Lecture 10 - Distributed Element Matching Networks

Microwave Active Circuit Analysis and Design

Clive Poole and Izzat Darwazeh

Academic Press Inc.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide1 of 74

 Intended Learning Outcomes

I Knowledge
I Understand the advantages and disadvantages of distributed element matching, when
compared to lumped element matching networks.
I Be aware that distributed element matching networks can be designed either analytically
(using a computer) or graphically using the Smith Chart.
I Understand the principles behind stub matching networks and be aware that there are
always two stub matching solutions to a given matching problem, which differ in terms of
their bandwidth.
I Understand the theory behind the quarter wave (λ/4) transformer, its properties and
applications.
I Understand bandwidth performance of distributed element matching networks.
I Skills
I Be able to design a single stub matching network to match an arbitrary load.
I Be able to design a double stub matching network to match an arbitrary load.
I Be able to design a quarter-wave transformer matching network to match an arbitrary
load.
I Be able to calculate the bandwidth of a single stub, double stub or quarter-wave
transformer matching network.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide2 of 74

Table of Contents

Impedance transformation with line sections

Single stub matching

Double Stub Matching

Forbidden regions for double stub matching

Triple Stub Matching

Quarter Wave Transformer Matching

Bandwidth of distributed element matching networks

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide3 of 74

 Impedance transformation with line sections
The simplest possible distributed matching network is just a single length of
transmission line, of characteristic impedance Zo , connected between load and source,
as shown in figure 1.

The transmission line section has the effect of transforming any load impedance into
some other impedance, determined by the length of the line section and the line
characteristic impedance Zo .

Transmission line section

Zo ZL

Zin l

Figure 1 : Single line matching

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide4 of 74

 Impedance transformation with line sections

The input impedance of the lossless transmission line section in figure 1, having
arbitrary length, l, characteristic impedance Zo , and terminated with the load ZL , is
given by equation (??):

ZL + jZo tan(βl)
 
Zin = Rin + jXin = Zo (??)
Zo + jZL tan(βl)
This can be expressed in admittance form as :

YL + jYo tan(betal)
 
Yin = Gin + jBin = Yo (1)
Yo + jYL tan(betal)
In order to match two impedances by using a length of transmission line connected
between them we need to transform YL into the desired input admittance, Yin by varying
the quantities Yo and l. The required relationship between YL and Yin referred to above
can be found by equating the real and imaginary parts of equation (1) and solving for l
and Yo . We equate the real parts of both sides of (1) and rearrange to obtain:
 
Gin − GL
tan(βl) = Yo (2)
Gin BL + GL Bin

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide5 of 74

 Impedance transformation with line sections

Equating imaginary parts of both sides of (1), substituting (2) and solving for Yo gives:
v
2 G − B2 G
u " #
u Bin L L in
Yo = GL Gin 1 +
t (3)
GL Gin (Gin − GL )

Since Yo must be real for a practical line, the condition for single line matching to be
realisable can be taken from (3) to be:
2 G − B2 G
Bin L L in
>1 (4)
GL Gin (Gin − GL )
The realisability condition of equation (4) can be represented by circles on the Smith
chart.

When condition (4) is not satisfied these circles have the advantage of revealing
whether the addition of a parallel stub will move the load admittance into an area where
single line matching is possible.

Due to this limitation the single line matching technique is rarely used in practice.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide6 of 74

Table of Contents

Impedance transformation with line sections

Single stub matching

Double Stub Matching

Forbidden regions for double stub matching

Triple Stub Matching

Quarter Wave Transformer Matching

Bandwidth of distributed element matching networks

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide7 of 74

 Single Stub Matching
I The ’single stub’ matching technique makes use of the fact that a length of
transmission line terminated in either an open or short circuit will behave as a pure
susceptance whose value depends on the nature of the termination (open or
short) and the length of the line section. Such a section of line is called a stub.
I The single stub matching network actually consists of two parts : the stub itself, of
length l, and a length of transmission line, length d, connected between the load
and the point at which the stub is attached.
I A typical single stub matching network implemented in microstrip is illustrated
schematically in figure 2.

Zo Zo

ZL (YL )
l Zo

Figure 2 : Open circuit stub matching implemented in microstrip

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide8 of 74

 Single Stub Matching

I Stubs are widely used in Zin (Yin ) p

d1 or d2
planar microwave circuits,
such as microstrip and MMIC
implementations as they are Zo ZL (YL )

extremely easy to fabricate.

In these media the stub is
connected in parallel with the open circuit p’

main line. l1 or l2

I Because a parallel connection

(a)
is used, the design procedure
is best carried out in terms of Zin (Yin ) p
d1 or d2
admittances.
I The open circuit stub Zo ZL (YL )
configuration is shown in
figure 3(a) and the short
circuit stub configuration is short circuit p’
shown in figure 3(b). l1 or l2

I The lines all have

characteristic impedance Zo (b)
(characteristic admittance
Yo ). Figure 3 : Generic single stub matching networks : (a)
Open circuit stub, (b) Short circuit stub

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide9 of 74

 Single stub matching : Analytical approach
We start by defining the input admittance of the line at the plane pp’ in figure 3(a) or
figure 3(b), at a distance ’d’ from the load as Yin . Applying equation (1), and
normalising by dividing by Yo , gives us the normalised admittance, yin , looking into the
line at the plane pp’ as:

yL + j tan(βd) y + jt
yin = = L (5)
1 + jyL tan(βd) 1 + jyL t
where:

t = tan(βd) (6)
2π
β= (7)
λ
In general yL will be complex, i.e.:

yL = gL + jbL (8)
so we can rewrite (5) as:

gL + j(bL + t)
yin = (9)
(1 − bL t) + jgL t

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide10 of 74

 Single stub matching : Analytical approach
Realising the denominator of (9) gives us the real and imaginary parts of yin as follows:

gL (1 + t 2 )
gin = (10)
(1 − bL t)2 + gL2 t 2

bL (1 − t 2 ) + t(1 − bL2 − gL2 )

bin = (11)
(1 − bL t)2 + gL2 t 2
The essential idea behind single stub matching is that at some point on the line
normalised susceptance looking into the line, gin , will be unity. At this point, a parallel
stub can be added to exactly cancel the line susceptance, bin , resulting in a perfect
match, i.e. yin = 1 + j0. The distance between the load and this point, pp’ in figure 3(a)
or figure 3(b), where the stub is to be attached, can be calculated by setting gin equal to
unity in equation (10), which results in the following quadratic in t:

t 2 (gL2 + bL2 − gL ) − 2bL t + (1 − gL ) = 0 (12)

This can be solved for t as follows :
q
bL ± gL ((1 − gL )2 + bL2 )
t= (13)
gL2 + bL2 − gL

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide11 of 74

 Single stub matching : Analytical approach

From (13) we can determine the value of d, as a fraction of a wavelength, as follows :

1

 tan−1 (t) if t ≥ 0
 2π


d
= (14)
λ
 1 (tan−1 (t) + π) if t < 0



2π
It can easily be shown that, in the special case of purely resistive loads (i.e. bL = 0)
there is a single solution to (12), namely :
s
1
t= (15)
gL
In the more general case of complex loads, we can see from (13) and (14) that there
are two values of line length, d, that satisfy the condition gin = 1. Let us refer to these
as d1 and d2 .

For each value of d determined from equation (13) and (14), the susceptance of the
stub required to cancel the line susceptance can be found by entering the value, d1 or
d2 (i.e. the values corresponding t1 or t2 ), into equation (11). The stub susceptance is
simply the same magnitude but opposite polarity of the bin value thus calculated.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide12 of 74

 Single stub matching : Analytical approach
The length of the stub needed to realise the susceptance bstub = −bin , is given by the
familiar equations for normalised susceptance of a lossless line terminated in open
circuit and short circuit respectively (which can be easily derived by setting yL = 0 or
yL = ∞ in (9)):

bopen = tan βlopen (16)

bshort = − cot βlshort (17)

Where lopen and lshort refer to the lengths of the stubs in figure 3(a) and figure 3(b),
respectively. Note that, since there are two values of d that satisfy (13), there will be a
different value of either open circuit or short circuit stub length corresponding to each
value of d. In other words, there will be four possible matching network solutions, as
follows :
1. d1 with an open circuit stub.
2. d1 with a short circuit stub.
3. d2 with an open circuit stub.
4. d2 with a short circuit stub.
If any of the equations above result in negative electrical lengths, or an electrical length
that is impractically short, we can simply add ±(nλ)/2 to get an equivalent positive
electrical length.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide13 of 74

 Single stub matching : Graphical approach

The procedure can be summarised in 4 steps as follows :

Step 1: Plot the normalized load impedance on the Smith chart
Step 2: Draw the contant VSWR circle and determine the normalised load admittance
Step 3: Move around the constant VSWR circle toward the source until you cross the unit
conductance circle. At this point, the normalised admittance looking into the line, yin , is 1 + jb
(or 1-jb, depending on which intersection with the unit conductance circle is chosen).
Step 4: Add in a shunt stub at this point with susceptance of equal magnitude and opposite sign to
±b. This cancels the transformed load susceptance at the attachment point, resulting in a
perfect match : yin = 1 + j0
The procedure is best illustrated by an example, in the following slides.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide14 of 74

 Single stub matching : Graphical example
0.115λ

d1

As an illustrative example, we will

consider the matching of a load 0.178λ
ZL = 25 − j50Ω to a 50Ω

1
0.9
0.8
transmission line system, using

0.7
b = 1.6

1.5
0.6
the graphical method, as follows:

5
0.

2
4
0.
yL

0.3 3

4
0.2 P

Normalise the load :

5
I

tor
zL = (25 − j50)/50 = 0.5 − j and

Genera
0.1

10
plot this point on the Smith Chart 20
d2

(indicated as ’zL ’ in figure 4). 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20

I Draw the constant VSWR circle Load −0.1

−20

through this point and thereby −10

determine the normalised load −0

.2

admittance as yL = 0.4 + j0.8,

−5
Q

.3 −4
which is 180o around the VSWR −0

circle (indicated as ’yL ’ in figure 4). −

0.
4
zL
−3

Note that from this point on, all −0

.5

points on the Smith Chart are to

−2
.6
−0

.5
−0.7
b = −1.6
be interpreted as normalised

−1
−0.8

−0.9

−1
0.321λ
admittances.

Figure 4 : Single stub matching: graphical method

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide15 of 74

 Single stub matching : Graphical example

I Starting at yL , we move around

the constant VSWR circle l1open

clockwise towards the generator

1
0.9
0.8
0.7
until we cross the unit

1.5
0.6
conductance circle at point ’P’ in

5
0.

2
figure 4. The normalised
4
0.

admittance looking into the line at 0 .3 3

point ’P’ is now 1+jb. Note that if 4

we keep going clockwise around 0.2 P

5
the constant VSWR circle we will 0.1
cross the unit conductance circle 10

again at point ’Q’. The normalised 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
20
S/C
0
admittance looking into the line at
point ’Q’ is 1-jb. The lengths of
O/C
−20
−0.1
line, from the load to the point of −10

stub attachment, in each of these −0

.2

cases is d1 and d2 respectively. In Q

−5

our example, we can determine −0

.3 −4

the line lengths, in terms of guided 0.

4 −3 l1short

wavelength, by reading off the

−

5 .
’wavelengths towards generator’

−0

−2
.6
−0
scale on the outer boundary of the

.5
−0.7

−1
−0.8

−0.9
Smith chart. This gives us the

−1
following :
d1 =(0.178 − 0.115)λ = 0.063λ 0.340λ

d2 =(0.321 − 0.115)λ = 0.206λ Figure 5 : Single stub matching: determination of stub

lengths for point ’P’
© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide16 of 74
 Single stub matching : Graphical example
0.115λ

d1

0.178λ

We now need to determine the

1
0.9
I

0.8
0.7
susceptance of the shunt stub to b = 1.6

1.5
0.6
5
be attached at either point ’P’ or

0.

2
point ’Q’. This should be equal in
4
0.
yL

magnitude and opposite in sign to 0.3 3

±b, as appropriate. The addition P

4

of the stub will cancel the

0.2
5

transformed load susceptance at

tor
Genera
0.1

the attachment point, resulting in 10

d2
a perfect match at that point, i.e. 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
20

yin = 1 + j0. We determine the

input susceptance of the line at Load −0.1
−20

the point of stub attachment by −10

finding the constant susceptance −0

.2

circle which intersects the

−5
Q

.3 −4
constant conductance circle at −0

points ’P’ and ’Q’ in figure 6. The −

0.
4
zL
−3

values in our example are b=1.6 −0

.5

and b=-1.6 respectively.

−2
.6
−0

.5
−0.7
b = −1.6

−1
−0.8

−0.9

−1
0.321λ

Figure 6 : Single stub matching: graphical method

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide17 of 74

 Single stub matching : Graphical example

I Now that we know the input

susceptance of the line where the stub l1open

is to be attached (points ’P’ or ’Q’), we

1
0.9
0.8
can use the Smith Chart to determine

0.7

1.5
0.6
the length of the short or open circuit

0.5

2
stub required. This procedure is
4
0.

illustrated for point ’P’ in figure 7 and 0.3 3

for point ’Q’ in figure 8. 0.2 P

4

5
I Note that the stub susceptance is of 0.1

opposite sign to the line input 10

S/C
susceptance. The length of the stub is 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
20

found by moving clockwise around the O/C

−20

chart starting at the open circuit −0.1

−10

(y = 0) or short circuit (y = ∞) points −0

.2

(labeled ’O/C’ and ’S/C’ respectively in Q

−5

figure 7 and figure 8) until the desired .3 −4

−0

location on the periphery of the Smith

−3 l1short
4
0.
−

chart is reached.

.5
−0

−2
.6
−0

.5
−0.7
This location is found by drawing a

−1
I

−0.8

−0.9

−1
radial through the point where the
constant reactance circle intersects the
boundary of the Smith chart. We can 0.340λ

then read off the value on the Figure 7 : Single stub matching: determination of stub
’wavelengths towards generator’ scale. lengths for point ’P’

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide18 of 74

 Single stub matching : Graphical example
0.160λ
l2open

1
0.9
0.8
0.7

1.5
0.6
The respective stub lengths are

5
0.

2
l1 open and l1 short in the case of
4
0.

point ’P’ (figure 7) and l2 open and 0 .3 3

l2 short in the case of point ’Q’ 0.2 P

4

(figure 8). In our example these

5

values have been determined with

0.1

10

reference to figure 7 and figure 8 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
20
S/C

as follows : O/C
−20
−0.1
−10

−0
.2
l1open =0.340λ Q
−5

.3 −4

l1short =(0.340 − 0.25)λ = 0.09λ −0

−3
4
0.
−
l2open =0.160λ

5 .
−0

−2
.6
l2short =(0.160 + 0.25)λ = 0.410λ

−0

.5
−0.7

−1
−0.8

−0.9

−1
l2short

Figure 8 : Single stub matching: determination of stub

lengths for point ’Q’
© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide19 of 74
 Single stub matching example : solutions

Zo = 50Ω Zo = 50Ω

0.063λ 0.063λ
0.340λ (25 − j50)Ω 0.090λ (25 − j50)Ω

o/c s/c

(a) (b)

Zo = 50Ω Zo = 50Ω

0.206λ 0.206λ
0.160λ (25 − j50)Ω 0.410λ (25 − j50)Ω

o/c s/c

(c) (d)

Figure 9 : Four possible microstrip stub matching networks for the load 25 − j50Ω: (a) Point ’P’ -
open circuit stub, (b) Point ’P’ - short circuit stub, (c) Point ’Q’ - open circuit stub, (d) Point ’Q’ - short
circuit stub

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide20 of 74

Table of Contents

Impedance transformation with line sections

Single stub matching

Double Stub Matching

Forbidden regions for double stub matching

Triple Stub Matching

Quarter Wave Transformer Matching

Bandwidth of distributed element matching networks

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide21 of 74

 Double Stub Matching
I One limiting characteristic of single stub matching is that the location of the matching stub is a
function of the load impedance. This may not be an issue in most cases, but in some practical
circumstances there may be constraints on the physical positioning of the stub.
I In such cases we can use an alternative matching technique called Double stub matching,
which employs two stubs, spaced a fixed distance apart.
I The advantage of the double stub matching approach is that, although the distance between
the two stubs is generally kept constant, the first stub may be placed at any distance from the
load.
I For this reason, the double stub technique is often used to implement variable tuners, where
the match can be adjusted by varying only the stub lengths, with the stub locations on the
main line remaining fixed.
yin b a
d

Yo yL

Stub 2 b’ Stub 1 a’

Figure 10 : Double stub matching, first stub at load

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide22 of 74
 Double stub matching : Analytical approach

Let us define ybb0 as the normalised input admittance looking into the line at the plane
bb’ in the absence of Stub 2. If the susceptance of Stub 1 has been correctly chosen
(using a method we will explain shortly) then ybb0 can be written as :

ybb0 = l + jbbb0
or
ybb0 = l − jbbb0
Where ±bbb0 is the residual susceptance of the line. We now add Stub 2 to cancel this
residual line susceptance. The line to the left of Stub 2 will then be matched to the load
since,

yin = ybb0 + jbStub2

or
yin = ybb0 − jbStub2
Where bStub2 is equal and opposite in sign to bbb0 . We therefore have a perfect match
at bb’, i.e. yin = 1.

As with single stub matching, we will outline both analytical and graphical design
approaches.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide23 of 74

 Double stub matching : Analytical approach

The normalised admittance at the load in figure 10, with Stub 1 attached is :

yaa0 = gL + j(bL + bStub1 )

Where bL is the load susceptance and bStub1 is the susceptance of Stub 1. After this
impedance has been transformed through the length of line, d, and prior to the
attachment of Stub 2, the normalised admittance at point bb’ in figure 10 is :

gL + j(bL + bStub1 + t)
ybb0 = (18)
1 + jt(gL + j(bL + bStub1 ))
where, again:

t = tan(βd) (19)
and:

2π
β= (20)
λ

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide24 of 74

 Double stub matching : Analytical approach

We know that the normalised conductance at point bb’ must be unity, since this is
where we will attach Stub 2 (which can only add susceptance) to achieve a perfect
match. We therefore set the real part of equation (18) equal to 1, which gives us the
following relationship between gL , bL and t :

1 + t2 (1 − t(bL + bStub1 ))2

gL2 − gL 2
+ =0 (21)
t t2
Re-arranging (21) yields the requisite value of bStub1 as :
q
1 ± gL (1 + t 2 ) − gL2 t 2
bStub1 = −bL + (22)
t

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide25 of 74

 Double stub matching : Analytical approach

The residual line susceptance at the point bb’, with Stub 1 attached, is given by the
imaginary part of (18) :

(1 − t(bL + bStub1 ))(bL + bStub1 + t) − gL2 t

bbb0 = (23)
(1 − t(bL + bStub1 ))2 + t 2 gL2
Now, substituting the value of bStub1 given by (22) into (23) gives us the residual line
susceptance at bb’ solely in terms of gL and t :
q
∓ gL (1 + t 2 ) − gL2 t 2 − gL
bbb0 = (24)
gL t
The susceptance of Stub 2 is chosen to cancel the residual line susceptance at bb’,
therefore :

bStub2 = −bbb0

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide26 of 74

 Double stub matching : Analytical approach

We are now in a position to define bStub2 solely in terms of gL and t as follows :

 q 
∓ g (1 + t 2 ) − g2 t 2 − g
L L L
bStub2 = −  (25)
 
gL t


Where the ∓ in (25) is paired with the ± in (22).

Once we have determined the required stub susceptances from (22) and (25), the
electrical length of the stubs can be calculated for either short or open circuited stub by
applying the familiar relationships (16) and (17).

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide27 of 74

 Double Stub Matching Analytical Example
You need to match load consisting of a 4nH inductor which has a series internal
resistance of 19.2 Ω to a 50 Ω lossless co-axial transmission line by means of a double
stub matching network, consisting of two short circuit stubs, spaced 0.375λ apart. The
stub nearest to the load is 0.1λ away from it. Determine the possible combinations of
stub lengths which are required to match the load to the line at the operating frequency
of 1835 MHz. You may assume all line sections and stubs are 50 Ω.
The load impedance is first calculated :

ZL =19.2 − j(2π × 1.835 × 109 × 4 × 10−9 )

ZL =19.2 + j46.17Ω

The normalised load impedance and admittance are therefore:

(19.2 + j46.17)
ZL = = 0.384 + j0.923
50

1
yL =
(0.384 + j0.923)
0.384 − j0.923
=
0.9994
= 0.384 − j0.923

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide28 of 74

 Double Stub Matching Analytical Example
Stub 1 is attached at a distance lx = 0.1λ from the load, so we need to work out the
admittance yaa0 at this point of attachment. We can employ equations (18) and (19)
which give us the conductance and susceptance of a load, yL , translated a distance, lx ,
towards the generator. In this case t = tan(0.2π) = 0.727, therefore :

0.384(1 + 0.7272 )
gaa0 =
(1 + 0.923 × 0.727)2 + 0.3842 × 0.7272
0.587
gaa0 =
2.870
gaa0 = 0.204

and

−0.923(1 − 0.7272 ) + 0.727(1 − (−0.923)2 − 0.3842 )

baa0 =
(1 + 0.923 × 0.727)2 + 0.3842 × 0.7272
−0.435
baa0 =
2.870
baa0 = −0.154

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide29 of 74

 Double Stub Matching Analytical Example

Using this value of yaa0 = 0.204 − j0.154 as the translated load, we now calculate the
required value of Stub 1 using equation (22), this time using t = tan(0.375 × 2π) = −1
:

1± 0.204(1 + (−1)2 ) − 0.2042 × (−1)2

p
bStub1 =0.154 +
−1
bStub1 = − 0.846 ± (−0.606)

We therefore have two solutions for Stub 1, namely :

bStub1 = −1.452

and
bStub1 = −0.240

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide30 of 74

 Double Stub Matching Analytical Example

We calculate the admittance of Stub 2 by applying equation (25) :

p !
∓ 2 × 0.204 − 0.2042 − 0.204
bStub2 =−
−0.204
∓0.605 − 0.204
 
bStub2 =−
−0.204

We therefore have two solutions for Stub 2, namely :

bStub2 = 1.967
and
bStub2 = −3.967

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide31 of 74

 Double Stub Matching Analytical Example

We can now apply equations (16) and (17) to determine the length of short circuit and
open circuit Stub 1 and 2 for both solutions as summarised in table 1.

Table 1 : Double stub matching: analytical solutions

Normalised Length Length

susceptance (s/c stub) (o/c stub)
Stub 1 = -1.452 Stub 1 = 0.096λ Stub 1 = 0.346λ
Solution 1 :
Stub 2 = -3.967 Stub 2 = 0.039λ Stub 2 = 0.289λ

Stub 1 = -0.240 Stub 1 = 0.212λ Stub 1 = 0.462λ

Solution 2 :
Stub 2 = 1.967 Stub 2 = 0.425λ Stub 2 = 0.175λ

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide32 of 74

 Double stub matching : Graphical approach
I Double stub matching networks can also be designed graphically using the Smith
Chart.
I Since we are adding elements in parallel so we will be using the Smith Chart is
being used in admittance mode.
Consider the arrangement shown in figure 11. The total admittance at the point aa’ is

yaa0 = yL + jbStub1
where yL is the admittance of the load and bStub1 is the susceptance of Stub 1.
yin b a
d

Yo yL

Stub 2 b’ Stub 1 a’

Figure 11 : Double stub matching, first stub at load

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide33 of 74

 Double stub matching : Graphical approach

I Since stub 2 can only contribute

susceptance, ybb0 must be some point

1
0.9
0.8
0.7

1.5
0.6
on the unit conductance circle on the

5
0.
Smith Chart (circle 1 in figure 13). We

2
4
0.

therefore deduce that yaa0 must lie on 3

a circle of equal radius but having its
0.3
H’ C

centre rotated d wavelength towards 0.2

I’ G’ B D 4

5
the load.

or
Generat
J’
0.1 F’ A E

I This principle can be illustrated by 10

considering figure 12 where we have 0 0.1 0.2 0.3

O’
0.4 0.5 1 1.5 2 3
O
4 5 10 20
20

chosen d to be λ/4. Picking an −20

arbitrary point, A, which lies on the unit

Load
−0.1
−10

conductance circle, we see that the E’ A’ F J

−0
.2
effect of adding the line d is to rotate D’ B’ G I
−5

this point around the constant VSWR −0

.3 C’ H
−4

circle through A, 0.25λ towards the −

0.
4 −3

load to a point A’.

.5
−0

−2
.6
Another point B on the unit

−0
I

.5
−0.7

−1
−0.8

−0.9
conductance circle is similarly rotated

−1
0.25λ to point B’. The same procedure
can be carried out for all the other Figure 12 : The principle behind double stub matching
points C to J in figure 12.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide34 of 74

 Double stub matching : Graphical approach

1
0.9
If the locus is then drawn through all

0.8
I d

0.7

1.5
0.6
the primed points in figure 12, it is

5
0.
found to be a circle of the same radius

2
0.
4 C

as the unit conductance circle whose 0.3 3

centre O’ has been rotated 0.25λ B’

towards the load from O. 0.2 4

circle 2
5

or
I A similar transformation exists for any

Generat
0.1

other value of d. in each case the 10

20
locus of points on the unit conductance 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20

circle maps into a circle of the same B

−20

radius, whose centre lies d

Load
−0.1
−10
circle 1
wavelengths counter-clockwise
−0
.2

towards the load from the centre of the −5

unit conductance circle. .3 −4

−0 A C’

4 −3
I Having established this principle, the −
0.

method of determining the stub lengths

5 .
−0

−2
.6
using the Smith Chart may be

−0

.5
−0.7

−1
−0.8

−0.9
illustrated with reference to figure 13.

−1
Figure 13 : Double stub matching : graphical method

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide35 of 74

 Double stub matching : Graphical approach

I In figure 14, circle 2 represents the unit

conductance circle transformed

1
0.9
through some distance d wavelengths

0.8
d

0.7

1.5
0.6
towards the load by means of the fixed

5
0.

2
line section, d, in figure 10. 0.
4 C

I Let us pick a load admittance 0.3 3

represented by the point A in figure 13. 0.2

circle 2
B’
4

The constant conductance circle (solid 5

or
circle) through point A cuts circle 2 at

Generat
0.1

two locations, B and C. Hence, to

10

20
achieve a match, we need to transform 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20

A to point B or C by choosing a value B

−20

Load
of Stub 1 which adds the correct −0.1
circle 1
−10

susceptance to move point A in the −0

.2

right direction along the constant −5

conductance circle to either point B or .3 −4

−0 A C’

point C. −
0.
4 −3

5
The first stub, therefore, must present

.
−0
I

−2
.6
−0
a normalised susceptance equal to the

.5
−0.7

−1
−0.8

−0.9
difference between the susceptance at

−1
point A and the susceptance at points
B or C. Figure 14 : Double stub matching : graphical method

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide36 of 74

 Double stub matching : Graphical approach

I We determine the susceptance values

by looking at which constant
susceptance circles the points A,B and

1
0.9
0.8
d

0.7
C lie on and subtracting one

1.5
0.6
susceptance value from another. With

5
0.

2
the addition of Stub 1 with a value thus 0.
4 C

calculated, the input admittance at 0.3 3

point B or C becomes yaa0 . 0.2

B’
4
circle 2
5
I The fixed line section, of length d, now

or
Generat
transforms yaa0 to ybb0 . The
0.1

10

admittance at B or C when translated d 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20

20

wavelengths towards the generator will

be given by points B’ or C’, B
−20

Load
−0.1

respectively, on the unit conductance

−10
circle 1

circle. This means that the admittance −0

.2
−5

at B’ is ybb0 = 1 + jbB0 and the .3 −4

−0 A C’

admittance at C’ is ybb0 = 1 + jbC0 . 4 −3

0.
−
I The second stub therefore must

5 .
−0
present a normalised susceptance

−2
.6
−0

.5
−0.7
equal and opposite to the susceptance

−1
−0.8

−0.9

−1
value at B’ and C’, these values being
determined by looking at the Figure 15 : Double stub matching : graphical method
susceptance scale on the Smith Chart.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide37 of 74

Table of Contents

Impedance transformation with line sections

Single stub matching

Double Stub Matching

Forbidden regions for double stub matching

Triple Stub Matching

Quarter Wave Transformer Matching

Bandwidth of distributed element matching networks

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide38 of 74

 Double Stub Matching : forbidden regions

There are limitations on the load admittance that can be matched with the configuration
shown in figure 10, as we will now show. Equation (21) is a quadratic in gL which has
the following solution :
s
1 + t2
" #
4t 2 (1 − t(bL + bStub1 ))2
gL = 1 ± 1 − (26)
t2 (1 + t 2 )2
We note that the term inside the square root in equation (26) is of the form (1 - x). In
order to meet the requirement that gL should be a real number, the value of x must lie
between 0 and 1. In the case of equation (26), this leads to the following boundaries on
the value of gL :

1 + t2
0 ≤ gL ≤ (27)
t2
With reference to standard trigonometric identities, (27) can be restated as :

1
0 ≤ gL ≤ (28)
sin2 βd

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide39 of 74

 Double Stub Matching : forbidden regions

1
0.9
0.8
The limiting conditions set by (28)

0.7
I

1.5
0.6
define a forbidden region on the Smith d = λ/8

5
0.

2
Chart within which loads are 0.
4

unmatchable with a double stub tuner. 0.3 3

The forbidden region is bounded by a 4

constant conductance circle whose
0.2
5
g=1 circle translated by λ/8
value depends solely on the electrical

or
Generat
0.1

stub separation, d/λ. 10

20
0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
I Let’s say, for example, we set 0

d = λ/8. Then βd = π/4 and forbidden region −20

Load
−0.1

sin2 βd = 0.5. This means that the

−10
g=2

forbidden region will be bounded by the −0

.2
−5
g = 2 circle, as shown in figure 16. .3
g=1
−4
−0

I In other words, only constant −

0.
4 −3

conductance circles lying outside the

5.
−0
forbidden region in figure 16, can

−2
.6
−0

.5
−0.7
intersect with the translated g=1 circle.

−1
−0.8

−0.9

−1
Figure 16 : Double stub matching: forbidden region for
d=λ/8

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide40 of 74

 Double Stub Matching : forbidden regions
λ
d= 8

λ
d= 12

circle 2
circle 2

g=1 g=1
circle 1 circle 1

(a) d = λ
8
λ
(b) d = 12

circle 1
g=1
circle 1
circle 2

λ
d= 4

circle 2

g=1

3λ
d= 8

(c) d = 3λ
8
(d) d = λ
4

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide41 of 74

 Double Stub Matching

yin b a
λ
8 lx

If gL lies within the forbidden

region (i.e. it is greater than the Yo yL

value set by (28)) then we can still

use the double stub technique with
the first stub located a minimum
Stub 2 b’ Stub 1 a’
distance, lx , away from the load
towards the generator, as shown in
figure 17.
Figure 17 : Double stub matching, first stub located lx from
the load

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide42 of 74

 Double Stub Matching

λ
8

I The added line length has the effect of

1
0.9
0.8
transforming the load admittance so

0.7

1.5
0.6
circle 2
that the transformed value of gL is

5
0.

2
reduced to a value below the limit set 0.
4

by (28). This procedure is illustrated in 0.3 3

figure 18, where point A represents the 4

normalised admittance of the load and
0.2
5

points A1 , A2 and A3 represent the

or
Generat
0.1

admittance of the load transferred to a 10

point on the line towards the generator 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
20

at a distance lx = l1 , l2 , or l3
respectively from the load.
−20

Load
−0.1
A −10

I Referring to figure 18, it can be seen −0

.2 A1
l1
that the constant conductance circle
−5

A2 −4
through point A, representing the load
.3
−0 circle 1

l2
admittance, never intersects the unit 4 −3
0.
−

conductance circle transferred towards

5 .
−0

−2
A3
the load by λ/8 (circle 2). Hence, this

.6
−0

.5
−0.7

−1
−0.8
load cannot be matched using this

−0.9
l3

−1
particular double stub tuner (i.e. with
d = λ/8) when the first stub is placed
at the load.
Figure 18 : Double stub matching : location of Stub 1
with respect to the load

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide43 of 74

 Double Stub Matching

I To achieve a match, the load conductance must be reduced to such a value that the constant
conductance circle through the new load point on the Smith Chart intersects circle 2. This can
be achieved by shifting the point of application of the first stub towards the generator, in other
words by adding the length of line lx in figure 17.
I The transformed value of load conductance is gradually reduced as the distance lx is
increased. This is equivalent to moving clockwise around the constant VSWR circle, starting
at A, in figure 18 and moving through points A1 , A2 , A3 , etc. It can be seen from figure 18 that
if the point of application of the first stub (points aa’ in figure 17) is at a distance equal to or
greater than l1 , an effective match can be achieved, as the constant conductance circles
through A1 , A2 , A3 , etc. all intersect circle 2.
I We conclude, therefore, that for double stub matching to be effective, Stub 1 must be
connected at a minimum distance l1 from the termination, where the value of l1 depends on
the value of the load conductance.
I Alternatively, by keeping the Stub 1 connected right at the load, the same result can be
achieved by altering the distance between the stubs. If Stub 1 is connected at the load, then
an effective match can be achieved by altering the distance between the stubs until the circle 2
cuts the constant conductance circle through the load admittance point.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide44 of 74

 Double Stub Matching Graphical Example
Design a double stub matching network to match a load of ΓL = 0.667∠90o to the a 50 Ω lossless
line. The two stubs are spaced 0.375λ apart. The stub nearest to the load is 0.1λ away from it.
Carry out two designs: one using open circuit and one using short circuit stubs.

1
0.9
We start by drawing the unit

0.8
I

0.7
0.176λ

1.5
0.6
conductance circle (circle 1) and circle 4

5
0.
the same circle rotated through

2
4
0.
A
3
8 λ towards the load (circle 2) in 0.3 circle 3 3

figure 19. circle 1 G=1+j4

4
0.2 G

I We then locate the load reflection 5

coefficient = 0.667∠90o as the 0.1

impedance point ’A’ on the Smith

10

20
Chart, and draw the constant 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
0.25λ

VSWR (=5) circle through this −20

point. The VSWR=5 circle is −0.1

C −10

designated as circle 3 in figure 19. −0

.2
−5
I We obtain the load admittance as .3
−4
−0
E

point B by rotating the point A

D

3 0.289λ
4
around circle 3 through 180o in
−
0.
− B
circle 2

either direction. From this point 0.

−
5

2
−
F

onwards, all coordinates on the .6

l = 0.1λ −0 E=1-j2

.5
7
−0.

−1
−0.8

−0.9
Smith Chart represent normalised

−1
admittance (g + jb).

Figure 19 : Double stub matching, Example 1

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide45 of 74

 Double Stub Matching Graphical Example
We are told that the distance to Stub 1 is 0.1λ from the load. We therefore locate point
C by rotating the load admittance 0.1λ towards the generator in figure 19. We note that
point C lies on the g=0.2 constant conductance circle (circle 4), which means that we
can move back and forth along this circle by adjusting the susceptance of Stub 1.

We see that the g=0.2 constant conductance circle (circle 4) intersects the translated
unit conductance circle (circle 2) at points D and F. Since there are two points of
intersection, there must be two values of Stub 1 that can provide a match. The
normalised susceptance corresponding to point D is found by looking at which constant
susceptance circles points C and D lie on and subtracting one from the other. This
gives us :

bStub1D = (−0.40) − (−0.15) = −0.25 (29)

We locate the -0.25 constant susceptance circle on the perimeter of the Smith Chart
and read off the value of 0.461λ on the ’wavelengths toward generator’ scale. For a
short circuit stub, we trace the stub length starting at the y = ∞ point on the Smith
Chart. The electrical length of the short circuit Stub 1 for point D is therefore :

lStub1Ds = 0.461 − 0.25 = 0.211λ (30)

The open circuit Stub 1 for point D is ±0.25λ away (we obviously chose ±0.25λ to give
us a positive electrical length) :

lStub1Do = 0.211 + 0.25 = 0.461λ (31)

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide46 of 74

 Double Stub Matching Graphical Example

We apply the same approach to determine the susceptance of Stub 1 for point F :

bStub1F = (−1.6) − (−0.15) = −1.45 (32)

Locating the point representing susceptance b = −1.45 on the ’wavelengths toward
generator scale’ gives us the following stub lengths:

lStub1Fs =0.346 − 0.25 = 0.096λ

lStub1Fo =0.096 + 0.25 = 0.346λ

The length of line separating the two stubs (d = 0.375λ) has the effect of translating
point D and F into the corresponding points E and G on the unit conductance circle in
figure 19. The susceptance of Stub 2 can be determined by looking at which constant
susceptance circle these two points lie on, and remembering that the susceptance of
Stub 2 must have equal magnitude but opposite sign in order to cancel the residual line
susceptance. Once again, since there are two intersections with the unit conductance
circle, we have two possible values of Stub 2 that can provide a match. Note that point
D is paired with point E and point F is paired with point G, so there are actually only two
possible solutions consisting of the stub pairs D,E and F,G.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide47 of 74

 Double Stub Matching Graphical Example

For points E and G we can read off the residual line susceptances as :

bE = − 2.00
bG =4.00

Locating the opposite polarity susceptance points on the ’wavelengths towards

generator’ scale on the outer boundary of the Smith chart gives the following electrical
lengths for Stub 2 using short circuit stubs :

lStub2Gs =0.289 − 0.25 = 0.039λ

lStub2Es =0.176 + 0.25 = 0.426λ

The equivalent open circuit stubs are ±0.25λ in length, therefore :

lStub2Go =0.039 + 0.25 = 0.289λ

lStub2Eo =0.426 − 0.25 = 0.176λ

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide48 of 74

 Double Stub Matching Graphical Example

So, we have four possible double stub matching network solutions, which are
summarised in table 2.

Table 2 : Double stub matching: graphical solutions

Normalised Length Length

susceptance (s/c stub) (o/c stub)
Stub 1 = -0.25 Stub 1 = 0.211λ Stub 1 = 0.461λ
Solution 1 : (D,E)
Stub 2 = 2.00 Stub 2 = 0.426λ Stub 2 = 0.176λ

Stub 1 = -1.45 Stub 1 = 0.096λ Stub 1 = 0.346λ

Solution 2 : (F,G)
Stub 2 = -4.00 Stub 2 = 0.039λ Stub 2 = 0.289λ

Notwithstanding the arbitrary assignment of "Solution 1" and "Solution 2" categories,
and rounding errors, the reader can see that these results are basically the same as
those obtained by numerical calculation in example ??.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide49 of 74

 Double Stub Matching Graphical Example
For clarity, a schematic representation of the two double stub matching network
solutions implemented in microstrip implementation is shown in figure 20 :

d=0.375λ l=0.1λ d=0.375λ l=0.1λ

Zo = 50Ω Zo = 50Ω

0.426λ 0.211λ ΓL 0.036λ 0.096λ ΓL

s/c s/c s/c s/c

(a) (b)

d=0.375λ l=0.1λ d=0.375λ l=0.1λ

Zo = 50Ω Zo = 50Ω

0.176λ 0.461λ ΓL 0.289λ 0.346λ ΓL

o/c o/c o/c o/c

(c) (d)

Figure 20 : Four possible double stub matching networks for the load ΓL = 0.667∠90o
implemented in microstrip : (a) Solution 1 (points D,E) s/c stubs, (b) Solution 2 (points F,G) s/c
stubs, (c) Solution 1 (points D,E) o/c stubs, (d) Solution 2 (points F,G) o/c stubs
© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide50 of 74
Table of Contents

Impedance transformation with line sections

Single stub matching

Double Stub Matching

Forbidden regions for double stub matching

Triple Stub Matching

Quarter Wave Transformer Matching

Bandwidth of distributed element matching networks

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide51 of 74

 Triple Stub Matching
I Double stub matching is not possible for certain values of load impedance and
stub placements. One solutions is to add a third stub.
I Triple stub matching is rarely used as a design technique in fixed media, such as
microstrip, but commercial "Triple Stub Tuners" are available as standard
components in waveguide or co-axial media.
I These tuners are frequently implemented with three stubs spaced at unequal
intervals, which ensures that any value of passive load can be matched. Short
circuit stubs are mainly used in co-axial or waveguide, for ease of adjustment.
I Examples of waveguide and co-axial triple stub tuners are shown in figure 21.

(a) Waveguide triple stub tuner (b) co-axial triple

stub tuner
© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide52 of 74
Table of Contents

Impedance transformation with line sections

Single stub matching

Double Stub Matching

Forbidden regions for double stub matching

Triple Stub Matching

Quarter Wave Transformer Matching

Bandwidth of distributed element matching networks

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide53 of 74

 Quarter-Wavelength Transformer (QWT)
Consider a situation where we have connected a length of transmission line of length λ/4 and
characteristic impedance Z1 in front of the load :

Zo Z1 ZL

λ/4
Zin

Figure 22 : Quarter Wave Transformer

From equation (??), the input impedance of any lossless line section, of characteristic impedance
Z1 , is given by :
ZL + jZ1 tan (βl)
 
Zin = Z1 (33)
Z1 + jZL tan (βl)
But in figure 22 we have set l = λ/4, so (33) now becomes:

2π λ
   
π
ZL + jZ1 tan · ZL + jZ1 tan
λ 4 2
Zin = Z1 ·  = Z1 · (34)
2π λ
  
π
Z1 + jZL tan · Z1 + jZL tan
λ 4 2
Since tan(π/2) = ∞ (34) reduces to:

Z12
Zin = (35)
ZL

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide54 of 74

 Quarter-Wavelength Transformer (QWT)

Equation (35) tells us that a quarter wave line section can act as a matching network
between a load, ZL , and a required input impedance, Zin , provided that the line section
has a characteristic impedance given by :
p
Z1 = ZL Zin (36)

Such a line section is called a Quarter Wave Transformer (QWT).

Equation (43) implies that a real QWT can only be used to match Resistive loads, but
we will show that the QWT can also be used to match complex loads with a little
modification.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide55 of 74

 Using a QWT match a complex load
I Equation (43) implies that, since the characteristic impedance of the QWT must be real, this
QWT can only be used to match purely resistive loads, (ZL = RL + j0). To match complex
loads, we would need a transmission line with a complex characteristic impedance, i.e. the
line would have to be lossy. This is generally undesirable.
I One alternative is to transform the complex load into a real quantity by inserting a transmission
line section between the load and the QWT, as illustrated, in microstrip form, in figure 23.
I For convenience we will assume that the added line section has the system characteristic
impedance, Zo .

λ/4

Line section, lx

Zo Z1 Zo

ZL

Figure 23 : Quarter Wave Transformer (QWT) with a complex load

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide56 of 74

 Using a QWT match a complex load

The effect of adding the extra line section, lx , in figure 23 can be demonstrated by
considering the input impedance, Zin of any line section terminated by a load ZL . Zin is
given by equation (??), which can be expressed in normalised terms as :

ZL + jt
zin = (37)
1 + jZL t
where, again:
 
lx
t = tan 2π
λ
Since we need the input impedance of the added line section in figure 23 to be purely
real, we set =(zin ) = 0 in (37), which results in :

(xL + t)(1 − xL t) − rL2 t

=0 (38)
(1 − xL )2 + rL2 t
where ZL = rL + jxL .

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide57 of 74

 Using a QWT match a complex load

1
0.9
0.8
0.7
I Figure 24 illustrates how such a

1.5
0.6
matching network can be

5
0.

2
designed using a Smith chart.
4
0.

3
We plot the normalised load
0.3
I
impedance, zL , on the chart at 0.2 A
4

5
point ’A’. We then draw the

or
lx

Generat
constant VSWR circle through the
0.1

10

point A, and we note that this 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20
20
0
circle crosses the real axis of the C B

chart at two points, B and C. At

p p
ZT = Zo × ZC ZT = Zo × ZB −20

Load
−0.1
both these points the input −10

impedance of the line section is −0

.2
−5
purely real. .3 −4
−0

I We note that at point B Zp > Zo −

0.
4 −3

and at point C Zp < Zo , the

.5
−0
lengths of series line required to

−2
.6
−0

.5
−0.7
achieve this are given by the

−1
−0.8

−0.9

−1
distances AB and AC=AB+0.25λ
in figure 24. Figure 24 : QWT with complex load : graphical solution

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide58 of 74

 Using a QWT match a complex load

1
0.9
0.8
0.7

1.5
0.6
Once the length of series line lx

5
I

0.

2
has been chosen, the QWT is
4
0.

then used to transform ZB or ZC 0.3 3

into Zo . Choosing point B will A

4

require a transformer with a

0.2
5

characteristic impedance greater

or
lx

Generat
0.1

than Zo and choosing point C will 10

result in a transformer with ZT 0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 3 4 5 10 20

20

less than Zo . C
ZT =
p
Zo × ZC ZT =
p
Zo × ZB
B
−20

Load
−0.1
I Whichever point is chosen, ZT is −10

computed using (??). −0

.2
−5

I If the QWT is being implemented −0

.3 −4

in microstrip medium, the lower −

0.
4 −3

value of ZT is generally preferred

.5
−0
because low impedance lines (i.e.

−2
.6
−0

.5
−0.7
thicker traces) are easier to

−1
−0.8

−0.9

−1
reproduce accurately.
Figure 25 : QWT with complex load : graphical solution

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide59 of 74

 Quarter-Wavelength Transformer (QWT)

Equation (38) implies:

(xL + t)(1 − xL t) − rL2 t = 0 (39)

Rearranging (39), and noting that |ZL |2 = rL2 + xL2 , results in the following quadratic in t:

xL t 2 − t(1 − |ZL |2 ) − xL = 0 (40)

We can now determine t as follows :
q
(1 − |ZL |2 ) ± (1 − |ZL |2 )2 + 4xL2
t= (41)
2xL
The two solutions embodied in (41) represent the two intersections of the constant
VSWR circle through ZL with the real axis. One of these intersections gives
Re(zin ) < Zo and the other gives Re(zin ) > Zo . With a given value of t, we determine
the electrical length of the line section, lx , from :

lx 1
= tan−1 (t)
λ 2π

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide60 of 74

 Quarter-Wavelength Transformer (QWT)

Zo Z1 ZL

λ/4 transformer
Zin

From the previous slide we have :

Z12
Zin = (42)
ZL
We can therefore us a QWT as a matching network between a load, ZL , and a required
input impedance, Zin , provided that the QWT has a characteristic impedance given by :
p
Z1 = ZL Zin (43)

This implies that a real QWT can only be used to match Resistive loads, ZL .

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide61 of 74

Table of Contents

Impedance transformation with line sections

Single stub matching

Double Stub Matching

Forbidden regions for double stub matching

Triple Stub Matching

Quarter Wave Transformer Matching

Bandwidth of distributed element matching networks

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide62 of 74

 Bandwidth of Single Stub Matching Networks

0.8 Solution 2

0.6

The fractional bandwidth

|Γin |
for the previous single Solution 1

0.4
stub matching example is
shown for open circuit
stubs in figure 26.
0.2

0
0.7 0.8 0.9 1 1.1 1.2 1.3
f /fo

Figure 26 : Single stub matching network bandwidth: open circuit

stubs

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide63 of 74

 Bandwidth of Single Stub Matching Networks

Solution 2

0.8

0.6

|Γin |
The fractional bandwidth
for the previous single
0.4
stub matching example is
shown for short circuit
stubs in figure 27.
0.2 Solution 1

0
0.7 0.8 0.9 1 1.1 1.2 1.3
f /fo

Figure 27 : Single stub matching network bandwidth: short circuit

stubs

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide64 of 74

 Bandwidth of Single Stub Matching Networks

In general, the single stub matching solution with the shortest combination of line and
stub lengths will provide the widest bandwidth [1]. This can be demonstrated, with
reference to example ?? in section ??, by considering an arbitrary value of |Γin |, say
0.2. At this value of reflection coefficient, the various fractional bandwidths of the single
stub matching network are shown in table 3.

Table 3 : Fractional bandwidth of a typical single stub matching network

Open circuit stub Short circuit stub

Solution 1 9% 19%
Solution 2 8% 6%

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide65 of 74

 Bandwidth of Double Stub Matching Networks

0.8 Solution 2

0.6
Solution 1

|Γin |
Figure 28 shows
fractional bandwidth for a 0.4
typical open circuit double
stub matching network.
0.2

0
0.85 0.9 0.95 1 1.05 1.1 1.15
f /fo

Figure 28 : Double stub matching network bandwidth: open circuit

stubs

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide66 of 74

 Bandwidth of Double Stub Matching Networks

Solution 2

0.8

0.6

|Γin |
Solution 1
Figure 29 shows
fractional bandwidth for a 0.4
typical short circuit double
stub matching network.
0.2

0
0.85 0.9 0.95 1 1.05 1.1 1.15
f /fo

Figure 29 : Double stub matching network bandwidth : short circuit

stubs

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide67 of 74

 Bandwidth of Double Stub Matching Networks

We can compare the fractional bandwidths of the double stub solutions of figure 28 and
figure 29 by adopting a similar approach to that taken with the single stub solutions,
only this time, because the bandwidth is narrower, we will choose a higher value of
|Γin | = 0.3. The results are shown in table 4.

Table 4 : Fractional bandwidth of a typical double stub matching network

Open circuit stubs Short circuit stubs

Solution 1 3% 4%
Solution 2 1% 2%

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide68 of 74

 QWT Bandwidth
A closed form expression for the fractional bandwidth of a single section transformer
has been derived[2] as follows :
 
4 2
p
∆f Γ m Z Z
L o 
= 2 − cos−1  q (44)

fo (1 − Γ2 ) |ZL − Zo |

π
m

Where Γm is the maximum reflection coefficient that we can accept as defining a ’good’
match.

We can see from (44) that the fractional bandwidth of the QWT increases as ZL
becomes closer to Zo , as outlined above. For a value of |Γin | = 0.1, the fractional
bandwidth of the QWT shown in figure 30 with various loads is summarised in table 5.

Table 5 : Fractional bandwidth of a typical quarter wave transformer

Load Fractional Bandwidth

ZL = 150Ω or 16.67Ω 22%
ZL = 100Ω or 25Ω 36%
ZL = 75Ω or 37.5Ω 66%

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide69 of 74

 QWT Bandwidth

0.4

I The Quarter Wave

0.3
Transformer is an inherently
narrow band matching ZL = 150Ω, 16.67Ω

network, since (by definition)

it is only exactly a quarter

|Γin |
0.2
wavelength at one frequency.
ZL = 100Ω, 25Ω
I It turns out, however, that the
closer RL is to characteristic 0.1
impedance Zo , the wider the
bandwidth of the quarter
ZL = 75Ω, 37.5Ω

wavelength transformer.
0
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4
f /fo

Figure 30 : Quarter wave transformer bandwidth

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide70 of 74

 Quarter Wave Transformer design example

Design a Quarter wave transformer to match a load of 10Ω to a 50Ω transmission line.
Determine the fractional bandwidth of this matching network if we define the VSWR for
a good match to be less than 1.6.
Solution : We can match this real impedance using a QWT with a characteristic
impedance of :
p
ZT = 10 × 50 = 22.36Ω
A VSWR of 1.6 corresponds to a reflection coefficient magnitude of :

VSWR − 1 1.6 − 1
Γm = = = 0.231
VSWR + 1 1.6 + 1
The fractional bandwidth can now be computed from (44) as :

" √ #
∆f 4 0.231 2 10 × 50
=2 − cos−1
(1 − 0.2312 ) |10 − 50|
p
fo π
∆f
=0.169
fo
The fractional bandwidth of this transformer is therefore 16.9%.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide71 of 74

 Multi-Stage Quarter Wave Transformers

I There is a natural bandwidth limitation of quarter-wave transformers, because you

only get exactly a quarter wavelength at one frequency
I Lower frequencies see less than a quarter-wave, higher frequencies see more.
I In order to achieve matching over a broader bandwidth, we add multiple
quarter-wave transformers in a series, so that the impedance mismatch that each
transformer is correcting becomes less and less.
I The bandwidth limitation of the single quarter-wave section can be overcome by
using multiple sections in series, with impedances chosen to provide the desired
response (e.g. maximally flat, Tchebyscheff etc.)
I The improvement in bandwidth carries a cost of increased physical size, which
may be a problem in some circumstances.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide72 of 74

 Example Multi-Stage Transformers

’2-quarter-wave’ transformer :
 1
λ/4 Z2 4
λ/4 ZA = Z1 (45)
Z1
Z1 ZA ZB Z2
 3
Z2 4
n=2
ZA = Z1 (46)
n=1
Z1
’3-quarter-wave’ transformer :

λ/4  1
λ/4 Z2 4
λ/4 ZA = Z1 (47)
Z1
Zo Z1 Z2 Z3 Z4
 3
Z2 4
n=3
ZA = Z1 (48)
n=2
Z1
n=1

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide73 of 74

 References

C W Davidson.
Transmission lines for communications.
Macmillan Press Ltd., London, 1978.

D M Pozar.
Microwave Engineering.
John Wiley and Sons Inc., New York, USA, 2 edition, 1998.

© Poole-Darwazeh 2015 Lecture 10 - Distributed Element Matching Networks Slide74 of 74