Vous êtes sur la page 1sur 15

4/20/2007

8_5 Filter Implementation

1/1

and

is to “realize” each individual lumped element with

approximations in our lumped element

8.5

all filters with

make

Filter

Implementations

Q:

So, we

now know

how to make any and

lumped elements—but this is a microwave

You said

that lumped elements where

difficult to

implement at microwave frequencies. You

said that

distributed elements were used

components.

elements!?!

to make microwave

A:

with distributed

There are many,

many

ways to make microwave

The first of these

realizations is:

engineering course!

filters

the

most straightforward

these

solutions.

or

microstrip

Kuroda’s Identities.

then insert

Perhaps

and

So how do we make a filter with distributed

 elements. transmission line sections,

HO: RICHARDS TRANSFORMATIONS

To easily

stripline circuit, we must apply one of

implement Richard’s Transforms

HO: KURODA

S IDENTITIES

in a

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Richards Tranformation

1/7

Richard’s

Recall the input impedances of

circuited

short-circuited and open-

.

transmission

line

stubs

s
Z
=
jZ
tan
β
in
0

o
Z
=−
j Z cot β
in
0

Note

lumped

that the input im

elements!

pedances

are

purely reactive—just

However,

have a much different

the reactance of lumped

like

inductors and capacitors

form

to that

of

mathematical

transmission

line

stubs:

Transformations

Z

0

Z

0

,β

,β

Z
=
j ωL
L
j
Z
=
C
ωC

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Richards Tranformation

2/7

we can say in

In other
words, the impedance of
lumped
elements
(capacitors
and
inductors)
are
functions with
respect
to frequency
general
that,
for
example:
s
o
Z
≠ Z
Z
≠ Z
in
L
in
C
However,
for a
given
lumped
element (
(with
a given
Z
and
length )
the
functions
will
0
precisely
one
frequency!
For
example,
satisfies
this equation
for
a given
L,Z ,
and
:
0
j ω L
= tan β
j
Z
c
0
c
ω c
= tan
j
Z
0
v p
or similarly
satisfies
this equation:

. Therefore,

there is one frequency—let’s call it

j

ω

c

j Z cot β

transmission line stubs

different

—that

L or C) and a given stub

be equal at

and

=−

 c ω c v p

ω C

 0 ⎡ Z cot ⎢ 0 ⎢ ⎣ v p π ω c
c
=−
j
To make things easier,
line stub
to
λ
8 , where:
c
2
λ
=
=
c
β c

let’s set the length of our transmission

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Richards Tranformation

3/7

Q:

Why

=λ

c

8

?

A:
Well, for one reason,
4 and
therefore
β c
tan
(
π
4
)
=
1.0
!
This of
course
greatly
simplifies
our earlier
results:
j
π ⎞
π
j ω L
= j
Z tan
= Z
j
cot
c
0
0
⎝ 4 ⎠
ω C
⎝ 4 ⎠
c
= j
Z
= Z
j
0
0
Therefore,
if we
wish to
build
a short-circuited stub
with the
same
impedance
as an
inductor L at
frequency ω , we
set
the
c
characteristic
impedance
of the stub
transmission
line
to be
Z
L :
0
c
s
Z
=
j ω L
=
Z
Z
=
ω L
L
c
in
0
c
λ
c
=
8
Likewise,
if we
wish to
build
an open-circuited stub
with the
same
impedance as
an capacitor
C at frequency
ω ,
we
set
the
c
characteristic
impedance
of the stub
transmission
line
to be
Z
= 1
ω C
:
0
c

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Richards Tranformation

4/7

=

λ

c

8

1
o
Z
=−
j
ω
C
=
Z
Z
=
C
c
in
0
ω C
c

We

call

these two results Richard’s

Transformations

.

However,

Transformations

lumped

inductors!

it is important

do not

to remember

that Richard’s

result in perfect

elements—the stubs

equal—at

transformation

only one

frequency (

ω ).

c

We

and

can use Richard’s transformations

capacitors

element filter

design.

lowpass filter

of a lumped design, the

frequency

ω

c

replacements

frequency.

for

is perfect—the impedances are

In fact, for

is the filter’s cutoff

do not behave like capacitors and

to replace the inductors

Using these stubs to

result in a filter

replace inductors and

that of

response similar to

capacitors will

the

lumped

element

design—a low pass filter

with cutoff frequency

ω

c

.

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Richards Tranformation

5/7

However,
the behavior
of the filter
in
the
stopband will be
very
For
at the

element

1 ) is:

λ 2

different from the lumped elemen

where the stub

response

will

t design.

of

ω=

(high) frequencies

, the filter

transmission!

be that

0 —near

Figure

8.37

(p.

411)

distributed-element low-pas filter

of Example 8.5.

Q:

So

why does

the filter response match the

the passband?

example,

lumped

A:

first

note that

Series

To see why, we

the Taylor

approximation for tan φ and

cot

φ when

φ is small

(i.e.,

perfect

φ

length becomes a multiple of

Amplitude responses of lumped-element and

1

tan φ

φ

and

cot φ

1 for

φ
φ

and

response

so well

in

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Richards Tranformation

6/7

The impedance
of our Richard’s transformation
some
arbitrary frequency
ω is:
λ
s
Z
(
ω )
= j Z tan
β
c
in
0
8 ⎠
⎛ ⎞
ω
λ
= tan
j
(
ω
L
)
c
c
v
8
⎜ ⎝ ⎠
p
⎛ ⎞
ω
= tan
j
(
ω
L
)
π
c
ω
4
⎝ ⎠
c
Therefore,
when
ω ω
(i.e., frequencies
in
c
low-pass filter!),
we can approximate this
ω π
s
Z
(
ω )
=
j (
ω L tan
)
in
c
ω
4 ⎠
c
ω π
≈ j ω L
c
ω
4
c
π
= j
ω
L
⎝ 4
⎞ ⎟ when ω

impedance as:

shorted stub

the passband of

at

a

ω

c

pass-

that the

Compare

this

to a lumped inductor impedance:

= j ωL
Z L
very close to
the lumped
less than
ω
(i.e., all
frequencies of
the
c
band)!
 Since the value π 4 is relatively close to one, we find Richard’s Transformation shorted stub has an input

low-pass filter

element inductor for all frequencies

impedance

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Richards Tranformation

7/7

j

ω

c

C

ω

c

pass-

capacitor

results are approximately the same for all

approximately:

Similarly, we
find that
the
Richard’s
circuit stub has an
input
impedance of
− ω
j
π
o
Z
(
ω
)
=
cot
⎛ ⎞
in
ω ω
C
4
⎝ ⎠
c
c
⎜ ⎝ ωπ
1
⎛ 4 ⎞
=
j ω C
π ⎠
Again, when compared
to the lumped element
impedance:
1
=
Z C
j ωC
we find
that
band
frequencies (i.e., when
ω
ω
).
c

when ω

transformation open-

Jim Stiles

4

ω

c

The Univ. of Kansas

Dept. of EECS

4/20/2007

Kurodas Identities

1/7

with its

change!

Kuroda’s

 We find that Kuroda’s Identities can be very the implementation of Richard’s

practicable.

Kuroda’s Identities essentially

port networks.

the

matrices.

By

equivalent, we mean that

same

In other

equivalent in

(e.g.,

a circuit, and

the

its scattering matrix) of

behavior and

the

 Q: Why would we want to do this? A: Because one of the equivalent may

implement!

circuit will not

Identities

transformations more

words, we can replace one two-port network

characteristics

be more practical to

useful in making

provide a list of equivalent two

they have precisely

For

example,

we can

use

Kuroda’s Identities

to:

 1) Physically separate transmission 2) Transform series stubs into shunt 3) Change impractical characteristic

more realizable

ones.

stubs.

line stubs.

impedances into

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Kurodas Identities

2/7

are

network, constructed with a

and

first

a very ambiguous

the

Four

Kuroda’s identities are

provided in

book.

confusing table (Table

two identities to

8.7) in your

be the most useful.

Port

1

We will find

Z
01
Z
02

Note

that the length of

the

stub and

the

 identical, but the characteristic impedance of each are different.
 Consider the following two-port length of transmission line, and an open-circuit shunt stub:

Port 2

transmission line

The first Kuroda identity

states that

the two-port

network

above is precisely the same two-port network

as this

one:

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Kurodas Identities

3/7

network:

Thus, we

can

Z
2
n
02
= 1
+
Z
Z
01
01
2
n
Z
02
2
n

Port 1

Port 2

replace

the first

Note

this equivalent

circuit uses a

short-circuited series

The second of

Kuroda’s Identities

states that this two

structure in some circuit with

port

stub.

 the one above, and the behavior that circuit will not change in the least!

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Kurodas Identities

4/7

Z 01

Z
02

Port 1

Port

2

this two-port network:

Port 1

2

n Z

01

 n 2 = 1 +

Z 02

Z 01

Port

2

n

2

Z 02

Is precisely identical to

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Kurodas Identities

5/7

With

useful when we replace the series

regard

to Richard’s Transformation

, these identities

with shorted

are

inductors

stubs.

To see

with distributed

example,

why

this is

useful when implementing a lowpass filter

consider

this

third

order filter

elements,

realized using Richard’s Transformations:

L
L
1
3
C
2
ω L
ω
L
c
1
c
3
1
ω C
c
2
λ
c
8
λ
c
8

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Kurodas Identities

6/7

Note

design!

that we have

a few

problems

in terms

of implementing this

of all

the

stubs are ideally

we build that?

We

could

infinitely

close

to each other—

physically separate them, but

First

how do

this would introduce

some
that would
mess
up our filter response!
stubs
are
difficult to
construct in
like
shunt stubs much better!
problems, we
a short length
of
line
( Z
and
8
) to the beginning
and
end
of
0
c
filter:
λ
ω L
ω L
λ
c
c
c
1
c
3
8
8
Z
Z
0
0
λ
λ
c
c
8
1
ω C
8
c
2
λ
c
8

transmission line

length between

them

Secondly, series

microstrip/stripline—we

To solve these

transmission

the

Jim Stiles

The Univ. of Kansas

Dept. of EECS

4/20/2007

Kurodas Identities

7/7

Note
lengths
only results in a phase
shift
in
the
filter response—the
transmission and
reflection functions will

remain

the

unchanged.

Now we can use the second of

with shunts

:

series stubs

Kuroda’s

Identities to

λ
λ
c
c
8
8
2
n ω L
2
n
ω
L
1
c
1
3
c
3
2
1
ω C
2
n Z
n
Z
c
2
1
0
3
0
λ
λ
λ
c
c
c
8
8
8

Z

Z

 2 = 1 + 0 n 3 2 = 1 + 0 1

n

ω

c

L

1

ω L

c

3

 where: Now this is separated,

a

realizable filter!

the

they are all shunt stubs.

Note

and

three stubs are

replace

Note

that a specific

numerical

example (example 8.5)

of this

procedure is

given

on pp.

409-411 of

your

book.

Jim Stiles

The Univ. of Kansas

Dept. of EECS