666 Design of Digits! Fitors Chap. g These formulas are extremely useful in obtaining a good estimate of the filter length required to achieve the given specifications Af, 6:, and é. The estimate is used to carry out the design and if the resulting 4 exceeds the specified 55, the Jength can be increased until we obtain a sidelobe level that meets the specifica. tions. 8.3 DESIGN OF IIR FILTERS FROM ANALOG FILTERS Just as in the design of FIR filters, there are several methods that can be used to design digital filters having an infinite-duration unit sample response. The tech- niques described in this section are all based on converting an analog filter into a digital filter. Analog filter design is a mature and well developed field, so Rot surprising that we begin the design of a digital filter in the analog domain and then convert the design into the digital domain. An analog filter can be described by its system function, a Last Bis) =O AG) (8.3.1) Yau! & where {a;} and {fj} are the filter coefficients, or by its impulse response, which is elated to H¥,(s) by the Laplace transform Hels) = f her 832) Alternatively, the analog filter having the rational system function H(s) given in (8.3.1), can be described by the linear constant-coefficient differential equation ftv _ SO, dtr) de 7 = ym an 83.3) where x(r) denotes the input signal and y(r) denotes the output of the filter. Each of these three equivalent characterizations of an analog filter leads to alternative methods for converting the filter into the digital domain, as will be described in Sections 8.3.1 through 8.3.4. We recall that an analog linear time- invariant system with system function H(s) is stable if all its poles lie in the left half of the s-plane. Consequently, if the conversion technique is to be effective, it should possess the following desirable properties: 1. The j2 axis in the s-plane should map into the unit circle in the z-plane. Thus there will be a direct relationship between the two frequency variables in the two domains. Scanned with CamScannerSec. 8.3 Design of II Filters From Analog Filters 687 2. The left-half plane (LHP) of the s-plane should map into the inside of the unit circle in the z-plane. Thus a stable analog filter will be converted to a stable digital filter. We mentioned in the preceding section that physically realizable and stable IIR filters cannot have linear phase. Recall that a linear-phase filter must have a system function that satisfies the condition HQ) = be" HE) (8.3.4) where z- represents a delay of W units of time, But if this were the case, the filter would have a mirror-image pole outside the unit circle for every pole inside the unit circle. Hence the filter would be unstable. Consequently, a causal and stable IIR filter cannot have linear phase. If the restriction on physical realizability is removed, it is possible to obtain a linear-phase IIR filter, at least in principle. This approach involves performing a time reversal of the input signal x(1), passing x(—n) through a digital filter 1/(z), time-reversing the output of H(z), and finally, passing the result through H(z) again. This signal processing is computationally cumbersome and appears to offer no advantages over linear-phase FIR filters. Consequently, when an application requires a linear-phase filter, it should be an FIR filter. In the design of IIR filters, we shall specify the desired filter characteristics for the magnitude response only. This does not mean that we consider the phase response unimportant. Since the magnitude and phase characteristics are related, as indicated in Section 8.1, we specify the desired magnitude characteristics and accept the phase response that is obtained from the design methodology. 8.3.1 IIR Filter Design by Approximation of Derivatives One of the simplest methods for converting an analog filter into a digital filter is to approximate the differential equation in (8.3.3) by an equivalent difference equa- tion. This approach is often used to solve a linear constant-coefficient differential equation numerically on a digital computer. For the derivative dy(1)/dt at time 1 = nT, we substitute the backward dif- ference [y(nT) — y(nT —1)]/T. Thus dy) _ YT) - ynT - T) dt |ear T * y(n) = yr = 1) 7 (83.5) where T represents the sampling interval and y(n) = y(nT). ‘The analog differ- entiator with output dy(t)/dt has the system function H(s) = s, while the digi tal system that produces the output [y(n) — y(n — 1)]/T has the system function H(z) = 1 —27)/T. Consequently, as shown in Fig. 8.29, the frequency-domain Scanned with CamScanner668 Design of Digital Filters Chap, g sa)= sD) Figure 829. Substitution of the backward difference for the derivative o implies the mapping s = (1—=-ly/r, ‘equivalent for the relationship in (8.3.5) is 1- s= (8.3.6) T 1)/dt* is replaced by the second difference, which The second derivative is derived as follows: &vo] _ a Fan Ta Near afr]. _ ber = yet = DVT = let = 1) - yet -2N VT T (2) —2ya—- 49-2 ya) (a = ) + y(n — 2) (8.3.7) In the frequency domain, (83.7) is equivalent to ide 1- ( 7 (838) ly follows from the discussion that the substitution for the kth derivative of »(0) results in the equivalent frequency-domain relationship ‘ ) (8.3.9) s Consequently, the system function for the digital IIR filter obtained as a result of the approximation of the derivatives by finite differences is HO) = Hal dha 83.10) where H,(s) is the system function of the analog filter characterized by the differ- ential equation given in (8.3.3). Let us investigate the implications of the mapping from the s-plane to the (8.3.11) Scanned with CamScannerSoc. 8.3 Design of IIR Filters From Analog Filters 669 ot 1 . OT = Teer tier As Q varies from —oo to oo, the corresponding locus of points in the z-plane is a circle of radius $ and with center at z= b as illustrated in Fig. 8.30. It is easily demonstrated that the mapping in (8.3.11) takes points in the LHP of the s-plane into corresponding points inside this circle in the z-plane and points in the RHP of the s-plane are mapped into points outside this circle. Con- sequently, this mapping has the desirable property that a stable analog filter is transformed into a stable digital filter. However, the possible location of the poles of the digital filter are confined to relatively small frequencies and, as a conse- quence, the mapping is restricted to the design of lowpass filters and bandpass filters having relatively small resonant frequencies. It is not possible, for exam- ple, to transform a highpass analog filter into a corresponding highpass digital filter. In an attempt to overcome the limitations in the mapping given above, more complex substitutions for the derivatives have been proposed. In particular, an Lth-order difference of the form ay at (83.12) (aT T) L 2a OTD 83.13) T has been proposed, where {as} are a set of parameters that can be selected to optimize the approximation. The resulting mapping between the s-plane and the zplane is now ie s=pme-2) (83.14) a Unit circle Figure 830 The mapping s = (1 — :“1)/T takes LHP in the s-plane into points inside the circle of radius } and center z= } in the z-plane. Scanned with CamScannerIm) ‘ z-plane Unit Circle Fig. 8.2 The Mapping of Eq. 8.5 into the z-plane It can be seen that the mapping of Eq. 8.5 takes the left-half plane of s-domain into the corresponding points inside the circle of radius 0.5 and centre at z = 0.5, and the right-half of the s-plane is mapped outside the unit circle. As a result, this mapping results in a stable analog filter transformed into a stable digital filter; however, as the locations of poles in the z-domain are confined to smaller frequencies, this design method can be used only for transforming analog low-pass filters and bandpass filters having smaller resonant frequencies. Neither a high-pass filter nor a band reject filter can be realised using this technique. ‘The forward difference can be substituted for the derivative instead of the backward difference. This gives, dy@) _ ynT +T)-yT) T yn+ D- yn) = ¥ntDa-y@) (8.12: Tr ) The transformation formula will be z-1 a 8.1: s25 (8.13) or, z=1+sT (8.14) The mapping of Eq. 8.14 is shown in Fig. 8.3. This results in a worse situation than the backward difference substitution for the derivative. When s = j ©, the mapping of these points in the s-domain results in a straight line in the z-domain with coordinates (Zseaiy Zimag) = (1,27). Consequently, stable analog filters do not always map into stable digital filters. Scanned with CamScannerInfinite Impulse Response (IIR) Filters 421 sa Im(z) A s-plane 1” zplane ° o o| i) Rew) , Po \ / Unit circle Fig. 8.3 Mopping of Eq. 8.14 into the z-plane The limitations of the two mapping methods discussed above are overcome by using a more complex substitution for the derivatives. An Mth order difference is proposed for the derivative, as shown. dy (t) 1x y(nT +kT)- yal —kT) = =2>y4,7"eeoOoemea (8.15) dé ne D> T where {a,} are a set of parameters selected so as to op approximation. The transformation from the s-plane to the z-plane is then, 1x s= aD an (24 - *) (8.16) me By choosing proper values for (a,}, the j 2 axis can be ma unit circle and the left-halfs-plane can be mapped into po’ unit circle in the z-plane. Pets feewe Use the backward difference for the convert the analog low-pass filter with system function H(s) st+2 Solution The mapping formula for the backward difference for the derivative is given in Eq. 8.5, i.e. Scanned with CamScanner422 Digital Signal Processing \ IfT = 1s, Use the backward difference for the derivative and convert the analog filter with system function H(s) = =——— 1-227 +274 1677 Ezra ‘An analog filter has the following system function. Convert this filter into a digital filter using backward difference for the derivative. 4 Hs) 7 (s+0.D? +9 Solution The system response of the digital filter is a z +o.) +9 72 H(@)= (1+0.2T7 + 9.017?) ios +017) 1 Zz? @+027 +9017) * *Groarse0iT) Scanned with CamScannerInfinite Impulse Response (IIR) Filters 423 If T= 1s, H(z)= 0.0979 1 0.2155 z"' + 0.09792 2? 8.3. IIR FILTER DESIGN BY IMPULSE INVARIANT METHOD In this technique, the desired impulse response of the digital filter is obtained by uniformly sampling the impulse response of the equivalent analog filter. That is, h(n)=hg (nT) 7 (8.17) where T is the sampling interval. The transformation technique can be well understood by first considering a simple distinct pole case for the analog filter's system function, as shown below. M H,(s)= 5 —4 imi = The impulse response of the system specified by Eq. 8.18 can be obtained by taking the inverse Laplace transform and it will be of the form (8.18) Pi a A, (t)= > Ape? v(t) (8.19) 4 where u, (t) is the unit step function in continuous time. The impulse response h(n) of the equivalent digital filter is obtained by uniformly sampling ,(t), i.e. by applying Eq. 8.17 Mt h(n) =hgT) = > Ape?!" uz (nT) (8.20) a ‘The system response of the digital system of Eq. 8.20, can be obtained by taking the z-transform, i.e. Hw = 5 hye ro Using Eq. 8.20, ~ > H@= > be Ae?"? uy wnmy]e" (8.2) zo Interchanging the order of summation, Mw > [= Ae"? uy on] aa nz a He) Scanned with CamScannerDiora Sion Process In this method, the design starts from the specifications of analog filter. We have to replace the analog filter H,(s) by a digital filter H(z), provided the frequency response of H(z) resembles| that of H,(s). If the impulse response h,(t) of H,(s) and the impulse response h(11) that of H(z) are invariant ie,, if h(n) = h(n) ---(9.39)| then H(e) and H,(Q) have similar responses. |. 9.4.1 The Transformation of H,(s) of H(z) for Invariant Impulse Response Let y HO)= > A -+(9.40)| int S~Pi } where p, are the poles of H,(s) and A, are the partial fraction coefficients, if Nis) | iQ) = ——_———————— ah “O* Spe Pa) = Pw) any Taking inverse Laplace transform of (9.40), we obtain Nn yo ht) = > A; e" ft wy h{nT) = h(n) =, A; e?" [replacing t by nT in ...(9.42)) A GF Toking vtransform on both the sides of (9.42), we get He)= Y hn" (by definition) for casual h(n) " Ms: (% Aen 0 (fe S nent en 4 Mz uA HO) 2 Tergt 0.43) which is the required digital filter H(z) obtained from H,(s). 9.4.2 Mapping of Poles, Relationship of s-plane to z-plane Thue, maps to Tart ie., a pole s = p, of H,(s) maps toa pole z= ¢PT of H(z). Scanned with CamScannerDesicn oF IIR Furrers Bl other words, in Te (9-48) $= 6+ jQand z= ro! r real - rei = e007 7 rel = eo | oft « rect, (9.45) o=9r ' Gis +ve,r= eT >1 ¥ sis-ve, t zplane unit circle Fig. 9.12. Mapping in invariant impulse response transformation Hence, L.HSS. of s-plane maps inside the unit circle ie.,a stable analog filter H,() results in a sable digitals circle H(z). RHS. of splane maps outside the unit circle | z | = 1. The jQ axis maps on the circumference dhe unit circle, As o=9T o=-% to+ = maps to @=-n to +n Q= ? to Fay) maps to =x to 3nie,,0=-z ton ‘ss¢* is periodic with period 2n) This implies that when | 2 | exceeds % then these frequencies overlap in frequencies ™pped by 1 2 |< = This causes ALIASING. °43 Drawbacks of Impulse Invariance Method 1 Analog frequency -4 sQs z covers e” from ~ x to x (+. @ = QT). Mapping of (i+ oe to (/+1)£ maps many times, (i+ 1) times, in #® domains. Thus, the mapping from analog frequency @ to digital frequency @ is one to many or many to one. The Mapping is not one to one. Scanned with CamScannerBLDicrasStena. Procesane 2. Due to aliasing, only limited frequency band “= 9 F filter can be transte without alias 15 in z domain, 3, The change in value of sampling time T has no effect to remove aliasing. Rule to Design a Digital Filter Using Impulse Invariance Method 1. With given specifications of digital filter H(z), obtain expression of H,(s) here = Frequency transformation is linear as @ is proportional to Q. 2. Express H,(s) in partial fractions (in first or record degree terms) 3. Obtain H(z) from above partial fractions using standard expressions viz. 1 1 ska sop Teel (s4ay +b? (stay +b? 4, We have, thus, obtained H(z) from H,(8). Note that | 2 | should be < Z to avoid in z-domain. 7 Standard Expressions for Invariant Impulse Response Transformation 1 1 — —> o-p 1-eZ A To show that Bea im 1-0" (cos bT) 2) os oo Fo bya! a tem (ray +b? T= 20" (cos dT) 24 TEE Proof: Poles of ware are at py = a + jb, py = a= jb a. sta a pba " Pota Grae 4h? © = py)+—P2) (= P+ = Pa) (—P24P2— PV) 1 Pea peta) ak jb sib . Mo PL s-—i SP} 2ib| “Pr a Scanned with CamScannerDesicn oF HR Fins a Divwigar tem) Tp) (OO WR oa a A __2-e"" Qeos bry 21-e" QeosiTyzte s+a UR (sta)? +b? (9.47) + b wfind IIR transformation for Grae trooft Inksof —4—e are OO (eka +b b 1 — > 2b (sr a)? +b? (8-1) p2) 1 1 =b})-———__ yd lz P(r = Pa) (= pr) (pa =a 5 (nm) [+ rl sal Pimps Jls=p sp Hence, LHS. transforms to: H(z) = (9.48) 5 samples/s, 5 sammples/s Scanned with CamScannerHL W@s2 Here, H() (0.67 + 0.818) = {0.818 + 0.67) + 0.548 == 0.148 =! * Jyrram sl hearSC(C« As. 1-1.488 2740548 = Burr 9.5: The system function of an analog filter is given by Hs) = —5*2 (+0.2P +9 7 ing T=1S" nalog filter into a digital filter using imspulse invariant technique, assuming Sowmos: Given Here, Ie“? (cos bT) He@)= tae" (cos tT) 1-2e“T (cos BT) 27 + 1c? cos (3T) 21 = 1. pear Z Scanned with CamScannercri irs RE (0.9899) = 1-2(08187) Wao 1~2 08187) (0.9999) 7 «0.6703 z= 1+ (08104 24 141.6208 27 Exum 9.6: Given an analog st Ans, tem HAs) : +0508 4054) Obtain a digital filter by using impulse invariant response (UR). transforms. Assume T = Nt AS Hs) = ——__________, B ( +05) (-O5¥ +05(-05)+2) (F +05s42) To find A,B As Compare Compare s? Hence, 0 ue _ OS fs+0.25]+ 0125 +025) + (2— 0.0625) 7 =(13919" $+025 {0238 Sr 0.25p +13? (13919 ] (s+ 0.25)" + (1.3919) = (0.5) 1-7 [cos 3N9TI=* 1=2e°=* cos (1.39197) AR, - 05) eS sin (1.3919 T)=7 ’ Se + (0.898) [~9TST cos (L3MIT|E +N = f (5) [1 - 0.1385 21} + 0.898) (0.7663 ="'] 1-027 = +0.606>* AG@)=1+= Ans. Scanned with CamScannerBE Dicrrat Stowat Processinc rhe sustem function of an analog filter is given as 6404 (401) +37 Exasarie 9. HE) Obtain the system function of the IR. digital filter by using invariant impulse response method 5+01 sta Soumone: Mi)= ayy age OM) = Es Here a=01,b=3 He l+e' (cos bT)z 1, po Se 26 “eos bT yz +6 2" (con BT) 2" = Ans. co BT) 2 * 4 OFF gn? WT = 1, for example, then NCOs 37) 2°" Hu) 1=e(con 30) z 1 2(0.8950) 2! + O.B187 2? 14 0.895 Ans 1-1,7913 2) + 0.6187 2? Fxamvrs 9.8: The system function of an analog filter is given by Hf) » Obtain the system function of IR digital filter by using invariant impulse response mets Bonin: We can write H,(6) as (640.3) 42 540. 144 H,() © eee eee We es 2 (540.3) +4" Porm —2*! 7 orm ——o 7 (ray ob? Gtapab Here a=03,be4 “1 (cos bT) Assume T= 1, use a= 0.3 and b = 4 Scanned with CamScanner611 u ind = 20 Costes ete 18 (0.6536) + 0.5 (-0.7568)) = 2(0.7408) (~0.6536) 27 + 0.5488 140.9683 2) 6. Ans. Bxatarts 9.9: Design a Butterworth low-pass digital filter, assuming passband magnitude constant within 1 dB for 0 to 0.2 x stopband magnitude (attenuation) is < 15 dit from 0.3 2 10% use impulse invariance transformation. Sowution: 1H,09) og Hye) 008 - 108 = 15 0B} see 00% 05 Tat fa) (b) Fig. Ex. 0.0. (0) Analog Bullernortns firs (yplea), (b) Speciiod digi The given data is 20 log, | Mle) 12-1 dB for 1150.2 ha 20 logy, | H(o™ 1-15 dB for03x4 1 ols Pulse invariance method of transformation, () Frequency j ‘i quency is mapped finea i) The 'y is mapped linearly LPF filter Tos 8 aliasing, problem: dhsign Ott ith, we aysume negligible effet of aliasing, and check such a performance after the ‘Ne here 7220 dee, i = Hy(s)|or,, = HE) Th oi He) = Help ae As in @ Butterworth filter we have 1 H,GQ) 1? = ——~ Sa Scanned with CamScannerO2e gives 9.=02R fori apa ; O35 gives Q.=03 xfer iS aBate a Wlog | H,G02n) | =-14B log! Hf03x) | =-15 6B sng 1-( 22) 21 j of» Dividing (8) by (A) goves (232 »ea (Gin, * Gay 6% = O5>* = 138.25 2M bog i5 = log 11828 = 20729 Fam Bien a |b maa - 54000 g, Fay 719102 2 67519 Scanned with CamScanner5 ae OED = 0.3670 5 = 0502) (= = 1.00. 3 | +0502) (s Joobtain H(c). we must put H,(s) in th. form noe O85 1 Hie) isi, Osus6. : Hie) 1562, 64208 . Out the system responce Hi we dae EN This is a Buttervor linear transformation, then ” Fig, Ex. 940. The desives tegea! tie hs : [or an analog Butterworth filter] Scanned with CamScannera In H(z) } doo) . As M(o) and @(«) are real quantities ign M(@) isimaginary and —Go._ is real quantity. 2H) «ee QED. Thus, Real part of (+ Sin He) which is same as (9.6). FERAL buinear TRANSFORMATION (INTEGRAL METHOD) We now derive the expression for bilinear transformation which will be used in the design of IR filters. Here, we integrate the differential equations of analog filter and then use numerical | approximation to the integral. Let : CUg(E) + equ elt) = dyx,(t) x +12) be the first order analog filter. Taking Laplace transform on both sides of (9.12), we get c,8Y (8) + ¢gY_(6) = dyX, (6) X@) _ _ do. As) X,6) @ +45 (9.13) We can write Yalt) = Yat) ~ Yalta) + Yalta) (Note this step) = ff yaa + yale) Let in —1)t,t=nT WM car = Seay VAD AE HY —DT) ie, 007) = Liycent) + yg(or-YT)] + y_{(1—DT) by Tras ndal Rule of | (9.14) We can write from (9.12), Putting 1 =nT there, Fig. 9.1. Explaining trapezoidal rule ot integration YAnT) = ~ Buoys x,(nT) eis) a Scanned with CamScannerDesten oF WR Fires Hl F G yl -1) T] =-2y,{(n-)T]+ * x,{(1-1)T] (Changing n ton—1 in (915)) q 1 Using these values of y/(nT) and y,{(v - 1) T] in (0-14), we have y(n) = F[-S win-nte - Th I 2a 4 40 fx, (nT) + allt -omi y{ln-IT] 936) Cy Defining y,(uT) 4 y(n) and x,(nT) £ x(n) (9.16) yields y(n) = yu =1) = F[-Su0- n= = 1+ Btn) ate wi Taking z-transform on both sides, we get Ye n-21= F[-Sreaee ne xe fe) =r 42 Sage) exeql aero or roa z a (l+z )]=x@F Barz ) Hes 2 X@) @az =. 35 ein ty ae z Comparing (9.18) with H,(¢) = + We find that +65 ai) Bilinear transformation “ a Scanned with CamScannerBDicna Stowns Process This is the Bilinear transformation i., in the given H,(s), if we replace s by ze obtain H(z). Hence, HAl)| Tass") Important Note: Though we have derived the Bilinear transformation formula based on the first ord! differential equation, cy,(t) + cay,(t) = dox,(t), the result thus obtained holds good also for N-t) differential equation of the generic form x at 2 MO = ¥ a xy (920) = a 1 i implies “Ct ; where the function i) implies . This is so because the above equation can be writtens a set of N first order linear equations of the form: CVAD) + coug(t) = dy x,(0) can be used for all order differenti equations. ) Also, it is well known that bilinear transformation is conformal. This important pron ensures that the ‘shape’ of the response H(s) will be similar to the shape of the response HG) we use BLT (bilinear transformation). Inshort, the transformation s 9.2.1 Mapping of s-Plane to z-Plane in Bilinear Transformation As 0 + jQ and the bilinear transformation is a) Scanned with CamScannerDesen or R Fivers (EE ci = gitar anny 7 @=2tan (¥) ; (ar = zt (Hf) ° SS =tan2 12(9.22) " - 2 2 a Putting s = jQ in (9.21), we get Iz | =1 for all values of Q, Thus, {2 axis maps to the unit circle | = | in plane. Also, from (9.22) when Q is +ve, @ is +ve and for Q-ve, @ is -ve. Hence +ve and —ve parts of jQ axis map to upper and lower halves, respectively, on the z-plane Fla. 8 unit circle | z 1 drcle 1 z | ‘Mapping of analog trequeney 91 on In fact, j2 == to+e maps the unit "INE Unt eel using Blinear anstration ONLY ONCE, as is apparent from Fig, 9.2. Mapping of LHP and RHP of s Plane 1+ +jQinz= ! ie 1+ T (6+ ja) 2. 1-5 is 5 (+0) Clearly, for (8.20) (0.238) Lover JQ axis maps to Tower hall of] z] = 1 (unit crete) Fla. 9.3, Mapping of ssplana to =nlane in Binet transkomation Hence, LHP splane maps to inside the unit citcle |= {= Land RUD plane maps to outside the unit circle [2 1 = 1. The stable analog filter HS) yields a stable digital filter HED, KNAUSS, ICH) has all its poles on left half of plane, ane these poles of stable £28) map insite the anil circle to gives stable HG). Scanned with CamScannerBE Dicrrat Sicwat Processine Also, the entire jQ axis, from —se to +62, maps to ONE TRACING of the unit circle. Thus, thee is no aliasing. Exams 9.1: An analog filter H,(s) lias a pair of poles at s = s,. What would be the transformed | pole pair =, =. if t0¢ use bilinear transformation to obtain H(z) from H,(s)? SowuTion: Let the given function be K.N() H,(s) = fe) +5) where N%s) <1 IN%s) means degree of N(s)]. “1 H@)= Poles of H(z) are given by ; fod -2)=5,7 0 +2) 20-2) 5,7 O42 [2 +5,N- 2-3, 2-30-2429) ie, T oT] [12457] 9 ie, 2-57, 2+5,T :. The poles are nm = 375,T = pT 24s,T We note that ta Exampte 9.2: Show that in a Bilinea’ in z-domain. the pole pair Scanned with CamScannerDestcw oF HR Fivtens sy st sgowrion: Let H,(s) have the pole pa HJ9)* Gaeta) (a typical case) to transform H,(s) to H(2) ie, Put Clearly, the poles are of the form =, and =, Note that the results obtained here are the same as in Example 9.t, where we found Vise i fa TYPICAL STANDARD FILTERS We shall now explain the basic properties and parameters of some of the analog filters used 8 designing the IIR digital filters. The most important typical filters are: (9 Butterworth Filter: Is monotonic in passband as well ay in stypband, (i) Chebyshev Filter: An equirippte characteristic i the pass band and mot stopband, (4) Elliptical Filter: Is equiripple in both the pasband as well as i st pogo ptund Scanned with CamScannerBDicra Sicxat Processinc During the bilinear transformation, the above properties are preserved. However, the line phaseness is lost ic. if we are interested in a digital lowpass filter with linear-phase characteris, we cannot obtain such a filter by applying bilinear transformation to analog low-pass fits having linear phase characteristics Note: Chebyshev filters can be further classified as the following: (a) Type L Chebyshev filters are all pole filters that are equiripple in the passband and monotan: Gharacteristic in the stopband. ©) Type Il. Chebyshev filters contain both poles and zeros and exhibit a monotonic behaviour inf passtand and an equiripple behaviour in the stopband. The zeros of this class of filters le on Se imaginary axis in the plane. 9.3.1 Digital Butterworth Filters Properties: (8 Butterworth filters are defined by the property that the magnitude response is maxim: flat in the passband. “order low-pass filter, the first 2N — 1 derivatives of | H,{j®)|? are zero z #) The approximation is monotonic in the passband as well as in the stopband. If H,(0) is the frequency response of the Butterworth filter then Fi T ; 1H.) 1? = ———. | Nt\order Butterworth low-pass filter, 1+(2 analog domain ...(924) iQ, N increases, the filter characteristics become sharper, as shown in Fig. 9.4. They rem to unity over more of the passband and become close to zero more rapidly in the s*7 a Scanned with CamScanner