 π x0  z
R = Redge cos 2

 L 
(The resistance varies as the
square of the modal field shape.) εr h
x
Advantages: y
 Simple
 Directly compatible with coaxial cables
 Easy to obtain input match by adjusting feed position ( x0 , y0 )
W
Disadvantages:
 Significant probe (feed) radiation for thicker substrates x
L
 Significant probe inductance for thicker substrates (limits
bandwidth)
 Not easily compatible with arrays
22
Feeding Methods
Inset Feed
Advantages:
 Simple Microstrip line
 Allows for planar feeding

 Easy to use with arrays
 Easy to obtain input match
Disadvantages:
 Significant line radiation for thicker substrates
 For deep notches, patch current and radiation pattern may show distortion
23
Feeding Methods
Inset Feed
Recent work has shown

that the resonant input
resistance varies as x0
Wf
S
2  π  2 x0  W
Rin A cos   − B
2 L 
L
The coefficients A and B depend on the notch width S but (to a good
approximation) not on the line width Wf .
Y. Hu, D. R. Jackson, J. T. Williams, and S. A. Long, “Characterization of the Input Impedance of

the Inset-Fed Rectangular Microstrip Antenna,” IEEE Trans. Antennas and Propagation, Vol. 56,
No. 10, pp. 3314-3318, Oct. 2008.
24
Feeding Methods
Proximity-coupled Feed
(Electromagnetically-coupled Feed)
Advantages:
 Less line radiation compared to microstrip feed
 Can allow for higher bandwidth (no probe inductance, so
substrate can be thicker)
Patch
Microstrip
Top view line
Microstrip line
Disadvantages:
 Requires multilayer fabrication
 Alignment is important for input match
25
Feeding Methods
Gap-coupled Feed
Advantages:
 Can allow for a match even with high edge impedances, where a notch
might be too large (e.g., when using high permittivity)
Gap Patch
Patch
Microstrip line Microstrip

Top view line
Disadvantages:
 Requires accurate gap fabrication
 Requires full-wave design
26
Feeding Methods
Aperture-coupled Patch (ACP)
Advantages:
 Allows for planar feeding Slot
 Feed-line radiation is isolated from patch radiation

 Higher bandwidth is possible since probe inductance is
eliminated (allowing for a thick substrate), and also a Microstrip
double-resonance can be created Top view line
 Allows for use of different substrates to optimize

antenna and feed-circuit performance
Patch
Disadvantages: Slot
 Requires multilayer fabrication
 Alignment is important for input match Microstrip line
27
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
28
Basic Principles of Operation
 The basic principles are illustrated here for a rectangular patch, but the
principles apply similarly for other patch shapes.
 We use the cavity model to explain the operation of the patch antenna.
h n̂
PMC
Y. T. Lo, D. Solomon, and W. F. Richards, “Theory and Experiment on Microstrip Antennas,”

IEEE Trans. Antennas Propagat., vol. AP-27, no. 3 (March 1979): 137–145.
29
Main Ideas:
 The patch acts approximately as a resonant cavity (with short-circuit (PEC)

walls on top and bottom, open-circuit (PMC) walls on the edges).
 In a cavity, only certain modes are allowed to exist, at different resonance

frequencies.
 If the antenna is excited at a resonance frequency, a strong field is set up inside

the cavity, and a strong current on the (bottom) surface of the patch. This
produces significant radiation (a good antenna).
Note: As the substrate thickness gets smaller the patch current radiates less, due to
image cancellation. However, the Q of the resonant mode also increases, making the
patch currents stronger at resonance. These two effects cancel, allowing the patch to
radiate well even for small substrate thicknesses.
(See next slide.)

30
A microstrip antenna can radiate well, even with a thin substrate.
 As the substrate gets thinner the patch current radiates less, due to image
cancellation (current and image are separated by 2h).
 However, the Q of the resonant cavity mode also increases, making the
patch currents stronger at resonance.
 These two effects cancel, allowing the patch to radiate well even for thin
substrates (though the bandwidth decreases).
z 1
Js ∝
h
εr Js h
x
31
Thin Substrate Approximation
On patch and ground plane: Et = 0 E = zˆ E z
Inside the patch cavity, because of the thin substrate,

the electric field vector is approximately independent of z.
Hence E ( x, y, z ) ≈ zˆ Ez ( x, y )
E z ( x, y )
h
32
Magnetic field inside patch cavity:
1
H=− ∇× E
jωµ
∇ × ( zEˆ z ( x, y ) )
1
=−
jωµ
=−
1
jωµ
( − zˆ × ∇Ez ( x, y ) )
33
( x, y )
H=
1
jωµ
( zˆ × ∇E ( x, y ) )
z
Note: The magnetic field is purely horizontal.

(The mode is TMz.)
E z ( x, y )
h
H ( x, y )
34
Magnetic-wall Approximation
On the edges of the patch: y

J s ⋅ nˆ =0 L
(Js is the sum of the top and bottom surface currents.)
Js n̂
On the bottom surface of the patch W
conductor, at the edge of the patch,
we have: tˆ
x
J bot
s ⋅ nˆ ≈ 0 ( assuming J bot
s >> J stop )
Also,
H tbot ≈ 0
J sbot = ( − zˆ × H ) h
H tbot ≈ 0
35
Since the magnetic field is approximately

independent of z, we have an approximate PMC y
condition on the entire vertical edge.
L
H t = 0 (PMC)
Js n̂
or W
nˆ × H ( x, y ) =
0 tˆ
x
PMC
H t
edge
≈0
n̂
h h
Actual patch PMC Model
36
y
nˆ × H ( x, y ) =
0
L
( x, y )
H=
1
jωµ
( zˆ × ∇E ( x, y ) )
z
Js n̂
W
Hence,
nˆ × ( zˆ × ∇Ez ( x, y ) ) = 0
tˆ
x
nˆ × ( zˆ × ∇Ez ( x, y=
) ) zˆ ( nˆ ⋅∇Ez ( x, y ) ) − ∇Ez ( x, y )( nˆ ⋅ zˆ )
zˆ ( nˆ ⋅∇Ez ( x, y ) ) = 0
∂Ez n̂
=0 (Neumann B.C.)
h
∂n PMC
37
Resonance Frequencies
y
∇ Ez + k Ez =
2
02
k= k=
1 k0 ε r
E z ( x, y )
From separation of variables:

 mπ x   nπ y 
Ez = cos   cos   W
PMC
 L   W 
(TMmn mode)
x
L
  mπ  2  nπ  2 
We then have  −   −  + k1  Ez =
2
0
  L   W  
  mπ  2  nπ  2 
Hence  −   −  + k1  =
2
0
  L   W  
38
We thus have
y
 mπ   nπ  E z ( x, y )
2 2
=k  1
2
 + 
 L   W 
PMC
W
Recall that
= 0 εr
k1 k= ω µ 0ε 0 ε r
x
ω = 2π f L
Hence
 mπ   nπ 
2 2
c
=f   +  c = 1/ µ0ε 0
2π ε r  L  W 
39
y
E z ( x, y )
Hence f = f mn
(resonance frequency of (m,n) mode)
PMC
W
x
L
 mπ   nπ 
2 2
c
f mn   + 
2π ε r  L   W 
40
Dominant (1,0) mode
y
This structure operates as a “fat planar dipole.”
Current
This mode is usually used because the
radiation pattern has a broadside beam.
c 1 W
f10 =  
2 εr  L 
x
πx 
Ez = cos   L
 L 
 1  π   π x 
H ( x, y ) = − yˆ     sin  
The resonant length L is about 0.5 guided
 jωµ   L   L  wavelengths in the x direction.
 −1   π   π x 
Js = x 
ˆ    sin  
 jωµ 0   L   L 
41
Resonance Frequency of Dominant Mode
The resonance frequency is mainly controlled by the

patch length L and the substrate permittivity.
Approximately, (assuming PMC walls)
 mπ   nπ 
2 2
This is equivalent to saying that
= 2
k1   +  the length L is one-half of a
 L   W  wavelength in the dielectric.
λ0 / 2
(1,0) mode: k1 L = π =L λ=
d /2
εr
Note: A higher substrate permittivity allows for a smaller antenna

(miniaturization) – but with a lower bandwidth.
42
The resonance frequency calculation can be improved by adding a
“fringing length extension” ∆L to each edge of the patch to get an
“effective length” Le .
y
Le = L + 2∆L
c  1  ∆L ∆L
f10 =  
2 ε r  Le  L
x
Le
Note: Some authors use effective permittivity in this equation.

(This would change the value of Le.)
43
Hammerstad formula:
 eff W 
( rε + 0.3 )  h + 0.264 
0.412 
∆L / h = 
 ε eff − 0.258  W + 0.8  
 ( r )  h 

−1/ 2
ε r +1  ε r −1    h 
ε reff = +  +  
2 
1 12
2    W 
Note: Even though the Hammerstad formula

involves an effective permittivity, we still use  c 1 
f10 =  
the actual substrate permittivity in the
2 ε r  L + 2 ∆L 
resonance frequency formula.
44
Note: ∆L ≈ 0.5 h
This is a good “rule of thumb” to give a quick estimate.
45
Results: Resonance Frequency
1
Hammerstad
NORMALIZED FREQUENCY
Measured
0.95
0.9
0.85
0.8
0.75
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
hh // λ
λ0
0
εr = 2.2 The resonance frequency has been normalized by the

zero-order value (without fringing):
W/ L = 1.5
fN = f / f0 46
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
47
General Characteristics
Bandwidth
 The bandwidth is directly proportional to substrate thickness h.
 However, if h is greater than about 0.05 λ0 , the probe inductance (for a

coaxial feed) becomes large enough so that matching is difficult – the
bandwidth will decrease.
 The bandwidth is inversely proportional to εr (a foam substrate gives a high

bandwidth).
 The bandwidth of a rectangular patch is proportional to the patch width W

(but we need to keep W < 2L ; see the next slide).
48
Width Restriction for a Rectangular Patch
W < 2L c 1
f 01 =  
2 εr  W 
c  mπ   nπ 
2 2
c 1
f10 =  
f mn   +  2 εr  L 
2π ε r  L  W 
c  2
f 02 =  
c 1 1  2 εr  W 
f 02 −=
f 01  − 
ε r  W 2L 
fc
W
f01 f10 f02
W = 1.5 L is typical.
L
49
Some Bandwidth Observations
 For a typical substrate thickness (h /λ0 = 0.02), and a typical substrate

permittivity (εr = 2.2) the bandwidth is about 3%.
 By using a thick foam substrate, bandwidth of about 10% can be achieved.
 By using special feeding techniques (aperture coupling) and stacked

patches, bandwidths of 100% have been achieved.
50
Results: Bandwidth
30
25
εεr r==10.8
10.8
BANDWIDTH (%)
20
15
10
5 2.2
2.2
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
hh/ /λλ0 0
The discrete data points are measured values.

The solid curves are from a CAD formula (given later).
εr = 2.2 or 10.8 W/ L = 1.5
51
Resonant Input Resistance
 The resonant input resistance is fairly independent of the substrate
thickness h unless h gets small (the variation is then mainly due to
dielectric and conductor loss).
 The resonant input resistance is proportional to εr.
 The resonant input resistance is directly controlled by the location of the

feed point (maximum at edges x = 0 or x = L, zero at center of patch).
(x0, y0)
W
L L
x
52
Resonant Input Resistance (cont.)
The patch is usually fed along the centerline (y0 = W / 2)

to maintain symmetry and thus minimize excitation of undesirable modes
(which cause cross-pol).
y
Desired mode: (1,0)
Feed: (x0, y0)
x
L
53
For a given mode, it can be shown that the resonant input resistance is
proportional to the square of the cavity-mode field at the feed point.
This is seen from the cavity-model eigenfunction analysis

(please see the Appendix).
( x0 , y0 )
y
Rin ∝ E 2
z
(x0, y0)
For (1,0) mode: W
 π x0 
Rin ∝ cos  2
 x
 L  L
54
y
Hence, for (1,0) mode:
 π x0  (x0, y0)
Rin = Redge cos  2

 L  W
x
L
The value of Redge depends strongly on the substrate permittivity

(it is proportional to the permittivity).
For a typical patch, it is often in the range of 100-200 Ohms.
55
Results: Resonant Input Resistance
200
The discrete data points are from a CAD
formula (given later.) y0 = W/2
INPUT RESISTANCE ( Ω )
150
εεr r==10.8
10.8 Region where loss is important x0 = L/4
100 y
2.2
2.2
50 (x0, y0)
W
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
hh // λ
λ0
0
L x
εr = 2.2 or 10.8 W/L = 1.5

56
Radiation Efficiency
 Radiation efficiency is the ratio of power radiated into

space, to the total input power.
Pr
er =
Ptot
 The radiation efficiency is less than 100% due to
 Conductor loss
 Dielectric loss
 Surface-wave excitation
57
Radiation Efficiency (cont.)
TM0
surface wave
Js
x
cos (φ) pattern
58
Hence,
Pr Pr
=
er =
Ptot Pr + ( Pc + Pd + Psw )
Pr = radiated power Pc = power dissipated by conductors
Ptot = total input power Pd = power dissipated by dielectric

Psw = power launched into surface wave
59
Some observations:
 Conductor and dielectric loss is more important for thinner substrates (the
Q of the cavity is higher, and thus more seriously affected by loss).
 Conductor loss increases with frequency (proportional to f 1/2) due to the

skin effect. It can be very serious at millimeter-wave frequencies.
 Conductor loss is usually more important than dielectric loss for typical
substrate thicknesses and loss tangents.
1 2
=Rs = δ
σδ ωµσ Rs is the surface resistance of the metal.
The skin depth of the metal is δ.
ωµ0
=Rs ∝ f
2σ
60
 Surface-wave power is more important for thicker substrates or for

higher-substrate permittivities. (The surface-wave power can be
minimized by using a thin substrate or a foam substrate.)
 For a foam substrate, a high radiation efficiency is obtained by making the

substrate thicker (minimizing the conductor and dielectric losses). There is no
surface-wave power to worry about.
 For a typical substrate such as εr = 2.2, the radiation efficiency is maximum for
h / λ0 ≈ 0.02.
61
Results: Efficiency (Conductor and dielectric losses are neglected.)
100
2.2
2.2
80
EFFICIENCY (%)
60
ε r10.8
= 10.8
40
exact
CAD
20
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
h // λ
h λ0
0
εr = 2.2 or 10.8 W/L = 1.5 Note: CAD plot uses Pozar formula (given later).
62
Results: Efficiency (All losses are accounted for.)
100
2.2
2.2
80
EFFICIENCY (%)
exact
60
CAD
ε rr == 10.8
10.8
40
tan δ = 0.001
20 σ 3.0 × 107 [S/m]
=
0
0 0.02 0.04 0.06 0.08 0.1
hh // λλ00
εr = 2.2 or 10.8 W/L = 1.5 Note: CAD plot uses Pozar formula (given later).
63
Radiation Pattern
y
E-plane: co-pol is Eθ
L H-plane: co-pol is Eφ
Js
W
x E plane
Probe
Note: For radiation patterns, it is usually more

convenient to place the origin at the middle of the patch
(this keeps the formulas as simple as possible).
H plane
64
Radiation Patterns (cont.)
Comments on radiation patterns:
 The E-plane pattern is typically broader than the H-plane pattern.
 The truncation of the ground plane will cause edge diffraction, which
tends to degrade the pattern by introducing:
 Rippling in the forward direction

 Back-radiation
 Pattern distortion is more severe in the E-plane, due to the angle

dependence of the vertical polarization Eθ on the ground plane.
(It varies as cos (φ)).
65
Radiation Patterns
Edge diffraction is the most serious in the E plane.
y Eθ varies as cos φ
Space wave
L
Js
W
E plane
H plane
66
Radiation Patterns
E-plane pattern
Red: infinite substrate and ground plane

Blue: 1 meter ground plane
0
Note: The E-plane pattern
30 -30 “tucks in” and tends to
-10 zero at the horizon due to
the presence of the infinite
60 -20 -60 substrate.
-30
-40 -30 -20 -10

90 -90
120 240
150 210
180 67
Radiation Patterns
H-plane pattern
Red: infinite substrate and ground plane

Blue: 1 meter ground plane
0
45 -10 -45
-20
-30
-40 -30 -20 -10

90 -90
135 225
180
68
Directivity
The directivity is fairly insensitive to the substrate thickness.
The directivity is higher for lower permittivity, because the patch is

larger.
69
Results: Directivity (relative to isotropic)
10
εεr r==2.2
2.2
8
DIRECTIVITY (dB)
10.8
10.8
6
4
exact
CAD
2
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
hh//λλ0
0
εr = 2.2 or 10.8 W/ L = 1.5

70
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
71
CAD Formulas
CAD formulas for the important properties of the
rectangular microstrip antenna will be shown.
 Radiation efficiency
 Bandwidth (Q)
 Resonant input resistance
 Directivity
 D. R. Jackson, “Microstrip Antennas,” Chapter 7 of Antenna Engineering Handbook, J. L.

Volakis, Editor, McGraw Hill, 2007.
 D. R. Jackson, S. A. Long, J. T. Williams, and V. B. Davis, “Computer-Aided Design of
Rectangular Microstrip Antennas,” Ch. 5 of Advances in Microstrip and Printed Antennas,
K. F. Lee and W. Chen, Eds., John Wiley, 1997.
 D. R. Jackson and N. G. Alexopoulos, “Simple Approximate Formulas for Input Resistance,
Bandwidth, and Efficiency of a Resonant Rectangular Patch,” IEEE Trans. Antennas and
Propagation, Vol. 39, pp. 407-410, March 1991.
72
CAD Formulas
Radiation Efficiency
erhed
er =
  Rsave  1    3   ε r   L   1 
1+ e
hed
 d +          
r
 π η
 0  h / λ0    16   1 W
p c   h / λ0  
where Comment: The efficiency becomes small as the substrate gets thin.
= =
 d tan δ loss tangent of substrate
1 ωµ
=
Rs =
surface resistance of metal =
σδ 2σ
=
Rsave (Rs
patch
+ Rsground ) / 2
Note: “hed” refers to a unit-amplitude horizontal electric dipole.
73
CAD Formulas
hed
P 1
=erhed =
sp
Psphed + Pswhed Pswhed

1 + hed
Psp
where
P hed
sp =
1
2 ( 0 )
λ0
kh
2
( c1 )
80π 2
 1 
3
1 3
2 ( 0 )
k h 60π 1 −  
3
= hed
Psw
λ0   ε r  
Note: “hed” refers to a unit-amplitude horizontal electric dipole.
Note: When we say “unit amplitude” here, we assume peak (not RMS) values.
74
CAD Formulas
Hence, we have
1
ehed
r = 3
3  1  1
1 + π ( k0 h )  1 − 
4  c1  ε r 
Physically, this term is the radiation efficiency of a

horizontal electric dipole (hed) on top of the substrate.
75
CAD Formulas
The constants are defined as
1 2/5
c1 =1 − +
εr ε r2
 3  1
( k0 W ) + ( a2 + 2a4 )   ( k0 W ) + c2   ( k0 L )
a2 2 4 2
p =+
1 2
10  560  5
 1 
+ a2 c2   ( k0 W ) ( k0 L )
2 2
 70 
c2 = − 0.0914153
a2 = − 0.16605
a4 = 0.00761
76
CAD Formulas
Improved formula for HED surface-wave power (due to Pozar)
ε r ( x − 1)
3/2
ηk 2 2
=
hed 0 0 0
P
8 ε r (1 + x1 ) + (k0 h) x02 − 1 (1 + ε r2 x1 )
sw
x02 − 1 −ε r2 + α 0α1 + ε r ε r2 − 2α 0α1 + α 02

x1 = x0 = 1 +
ε r − x02 ε r2 − α12
α 0 = s tan ( k0 h ) s 
1
−  tan ( k0 h ) s  +
α1 =
( k0 h ) s 

s  cos ( k0 h ) s  
2
=s ε r −1
D. M. Pozar, “Rigorous Closed-Form Expressions for the Surface-Wave Loss of Printed
Antennas,” Electronics Letters, vol. 26, pp. 954-956, June 1990.
Note: The above formula for the surface-wave power is different from that given in Pozar’s paper by
a factor of 2, since Pozar used RMS instead of peak values.
77
CAD Formulas
Bandwidth
1   Rsave   1   16   p c1   h   W  1 
BW =+  d    +      hed 
2   π η0   h / λ0   3   ε r   λ0   L   er  
1
Q= Comments: For a lossless patch, the bandwidth is
approximately proportional to the patch width and
2 BW to the substrate thickness. It is inversely
proportional to the substrate permittivity.
For very thin substrates the bandwidth will
increase, but as the expense of efficiency.
BW is defined from the frequency limits f1 and f2 at which SWR = 2.0.

f 2 − f1
BW = (multiply by 100 if you want to get %)
f0
=
Rsave (R s
patch
+ Rsground ) / 2
78
CAD Formulas
Quality Factor Q
Us
Q ≡ ω0 U s = energy stored in patch cavity
P
P = power that is radiated and dissipated by patch
1 P
=
Q ω0U s
P = Pd + Pc + Psp + Psw
1 1 1 1 1
= + + +
Q Qd Qc Qsp Qsw
79
CAD Formulas
Q Components
Qd = 1 / tan δ
 η 0   ( k0 h) 
Qc =    ave  =
Rsave (Rs
patch
+ Rsground ) / 2
 2   Rs 
3  εr  L  1 
Qsp ≈     The constants p and c1 were defined previously.
16  pc1   W   h / λ0 
 erhed  erhed =
1
Qsw = Qsp  hed 
3
 1 − e  3  1  1
r
1 + π ( k0 h )  1 − 
4  c1  ε r 
80
CAD Formulas
Resonant Input Resistance
Probe-feed Patch
 π x0 
=
R R= Redge cos 
max
in 
2
 L 
 4η0   L   h 
   
 π   W   λ0 
Redge =
 R   1   16   p c1   W  h  1 
d +  s   +   L     hed 
π η
 0  h / λ0   3  ε
 r    λ0   er 
Comments: For a lossless patch, the resonant resistance is approximately

independent of the substrate thickness. For a lossy patch it tends to zero as the
substrate gets very thin. For a lossless patch it is inversely proportional to the
square of the patch width and it is proportional to the substrate permittivity.
81
CAD Formulas
Directivity
 3  εr 
D= ( ( k1h ) )
2
 tanc
 pc1   ε r + tan ( k1h ) 
2
k1 = k0 ε r
where
tanc ( x ) ≡ tan ( x ) / x
The constants p and c1 were defined previously.
82
CAD Formulas
Directivity (cont.)
For thin substrates:
3
D≈
p c1
(The directivity is essentially independent of the substrate thickness.)
83
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
84
Radiation Pattern
There are two models often used for calculating the radiation pattern:
 Electric current model
 Magnetic current model
Note: The origin is placed at the center of the patch,

at the top of the substrate, for the pattern calculations.
Patch
x
h Probe εr
Coax feed
85
Radiation Pattern
Electric current model:
We keep the physical currents flowing on the patch (and feed).
Patch
x
h Probe εr
Coax feed
= J top
J spatch s + J bot
s J spatch
x
h J sprobe εr
86
Radiation Pattern
Magnetic current model:
We apply the equivalence principle and invoke the (approximate) PMC condition
at the edges.
Equivalence surface Patch
x
J se = nˆ × H
h Probe εr
M se =−nˆ × E
Coax feed
The equivalent
surface current is
approximately zero
on the top surface
x
(weak fields) and h M se M se εr
the sides (PMC).
We can ignore it on
the ground plane (it
does not radiate). M se =−nˆ × E
87
Radiation Pattern
Theorem
The electric and magnetic models yield identical patterns
at the resonance frequency of the cavity mode.
Assumption:
The electric and magnetic current models are based on the fields of a
single cavity mode, corresponding to an ideal cavity with PMC walls.
D. R. Jackson and J. T. Williams, “A Comparison of CAD Models for Radiation from Rectangular
Microstrip Patches,” Intl. Journal of Microwave and Millimeter-Wave Computer Aided Design, vol. 1,
no. 2, pp. 236-248, April 1991.
88
Radiation Pattern
Comments on the Substrate Effects
 The substrate can be neglected to simplify the far-field calculation.

 When considering the substrate, it is most convenient to assume an infinite
substrate (in order to obtain a closed-form solution).
 Reciprocity can be used to calculate the far-field pattern of electric or
magnetic current sources inside of an infinite layered structure.
 When an infinite substrate is assumed, the far-field pattern always goes to
zero at the horizon.
no. 2, pp. 236-248, April 1991.
89
Radiation Pattern
Comments on the Two Models
 For the rectangular patch, the electric current model is the simplest since
there is only one electric surface current (as opposed to four edges).
 For the rectangular patch, the magnetic current model allows us to classify
the “radiating” and “nonradiating” edges.
 For the circular patch, the magnetic current model is the simplest since there
is only one edge (but the electric surface current is described by Bessel
functions).
y
πx 
J s = xˆ A10 cos   πx 
 L  “Radiating edges” Ez = − sin  
 L 
On the nonradiating edges, the

Js magnetic currents are in opposite
W x directions across the centerline (x = 0).
M se =−nˆ × E
M se
L “Nonradiating edges”
90
Radiation Pattern
Rectangular Patch Pattern Formula
(The formula is based on the electric current model.)
L
x
h εr
Infinite ground plane and substrate
H-plane
The origin is at the
center of the patch. y
(1,0) mode
W x E-plane
 πx 
J s  x cos  
ˆ
 L 
The probe is on the x axis.
L
91
Radiation Pattern
The far-field pattern can be determined by reciprocity.
  ky W    kx L  
 sin     cos   
 π WL    2    2  
Ei (r , θ , φ ) = Ei ( r , θ , φ ) 
hex

 2   ky W    π 2  k L 2 
    − x  
 2   2   2  
y
i = θ or φ
k x = k0 sin θ cos φ
k y = k0 sin θ sin φ Js
W x
The “hex” pattern is for a
horizontal electric dipole in the x direction,
sitting on top of the substrate.
L
no. 2, pp. 236-248, April 1991.
92
Radiation Pattern
Eφhex ( r , θ , φ ) = − E0 sin φ F (θ )
Eθhex ( r , θ , φ ) = E0 cos φ G (θ )
where
 − jω µ0  − jk0 r
E0 =  e
 4π r 
2 tan ( k0 h N (θ ) )
F (θ ) = 1 + Γ TE
(θ ) =
tan ( k0 h N (θ ) ) − j N (θ ) sec θ
2 tan ( k0 h N (θ ) ) cos θ
(θ ) cos θ (1 + ΓTM=
G= (θ ) ) εr
tan ( k0 h N (θ ) ) − j cos θ
N (θ )
N (=
θ) ε r − sin 2 (θ ) Note: To account for lossy substrate, use
ε r → ε rc = ε r (1 − j tan δ )
93
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
94
Input Impedance
Various models have been proposed over the years for calculating the
input impedance of a microstrip patch antenna.
 Transmission line model

 The first model introduced
 Very simple
 Cavity model (eigenfunction expansion)

 Simple yet accurate for thin substrates
 Gives physical insight into operation
 CAD circuit model

 Extremely simple and almost as accurate as the cavity model
 Spectral-domain method
 More challenging to implement
 Accounts rigorously for both radiation and surface-wave excitation
 Commercial software
 Very accurate
 Can be time consuming
95
Input Impedance
Comparison of the Three Simplest Models
Circuit model of patch Transmission line model of patch
Cavity model (eigenfunction expansion) of patch

ε r = 2.2 L = 6.255 cm x0 = 6.255 cm
tan δ = 0.001 W / L = 1.5 y0 = 0
h = 1.524 mm σ 3.0 × 107 S/m
= a = 0.635mm
Results for a typical patch show that the first

three methods agree very well, provided the
correct Q is used and the probe inductance is
accounted for.
96
Input Impedance
CAD Circuit Model for Input Impedance
The circuit model discussed assumes a probe feed.

Other circuit models exist for other types of feeds.
Note: The mathematical justification of the CAD circuit model comes

from the cavity-model eigenfunction analysis (see Appendix).
The transmission-line model and the cavity model are

discussed in the Appendix.
97
Input Impedance
Probe-fed Patch
 Near the resonance frequency, the patch cavity can be approximately modeled
as a resonant RLC circuit.
 The resistance R accounts for radiation and losses.
 A probe inductance Lp is added in series, to account for the “probe inductance”
of a probe feed.
Probe Patch cavity
L
Lp
Zin R C
98
Input Impedance
R
Z in ≈ jω Lp +
 f f0 
1 + jQ  − 
 f0 f 
R 1
Q= BW = BW is defined here by SWR < 2.0 when the
ω0 L 2Q RLC circuit is fed by a matched line (Z0 = R).
1
ω0 2=
= π f0
LC
L
Lp
Zin R C
Z=
in Rin + jX in
99
Input Impedance
R
Rin = 2
  f f0  =
R R= max
R
1 + Q  −   in f = f
in 0
  f0 f 
R is the input resistance at the resonance of the patch cavity
(the frequency that maximizes Rin).
f = f0 (resonance of RLC circuit)

Lp
L
max C
R in R
f = f0
100
Input Impedance
R
Z in ≈ jω Lp +
 f f0 
1 + jQ  − 
 f0 f 
The input resistance is determined once we know four parameters:
 f0: the resonance frequency of the patch cavity
CAD formulas
for the first three  R: the input resistance at the cavity resonance frequency f0
parameters
have been given  Q: the quality factor of the patch cavity
earlier.
 Lp: the probe inductance
Lp (R, f0, Q)
L
C
Zin R
101
Input Impedance
Results: Input Resistance vs. Frequency
Note: “exact” means the cavity model will all infinite modes.
80
Rectangular patch
70
60
CAD
50 exact
Rin ( Ω )
40
Frequency where
30 the input resistance
20
is maximum (f0):
Rin = R
10
0
4 4.5 5 5.5 6
FREQUENCY (GHz)
εr = 2.2 W/L = 1.5 L = 3.0 cm 102

Input Impedance
Results: Input Reactance vs. Frequency
Note: “exact” means the cavity model will all infinite modes.
Frequency80 where the input
resistance is maximum (f0) Rectangular patch
60
CAD
exact
40
Xin ( Ω )
20
Xp
Shift due to probe reactance

-20
-40
4 4.5 5 5.5 6
FREQUENCY (GHz)
Frequency where the
input impedance is real
εr = 2.2 W/L = 1.5 L = 3.0 cm 103

Input Impedance
Approximate CAD formula for probe (feed) reactance (in Ohms)
a = probe radius h = probe height
η0   2 
=Xp ( k0 h )  − γ + ln   
2π   ε r ( k0 a )  
This is based on an infinite parallel-plate model.
εr 2a h
X p = ω Lp
γ  0.577216 (Euler’s constant)
=η0 µ0 / ε 0 376.7303 Ω
=
104
Input Impedance
 Feed (probe) reactance increases proportionally with substrate
thickness h.
 Feed reactance increases for smaller probe radius.
η0   2 
=Xp ( k0 h )  − γ + ln  

2π
  ε r ( k0 a )  
Important point:
If the substrate gets too thick, the probe reactance will make it difficult
to get an input match, and the bandwidth will suffer.
(Compensating techniques will be discussed later.)
105
Input Impedance
Results: Probe Reactance (Xf =Xp= ωLp)
40
35 CAD
exact Rectangular patch
30
εr = 2.2
25
Xf ( Ω )
y
20
W/L = 1.5
15
Note: “exact” means the cavity model with all infinite modes. (x0, y0) W
10 h = 0.0254 λ0
5
L L x a = 0.5 mm
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Center Xxr
r Edge
The normalized feed location ratio xr is zero at the center

xr = 2 ( x0 / L) - 1 of the patch (x = L/2), and is 1.0 at the patch edge (x = L).
106
Design Example
Design a probe-fed rectangular patch antenna on a substrate having a relative permittivity of 2.33 and
a thickness of 62 mils (0.1575 cm). (Rogers RT Duroid 5870). Choose an aspect ratio of W / L = 1.5.
The patch should resonate at the operating frequency of 1.575 GHz (the GPS L1 frequency). Ignore
the probe inductance in your design, but account for fringing at the patch edges when you determine
the dimensions. At the operating frequency the input impedance should be 50 Ω (ignoring the probe
inductance). Assume an SMA connector is used to feed the patch along the centerline (at y = W / 2),
and that the inner conductor of the SMA connector has a radius of 0.635 mm. The copper patch and
ground plane have a conductivity of σ = 3.0 ×107 S/m and the dielectric substrate has a loss tangent of
tanδ = 0.001.
1) Calculate the following:
 The final patch dimensions L and W (in cm)

 The feed location x0 (distance of the feed from the closest patch edge, in cm)
 The bandwidth of the antenna (SWR < 2 definition, expressed in percent)
 The radiation efficiency of the antenna (accounting for conductor, dielectric, and surface-
wave loss, and expressed in percent)
 The probe reactance Xp at the operating frequency (in Ω)
 The expected complex input impedance (in Ω) at the operating frequency, accounting for the
probe inductance
 Directivity
 Gain
2) Plot the input impedance vs. frequency.

107
Design Example
y
Results from CAD formulas
Feed at (x0, y0)
1) L = 6.07 cm, W = 9.11 cm
2) x0 = 1.82 cm
3) BW = 1.24% Y0 = W/2
4) er = 81.9% W
5) Xp = 11.1 Ω
6) Zin = 50.0 + j(11.1) Ω x
7) D = 5.85 (7.67 dB) L
8) G = (D)(er) = 4.80 (6.81 dB)
108
Design Example
R
Z in ≈ jX p +
 f f 
Results from CAD formulas 1 + jQ  − 0 
 f0 f 
f0 = 1.575 ×109 Hz
R = 50 Ω
60
Q = 56.8 50.255
Xp = 11.1 Ω 50
Rin
40
Ω 30
Rin( fghz)
20
Xin( fghz)
10
Xin
− 10
− 13.937
− 20
1.5 1.525 1.55 1.575 1.6 1.625 1.65
1.5 fghz 1.65
f (GHz)
109
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
110
Circular Polarization
Three main techniques:
1) Single feed with “nearly degenerate” eigenmodes (compact but small

CP bandwidth).
2) Dual feed with delay line or 90o hybrid phase shifter (broader CP
bandwidth but uses more space).
3) Synchronous subarray technique (produces high-quality CP due to

cancellation effect, but requires even more space).
The techniques will be illustrated with a rectangular patch.
111
Single Feed Method
(0,1)
The feed is on the diagonal.
The patch is nearly
(but not exactly) square. W
(1,0)
L ≈W
L
Basic principle: The two dominant modes (1,0) and (0,1) are
excited with equal amplitude, but with a ±45o phase.
112
Design equations: y
fx + f y
f CP =
2
(0,1)
The optimum CP frequency is the
1
average of the x and y resonance BW =
frequencies. 2Q W
(1,0)
(SWR < 2 )
 1  x0 = y0
f x = f CP  1   
 2Q   x
Top sign for LHCP,
 L
 1  
bottom sign for RHCP.
=f y f CP  1 ±  
 2Q 
The frequency f0 is also the resonance frequency: Z=

in R=
in R=
x Ry
The resonant input resistance of the CP patch is the same as what a
linearly-polarized patch fed at the same position would be.
113
Other Variations
Note: Diagonal modes are used as degenerate modes
y y
L L
x x
L L
Patch with slot Patch with truncated corners

114
Here we compare bandwidths (impedance and axial-ratio):
1
=
Linearly-polarized (LP) patch: LP
BWSWR (SWR < 2)
2Q
Circularly-polarized (CP) single-feed patch:
=
BW CP
SWR
Q
2
=
(SWR < 2) BW CP
AR
0.348
Q
( AR < 2 (3dB) )
The axial-ratio bandwidth is small when using the single-feed method.
W. L. Langston and D. R. Jackson, “Impedance, Axial-Ratio, and Receive-Power Bandwidths of

Microstrip Antennas,” IEEE Trans. Antennas and Propagation, vol. 52, pp. 2769-2773, Oct. 2004.
115
Dual-Feed Method
y RHCP
Phase shift realized with delay line: L
P
L
P+λg/4
116
Phase shift realized with 90o quadrature hybrid (branchline coupler)
RHCP
Z0 Z0 / 2 Z0
Feed
λg/4 Z0
50 Ohm load
λg/4
This gives us a higher bandwidth than the simple power divider,

but requires a load resistor.
117
Synchronous Rotation
Multiple elements are rotated in space and fed with phase shifts.
-180o
-90o
-270o
0o
Because of symmetry, radiation from higher-order modes (or probes)
tends to be reduced, resulting in good cross-pol.
118
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
119
Circular Patch
a x
h εr
120
Circular Patch
Resonance Frequency
From separation of variables:
Ez = cos ( mφ ) J m ( k1 ρ ) PMC
a
k1 = k0 ε r
Jm = Bessel function of first kind, order m.
∂Ez
=0 J m′ ( k1a ) = 0
∂ρ ρ =a
121
Circular Patch
Resonance Frequency
J m′ ( k1a ) = 0
a PMC
This gives us
′
k1a = xmn
(nth root of Jm′ Bessel function)
c
f mn = ′
xmn c=
1
2π ε r µ 0ε 0
122
Circular Patch
Resonance Frequency
′
Table of values for xmn
n /m 0 1 2 3 4 5
1 3.832 1.841 3.054 4.201 5.317 5.416
2 7.016 5.331 6.706 8.015 9.282 10.520
3 10.173 8.536 9.969 11.346 12.682 13.987
Dominant mode: TM11
c
f11 = x11′ ′ ≈ 1.841
x11
2π a ε r
123
Circular Patch
Dominant mode: TM11
y y
x x
Circular patch Square patch

W=L
The circular patch is somewhat similar to a square patch.
124
Circular Patch
Fringing extension
ae = a + ∆a
c a PMC
f11 = x11′
2π ae ε r
a + ∆a
“Long/Shen Formula”:
2h  πa   h  πa  
ae =
a 1+   2h 
ln + 1.7726  or ∆a ≈  ln   + 1.7726 
π aε r     πε r   2h  
L. C. Shen, S. A. Long, M. Allerding, and M. Walton, "Resonant Frequency of a Circular Disk Printed-
Circuit Antenna," IEEE Trans. Antennas and Propagation, vol. 25, pp. 595-596, July 1977.
125
Circular Patch
Patterns
(The patterns are based on the magnetic current model.)
2a
x
h εr
Infinite GP and substrate
H-plane
The origin is at the y

center of the patch. x E-plane
a The probe is on the x axis.
In patch cavity:
 J1 ( k1 ρ )   1 
Ez ( ρ , φ ) = cosφ     k1 = k0 ε r
 J1 ( k1a )   h 
(The edge voltage has a maximum of one volt.)
126
Circular Patch
Patterns
E0
Eθ ( r , θ , φ ) = 2π a
R
tanc ( k z1h ) cosφ J1′ ( k0 a sinθ ) Q (θ )
η0
E0  J1 ( k0 a sinθ ) 
Eϕ ( r , θ , φ ) = −2π a
R
tanc ( k z1h ) sinφ   P (θ )
η0  k0 a sinθ 
where
tanc  x  tan  x / x
 −2 jN (θ ) 
P (θ ) cosθ (1 − Γ =
= (θ ) ) cosθ 
TE

 tan ( k0 hN (θ ) ) − jN (θ ) secθ 
 εr   − jω µ0  − jk0 r
−2 j   cosθ E0 = 
 N (θ )  e
Q (θ ) = 1 − ΓTM (θ ) =  4π r 
εr
tan ( k0 h N (θ ) ) − j cosθ
N (θ ) N (=
θ) ε r − sin 2 (θ )
ε r ε r′ (1 − j tan δ )
Note: To account for lossy substrate, use=
127
Circular Patch
Input Resistance
ρ0
 J12 ( k1 ρ0 ) 
Rin ≈ Redge  2 
 J1 ( k1a ) 
k1 = k0 ε r
128
Circular Patch
Input Resistance (cont.)
 1 
Redge =  er
 2 Psp 
where er = radiation efficiency
π /2
π
( k0 a ) ∫ tanc2 ( k0 hN (θ ) )
2
Psp =
8η0 0

⋅  Q (θ ) J1 ( k0 a sin θ ) + P (θ ) J inc
2
′ 2 2 2
( k0 a sin θ )  sin θ dθ
 
J inc ( x ) = J1 ( x ) / x
Psp = power radiated into space by circular patch with maximum

edge voltage of one volt.
129
Circular Patch
Input Resistance (cont.)
CAD Formula:
π
Psp = ( k0 a ) 2 I c
8η0 e0 = 1
e2 = −0.400000
e4 = 0.0785710
6
4 pc = ∑ ( k0 a ) e2 k −7.27509 × 10−3
2k
e6 =
I c = pc
3 k =0 =e8 3.81786 × 10−4
−1.09839 × 10−5
e10 =
=e12 1.47731 × 10−7
130
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
131
Improving Bandwidth
Some of the techniques that have been successfully

developed are illustrated here.
The literature may be consulted for additional designs and variations.
132
Improving Bandwidth
Probe Compensation
L-shaped probe: As the substrate
thickness increases the
probe inductance limits
the bandwidth – so we
compensate for it.
Top view
Capacitive “top hat” on probe:
133
Improving Bandwidth
SSFIP: Strip Slot Foam Inverted Patch (a version of the ACP).
 Bandwidths greater than 25% have been achieved.
 Increased bandwidth is due to the thick foam substrate and
also a dual-tuned resonance (patch+slot).
Note: There is no probe inductance to worry about here.
Patch substrate Patch
Foam
Microstrip
substrate
Microstrip line Slot
J.-F. Zürcher and F. E. Gardiol, Broadband Patch Antennas, Artech House, Norwood, MA, 1995.
134
Improving Bandwidth
Stacked Patches
 Bandwidth increase is due to thick low-permittivity antenna

substrates and a dual or triple-tuned resonance.
 Bandwidths of 25% have been achieved using a probe feed.
 Bandwidths of 100% have been achieved using an ACP feed.
Patch substrates Coupling patch

Top Patch
Microstrip
substrate
Microstrip line
Slot
135
Improving Bandwidth
Stacked Patches
-5
-10
Return Loss (dB)

-15
Measured
-20
Computed
-25
-30
-35
-40
3 4 5 6 7 8 9 10 11 12
Frequency (GHz)
Stacked patch with ACP feed
Bandwidth (S11 = -10 dB) is about 100%

136
Improving Bandwidth
Stacked Patches
90 80
100 70
110
0 60
12
0 50
13
40
14
0
30
15
160
20
170
10
0
180
0
0.2
0.5
10
-170
-10
-160
-20
13 GHz
-30
-15
40
-4
-1
0
-5
30
4 GHz
0
-1
0 -60
-12 -70
-110 -80
-100 -90
Stacked patch with ACP feed
Two extra loops are observed on the Smith chart.

137
Improving Bandwidth
Parasitic Patches
Radiating Edges Gap Coupled

Microstrip Antennas Mush of this work
(REGCOMA). was pioneered by
K. C. Gupta.
Non-Radiating Edges Gap

Coupled Microstrip Antennas
(NEGCOMA)
Four-Edges Gap Coupled

Microstrip Antennas
(FEGCOMA)
Bandwidth improvement factor:

REGCOMA: 3.0, NEGCOMA: 3.0, FEGCOMA: 5.0?
138
Improving Bandwidth
Direct-Coupled Patches
Radiating Edges Direct

(REDCOMA).
Non-Radiating Edges Direct

(NEDCOMA)
Four-Edges Direct Coupled

Microstrip Antennas
(FEDCOMA)
Bandwidth improvement factor:

REDCOMA: 5.0, NEDCOMA: 5.0, FEDCOMA: 7.0
139
Improving Bandwidth
U-Shaped Slot
The introduction of a U-shaped slot can give a

significant bandwidth (10%-40%).
(This is due to a double resonance effect, with two different modes.)
“Single Layer Single Patch Wideband Microstrip Antenna,” T. Huynh and K. F. Lee,
Electronics Letters, Vol. 31, No. 16, pp. 1310-1312, 1986.
140
Improving Bandwidth
Double U-Slot
A 44% bandwidth was achieved.
Y. X. Guo, K. M. Luk, and Y. L. Chow, “Double U-Slot Rectangular Patch Antenna,”

Electronics Letters, Vol. 34, No. 19, pp. 1805-1806, 1998.
141
Improving Bandwidth
E Patch
A modification of the U-slot patch.
A bandwidth of 34% was achieved (40% using a capacitive “washer” to

compensate for the probe inductance).
B. L. Ooi and Q. Shen, “A Novel E-shaped Broadband Microstrip Patch Antenna,”

Microwave and Optical Technology Letters, vol. 27, No. 5, pp. 348-352, 2000.
142
Multi-Band Antennas
A multi-band antenna is sometimes more desirable than a broadband

antenna, if multiple narrow-band channels are to be covered.
General Principle:
Introduce multiple resonance paths into the antenna.
143
Multi-Band Antennas
Low-band
Low-band
Feed High-band Feed
Low-band
High-band
Dual-band E patch Dual-band patch with parasitic strip
144
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
145
Miniaturization
• High Permittivity
• Quarter-Wave Patch
• PIFA
• Capacitive Loading
• Slots
• Meandering
Note: Miniaturization usually comes at a price of reduced bandwidth!
Usually, bandwidth is proportional to the volume of the patch cavity,

as we will see in the examples.
146
Miniaturization
High Permittivity
εr = 1
εr = 4
Size reduction
W W′ =W / 2
(Same aspect ratio)
L′ = L / 2
The smaller patch has about one-fourth the bandwidth of the original patch.
(Bandwidth is inversely proportional to the permittivity.)
147
Miniaturization
Quarter-Wave patch
Ez = 0
Short-circuit vias
W W
L L′ = L / 2
The new patch has about one-half the bandwidth of the original patch.

U s′ = U s / 2
Neglecting losses:
Us 
Q ′ = 2Q
Q = ω0 
Pr Pr′ = Pr / 4 

Note: 1/2 of the radiating magnetic current
148
Miniaturization
Smaller Quarter-Wave patch
A quarter-wave patch with the same aspect ratio W/L as the original patch
W′ =W / 2
Ez = 0 Short-circuit vias
Width reduction
W W W′
L′ = L / 2
L L′ = L / 2
The new patch has about one-half the bandwidth of the original quarter-
wave patch, and hence one-fourth the bandwidth of the regular patch.
(Bandwidth is proportional to the patch width.)
149
Miniaturization
Quarter-Wave Patch with Fewer Vias
Use fewer vias
W W
L′ = L / 2 L′′
L′′ < L′
Fewer vias actually gives more miniaturization!
(The edge has a larger inductive impedance: explained on the next slide.)
150
Miniaturization
Quarter-Wave Patch with Fewer Vias
Inductance
Short Open
The Smith chart provides a simple explanation for the length reduction.
151
Miniaturization
Planar Inverted F (PIFA)
Feed
Shorting strip or via
Top view
A single shorting strip or via is used.
This antenna can be viewed as a limiting case of the via-loaded patch,

or as an LC resonator.
152
Miniaturization
PIFA with Capacitive Loading
Feed
Shorting plate Top view
The capacitive loading allows for the length of the PIFA to be reduced.
1
ω0 =
LC
153
Miniaturization
Circular Patch Loaded with Vias
Feed c
2a
Patch Metal vias
The patch has a monopole-like pattern

The patch operates in the (0,0) mode, as an LC resonator
(Hao Xu Ph.D. dissertation, University of Houston, 2006)

154
Miniaturization
Circular Patch Loaded with Vias
Example: Circular Patch Loaded with 2 Vias
0
0
315
-10
45
-20
-5
-30
S11[db]
-10
-20
-30
-40
-10
270
90
-15
E-theta
E-phi
-20
225
135
0.5 1 1.5 2 2.5
Frequency [GHz]
180
Unloaded: resonance frequency = 5.32 GHz.
(Miniaturization factor = 4.8)
155
Miniaturization
Slotted Patch
Top view
0o ±90o
Linear CP
The slot forces the current to flow through a longer path,

increasing the effective dimensions of the patch.
156
Miniaturization
Meandering
Via
Feed
Feed Via
Meandered quarter-wave patch Meandered PIFA
Meandering forces the current to flow through a longer path,

increasing the effective dimensions of the patch.
Meandering also increases the capacitance of the PIFA line.
157
Outline
 Feeding methods
 CAD Formulas
 Input Impedance
 Circular patch
 Miniaturization
158
Reducing Surface and Lateral Waves
Reduced Surface Wave (RSW) Antenna
z
Feed
b
Shorted annular ring ρ0
ρo
a
x a b
h
Ground plane Feed
SIDE VIEW TOP VIEW
D. R. Jackson, J. T. Williams, A. K. Bhattacharyya, R. Smith, S. J. Buchheit, and S. A. Long,

“Microstrip Patch Designs that do Not Excite Surface Waves,” IEEE Trans. Antennas Propagat.,
vol. 41, No 8, pp. 1026-1037, August 1993.
159
Reducing surface-wave excitation and lateral radiation reduces
edge diffraction.
Space-wave radiation (desired)
Lateral radiation (undesired)
Diffracted field at edge
Surface waves (undesired)
160
y Principle of Operation
TM11 mode:
 −1 
a x Ez ( ρ , φ ) = V0   cosφ J1 ( k1 ρ )
 hJ1 ( k1a ) 
M sφ
k= k= k0 ε r V0
Ez = − cos φ
1
At edge:
h
M s =− nˆ × E =− ρˆ × ( zE
ˆ z)
M sφ (φ ) = Ez ( a, φ )
V0
M sφ = − cos φ
h
161
y
Substrate
V0
a x M sφ = − cos φ
h
M sφ
Surface-Wave Excitation:
( )
EzTM 0 = ATM 0 cos φ H1( 2) βTM 0 ρ e − jk z 0 z
(z > h)
(
ATM 0 = AJ1′ βTM 0 a )
Set ( )
J1′ βTM 0 a = 0
162
y
Substrate
βTM a = x1′n
0
a x
For TM11 mode: x11′ ≈ 1.841
M sφ
Hence βTM a = 1.841

0
Patch resonance: k1a = 1.841
Note: βTM < k1

0
The RSW patch is too big to be resonant.
163
z
Feed
b
Shorted annular ring ρ0
ρo
a
x a b
h
Ground plane Feed
SIDE VIEW TOP VIEW
βTM b = 1.841
0
The radius a is chosen to make the patch resonant:

 k1 x11
′ 
J 1′  
J 1 ( k1a )  βTM 0 
= 
Y1 ( k1a )  k1 x11
′ 
Y1′
 βTM 
 0 
164
Reducing the Lateral Radiation
y Ground plane
V0
M sφ = − cos φ
h
a x
Assume no substrate outside of patch
M sφ (or very thin substrate):
 1  − jk0 ρ
Space-Wave Field: E SP
= ASP cos φ   e (z = h)
ρ
z
ASP = CJ1′ ( k0 a )
Set J1′ ( k0 a ) = 0 k0a = 1.841

165
y
Substrate
For a thin substrate:
a x βTM ≈ k0
0
M sφ
The same design reduces
both surface-wave fields and
lateral-radiation fields.
Note: The diameter of the RSW antenna is found from
k0 a = 1.841
2a Note: The size is approximately independent of the

= 0.586 permittivity (the patch cannot be miniaturized by
λ0 choosing a higher permittivity!).
166
E-plane Radiation Patterns
Measurements were taken on a 1 m diameter circular ground plane at 1.575 GHz.
Measurement
0
Theory (infinite GP) 0
30 -30 30 -30
-10 -10
60 -20 -60 60 -20 -60
-30 -30
-40 -30 -20 -10 -40 -30 -20 -10

90 -90 90 -90
120 240 120 240
150 210 150 210
180 180
Conventional RSW
167
Reducing surface-wave excitation and lateral
radiation reduces mutual coupling.
Space-wave radiation
Lateral radiation
Surface waves
168
Reducing surface-wave excitation and lateral radiation reduces mutual coupling.
-10 E-plane RSW - Measured

RSW - Theory
-20 Conv - Measured
-30 Conv - Theory
-40
S12 [dB]
-50 1/r
-60
-70
-80 1/r3
-90
-100
0 1 2 3 4 5 6 7 8 9 10
Separation [Wavelengths]
“Mutual Coupling Between Reduced Surface-Wave Microstrip Antennas,” M. A. Khayat, J. T. Williams, D. R.

Jackson, and S. A. Long, IEEE Trans. Antennas and Propagation, Vol. 48, pp. 1581-1593, Oct. 2000.
169
References
General references about microstrip antennas:
Microstrip Patch Antennas, K. F. Fong Lee and K. M. Luk, Imperial

College Press, 2011.
Microstrip and Patch Antennas Design, 2nd Ed., R. Bancroft, Scitech

Publishing, 2009.
Microstrip Patch Antennas: A Designer’s Guide, R. B. Waterhouse,

Kluwer Academic Publishers, 2003.
Microstrip Antenna Design Handbook, R. Garg, P. Bhartia, I. J. Bahl,

and A. Ittipiboon, Editors, Artech House, 2001.
Advances in Microstrip and Printed Antennas, K. F. Lee, Editor, John
Wiley, 1997.
170
References (cont.)
General references about microstrip antennas (cont.):
CAD of Microstrip Antennas for Wireless Applications, R. A. Sainati,

Artech House, 1996.
Microstrip Antennas: The Analysis and Design of Microstrip Antennas

and Arrays, D. M. Pozar and D. H. Schaubert, Editors, Wiley/IEEE
Press, 1995.
Millimeter-Wave Microstrip and Printed Circuit Antennas, P. Bhartia,
Artech House, 1991.
The Handbook of Microstrip Antennas (two volume set), J. R. James

and P. S. Hall, INSPEC, 1989.
Microstrip Antenna Theory and Design, J. R. James, P. S. Hall, and

C. Wood, INSPEC/IEE, 1981.
171
References (cont.)
More information about the CAD formulas presented here

for the rectangular patch may be found in:
Microstrip Antennas, D. R. Jackson, Ch. 7 of Antenna Engineering

Handbook, J. L. Volakis, Editor, McGraw Hill, 2007.
Computer-Aided Design of Rectangular Microstrip Antennas, D. R.

Jackson, S. A. Long, J. T. Williams, and V. B. Davis, Ch. 5 of Advances
in Microstrip and Printed Antennas, K. F. Lee, Editor, John Wiley, 1997.
172
References (cont.)
References devoted to broadband microstrip antennas:
Compact and Broadband Microstrip Antennas, K.-L. Wong,

John Wiley, 2003.
Broadband Microstrip Antennas, G. Kumar and K. P. Ray,

Artech House, 2002.
Broadband Patch Antennas, J.-F. Zürcher and F. E. Gardiol,

Artech House, 1995.
173
174
 Transmission line model
 Cavity model (eigenfunction expansion)
175
Input Impedance
Transmission Line Model for Input Impedance
 The model accounts for the probe feed to improve accuracy.

 The model assumes a rectangular patch
(it is difficult to extend to other shapes).
We think of the patch as a wide transmission line resonator (length L).
176
Input Impedance
y
Physical patch dimensions (W × L)
Js x0e= x0 + ∆L
We PMC y0e= y0 + ∆W
( x0e , y0e )
x
Le ε rceff
Denote
ke = k0 ε rceff Note:
∆L is from Hammerstad’s formula
ε r′ (1 − j leff )
∆W is from Wheeler’s formula
ε=
eff
rc
1 1 1 1 1
leff = = + + + A CAD formula for Q has been given earlier.
Q Qd Qc Qsp Qsw
177
Input Impedance
Commonly used fringing formulas
 ln 4 
∆W =
h  (Wheeler formula)
 π 
W 
+ 0.262
 ε reff + 0.300   h 
∆L =0.412h  eff   (Hammerstad formula)
εr − 0.258   W + 0.813 
 h 
 ε +1   ε −1  1
ε reff  r  +  r 
=
 2   2   h
1 + 12  
W 
178
Input Impedance
x=0 x = x0e x = Le
Z 0effc , kceff
Lp
X p = ω Lp Y0effc = 1 / Z 0effc
eff h 1 h
Z 0effc η=
= η
Z= in + jX p
c 0
in Z TL We ε rceff We
Zin
= / Z inTL jY0effc tan ( kceff x0e )
YinTL 1=
where
(
+ jY0effc tan kceff ( Le − x0e ) )
η0   2  γ  0.57722
=Xp ( k0 h )  − γ + ln   
2π   ε r ( k0 a )   (Euler's constant )
(from a parallel-plate model of probe inductance)
179
Input Impedance
Cavity Model
Here we use the cavity model to solve for the

input impedance of the rectangular patch antenna.
 It is a very efficient method for calculating the input impedance.

 It gives a lot of physical insight into the operation of the patch.
 The method is extendable to other patch shapes.
Y. T. Lo, D. Solomon, and W. F. Richards, “Theory and Experiment on Microstrip Antennas,”

IEEE Trans. Antennas Propagat., vol. AP-27, no. 3, pp. 137-145, March 1979.
180
y
Input Impedance
Physical patch dimensions (W × L)
x0e= x0 + ∆L
We ( x0e , y0e ) PMC
y0e= y0 + ∆W
x
Le
Denote ε rceff
ke = k0 ε rceff Note:
∆L is from Hammerstad’s formula
ε= ε r′ (1 − j leff )
eff
rc ∆W is from Wheeler’s formula
1 1 1 1
leff = + + + A CAD formula for Q has been given earlier.
Qd Qc Qsp Qsw
181
Input Impedance
Commonly used fringing formulas
 ln 4 
∆W =
h  (Wheeler formula)
 π 
W 
+ 0.262
 ε reff + 0.300   h 
∆L =0.412h  eff   (Hammerstad formula)
εr − 0.258   W + 0.813 
 h 
 ε +1   ε −1  1
ε reff  r  +  r 
=
 2   2   h
1 + 12  
W 
182
Input Impedance
Next, we derive the Helmholtz equation for Ez.
∇ × H= J i + jωε ceff E (Ampere’s law)
∇ × E = − jωµ H (Faraday’s law)
Substituting Faradays law into Ampere’s law, we have:
1
− ∇ × ( ∇ × E=
) J i + jωε ceff E
jωµ
∇ × ( ∇ × E ) = − jωµ J i + ke2 E
∇ ( ∇ ⋅ E ) − ∇ 2 E = − jωµ J i + ke2 E
∇ 2 E + ke2 E =jωµ J i
183
Input Impedance
Hence
jωµ J zi
∇ 2 Ez + ke2 Ez =
Denote ψ ( x, y ) = E z ( x, y )
Then ∇2ψ + ke2ψ =f ( x, y )

(scalar Helmholtz equation)
where
f ( x, y ) = jωµ J zi ( x, y )
184
Input Impedance
Introduce “eigenfunctions” ψmn (x,y):
∇ ψ mn ( x, y ) =
2
−λ ψ mn ( x, y )
2
mn
∂ψ mn
=0C
∂n
For rectangular patch we have, from separation of variables,
 mπ x   nπ y 
ψ mn ( x, y ) = cos   cos  
 e 
L  e 
W
 mπ  2  nπ  2 
= λmn
2
  +  
 e   e  
L W
185
Input Impedance
Assume an “eigenfunction expansion”:
ψ ( x, y ) = ∑ Amnψ mn ( x, y )
m,n
This must satisfy ∇ 2ψ + ke2ψ =f ( x, y )
Hence
∑ mn mn e ∑ Amnψ mn =
A
m,n
∇ 2
ψ + k 2
f ( x, y )
m,n
Using the properties of the eigenfunctions, we have
∑ mn e mn )ψ mn ( x, y) =
A
m,n
( k 2
− λ 2
f ( x, y )
186
Input Impedance
Next, we multiply by ψ m′n′ ( x, y ) and integrate.
Note that the eigenfunctions are orthogonal, so that
∫ψ
S
mn ( x, y )ψ m′n=
′ ( x, y ) dS 0, (m, n) ≠ (m′, n′)
Denote
< ψ mn ,ψ mn > =∫ψ mn
2
( x, y ) dS
S
We then have
Amn ( ke2 − λmn

2
) < ψ mn ,ψ mn > = < f ,ψ mn >
187
Input Impedance
Hence, we have
< f ,ψ mn >  1 
Amn =  2 2 
< ψ mn ,ψ mn >  ke − λmn 
For the patch problem we then have
 < J zi ,ψ mn >   1 
Amn = jωµ   2 2 
 < ψ mn ,ψ mn >   ke − λmn 
The field inside the patch cavity is then given by
Ez ( x, y ) = ∑ Amnψ mn ( x, y )
m,n
188
Input Impedance
To calculate the input impedance, we first calculate the complex

power going into the patch as
1
Pin = − ∫ E z ( x, y ) J zi*dV
2V
1
= − h ∫ E z ( x, y ) J zi* dS
2 S
1
= − h ∫ ∑ Amnψ mn J zi* dS
2 S m ,n
1
− h ∑ Amn < ψ mn , J zi* >
=
2 m ,n
189
Input Impedance
Hence
1
− h ∑ Amn < ψ mn , J zi* >
Pin =
2 m ,n
1   < ψ mn , J zi >   1 
− h ∑  jωµ 
=  2 2 
< ψ , J z >
i*
 < ψ mn ,ψ mn >   ke − λmn  

mn
2 m ,n 
1  <ψ , J i > 2  1 
= − h∑ jωµ  
mn z
 2 
2 m ,n  < ψ mn ,ψ mn >   ke2 − λmn 
 
1
Pin = Z in I in
2
Also,
2
2 Pin
so Z in = 2
I in
190
Input Impedance
Hence we have
 <ψ , J i > 2   
1 1
2 ∑
Z in = − jωµ h  mn z
  2 
I in m ,n  < ψ mn ,ψ mn >   ke − λmn 
2
 
∞ ∞
where ∑
m,n
=∑
=m 0=
∑
n 0
191
Input Impedance
Rectangular patch:
 mπ x   nπ y 
ψ mn = cos   cos  
 Le   We 
2 2
 mπ   nπ 
λ
=   +
2
mn 
 e   e
L W
ke = k0 ε rceff
where ε=
eff
rc ε r′ (1 − j leff )
2  mπ x   nπ y 
Le We
ψ mn ,ψ mn = ∫ cos   dx ∫0 cos  We  dy
2
0  Le 
192
Input Impedance
so
 We   Le 
ψ mn ,ψ
= mn     (1 + δ m 0 )(1 + δ n 0 )
 2  2 
1, m = 0
δ m0 =
0, m ≠ 0
To calculate ψ mn , J zi , assume a strip model as shown below.
ap Wp
( x0e , y0e )
( x0e , y0e )
Actual probe Strip model

193
Input Impedance
For a “Maxwell” strip current assumption, we have:
I in  e Wp e Wp 
J sz , y ∈  y0 − , y0 + 
 Wp 
2
 2 2 
 − ( y − y0 )
e 2
π 
 2 
Wp
W p = 4a p ( x0e , y0e )
Note: The total probe current is Iin.
194
Input Impedance
For a uniform strip current assumption, we have:
I in  e Wp e Wp 
=
J sz , y ∈  y0 − , y0 + 
Wp  2 2 
3
W p = a p e  4.482 a p
2
Wp
( x0e , y0e )
Note: The total probe current is Iin.
D. M. Pozar, “Improved Computational Efficiency for the Moment Method Solution of

Printed Dipoles and Patches,” Electromagnetics, Vol. 3, pp. 299-309, July-Dec. 1983.
195
Input Impedance
Assume a uniform strip current model
(the two models gives almost identical results):
Wp
y0e +
 mπ x0e 
2
 nπ y  1
ψ mn , J i
z = I in ∫ cos   cos   dy Use
e Wp
y −
 Le   We  W p
0
2 y y0e + y′
=
Wp
+
I in  mπ x0e  2  nπ e 
cos   ∫ cos   y0 + y′  dy '
Wp  Le  −Wp  We 
2 Integrates to zero
Wp
+
I in  mπ x  2   nπ y0e 
e
 nπ y0e  
cos   ∫  
0
cos  cos ( )
y ' − sin   ( )  dy ′
sin y ′
Wp  Le  −Wp   We   We  
2
I in  mπ x0e    nπ y0e   nπ W p 
= cos   
cos  p
W sinc  
Wp  e   e 
L W  2We 
196
Input Impedance
Hence
 mπ x0e   nπ y0e   nπ W p 
ψ mn , J i
z = I in cos   cos   sinc  
 e 
L  e 
W  2W e 
 nπ W p 
Note: It is the sinc   term that causes the series for Zin to converge.
 2 We 
We cannot assume a prove of zero radius, or else the series

will not converge – the input reactance will be infinite!
197
Input Impedance
Summary of Cavity Model
 <ψ , J i > 2   
1 1
2 ∑
  2 2 
 < ψ ψ >  − λ
 e mn 
I in m ,n , k
 mn mn
 mπ x0e   nπ y0e   nπ W p 
ψ mn , J i
z = I in cos   cos   sinc  
 e 
L  e 
W  2W e 
W  L 
ψ mn ,ψ
=     (1 + δ m 0 )(1 + δ n 0 )
e e
mn
 2  2 
2 2
 mπ   nπ 
λmn 
=2
 + 
 e   We 
L
ke = k0 ε rceff
198
Input Impedance
Summary (cont.)
ε=
eff
rc ε r′ (1 − j leff )
1 1 1 1 1
leff = = + + +
Q Qd Qc Qsp Qsw
A CAD formula for Q has already been given.
199
Input Impedance
Probe Inductance
 <ψ , J i > 2   
1 1
2 ∑
  2 2 
I in m ,n  < ψ mn ,ψ mn >   ke − λmn 
 
Note that
(1,0) = term that corresponds to dominant patch mode
(which corresponds to the RLC circuit).
Hence
 ψ ,Ji 2 
1    1 
2 ∑ 
jX p = − jωµ h
mn z
  2 2 
ψ ψ
I in ( m ,n )  mn mn   e
, k − λmn 
≠ (1,0)  
200
Input Impedance
RLC Model
The input impedance is in the form

∞ ∞
Z in = ∑ Z m,n
in where ∑ =∑ ∑
m,n m,n =m 0=n 0
where
1  < ψ mn , J z >  
 2

i
1
Z inm ,n = − jωµ h  2 2 
2
 < ψ ψ >  − λ
 e mn 
I in , k
 mn mn
After some algebra, we can write this as follows:
201
Input Impedance
Rmn
Z m,n
=
 1 
in
f rmn + jQ  f rmn − 
 f rmn 
where
1
Q=
leff and
f 2
f rmn = 1 ψ mn , J i
Pmn = µ h
z
f mn
I in
2
ψ mn ,ψ mn
 Pmn 
Rmn = 2
k l  ωmn
 mn eff 
202
Input Impedance
2
For f rmn ≈ 1 , we have
Rmn
Z m ,n
≈
 1 
in
1 + jQ  f rmn − 
 f rmn 
This is the mathematical form of the

input impedance of an RLC circuit.
203
Input Impedance
Circuit model
Note: The (0,0) mode has a uniform electric field,

and hence no magnetic field. It also has negligible
radiation (compared to the (1,0) mode).
(1, 0) (0,1)
(0, 0)
Zin
Note: This circuit model is accurate as long as we are near the

resonance of any particular RLC circuit.
204
Input Impedance
RLC Model
Approximately, Rmn ≈ 0 (m, n) ≠ (1, 0)
Hence, an approximate circuit model for operation near the

resonance frequency of the (1,0) mode is
(1, 0)
jX p
Zin
205

Introduction To Microstrip Antennas PDF

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Introduction To Microstrip Antennas PDF

Încărcat de

Drepturi de autor:

Formate disponibile

Introduction to Microstrip Antennas

 Provide an introduction to microstrip antennas.

 Some basic references are provided at the end of these viewgraphs.

ECE 6345: Microstrip Antennas

λ0 = wavelength of free space λ0 = c / f

k0 = wavenumber of free space =k0 ω=

ε r = relative permtitivity (dielectric constant) of substrate

 One of the most useful antennas at microwave frequencies (f > 1 GHz).

Microstrip line feed Coax feed

Rectangular Square Circular Annular ring

 Invented by Bob Munson in 1972 (but earlier work by Dechamps

 Became popular starting in the 1970s.

G. Deschamps and W. Sichak, “Microstrip Microwave Antennas,” Proc. of Third

R. E. Munson, “Microstrip Phased Array Antennas,” Proc. of Twenty-Second Symp.

R. E. Munson, “Conformal Microstrip Antennas and Microstrip Phased Arrays,” IEEE

 Low profile (can even be “conformal,” i.e. flexible to conform to a

 Low bandwidth (but can be improved by a variety of techniques). Bandwidths of

 Efficiency may be lower than with other antennas. Efficiency is limited by

 Cannot handle extremely large amounts of power (dielectric breakdown).

* Conductor and dielectric losses become more severe for thinner

 Cell phone antennas

Linear array (1-D corporate feed)

2-D 8X8 corporate-fed array 4 × 8 corporate-fed / series-fed array

The substrate is so thin that it can be bent to “conform” to the surface.

(Photo courtesy of Dr. Rodney B. Waterhouse)

Note: L is the resonant dimension (direction of current flow).

Some of the more common methods for

The feeding methods are illustrated for a rectangular patch,

Feed at (x0, y0) x

 Allows for planar feeding

Recent work has shown

Y. Hu, D. R. Jackson, J. T. Williams, and S. A. Long, “Characterization of the Input Impedance of

Microstrip line Microstrip

 Feed-line radiation is isolated from patch radiation

 Allows for use of different substrates to optimize

Y. T. Lo, D. Solomon, and W. F. Richards, “Theory and Experiment on Microstrip Antennas,”

 The patch acts approximately as a resonant cavity (with short-circuit (PEC)

 In a cavity, only certain modes are allowed to exist, at different resonance

 If the antenna is excited at a resonance frequency, a strong field is set up inside

(See next slide.)

On patch and ground plane: Et = 0 E = zˆ E z

Inside the patch cavity, because of the thin substrate,

Magnetic field inside patch cavity:

Note: The magnetic field is purely horizontal.

On the edges of the patch: y

Since the magnetic field is approximately

Actual patch PMC Model

From separation of variables:

The resonance frequency is mainly controlled by the

Approximately, (assuming PMC walls)

Note: A higher substrate permittivity allows for a smaller antenna

Note: Some authors use effective permittivity in this equation.

Note: Even though the Hammerstad formula

This is a good “rule of thumb” to give a quick estimate.

εr = 2.2 The resonance frequency has been normalized by the

 The bandwidth is directly proportional to substrate thickness h.

 However, if h is greater than about 0.05 λ0 , the probe inductance (for a

 The bandwidth is inversely proportional to εr (a foam substrate gives a high

 The bandwidth of a rectangular patch is proportional to the patch width W

Some Bandwidth Observations

 For a typical substrate thickness (h /λ0 = 0.02), and a typical substrate

 By using a thick foam substrate, bandwidth of about 10% can be achieved.