KEMBAR78
Chapter5 Part2 | PDF | Electrical Engineering | Electromagnetism
0% found this document useful (0 votes)
65 views19 pages

Chapter5 Part2

The document describes the design of multi-section quarter-wave transformers used for impedance matching. It discusses the binomial approach where the response is flat around the design frequency. The design procedure involves choosing the number of sections N, then determining the reflection coefficients Γn at each section to synthesize the desired response Γ(θ). This provides a flat response and maximizes the bandwidth for a given number of sections. An example is given of designing a 3-section binomial transformer to match a 50 ohm load to a 100 ohm line and calculating its bandwidth.

Uploaded by

Grant Heileman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
65 views19 pages

Chapter5 Part2

The document describes the design of multi-section quarter-wave transformers used for impedance matching. It discusses the binomial approach where the response is flat around the design frequency. The design procedure involves choosing the number of sections N, then determining the reflection coefficients Γn at each section to synthesize the desired response Γ(θ). This provides a flat response and maximizes the bandwidth for a given number of sections. An example is given of designing a 3-section binomial transformer to match a 50 ohm load to a 100 ohm line and calculating its bandwidth.

Uploaded by

Grant Heileman
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Double-Stub Tuner

It is a matching network consisting of two


adjustable stubs fixed in position on the
t
transmission
i i line
li

The spacing of the stubs is not critical


An odd number of
impedances

/8

will match a wide range of

In our examples we will use a stub of separation of

/8

For proper matching the input normalized


admittance to the left of junction 2-2 should be unity.

Since stub2 adds susceptance only to the line, the


normalized admittance to the right of junction 2-2 must be
of the form 1+jb.
That means that the admittance to the right of 2-2 junction
must appear on the dashed circle A below.
below

The / 8
transformer (between the two stubs) will transform
all the admittances that lie on circle A to points on circle B which
is displaced toward the load from circle A.
Basically, the idea is:
Get stub1 to transform the input impedance of the line and load
to the right of junction 1-1 into an admittance with a locus
on circle B
1) The / 8 transformer will transform the admittance value
(just to the right
(j
g of 2-2)) on the locus circle A with a value
of 1+jb.
2) Stub2 is then used to cancel the +jb component to end up
wit ha normalized admittance of 1.

Example
The normalized admittance of the load on a line is 0.3-j2.0.
This is to be matched to the line by a double stub spaced / 8
apart with the nearest stub 0.1 from the load.
apart,
load

Solution:
1) The admittance seen to the right of stub1 is found by rotating
the normalized admittance point (0.3-j2) a distance 0 . 1
towards the generator.
That value is 0.08-j0.53 (point 1 on the chart)
2) Stub 1 adds susceptance to the line so that the admittance lies
on circle B
That corresponds to point 0 on the smith chart and has the value
of 0.08+j0.6
That means that stub 1 must add a normalized susceptance of
(0.08+j0.6) (0.08-j0.53) = j1.13

The length L1 of stub 1 is found to be 0 . 385 which can be


evaluated by rotating along the generator circle from the infinite
admittance point until an admittance of j1.13 is found
3) Next
N
bby moving
i a / 8 distance
di
from
f
point
i 1-1
1 1 to point
i 2-2
22
we reach point 2 on circle A.
That point corresponds to the value 1+j3.9 on the chart
Stub 2 must be adjusted to yield a susceptance of -j3.9 to
achieve the final match.
match
From the Smith Chart L2 is found to be equal to 00. 04
. 04 by
rotating towards the generator from the infinite admittance point
until an admittance of j3.9 is found

Another Example : Double-stub Matching


The terminating impedance Zl is 100 +j100 and the
characteristic impedance Z0 of the line and stub is 50 . The first
stub is placed at 0.40 away from the load. The spacing between
the two stubs is 3/8. Determine the length
g of the short-circuited
stubs when the match is achieved. What terminations are forbidden
for matching the line by the double-stub device?

Solution
1. Compute the normalized load impedance Zl and enter
it on the Smith Chart :
Zl =

100 + j100
= 2 + j2
50

2. Plot a SWR circle and read the normalized load

admittance 180o out of phase with Zl on the SWR


circle:

y l = 0 .25 j 0 .25

3. Draw the spacing circle of 3/8 by rotating the constantconductance unity circle (g=1) through a phase angle of
2d=23/8=3/2 toward the load. Now y11 must be on this spacing
circle, since yd2 will be on the g=1 circle (y11 and yd2 are 3/8 apart).
44. Move
M
yl for
f a distance
di t
off 0.40
0 40 from
f
0 458 to
0.458
t 0.358
0 358 along
l
the
th
SWR circle toward the generator and read yd1 on the chart:
yd1=0.55-j1.08
5. There are two possible solutions for y11. They can be found by
carrying yd1 along the constant-conductance (g=0.55) circle that
intersects the spacing circle at two points:
. y11=0.55-j1.01

y11=0.55-j1.88

6. At the junction 1-1,


y11=yd1+ys1
Then
ys1=y11-yyd1=(0.55-j0.11)-(0.55-j0.08)=+j0.97
=(0 55 j0 11) (0 55 j0 08)=+j0 97
Similarly,
ys1= -j0.80
7. The lengths of stub 1 are found as
l1= ((0.25+0.123)) =0.373
l1= (0.25-0.107) =0.143
8. The 3/8 section of line transforms y11 to yd2 and y11 to yd2 along
their constant standing-wave circles, respectively. That is:

yd2=1-j0.61
yd2=1+j2.60
9. Then stub 2 must contribute
ys2=+j0.61
ys2=-j2.60
10. The lengths of stub 2 are found as
l2= (0.25+0.087) =0.337
l2= (0.308-0.25) =0.058
11. It can be seen from the Smith Chart that a normalized yl located
inside the hatched area cannot be brought to lie on the locus of y11 or
y11 for a possible match by the parallel connection of any shortcircuited stub because the spacing circle and g=2 circle are mutually
tangent.

Thus the area of a g=2 circle is called the forbidden region of the
normalized load admittance for possible match.

The Quarter Wave


Transformer
If only a narrow band impedance match is required, a
single-section
i l
ti transformer
t
f
is
i goodd enough.
h
One drawback: it can only be used to match a real load
impedance.
Multi-section /4 Transformers can be used to achieve
optimum matching characteristics over a desired
frequency band.

Consider the following quarter lambda Transformer

l = =

Z in = Z1
With l = =

Z L + jZ1 tan l
Z1 + jZ L tan l

(1)

at the design frequency fo

Since
and Z

2
1

Z
Z

= Z

in

Z
+ Z

in

at the point where Zo and Zin meet

we can show that:

ZL Z0
Z L + Z 0 + j 2 tan l Z 0 Z L

(2)

This equation is valid for a single /4 Transformer only. The


magnitude of the reflection coefficient is:

If

and since
becomes:

{1 + [4 Z

then

fo

(sec
Z

cos
0

y2

(3)

1 )

and

Z L / (Z L Z 0 ) sec
2

equation (3)

for

(4)

The next Figure shows the approximate behavior of


the single l transformer.
4

=2(/2-m) is the bandwidth within which < m


At =m we get = m
For TEM lines:
= l =

and

fo

fo

(5)

Fractional bandwidth:

f
2( fo fm
=
fo
fo
f
4
= 2
cos
fo

)=
1

2 fm
4
= 2
fo

m
1 m2

2 Z oZ L

ZL Zo

(6)
(7)

10

Reflection coefficient magnitude as a function of f


various load mismathes

for

The Theory of Small Reflections


For more bandwidth, multi-section /4 transformers are
used.
To design a multiple /4 let us look at the reflections that occur
in a T.L. Consider the figure below:

11

The partial and T are:

1 =

Z 2 Z1
Z Z2
; 2 = 1 ; 3 = L
Z 2 + Z1
ZL + Z2

T21 = 1 + 1 =

2Z 2
2Z1
; T12 = 1 + 2 =
Z1 + Z 2
Z1 + Z 2

The total reflection seen at the feed line is:

= 1 + T12T213e 2 j + T12T21232 e 4 j + ...


= 1 + T12T213e

2 j

e
n =0

n n 2 j
2 3

(8)

using the geometric series:

n = 0

1
1 x

for

equation (8) becomes:

T 12 T 21 3 e 2
= 1 +
1 23e 2

j
j

(9)

using 2= -1, T21=1+ 1, and T12= 1-1

1 + 3 e 2 j
We get: =
1 + 1 + 3 e 2 j

(10)

12

For

j 2

(11)

are the dominant components

The same concept can be extended to the

Multisection Transformer.

0 =

Z1 Z0
Z Z
Z Z
; n = n+1 n ; N = L N ;
Z1 + Z0
Zn+1 + Zn
ZL + Z N

We can approximate the total as:

( ) 0 + 1e 2 j + 2 e 4 j + ... + N e 2 jN

( )
(12)

If the Transformers are made symmetrical then:

13

for N even
( ) = e

jN

0 cos N + 1 cos(N 2) + ... + n cos( N 2n) + ... +

N / 2
2
2

or

(13)

for N odd
0 cosN + 1 cos( N 2) +...+ n cos(N 2n) +...+
( ) = 2e jN

( N1)/ 2 cos

(14)

The idea is to use these equations to synthesize any


desired as a function of f, by choosing the
appropriate # of sections (N) and reflections ns.
There are two main approaches:
The binomial (maximally flat) or
the Chebystev (equal ripple) approach.

14

Binomial Transformer
Given a certain number of sections, the response () is
flat around the design frequency.
Basically we set:

= 2 N A cos

(15)

We can determine A by setting:


f

0 = l = 0 equation (15) becomes:

(0 )

= 2

Z
Z

L
L

Z
+ Z

(16)

0
0

at f=0, all sections are of zero length and l=0 for


all of them.

A = 2N

ZL - Z0
ZL + Z0

(17)

() can be expanded also as:

)=

n=0

N
n

N
n

N!
( N n )! n !

e 2

jn

Binomial coefficients

(18)
(19)

CnN = C NN n , C0N = 1, C1N = N = C NN1

15

The design procedure is as follows:


Choose a and match it to the:
N

( ) = A

C nN e 2

= 0 + 1 e 2

n=0

jn

+ ... + N e 2

jN

(20)

That also means:

1 = ACnN ; 2 = AC2N ; 3 = AC3N ...


n = AC nN

or

(21)

at each interface we have:


n =

Z n +1 Z n
Z
1

ln n + 1
Z n +1 + Z n
Zn
2

(22)

Matching equation (22) and (21) yields:

ln
also

Z n +1
= 2 AC
Zn

N
n

2(f0 fm
f
=
f0
f0
= 2

cos

ZL
Z0

2 N C nN ln

(23)

)=

1 m
2 A

1
N

16

Example
Design a 3 section binomial transformer to match a
50 load to a 100 line and calculate the
b d idth for
bandwidth
f
Pl t the
th results
lt for
f 1,2,3,4,
1234
m = 0.05 . Plot
and 5 sections

Solution:
For N=3 , ZL=50 Ohm, Zo=100 Ohm we get:

A = 2N

1
ZL - Z0
Z
N + 1 ln L = 0 . 0433
2
ZL + Z0
Zo

The bandwidth is then:


f
4
= 2
cos
f0

= 2

cos

1
2

1 0 . 05
2 0 . 0433

3
= 0 . 70 , or 70 %

1
N

The necessary binomial coefficients are:


C

3
0

C 13
C

3
2

3!
= 1
0 )!0 !
3!
=
= 3
(3 1 )!1!
3!
=
= 3
(3 2 )! 2 !
=

(3

17

The characteristic impedances are:

n = 0 ln Z 1 = ln Z
= ln 100 + 2 3 ( 1 ) ln

+ 2 3 C

50
100

3
0

ln

Z
Z

= 4 . 518

Z 1 = 91 . 7

n = 1 ln Z

= ln Z 1 + 2 3 C 13 ln

= ln 91 . 7 + 2 3 ( 3 ) ln
Z

= ln Z

= ln 70 . 7 + 2 3 ( 3 ) ln
3

= 4 . 26

= 70 . 7

n = 2 ln Z

50
100

Z
Z

+ 2 3 C

50
100

3
2

ln

Z
Z

= 4 . 00

= 54 . 5

18

19

You might also like