2-digital filters (iir).ppt

28
  1 AG C DS P AG C DS P Professor A G Constantinides IIR Digital Filter Design IIR Digital Filter Design Standard approach (1) Convert the digital flter specifcations into an analogue prototype lowpass flter specifcations (2) Determine the analogue lowpass flter transer unction (!) " rans orm #y replacin g the comple$ varia#le to the digital transer unction ) (  s  H a ) (  z G ) (  s  H a

Upload: boopathi-sambandha-moorthi

Post on 04-Oct-2015

21 views

Category:

Documents


0 download

TRANSCRIPT

  • 5/18/2018 2-Digital filters (IIR).ppt

    1/28

    1

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    IIR Digital Filter DesignIIR Digital Filter Design

    Standard approach

    (1) Convert the digital flter specifcations

    into an analogue prototype lowpassflter specifcations

    (2) Determine the analogue lowpassflter transer unction

    (!) "ransorm #y replacing thecomple$ varia#le to the digital transerunction

    )(sHa

    )(zG

    )(sHa

  • 5/18/2018 2-Digital filters (IIR).ppt

    2/28

    2

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    IIR Digital Filter DesignIIR Digital Filter Design

    "his approach has #een widely usedor the ollowing reasons%

    (1) Analogue appro$imation

    techni&ues are highly advanced(2) "hey usually yield closed'ormsolutions

    (!) $tensive ta#les are availa#le or

    analogue flter design() *ery oten applications re&uiredigital simulation o analogue systems

  • 5/18/2018 2-Digital filters (IIR).ppt

    3/28

    3

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    IIR Digital Filter DesignIIR Digital Filter Design

    +et an analogue transer unction#e

    where the su#script ,a- indicates

    the analogue domain A digital transer unction derived

    rom this is denoted as

    )(

    )()(

    sD

    sPsH

    a

    aa =

    )(

    )()(

    zD

    zPzG =

  • 5/18/2018 2-Digital filters (IIR).ppt

    4/28

    4

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    IIR Digital Filter DesignIIR Digital Filter Design .asic idea #ehind the conversion o

    into is to apply a mapping rom thes'domain to thez'domain so that essentialproperties o the analogue re&uency

    response are preserved "hus mapping unction should #e such that

    /maginary ( ) a$is in the s'plane #emapped onto the unit circle o thez'plane

    A sta#le analogue transer unction #emapped into a sta#le digital transerunction

    )(sHa)(zG

    j

  • 5/18/2018 2-Digital filters (IIR).ppt

    5/28

    5

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    IIR Digital Filter: The bilinearIIR Digital Filter: The bilinear

    transformationtransformation

    "o o#tain G(z)replace s#y f(z) inH(s)

    Start with re&uirements on G(z)G(z) Availa#le H(s)

    Sta#le Sta#le

    0eal and 0ational inz 0eal and0ational in s

    rder n rder n

    +P (lowpass) cuto3 +P cuto3 Tcc

  • 5/18/2018 2-Digital filters (IIR).ppt

    6/28

    6

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    IIR Digital FilterIIR Digital Filter

    4ence is real and rational inzo order one

    ie

    5or +P to +P transormation were&uire

    "hus

    )(zf

    dcz

    bazzf

    ++=)(

    10 == zs 00)1( =+= baf

    1 == zjs 0)1( == dcjf

    1

    1.)(

    +

    =

    z

    z

    c

    azf

  • 5/18/2018 2-Digital filters (IIR).ppt

    7/28 7

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    IIR Digital FilterIIR Digital Filter

    "he &uantity is f$ed rom

    ie on

    r

    and

    ca

    ccT

    2

    tan.)(1: T

    j

    c

    azfzC

    c

    ==

    2tan. T

    jc

    aj cc

    =

    1

    1

    1

    1.

    2

    tan

    +

    =

    z

    z

    Ts

    c

    c

  • 5/18/2018 2-Digital filters (IIR).ppt

    8/28 8

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Bilinear TransformationBilinear Transformation "ransormation is una3ected #y scaling

    Consider inverse transormation withscale actor e&ual to unity

    5or

    and so

    ssz

    +=11

    oo js +=22

    222

    )1(

    )1(

    )1(

    )1(

    oo

    oo

    oo

    oo zj

    jz

    +++=

    ++=

    10 == zo10 > zo

  • 5/18/2018 2-Digital filters (IIR).ppt

    9/28 9

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Bilinear TransformationBilinear Transformation 6apping o s'plane into thez'plane

  • 5/18/2018 2-Digital filters (IIR).ppt

    10/28 10

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Bilinear TransformationBilinear Transformation

    5or with unity scalarwehave

    or

    )2/tan(1

    1

    je

    ejj

    j

    =+=

    j

    ez=

    )2/tan(=

  • 5/18/2018 2-Digital filters (IIR).ppt

    11/28 11

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Bilinear TransformationBilinear Transformation

    6apping is highly nonlinear Complete negative imaginary a$is in

    the s'plane rom to is mapped

    into the lower hal o the unit circle inthez'plane rom to

    Complete positive imaginary a$is inthe s'plane rom to is

    mapped into the upper hal o the unitcircle in thez'plane rom to

    = 0=

    0= =

    1=z 1=z

    1=z 1=z

  • 5/18/2018 2-Digital filters (IIR).ppt

    12/28 12

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Bilinear TransformationBilinear Transformation 7onlinear mapping introduces a

    distortion in the re&uency a$iscalled frequency warping

    3ect o warping shown #elow

  • 5/18/2018 2-Digital filters (IIR).ppt

    13/28 13

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Spectral TransformationsSpectral Transformations"o transorm a given lowpass

    transer unction to another transer

    unction that may #e a lowpass8highpass8 #andpass or #andstop flter(solutions given #y Constantinides)

    has #een used to denote the unit

    delay in the prototype lowpass flterand to denote the unit delay in

    the transormed flter to avoidconusion

    )(zGL

    )(zGD

    1

    z

    1

    z )(zGL

    )(zGD

  • 5/18/2018 2-Digital filters (IIR).ppt

    14/28 14

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Spectral TransformationsSpectral Transformations 9nit circles inz' and 'planes

    defned #y

    8

    "ransormation romz'domain to

    'domain given #y

    "hen

    z

    z

    jez= jez=

    )(zFz=

    )(!)( zFGzG LD =

  • 5/18/2018 2-Digital filters (IIR).ppt

    15/28 15

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Spectral TransformationsSpectral Transformations 5rom 8 thus 8

    hence

    "hereore must #e a sta#leallpass unction

    )(zFz= )(zFz=

    >

    1if"1

    1if"1

    1if"1

    )(

    z

    z

    z

    zF

    )(/1 zF

    1"

    1

    )

    (

    1

    1

    #

  • 5/18/2018 2-Digital filters (IIR).ppt

    16/28

    16

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-LowpassLowpass-to-Lowpass

    Spectral TransformationSpectral Transformation "o transorm a lowpass flter with a

    cuto3 re&uency to another lowpassflter with a cuto3re&uency 8 the transormation is

    n the unit circle we have

    which yields

    )(zGL

    )(zGDc

    c

    == z zzFz

    1

    )(11

    1 j

    jj

    eee

    =

    )2/tan(

    1

    1)2/tan(

    +=

  • 5/18/2018 2-Digital filters (IIR).ppt

    17/28

    17

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-LowpassLowpass-to-Lowpass

    Spectral TransformationSpectral Transformation Solving we get

    $ample' Consider the lowpass digital

    flter

    which has a pass#and rom dctowith a :;d. ripple

    0edesign the a#ove flter to move the

    pass#and edge to

    ( )( )2/)(sin

    2/)(sin

    cc

    cc

    +=

    )3917.06763.01)(2593.01(

    )1(0662.0)(

    211

    31

    ++=

    zzz

    zzGL

    25.0

    35.0

  • 5/18/2018 2-Digital filters (IIR).ppt

    18/28

    18

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-LowpassLowpass-to-Lowpass

    Spectral TransformationSpectral Transformation 4ere

    4ence8 the desired lowpass transer

    unction is

    1934.0)3.0sin(

    )05.0sin( ==

    1

    11

    1934.01

    1934.0)()(

    ++==

    z

    zzLD

    zGzG

    0 0.2 0.4 0.6 0.8 1-40

    -30

    -20

    -10

    0

    /

    Gain,

    dB G

    L(z) G

    D(z)

  • 5/18/2018 2-Digital filters (IIR).ppt

    19/28

    19

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-LowpassLowpass-to-Lowpass

    Spectral TransformationSpectral Transformation"he lowpass'to'lowpass

    transormation

    can also #e used as highpass'to'

    highpass8 #andpass'to'#andpassand#andstop'to'#andstoptransormations

    ==

    z

    z

    zFz

    1

    )(

    11

  • 5/18/2018 2-Digital filters (IIR).ppt

    20/28

    20

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-HighpassLowpass-to-Highpass

    Spectral TransformationSpectral Transformation Desired transormation

    "he transormation parameter is given #y

    where is the cuto3 re&uency o thelowpass flter and is the cuto3 re&uencyo the desired highpass flter

    1

    11

    1

    +

    +=z

    zz

    ( )

    ( )2/)($os

    2/)($os

    cc

    cc

    +=

    c

    c

  • 5/18/2018 2-Digital filters (IIR).ppt

    21/28

    21

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-HighpassLowpass-to-Highpass

    Spectral TransformationSpectral Transformation $ample'"ransorm the lowpass

    flter

    with a pass#and edge at to ahighpass flter with a pass#and edge

    at 4ere

    "he desired transormation is

    )3917.06763.01)(2593.01(

    )1(0662.0)(

    211

    31

    ++=

    zzz

    zzGL

    25.055.0

    3468.0)15.0$os(/)4.0$os( ==

    1

    11

    3468.01

    3468.0

    =

    z

    zz

  • 5/18/2018 2-Digital filters (IIR).ppt

    22/28

    22

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-HighpassLowpass-to-Highpass

    Spectral TransformationSpectral Transformation

    "he desired highpass flter is

    1

    11

    3468.01

    3468.0)()(

    ==

    z

    z

    zD

    zGzG

    0 0.2 0.4 0.6 0.8

    80

    6040

    20

    0

    Normalized frequency

    Gain,

    dB

  • 5/18/2018 2-Digital filters (IIR).ppt

    23/28

    23

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-HighpassLowpass-to-Highpass

    Spectral TransformationSpectral Transformation"he lowpass'to'highpass

    transormation can also #e used to

    transorm a highpass flter with acuto3 at to a lowpass flter with acuto3 at

    and transorm a #andpass flter with acenter re&uency at to a #andstopflter with a center re&uency at

    cc

    oo

  • 5/18/2018 2-Digital filters (IIR).ppt

    24/28

    24

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-BandpassLowpass-to-Bandpass

    Spectral TransformationSpectral Transformation Desired transormation

    11

    2

    1

    1

    1

    11

    2

    12

    12

    1

    ++

    +

    +

    ++=

    zz

    zzz

  • 5/18/2018 2-Digital filters (IIR).ppt

    25/28

    25

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-BandpassLowpass-to-Bandpass

    Spectral TransformationSpectral Transformation "he parameters and are given #y

    where is the cuto3 re&uency o thelowpass flter8 and and are thedesired upper and lower cuto3 re&uencieso the #andpass flter

    ( ) )2/tan(2/)($ot 12 ccc =

    ( )

    ( )2/)($os

    2/)($os

    12

    12

    cc

    cc

    +

    =

    c1c 2c

  • 5/18/2018 2-Digital filters (IIR).ppt

    26/28

    26

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-BandpassLowpass-to-Bandpass

    Spectral TransformationSpectral Transformation Special Case'"he transormation

    can #e simplifed i"hen the transormation reduces to

    where withdenoting the desired centerre&uency o the #andpass flter

    12 ccc =

    o $os= o

    1

    111

    1

    =z

    zzz

  • 5/18/2018 2-Digital filters (IIR).ppt

    27/28

    27

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-BandstopLowpass-to-Bandstop

    Spectral TransformationSpectral Transformation Desired transormation

    1

    1

    2

    1

    1

    11

    12

    12

    12

    1

    +

    +

    +

    +++=

    zz

    zz

    z

  • 5/18/2018 2-Digital filters (IIR).ppt

    28/28

    28

    AGC

    DSP

    AGC

    DSP

    Professor A G Constantinides

    Lowpass-to-BandstopLowpass-to-Bandstop

    Spectral TransformationSpectral Transformation"he parameters and are given

    #y

    where is the cuto3 re&uency othe lowpass flter8 and and arethe desired upper and lower cuto3re&uencies o the #andstop flter

    c

    1c 2c

    ( )

    ( )2/)($os

    2/)($os

    12

    12

    cc

    cc

    +=

    ( ) )2/tan(2/)(tan 12 ccc =