Spring 2015 Notes 5 ECE 6345 Prof. David R. Jackson ECE Dept. 1
Spring 2015
Notes 5
ECE 6345
Prof. David R. Jackson ECE Dept.
1
Overview
This set of notes discusses improved models of the probe inductance of a coaxially-fed patch (accurate for thicker
substrates). A parallel-plate waveguide model is initially assumed (at the end of the notes we will also look at the actual finite patch).
z
h ,r rε µ2a
x2b
2
Overview (cont.)
The following models are investigated:
Cosine-current model
Gap-source model
Frill model
H. Xu, D. R. Jackson, and J. T. Williams, “Comparison of Models for the Probe Inductance for a Parallel Plate Waveguide and a Microstrip Patch,” IEEE Trans. Antennas and Propagation, vol. 53, pp. 3229-3235, Oct. 2005.
Reference:
3
Derivations are given in the Appendix. Even more details may be found in the reference below.
Cosine Current Model
cosI z k z h
Note: The derivative of the current is zero at the top conductor (PEC).
2
2
0c
inPZ
I
Pc = complex power radiated by probe current
z
h ,r rε µ2a
x
I (z)
We assume a tube of current (as in Notes 4) but with a z variation.
4
*12
p
c z szS
P E J dS= ∫
5
( ) 222 (2)0 0 0 0 0 0
0
1 1( ) sec (1 ) ( ) ( )8
min r m m m mmr
Z k h k h I k H k a J k aρ ρ ρη ε δε
∞
=
= +
∑
Cosine Current Model (cont.)
Final result:
where
0
1, 00, 0m
mm
δ=
= ≠
2 2
2 ( ) sin( )1 ( ) ( )m
mo
khI khkh mδ π
= + −
1/2221m
mk khρπ = −
0/m mk k kρ ρ=
Note: The wavenumber kρm is chosen to be a positive real number or a
negative imaginary number.
Gap Source Model
An ideal gap voltage source of height ∆ is assumed at the bottom of the probe.
z
h ,r rε µ2a
x1V + - ∆
1
inZI
6
Gap Source Model (cont.)
7
(2)0
(2)00 0 0
( )1 24 sinc(1 ) ( )
min r
m m m m
H k aa mY jh k H k a h
ρ
ρ ρ
ππ εη δ
∞
=
′ ∆ = + ∑
Final result:
where 1/22
21m
mk khρπ = −
Note: The wavenumber kρm is chosen to be a positive real number or a
negative imaginary number.
0
1, 00, 0m
mm
δ=
= ≠
0/m mk k kρ ρ=
A magnetic frill of radius b is assumed on the mouth of the coax.
s ρˆˆ ˆM z E z ρ E
z
h ,r rε µ2a
xb
1 1
lnρEρ b / a
sM Eφ ρ= −
(TEM mode of coax, assuming 1 V)
10inZ
I
Frill Model
8
Choose:
Frill Model (cont.)
9
( )
( ) ( )
( )
2 20 0
2200 0 0 0
( ) ( )1 1 1 4ln / ( ) (1 ) ( )
m min r
m m m m
H k b H k aY j
k h b a k H k aρ ρ
ρ ρ
πεη δ
∞
=
− = +
∑
Final result:
where
0
1, 00, 0m
mm
δ=
= ≠
Note: The wavenumber kρm is chosen to be a positive real number or a
negative imaginary number.
1/2221m
mk khρπ = −
0/m mk k kρ ρ=
Comparison of Models
Next, we show results that compare the various models, especially as the substrate thickness increases.
z
h ,r rε µ2a
x2b
10
Comparison of Models
0 0.005 0.01 0.015 0.02 0.025h (m)
-300
-200
-100
0
100
200
300
Xin
(Ω)
Uniform current modelFrill modelCosine current modelHFFS simulation
Models are compared for changing substrate thickness.
εr = 2.2 a = 0.635 mm f = 2 GHz Z0 = 50 Ω (b = 2.19 mm)
11
z
h ,r rε µ2a
x2b
(λ0 = 15 cm)
Comparison of Models (cont.)
0 0.005 0.01 0.015 0.02 0.025h (m)
0
50
100
150
200
250
300
350
400
450
500
Rin
(Ω)
Uniform current modelFrill modelCosine current modelHFFS simulation
εr = 2.2 a = 0.635 mm f = 2 GHz Z0 = 50 Ω (b = 2.19 mm)
/ 4 0.025 [m]dλ =Note:
Models are compared for varying substrate thickness.
12
z
h ,r rε µ2a
x2b
(λ0 = 15 cm)
For the gap-source model, the results depend on ∆.
0 0.005 0.01 0.015 0.02 0.025h (m)
-300
-200
-100
0
100
200
300
Xin
(Ω)
Delta=0.01mmDelta=0.1mmDelta=1mm
Comparison of Models (cont.)
εr = 2.2 a = 0.635 mm f = 2 GHz Z0 = 50 Ω (b = 2.19 mm)
13
z
h ,r rε µ2a
x1V+-∆
(λ0 = 15 cm)
The gap-source model is compared with the frill model,
for varying ∆, for a fixed h.
0.0001 0.0002 0.0003 0.0004 0.0005 0.0006 0.0007 0.0008 0.0009 0.001
∆ (m)
-300
-200
-100
0
100
200
300
400
500
Rin
or X
in (Ω
)
Gap source model reactance Frill model reactance Gap source model resistance Frill model resistance
Comparison of Models (cont.)
εr = 2.2 a = 0.635 mm f = 2 GHz Z0 = 50 Ω (b = 2.19 mm)
0.52mm3
b a−=
h = 20 mm
R
X
14
z
h ,r rε µ2a
x1V+-∆
(λ0 = 15 cm)
These results suggest the “1/3” rule: The best ∆ is chosen as
3b a−
∆ =
This rule applies for a coax feed that has a 50 Ω impedance.
Comparison of Models (cont.)
15
z
h ,r rε µ2a
x1V+-∆
z
h ,r rε µ2a
x2b
Comparison of Models (cont.)
0 0.005 0.01 0.015 0.02 0.025h (m)
-300
-200
-100
0
100
200
Xin
(Ω)
Gap source modelFrill model
The gap-source model is compared with the frill model, using the
optimum gap height (1/3 rule).
εr = 2.2 a = 0.635 mm f = 2 GHz Z0 = 50 Ω (b = 2.19 mm)
16
z
h ,r rε µ2a
x2b
Reactance
(λ0 = 15 cm)
Comparison of Models (cont.)
The gap-source model is compared with the frill model, using the
optimum gap height (1/3 rule).
0 0.005 0.01 0.015 0.02 0.025h (m)
0
100
200
300
400
500
Rin
(Ω)
Gap source modelFrill model
εr = 2.2 a = 0.635 mm f = 2 GHz Z0 = 50 Ω (b = 2.19 mm)
17
z
h ,r rε µ2a
x2b
Resistance
(λ0 = 15 cm)
Probe in Patch A probe in a patch does not see an infinite parallel-plate waveguide.
Exact calculation of probe reactance:
L
W
x
y
(x0, y0)
cavityin p inZ jX Z
0
Imp in fX Z
f0 = frequency at which Rin is maximum
Zin may be calculated by HFSS or any other software, or it may be measured.
0cavityinX
18
Probe in Patch (cont.)
Using the cavity model, we can derive an expression for the probe reactance (derivation given later)
This formula assumes that there is no z variation of the probe current or cavity fields (thin-substrate approximation), but it does accurately account for the actual patch dimensions.
Le
We
x
y
PMC
(x0, y0)
Cavity Model
19
2 2 20 0
0 2 21 0 20 0
cos cos sinc2
1 14
p
e ep r
m,n , e em n
e e
nπwmπx nπyL W
X ω μ μ hW L mπ nπδ δ k
L W
3 2/pw e a
a = probe radius
(x0, y0) = probe location
Probe in Patch (cont.)
20
Final result:
Using image theory, we have an infinite set of “image probes.”
Probe in Patch (cont.)
Image theory can be used to improve the simple parallel-plate waveguide model when the probe gets close to the patch edge.
Image Theory
21
A simple approximate formula is obtained by using two terms: the original probe current in a parallel-plate waveguide and one image. This should be an improvement when the probe is close to an edge.
( ) ( ) ( ) ( )two 1 10 0 0 04 4 2inX = khY ka J ka khY k J ks aη η− −
0
0 /r
r
k k ε
η η ε
=
=
Probe in Patch (cont.)
Original
Image
two probe imagein in inX = X X+
22
As shown on the next plot, the best overall approximation in obtained by using the following formula:
( )probe twomax ,in in inX X X=
Probe in Patch (cont.)
“modified CAD formula”
23
Probe in Patch (cont.) Results show that the simple formula (“modified CAD formula”) works fairly well.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Xr
0
5
10
15
20
25
30
35
X f (Ω
)
Uniform current modelCavity modelUniform current model (with one image)Modified CAD formulaHFFS simulation
L
W
x
y
(x0, y0)
0 / 2/ 2r
x LxL−
=
24
( )probe twomax ,in in inX X X=
center edge
Next, we investigate each of the improved probe models in more detail:
Cosine-current model Gap-source model Frill model
25
Appendix
Cosine Current Model
Assume that ( ) cos ( )I z k z h= −
Note: (0) cos ( )I kh=
z
h ,r rε µ2a
x
I (z)
26
21 (0)2c inP Z I=
Cosine Current Model (cont.)
Circuit Model: (0)I
coax feed inZ
2
2
0c
inPZ
I
27
Represent the probe current as: 0
( ) cosmm
m zI z Ihπ∞
=
=
∑
2
2
0c
inPZ
I
This will allow us to find the fields and hence the power radiated by the probe current.
z
h ,r rε µ2a
x
I (z)
28
Cosine Current Model (cont.)
Hence
00 0
( ) cos cos cosh h
mm
m z m z m zI z dz I dzh h hπ π π∞
=
′ ′ =
∑∫ ∫
( )
( )
00
00
( )cos 12
2 ( )cos1
h
m m
h
mm
m z hI z dz Ih
m zI I z dzh h
π δ
πδ
= + ⇒ = +
∫
∫
Using Fourier-series theory:
Cosine Current Model (cont.)
The integral is zero unless m = m′.
29
Result:
( ) 0
2 cos ( ) cos1
h
mmo
m zI k z h dzh h
πδ
= − + ∫or
2 2
2 ( ) sin( )1 ( ) ( )m
mo
khI khkh mδ π
= + −
(derivation omitted)
Cosine Current Model (cont.)
30
(Time-Harmonic Fields)
Note: We have both Ez and Eρ To see this:
so
0E∇⋅ =
1 1( ) 0zE EEz
φρρ
ρ ρ ρ φ∂ ∂∂
+ + =∂ ∂ ∂
0Eρ ≠
Cosine Current Model (cont.)
31
For Ez, we represent the field as follows:
where
( )00
cosz m mm
m zE A J kh ρπ ρ
∞− −
=
=
∑aρ <
aρ > ( )(2)0
0cosz m m
m
m zE A H kh ρπ ρ
∞+ +
=
=
∑
( )
1/ 222
1/ 22 2
m
zm
mk kh
k k
ρπ = −
= −
Cosine Current Model (cont.)
32
At
so
aρ =
z zE E+ −= (BC 1)
( ) ( )(2)0 0m m m mA H k a A J k aρ ρ
+ −=
Cosine Current Model (cont.)
33
Also we have
To solve for Eρ , use
(BC 2) 2 1 szH H Jφ φ− =
1 zE EHj z
ρφ ωµ ρ
∂ ∂−= − ∂ ∂
EjH ωε=×∇
Cosine Current Model (cont.)
where
34
so 1
1
z HHj Ez
HE
j z
φρ
φρ
ωερ φ
ωε
∂∂= −
∂ ∂∂
= −∂
Hence we have 2
2
1 1 zH EHj j z
φφ ωµ ωε ρ
∂ ∂= − − −
∂ ∂
For the mth Fourier term:
( )( )
( ) 2 ( )1 1 mm m z
zmEH k H
j jφ φωµ ωε ρ ∂
= − − − − ∂
Cosine Current Model (cont.)
35
Hence
( )2 ( ) 2 ( )
mm m z
zmEk H k H jφ φ ωερ
∂− = −
∂
2 2 2zm mk k kρ− =
so that
( )( )
2
mm z
m
EjHkφρ
ωερ
∂= −
∂
where
Cosine Current Model (cont.)
36
where ( )( )
2
mm z
m
EjHkφρ
ωερ
∂= −
∂
2 1 2szIH H J
aφ φ π− = =
( ) ( ) ( )2 1 2m m m m
szIH H J
aφ φ π− = =
For the mth Fourier term:
Cosine Current Model (cont.)
37
Hence
we have
( ) ( )(2)0 02 ( )
2m
m m m m mm
Ij k A H k a A J k ak aρ ρ ρρ
ωεπ
+ − − ′ ′− =
(2)0(2)
0 00
( )( ) ( )
( ) 2m mm
m m m pmm
H k a kIA H k a A J k aJ k a a j
ρ ρρ
ρ π ωε+ +
′ ′− = −
Cosine Current Model (cont.)
( ) ( )(2)0 0m m m mA H k a A J k aρ ρ
+ −=Using (BC 1)
38
(2) (2)0 0 0 0 0( ) ( ) ( ) ( ) ( )
2mm
m m m m m m
kIA J k a H k a H k a J k a J k aa j
ρρ ρ ρ ρ ρπ ωε
+ ′ ′− = −
02 ( )
2mm
m mm
kIA j J k ak a a j
ρρ
ρπ π ωε+ − = −
( )2
014
mm m m
kA I J k aρ
ρωε+
= −
Hence
or
or
(using the Wronskian identity)
Cosine Current Model (cont.)
39
*
2*
0 0
*
0
*
0
(2) *00
0 ' 0
12
1 ( )2
( )
( ) ( )21 ( )cos cos2
scs
h
z sz
h
z sz
h
z
h
m m mm m
P E J dS
E a J a dz d
a E a J dz
a E a I z dza
m z m zA H k a I dzh h
π
ρ
φ
π
ππ
π π∞ ∞+
′= =
−= ⋅
−=
= −
= −
′ = − ⋅
∫
∫ ∫
∫
∫
∑ ∑∫
We now find the complex power radiated by the probe:
Cosine Current Model (cont.)
40
( )
( )
* (2)0 0
0
2(2) *
0 0 00
1 ( ) 12 2
11 ( ) ( )4 4
c m m m mm
mm m m m m
m
hP A I H k a
kh H k a I J k a I
ρ
ρρ ρ
δ
δωε
∞+
=
∞
=
= − +
= − + −
∑
∑
Integrating in z and using orthogonality, we have:
Hence, we have:
2 2 (2)0 0 0
0
1 (1 ) ( ) ( )16c m pm m m m
m
hP I k H k a J k aρ ρδωε
∞
=
= + +
∑
Am+ coefficient
Cosine Current Model (cont.)
41
Therefore,
2
2cos ( )
cin
PZkh
=
( )22 2 (2)0 0 0
0
1 sec ( ) 1 ( ) ( )8in m m m m m
m
hZ kh I k H k a J k aρ ρ ρδωε
∞
=
= +
∑
Define: 0
2
0
mm
r r
kk
k
mk h
ρρ
πε µ
=
= −
42
Cosine Current Model (cont.)
We then have
Also, use r
kεη
ωεεµω
ωε0000 ==
( ) 222 (2)0 0 0 0 0 0
0
1 1( ) sec (1 ) ( ) ( )8
min r m m m mmr
Z k h k h I k H k a J k aρ ρ ρη ε δε
∞
=
= +
∑
The probe reactance is: Im( )p inX Z=
43
Cosine Current Model (cont.)
Keep only the m = 0 term
( ) 222 (2)0 0 0 0 0 0
0
1 1( ) sec (1 ) ( ) ( )8
min r m m m mmr
Z k h k h I k H k a J k aρ ρ ρη ε δε
∞
=
= +
∑
0 1:k h <<
(2)0 0 0 0
1 ( ) ( ) ( )4in rZ k h J ka H kaη µ≈
(same as previous result using uniform model)
Thin substrate approximation
2 2
2 ( ) sin( )1 ( ) ( )m
mo
khI khkh mδ π
= + −
The result is
44
Cosine Current Model (cont.)
Gap Model z
h ,r rε µ2a
x1V + - ∆
( ) 1/ , 0,
0, otherwise.z
zE z a
− ∆ < < ∆=
( ) ( ) ( )20
0, cosz m m
m
m zE z B H khρπρ ρ
∞
=
=
∑
Note: It is not clear how best to choose ∆, but this will be re-visited later.
45
Gap Model (cont.) z
h ,r rε µ2a
x1V + - ∆
( ) ( ) ( )20 0
2 sinc1m
m m
mBhh H k aρ
πδ
− ∆ = +
From Fourier series analysis (details omitted):
( ) ( ) ( )20
0
1 / , 0, cos
0, otherwise.z m mm
zm zE z a B H k ahρπ∞
=
− ∆ < < ∆ = =
∑
At ρ = a:
46
Gap Model (cont.)
( )2in szY a Jπ= ∆ ( ) ( )szJ z H zφ=where
The magnetic field is found from Ez , with the help of the magnetic vector potential Az (the field is TMz):
1 zAHφ µ ρ∂
= −∂
22
2
1z zE k A
j zωµε ∂
= + ∂
( ) ( ) ( )20
0, cosz m m
m
m zA z A H khρπρ ρ
∞
=
=
∑
where
z
h ,r rε µ2a
x1V + - ∆
Use:
Setting ρ = a allows us to solve for the coefficients Am.
47
Gap Model (cont.)
(2)0
(2)0 0 0
( )1 24 sinc(1 ) ( )
min
m m m m
H k aa mY j kh k H k a h
ρ
ρ ρ
ππη δ
∞
=
′ ∆ = + ∑
Final result:
z
h ,r rε µ2a
x1V + - ∆
48
z
h ,r rε µ2a
xb
Frill Model
To find the current I (z) , use reciprocity.
10inZ
I
Introduce a ring of magnetic current K = 1 in the φ direction at z (the testing current “B”).
( )1 1
ln /sMb aφ ρ
= −
F
a b
V
b a
V
bs
S
I z H M dV A,B B,A
H M dV
H M dS
A
B SF
z
1V frill
49
Frill Model (cont.)
( )
( )
( )
( ) ( )
( ) ( )
2
0
,0
2 ,0
1 12 ,0ln /
2 ,0ln /
F
bs
S
bb
sa
bb
sa
bb
a
bb
a
I z H M dS
H M d d
H M d
H db a
H db a
π
φ φ
φ φ
φ
φ
ρ ρ ρ φ
π ρ ρ ρ
π ρ ρ ρρ
π ρ ρ
= ⋅
=
=
= −
= −
∫
∫ ∫
∫
∫
∫
A B
SF z
50
( ) ( ) ( )2 ,0ln /
bgap
a
I z H db a φπ ρ ρ= − ∫
z
b
The magnetic current ring B may be replaced by a 1V gap source of zero height (by the equivalence principle).
Let z → 0:
The field of the gap source is then calculated as was done in the gap-source model, using ∆ = 0.
Frill Model (cont.)
z
b
51
Final result:
( )
( ) ( )
( )
2 20 0
220 0 0
( ) ( )1 1 4ln / ( ) (1 ) ( )
m min
m m m m
H k b H k aY j k
h b a k H k aρ ρ
ρ ρ
πη δ
∞
=
− = +
∑
Frill Model (cont.)
z
h ,r rε µ2a
xb
52