CHAPTER 3 OA-C FILTERS ; DIRECT FORM SYNTHESIS^ 3 .0 In trod u ction This ch apter i s concerned w ith the r e a liz a t io n o f secon d - and h ig h er-ord er OA-C f i l t e r s , using the d ir e c t form syn th esis approach. This syn th esis technique should be such th a t i t r e a liz e s the fin a l network in the form s u ita b le f o r the m icrom in iaturization in the MOS technology^® ’ ^ . This im p lies that the c i r c u i t s should con ta in only the a ctiv e d e v ic e , v i z . , op era tion a l a m p lifie r s , and ca p a cito rs in t h e ir 70-7^ r e a liz a t io n s . Moreover, a l l the im portant parameters o f th e network should be a fu n ction o f ra tio e d -C s. Three techniques are proposed here f o r such a purpose and are c r i t i c a l l y stu d ied . These are ; i ) L-Based Approach to OA-C sy n th esis ; i i ) FDNR-3ased Approach to OA-C s)Tithesis ; and iii) FDNCAP-Based Approach to OA-C sy n th esis. In any d ir e c t form sy n th esis technique, th e s ta r tin g p oin t is the p assive RLC-prototype network, whose elements are to be r e a liz e d in a manner so that the fin a l r e a liz a t io n contains only those elem ents, which are s u ita b le f o r convenient m icro­ m in ia tu riza tion . In the presen t ca s e , only the OAs and Cs t The m aterial presented in th is ch apter has led to the p u b lica tio n s o f the au th or’ s paper no. 3, and k l i s t e d on page v ii. 8 8 89 are perm itted in the f in a l r e a liz a t io n s . The r e s t r i c t i o n on the use o f only OAs and c a p a c ito r s requ ires the sim ulation o f two d if fe r e n t types o f c i r c u i t elements in each o f the th ree proposed techniques. The a p p lic a tio n o f a p a r tic u la r technique w i l l la r g e ly depend upon the c o n fig u ra tio n o f the p ro to ty p e . Each technique w i l l be d iscu ssed in d e t a ils in the subsequent s e c tio n . Here we only presen t t h e ir o u tlin e s . 0 In the L-based Approach to Oa -C S y n th e sis, the R and L elements are sim ulated and d ir e c t ly su b stitu te d in the p ro to ­ type to giv e an equivalent r e a liz a t io n in the OA-C form. 0 In the FDNR-based Approach t o OA-C S y n th esis, the admittance o f the p rototyp e network (N) i s f i r s t sca led by the complexfrequency v a r ia b le (s ) to g iv e the transform ed network (N ^ . The transform ation re s u lts in con v ertin g an RLC-network in to a corresponding CRD-network^^. The f i n a l r e a liz a t io n i s then obtained by sim ulating the R and D elements by the OA-C sim u lators. I t may be noted th at the tran sform ation does not a f f e c t the tr a n s fe r f m c t i o n o f the networic. 0 In the Fl^CAP-based Approach t o OA-C S y n th esis, the admitt­ ance o f the p rototyp e RLC-network (N) are sca le d by the func­ t io n , s^, to give a corresponding CDF-network (N^^), without a ffe c t in g the tra n s fe r fu n ctio n . The fin a l r e a liz a t io n b o ils down to the sim ulation o f D and F elements in the OA-C form. For a l l the th ree tech n iqu es, the OA-C sim ulators, 90 already studied in Chapter 2, are used. As the sim ulation o f two d if fe r e n t type o f elements g e n e ra lly requ ires a la r g e count o f a c tiv e and p a ssive components, methods are a ls o suggested f o r the op tim ization o f component count. The three approaches are c r i t i c a l l y stu died f o r the r e a liz a t io n o f secon d -ord er and h ig h e r-o rd e r f i l t e r s , res­ p e c t iv e ly in S ection s 3 .1 , 3.2 and 3 .3 . Suggestions f o r 't h e op tim iza tion o f component count and s u i t a b i l i t y o f stru ctu res f o r a p a r tic u la r technique are c l e a r ly examined. In S ection 3 .^ , experimental r e s u lts on some c i r c u i t s are inclu ded. The con clu d in g remarks are given in S e ctio n 3 .5 . 3.1 FDNR-Based Approach t o OA-C F i l t e r R e a liza tio n s In the Frequency Dependent N egative R esistan ce (FDNR)- based approach t o OA-C s y n th e sis , the p a ssive RLC prototype c i r c u i t d escrib in g the tr a n s fe r fu n ction i s f i r s t transformed in t o an equ ivalent network by s ca lin g each admittance o f the o r ig in a l network (N) by the c o m p le x -v a r ia b le (s ), as shown in Fig. 3 .1 . The new network (N^) i s obtained by con vertin g a r e s is t o r to a c a p a c it o r , an in d u ctor t o a r e s is t o r and a c a p a c it o r to a frequency dependent n ega tive re s is ta n ce element^^. The FES^R has an impedance o f the form , 1 /s D, where D is para­ meter o f the element w ith the u n its o f square farad-ohm (f , ) . Thus, the RLC-network (N) i s con verted to a CRD- netwoiic (N^ ) without a ff e c t in g i t s tr a n s fe r fu n ction and the 91 NETWORK (N) Vr o--- V nAA— o ^ N E T WO R K (N ) 5C‘ II o; VsL o----- -o sC o-------- II------ o Fi'g.3.1 c=Vr Vr' ■AAt^----- o i R = L Transform a­ tion C,RD e q u i v a l e n t s of R L ^ s^D elements 92 sy n th esis problem b a s ic a lly becomes that o f the r e a liz a tio n o f r e s is t o r s and FDNRs in the OA-C form. I t may be noted that in t h is tech n iqu e, the fin a l n et­ work r e a liz a t io n requ ires the sim ulation o f two d iffe r e n t types o f c i r c u i t element, v i z . , FDNRs and Rs. This can be achieved by e ith e r rep la cin g each component by i t s id e a l Oa-C immittance sim ulator or by re p la cin g a s e t o f components by the equ ivalen t n on -id ea l OA-C component sim u lators. Generally the c le v e r use o f the second approach proves out to be more a t t r a c t iv e from the co n sid e ra tio n s o f minimizing a c tiv e and p a ssiv e componenxs and in the r e a liz a t io n s having a tt r a c tiv e s e n s i t i v i t y p ro p e rtie s . In th is s e c tio n , the FDNR- based technique i s used f o r th e r e a liz a t io n o f ( i ) standard secon d -ord er OA-C f i l t e r s and (ii) h ig h er-ord er OA-C f i l t e r s . In S e ctio n a standard secon d -ord er OA-C band-pass f i l t e r i s r e a liz e d by employing id e a l FDNR sim ulator. This i s follow ed by the r e a liz a tio n s o f standard secon d -ord er OA-C band-elem ination and a ll-p a s s f i l t e r s using the n on -id ea l FEWR-based component sim ulators. The advantages a sso cia te d w ith the use o f n on -id ea l sim ulators are c le a r ly demonstrated. The technique i s a ls o employed in S ection 3 .1 .3 in the r e a liz a t io n o f h ig h er-ord er f i l t e r . 93 3 .1 .1 R e a liza tio n o f a secon d -ord er s e c tio n s w ith id e a l FDNR-simulator The importance o f standard secon d -ord er f i l t e r s e ctio n s in a c tiv e syn th esis hardly needs any e la b o ra tio n . In Section s 3 .1 .1 and 3 .1 .2 the r e a liz a t io n o f secon d -ord er f i l t e r s , using th e FDNR-based approach, i s demonstrated. These employ the FDNR-based sim u lator o f Table 2 .2 . The r e a liz a t io n o f a secon d -ord er band-pass f i l t e r i s f i r s t demonstrated by employing an id e a l FDNR sim ulator. C onsider the p a ssiv e p rototyp e secon d -ord er band-pass (BP) f i l t e r , shown in F ig. 3 .2 ( a ) . I t s s c a lin g by the complex frequency ' s ’ leads to the FDNR-based v e r s io n , shovm in Fig. 3.2 ( b ) . For the fin a l r e a liz a t io n o f the BP f i l t e r in the OA-C form, the elements in the shunx arm are to be replaced by the appropriate component sim u lators. In t h is ca se ,th e id e a l FDNR i s replaced by the id e a l OA-C FDNR sim ulator o f Fig. 2 .7 and R;C shunt-br^nch i s rep la ced by the immittance sim ulator o f Fig. 2 .1 8 . in Fig. 3 . 2 ( c ) t . The com plete r e a liz a t io n i s shown D ire ct a n a ly sis o f the c i r c u i t , y ie ld s the tr a n s fe r fu n ction as : ^ '’ P w ^ ° )S s ^ + (^ )s t w 2 w H, „ ( t ~ ) s = _&E_9----- (3.1) D(s) As an a lte r n a tiv e , the r e s is ta n ce o f the shunt branch may be replaced by id e a l R -sim ulator o f F ig. 2 .1 6 . This however req u ires matching c o n s tr a in t and la r g e component count. 9^ -TO BE SIMULATED + o- t \ _ ii 1 ''i r, C - R ^ | | D ± - o L , ‘__ I l__ -o - (b) (a) REQUIRED O A - C SIMULATORS (C ) -o + OUT ---- o IN o- IN o— OUT -o _o - (d ) Fig. 3-2 (a) P a s s i v e R L C prototype b a n d - p a s s filter ( B P F ) . (b) F D N R - v e r s i o n of p r o t ot y p e BPF. (c) O A — C v er s i on of F i g . 3 - 2 ( b ) . (d) C - M O S inverter. 95 where, and Q are, r e s p e c t iv e ly , the mid-band gain, the p ole-freq u en cy and the p ole-Q o f the BP f i l t e r , having the follow in g r e la tio n s h ip s : C-.C, 1 /2 ] ^ ol ° ° ^ and C, B„ Q = ^ C^ . Cj ^ ol where, C ,. ^k* 1 /2 j (3 .2 ) output has not been taken at the output term inal o f an OA, the response w i l l be load s e n s it iv e . In order t o ob v ia te th is problem , i t i s recommen­ ded to use a b u ffe r stage co n s itu te d by C-MOS in v e rte r shown in Fig. 3 .2 (d ) . An im portant p o in t t o be noted is that the derived c i r c u i t may p ro v id e , in a d d itio n , standard responses the r e a liz a tio n . at the other outputs o f the OAs employed in For example, the BPF o f Fig. 3 .2 ( c ) r e a liz e s standard secon d -ord er low pass response a t the output o f OA^, w ith the tra n s fe r fu n ctio n given by T, ( s ) = ^ V. "°D(s) (3 .5 ) where, the low pass gain = (5 .4 ) From (5 .2 ) to (3 .^ ) i t is seen that a l l the f i l t e r para­ meters are in terms o f c a p a c ito r s r a t i o s , which is a highly 96 d e s ir a b le fea tu re fo r the implem entation o f the c i r c u i t in MOS-technology^^"^^. The c i r c u i t p ossesses independent tuning o f the p o le -fre q u e n cy by b ia s -v o lta g e c o n tr o l. The p ole-Q i s independently tunable by c a p a c ito r C^, without d istu rb in g the p o le -fre q u e n cy . The s e n s it iv it y study o f the f i l t e r was a ls o ca rrie d out. I t y ie ld e d the s e n s i t i v i t y fig u r e s o f f i l t e r parameters o f in t e r e s t , as given in Table 3 .1 . The ejqiressions c le a r ly demonstrate the s e n s it iv it y fig u r e s to be a t t r a c t iv e , being no more than unity in magnitude. However, i t may be r e c a lle d th a t the r e a liz a tio n o f id e a l FDNR sim ulator i t s e l f requ ires c r i t i c a l component matching in Bs, i . e . , = B^. This i n - turn makes the o v e r a ll r e a liz a t io n f a i r l y s e n s itiv e to B2 and B^, which i s not evident from Table 3 .1 . The study o f the s e c t io n demonstrates how a second-order OA-C f i l t e r may be r e a liz e d using id e a l FDNR sim ulator. The problem a ssocia ted w ith the use o f id e a l FDNR are ; ( i ) the c i r c u i t s normally tend to re q u ire a h igher component count. The present c i r c u i t uses fo u r OAs, f i v e ca p a cito rs and a b u ffe r and ( i i ) the id e a l sim u lator req u ires matching o f components f o r i t s r e a liz a t io n , making the o v e r a ll c i r c u i t s e n s itiv e to such parameters. 97 98 3 .1 .2 R ea liza tion o f secon d -ord er s e c tio n s w ith n on -ideal FDNR sim ulators The s e ctio n demonstrates the a t t r a c t iv e featu res o f f i l t e r r e a liz a tio n using the n on -id ea l FDNR sim ulators. The r e a liz a t io n o f two standard secon d -ord er f i l t e r s are con si­ dered as examples, v i z . , band-elem ination (BE), and a ll-p a s s ( aP) f i l t e r s . The d e t a ils o f the r e a liz a t io n and study of these f i l t e r s are d iscu ssed as fo llo w s ; 0 Band elem ination f i l t e r - The p rototyp e secon d -ord er bandelem ination f i l t e r is shown in Fig. 3 .3 (a ) and i t s transformed FDNR v ersion i s given in Fig. 3 .3 ( b ) . To r e a liz e i t in the OA-C form, the shunt brancn i s to be sim ulated. The n on -ideal FDNR sim iilator o f Fig. 2 .1 0 has been used in the d ir e c t simu­ la t io n o f the e n tire shunt arm (shown w ith in the dotted lin e s ) o f th e transformed network. The f i n a l r e a liz a t io n o f the BE f i l t e r in the OA-C form i s shown in Fig. 3 .3 ( c ) . The analysis o f the c i r c u i t y ie ld s the tr a n s fe r fu n ction as : It, ( s ) = ^ 3 = ° ( 3 .5 ) where C2 1/2 ^be ^ ^o^^ol * '^o ^ '^n ^ (% B 2. and Cn Q = (1 + ^ )■( - ^ . - ^ ) 1 \ ^23 1/2 (3 .6 ) 99 Ro -0 + R' ci Vf Vb (a) r -N O N -ID EA L FDNR 0 _______ It _ / / / R 7 / / ± c' D^ T_ / Vb, / z____________ (b) r - O A - C SIMULATOR OF N O N -ID E A L FDNR o- -t- r* cv C2 V lp Vbe "TT (c) F i g . 3.3 (a) Prototype-RLC band-€lemination filter(BEF) (b) S e a Led v e r s i o n of B E F (c) O A - C v e r s i o n of B E F of Fig 3.3 (b) 100 N ote, once again the f i l t e r parameters o f in t e r e s t are having the a t t r a c t iv e featu re o f being in the form o f c a p a c it o r r a t io s . The c i r c u i t en joys the m u ltifu n ctio n a l c a p a b ilit y and p rovid es two a d d ition a l standard responses o f band-pass (b p ) and low -pass (LP) a t the outputs o f OA^ and OA2 , respec­ tiv e ly . The tr a n s fe r fu n ctio n o f the BP and LP f i l t e r s are : D(s) and T (s ) = ^ = J^_2_ D (s) (3 .3 ) where, the re s p e ctiv e f i l t e r s gain are given by Hbp = , and I^.p (3 .9 ) ^ol These are a ls o in the form o f C -r a tio s . O bviously, the e^qare- ssions f o r the w and Q remain lanaltered f o r the a d d ition a l o respon ses. The c i r c u i t p ossesses independent b ia s -v o lta g e tuning o f p ole-freq u en cy w^ (= w^^). A lso, independent p assive tuning o f p ole-Q i s p o s s ib le w ith a n d /or C^. The s e n s it iv it y fig u r e s o f the c i r c u i t were d erived out and the f in a l r e s u lts are inclu ded in Table 3 .2 . I t i s seen that a c tiv e and p assive s e n s i t i v i t i e s are, once again, w ith in unity in magnitude. Note, 101 ■I fc t . J 102 high-valued r e s is t o r s are requ ired a cross c a p a c ito r s , and C^ , to p rovide dc b ia s in g to the in v ertin g -term in a ls o f the OAs. 0 A ll-p a s s f i l t e r - The secon d -ord er p rototyp e o f an a ll-p a s s ( aP) f i l t e r i s shown in Fig. 3 -M a ) sion i s given in F ig. 3 .M h ). and i t s transformed v er­ For the OA-C r e a liz a t io n , the com plete shunt arm, shown w ith in the d otted l i n e s , may be sim ulated by d ir e c t ly using the n o n -id e a l FDNR-simulator o f F ig. 2 . 3 . Fig. 3 . 4 ( c ) . The fin a l OA-C v e rs io n o f the f i l t e r i s given in The use o f n on -id ea l (FDNR) re s u lts in an appre­ c ia b le op tim ization o f a c tiv e and p a ssiv e components. A nalysis o f th e f i l t e r y ie ld s the tr a n s fe r fu n ctio n : where. ^o a p p ' o '"ol o S 1 2 p ^23 ^4 p ^45 and th e AP-response i s r e a liz e d by s a t is fy in g the component c o n d itio n ; C C 1 C a-, = b. = - ^ , when -2 -= - ( - 1 - 1 ) . r* p r ^23 ^1 ^2 (3 .1 1 ) This c o n d itio n can con v en ien tly be s a t i s f i e d , p a r tic u la r ly i f the c ir c x iit i s implemented in m on olith ic form in the MOS tech n olog y . As in e a r l ie r ca s e , the c i r c u i t has m u ltifu n ction al 103 -JVy/v- -o Ro Vi + 1 Vo Ri -R C -o (a) — N O N - I D E A L FDNR 1 Co - S I M U L A T E D N O N - I D E A L FDNR ------------------------------------- 0 + -\h - C3 '+ T Vip -»l-----C4 C5 Vbp (C) F i g . 3-4 (a) Prot ot ype all-pass f i lt e r ( A P F ) (b) T r a n s f o r m e d F D N R - v e r s i o n of A P F (c) O A - C f o r m of A P F Vap 104 c a p a b ilit y and p rovid es a d d itio n a l standard secon d-order BP and LP responses simuiltaneously at the outputs o f OA^ and 0A2 » r e s p e c tiv e ly . The v a riou s p e rtin e n t parameters o f LP and BP responses are given by ; Hip = f^ol - V “ - r^1 C. = (B,B^. ^ ■ C, ) 1 /2 . (3 .1 2 ) and U = 2( 1 + Co Cl B? ~ B^ C, C, C25 ) 1 /2 . ( 5 . 12 ) Once again, the f i l t e r parameters have the a tt r a c tiv e fea tu re o f being in terms o f C -r a tio s . v a riou s filte r The s e n s i t i v i t y fig u re s f o r parameters w ith resp ect to the a c tiv e and p a ssiv e eleraenxs are analysed and the f i n a l r e s u lts included in Table 5 .3 . The ta b le shows th a t the c i r c u i t enjoys a ttr a c ­ t iv e s e n s it iv it y perform ance, being l e s s than or equal to unity in magnitude. The c i r c u i t a ls o enjoys independent e le c tr o n ic tuning o f p o le -fre q u e n cy . This f i l t e r can r e a liz e high-Q values due to the a d d itio n a l m u ltip lic a tio n factor^ (1+ C ^ /C i), in v olv ed in the exp ression o f (A. To p rovide the dc bias at the in v e rtin g -in p u ts o f both the OAs, high-valued r e s is t o r s are required a cross the ca p a cito rs and C^’. I t i s important to n ote that both the c i r c u i t s , re a liz e d w ith the n on -id ea l based sim u la tors, have very a ttr a c tiv e 105 106 s e n s i t i v i t y p ro p e rtie s . In ad dition ^ th ey norm ally do not req u ire component matching in t h e ir rea liza tion s'*’ , and t h e ir t o t a l a c tiv e and p a ssiv e components are con sid erably low er. This i ll u s t r a t e s th a t, whenever p o s s ib le , c le v e r use o f n on -id ea l sim ulators should be made in the r e a liz a ­ tio n o f networks based on the FDNR approach. 3 . 1 .3 R ea liza tion o f h ig h e r-o rd e r ladd ers H igher-order f i l t e r s are freq u en tly r e a liz e d in the form o f lad d er stru ctu res^ ^ ’ ^ ^ I n t h is s e c t io n , the FDNR based approach i s applied to the a ctiv e -C r e a liz a t io n o f an n -th order a l l - p o l e la d d er. The v o lta g e tr a n s fe r fu n ction of the LC la d d er, shown in Fig. 3.5 ( a ) , i s given by T ( s ) = - j -------------- ^ s + b_n-1T s K bo -------------------------+ . . . + bTS+bo 1 (3 .1 3 ) where, bo, b^ — , b^_^ and K are the a p p rop ria tely choosen con sta n ts, depending upon the f i l t e r c h a r a c t e r is t ic s . Low- pass f i l t e r s , l i k e , Butterworth, Chebyshev, e t c . , have tr a n s fe r fu n ction s o f th is typ e, d if f e r in g only in the ch o ice o f the c o e ffic ie n ts . The LC-ladder may be r e a liz e d in the OA-C form by using id e a l flo a t in g inductance sim u lators, along with grounded p a ssive Cs. This i s , however, n ot a t t r a c t iv e from The C-matching in the AP-case has been imposed by the b a s ic stru ctu re from which the A P - f i lt e r is re a liz e d . 107 L2 Rs L4 — •- O+ --C 3 C,: o -- 1--------- •- 0----------- n EVEN n ODD (a) + Cs R2 » — —. o3------- 11----' Dl == R: ---- •------- ----- 'n D3: 1— 0 =C l Dn^ ^C l ------------- 9-------- •------------ 1--- 0 ■ ■ 0- 0— 1 n EVEN n ODD — o>- (b) Fig. 3-5 (a) (b) nth-order pr o t ot yp e FDNR-version ladder. LC l o w - p a s s ladder. of t h e p r o t o t y p e l o w - p a s s 108 the co n sid e ra tio n o f p r a c t ic a l implem entation o f the c i r c u i t . The id e a l flo a t in g Ls req u ire a la r g e count o f a c tiv e and p assive components, b esid es the requirement o f c r i t i c a l matching c o n d itio n s . T h erefore, the FDNR-based r e a liz a tio n o f the la d d er is con sid ered . by the complex frequency The la d d e r network i s scaled *s*, which transform s i t in t o the CRD-version, shown in Fig. 3 .5 ( b ) , w ithout a ffe c t in g the tr a n s fe r fu n ction . The problem then reduces t o the r e a l i ­ z a tio n o f id e a l grounded FDNRs and flo a t in g Rs in the OA-C form. However, as i t has been p oin ted out in Chapter 2, the id e a l sim ulators req u ire la r g e component coiant. T herefore, the a ctiv e-C v e rs io n o f the ladd er i s r e a liz e d in th is s e c tio n by employing n on -id ea l immittance sim u la tors, which within c e r ta in c o n s tr a in ts , r e a liz e t h e ir id e a l v e r s io n s . I t may be noted th at these c i r c u i t s were not in clu ded in Chapter 2. Consider the n o n -id e a l grounded FDNR and the flo a t in g R sim u lators, r e s p e c tiv e ly shown in F ig. 3 .6 (a ) and ( b ) . The n on -id ea l grounded FDNR b a s ic a lly r e a liz e s the d r iv in g -p o in t impedance fu n ction s f B 1 B Z (s ) = — p— = ----- + ^ • s^C sC s^C (3 .1 4 ) This c i r c u i t r e a liz e s a s e r ie s com bination o f an FDNR and a c a p a c ito r . However, in the frequency range, w << B, the c i r c u i t r e a liz e s an e f f e c t i v e ideal-FDNR with 109 . I. 4- ^ ■_ 1 o- 10 " Z (s)- L '1 a T (a) 1o 1 o- ■ - A A /-------- -o * T (b) F i g . 3*6 Tm O A - C c i r c u i t s r eal i zi ng (a) i d e a l graunded F D N R (b) i deal f l oat i ng R ; in the f r e q u e n c y r a n g e « « B - ■^2 110 (3 .1 5 ) S im ila rly , a n a ly sis o f the flo a tin g -R sim ulator o f Fig. 3 .6 (b ) r e a liz e s the adm ittance-param eter m atrix : (s-^B) - B ( 3 . 16 ) Y = C (s+B) -B In the frequency range, w << B, the m atrix reduces to 1 -1 Y = CB (3 .1 7 ) -1 1 and r e a liz e s an e f f e c t i v e i d e a l - f l o a t i n g r e s is t o r with R = 1/CB. These low-component sim ulators are subsequently used in the a ctiv e-C r e a liz a t io n o f the la d d er, employing the trans­ formed c i r c u i t o f Fig. 3 .5 ( b ) . la d d er i s given in Fig. 3 .7 . The fin a l OA-C v e rs io n o f the I t may be noted th at the c i r c u i t uses a low component count o f only c a p a cito rs f o r even n, and o f to r s f o r odd n la d d e rs . n OAs and ( ^ n * 2 ) ( 3 n - l ) Oas and |-(n+l) ca p a ci­ The on ly r e s t r i c t i o n w ith the c i r c u i t i s to use i t in the lim ite d frequency range, w << B, which r e s t r i c t s i t s use to the AF-range. This s e c tio n demonstrates how component saving may be achieved at the c o s t o f some co n stra in ts imposed on the working frequency-range o f f i l t e r s . Ill Q Q O uO ro <u T3 -O ro if) in ro a. I o c 0 ‘in (U > O 1 < o CO CD 112 5 .2 L~Based Approach to OA-C Synthesis In the inductance (L )-b a sed approach to OA-C synthe­ s i s , the R and L elements o f the p rototyp e f i l t e r d escrib in g the tr a n s fe r fu n ction are required t o be sim ulated in the OA-C form and then su b stitu te d f o r the corresponding passive components. This r e s u lts in the OA-C v e r s io n o f the f i l t e r . As has been seen in the preceding s e c tio n , the use o f id e a l component sim ulators norm ally requ ires a la rg e count o f a c tiv e and p assive components. In a d d itio n , the use o f such compo­ nent sim u lators, p a r tic u la r ly the flo a t in g ones, may re s u lt in u ndesirable s e n s it iv it y p r o p e r tie s and the c i r c u i t having unaccountable p a r a s ite s , when the matching con d ition s are not s t r i c t l y s a t i s f ie d . A lte r n a tiv e ly , the use o f n on -id ea l component sim ularors f o r rep la cin g a s e t o f components is more a t t r a c t iv e from the co n sid e ra tio n s mentioned above. In th is s e c tio n , the second approach i s th e re fo re used in the OA-C r e a liz a tio n s o f ( i ) standard secon d -ord er f i l t e r s and ( i i ) h ig h er-ord er la d d ers. In S ection 3 .2 ,1 , a standard secon d -ord er OA-C high-pass (HP) f i l t e r i s re a liz e d by employing n on -id ea l L -sim iilator. The technique i s then ap plied in S ection 5 . 2 . 2 . in the r e a liz a t io n o f h ig h er-ord er OA-C la d d ers. 113 5 .2 .1 R ea liza tion o f se con d -ord er f i l t e r with n on -ideal L -sim ulator This s e c tio n demonstrates the r e a liz a tio n o f a second ord er high-pass f i l t e r from i t s p rototy p e, by using a non­ id e a l OA-C grounded L -sim u la tor. i s shown in Fig. 3 .8 ( a ) . RLC-tuned c i r c u i t i s For Its the The re s u ltin g c i r c u i t i s shown a n a ly sis g iv es the tr a n s fe r fu n ction V. V ^ > obtain in g i t s OA-C v e rs io n , d ir e c t ly sim ulated and replaced by the OA-C sim ulator o f Fig. 2 ,3 in Fig. 3 .8 (b ). The p a ssiv e HP-prototype H, s^ : N(s) = ^ where C hp C^l • C^ o C, 1 /2 1 2 and C„ B, C Q » pS [ ^ . '-1 “l ^2 C. 1 /2 . =2- J . ''34 (3 .1 9 ) This f i l t e r enjoys the m u ltifu n ctio n a l c a p a b ilit y and p rovides a d d itio n a l standard secon d -ord er BP and LP responses at the outputs o f OA^ and 0A2 » r e s p e c t iv e ly . The BP and LP gains are given by : ^ (3.20) and the other parameters o f in t e r e s t , such as, pole-w^ and pole-U remain unaltered. The f i l t e r parameters have the 114 ■TO B E SIMULATED 1 R i L (a ) r REQUIRED O A - C Co sim u la to r ----------- 1 1 I ^^3 Vhp _ Vbp Vlp __L zr (b) Prototype + tC 2 Vi Fig. 3-8 (a) ^ h i g h - p a s s filter. (b) O A - C version of the prototype HPF. 115 a t t r a c t iv e fe a tu re o f being in terms o f C -r a tio s . The s e n s i­ t i v i t y fig u r e s , f o r v a riou s f i l t e r parameters w ith resp ect to the a c tiv e and p assive elem ents, are analysed and the fin a l r e s iilts are in clu ded in Table 3.^ . Expressions in the ta b le show that the f i l t e r enjoys a t t r a c t iv e s e n s it iv it y p ro p e rtie s ; again a l l the s e n s i t i v i t i e s being l e s s than or equal to u nity in magnitude. The c i r c u i t a ls o enjoys independent e le c t r o n ic tuning o f p o le -fre q u e n cy . However, independent p a ssiv e co n tro l o f f i l t e r parameters i s not p resen t in th is ca se. In order to p rovid e dc b ia sin g and s t a b i l i t y , high­ valued r e s is t o r s are requ ired a cross the ca p a cito rs C2 and C^. 3 .2 .2 R e a liza tio n o f h ig h e r-o rd e r lad d ers Inductance-based sim ulators o f Table 2 .1 may a ls o be employed in the r e a liz a t io n o f v a riou s la d d e r stru ctu res in the OA-C form, th is technique w ill p a r t ic u la r ly be su ita b le fo r t o p o lo g ie s , which have inductance in the shunt-arms. Two examples are f i r s t con sid ered f o r the r e a liz a t io n o f OA-C ladd ers having the Ls in th e shunt-arms. ' Consider the n -p o le band-pass f i l t e r o f Fig. 3 .9 (a ) 86 I t s r e a liz a t io n in the OA-C form requ ires the replacement o f the grounded LC- arms by t h e ir sim ulated v e r s io n s . The grounded LC-sim ulator o f Fig. 2 .2 i s used in the r e a liz a tio n o f a ctiv e-C v e rsio n o f the n -p o le BP f i l t e r , as shown in Fig. 5 . 9 ( b ) . The r e a liz e d f i l t e r only requ ires 2n OAs and (^ n - 1 ) ca p a c ito r s . 115 H c\j \ H I CM rH r-H o vV _v + CNJ •> CM r—( O o o (—1 O u CM I—1 o CM U rHiCvJ (\J rvj N/ CM CM H CJ CM + O Hu O ^IfVJ fT\ U U rHifVJ I I cv C\J CM r\y — < H V (V rH CM f— 1 rH o CN J o ^ o o HiCM o o Hicvj I -d- fo o u H|C\J fV CM H N/ CM -d’ iH r— ( o O -J" o u o u h Ioj HlfM <r o O r-flCvl I I P. H af I a H N/ #> CM CM r-l o * rH O u rH u U CM cm ' rH O H O CM u o 4-> OQ 1s CO f\J CQ O o CM U u o 117 — TO B E sim u l a t e d Cn-1 ------- II— -------------------- o + Q { -o- n (a) F i g . 3-9 (a) Prototype n - p o l e B P - f i l t e r (b) O A - C versfon of n - p o l e B P filter. ■o- 118 As a second example, c o n s id e r the n -th order lin e a r phase BP f i l t e r , shown in Fig. 5 .1 0 (a ) 86 . I t s active-C r e a liz a t io n requ ires the sim u lation and corresponding rep la­ cement o f the p assive RLC shunt-arms. The grounded tuned c i r c u i t sim ulator o f Fig. 2 .3 i s employed f o r the purpose. The fin a l r e a liz a tio n i s shown in Fig. 3 .1 0 (b ), where an n -th order f i l t e r requ ires 2n OAs and (5 n -l) ca p a cito rs in i t s r e a liz a t io n . The inductance-based sim u lation technique may a ls o be employed a t t r a c t iv e ly , w ith c e r ta in c o n s tr a in ts , in r e a liz in g s tru ctu re s, which have Ls in the flo a t in g arms, provided m u ltip le component replacem ent i s p o s s ib le . To demonsxrate t h is , an n -th order Cauer-Chebyshev (CC) low -pass ladder o f Fig. 3 .1 1 (a ) i s con sid ered . I t s r e a liz a t io n in the Oa-C form req u ires the sim ulation and subsequent replacement o f the flo a t in g LC-tuned c i r c u i t . The flo a t in g LC-sim ulator o f Fig. 2 .2 5 (b ) already stu d ied in S e ctio n Job. 2.5 i s used f o r the The resu ltin g Oa -C v e r s io n o f the CC low -pass la d d er i s shown in Fig. 3.11(t»). 2 ( n - l ) OAs and This c i r c u i t uses a low count o f only ( 3 n - l) c a p a c it o r s f o r an n -th order la d d er. The b a sic lim it a tio n o f the c i r c u i t i s in i t s use over a r e s t r ic t e d frequency range o f op era tion , w << B, as discu ssed in S ection 3 .1 .3 ^ . Note ; In the Oa-C la d d er r e a liz a t io n s , high-valued r e s is to r s may a p p rop ria tely be used f o r providin g dc bias to the OAs. ■ 119 TO BE s i m u l a t e d 1 I n (a) Fig-3.10 (a) Prototype (b) O A - C o r d e r l i near p h a s e B P f i l t e r v e r s i o n of l i n e a r - p h a s e B P filter. 120 L2 HI— ^ >-n-1 t^n-1 f ^3 T^n (a) r> -^ < 3 =rCi a n -1 (b) F i g . 3-11 (a) (CO order Ca a e r Chebyshev type l o w - p a s s filter. (b) OA — C v e r s i o n of C C - t y p e L P filter. n 121 3.3 FDNCAP-Based Approach t o OA-C syn th esis In th is s e c tio n , a new approach to d ir e c t form syn­ t h e s is , c a lle d the FDNCAP-Based Approach, i s con sidered f o r the r e a liz a tio n o f secon d -ord er and h ig h er-ord er f i l t e r s in the OA-C form. In th is tech n iqu e, the p a ssive FlLC-prototype network d e scrib in g the tr a n s fe r fu n ction i s f i r s t transformed in t o an equ ivalent network by s ca lin g each admittance o f the o r ig in a l network (N) by the square o f the com plex-variable ( s ^ ) , as shown in Fig. 3 .1 2 . The re s u ltin g network (N*) i s thus obtained by con vertin g a r e s is t o r to an FDNR, an induet o r to a c a p a c ito r and a c a p a c it o r to a frequency dependent n egative ca p a citan ce (FDNCAP) element^^. The FDNCAP has impedance o f the form, 1/s^ F , where F i s parameter o f the element having the u nits o f ( farad) ^. ( ohm) ^ (f ^ . il^ ). Thus, the RLC-network (N) i s cen verted to a DCF-network (N^) and sy n th esis problem becomes th a t o f the r e a liz a t io n o f FDNRs and FDNCAPs in the OA-C form. As w i l l be seen in the subsequent s e c t io n , the technique w i l l prove out t o be very a t t r a c t iv e in the r e a liz a t io n o f OA-C v e rsion s o f those s tru ctu re s which have ex cessiv e number o f inductances in the s e r ie s , as w ell as, the shunt-arms. In t h is method a ls o , e ith e r id e a l or n o n -id e a l FDNCAP c ir c u it s may be used in the r e a liz a t io n o f OA-C c i r c u i t s . However, as has been pointed out e a r l ie r , the use o f n on -id eal FDNCa Psim ulators f o r m ultiple-com ponent replacem ent w ill prove out 122 NETWORK-N network-N D o-- VV'- -- o ,V r r SC V l VsL J sc Fig. 3-12 -o i D =• V -o J C —V -o » F = C V R L C e l e m e n t s t r a n s f o r m a t i o n to D C F version by s ^ - s c a l i n g on each a d m i t t a n c e s . 123 t o be much more a t t r a c t iv e . In S ection 3 .3 .1 , the technique i s used, in the r e a liz a t io n o f secon d -ord er low p a s s - f i l t e r and in S ection 3 .3 .2 f o r the r e a liz a t io n o f h igh er-ord er la d d ers. 3 .3 .1 R ea liza tion o f secon d -ord er s e c tio n with n on -id eal FDNCAP-simulators In th is s e c tio n the r e a liz a t io n o f standard second- ord er low -pass f i l t e r i s d escrib ed using an FDNCAP-based sim u lator. The p rototyp e secon d -ord er LP f i l t e r i s shown in Fig. 3 .1 3 (a ). I t s s ca lin g by s^ leads t o the FDNCAP-based v e rs io n o f the c i r c u i t , shown in Fig. 3 .1 3 (b ). I t may be n o tic e d that the transformed network may com pletely be simu­ la te d by the FDNCAP-based. sim u lator o f Fig. 2.15 by taking the responses a t the output o f Oa.^. In the re s u ltin g c i r c u i t , c a p a c it o r C i s s p l i t as a p a ra lle l-co m b in a tio n o f i.e ., C = and C^, fin a l OA-C r e a liz a tio n i s given in Fig. 3 .1 3 ( c ) . A nalysis o f the c i r c u i t y ie ld s the tra n s fe r fu n ction : (s ) = V, Vi H, w^ N(s) ^ D(s) ( 5. 2 1 ) where 1 /2 h p " % - Cb^.B2 . . ^12 and ] ^3 ^ 124 ( b) (c ) F i g . 3.13 (a) P r o t o t y p e R L C L o w _ p a s s f i l t e r ( L P F ) . (b) F D N C A P - v e r s i o n (c) O A - C - v e r s i o n of L P F . of LPF. 125 Q = [ . ®2 '^12 . (3 .2 2 ) '^3 Once again, the f i l t e r parameters have the a tt r a c tiv e feature o f being in terms o f C -r a tio s and the p o le -fre q u e n cy being independently tunable w ith the b ia s -c o n t r o l. However, inde­ pendent p assive c o n tr o l o f f i l t e r parameters i s n ot presen t. AS the f i l t e r ’ s output is available at the output of OA, addi­ tional buffer is not needed, which is generally required in the circu its obtained by the direct form approach. The s e n s it iv it y fig u r e s f o r variou s f i l t e r parameters, w ith resp ect to a c tiv e and p a ssiv e elem ents, are analysed and th e f in a l r e s u lts are in clu ded in Table 3 .5 . The ta b le shows th at f i l t e r enjoys a t t r a c t iv e s e n s it iv it y p r o p e r tie s , being le s s than or equal t o unity in magnitude. In order t o p rovide dc b ia sin g t o the OAs, high-valued r e s is t o r s are required a cross 3 .3 .2 and C^. R e a liza tio n o f h ig h e r-o rd e r ladders Sometimes in the la d d e r r e a liz a t io n s , the L-based and the FDNR-based techniques do not prove out to be a t t r a c t iv e , p a r t ic u la r ly when the stru ctu res con tain inductances in both the s e r ie s and the shunt-arms. In such ca s e s , the FDNCAP- based approach g e n e ra lly lea d s t o a t t r a c t iv e r e a liz a t io n o f the c i r c u i t in the OA-C form. In th is s e c t io n , we s h a ll co n sid e r two examples f o r the r e a liz a t io n o f the n -th order 126 Table 3.5 ; S e n s it iv it y fig u r e s f o r OA-C f i l t e r o f F ig .5 .1 3 (c ) 127 la d d e rs , based on the FDNCAP approach. The f i r s t example i s based on the use o f n o n -id e a l FDNCAP f o r the sim ulation o f a s e t o f components. T h erefore, i t re s u lts in a low component r e a liz a t io n w ith no matching and frequency r e s t r i c t i o n s . In the second r e a liz a t io n , an id e a l grounded FDNCAP i s required. However, rath er than using an id e a l grounded FDNCAP c i r c u i t w ith a la rg e component count and matching c o n d itio n s , the use o f n on -id ea l FDNCAP as an id e a l one w ithin c e r ta in frequency c o n s tr a in ts , i s suggested. This once again r e s u lts in the op tim iza tion o f components and elem ination o f c r i t i c a l matching c o n d itio n s . In the f i r s t ca se , co n s id e r the e l l i p t i c la d d er f i l t e r o f Fig. 3 . l H a ) . F ig. 3 . 1 ^ b ) . I t s transform ed FDNCAP-version i s shown in For the r e a liz a t io n o f the la d d er in the OA-C form, two types o f n o n -id e a l grounded simulators^ I and I I , are requ ired. The f i r s t type o f sim ulators ( l ) are used f o r the replacement o f the s e r ie s C;F shunt-anns, except f o r l a s t arm. The FDNCAP- based sim u lator r e a liz in g grounded C:F- fu n ction i s given in Fig. 2 .1 3 . This c i r c u i t i s used f o r the replacement o f the shunt-arms in the transformed c i r c u i t . Rather than sim ulating an id e a l FDNR f o r the l a s t shunt-anm, the com plete L -s e c tio n c o n s is tin g o f C and D i s sim ulated by the n on -id ea l FDNR-simulator ( I I ) o f Fig. 2 .9 . The fin a l OA-C v ersion o f the la d d er i s shown in Fig. 2 .1 4 ( c ) . I t is seen th at the r e a liz a t io n does not requ ire any matching or frequency c o n s tr a in ts , and moreover, i t uses a low count o f 128 (a) TYPE-II TYPE-I OA-C Cl sim ulato r sim u la t o r Cn-2 C3 Cn - i) - C2 ± I I — *^n-l I ____ I L ± c T •s " (b) (c) Fi g . 3- U (a) Prototype ell iptic l a d d e r filter. (b) F D N C A P - v e r s i o n of e l l i p t i c l a d d e r filter. (c) O A - C v e r s i o n of e l l i p t i c l a d d e r filter. 129 only n OAs and 2n Cs. In the second example, the low -pass LC la d d er o f Fig. 3 .1 5 (a ) i s con sid ered . I t s transformed FDNCAP-version i s shown in Fig. 3 .1 5 (b ). Now t o r e a liz e the c i r c u i t in the OA-C form, the grounded id e a l FDNCaPs are requ ired to be rep la ced by the corresponding OA-C s iu u la to r s . The n on -id eal FDNCaP o f Fig. 2.13 in the frequency range, w << B, a cts as an id e a l grounded FDNCAP-Simulator, w ith the impedance func­ t io n , Z (s ) = l/s ^ F and F = C^/B^B^. T herefore, f o r the r e a l i ­ z a tio n o f the la d d er in the GA-C form, t h is FDNCAP-simulator i s used f o r the replacem ent o f a l l the shunt-arms, except f o r the l a s t arm, which i s replaced by the FDNR-simulator o f Fig. 3 .6 ( a ) . Fig. 3 .1 5 ( c ) . The OA-C v e rs io n o f the Ladder i s shown in I t i s seen th a t the OA-C la d d er uses only (n+1) OAs and (2 n + l) Cs f o r e v e n -n , w hile n OAs-and 2n Cs f o r odd-n f i l t e r s . Matching co n s tra in ts are n ot p resen t in th is r e a liz a t io n , but frequency r e s t r i c t i o n , w << B, i s present in the c i r c u i t . 3. ^ S3q)erimental Results In th is s e c tio n es^Jerimental v e r i f i c a t i o n s o f the three OA-C d ir e c t form syn th esis tech n iqu es, stu died in the preceding s e c tio n s , have been in clu d ed . This i s done by e:Q)erimentally v e r ify in g , one c i r c u i t each, r e a liz e d through the d if fe r e n t d ir e c t forro syn th esis approaches. Since the f i l t e r s r e a liz e d in t h is ch apter are based on the component sim ulators studied 130 L3 -n»- f -----HfV ‘-n -l 00— HiV Cn== >R T i-e vc n TYPE-I \ r T Y P E - II SIMULATOR sim u la t o r ^3 ^n\ -II- - - - - - -^n-2 - - I-- pHI \ . ^n-1 -o • — 11- I 1 = F Fn= F n -l^ xD _ J > --o •- -n-od d T i- even (b ) TY P E -I O A - C SIMULATO TYPE-ir sim u la t o r (C) Fig. 3*15 (a) P rot ot y pe s i ng l y ladder. terminated low-pass (b) F D N C A P - v e r s i o n of the L P - l a d d e r . (c) O A - C v e r s i o n of the L P - l a d d e r , 131 in Chapter 2, the experim ental re s u lts a ls o p rovide the v e r iX ic a tio n s on some o f the component sim ulators o f the previous ch apter. Lacking IG fa b r ic a tio n f a c i l i t i e s in the Department, the experim ental v e r i f i c a t i o n s were done on the d is c r e t e v e rsion s o f the c i r c u i t s . The reported re s u lts have been obtained without re s o rtin g to tuning. 3 .^ .1 Design and te s t in g o f B S - f i l t e r (FDNR-Based Approach) The d is c r e t e v e rs io n o f the secon d -ord er band-elem ina- tio n f i l t e r (BEF) o f Fig. 3 .3 ( c ) , r e a liz e d through the FDNRbased technique, was e 35)erim en ta lly v e r i f i e d in the la b ora tory. I n it ia lly ^ t h e BE f i l t e r was designed f o r a ce n tre frequency ( f ^ ) o f 32.9 KHz and q u a lity f a c t o r o f 10.5 w ith OAs o f 741type and d is c r e t e Cs o f 1 p ercen t to le ra n ce . The use o f the design equation ( 3 . 6 ) , along w ith the measured value o f B o f each OA as 0.9 MHz at Vc c = + 9 v o l t s , y ie ld e d the fo llo w in g component values : = 65 nF, = 226 pF, = 99.5 pF, and = 74 nF. The designed c i r c u i t o f Fig. 3 .1 6 (a ) was te s te d w ith frequency response shown in Fig. 5 .1 6 (b ). The e x p e ri­ mental r e s u lts show good con form ity with design. The m u ltifu n ction a l c a p a b ilit y o f the f i l t e r was a ls o v e r i f i e d experim en tally. The previous f i l t e r , with the given s p e c ific a t io n s and d esign , a d d itio n a lly p rovides the band­ pass c h a r a c t e r is t ic s at the output o f OA^. The frequency response o f th is c i r c u i t , along w ith the designed and observed valu es o f f^ and Q, are in clu d ed in Fig. 3.17. The low -pass 132 u • ro ^ ro CJ) i r <£> PO 6^ « <u Li. 4-» o iZ ^ c o Ll I ■w CD ro c o E (U in UJ c o I CL ■a tn c o nj CD >N *o u <u c c OJ 13 V i O<u u a TO U3 o rn cn 153 3 » —N ^ ro i-L n v4— • o cr» (U {/) c o Q. O) Ql CD >% u (U D ain oj <u l_ •o (U c Q Li. 03 rn f Li_ response i s a ls o a v a ila b le a t the output term inal o f OA2 . The f i l t e r was then s li g h t l y redesigned t o y i e l d the B utterworth response at the output o f OA2 f o r previou s p ole frequency, f^ = 32-9 KHz. become : = 5 .3 nF and remaining u naltered. 3 .1 8 (a ). The m odified valu es o f ca p a cito rs = 290 pF ; the oth er C-values The low -pass f i l t e r i s shown in Fig. I t s experim ental frequency response, along with the th e o r e tic a l and observed valu es o f paramerers, are in c lu ­ ded in Fig. 3 .1 8 (b ). The agreement between tn eory and ejqjeriment i s obviou s. The v a r ia tio n o f p o le - f ^ and p ole-Q w ith b ia s v olta g e (V OQ) f o r the BEF o f Fig. 3 .1 6 (a ) was a ls o experim entally v e r ifie d . The r e s u lts are shovm in Fig. 3 .1 9 . I t demonstrates tne independent e le c t r o n ic tuning o f p o le -fre q u e n cy with b ia s v o lta g e over a con sid era b le range o f about 81 per ce n t, with Q p r a c t i c a ll y remaining u naltered. In a l l the ca se s, the experimental valu es were foxmd to be in c lo s e con form ity w ith the th eory. N ote, high­ valued r e s is t o r s , each o f 2.2 Mohm, were employed across the ca p a cito rs and C2 to p rovide the dc b ia s and s t a b i l i t y to the OAs. 3.^ .2 Design and te s tin g o f H P -filt e r (L.-Based Approach) The secon d -ord er high-pass f i l t e r o f Fig. 3 .3 (b ), r e a liz e d through the L-based tech n iqu e, was designed and - •, r -- 155 N X cc Ixl Q. ll. CL m X II u >— u cr o Ll_ Q_ UJ X II fN O < O u. O c o» rg N X <X) <M m o o u. U. Q . c in n« • O) ir» (T> II II n U O O PO no •cn U_ •r: nj U CO 'u (U \£ o in <u «/» O) £X o 4. ^ 0» ■D <U cn in C a iP oo ro m il C V 13 cr OJ Li. X] ' 136 Q 25 20 15 10 BIAS VOLTAGE (VOLTS) Fig.3-19 Variation of p o l e - f r e q u e n c y ( U a n d p o l e - Q w i t h b l a s - v o L t a g e (V^c) f or B E F of Fig.3-16(a). 137 experim entally v e r i f i e d by employing Bi-MOSFET input OAs, CA 3 1 ^ . The measured valu e o f B a t a bias o f + 9V was found to be 4.2 MHz. worth response a t a The H P -filt e r was designed f o r B u tterc u t - o f f frequency, f^ = 420 KHz. Since the op era tion al frequency was in hundred o f k ilo h e r tz range, i t was e s s e n tia l t o use the p r e d is to r tio n tech n iqu e, d iscu ssed in Chapter 4. The c i r c u i t design in corp ora tin g p r e d is to r tio n , along w ith ( 3 . 9 ) , y ie ld e d the fo llo w in g ca p a cito rs valu es ; = 0.97 nF, = 1.02 nF, = 71.2 nF. = ^.6 nF, = 0.99 nF, and This c i r c u i t i s shownin Fig. 3 .2 0 (a ). I t was experim entally v e r i f i e d and the frequency response i s given in Fig. 3 . 20 ( b ) . The r e s u lts once again e x h ib it good c o n fo r ­ mity w ith the th eory. 3 .4 .3 Desifoi and te s tin g o f L P - f i l t e r ( FDNCAP-Based Approach) The d is c r e t e v e rs io n o f the secon d -ord er low -pass f i l t e r o f F ig. 3 .1 3 ( c ) , r e a liz e d through the FDNCAP-based tech n iqu e, was designed and experim en tally v e r i f i e d employ­ ing Bi-MOSFET input OAs, CA 3140. at a bias o f + 15 V was 4.5 MHz. The measured value o f B The low -pass f i l t e r was designed f o r Butterworth response a t c u t - o f f frequency, f^ = 100 KHz. Using ( 3 .2 2 ) , the design y ie ld s the fo llo w in g capa­ c i t o r s values : C^ = 220 nF, C^ = 100.8 nF. = 112 pF, C^ = 3.2 nF, and The designed c i r c u i t i s given in Fig. 3 .2 1 (a ). The LPF was then eijqjerimentally v e r i f i e d and the frequency response i s shown in Fig. 3 .2 1 (b ). Once again, the re s u lts 138 _J < \— N z X UJ z lT> cr o u fsj Q _ X UJ Q r lO H UJ X z (S> o in rsi UJ nJ o OO rn o M — ‘uo o ° ro N X o o oo <u ±i M- — o o (D o o v3- o o (N o {/) OJ if) ^ CL O I Q - ^ 4J Ic *D (U C cn .— O OJ ^ O" C a u. no xj o fM ro >(l ’l l 139 DE SI GNED 9 8 -4 KHZ lOOKHz ^0 Vv( OBSERVED 1-0 0-8 C^s220nF 0-6 + Vl 0-2 -H h t— W T C2*112pF --------- c + IF - - r3*2nFIL - X \ Vo . C4,sl00*8nF .. (a) X 10 0-0 40 100 (b) Fig. 3-21 (a) (b) De s i g n e d 200 f (KHz) LPF of F i g . 3-13 (c)- Freq u en cy r e s p o n s e curve of LPF of Fig. 3.21 (a). lUO were found in c lo s e con form ity with the th eory. 3.5 ConcludlnR Remarks In th is ch ap ter, the d ir e c t form OA-C syn th esis has been con sidered in d e t a ils . ( i ) L-Based Approach, ( i i ) (iii) Three sjm th esis tech n iqu es, v i z . , FDNR-Based Approach, and ( i i i ) FDNCAP-Based Approach, have been d e scrib e d in d e t a il and c r i t i c a l l y stu d ied f o r the r e a liz a t io n o f secon d -ord er and h ig h er-or^ er networks in the OA-C form, s ta r tin g from the p assive RLC- p ro to ty p e . For the sim ulation o f components, id e a l and n o n -id e a l immittance sim iolators, already studied in Chapter 2, were employed. In a ll the th ree approaches, two d if fe r e n t types o f component sim ulators were requ ired. I t has been shown that use o f id e a l sim u la tors, p a r t ic u la r ly in the flo a tin g mode, lead s t o c i r c u i t r e a liz a t io n s with a number o f u ndesirable fe a tu re s . These are : ( i ) ‘ use o f la r g e count o f a c tiv e and p a ssive components, ( i i ) requirement o f c r i t i c a l component matching, and ( i i i ) in some c a s e s , a r e s t r ic t e d working range o f frequency. T herefore, in cases where id e a l component sim ulators are e s s e n tia l, i t has been suggested to use th e corresponding n on -id ea l sim ulators w ithin a frequency range, w << B, where the n on -id ea l c i r c u i t s can be assumed to p rovid e the behaviour o f id e a l sim u lators. The use o f such n on -id ea l sim u lators, w ith in the r e s t r ic t e d frequency range, p rovid es c i r c u i t r e a liz a tio n s w ith low component count and w ithout requiring 141 c r i t i c a l matching c o n s tr a in ts . An a lte r n a tiv e technique o f using n on -id ea l component sim u la tors, without frequency r e s t r ic t io n , proves out t o be s t i l l more a t t r a c t iv e . In th is method, the n on -id ea l compo­ nent sim ulators used f o r the replacement o f m u ltip le-elem en ts, e it h e r o f the p rototyp e or i t s transformed v e r s io n . Such r e a liz a t io n s use a low count o f a c tiv e and p a ssive components, and do not have matching and frequency r e s t r i c t i o n s . The s u i t a b i l i t y o f a p a r tic u la r OA-C d ir e c t form syn th esis approach, as i s the case w ith con ven tion al active-RC s y n th e s is , depends on the stru ctu re o f the b a sic p rototyp e network. For example, the L-based technique proves out t o be a t t r a c t iv e when stru ctu res have C (s) in the flo a tin g -a r m (s ) and L or LC or RLC-network(s) in the shunt arm (s). The FDNR based approach, g en era lly le a d s to the r e a liz a t io n s with a la r g e r component count. However, the use o f n on -id ea l component sim u la tors, w ithin the r e s t r ic t e d frequency range, sometimes r e a liz e s f i l t e r s w ith low component count. Struc­ tu res which req u ire, both flo a t in g and grounded in d u ctan ces, are g e n e ra lly s u ita b le to r e a liz e w ith the d ir e c t form FDNCAP approach. This i s p a r tic u la r ly so i f m ultiple-com ponent sim ulator i s employed. The re s u ltin g OA-C networks have low component count and a ls o do not s u ffe r from matching cons­ tr a in ts and frequency r e s t r ic t io n s . The d iscu ssed techniques provided a u sefu l method f o r 142 the r e a liz a t io n o f b iq u a d ra tic, as w ell as, h ig h er-ord er f i l t e r s , s u ita b le f o r m on olith ic implementation in the MOS tech n ology. The paramexers o f a l l the re a liz e d f i l t e r were found to be in terms o f ra tioed -C . c t iv e featu re f o r MOS im plem entation. This i s very a ttra ­ Tne re a liz e d second order f i l t e r s were g e n e ra lly found to p rov id e a d d ition a l standard responses, besid es the one ibr wnich i t was re a liz e d . The p o le -fre q u e n cie s o f a l l secon d -ord er f i l t e r s were found to be tunable e l e c t r o n ic a lly w ith the b ia s -v o lta g e co n tro l without a ffe c t in g the u o f the f i l t e r s . A lso in many ca ses, the c i r c u i t in corp ora tes independent p assive c o n tr o l o f pole-Q . The s e n s it iv it y stu d ies o f the secon d -ord er f i l t e r s showed them to have a tt r a c tiv e s e n s i t i v i t y perform ance. F in a lly , the ejcperimental re s u lts on the f i l t e r s , r e a liz e d by the three syn th esis tech n iqu es, gave experim ental r e s u lts which were in good con form ity with the th eory, w ithout re s o rtin g to tuning.