3A. Impedance Transformation and Impedance Matching: References

3A.
Impedance
Transformation and
Impedance Matching
The information in this work has been obtained from sources believed to be reliable.
The author does not guarantee the accuracy or completeness of any information
presented herein, and shall not be responsible for any errors, omissions or damages
as a result of the use of this information.
August 2007  2006 by Fabian Kung Wai Lee 1
References
• [1] R. Ludwig, P. Bretchko, “RF circuit design - Theory and

applications”, 2000 Prentice-Hall.
• [2] D.M. Pozar, “Microwave engineering”, 2nd edition, 1998 John-
Wiley & Sons (3rd edition, 2005 by John-Wiley & Sons is also
available).
• [3] R.E. Collin, “Foundation for microwave engineering”, 2nd edition,
1992, McGraw-Hill.
1
1.0 Lumped Impedance
Transformation Network
Impedance Transformation
• An impedance transformation network is a two-port network that when
connected in series with an impedance ZL at one port, will result in Zs
being seen on another port.
• ZL is usually not equal to Zs (otherwise there will be no need for
transformation). Zs is known as the image impedance of ZL.
• We immediately notice that the transformation network is a 2-port
network.
Transformation
Zs
Network
ZL
2
Why Impedance Transformation is
Needed?
• Maximum power is delivered when load is matched to the Tline
(assuming generator is matched).
• Impedance matching on sensitive receiver components (antenna, low-
noise amplifier etc.) improves the signal-to-noise ratio of the system.
• Impedance matching in a power distribution network (such as antenna
array feed network) will reduce amplitude and phase errors.
The Theory of Maximum Power

Transfer
IL Time averaged power dissipated across
Zs
load ZL: 1 *
{
PL = 2 Re V L I L }
where Vs Z L Vs
VL = Zs +ZL
IL = Zs +ZL
Vs ZL VL
 V Z Vs    1  Vs Z L 
* 2
Z s = Rs + jX s PL = 1 Re s L ⋅    = Re 
2  Zs +Z L  Zs +Z L   2
Z L = RL + jX L    Z s + Z L 2 
2
Vs R L
⇒ PL = 1
2 (R + R )2 + ( X + X )2
s L s L
Letting ∂PL ∂P
= ∂XL = 0
∂RL L PL = PL (RL , X L )
We find that the value for RL and XL
that would maximize PL is To maximize power transfer to the load
RL = Rs, XL = -Xs. impedance, ZL must be the complex
In other words: ZL = Zs* conjugate of Zs, a notion known as
Conjugate Matched.
3
Impedance Transformation and
Matching
• An example of a RF system where every modules are internally
matched to 50Ω
RS
Zo
~
Det
50Ω 50Ω 50Ω 50Ω 50Ω 50Ω 50Ω 50Ω 50Ω 50Ω
~
Standardizing to a fixed impedance simplifies
system design, as each module can be replaced
with module of similar functions and we can be assured
the power flow along the path is not affected.
Types of Transformation Network
• Single lumped element (either L or C)
• Dual lumped elements (L impedance matching network)
• Triple lumped elements (Pi or T impedance matching network)
• More lumped elements (ladder type)
• Distributed elements (consists of section of Tlines)
• Hybrid - Consists of both Tline and lumped elements
4
Impedance Transformation Using
Lumped Elements
• Lumped components such as surface mounted device (SMD) inductor
and capacitor can be easily purchased nowadays.
• SMD capacitors have a range from 0.47pF to greater than 10000pF.
With tolerance less than ± 5% and operating temperature between -
55oC to 125oC.
• SMD inductors have a range from 1.0nH to greater than 4000nH.
With tolerance from ± 2% to± 10%, operating temperature from -40oC
to 125oC and Q factor from a minimum of 15 to greater than 45.
• The inductors come in a variety of form, from coil-type, thin-film, to
spiral inductors mounted in SMD package. Self-resonance frequency
ranges from 200MHz to greater than 5GHz for L<100nH (thin-film).
Ultra High Frequencies Passive

Components (>250MHz)
Multilayer electrolytic Multilayer ceramic capacitor 80mils
capacitor (1812) (0603), NPO dielectric
Aluminium oxide 0805 50mils
(0805) Thick film

resistor
Other sizes from
(0603), (1210) JEDEC Solid State
(1210), NPO dielectric Technology Association
Y5V dielectric - 0402, even smaller
than 0603.
- 1210, bigger than
0805.
-Nowadays there are
Thin film inductor Coil inductor also 0201 and 01005
(0603) (0805)
5
Medium Frequency Passive
Components (up to 250MHz)
Coil inductor Multilayer

with Ferrite core ceramic capacitor
Coil inductor
with air core
Carbon/metal
film resistor
Passive Lumped Components for

Incorporation into PCB and other Substrates
Add the corresponding schematics for these
Microstrip Line Air Bridge
Interdigital Capacitor
Series Single-Loop Spiral Inductor Series Multi-Loop Spiral Inductor
High εr dielectric Via

Deposited carbon film,
or semiconductor
Metal-Insulator-Metal (MIM) Low resistance High resistance

Capacitor
Shunt Multi-Loop Spiral Inductor Resistors
6
Single Lumped Element
Transformation Network (1)
jX
Zs = ZL + jX ZL = RL+ jXL
ZL
2 port network
Single Lumped Element

Transformation Network (2)
Ys = YL + jB
jB ZL = RL+ jXL
ZL
2 port networks
7
Dual Lumped Elements Transformation
Network (1)
jX
1 1
Ys = = jB +
Zs RL + j ( X L + X )
RL + jXL
jB
If Zs = Rs + jXs is given,
we could solve for X and
B by equating the real and
imaginary parts of the above expression: This configuration
is only applicable
R for Rs > RL
X = − X L ± RL (Rs − RL ) + L X s2
Rs
Rs − RL (1.1)
B=
Rs X L + RL X s + Rs X
Dual Elements Transformation

Network (2)
jX
1
Z s = jX +
1
jB +
RL + jX L jB RL + jXL
If Zs = Rs + jXs is given,
again we could solve for
X and B by equating the This configuration
real and imaginary parts is only applicable
of the above expression: for RL> Rs
R
X = X s ± Rs (RL − Rs ) + s X L2
RL
Rs − RL (1.2)
B=
RL X s + Rs X L − RL X
8
Example 1.1
• Transform ZL=100 + j80 to 50 + j40 at 410MHz.
R RL>Rs
X = X s + Rs (RL − Rs ) + s X L2 = 115.498
RL
Rs − RL
B= = 0.014
RL X s + Rs X L − RL X
Since X is +ve, an inductor can be used to realize it:
X
L=
(
2π 410 × 106 ) = 44.83nH
Since B is +ve, a capacitor can be used to realize it:
B
C=
(
2π 410 × 106 ) = 5.468 pF
Synthesizing Reactance (X) and

Susceptance (B) Using Lumped Elements
• For X:
X > 0 : Use inductor to synthesize it. X < 0 : Use capacitor to synthesize it.
jωo L = jX 1 = j  − 1  = jX
jω o C  ω C
L
⇒L= X = X
C  o 
ωo 2πf o
⇒C = 1 = 1
ωo X 2πf o X
• For B
B > 0: Use capacitor to synthesize it. B < 0: Use inductor to synthesize it.
L= 1 = 1
C= B = B L ωo B 2πf o B
C ωo 2πf o
9
Example 1.1 Cont...
• The final schematic:
44.83nH
100+j80
5.47pF
50+j40
At 410MHz Only!
Exercise
• Transform ZL= 50 + j100 to 300 - j10 at 900MHz using 2 lumped

element matching networks.
10
Example 1.2
• Repeat Example 1.1 using Smith chart.
Exercise 2
• Repeat Example 2 using Smith chart.
11
2.0 Nodal Q-Factor and
Approximate Bandwidth of
Network
Bandwidth of the Matching Network (1)
• Suppose in Example 1.1 the load ZL is actually given by an inductor in

series with a resistor, so that at 410MHz we obtain ZL = 100+j80.
44.83nH
I
31.05nH
Zs= 50+j40 V
5.47pF 100Ω
Vs at 410MHz
ZL
• We input the above schematic in a circuit simulator (PSPICE) and run

a frequency sweep (change the frequency of the source Vs while
measure I & V) from 100MHz to 800MHz.
12
Re(Zs)
We see that the matching
network only transform ZL
to 50+j40 at 410MHz, at other
frequencies, Zs gradually
deviates from the desired
Im(Zs) value.
410MHz
• Within a range of frequencies near to the operating frequency fo=

410MHz, Zs= Rs + jXs is quite near the desired value. We will call this
range of frequency the bandwidth (BW) of the transformation network.
• To examine this closer, we plot Zs in terms of its magnitude and

phase.
Following the theory
of series RLC
271.76MHz network, we define
|Zs| the 3dB BW as the
range of freq. Where
|Zs| is less than 2 Z o
2 Z o ≅ 90.4
, where Zo is the
magnitude of the
f o = 410MHz
impedance at the
Z o = 63.9Ω
operating freq.
Arg(Zs)
fo = 410MHz.
We see that the
‘measured’ BW is:
|Zs| and Arg(Zs) is very close to BW = 271.76MHz
the pattern of series RLC circuit
near operating
August 2007 frequency fo  2006 by Fabian Kung Wai Lee 26
13
• Now consider the circuit of Example 1.1 again. We could compute a

quantity known as the Nodal Q factor, Qn as follows:
Last element
44.83nH X=115.498, B=0.014
Zs= 50+j40 31.05nH jX

at 410MHz ZL=
5.47pF 100Ω jB 100+j80
ZL
Compute the impedance or admittance

without the last element of the network
X=115.498
B'
0.0091 X ' 75.50 jX Y’ = G’ + jB’
Qn = = = = ≅ 1.5 0.0061+j0.0091
G ' 0.0061 R ' 50.06 or
Z’ = 50.06-j75.50
• We could calculate the BW of the system using the equation in (1.4):

f o 410 MHz 271.76MHz
BW ≅ = ≅ 273MHz
Qn 1.5
• Surprisingly this is quite near the measured value using simulation. Both
measured and calculated BW using this method will match even closer if
Zs is real, or Xs=0. This applies to all lumped element transformation
network as well (3 elements or more).
• When Xs is not 0, there is an error, the larger |Xs|, the greater the error.
However this does illustrate that we could in general compare the BW of
various transformation network merely by calculating Qn.
• Generally high Qn denotes narrow BW, low Qn denotes wide BW.
14
Nodal Q Factor, Qn (1)
• Qn for a few favorite transformation networks.
jX1 jX1
jB1 ZL ZL
jB2 jB1
Z’ = R’ +jX’ B' X' Z’ = R’ +jX’

Y’ = G’ +jB’ Qn = =
G' R' Y’ = G’ +jB’
jX2 jX1 TL1
ZL jB1 ZL
jB1
Z’ = R’ +jX’
Z’ = R’ +jX’
Y’ = G’ +jB’
Y’ = G’ +jB’
Nodal Q Factor, Qn (2)
• The previous slides only illustrate the concept of using nodal Q factor
to estimate and compare bandwidth between transformation networks
heuristically. A more formal argument and derivation can be found
from various materials:
– R. Ludwig, P. Bretchko, “RF circuit design - Theory and
applications”, 2000, Prentice-Hall.
– J.R. Smith,”Modern communication circuits”, 2nd edition 1998,
McGraw-Hill.
– EEN3096 (Communication Electronics) notes, year 2006 of MMU.
– Unpublished works of F. Kung, 2003.
15
Example 2.1
• Transform the load ZL = 200 - j40 to 50 + j20 at 2.4GHz. Find the
nodal Q factor and estimate the bandwidth of the circuit. Use Smith
chart to aid the design.
jX
Zs=50+j20 ZL=200-
jB
j40
X=109.2 B=0.00756
109.2
L= = 7.24nH
Zs 2π (2.4 × 109 )
0.00756
C= = 0.50 pF
2π (2.4 × 109 )
89.23
Qn = = 1.764
ZL 50.58
X B
2.4GHz
BW = = 1.36GHz
1.764
Z’
Constant Qn Circles (1)
• Given a certain Qn, we can find the locus or curve of the

corresponding impedance point on the Smith Chart. It turns out this
locus is a circle, known as Constant Qn Circle.
• Qn depends on the point location on the Smith chart. We could joint
all points on the Smith chart giving a similar Qn to form a curve or
locus. It happens that this locus is a circle, known as Constant Qn
circles.
• The center and radius for the circles can be derived as follows.
• From the theory on Smith chart:
1 + U + jV 1−U 2 −V 2 2V
r + jx = = +j 1
1 − U − jV (1 − U )2 + V 2 (1 − U )2 + V 2 Γcenter = 0 m j
Qn (2.1)
r x
2
 1  1 1
x 2V U 2 + V ±  = 1 + Radius = 1 +
Qn = =  Q n Qn2 Qn 2
r 1−U 2 −V 2
16
Constant Qn Circles (2)
Qn=3.0
Qn Radius Center
Qn=1.0 0.5 2.2360 0±j2.000
1.0 1.4142 0±j1.000
2.0 1.1180 0±j0.500
Qn=0.5 3.0 1.0541 0±j0.333
5.0 1.0198 0±j0.200
Qn=0.5
Qn=1.0
Qn=3.0
Limitation of 2 Lumped Elements

Network
• By now it is obvious of the limitation of the 2 elements network. For
instance in Example 3 there are only two ways to transform ZL= 200 -
j40 to Zs= 50 + j20.
• Therefore we cannot control the nodal Q factor of 2 elements network,
it is determined by the values of ZL and Zs.
• Using an extra element, we have extra degree of freedom and we can
control the value of Qn in addition to performing impedance
transformation/matching. This is the advantage of using the T or Pi
networks.
17
Three or More Lumped Elements
Transformation Network
• For more than 3 lumped elements, analytical method such as shown
in previous slides is very cumbersome to apply.
• It is easier to perform 3 elements transformation network design with
the aid of Smith Chart.
• As oppose to 2 elements network, 3 or more elements network does
not suffer from blind spot. It can transform any passive load ZL to any
required impedance value.
Example 2.2
• Repeat Example 2.1 using 3 elements impedance transformation
network, either T or Pi, with the aid of Smith chart. It is required that Qn
be equal to 3. (ZL= 200-j40, Zs= 50+j20 at fo = 2.4 GHz).
Qn=3
jX2 jX1
Zs jB1
ZL X1
Z’ X1 = 294.0
B1 = 0.0083
B1
X2 = 174.6 X2
L1= 19.5nH
Z’
C1 = 0.55pF
L2 = 11.58nH Qn=3
18
Example 2.3
• Repeat Example 2.2 using 3 elements transformation network, either T
or Pi, with the aid of Smith chart. It is required that Qn be equal to 5.
(ZL=200-j40, Zs=50+j20 at 2.4 GHz).
Qn=5
jX2 jX1
Zs jB1
X1
ZL
Z’ X1 = 488.6
B1
B1 = 0.00584
X2 = 261.0 X2
Z’
L1= 32.4nH
Qn=5
C1 = 0.387pF
L2 = 17.3nH
Zs Versus f from PSPICE Simulation
Rs
Both circuits from Example
4 and 5 are fed into PSPICE.
Xs Qn=3 AC simulation is run from
1.8GHz to 2.8GHz and the
Resonance results are compared. It is
Frequency seen that the T network with
higher nodal Q factor has
Rs narrower BW, characterized by
more rapid deviation from
fo = 2.4GHz.
Xs Qn=5
19
Exercise
• Repeat Example 2.3 using 3 elements T transformation network, with

the aid of Smith chart. It is required that Qn be equal to 1. (ZL=200-
j40, Zs=50+j20). Can you synthesize the T network? Suggest a
solution to this.
Exercise
• Repeat Example 2.3 using 3 elements Pi transformation network, with

the aid of Smith chart. It is required that Qn be equal to 3. (ZL=200-
j40, Zs=50+j20). Can this impedance transformation be realized ?
Discuss the result.
20
Pros & Cons of Lumped Element
Network
• Lumped element network is compact, small in size.
• Suitable for use up to frequency of 2.5GHz.
• Not every values of inductance and capacitance are available.
• Stability, value changes with temperature.
• Tolerance of components.
• Difficult to tune.
• Higher cost.
Appendix 1 – Series and

Parallel RLC network, Nodal
Q-Factor Concepts
21
Q Factor
• The Q Factor of a series or parallel impedance is defined by:
Taken from the

Q =ω
[Maximum energy stored] notes of EEN3096,
power dissipation Communication
Electronics,
I(ω) F. Kung 2000.
I(ω)
R
V(ω) jX V(ω) G jB
X B
Qs = (A.1a) Qp = (A.1b)
R G
Q Factor and Series/Parallel RLC

Network
Rs Ls
Cp Lp
Cs Z Rp
Z
Parameter Series RLC network Parallel RLC network

Input impedance 1  1 
−1
Rs + jωLs + 1
 
jωC s  R + jωL + jωCP 
 P P 
Resonance frequency 1 1
ωo = ωo =
LsCs L pC p
(A.2)
Quality factor, Q at resonance ω o Ls 1 RP
Qs = = Qp = = ω o RPC p
frequency ωo Rs ω o Rs Cs ωo Lp
Bandwidth BW (note that this ω o ωo
is just an approximation) Qs Qp
22
Frequency Response of Series/Parallel
RLC Network
L C R
L1 C1 R1 L R
C
L1 R1
|Z(ω)|
C1
|Z(ω)|
1 R
2
3dB Bandwidth
2R
R 3dB Bandwidth
ω1 ωo ω2 ω
ω1 ωo ω2 ω
Arg(Z(ω))
+90o Arg(Z(ω))
+90o
0
0
-90o ω
-90o ω
Parallel RLC Series RLC

R Q R Q
R Q R Q
Poles and Zeros of Series and Parallel

RLC Network
L C R
L1 C1 R1
Z (ω ) = R + jωL + jω1C jω
For series RLC:
1+ jωRC −ω LC 2 2 complex conjugate zeros Resonance
= on s-plane
Resonance frequency is jωC Frequency
Z (s ) = 1+ sRCsC+ s
2
the frequency where LC
input impedance to a x σ
passive RLC network Pole
becomes real. 1 pole on s-plane
L C R
L1 C1 R1
Zero
jω
For parallel RLC: (

Z (ω ) = R1 + jω1 L + jωC
−1
) x
Resonance
Frequency
jωRL
= σ
1 zero
R −ω 2 RLC + jωL
Z (s ) = sRL
R + s 2 RLC + sL
x
2 complex conjugate poles

23
Resonance Frequency of Higher Order
Systems (1)
• For a system with more than one L and C, there will be higher order
poles and zeros. These will distort the location of the fundamental
resonance frequency of the network and introduce higher order
resonance frequencies.
jω
x fresonance2 (higher
Higher order order)
poles and zeros fresonance3 (higher
order)
x
fresonance1
σ
Z x
August 2007  2006 by Fabian Kung Wai Lee x 47

Systems (2)
• Since each resonance frequency is still due to the dominant poles and
zeros, the concept of Q factor with regards to 3dB bandwidth can still be
applied to higher order network.
Dominant poles and
zero contributing to jω
(parallel) fresonance2
x fresonance2 (higher
order)
Dominant poles and
fresonance3 (higher
zero contributing to
order)
(series) fresonance3
x
fresonance1
Dominant poles and σ
zero contributing to
(parallel) fresonance1 x
x
24
Systems (3)
Ceq1 Req1
Leq1
|Z|
Leq3 Req3
Ceq3
Leq2 Ceq2 Req2
f
0
fresonance1 fresonance3 fresonance2
Impedance Transformation Network as

a Resonating Network
Impedance
Transform
RL
Network
ZI (Lossless) jXL
Augmented Network
If ZI = Rs , then the augmented
network is actually under
resonance during normal Impedance
operation. The concept of Q Transform jXL
Network
factor can be applied. If ZI is ZI (Lossless) RL
complex, the concept of Q factor
can still be applied if the Xs is
small.
25
3.0 Distributed and Hybrid
Network
Distributed Transformation Network
• Single Stub transformation network.

l
TL1
−1 Zc , β
  Zs ZL
 1  jB1
Z s =  jB1 +  (3.1)
 Z L + jZ c tan (βl ) 
Zc
 Z c + jZ L tan (βl )  Z’ = R’ +jX’
 Y’ = G’ +jB’
• jB can be implemented using a Tline with open/short circuit at one end.

Can also use lumped elements such as SMD capacitors. In this case
the network is known as hybrid network.
• No blind spot.
26
Example 3.1
• Transform the load ZL = 200 - j40 to 50 + j20 at 2.4GHz. Find the nodal
Q factor and estimate the bandwidth of the circuit. Use Smith chart to
aid the design. Synthesize the circuit.
TL1
jB1 βl = θ ZL SWR
Zs
circle
Z’ = R’ +jX’ ZL
Y’ = G’ +jB’ B1
θ
θ = βl = 1.013 Z’
B = −0.0356
Example 3.1 Cont...
• Use a microstrip line to implement the circuit, Zc = 50Ohm. Dielectric

constant = 4.6, and d=1.57mm.
• Step 1 - Synthesize Tline.
• From microstrip Tline design example of Chapter 1, we see that the
required W must be roughly 2.88mm.
ε eff = 3.55
β = ω ε oε eff µ
( )
= 2π 2.4 × 109 3.55ε o µo = 94.77
θ 1.013
l= = = 0.011 m = 1.1cm
β 94.77
27
Example 3.1 Cont...
• Step 2 - Synthesize jB1.

• We can use an inductor for B1:
1
L= = 1.863nH
( )
2π 2.4 ×109 ⋅ 0.0356
• Or we can use another short circuit Tline to generate B1:

1 1
Z in (l ) = jZ c tan ( βl ) = = j 
− jB B
 1 
 = 1 tan −1 
1 1
l= tan −1  = 0.0054 m
β Z B 94.77  0. 0356 × 50 
 c 
Example 3.1 Cont...
• Thus the final circuit...

2.88mm
Other circuit ZL
here
5.4mm 11mm
Or...
2.88mm
Other circuit
here
11mm
28
Double-Stub Distributed Network (1)
• The single-stub network suffers from the disadvantage of requiring a

variable length of Tline between the load and the stub. This may not
be a problem for fixed transformation network, but would pose some
difficulty if an adjustable tuning network is desired.
• To overcome this disadvantage a double-stub transformation network
is used. ZL’
Optional
TL2 TL1
Zs
jB2 jB1 ZL
−1
 
λ 3λ  
l = to  
8 8  
 1 
Z s =  jB2 +
 Z ' // 1  + jZ tan (β l ) 
  L jB  c 
Can be adjusted  Zc  1  
 Z c + j  Z L // 1  tan (βl ) 
 
  jB1  
Double-Stub Distributed Network (2)
TL2 TL1
Zs ZL
jB2 jB1 ZL
B2
Zs TL1
Suppose we want to
transform ZL to
Zs=50Ω
B1
The stubs TL2
Maury Microwave Corp.
29
Quarter-Wave Transformer
• A quarter-wave transformer is a simple and useful circuit for matching a
real load impedance to a transmission line. An additional feature is that it
can be extended to multi-section design for broader bandwidth.
• Consider a terminated lossless Tline again, using the input impedance
relation of Terminated Tline: λ
l=
Tline1 4
λ 2π λ π
l= βl = ⋅ = Zc Zin ZL
4 λ 4 2 Z1
Z in (l ) = Z1
(2 ) = Z12
Z L + jZ1 tan π l z=0
Z1 + jZ L tan (π ) Z L
2
Z2 To match Tline1 to ZL:
Z in (l ) = 1 (3.2a)
ZL Z1 = Z L Z c (3.2b)
Example 3.2
• Design a quarter-wave transformer to transform a 200Ω load into 50Ω

at 2.4 GHz using a microstrip line constructed on a dielectric with
dielectric constant of 4.6 and thickness of 1.57 mm.
Z1 = Z c RL = 50 × 200 = 100
Using the microstrip design equations of “Chapter 1 - Advance Transmission Line”:
s = w/d
30
Example 3.2 Cont...
In this analysis the reactance associated

50Ω 2.88mm with the discontinuities must be taken into
Microstrip 0.66mm
account. Proper compensation technique
Line
must be used.
200
18.0mm
Top view
Limitation of Quarter Wave Transformer
• Although quarter-wave transformer can in theory used to match complex

impedance, it is more common to use it to match real impedance.
• However bear in mind that a complex load impedance can always be
transformed to a real impedance by adding the correct series or shunt
reactive component.
• At the operating frequency fo , the electrical length of the matching
section is λ o/4. But at other frequencies the length is different, so a
perfect match is no longer obtained. The quarter wave transformer has
a limited bandwidth, like other transformation methods.
• Writing Zin as:
This parameter is
t = tan θ , θ = βl frequency dependent
Z + jZ1t
Z in = Z1 L (3.3)
Z1 + jZ Lt
31
BW of Quarter Wave Transformer (1)
• Using (3.2a) and (3.3): Z − Zc Z L − Zc

Γ = in =
Z L − Zc Z in + Z c Z L + Z c + j 2t Z c Z L
Γ=
[(Z 2 2
L + Z c ) + 4t Z o Z L ] 1
2
1
=
{((Z L + Zc ) /(Z L − Zc )) + [4t 2Zc Z L /(Z L − Zc )2 ]}
2
1
2
1
=
{1 + [4Zc Z L /(Z L − Zc )2 ]sec2 θ } 1
2
(3.4a)
• For frequency near fo , l ≅ λo /4, sec2θ >>1, and this simplifies to:
Z L − Zc
Γ =ρ≅ cos θ (3.4b)
2 Zc Z L
• Assuming the maximum mismatch we can tolerate is |Γ|max = ρm= 0.1,

or return loss of -20log10(0.1) = -20dB.
• Plotting |Γ| = ρ versus θ = βl, we can determine the bandwidth of the
quarter wave transformer.
0.4
ρ θ
i
0.2
ρm
0
1 θm 1.5 θm ' 2 2.5
π θ
i
2
32
• Putting ρm into (3.4a) and solve for θm :

ρm 2 Zc Z L
• Assuming TEM or quasi-TEM mode: cos θ m = ⋅
1 − ρm2 Z L − Zc
Quarter wavelength
β
• And the bandwidth is given by:
2πf m v p πf m
θ m = βl = =
v p 4 fo 2 fo
2θ m f o
⇒ fm =
π
BW = 2 f o − f m
 
4 ρm 2 Z L Z c  (3.5)
= 2 fo − cos −1  ⋅
π  2 Z L − Zc 
 1 − ρ m 
Example 3.3
• Design a single-section quarter wave transformer to match a 10Ω load

to a 50Ω Tline, at fo=2.4GHz. Determine the bandwidth for which
VSWR<1.3. Use the microstrip line design example of Chapter 1 to
physically realize it.
Z1 = 50.10 = 22.361
VSWR − 1 From Example 5.1 of Chapter 1
ρm = = 0.13
VSWR + 1
λ 2.4GHz = 2π / β = 6.6cm
β 2.4GHz = 94.77 λ 2.4GHz
= 1.7cm
4
 2 − 4 ⋅ acos  ρm ⋅ ( 2⋅ 50⋅ 10)   ⋅ fo = 4.511 × 108
BW = 451.2 MHz
 π  1 − ρm2 10 − 50  
  
33
Pros & Cons of Distributed Network
• Easy to fabricate and incorporate into microwave circuit. Utilize the

PCB itself.
• Cheap and stable, good tolerance if mechanical tolerance is properly
controlled.
• Easier to tune than lumped element network.
• Modern manufacturing facilities use LASER to trim the transmission
line dimension during tuning.
• At low frequency, the length of the Tline can be unwieldy large.
Increasing Bandwidth of Distributed

Matching Network
• For applications requiring more bandwidth than a single quarter wave
section can provide, multi-section transformers can be used.
Zn must increase or
decrease
Γ
Γ1 Γ2 3 monotonically
ZL must be real.
Z0 Z1 Z The theory of multi-section transformer

2
Z3 ZL is beyond the time frame of this course.
Interested students please refer to
Section 5.10-12 of reference [3].
• We can synthesize any desired reflection coefficient response as a

function of frequency, by properly choosing Γn and using enough
sections.
34
Multisection Quarter-Wave
Transformer
|ρ(f)| |ρ(f)|
2-section transformer transformer ratio = 4
(Binomial) (Binomial)
Transformer
n=6 4 2 k=1 2 3
ratio, # section,
0.2 0.2
f / fo f / fo
0.6 0.8 1 1.2 1.4 0.6 0.8 1 1.2 1.4
1/4 λ
1/4 λ
|ρ(f)|
transformer ratio = 4
Zo Z1 Z2 Z3=ZL ZL (Chebyshev)
2-section impedance
transformer k=2
1+ ρ
resistive VSWR = = 15
.
thin film 1− ρ
Zo
0.2 Return loss = −20 log ρ
= 13.98dB
Matched Termination
f / fo
August 2007  2006 by Fabian Kung0.6
Wai Lee 0.8 1 1.2 1.4 69
Binomial and Chebyshev Transformers
Binomial Transformer
• impedance of consecutive 1/4 wave lines are proportional to binomial
coefficients.
• gives maximally flat passband characteristic.
Chebyshev Transformer
• wider bandwidth than Binomial Transformer for the same number of 1/4
wave sections.
• ripple over passband.
Tapered Transition
• characteristic impedance varies continuously in a smooth fashion.
• taper length of 0.5 - 1.5 wavelength required.
35
THE END
36

3A. Impedance Transformation and Impedance Matching: References

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

3A. Impedance Transformation and Impedance Matching: References

Uploaded by

Copyright:

Available Formats

3A.

August 2007  2006 by Fabian Kung Wai Lee 1

• [1] R. Ludwig, P. Bretchko, “RF circuit design - Theory and

August 2007  2006 by Fabian Kung Wai Lee 2

August 2007  2006 by Fabian Kung Wai Lee 3

August 2007  2006 by Fabian Kung Wai Lee 4

August 2007  2006 by Fabian Kung Wai Lee 5

The Theory of Maximum Power

Types of Transformation Network

• Single lumped element (either L or C)

• Dual lumped elements (L impedance matching network)

• Triple lumped elements (Pi or T impedance matching network)

• More lumped elements (ladder type)

• Distributed elements (consists of section of Tlines)

• Hybrid - Consists of both Tline and lumped elements

August 2007  2006 by Fabian Kung Wai Lee 8

August 2007  2006 by Fabian Kung Wai Lee 9

Ultra High Frequencies Passive

(0805) Thick film

August 2007  2006 by Fabian Kung Wai Lee 10

Coil inductor Multilayer

August 2007  2006 by Fabian Kung Wai Lee 11

Passive Lumped Components for

High εr dielectric Via

Metal-Insulator-Metal (MIM) Low resistance High resistance

August 2007  2006 by Fabian Kung Wai Lee 12

August 2007  2006 by Fabian Kung Wai Lee 13

Single Lumped Element

August 2007  2006 by Fabian Kung Wai Lee 14

Dual Elements Transformation

• Transform ZL=100 + j80 to 50 + j40 at 410MHz.

Synthesizing Reactance (X) and

August 2007  2006 by Fabian Kung Wai Lee 18

• The final schematic:

August 2007  2006 by Fabian Kung Wai Lee 19

• Transform ZL= 50 + j100 to 300 - j10 at 900MHz using 2 lumped

August 2007  2006 by Fabian Kung Wai Lee 20

• Repeat Example 1.1 using Smith chart.

August 2007  2006 by Fabian Kung Wai Lee 21

• Repeat Example 2 using Smith chart.

August 2007  2006 by Fabian Kung Wai Lee 22

August 2007  2006 by Fabian Kung Wai Lee 23

Bandwidth of the Matching Network (1)

• Suppose in Example 1.1 the load ZL is actually given by an inductor in

• We input the above schematic in a circuit simulator (PSPICE) and run

August 2007  2006 by Fabian Kung Wai Lee 24

• Within a range of frequencies near to the operating frequency fo=

Bandwidth of the Matching Network (3)

• To examine this closer, we plot Zs in terms of its magnitude and

• Now consider the circuit of Example 1.1 again. We could compute a

Zs= 50+j40 31.05nH jX

Compute the impedance or admittance

Bandwidth of the Matching Network (5)

• We could calculate the BW of the system using the equation in (1.4):

August 2007  2006 by Fabian Kung Wai Lee 28

• Qn for a few favorite transformation networks.

Z’ = R’ +jX’ B' X' Z’ = R’ +jX’

jX2 jX1 TL1

Nodal Q Factor, Qn (2)

August 2007  2006 by Fabian Kung Wai Lee 30

Constant Qn Circles (1)

• Given a certain Qn, we can find the locus or curve of the

August 2007  2006 by Fabian Kung Wai Lee 32