Active RC networks utilizing the voltage follower by James Llafet Hogin A thesis submitted to the Graduate Faculty in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY in Electrical Engineering Montana State University © Copyright by James Llafet Hogin (1966) Abstract: The (RCVF) is defined as an active network whose passive portion consists of a three-port grounded RC structure and whose active portion consists of the three-terminal grounded voltage follower. The necessary and sufficient conditions on a voltage transfer function for its realization by the (RCVF) are developed. Synthesis methods for the passive portion of the (RCVF) are reviewed, developed, and discussed. A digital computer program is presented which demonstrates the feasibility of optimizing (RCVF) networks by the method of steepest descent. A discussion and analysis of practical voltage followers and the practical considerations involved with their use is presented. In addition, a useful three stage voltage follower is developed which has very high input impedance and a gain very close to unity. Illustrative examples and experimental verification are given. It is concluded that the (RCVF) is a highly useful and practical network. A C T I V E RC NETWORKS U T I L I Z I N G T H E VOLTAGE F O L L O W E R by JAMES LLAFET H O G I N A thesis submitted to the Graduate Fac u l t y in partial fulfillment of the requirements for the degree of D O CTOR OF PHILOSOPHY in Electrical Engine e r i n g Approved: H e a d „ Ma j o r Department D e a n 9 Graduate Division MON T A N A S T A T E UNI V E R S I T Y B o z e m a n 9 Mon t a n a M a r c h 9 1966 ACKNOWLEDGMENTS The author p a r t i c u l a r l y wishes to thank Prof e s s o r B 0 J c Bennett for his encouragement and helpful suggestions. is also grateful for the f inancial support provi d e d b y a Douglas Aircraft C o m p a n y scholarship. iii He TABLE OE CONTENTS Page Vita ' Acknowle d g m e n t s ii iii Table of Contents iv List of Tables vi List of Figures vii List of Symbols x Abstract I, xii Int r o duction I A0 The Active Ne t w o r k 2 B0 The Voltage .Follower . 4- C0 Comparison of Active and Passive Transfer F u n c t i o n Synthesis II0 5 Statement of Pro b l e m 11 IIIo N e c e s s a r y and Sufficient Conditions for the R e a l i z a t i o n of the Voltage Transfer F u n c t i o n T^ A0 IVo Surplus F a ctor U t i l i z a t i o n 21 (RCVF) R e a l i z a t i o n Techniques A0 26 A P ictorial R e p r e s e n t a t i o n of the Passive Synthesis P r o b l e m B0 13 , R e v i e w of P r e v i o u s l y Proposed Configurations iv 26 $0 Page Co A R e a l i z a t i o n Technique f o r the Biquadratic Transfer F u n c t i o n B0 Jl Ladder Ne t w o r k R e a l i z a t i o n for Low-pass and High-pass T r a nsfer F unctions E0 Fo Vo VIo J6 A N etwork Co n f i g u r a t i o n f o r Fourth-'order Band-pass T r a n s f e r Functions 47 Examples 51 Net w o r k O p t i m i z a t i o n 59 A, Opt i m i z a t i o n Parameters 59 B0 Computer P r o g r a m D e s c r i p t i o n 67 Co Example N etwork 85 P r a c t i c a l Considerations 90 A0 The Non-ideal Voltage F o l l o w e r 90 B0 Voltage F o l l o w e r Circuits 91 Co Experimental Results 96 VIIo Con c l u s i o n and Rec o m m e n d a t i o n s 107 Ao Con c l u s i o n 107 B0 Recommendations 107 I, 111 Appendix A p p e n d i x 11, O p t i m i z a t i o n C o m puter Program L i s t i n g Opti m i z a t i o n C o m p u t e r Program Input and Output Data for Example Problem 121 A p p e n d i x IIIo N umerical Computations 125 Literature Consulted IJO v LIST OF TABLES F r e q u e n c y - a t t e n u a t i o n t a b u lation of experimental and t h eoretical values for network of Figure 6.8. F r e q u e n c y - a t t e n u a t i o n t a b u lation of experimental and t h eoretical values f o r n etwork of Figure 6.9» F r e q u e n c y - a t t e n u a t i o n t a b u lation of experimental and theoretical valu e s for.n e t w o r k of Figure 6.10. LIST OB’ FIGURES 1.1 The active RC n e t w o r k u t i l i z i n g the v oltage f o l l o w ­ er. 2 . / I , = # s _ +2.8280+4 ^ ^ h out in s ^ +1.4 1 4 s +1 less net w o r k terminated h y r e s i s t o r s . 1.2 R e a l i z a t i o n of E 1.3 R e a l i z a t i o n of E ^ / E ^ ==0.2 a loss- s +2.8288+4 s^+1.414s+l P a n t e l l 1s method. 1.4 Simple r e a l i z a t i o n of E 1.5 R e a l i z a t i o n of . / I . = - +2?828s+4 ^ out in s2 +lo4l4s+1 2 X Sg + 2 °' 8' 2 8s+4 w i t h the s +1.414s+l (RCYP) 3.1 The "A" and 3.2 Interco n n e c t i o n of "A" and 3.3 The "A" and "B" two-port unbalanced n e t w o r k s . "B" networks. "B" n e t works connected to the Voltage follower. 4.1 A p o s sible current distribution. 4.2 ■Current d i s t r i b u t i o n of passive p o r t i o n of (RCYP) 2 apS +a^s+a n e t w o r k to realize T = --- g-------- . P 2 3 +Pi8+P 0 4.3 Net w o r k co n f i g u r a t i o n w i t h input current d i s t r i b u ­ t i o n of Figure 4.2. 4.4 Current d i s t r i b u t i o n for case of a-, / a s r n /r 4.5 Current d i s t r i b u t i o n for case of r ^ / r Q s a^/a^ 4.6 The basic ladder network. J- O J- O . . 4 .7 The basic decomposed ladd e r network. 4.8 Decomposed ladder n e t w o r k wit h shunt arm capacitors and series arm resistors. 4=9 Current d i s t r i b u t i o n for bas i c . l a d d e r ne t w o r k r e a l ­ ization of y - Q . 4.10 P ole-zero d i s tributions for to y ^ ^ \ 4.11 Three basic b u i l d i n g bloc k s for the f o u r t h-order band-pass t r a nsfer f u n c t i o n realization. 4.12 Current d i s t r i b u t i o n for network c o r responding to t r a nsfer f u n c t i o n of Eq.. (4.23)• 2 4.13 4.14 Net w o r k r e a l i z a t i o n for T = v % . ■ s^ +1»4 1 4 s +1 Net w o r k r e a l i z a t i o n for T ^ = I / sZ*"+2,6132s^+3<>4143s^ +2.6132s+l. 4.15 Net w o r k r e a l i z a t i o n f o r T^=2.172s^/s^+1.414s^+3s^ +1.4!4s + 1 0 '3.1. P l o w chart f o r n e t w o r k optimization c o m puter p r o ­ gram. 5.2 5 =3 Config u r a t i o n of example network. Optimized n e t w o r k for realiz a t i o n of T^l/s^" +2»6132s 5+3.4143s 2+2.6132s +1. 6=1 Current d i s t r i b u t i o n di a g r a m for n o n u n i t y voltage follower gai n k. 6.2 Voltage f o l l o w e r 6.3 Altered network. of Refe r e n c e 32. viii 6<>4- S y s t e m d i a g r a m f o r n e t works of Figures 6.2 and 6.3. 6,5 Three stage voltage f o l lower network. 606 6.7 Current d i s t r i b u t i o n f o r twin-T network. p Net w o r k for r e a l i z a t i o n of T ^ = O .ls/s +.ls+1. 6.8 (HCVF) for r e a l i z a t i o n of second-order band-pass t r a nsfer f u n c t i o n d e n o r m a l i zed to 30Q 9000 ohms and 150 cps. 6.9 (RCVF) for r e a l i z a t i o n of fourth-order low-pass t r a nsfer f u n c t i o n d enormalized to I m e g o h m and 85 cps. 6.10 (RCVF) for r e a l i z a t i o n of fourth-order band-pass t r a nsfer f u n c t i o n d enormalized to 500»000 ohms and 39»8 cps. ix LIST OF SYMBOLS EC A b b r e v i a t i o n for the t e r m "resistance-capacitance"., EOVF A b b r e v i a t i o n for the active network composed of resistors, capacitors, and a single grounded v o l t ­ age f o l l o w e r o E I i I y Ij Voltage p a r a m e t e r o Current p a r a m e t e r c Short-circuit admittance p a r a m e t e r s „ Voltage t r a n s f e r function. Complex f r e q u e n c y v a r i a b l e „ N nume r a t o r po l y n o m i a l of y ^ for i=j. Negative of n u m e r a t o r po l y n o m i a l for i / j. D(G) D e nominator p o l y n o m i a l of y^^. A(G) Nume r a t o r p o l y n o m i a l of T^, P(8) Denomi n a t o r p o l y n o m i a l of T^, B( G ) Polynomial equal to P ( s ) - A ( s ) , E(G) A u g m e n t i n g po l y n o m i a l such that P(s)+E(s) h a s - n e g a ­ tive real zeros. ai P1 ri ei Coefficients of A(s). Coefficients of P(s) „ Coefficients of E.(s ) .■ Coefficients of E(s). Coefficients of D ( s ). Admittance parameter, x Impedance parameter= R e s i stance parameter= Capacitance parameter* 7H 6± Zeros of admittance* Poles of admittance* Opt i m i z a t i o n parameter. S e n s i t i v i t y of T w i t h respect to k* Polynomial root s e n s i t i v i t y (p = any p o l y nomial r o o t )* Coefficients of T^. Capacitance d i s t r i b u t i o n parameter. Et Resistance d i s t r i b u t i o n parameter. Capacitance summ a t i o n parameter. yR Capacitance ratio parameter. Resistance ratio parameter. Opt i m i z a t i o n p a r a m e t e r weig h t i n g functions. 2(8) Sec o n d - o r d e r po l y n o m i a l with zeros lying on unit circle in the RHP* Angle zeros of F(s) mak e w i t h the p o s i t i v e real axis* B(s) Au g m e n t i n g po l y n o m i a l for F ( s ) . Zeros of B ( s ) » A Coefficients of B ( s )* F u n c t i o n of the coefficients of B ( s ) . xi AB S T R A C T The (RCVB) is defined as an active n e t w o r k whose p a s ­ sive p o r t i o n consists of a three-port grounded RC structure and whose active p o r t i o n consists of the three-terminal grounded voltage follower* The n e c e s s a r y and sufficient conditions on a voltage t r a n s f e r function for its r e a l i z a ­ tio n "by the (RCVF) are developed* Synthesis met h o d s for the p a s s i v e p o r t i o n of the (RCVF) are r e v i e w e d $ developed, and discussed* A digital c o m puter p r o g r a m is pres e n t e d which demonstrates the f e a s i b i l i t y of optimizing (RCVF) networks b y the me thod of steepest descent* A dis c u s s i o n and a n a l y ­ sis of p r a c t i c a l v oltage f ollowers and the p r a c t i c a l c o n ­ siderations involved w i t h th e i r use is presented* In a d d i ­ tion, a u s e f u l three stage vol t a g e follower is developed w h i c h has v e r y h i g h input impedance and a gain v e r y close to unity* Illustrative examples and experimental v e r i f i c a t i o n are given* It is concluded that the (RCVF) is a h i g h l y u s e ­ ful and p r actical network* xii I. INTRODUCTION A c t i v e RC networks are networks containing only r e s i s ­ tors g capacitors and active e l e m e n t s » These networks ax-e attractive at the lower frequencies where m a g netic elements are large, expensive and d ifficult or impossible to realize. In addition, the advent of mic r o c i r c u i t technology, where a stable inductance p a r a m e t e r is difficult to attain, has r e ­ sulted i n an increased interest in this type of network. The need for a p o w e r supp l y f o r the active elements is often u nimport a n t if, as f r e q u e n t l y happens, the n e t w o r k is a part of a system for w h i c h p o w e r is required regardless of the network. The purpose of this study is to develop synthesis p r o ­ cedures concerned w i t h a p a r t i c u l a r type of active RC n e t ­ work: an active RC n e t w o r k whose active element consists of a single ideal voltage follower. The ideal vol t a g e follower is defined as a 2-port net w o r k whose input admittance and output impedance are zero and whose ratio of output to in ­ put volt a g e is unity. The cathode, emitter, or source f o l ­ lower circuit m a y be considered a practical approx i m a t i o n to the ideal voltage follower. In addit i o n to de v e lo p i n g synthesis procedures, this study is concerned w i t h the opti m i z a t i o n of the synthesized ™2™ networks b y means of the method of steepest descent as ap ­ plied b y the digital C o m p u t e r 0 F i n a l l y 9 the study, is c o n ­ cerned wit h the p r a c t i c a l d e m o n s t r a t i o n of the synthesis methods and w i t h the p r o b l e m of obtaining a vol t a g e f o l l o w ­ er w h i c h clo s e l y a p p r o x i m a t e s .the I d e a l 0 A« THE ACTIVE HETWORK The active RC net w o r k c o n f i g u r a t i o n of c o n c e r n in this r e s e a r c h is shown in Figure I 0I 0 The passive p o r t i o n of. the n e t wo r k consists of a 3-port grounded RC structure. The active p o r t i o n consists of the 3™terminal grounded voltage f o l l o w e r defined above. F o r convenience, this type of ac ­ tive n e t w o r k will be referred to as the ( R CVF)0 I A n important c o n s i d e r a t i o n in the d e s i g n of active n e t ­ works is the sensi t i v i t y of the desired net w o r k function to variations of the active element p o r t i o n of the active n e t ­ work. This co n s i d e r a t i o n is in addition to that of the s e n ­ siti v i t y associated w i t h va r i a t i o n s of the p a s s i v e n etwork elements. It should be pointed out, however, that from a pra c t i c a l po i n t of view, the g a i n of the voltage follower proposed as the active element in this study c a n be made more stable t h a n alternate types of active devices which r e ­ quire a gai n other t h a n unity. This observation is based RG P A S S I V E NETWORK I I VM P A S S I V E POR T I O N OF NETWORK Figure 1.1. ACTI V E PORTION OF NETWORK The active RC network u t i l i z i n g the voltage follower (RCVF) -4on the fact that the voltage f o l l o w e r is c o n c e r n e d ■wit h the direct c o m p arison of input vol t a g e and output voltage r a ther t h a n wit h their indirect comparison. Thus 9 a direct c o m p a r ­ is o n of active networks on the basis of the magn i t u d e of the sensit i v i t y f u n c t i o n associated w i t h the active element could be misleading. It is more informative to specify network f u n c t i o n variations caused b y expected variations in the a c t ­ ive device. In conclusion, the active element p r o p o s e d for this stu d y m a y be m ade h i g h l y stable and at least p a r t i a l l y compensate for the fact that the networks are quite sensi­ tive to changes in the active element. A n additional r e a s o n f o r considering the voltage f o l l o w ­ er as an active device is that, in its elementary f o r m s , it is an extremely simple and w i d e l y used device. Bo THE VOLTAGE TRANSFER FUNCTION The development of the v o l t a g e transfer f u n c t i o n for the (RCVF) proceeds as follows. Referring to the passive p o r t i o n of the net w o r k shown in Figure 1.1 and observing the conventions shown, the first of the short-circuit admittance equations is F or the p a r t i c u l a r c o n f i g u r a t i o n shown in Figure 1.1, I^=O -5“ and E^=E^o E q u a t i o n (1.1). t h e n "becomes yI1E1+712E2 +yl3E1=0 (1-2) and the voltage t r a n s f e r f u n c t i o n T ip _E ont_E l._ m a y "be solved as -y 12 V Eiii E2 yi:L+y15 (1.3) Since the poles of the short circuit admittance p a r a ­ meters .are identical (provided c o m m o n factors have not "been cancelled), we m a y c o n v e n i e n t l y define for i = j (1.4) ?ij=< " N i,1 I D, . for i/j where the N 0s' and D are polynomials, and write the voltage transfer f u n ction as ,m V E , N 12 .„ O U t . ■ n II-n I3 „ A(s) "■ F < s ; . (1.5) This m a n n e r of d e f ining the y^j allows us to consider all of the coefficients of the as h a v i n g the same sign (which we a r bitrarily choose positive). Go COM P A R I S O N OE A C T I V E A N D P A S S I V E T R A N S F E R FUNCTION. SYNTHESIS The m a i n difference b e t w e e n active and pas s i v e transfer f u n c t i o n synthesis is t h a t p a s s i v e synthesis is g enerally -6concerned w i t h the direct r e a l i z a t i o n of a g i v e n transfer quan t i t y "by means of one of the transfer i m m i t t a n c e s „ Ac­ tive synthesis is g e n e r a l l y concerned w ith the indirect r e ­ aliz a t i o n of the same t r a n s f e r q u a ntity b y means of the sum or difference of two t r a nsfer immittances =, The active ele­ ments of the active n e t w or k serve as sum or difference t a k ­ ing d e v i c e s o As an e x a m p l e 9 let us c o n sider a given t r a n s f e r func- , t i o n Tv and examine some of the ways this f u n c t i o n m a y be realized w i t h passive circuitry. T h e n it will be shown h o w this same f u n c t i o n m a y be r e a lized w i t h the '(EClfB1) active network. Con sider the t r a n s f e r f u n c t i o n T = K where K is an a rbitrary posit i v e constant. s + 2 o8 2 8 s »4 s ^ +I.. 414s +1 This function m a y be realized b y a passive n e t w o r k since its poles lie in the left-half of the com p l e x f r e q u e n c y plane (LHP). The f u n c t i o n is a m i n i m u m phase f u n c t i o n since its zeros also lie in the LHP. G uillemin (25) shows that a m i n i m u m phase tran s f e r f u n c t i o n cannot be realized b y a lossless network terminated in a single resistor. He shows $ h o w e v e r 9 a r e a l ­ i z a tion co n s i s t i n g of a lossless network terminated b y more th a n one resistor. A p p l y i n g this method to the present case results in the n e t w o r k of Figure 1.2. (See A p p e n d i x III-A -7O530 H 177 H 707 H 707 H R e a l i z a t i o n of E ,/I. =j4(s^+2o828s+4/s^+lo414s + I) w i t h a lossless ne t w o r k terminated b y resistors® 106 H yTlfTv Pigure 1.3<> R e a l i z a t i o n of E outZ E ^ n = O ®2 (s2 +2®8 2 8 s + 4 / s 2 + l®414s+l) b y P a n t e l l 1s method ® -8f or n ume r i c a l c o m p u t a t i o n s ) , The chief disadvantage of this realization, besides the large n u m b e r of elements, is that the input and output do not share a common ground. Other p o s sibilities for the realization, of T^ include its re a l i z a t i o n as a driving po i n t impedance or admittance (T^ is p o s itive real), its r e a l i z a t i o n in the f o r m of a c o n ­ stant resistance ladder (29), or b y P a n t e l l 1S meth o d T v (41). has b e e n realized w i t h the r esultant net w o r k shown in Figure 1.5 (see A p p e n d i x III-B for numerical computations). Note that P a n t e l l 1s method has the advantage of a driving and t erm i n a t i n g resistor. Finally, about t e r m i n a t i n g resistors, if one does not care etc. and a simple n e t w o r k is sought w h i c h realizes Tv , the c o n figuration of Figure 1.4 is among the simplest available. This n etwork was not syn­ thesized b y any rigorous method, but b y the applic a t i o n of heuristic means. To realize Tv b y means of the ( R CVF), we f o l l o w m e t h ­ ods to be pres e n t e d later in this thesis to develop the n e t ­ w o r k shown in Figure I . 5. Note that the n u m b e r of passive elements required iq less t h a n the number required in the m a j o r i t y of thq passive networks presented above. Also note that the same n u m b e r of reactive and only one additional r e ­ sistive element is required b y the (RCVF) r e a l i z a t i o n t h a n "by the simplest of the p a s s i v e networks -10- 4.248H Figure lo4„ Simple r e a l i z a t i o n of a 'out s £+2^828 s +4 o ‘in s 2+1.414s +1° 4F. 547F, I «64F c .1390 ----AA/'---__ ^ < :.646Q < VO L T A G E FOLLOWER (RCVF). jOUt -O O- Figure l.$o -q R e a l i z a t i o n of 3^+2.828 8 +4 wit h the s^+1.414s+l II. S TATEMENT OE PRO B L E M A t t e n t i o n will n o w Be given to a statement of the p r o b ­ lem for this research. The basic p r o b l e m of d e v e loping s y n ­ thesis procedures for the (RCYE) is b r o k e n into the f o l l o w ­ ing five' parts I. Determine the n e c e s s a r y and sufficient conditions on a voltage t r a n s f e r f u n c t i o n T =A(s)/P(s) such that it m a y be realized in the configu r a t i o n of the (RCYE) as shown in Figure I.I. 2p G-iyen the desired vol t a g e transfer f u n c t i o n T ^ = A ( s )/ P(s) whose n u m e r a t o r and denominator polynomials meet the n e c e s s a r y and sufficient conditions s pec­ ified above, determine the c o r responding network functions of the p a s s i v e RC po r t i o n of the 3P ( R CVE). Realize the determined n e t w o r k functions in the for m of a f o u r-terminal p assive RC n e t w o r k such that w h e n the n e t w o r k is used in c o n j u n c t i o n w i t h the voltage f o l l o w e r as in Figure 1.1, the overall n e t w o r k p ossesses the v o l t a g e transfer function T^= A(8)/P(s). 4-» Take advantage of the digital computer topological methods of circuit analysis p r e s e n t l y available and the method of steepest descent in order to i m ­ prove and optimize the realized networks. 5P Demonstrate the p r a c t i c a l i t y of the above items b y -12W i l d i n g and tes t i n g some example n e t w o r k s „ Item 2 above is complicated b y the fact that if there i s 'one w a y t h e n in general there are an infinite n u mber of ways that Ty m a y be b r o k e n up into the ne t w o r k functions of a p a s siv e EG f o u r-terminal network. Thus, we are faced wit h the p o s s i b i l i t y of s electing the one set of n e t w o r k f u n c ­ tions w h i c h optimizes some attribute or combi n a t i o n of a t ­ tributes of the network. S u c h attributes include m i n i m i z i n g the spread of element values, m i n i m i z i n g the sensitivity of to va r i ations in circuit values, etc. A significant p o r ­ t i o n of this r e s e a r c h is devoted to this opt i m i z a t i o n p r o ­ blem III. N E C E S S A R Y A N B S U R E I CIENT CONDITIONS E O R T H E R E A L ­ I Z A T I O N OE T H E V O L T A G E T R A N S F E R F U N C T I O N Tyo Considering Tv as the r a t i o .of polynomials A(s) to P (s ) $ this section will "be concerned with the e s tablish­ ment •of the n e c e s s a r y and sufficient conditions on A(s) and. P(s) such that a desired stable voltage t r a n s f e r f u n c t i o n Tv m a y be realized with the (RCVE) configuration. Since we are onl y interested in stable t r a n s f e r functions, it follows i m m e d i a t e l y that the zeros of P(s) are confined to the lefthal f of the com p l e x f r e q u e n c y pl a n e wit h the p o s sible excep­ t i o n of a single zero at the origin. Next we call on the E i a l k o w c ondition (29) which states F or an u n b a lanced two-port w i t h no mutual inductance the n u m e r a t o r coefficients of -Ypgs ^ l l 9 an^ Y g 2 are all n o n n e g a t i v e , the coefficients of -y^g "being no greater t h a n the c o r r e s p o n d i n g ones in y ^ or J22° This c o n d i t i o n can obvi o u s l y b e extended to the threeport unb a l a n c e d p assive RC n e t w o r k of Figure 1.1 w ith the res u l t i n g statement t h a t : F o r an unbalanced RC three-port with no m u t u a l i n d u c t ­ ance the n u m e r a t o r coefficients of - y ^ s “Y i g 9' an^ Y n are all nonnegative, the coefficients of - y ^ be i n g no gre a t e r t h a n the c o r r e s p o n d i n g coefficients in (y-^+y-^) -14This latter cond i t i o n follows f r o m the fact that if we f orm a two-port n e t w o r k f rom the three-port p assive n e t w o r k of f igure I 0I b y joining terminals 2 and 3 together, t hen the short-circuit t r a n s f e r admittance of the two-port n etwork is the sum of the short-circuit admittance p a r a meters y^^) of the three-port network* Cy^+ The second p art of the f i a l k o w c o n d i t i o n for u n b a lanced two-ports m a y t h e n be r e ­ stated as "the coefficients b e i n g no greater t h a n the corres p o n d i n g ones in y ^ " » If y-^ is added to each side of the inequality, the p r e v i o u s l y stated cond i t i o n is verified* A n o t h e r w a y of e x p r essing this cond i t i o n is that the coefficients of of equation (1*5) must be greater t h a n or equal to the corresp o n d i n g coefficients of ELo* Since the n u m e r a t o r A(s) of the transfer f u n ction T^ is the n u m e r a t o r of the t r a n s f e r admittance of a passive RG unbalanced t w o - p o r t , it follows that the zeros of the tran s f e r f u n c t i o n m a y appear anywhere except on the positive real axis* The f ollowing n e c e s s a r y c o n d i t i o n for the voltage trans fer f u n c t i o n of the (RCVR) m a y n o w be s t a t e d : The voltage t r a n s f e r f u n c t i o n of an unb a l a n c e d R O three-port n e t w o r k used in c o njunction w i t h a voltage -15fol l o w e r n e t w o r k as shown in I1Igure 1.1 must be a. ratio of nonne g a t i v e coefficient polynomials in s with the coefficients of the n u m e r a t o r polynomial b e i n g no greater t h a n the c o r r e s p o n d i n g coefficients in the d e ­ nom i n a t o r polynomial. In addition, f o r stable t r a n s ­ fer functions, the zeros of the denominator polynomial must be located in the L H P w i t h the p o s sible exception of a single zero at the origin= The zeros of the n u ­ m e r a t o r po l y n o m i a l m a y appear anywhere except on the pos i t i v e real axis. A synthesis proc e d u r e will n o w be given wh i c h shows that a voltage t r a n s f e r f u n c t i o n fulfilling the above n e c ­ essary c ondition m a y always be realized. T h u s , the stated c onditio n on a t r a nsfer f u n c t i o n will be shown to be suffi­ cient as well as n e c e s s a r y for its realization. Consider the two-port un b a l a n c e d EC networks, "Bfl, shown in Figure $.I. "A" and A s sume that the short-circuit a d ­ mittance parameters of bot h n e t works have the identical d e ­ nomi n a t o r p o l y nomial .D(s) and that their n u m e r a t o r p o l y n o ­ mials are defined as i n equation (1.4) with the addition of the subscript A or B to denote net w o r k "A" or “B" r e s p e c t i v e ­ ly. If B volts are placed on the input terminals of each n e t w o r k and the output terminals g r o u n d e d , t h e n f rom the —16— N "12A IlA N11B N12B 11A~ 12A Figure $.1. The "A" and "B" two-port unbalanced RC n e t w o r k s „ Figure $.2. I n t e rconnection of "A" and "B" networks -17short-clrcuit admittance equations we k n o w that currents, are established as shown in Figure 3*1° Now if the "A" and riB tl networks are interconnected as shown in Figure 3 * 2 ? the same currents will f l o w since the conditions p r e v i o u s l y imposed still apply* (terminal nal The int e r c o n n e c t i o n forms a n e w three-port (I) to ground, termi n a l (2) to g r o u n d , and t e r m i ­ (3) to g r o u n d ) unb a l an c e d RC net w o r k whose short-circuit admittance parameters are seen to be jIl" n IIA+nIIB „ D = y I l A +yIlB .-~N 12A J12" y I 3~ I '='y12A (3*1) N1 2 A +N1 2B"N 11A n IIB ; ■*■■ r ■ D If this n e t w o r k is used for the p assive p o r t i o n of the .(RCVF), the overall voltage t r a n s f e r function will be ,12A “y12 y l l +y13 n = v - • '12A N 1 2 A +N12B N 1 1 A +N1 1 B +N1 2 A +N12B"N 1 1 A _ N 11B = A(s) (3.2) ■ This equation is intere s t i n g in that its element .poly­ nomials are polynomials wh i c h are the numerators of RC transfer admittances and not numerators of RC driving point V admittanceso The e q u ation suggests the f ollowing synthesis / procedur e for a t r a n s f e r f u n c t i o n which meets the p r e v i o u s l y -18stated n e c e s s a r y conditions. 1. Ghoose a p o l y n o m i a l D whose zeros are restricted to the negat i v e real a x i s . 2. Synthesize an E G "A" ne t w o r k whose t r a n s f e r admit­ tance is _ -A(s) D(s) J12A~ This m a y always be done provided the coefficients of A(s) are posit i v e (the n e c e s s a r y cond i t i o n g u a r ­ antees that t h e y are) and that the order of D and the magnitudes of its coefficients are chosen s u f ­ f i c i e n t l y large 5» Synthesize an EG (see S e c t i o n 15-3 of reference 25)® "B" n e t w o r k whose t r a nsfer a d m i t ­ tance is: y 12B _ -(P(s)-A(s)) Zttsl A g a i n the coefficients of (P(s)-A(s)) are g u aran­ teed to be p o s itive b y the nece s s a r y c o n d i t i o n ^ and the nB rt n e t w o r k m a y always be realized provided the order of D and the magnitudes of its c o effi­ cients are c h osen s u f f i ciently large. 4-o A f t e r b e i n g synthesized, the "A" and "B" networks are connected as i n .Figure J.2 and joined to a . v o l t ­ age follower as shown in Figure 1.1. This new overall c o n n e c t i o n is sh o w n in Figure 3®3® -19- Figure 3»5« followero "A" and l1B ri n e t works connected to the voltage -20The voltage t r a n s f e r f u n c t i o n of the active c o n f i g u r a ­ t i o n will t h e n he T y= : Thus, the p r e v i o u s l y quoted con d i t i o n is sufficient as well as necessary. The synthesis proc e du r e outlined above will, in general, p r o duce configurations with an excessive n u m b e r of e l e m e n t s . F o r example, consider the band pass voltage t r a nsfer f u n c ­ tion ,Ools -Ly.-' "7r^T .. v s^+.ls+l (3o3) If we applied the suggested synthesis p r o c e d u r e , we could separate the f u n c t i o n as y 12A= -Ools -Di and •(s2 +l) y12B= (3.4) y 1 2B m a y ^3e realized in the f o r m of a twin-T RC network w h i c h requires 3 capacitors and 3.resistors for a total of 6 elements. y]_2A co"t:i^^ t h e n be realized w i t h 2 elements. It is known, however, that it is possible to realize the transfer f u n c t i o n w ith fewer elements (reference 14) tha n 8. Thus the suggested synthesis procedure, although p o w e r f u l , in general requires an excessive numb e r of elements. For this reason, it is desirable to examine other methods of synthesis. The next chapter will be concerned w i t h this problem. / -21A. • SUBPLUS F A C T O R U T I L I Z A T I O N Before going on to other synthesis methods 9 it is a p ­ propriat e to examine the p o s s i b i l i t y of u t i l i z i n g surplus factors to convert a t r a n s f e r funct i o n T^ w h i c h apparently does not meet the n e c e s s a r y and sufficient conditions of this chapter into a t r a n sf e r f u n c t i o n which does meet these conditions, o It is well knoTim in circuit t h e o r y that there are cases w h e n it is p o s sible to m u l t i p l y a po l y n o m i a l with negative coefficients b y a p o l y n o m i a l with L H P zeros such that the resultant p o l y n o m i a l has nonnegative coefficients. T h u s , if we have a T whose poles are restricted to the LHP and its n u m e r a t o r p o l y nomial contains some negative coef f i ­ cients, there still exists a p o s s i b i l i t y that surplus f a c ­ tors mig h t be utilized so that T ^ meets the prescribed n e c ­ essary and sufficient conditions. In the literature, the overall pro b l e m is b r o k e n d o w n into the fundamental p r o b l e m of fin d i n g a p o l y n o m i a l B(s) w i t h L H P zeros such that B(s)<F(s) has n o nnegative coeffi­ cients where F ( s ) =s^-(2 c o s 0 ) s + l and „ This is a valid simpl i f i c a t i o n of the p r o b l e m since factors of the fo r m of F(s) m h y t>e separated out of any po l y n o m i a l with - negative coefficients 22- (provided an appropriate f r e q u e n c y t r a n s f o r m a t i o n is m a d e ) „ These factors m a y t h e n be ind i ­ v i d u a l l y operated u p o n to pr o d u c e nonnegative coefficient polynomials. One appro a c h has b e e n to sho w that a po l y n o m i a l of the type B(s)=(s+l)n could pr o d u c e the desired coefficients and to r e a s o n t h e n that for s u f f i c i e n t l y small and distinct n the surplus fact o r B(s)= Tf ( s + l + 6\ ) will a p p roach the i=l 1 same result (28,40). % The smaller the value of (f) , the larger the vhlue of n required to produce h o nnegative coefficients. es zero, the required n approaches o o , of As 0 a p p r o a c h­ The optimum choice f or B(s) = (s+y3 )n will be considered so that for a given n, 0 m a y take on the smallest value p o s sible tionsis- tent wi t h the n o nnegative coefficient r e s t r i c t i o n on . #(s)B(s). odd is It. is s h o w n that the optimum choice of JJq^ = I , but that for n eve n The corres p o n d i n g values for 0 c o a "1" JJ for n or' m i n are cbs”"^ ■n and ^ .Consider B(s) = (s 0})n =bo sn +b 1 sn ' +b “•s+b n (3.5) -23where -K = — (g= 0 sl s2 ,»o o o ?n) j (3 »6 ) The p r o d u c t of B(s) and F(s) m a y t h e n he w r i t t e n as: Y)-f-"] o B(s)F(s)= ‘Y, ( h . T + b . ^ - 2 G o s 0 b - ) s n + 1 ™3 (3-7) j=-l 3-13+1 where Thus, it is seen that the f o l l o w i n g set of inequalities m ust he satisfied in order that B(s)F(s) have n o nnegative c o e f f i ­ cients hj-i+ho-t Gj= ■l J -b ^ ' - 2 cos, (j=0,l,...,n) ■ (3-8) J Substi t u t i n g the values from Eq. S 4- — 3 (3-6) into (3-8) -JLf- /](n-a+l) (j+1) (5.9) The basic p r o b l e m m a y n o w be equated to that of choosing a n onnegat i v e value of J3 such that the minimum of the set (Gq , G ^ , p « --,Gn ) is maximized- Call this value of G, G m i n max' As seep from Eqeontinous functions of (3-9), f o r arbitrary h, the G^ are J3 w h i c h possess n o m ore t h a n one m i n i m u m and no m a x i m a in the range ©-= j!] <co» In addition, if g ( y O ) is defined as m i n (Gq 9G 1 , -- - - ,Gn ) , t h e n g =0 for ^3=0 and co ■- J3 corresponds to The values of J3 for T h u s , the optimum value of inter s e c t i o n of two of the G curvee» w h i c h Go and G . intersect m a y be computed b y me a n s of U ■ an equation (3°90 as P C'i)3—© jI j o o o o jn 51^ 3°) - : - (3 .10 ) A n exami n a t i o n of this equation shows that f o r a given 3% the value of i =* 3 w h i c h results in a mi n i m u m value for i=g’4-l„ P is Since S=G q for yf}=09 it follows that as yQ is in ­ creased fro m 0 , g corresponds successively to ©"os©i$ <>»»»^Gn , Substi t u t i n g (3 +1) for I i n e q u ation (3»10)„ the optimum value of P> is that element of the set Ch4'3 R n - 3 + 1 J (.3~O.5:00® »n—I) w h i c h produedp the largest value of G. from E q e U (3 .11 ) ( 3 . 9 ). That is 9 what permi s s i b l e Value of 3 maximizes (nr 3OC 3 +jT (5.12) Differen t i a t i n g wit h respect to 3 : 9G3 = (n-2i+l)(n+2) 372 37^ (n-g) ■(3 + 1 )7^Cn-3 +Ijpz^(g+ 2) (5.13) n-l wh i c h is seen to be p o s i t i v e „ negative, and zero for 3 < — g— » 5 and 3 n-l respectively0 It follows that the opti m u m value of y0 (yQ 0p^) and the corresponding value of m i n i mum(^)=cos""^ .nLa? ' is found b y substituting 3=2^1 for n odd and 3 = 2^1..+ y2 f o r n even into E q s » (3.11) and ( 3 . 1 2 ) e E o r the case of n even, it is found that the two possible values of 3 result in identical values of Gm i n ma3C» Thus, -25- \ /^opt ^ \ n odd (3.14) 0 min=cos 1 / and \ / n even (3.15) / The above shows the opt i m u m surplus factor of the form (s+yO)n . In the pre s e n t case, we are not limited to surplus factors wh i c h lie on the n e g ative real axis. I H P complex conjugate zeros are p e r m i s s i b l e and, in g e n e r a l , the y are superior to negative real axis zeros. The p r o b l e m of f i n d ­ ing thei r optimum l o c ation appears to be much, mor e difficult ahd ho solut i o n is offered to this problem. however, is left as a p r o b l e m c h a l l e n g e . ; The question, ■IV. (HCVF) HEALIZATIOH TECHNIQUES This sec t i o n is concerned wit h the synthesis of special n e t w o r k configurations to he used as the pas s i v e p o r t i o n of the ( H CV F ) . These configurations are of interest since, in general, t h e y require fewer elements than the c o r responding networks realized b y the general synthesis p r o c e d u r e of Chapter I I I e. S e c t i o n A of this chapter gives a pictorial r e p r e s e n t a t i o n of the pas s i v e synthesis p r o b l e m which s i m p l i­ fies the v i e w i n g of the p r o b l e m and offers insight into the sensitivity, problem. S e c t i o n B reviews n e t w o r k c o n f i g u r a ­ tions pres e n t e d b y other authors. This mater i a l is solely concerned w i t h second-order t r a n s f e r functions. Sec t i o n C presents a n e w l y developed synthesis procedure for the b i ­ quadratic transfer function. S e c t i o n B presents a ladder n e t w o r k c o n f i g u r a t i o n for low-pass and high-pass transfer functions. S e c t i o n E presents a special n e t w o r k c o n f i g u r a ­ t i o n for f o u r t h-order band pass t r a nsfer functions. Finally, S e c t i o n F presents several n u m e r i c a l examples of the syn­ thesis procedures pres e n t e d in this chapter. A. A P I C T O R I A L R E P R E S E N T A T I O N OF T H E PASSIVE SYNTHESIS PROBLEM There are a n u mber of ways to state the requirements on the pass i v e p o r t i o n of the (ECVF). A simple w a y of vie w i n g -27t?he p r o b l e m and one w h i c h offers some insight into the s e n ­ sitiv i t y p r o b l e m is offered below* Con sider a (RCVF) vol t a g e trans f e r f u n c t i o n Tv m e e t i n g the n e c e s s a r y and sufficient conditions of Cha p t e r III* F r o m equ ation (1*5) rn _ A(s) _ :.A(s) _ ■ %2 V - F t s T - R(a)+I(s) " Ir11- H 1 , ■ (4.1) where R(s) = 'P(s)-A(s) = rn sn + *.* * * * * * *»+r^s>rQ has nonnegative coefficients since the coefficients of A(s) are required to be less t h a n or equal to the c o r responding coefficients of P ( s ) , ■Since the p o lynomials and 5 are the n u m e r ­ ators of short-circuit RC admittance p a r a m e t e r s 9 it is p o s s i ­ ble to associate the p o lynomials A ( s ) s P(s), and R(s) with the currents f l o w i n g in a f o u r-terminal RC n e t w o r k whose short-circuit admittance pa r a m e t e r s possess t h e s e nume r a t o r polynomials* This p o s sible current dist r i b u t i o n is shown in Figure 4*1 and is onl y valid w h e n terminal (I) is driven b y a voltage source equal to the characteristic polynomial B(s) of the network, and the other terminals are grounded* Since — is a d r i v i n g point admittance, the zeros of its n u m e r a t o r and denom i n a t o r must be interlaced -28- Figure 4.1. A p o s sible current distribution. A ( S ) = S - S ^ a n s+a P(s)+E(s)= o P 2 S +(p 1 +e 1 )s+pl R(s)=r-s +r-, s+r Figure 4.2. Current d i s t r i b u t i o n of passive p o r t i o n of • 2 2 (RCVF) net w o r k to realize T ^ = ( a 2 s +a 1s+a 0 )/(p2 s +p^s+p^). -29™ along the negative real axis* Thus E(s) m a y be thought of as a pol y n o m i a l whose coefficients serve the sole purpose of augmenting the co r r e s p o n d i n g coefficients of P(s) in such a m a n n e r that the zeros of the resulting po l y n o m i a l are shifted onto the negative real a x i s . Since E(s) is also the num e r a t o r po l y n o m i a l of the s e nsitivity f u n c t i o n associated wi t h the g ain of the vol t a g e follower (as will be shown l a t e r ) „ it is desirable to choose E(s) so that its effect on this s e nsitivity is minimized* This p r o b l e m is d i s c u s s ­ ed fu l l y in Chapter V* In conclusion, Figure 4-*I is a pictorial r e p r e s e n t a ­ ti o n of the f ollowing synthesis problem* Synthesize a fo u r - t e r m i n a l HG n etwork wh o s e c haracter­ istic polynomial D poss e s s e s arbitrary negative real axis zeros, suc h that w h e n terminal (I) is driven b y D volts and the other terminals are shorted to ground, A(s) amps f l o w t h r o u g h t e r m i n a l t h r o u g h terminal min a l (4)* (2), E(s) amps flow ($), and R(s) amps f low thr o u g h t e r ­ Once this n e t w o r k is realized, nected as shown in Fi gure 1*1 w i t h terminal ing the ground terminal* it is c o n ­ (4-) b e c o m ­ The resulting v oltage t r a n s ­ fer f u n c t i o n is then, of course, Tv = A ( s ) / P ( s ) » This pict o r i a l r e p r e s e n t a t i o n as discussed above and shown in Figure 4*1 will be used extens i v e l y in this chapter* _ JO— ■B „ R E V I E W OF P R E V I O U S L Y P R O P O S E D O O H F IGURATIOES Before p r e s e n t i n g n e w synthesis techniques, a revi e w of existing techniques will he given, S a l l e n and K e y (14) p r e sent a catalog of 18 circuits which, w h e n used in c o n ­ junction w i t h an a mplifier of g a i n K q , produce second order transfer functions of the general form a 9Sv +a 1s+an V " — -4 — P 2 S ^ p 1 Stp 0 0 .2 ) A l t h o u g h the a mplifier gain K q of these circuits is neither restricted to n o r required to he u n i t y in all cases, m any of the cataloged circuits are usef u l at this va l u e of gain, S a l l e n and Key, in fact, point out the advantage of using this value. W h e n u n i t y g a i n is permissible, the networks of S a l l e n and K e y reduce to the (RGVF) case. wh i c h are permi s s i b l e are 10 in number, Those networks A study of them shows that there is a wide choice of network element values for the r e a l i z a t i o n of a specific n etwork f u n c t i o n b e longing to a giv e n n e t w o r k f u n ction form. It is i n t e r e s t i n g to note that none of the networks of S a l l e n and K e y are valid for the real i z a t i o n of the n e t w o r k f u n ction of the f o r m T^ = A(S)ZP(S ) = U g S + a ^ s + a g / p g S + p ^ s + p Q in the (RCVF) conf i g u r a ­ tion, It is also i n teresting to note that the wor k of Bal a b a n i a n and Patel (IJ) w h i c h is an extension of that of S a l l e n and K e y is co m p l e t e l y inapplicable to the (RGVF) -31case in that the g ain of the active element is in the v i c i n ­ ity of u n i t y but that a gain of exactly u n i t y is i n admissi­ ble. F o r the synthesis of the networks of S a l l e n and K e y 9 the r ead e r is referred to t h e i r report C. (14)„ A R E A L I Z A T I O N T E C H N I Q U E F O R T H E BIQUADRATIC TRANSFER FUNCTION Consider the b i quadratic t r a nsfer function ' 2 a 0s +a.-, s +3/-, T K - 2 - 1 — 0 (4.3) P 28 +P 1 S +P 0 whose coefficients s atisfy the n e c e s s a r y and sufficient c o n ­ ditions of the (RCVF )i0 Assume all coefficients are greater t h a n zero and K is mad e s u f f i c i e n t l y small such that the r e ­ sulting n u m e r a t o r coefficients are less than the co r r e s p o n d ­ ing deno m i n a t o r c o e f f i c i e n t s » .In order to rea l i z e T^ as the voltage t r a nsfer f u n c t i o n of the (RCYF )9 it is sufficient to reali z e a four-terminal n e t w o r k whose current distr i b u ­ t i o n as defined in S e c t i o n IY A is shown in F i gure 4.2» Since the c o m p l e x left-half plane zeros of a quadratic pol y n o m i a l m a y be shifted onto the negative real axis b y i n ­ creasing the coefficient of the first order term,"it is p e r ­ missible that E(s) be no more complicated t h a n E ( s ) = e ^ s e- -32It will n o w be shown that the biquadratic 3? w i t h the p r e v i o u s l y m entioned restrictions m a y be r e a lized with no more t h a n three capacitors and f our resistors <> The effect of r e m o v i n g the restrictions on the coefficients will then be examined and it wil l be shown that the r e q u i r e d number of elements m a y o f t e n be r e d u c e d o A ss u m e the characteristic polynomial, D ( s ) , of the four terminal p assive n e t w o r k to be realized is D(s)=d^s+l where d ^ m i n C a 1Z a ^ r 1Zr0 ) and is greater t h a n zero since aQ , S 1 „ r 0 , r 1 are all greater t ha n zero, BTq w consider the net w o r k c o n figuration s h o w n in Figure 4,3, It is easily verified that this c o n figuration has an input current d i s t r i b u t i on the same as that of Figure 4,2 and that the characteristic p o l y n o m i a l is of the form D(s) = d 1 s+l» Thus, the dri v i n g po i n t current m a y be considered Q as PgS +(P 1 ^e 1 )Sfp 0 as shown in Figure 4,3, To calculate the current d i s t r ibution for the r e mainder of the network, it is p o s s i b l e to p e r f o r m a continued f r a c ­ ti o n expansion of the dri v i n g p o int.admittance to obtain the values of E 1 and G 1 as E 1_ I pO 0i"P1 +ei- <a1Po -33- d-, s +1 2 2 Ia =P 28 +(P 1 H-B1 )S^-P0 Ib =P 28 +(P 1 +e 1- d 1p 0 )s Figure 4.3. Network c o n f i g u r a t i o n with input current distri b u t i o n of Figure 4.2. oH : c u ------- T 1C^I -O I P^xj -- W -------+ aI -—s+1 R1 aO 1 I f Ib I - Cr0 aI 7 aO )s+r0 a I b =alSpa0 I =r0 s^ c ^ 2 I — d.^S d I e - (rl-r 0 al/a 0 )s I = B1 S f :: °> 1C j I r3 T 4 f 4 R 1=(IZr0 ) Q Rg=(IZa0 ) Q (ai-a 0P 2Z C 1 )Z(a 0ri-r0 ai ^al^l“P 2a 0 ^ ^Cl aO el ^ C l = ^rp l +el“ alp 0 //a0 ^) F. F< T g Z C a 1Z aQ-RgZCi) S 2Z C a 1Za 0 - P 2 Z C 1 ) Figure 4.4. Current d i s t r i b u t i o n for case of Ft (S 1Za 0 )=(T 1Zr 0 ) o - 34 ™ The current flowing thr o u g h is t h e n seen to he (d^QS+Pg,) p amperes and the current t h r o u g h is p^s +G-^se The eontinued f r a c t i o n e x pansion is continued to obtain p2 — and ' d-, - p2 E 0= — % 1 The voltage in Figure 4,3° P2 S1 - C, and the currents th r o u g h Gg and Eg are shown It is easily s een that G^, O g , and Eg are posi t i v e provided e^ is c h o s e n sufficiently l a r g e „ The elements E 1 , C 0 and E 0 are n o w divided in such a m a n n e r that paths for the currents rOaI (— — — s+rD ) 9 (an s+an ) $ 2 2 rOaI 0 (rgS ; ^ (a.gS^), (r^- ■a ;,'J~)s, and e^s are provided for the case of Ca 1Za 0 )-Cr 1Zr 0 ) o F o r the case of Cr 1Z ^ 0 )-Ca 1Za 0 )9 the elements are divided in s uch a manner that paths for the a0r 2 2 2 aGr I currents C^1S +rQ ) 9 C ™ — 8 +a0 ), C^gG ), (agB ), Ca1)s 0 and S1 S are provided. V The current distributions for these two cases are shown in Figures 4-.4- and 4.5$ respectively. It is r e a d i l y seen that the networks of Figures 4.4 and 4.5 m a y be put into the f orm of Figure 4.2. Thus, these two networks will realize in the (BCVF) config u r a t i o n ag s 2 +a1 s +a 0 any biquadratic voltage t r a n s f e r function T = — — g— 1 -----P2S +P1 S +p0 to w i t h i n a constant m u l t i p l i e r K provided all coefficients are greater t h a n zero. -35- O 'TT I Bi O t I, 1S = crI V rO ^ +aO 1I==r I s Ir O S1-(Va0)Q E2=(Vr0)Q 1C = a S s 2 Crl-ro V 0l)/CrOal-aprl>|ft ^a=r S s crI0I-PsrO v c c IrOeI V 1C = caI=aOr Iz r O ^ 1I = eIs Figure 4.5» 0I = cP I +eI-r Ip O z r O 5 P a 2/(r i/r o - p 2^ C l ^ r 2/ (^1Zr 0 - P 2Z C 1 ) Current d i s t r i b u t i o n for case of F, F, ( r r Q )=(a ^ / a0 ) -36The p r e c e d i n g was based on the fact that K be chosen s u f f i c i e n t l y small that the n u m e r a t o r coefficients of T^ be less t h a n the co r r e s p o n d i n g denominator c o e f f i c i e n t s <, Since m a x i m u m gai n is u s u a l l y desired, it is desirable to increase K until at least one of the n umerator coefficients is- equal to its c o r r e s p o nd i n g denominator coefficient. In other words, K is increased un t i l at least one of the' c o e f f i ­ cients of the p o l y n o m i a l R(s) is zero. A s s u m i n g that o nly one of the coefficients of R.(s) is reduced to zero, three pos s i b i l i t i e s will be e x a m i n e d , If rg is z e r o , it is noted that the n u mber of r e q uired c a p a c i ­ tors is reduced b y one. If Tq is zero, the number* of re- . quired resistors is reduced b y one, Ior the ease of r-^=G, the prop o s e d net w o r k c o n fi g u r a t i o n no longer h o l d s , and one must either b a c k off on the value of K or attempt a different n e t w o r k configuration. The four remaining cases, w hen two or three of the coefficients of R ( s ) simultaneously vani s h as K is increased, are s imilarly analyzed. D. L A D D E R N E T W O R K R E A L I Z A T I O N E O R LOW-PASS' A N D HIGH-PASS T R A N S F E R FUNCTIONS A r e v i e w of a synthesis p r o c e d u r e proposed b y the p r e ­ sent author (16) for the synthesis of a rbitrary order low- -37pass active filters in the giveno (EGVF) configu r a t i o n will "be As an e xtension of this work, it will "be p r o v e n that the 4 t h order low-pass t r a n s f e r function m a y always "be realized "by this method. In addition, the method will "be extended to include h igh-pass filters. Consider the "basic ladder n e t w o r k shown In Figure 4,6 and th e n consider each Y . (j=l, 2 ,,*,,n) decomposed into two 9 n subnetworks whose admittances are Y. and Y . ' such that 9 it U n U Y.+Y,=Y.o Eac h of the Y, ground terminals m a y be d i sconnectJ J J J ed fro m ground and soldered t o g e t h e r to form a n e w terminal as shown in Figure 4,7* ,This altered net w o r k m a y be thought of as a decomposed ladder network. The n o t a t i o n has b e e n changed in Figure 4,7 to agree w i t h the current dis t r i b u t i o n n o t a t i o n used in previous sections. It is s een that the p u r ­ pose of decomp o s i n g the admittances is to shunt the currents B(s) and E(s) to separate terminals. The success of the s y n ­ thesis proc e d u r e p resented in this section depends u p o n the ability to p e r f o r m this separation, added in series w i t h terminal f r o m terminal Eote that an impedance (I) or an admittance placed (2 ) to the refe r e n c e node has no effect on the voltage transfer f u n c t i o n of the corresponding (EGVF), A synthesis p r o c e d u r e for low-pass t r a nsfer functions in the for m of the decomposed ladder will n o w be given w i t h -38- Figure 4.6. The basic ladder network. P(s)+E(s) Figure 4.7. The basic decomposed ladder n e t w o r k no guarantee that the r e su l t i n g p assive elements hav e non nega t i v e v a l u e s o Experience has shown that all of the n e t ­ w o r k functions so f a r encountered of the low-pass form (4.4) V ® m sn+V l sIl“ 1+' m a y be r e a l i z e d 0 — +PjS+1 P(s) A p r o o f is offered that all functions of this fo r m for the case n = 4 m a y be realized b y the decomposed ladder n e t w o r k » The p r o p o s e d synthesis p rocedure n o w follows w i t h the shunt arms of Figures 4-»6 and 4<>7 assumed to be single capacitors and the series arms assumed to be single resistors as shown in Figure 4,8» 1, Choose a.polynomial-E(s) of degree less t h a n or equal to n and a p o l y n o m i a l D(s) of degree (n-1) such that y 11='fp(sO+E(s)J /D(s) is a dri v i n g point admittance conforming to the config u r a t i o n of Figure 4,6» E(O)-D(O)=Io Magn i t u d e scale y 11 so that P(O)+ Since. P ( O ) = I 9 t h e n E ( O ) = O o , Eote that the zeros of . [p(s)+E(s)j and D(s) must be interlaced along the negative r eal axis of the com p l e x f r e q u e n ­ cy plane, 2, P e r f o r m a continued f r a c t i o n expansion on Jp(s)+E(s)] /D(s) to determine the element values for the c o n f i g u r a t i o n s h o w n in Figure 4.6» step of the expansion, At each, label the denom i n a t o r p o l y n o ­ mia l of the ratio of polynomials used to determine — 40 - — Figure 4.8, Decomposed ladder network with shunt arm capacitors and series arm resistors. D ( S ) = K ^ (s +S1 ) (s +(5^) (s +66 ) 1S=k I s (S +61 )(S +64 )(S +6 5 ) cs +62 x s +S5 ) Vb =K^ C s +5 5 ) Ic=k$ 8 (s+Sy) Vc=^4 kj=CjEj Id~k4 s (j=l,2,),4) Figure 4.9« Current d i s t r i b u t i o n for basic ladd e r network realiz a t i o n of . - 41 - - C , as D.* d: 3o u Tl To determine the 0. of Figure 4*8, the f o llowing tJ proc e d u r e is carried out* A continued fraction ft expansion of E(s)/D(s) is b e g u n to o b tain C ^ s * It Rote that O^s is E(oo)/D(oo)* mined as [e (s )-C^s D (s )J Eg(Co)ZDg(Co) * EgQs) is next d e t e r ­ and C^s is determined as This pro c e s s is continued to determine ft the r e m a i n i n g C .* If E(s) and D(s) are considered Jt as E 1 (s) and D 1 (s) respectively, t h e n C . s = E . (s)/ Jo 1 3. V D. 4-o (s)* o Finally, the O . of F i gure 4.8 are d e t e r m i n e d as I „ J CL=C,-C . * u d u The a p p endix of r eference 16 contains a p r o o f that the synthesis p rocedure outlined above is valid* The reference does not con t a i n any guarantee that the element values so realized are nonne g a t i v e * It is shown b e l o w that the 4-th o r ­ der low-pass t r a n s f e r functions m a y always be realized w i t h n onnegat i v e elements b y this m e t h o d » Figure 4*9 shows the basic three-terminal passive RC,' n e t w o r k w h i c h must be decomposed in order to realize a 4 t h order low-pass t r a n s f e r f u n c t i o n b y the method outlined in this section* A s ■explained previously, this n e t w o r k has b e e n synthesized fro m an RC d r i v i n g point admittance -424 (p.+e^, )s"l'+(p,+e^)s:?+(pp+ep)s 2 +(p -1+en )s+l J 1 1 = ----- ^ 2.------- p ~ ^ ------ :----- ---- d 3 s +d2 s +di s+1 (4,5) where the trailing coefficient of the numerator and d e n o m i ­ n a t o r po lynomials have b e e n n o r m alized to unity. If the current and voltage d i s t r i b u t i o n of figure 4,9 is calculated 9 it is seen to have the for m shown. Since the object of a continued fraction e x pansion of y-^ is to reduce the order of the p o lynomials associated w i t h the current and voltage distributions b y one at each step of the expansion, the d i s ­ tributions of figure 4,9 are valid. The p ole-zero p a t t e r n of the continued f r a c t i o n ex-, p e n s i o n of y-^ will n o w be r e v iewed to show t h a t 6 / ° It is well k n o w n that the poles and zeros of y - ^ must be interlaced along the n e g ative real axis. Thus, y-^ is of the form (s+7^ ) (s+7^) (s+7^) (s +7|7 ) Jll= =E (4.6) ( S ^ 1 ) (S4<54 ) (s+(5g) £) < TJjz & 2± 7] 6 ^ where 7] < bers. The p ole-zero d i s t r i b u t i o n of y-^ is shown in figure 4’ .10.( a). 7] ^ are all p o s i t i v e real n u m ­ The first step of the continued f r a c t i o n expansion is to remove the pole of y ^ at infinity. - The admittance r emainin g after this removal is y-^^^ ^ r2SrI i ^ ^ " " ^ l s where is the original y^^° Since is guaranteed positive, -43- ^7 4$ % ^4 c Cb ;in 5 < I I I I I I I I y.- - - - - - - - (a) ^3 4 7?/ ^--------- y ------- -------y ---I I I i i i I i i I i I I I ft — I x o-l I I I I I i I I i -------A ---------V ^ -— « -------- I ^ (b) (c) (a) 4— o — I 4 — JrU 5 ©- (e) 6 - * ---- -e------ x "* ' — e- ^ 5 (f) 4 ---e- JrI l 5 (s) ----- Figure 4,10, yg P ole-zero distributions for y^l^ ^ l l ^0 5 -44it is easily verified that the effect of the above pole r e ­ moval is to shift all of the zeros to the l eft except for the most -negative zero w h i c h d i s a p p e a r s o changed. The poles are unfl) The resu l t i n g pole - z e r o pa t t e r n for is shown in Figure 4,10 (b)„ The nex t step in the expansion is to shift the most n e g ative pol e of to infinity. This is accomplished b y s u btracting a positive constant from /'-lb° =rIi The r e s u l t i n g admittance is y , , ^ ^ = ^— } — — ■ — , I CTT -H i y Il The effect of this extraction is to leave the zeros u n ­ touched a n d .to shift all poles to the left, is chosen so as to shift the most negative pol e to infinity. The r esult(o) ing pole - z e r o p a t t e r n for J s^10im Fi.gu.re 4,10 ( c ), The p r e c e d i n g two steps are repeated until the order of the admittance is reduced to zero. The pole-zero patterns for each of these subsequent steps is shown in Figure 4,10, Notice that the poles of y ^ - ^ ^ for k even b e come the zeros of the currents flowing thr o u g h the capacitors in the eprrent d i s t r i b u t i o n dia g r a m of Figure 4,9, A study of the regions to w h i c h these poles are restricted leads to a subscript n o t a t i o n for the ($'a such that 6 2 .^ ~~— — — one exception to this ine q u a l i t y is that (5 t h a n 64.0 The m a y be larger This e x ception is of no consequence in the f o l l o w ­ ing development. -45If the proposed synthesis proc e d u r e is to he s u c c e s s f u l 9 |p(s)+E(s)J and D(s) must he c h o s e n to have p r o d u c e d currents t h r ough the capacitors of E i gure 4o9 such that a set of IVs m a y he c h o s e n to s atisfy the equality h^s(s+(5^ ) ( s + 64 ) (s + S g )+h-2 s (s +62 )(s +6 ^ ) +h^ s (s p Zt_s4 +p 5s 5 +p 2s 2 +p1 s where O-h.-k. J J ) +h ,8= 3y (4.7) ( j = l 52 5394 )o E q u a t i n g coefficients leads to the foll o w i n g set of equalities Ii1=P4 (4.8) hi(6i+64^g)+h.2:=P3 (4.9) Ii1 (6 264 (4 *10) ) +bg (6 2 +65 ) +Ii^ =P 2 Ii1 ( ( S ^ ^ )+h2 (62$^) + h ^ ^h4=P1 (4.11) E r o m the pol y n o m i a l D ( s ) = d ^ s ^ + d 2s 2 +d 1s+l=K 1 (s+ 61 )(s+ 64 )(s+ 6 g) we have (4 .12 ) 6164 +54<5q =^ ^ (4.13) (4.14) If (S 1 -^ 4 -KS6 ) is c h o s e n equal to P ^ Z p 49 the n a solution of equations (4.8) t h r o u g h (4.14) acquires the h's in terms of the coefficients of P(s) and D(s) and the q u a n t i t y h1=p4 h g =0 dI h 3 -p2 =p4 d, P4 2 h/4,=p =Pnl ” - d7 Z)-z(Po-P;i “V -—) “ 6 5 (P2 -p 4 d 7 ' ) 3 3 Sj<> (4.15) (4.16) (4.1?) (4.18) -46Since the k's i n the n e c e s s a r y conditions O - h . - k . J J (3= I 92-95 $4) m a y be made a r b i t r a r i l y large b y choosing the K of E q u a t i o n (4o6) a r b i t r a r i l y large, the conditions a c t ually reduQe to 0 - h. (3= 1 ,2.,5»4)„ These conditions are reflected U in Equations (4ol7) and (4«18) as p 2"-p 4 I ™ * “ 0 (4.19) p4 r , p l~ d^- . - 6 5 (p 2-p 4 d ^ " ) (4.20) Since gg sund (6 g reater t h a n zero and the quantity )=<a1/^39 m a y be mad e as small as desired b y m a k i n g S 1 andS^ (for example) arbit r a r i l y small while h o l d i n g (S1 - K^4S g ) = dg/d^ c o n s t a n t , the c ondition of E q u a t i o n (4.19) m a y always be fulfilled. L i k ewise (S1S ^ S g ) = m a y also be made as small as desired b y m a k i n g S 1 and S ^ arbitrarily small. The q u a ntity (S1 -Sx) m a y be made as small as desired b y choosing the zero of [p(s)+E(s)j ar b i t r a r i l y close to the S 1 term. c hosen as small as desired, as small as desired. nearest the origin ( ^ 1 ) Since S 1 Ir t u r n m a y be it follows that S ^ m a y be c h o s e n T h u s , the c o ndition of E q u a t i o n (4.20) m a y always be fulfilled, and it follows that j]?(s)+.E(s)J and D(s) m a y always be c h osen suc h that the t r a nsfer f u n c t i o n T^ of Eq u a t i o n (4.4) wit h n = 4 m a y always be realized. A means for r e a l i z i n g h i gh-pass functions of the form -47- V _n— I Prn~ v,S11+P t -"n-l" '+— --+P1S+! (4.21) based on the methods of this s e c t i o n will n o w be given. The method is based on the foll o w i n g steps. 1. P e r f o r m a high-pass to low-pass t r a n s f o r m a t i o n on the T of E q u a t i o n (4.21) to obtain a T^ of the form of E q u a t i o n (4=4). 2. Realize this t r ansformed T^ b y the methods of this section. 5» Divide the admittance of each of the elements of the realized n e t w o r k b y s. This is equivalent to m u l t i p l y i n g the characteristic po l y n o m i a l b y s. T h u s 5 the current d i s t r i b u t i o n of the ne t w o r k is unchanged. 4. N o w p e r f o r m a low-pass to h i g h ^pass t r a n s f o r m a t i o n on the n e t w o r k to o b t a i n the high-pass r e a l i z a t i o n of the original T^. It is easily verified that the proposed high-pass synthesis method is valid. A N E T W O R K CONFIGU R A T I O N EOR E O U R T H - ORDER BAND-PASS TRAN S F E R FUNCTIONS. A p assive n e t w o r k c o n f i g u r a t i o n is n o w pres e n t e d which m a y be used in c o n j u n c t i o n w i t h the (ROVE) to realize a — 48 - — fourth-o r d e r hand-pass', t r a nsfer f u n ction of the form ap sd ,2 . ZS-^tp2 S' +P1 S+ ! (4.22) where the coefficients are assumed to meet the n e c e s s a r y c o n ditions of Chapter I I I 0 W i t h r espect to active device s e n ­ sitivity and spread of element values, this n e t w o r k r e p r e ­ sents the best of a considerable n u mber of p o s s i b l e con f i g u ­ rations i n v e s t i g a t e d o The three basic building, bl ocks of the proposed c o n f i g ­ u r a t i o n are shown in Figure -4-.11«, placed in parallel, These basic blocks are and elements are disconnected from ground and joined t o g ether to f orm terminals the sought after three-port (2 ) and (EGVF) passive network. terminals are p r o p e r l y marked in Figure 4.11. (3 ) of These Since the, three networks are to be paralleled, they m a y be assumed to be d rive n b y the same v oltage source (D(s)) for the purpose of current dis t r i b u t i o n calculation. Thus, it is seen that the present co n f i g u r a t i o n was obtained b y an examination of the current d i s t r i b u t i o n for v arious subnetworks. A d i s c u s s i o n of the choice of network parameters to realize a transfer f u n c t i o n of the form of E q u a t i o n (4.22) will n o w be given. First choose a third-order denominator 'a" net w o r k "b" network 'c " network 1 vO I 0 Ia =(7I l ^ i ) F. E i b =i Q E2b=((62-/72)/62) Rla=(( 7I l A ) Z 7I f ) Q Ia 2 =S 2 D, V = ( V A - 7I2 )) R- V ( ( 7Ii A ) A ) Vb = ^ 2~^ 2')I)2 =7Ii s d d Is Ib 2-sD2 Figure 4.11. Three basic b u i l d i n g blocks for the fourth-order band-pass transfer function realization. —50p o l y nomi a l D(S) = ( S ^ 1 ) (s+ 6 2 ) (s+ 6 j) (4.25) where 5 1 , (j2 ? 6 3 ? are un e q u a l p o s i t i v e real quantities. f e r ring to Figure 4 , I l 9 D 1 is see n to be Dg, is (s^Ks+fij)- (s+(52 ) (s+(5j) ■and Notice that the current f l o w i n g through the grounded c a pacitor of the <52 (53 s 2 , Re­ "a" n etwork is sZf"+(S2 +63 )s^+ If (62 +(53 ) is c h o s e n equal to P 3Zp^ and the ad m i t ­ tance of the "a" n e t w o r k is magn i t u d e scaled "by the factor P ^ 9 this current becomes p Z|_sZt"+p3s^+ 62 6 3p^s^» The "a" n e t ­ w o r k is s e e n to take care of the realiz a t i o n of the two l e a d ­ ing coefficients of the po l y n o m i a l P ( s ), Also notice that the current flowing t h r o u g h the grounded r e s i s t o r of the "b" n e t w o r k i s ^ s 2 + ^ (Sp+S 3 ) S-^ 2 S 1 S 3 - If (Sp-I-S3 )ZS 1 6 3 is c h o s e n equal to P 1 and the admittance level of the "b" net wor k is scaled b y IZTlp S j S 3 , this current becomes (IzS 1 S 3 )G +PpSHl and m a y be considered to realize the two trailing c o ­ efficients of P ( s ) , The "c " net w o r k accomplishes the r e a l i z a t i o n of the Ty nume r a t o r and d e nominator quadratic term, ure 4,11 (c) are c h osen to interlace with the an RG admittance function. and 7^^ of F i g ­ S ’s to produce The admittance T = s (s +T^3 ) (s +7^.)Z (s+Sp)(s+S 2 )(s+S3 ) is t h e n r e a lized in the config u r a t i o n shown, A current d i s t r i b u t i o n c o mputation for the "c" n e t ­ w o r k shows that the current t h r o u g h the final r e s i s t o r is -51The admittance level of the ne t w o r k m a y t h e n be scaled, Since a quadratic t erm current. ( has 6163 already b e e n ' r e a l i z e d , the s caling factor is c h o s e n such that the r e s u l t i n g current f l o w i n g through the final r e s i s ­ t or is (p2 - --- 6 2 6 ^ It is seen that this current is the A(s) p o r t i o n of the r e a l i z a t i o n and is equal to ap. Thus, the proposed n e t w o r k realized the desired band-pass transfer f u n c t i o n to w i t h i n a constant multiplier* It is apparent that the success of the method depends on. the a b i l ­ i ty to choose the - C r 2 6 's such that the quantity Cp0- v— 7— is positive. 0103 N o tice that no r e s t r i c t i o n has b e e n p la c e d on the relative m a g n itudes of the ($'s. The element values of the "c" n e t w o r k are hot shown as they b e ­ come quite complicated in terms of the (51s a n d is far easier to w o r k w i t h n u m e r i c a l values. 7]’s » The synthesis method as presented is only valid w h e n ( 6 2 +63 ) ma N sen equal to p ^ / p ^ and to equal p^. It cho­ (5163 s i m u l t aneously chosen It is obvious that there are cases where this A a y not be done. N o r those cases it is suggested that a current d i s t r i b u t i o n dia gr a m applied to the subnetworks might suggest alternative realizations. N. EXAMPLES This section will present n umerical examples of the syn -52thesis p r o c edures p r esented in the previous sections of Ghap- ter IVo Each example p r o b l e m will be presented in the c u r ­ rent d i s t r i b u t i o n for m of Figure 4.1. ,F or the first example, c o n s i d e r the biquadratic t r a n s ­ fer func t i o n of S e c t i o n I eC e 's^+2,o.828s+4 (4*23) s ,+Io 4 1 4 s +1 / This tra nsfer f u n c t i o n will be realized b y the methods of Section IVoC0 The current d i s t r i b u t i o n for this T in Figure 4*12 for the case of V is shown K=1 A- (maximum value allowable), aI rI Since — — < — — 9 we take the n e t w o r k of Figure 4 » 4 and easily ”/ 'r 0x xr O\ calculate the element v a l u e s o The resultant n e t w o r k is shown in Figure 4,13 for the case of e-^ h a ving b e e n c h o s e n equal to 3.o293.<> Note that e^ could have b e e n chosen c o n s i derably less since the only requi r e m e n t o n e ^ is that the polynomial nected to the voltage follower is shown in Figu r e l»3o The next example will demonstrate the low-pass ladder n e t work r e a l i z a t i o n of Se c t i o n I V 0D 0 Consider the Butter- wo r t h low-pass t r a nsfer f u n ction I (4.24) s4 +2 o6131s 5+3 o4 1 4 2 s 2 +2»6131s +1 The first step of the proc e d u r e consists of c h o o s i n g a po l y - -53- A (s )= X s ^ + o? 07 s +1 P(s)+E(s)=s + (lo4-14-+e-, )s +1 Figure 4.12. Current d i s t r i b u t i o n for network corresponding to transfer f u n c t i o n of E q . (4.23)« Figure 4.13° Network r e a l i z a t i o n for T =X s + 2 ° 8 2 8 s + 4 ^ v s^+1„414s+l -54nomial E(s) ©f degree less t h a n or equal to four and a p o l y ­ nomial D(s) of degree three such that y ^ ^ = jp(s)+E(s)]/D(s) is a dri v i n g point admittance conforming to the c o n f i g u r a ­ t i o n of Figure 4.6. and The p o lynomials E ( s ) = 1 2 . 9 7 1 0 s 5 +25. 1 9 1 9 s 2 +ll.1407s (4.25) D ( s ) = 2 . 6 5 5 s 5 +6.9391s 2 +5.2260s+l (4.26) fulfill these requirements. A continued f r a c t i o n e x pansion of jp(s)+E(s)J/D(s) is performed to determine the element values for the c o n f i g u r a ­ tio n shown in Figure 4.6. The D. polynomials are identified U d uring this p rocedure as sh o w n below. C1 S=O .57 6 5 s D 1 = 2 .6 5 5 s 5 + 6 .95 9 1 s 2 + 5 « 2 2 6 0 s + l / s ^ + 1 5 .5842s5 + 2 8 .6 0 6 2 s 2 +!5. 7559s +1 s 4 + 2 .6 1 5 2 s ^ + 1 .9 6 7 6 s 2 + .5 7 6 5 s E 1=O. 2047. 1 2 .9 7 1 0 s 5 + 2 6 .6 5 8 6 s 2 + 1 5 » 5 7 7 4 s + l / 2 . 6 5 5 s 3 + 6 .9 5 9 1 s 2 + 5»2260 s +1 2 . 6 5 5 s ^ + 5 » 4 5 2 9 s 2 + 2 .7 5 8 4 s + .2 0 4 7 CgS=S ..728s D 2 =I.4 8 6 2 s 2 +2.4876s+.7955 /l2.97l0s 5+26.6586s 2 +15-5774s+l 12.9710 s ^ + 2 1 . 7 1 2 6 s 2 + 6.9405s E 2 = O . 5017 4 . 9 2 6 0 s 2 +6 . 4 5 6 9 s +1 / 1 . 4 8 6 2 s 2 +2 .4 8 7 6 s + 0 . 7955 1 .4 8 6 2 s 2 + ! o9 4 2 0 S + Q .5017 -55C =s= 9 .027s D 5 = O »54-56s+0.o4-956 /4 » 9 2 6 0 s 2 +6»4-369 s +1 4 .9 2 6 0 s 2 +4-»454-8s E 5=O.2753 l»9821s+l |0»5456s+0.4-936 0.54568+0.2753 0^ 3 = 9.0843 D4=O »2183 /l.9821s+l 1 .9821s I E4=O.2183 /0o2183 0.2185 The C . m a y n o w he determined as U 028=8.7288 D2=Io4862s2 +2.48 7 6 s + 0 o 7953 /12 .9 7 1 0 s 3 + 2 5 .1919s2+ll.1470s 12o9710s^+21.7126s2+ 6»9405s 0^ 8 = 6.5778 D 5= O.5456s+0.4936 /3 .4793s2 +4o2002s 5.4795s2 +5.1477s C4s =4o828s D4 =O.2183 /1I0525 S 1.05258 0 and O 0 determined as T The above calculations -56restilt in the n e t w o r k shown in Figure 4» 14« 'The final example is concerned with a demons t r a t i o n of the four t h-order band-pass t r a n s f e r function synthesis of S e c tion I-V 0Ec This example f u n c t i o n is _________ sf______ , V (4.27) e- sV l o414s V S 2 +1 o4 1 4 s +1 The first step is to choose the zeros of D ( s )0 c h osen as (5-1= 2 042, S p = O 04 1 4 9 S % = I * 0 0 o t y (p2 - g r These are Note that the qu a n t i- -6263P 4 ) is equal to 2 .172 , T h u s 9 the above choice is p e rmissible and leads to a value of K = 2„172 and D ( s ) = ( s + 2 04 2 ) ( s + 0414)(s+1.00) The (4.28) T^tS of the n e t w o r k are c h o s e n as 7|]_=2. 6 6 7 5 4 3 s T^j = O 06 0 0 5 and »667» These values lead to the network r e a l i z a t i o n of Figure 4 .15 w i t h the terminals for the c o n ­ fig u r a t i o n of Figure 4.1 p r o p e r l y marked. A(s)=l P(s)+E(s) 3o414s +2c6132s Figure 4,14. Net w o r k r e alization for T v s4 +2.6132s 5+3o4143s 2 +2.6132s+l' © 3.85 F, 5850 , W v ----1 (- O + (- 1.71 F, 1.109 F 1 .550^ .05520K 14.232. 1720 Iigure .15» N etwork r e a l i z a t i o n for T > 2 a 72s 2 s4+ 1 . 4 1 4 s 3 +3s2 +l.4143-1-1* V, Z E T W O B K OPT I M I Z A T I O N This chapter investigates the p o s s i b i l i t y of util i z i n g the digital computer to optimize a given n e t w o r k conf i g u r a ­ tion= A p a r t i c u l a r class of networks is selected for use as an example of the o p t i m i z a t i o n process and a computer p r o ­ gram is developed w h i c h optimizes any low-pass or, b y a suitable transformation, any h igh-pass transfer function realized b y the decomposed ladder method of S e c t i o n I V o D 0 S e c t i o n A of the pre s e n t chapter is devoted to a discussion of the n e t w o r k p a r a meters to be o p t i m i z e d « S e c t i o n B is concerned wit h a d e s c r i p t i o n of the optimization computer p r o g r a m and the results of the p r o g r a m on an example n e t ­ work, The prime pur p o s e of S e c t i o n A is to define the op ­ t i m i z a t i o n p a r a m e t e r Q while the latter section will be c o n ­ cerned w i t h the m echanics involved in m i n i m i z i n g this p a r a ­ meter, A0 OPTI M I Z A T I O N PABAMETEBS The first p a r a m e t e r to be discussed is the sensitivity parameter. Bode first introduced the term "sensitivity" in c o n n ect i o n with f e e d b a c k analysis (30), Since t h e n the t e r m has b eco m e c o m monly defined as the logarithmic derivative of T(s,k) w i t h respect to the l o g a r i t h m of k where T(s,k) is a n e t w o r k f u n c t i o n of interest (in the present case the voltage -60tran s f e r f u n c t i o n T ) and k is a variable element, Thus, the sensitivity p a r a m e t e r is qT 9 (I n T ) k . 9T k - g(lhk7 “ T gk (5.U The above de f i n i t i o n of s e n s i t i v i t y m a y be looked u p o n as r e p r e s e n t i n g the ratio of a pe r c e n t a g e changb in a given n e t ­ wo r k fun c t i o n to a pe r c e n t a g e change in a variable element of the n e t w o r k function. T h u s , ignoring phase a n g l e s 9 if the n e t w o r k f u n c t i o n varies b y 0,01 db for a I db v a r i a t i o n in k, we say the s e nsitivity is 0,01» As an example of the use of the s e nsitivity p a r a m e t e r in the present investigation, consider that the g ain of the active device departs from u n i t y and is represented b y the variable k. This represents the actual s ituation since in a n y practical case, the active device will depart from the ideal b y some amount. Ex p r e s s i on (1 .5 ) w h i c h is the expression for Tv in terms of the a p p l i ­ cable admittance polyn o m i a l s must the n be altered to T__= , = MsI v N 11- ^ 13 p(.s) (5 .2 ) A p p l y i n g the d e f i n i t i o n of (5*1) to this e q u ation results in 5®v_ k ' klL ■15 (5.5) N ll“k % 3 Since k is approx i m a t e l y u n i t y and (B-^-kir^) is P ( s ) , E q u a t i o n (5*3) m a y be w r i t t e n as SkV ~ PTsJ (5*4) -61Since the n e t w o r k r e a l i z a t i o n has no control of P ( s )9 this equation shows that the only p o s s i b i l i t y of sensitivity m i n i m i z a t i o n lies in the choice of wh i c h is the E(s) of the p i c t o r i a l r e p r e s e n t a t i o n of Figure 4.1. A n alternate approach to sensitivity is to utilize the pol e - z e r o s e nsitivity w h i c h follows w h e n the s e nsitivity of T ( s $k) is obtained in terms of its poles and zeros Let TC (31)° (s+Zi ). i=l T ( s $k ) - g n (5.5) TT (s+p.) 1 i=l where the p ^ ? and g are 9 in g e n e r a l $ functions of some p a r a m e t e r k. Taki n g the loga r i t h m of each side $ d i f f e r e n ­ t i a ting w i t h respect to k, and m u l t i p l y i n g both sides b y k i k where -I. sI + .E 1 '9 k s+zi ’ i»i sEi "k" s +P 1 (5,6) ~sp 1 5 k f h k gk are defined r e s p e c t i v e l y as the zero and pole sensitivity of the n e t w o r k f u n ction T. Still another type of s e nsitivity is "coefficient senn sitivity" (18). If T ( s ?k)= t . (k)s 1 9 t hen coefficient i=Q sensit i v i t y is defined as Y - 4.. A t 3- A j s?1= 62 - A(Int1 ) -----— Ak/k ■ d (Ink) wh i c h is seen to be the ratio of a percentage change in a gi v e n p o IynomiaJl coefficient to a percentage change in the variable element k. In the present investigation, we are interested in the variable element k r e p r e s e n t i n g the values of the components in the passive p o r t i o n of the n e t w o r k as well as the gain of the active p o r t i o n of the ( E C O ’) . Other net w o r k p a r a meters w h i c h will be of interest are element d i s t r ibution (33 )» element sum, and element ratioo Capacitance and r e s i stance distribution are de ­ fined re s p e c t i v e l y as Op= v and E b = ^ (O 1-H)Z i=l x y) f r (E 1- E )2 i=l ^ (5 .8 ) where H and \E are the average values of net w o r k capacitance and resi stance respectively. Capacitance sum C q .is the t o ­ tal capacity of the n etwork n„ S 1 (5.9) This qua n t i t y is of interest since it is p r o p o r t i o n a l to the —63~ volume and cost of the resu l t a n t network. Element ratio is defined as the ratio of m a x i m u m ele­ men t value to m i n i m u m element v a l u e » T h u s 5 capacitance and resistan c e ratio are defined r e s p e c t i v e l y as max R “ '0; min max and TT B (5.10) min It is d e s i r a b l e 5 in p r a c t i c e 9 to minimize each of the parameters listed a b o v e „ In g e n e r a l s h o w e v e r 5 it is im ­ possible to minimize t h e m simultaneously. In order to r e ­ duce a m ultitude of o p t i m i z a t i o n parameters to a single op ­ tim i z a t i o n p a r a m e t e r to be mini m i z e d $ an overall o p t i m i z a ­ t i o n par a m e t e r w ill n o w be defined together w i t h its a s s o ­ ciated o p t i m ization problem. The optimization para m e t e r Q will be defined as SJi k + Zw1 i Sgi + Z Zw0 i j 'j'Pi C j 8 .•^ k S ai S^i + T s 7 % %&i . 7 + / .W I ^ i Z Z i 'I i-j + ZZw.R j 5P i 8R i • U i. j wOB0B+wHEEI)+Wc,^Os+wc,^Oji+Wjj^SH where all quantities have p r e v i o u s l y b e e n defined gai n of the v o l t a g e .follower) Ed (5.11) (k is-the except for the w " s which are arbitr a r i l y assigned w e i g h t i n g factors. These m a y be used "-64"— to upgra d e or downgrade the r e l a t i v e importance of the c o m ­ p onent n e t w o r k parameters. The corresponding optimization p r o b l e m is to m i n imize Q for a gi v e n set of weig h t i n g f a c ­ tors w subject to the u s u a l n e t w o r k constraints of n o n n e g a ­ tive element values. A n important c o n s i d e r a t i o n in optimization problems is that of the constraints. It m a y be reasoned in the present p r o b l e m that the constraints are, in effect, absent. This is true since the n e t w o r k element values are continous f u n c ­ tions', of :the independent parameters. The element values of the initial n e t w o r k are all positive. becomes negative, zero. If. an element value it mus t have p a ssed through a value of This means that either the value of t h r o u g h infinity. or passed As long as W r, and W d have finite valitR u e s , the c o n d i t i o n of n e g ative element value will either never occur or else signify that too large a step size has b e e n t a k e n in the o p t i m i z a t i o n process. This latter c o n ­ d i t i o n is handled b y t a king the step over a g a i n with r e ­ duced step size. Thus, in the present problem, the con­ straints are e f fectively absent. A n e x amination of the opt i m i z a t i o n p a r a m e t e r Q shows t h a t , once a n e t w o r k has b e e n synthesized, all of the c o m ­ ponents of Q m a y easily be evaluated wit h the p o s s i b l e ex- _ g Ex­ c e p tion of the component s e n s i t i v i t y p a r a m e t e r s . The eva l ­ u a t i o n of the component sensi t i v i t y parameters involves i n ­ creasing each element in t u r n "by one^percent, appropriate t r a n s f e r admittances, evaluating the and deter m i n i n g the change in the coefficients of the t r a n s f e r function T - Fo r t u n a t e - V ly, there are digital computer routines available wh i c h e val­ uate the transfer admittances of a net w o r k fro m a d e s c r i p t i on of its topology. There are a n u m b e r of ways a solution to the o p t imiza­ tio n p r o b l e m m a y be attempted. It should be p ointed out at' the outset that the opti m i z a t i o n pr o b l e m is extremely c o m p l i ­ cated and that no exact s o l u t i o n is attempted. One p o s s i ­ b i l i t y for an approximate s o l ution is to p e r f o r m a p u r e l y r a n d o m choice of the independent parameters of the subject n e t w o r k a large n u mber of times w i t h Q.being evaluated for each of the resultant networks. That n etwork w h i c h c o r r e ­ sponds to the lowest value of Q is. the n selected as the final design. A n o t h e r possibility, and the one used in this i nvesti­ gation, is the method of steepest descent (35)» This m e thod is often, compared in the literature to the p r edicament of b e i n g lost in the woods. Since civilization is u s u a l l y c o n ­ centrated at the lower elevations, one p o s s i b i l i t y for —66 — .,reaching shelter is to travel at all times in that dire c t i o n w h i c h ma ximizes one's descent. In optimization problems this corresponds to c o m p u t i n g gradients at each step in the developm e n t of the solution and taking a n e w step in the dire c t i o n of steepest descent. This process is continued u n t i l no f urther significant improvement is possible. The method consists of seeking a local minimum' and it is seen that other m i n i m a are i g n o r e d . Since the compu t a t i o n of gradients can r equire considerable computational effort <, a n umber of s t e p s 'm a y be t a k e n in a given dire c t i o n in order to f u l l y exploit a g i v e n gradient computation. This p o s s i ­ b i l i t y is t a k e n advantage of in the present investigation. A p r e l i m i n a r y step in the present o p t i m ization p r o b l e m is the de t e r m i n a t i o n of a r e a s onable network t o p o l o g y and initial element values for a g i v e n transfer f u n c t i o n T^. This t o p o l o g y m a y either be selected from those presented in this report or devised b y other mieans. The experience and ability of the n e t w o r k d e s i g n e r are utilized in this step. A "reasonable" n e t w o r k is defined here as one w h i c h appears ; to be amenable to optimization. The net w o r k synthesized in this step is called the initial network. The computer p r o g r a m to be described in the next sec t i on accepts the initial n e t w o r k and optimization p a r a m e t e r w e i g h t i n g factors and, b y apply i n g the method of steepest d e s c e n t , determines improved n e t w o r k s , Bo COMP U T E R P R O G R A M DESCRIPTION. T h e .c o m puter p r o g r a m for n e t w o r k optimization has b e e n w r i t t e n in F O R T R A N for the IBM 1620 storage capability. (60K) computer with dis k As explained previously, the computer p r o g r a m is not w r i t t e n for a general network but for a p a r ­ t i c ular class of n e t w o r k s . This class of networks is the low-pass decomposed ladder n e t w o r k of Section IVoD. Since m a n y of the subroutines of.the computer p r o g r a m m a y be used if similar computer programs are wri t t e n for other classes of networks, and jsince the present computer p r o g r a m is valid for a w i d e l y used class of n e t w o r k f u n c t i o n s , the FOR T R A N statements for the computer p r o g r a m are printed in A p p endix I, In addition, the method and purpose of each subroutine are explained in this section. This should enable the read-' er to recognize and adapt those routines w h i c h are of in t e r ­ est to him. The network o p t i mization computer p r o g r a m consists of 16 subroutines and one main, program. The m a i n pr o g r a m has b e e n b r o k e n into five linked subprograms b ecause of the limited core storage of the 1620 computer. These five sub- -68programs, which have b e e n named CHOKE, C H T W O , C H T R I , CHFOUR and C H F I V E 9 are t o g ether considered as the m a i n program. The f l o w chart for this m a i n p r o g r a m is shown in Fignre $.1. The circuit analysis p o r t i o n of the pro g r a m is accomplished in terms of topological formulas based on linear graph t h e o ­ ry. Ho attempt is made here to present the basic theory or to explain the terminology. The interested read e r is r e ­ ferred to references $6 and 37 for an exposition of the t h e ­ ory and f a m i l i a r i z a t i o n w i t h the terminology. Before d i s ­ c u s sing the f l o w c h a r t 9 each of the 16 subroutines will be explained. S U B R G U T IHE R D I H . This subroutine is concerned w i t h reading in input data b y means of the card reader and c o mputing some constants which will be used later in the program. The in­ put data consists of the f o l l o w i n g quantities. HOD Humber of .branches in a complete tree of the network. HEL P Humb e r of n e t w o r k elements exclusive of the auxi l i a r y elements. HO Order of the denomi n a t o r of T^. HTT P A r r a y for identifying the type of each ele­ ment- of the network. +1 for a Capacitor, 0 for a resis t o r and -I for an inductor (induc­ tors are not used in the present c a s e ) . -69- ( START GALL R D I N I N C REASE ELEMENTS OF IFR & ITO _________ ARRAYS BY ONE__________ CALL T R G O M A L P H ( I ) = A L B A R ( I ) ,1=1,NO B E T A (I)= B T B A R ( I ) ,1=1,NOMT CALL SYNTHZ CALL E V LVTR CALL SENSIT CALL EVOB VOB A R = V O B J PRINT H E A D I N G F O R INITIAL NETWORK CALL DMPOT Figure 5»1» program. F l o w chart for net w o r k optimization computer —yo— Figure 5,1. (Continued) -71- ST p =STBES CAL L TKSTP A L B A R ( I ) = A L P H ( I ) ,1=1,NO B T B A R ( I ) = B E T A ( I ) ,1=1,NOMT VOB A R = V O B J SENSE SWITCH D E C S = ( (VORJ-VOBJ)/ V O R J )*10000 D E C S = A B S F (DECS) PRINT OUTPUT H E A D I N G F O R FINAL N E T W O R K CALL DEPOT CALL EXIT Figure 5« I (Concluded) -72- IB Identifying n u m b e r of an element connected b e ­ t w e e n the input and reference node of the s u b ­ ject network. IJ Identifying n u mber of an element connected b e ­ tw e e n the input and output node of the n e t ­ work. KB Identifying n u m b e r of an element connected b e ­ t w e e n the output and reference node of the n e t ­ work . IBR A r r a y for ident i f y i n g the node "from" which each element is connected. ITO A r r a y for i d entifying the node "to" which each element is connected. SP A r r a y for the coefficients of the denominator (P(s)) of T . SP(I) is the coefficient of the s t erm raised to the AIiBAR • (I-I).power.. A r r a y for zeros of the initial (P(s)+E(s)) p o ­ lynomial . BTBAR A r r a y for zeros of the initial D(s) polynomial, WK WX A r r a y f o r the w e i g h t i n g factors wJgpi1 A r r a y for the w e i g h t i n g factors w. WELH A r r a y for the w e i g h t i n g factors w n and ,1?Pi .-R-j WCD », Set equal to the weighting factor W q -73WED WCS Set equal to the w eighting factor W to . -D Set equal to the w e ighting factor W r, . WCR Set equal to the ESP E u m h e r of coefficients in P(s) ETT . E u mher of branches in a 2-tree w eighting f a ctor W n . uE WEE Set equal to the w eighting factor wTO . ltE The subroutine computes the foll o w i n g q u a n t i t i e s „ EVB (EO.+l) „ (EOD - I )0 Eumher of independent variables associated wit h the n e t w o r k (2 x E Q - 2 )0 EOMI (RO-I)i EOM T (NO-2). EPAE Eumber of core storage locations required for the storage of one tree (5 elements m a y be stored in one core storage l o c a t i o n ) „ S E B R O E T IEE T E C O M . Determines and stores those 2-trees of the n e t w o r k w h i c h are u t i l i z e d in the determ i n a t i o n of the num e r a t o r p o l y nomial of y^g - E p j « U pon co m p l e t i o n of this s u b r o u t i n e , the applicable 2-trees are stored in the ITE array with each m e m b e r of the a r r a y containing as m a n y as five n e t w o r k element i d e n t i f i c a t i o n n u m b e r s » The quantity ETS is set equal to the n u m b e r of 2-trees stored in the ITR array. S E B E O E T IEE T E D E T . This subroutine is called b y T R C O M for - rJQ- the purp o s e of d e t e r m i n i n g if a given set of ne t w o r k b r a n c h ­ es forms a network tree. Given a set of BTOD integers (LD array) identi f y i n g a set of b r a nches of a n e t w o r k c o n s i s t ­ ing of N O D independent node p a i r s ? it is determined if the set forms a tree of the n e t w o r k . The subroutine returns the integer variable N E L as 0 if the set forms a tree. Other­ wise NEL- is set equal to K to denote that m e m b e r of the LD array w h i c h represents the b r a n c h which first forms a c i r ­ cuit with those branches p r e v i o u s l y examined (the branches are examined b e g i n n i n g wit h b r a n c h number L D ( N O D ) ). The method of determ i n i n g whe t h e r or not a given set of b r a n c h ­ es constitutes a tree is the following. 1, Take each b r a n c h b e g i n n i n g with the b r a n c h i d e n t i ­ fied b y LD(NOD) and pr o c e e d thr o u g h L D ( N O D - I )9 e t c o 9 to L D ( I ) . As each b r a n c h is s e l e c t e d 9 d e t e r ­ mine if the nodes of this branch w h e n considered w ith the nodes of those branches alr e a d y selected: 2. a. Adds no n e w nodes. b. Adds one n e w node. c. Adds two n e w nodes. If two n e w nodes are added 9 the b r a n c h selected forms a subgraph w h i c h is not connected to any w h i c h m a y have alr e a d y b e e n c o n s i d e r e d . 3° If one n e w node is a d d e d 9 the branch b e i n g c o n s i d ­ ered forms an e xtension to one of the subgraphs -75p r e v i o u s l y g e n e r a t e d ». No circuits are generated in this case., 4o If no n e w nodes are generated, there are two p o s s i ­ bilities . a0 Two uncon n e c t e d subgraphs hav e b e e n c o n ­ nected to f orm a single subgraph. No cir- • cults are generated in this case, b« A circuit has b e e n generated b y the a d d i ­ t i o n of this node in wh i c h case the q u a n ­ t i t y N E L is set equal to the subscript of the m e m b e r of the LD a r r i y w h i c h is c u r r e n t l y involved, and the examination of branches is terminated since the set of branches b e i n g examined cannot p o s s i b l y form a tree. 5o If all branches are examined without encountering the c ondition of 4b, the set of branches c o nsti­ tutes a tree and the q u a n t i t y N E L is set equal to zero o- SUBROTJTINE T R G L . of six categories. This subroutine classifies a tree into one O nly one of these categories is of in ­ terest in the present case, but since the toiitine is of use for other c o n f i g u r a t i o n s , it is discussed here. routine will accept a set of N O D integers The s u b ­ (LD array) i d e n t i ­ -76fying the branches of a tree of a net w o r k and assign the tree to one of six classifications b y setting the integer variable ICLS equal to the c l a s s i fication n u m b e r of the tree* . .Before d e f ining the clas s i f i c a t i o n n u m b e r s ? it is n e c e s s a r y to define several t e r m s » A tree w h i c h contains the b r a n c h I B 0 IHPTJT THEE GUTPUT TREE A tree w h i c h contains the b r a n c h K B 0 PLU S TRE E A n input tree w h i c h forms a tree whe n b r a n c h K B is substituted for IB but does not f o r m a tree w h e n b r a n c h IJ is substituted for I B 0 HIH U S TREE A n input tree w h i c h forms a tree bot h w h e n b r a n c h K B is substituted for IB and again w h e n IJ is substituted for I B 0 The ICLS integers m a y n o w be d e f i n e d 0 ICLS =. I .the tree does not fall into any of the c a t e ­ gories given b e l o w 0 =- 2 ? the tree is an input t r e e 0 = 3 » the tree is an output t r e e 0 = .4 ? the tree is b o t h an input and output tree, ■= 5 ? the tree is b o t h an input and minus tree, = 6 , the S U B R O U T IHE S E Q G E H 0 tree is. b o t h an input and plus tree. This subroutine is an aid in the generai t i o n of all p o s sible combinations of H things t a k e n M at a — 77 ~ In .the pre s e n t c a s e 9 the objective is to generate all time. possible trees f r o m a given set of elements and n o d e s » S p ecific a l l y , . t h e subroutine takes a set of integers, LD(M) i LD(M-I) .......... . LD(2), LD(I) where LD(I) LD(2) LD(M-I) ^ LD(M) and LD(I)Si N, and generates a n e w set of integers $ L D 8(M), L D 8(M-I), ....... L D 8(2), L D 8(I) subject to the same constraints given' above and the a d d i t i o n ­ al constraint that the quant i t y Y (LD8(J)-LD(J))Nj J=I' is posit i v e and the m i n i m u m value possible. In other words, the next highest in r a n k of N integers ta k e n M at a time has n o w b e e n generated* A n additional feature of this subroutine is the ability to skip over combinations. Suppose we have the sequence LD(M) , LD(M-I) *...... . L d C k +1) , 'LD(K)', LD(E-I) >,, , .LD(I) and, instead of generating the next ranking sequence, we d e ­ sire to skip to the next ran k i n g sequence after LD(M) ,.'LD(M-I) ,LD (K+l)., LD(K) ,(N+2-K) (-N-1) ,N» This is accomplished b y the incl u s i o n of the integer variable N E L in the input ar r a y of the subroutine where N E L is set equal to K. occurs. If N E L is set less t h a n 2, normal sequencing U p o n r e t u r n i n g f rom the subroutine, if N E L is zero, -78the sequence presented to the subroutine was the highest r a n k i n g p o s sible and the g e n e r a t i o n of all p o s s i b l e c o m ­ bination s was p r e v i o u s l y completed, For any other condi- .tion, K will be u n changed if its initial value was great­ er th a n I or set to 2 if.its initial value was less than 2 , SUBROUTINE SYNT H Z , This subroutine performs the low-pass synthesis p rocedure of S e c t i o n I Y ,B, of P(s) Given the coefficients (SR array), the NO zeros of the pol y n o m i a l +E(s)),' and the (P(s) (NO-2) zeros of the polynomial D(s), this . subroutine computes the ($xNQ-2) ■ of the r esultant low-pass filter. element values (ELEM array) The first NO element v a l ­ ues correspond to R^ thr o u g h R n res p e c t i v e l y (see Figure ff 4,8), The next two element values correspond to respectively. The r e m a i n i n g element values a l ternately correspond to G= and G= ^ tl and U ' (0= 3 )4 ,*,.,NO), Subroutine SYNTHZ u calls on subroutines P O A D and P O H U L , These two subroutines are explained below. SUBROUTINE P O A D , The input a r r a y of this subroutine is (POO, NOG, POD, NOD, R P Q , N P O ). NOD coefficients mial The subroutine adds the (POD array) of a (N0D-1) order p o l y n o ­ (POD(I) is the t r a i l i n g coefficient) to the NOG c o ­ efficients (P0C array) of a (NOC-I) order polynomial. The coefficients of the r e s u l t i n g pol y n o m i a l are stored in the -79“ E P O a r r a y and the q u a ntity EPO is set equal to the numb e r of coefficients of the r e s u l t i n g polynomial. STJBROTJTIEE POMUL. The input a r r a y of this subroutine is ( A E ?I A E ?P © B 9E O L sP 0 G 9E 0 0 ?EERR).<> The subroutine multiplies an (IAE-I) order p o l y n o m i a l ? whose IAE coefficients are stored in the A E a r r a y , b y a (EOD-I) order po l y n o m i a l whose coefficients are stored in the P O D array. The coefficients of the r e s u l t i n g po l y n o m i a l are stored in the POO array and the quan t i t y EOG is set equal to the numb e r of coefficients of the r esulting polynomial. S E B E O U T IEE E V L Y T E . This subroutine evaluates the 2-tree admittance products of the ETS 2-trees stored in the. ITE array. The admittance p roduct for 2-tree n u m b e r I is stored at l o c ation V T E ( I ) . S U B R O U T IEE S E E S I T . This subroutine evaluates the se n s i ­ tivities of the coefficients of P(s) to changes in the element values of the n e t w o r k synthesized b y subroutine SYETHZ. These sensitivities are stored in the S E L M array. S E L M ( M 11J) gives the p e r c entage change in coefficient p^_^ of the po l y n o m i a l P(s) due to a lj% increase in the element value of the low-pass n e t w o r k whose value is stored at l o ­ c a t i o n ELEM(J).. -80SUBROUTIETE E T O B 0 The sole pur p o s e of this subroutine is to evaluate the o p t i mization p a r a m e t e r Q of E q u a t i o n (5°11 )0 This qua n t i t y is stored at l o c a t i o n T O B J 0 SUBROUTINE.-DMPOT o This is an output subroutine which h a n ­ dles the data associated w i t h the initial n e t w o r k and the final n e t w o r k as explained i n the previous section, .subroutine tests sense switch I, l Hhe If -sense switch I is off, the output is printed on the on line p rinter only. If on, output is also punched on cards, SUBROUTINE E T G R A D „ The independent variables of the p r e s ­ ent opti m ization p r o b l e m are the NO zeros of (P(s)+E(s)) and (N"0-2) zeros of D ( s ) <> variables is N T B = 2 x N 0 - 2 „ T h u s , the number of independent The p u r p o s e of the present sub­ routine is to evaluate the N T B gradients associated w i t h the o p timiza t i o n problem. The cqmputed gradients are stored in the GRAD a r r a y and are weighted in direct p r o p o r t i o n to the magnitud e of the associated p o l y n o m i a l zero, SUBROUTINE STCO M P . This subroutine determines the size of step to be t a k e n in the method of steepest descent. The step size is stored at l o c ation S T P 0 SUBROUTINE T E B T P » This subroutine takes the step of value ■81S TP and computes the r esultant optimization p a r a m e t e r V 6BJ» S U B R O U T I N E OOUSTR. T h i s subroutine checks to see that a given set of zeros for .(P(s)+E(s)) gi v e n set of zeros for D(s) terlaced » (ALPH array) and a • (BETA array) are p r o p e r l y in ­ U p o n r e t u r n i n g f r o m the subroutine, the integer variable N O O U is zero if the zeros are p r o p e r l y interlaced and set to one if the y are not p r o p e r l y interlaced. This completes the e x p l a n a t i o n of the 16 subroutines. The m a i n p r o g r a m of Figure 5,1 w ill n o w be discussed. First5 subroutine R D l U is. called to read the input dat a fro m cards. Since the elements of the IFR and ITO arrays are to be used later as subscripts and since these elements c o n t a i n the i n ­ teger O 9 each of the elements are incremented b y I, Uext9 subroutine TROOP? is called to determine and store the .appli­ cable 2-trees of the input network. The A L P H and BETA a r ­ rays are set equal to the A L B A R and BitBAR arrays r e s p e c t i v e ­ ly, Subroutines S T U T H Z 9 E V L V T R , SEUSlT and E V O B are then called in turn, and it is seen that at this po i n t the value of the objective f u n ction of the ne t w o r k init i a l l y input has b e e n ■computed along w i t h the element values of the n e t ­ work, V O B A R is set equal to t h e current value of the ob ­ jective f u n c t i o n and an output h e a d i n g for the initial n e t ­ w o r k is printed. Subroutine DEPOT is next called to print —82— out the computed parameters of the initial network. The r e ­ m a i n i n g p o r t i o n of the m a i n p r o g r a m is concerned with the improvement of the initial n e t w o r k "by the method of steep­ est d e s c e n t o As each improvement to the net w o r k is accom­ plished, the value of the r e s u l t i n g objective function is typed out if sense switch I is turned on. The pr o g r a m is terminated automat i c a l l y if the objective f u n c t i o n im p r o v e ­ ment obtainable for a given set of gradients is less t han 0 , 0 5 % of its previous value. If it is desired to terminate the p r o g r a m before this c o n d i t i o n is f u l f i l l e d , sense switch 3 is turned on. U p o n termination, subroutine DMPOT is c a l l ­ ed to print out the parameters of the final improved n e t ­ wor k and the program,is terminated. The method of p r e p a r i n g the input data will n o w be ex­ plained, R e f e r to the expla n a t i o n of subroutine RDIET for a def i n i t i o n of the variables to be discussed below. The first input card is used to input the 7 integer variables ETOD9 ETELP9 ETO9 I B 9 I J 9 KB and ETELM, These are input as 7-two digit, numbers w h i c h o c cupy the first 14- columns of the first input card. ETTYP array, Card n u m b e r 2 is used to input the A m a x i m u m of 30-two digit numbers m a y occupy the first 60 columns of this card. The same format is used to input the IPR ar r a y on card numb e r 3 and the ITO array on card n um b e r 4-, The first 50 columns of the next card are -83used to input five F l O „2 numbers corresponding to the values of W O D 9 E R D 9 E C S 9 W O E and W E E r e s p e c t i v e l y <> N e x t 9 the E E L M a r r a y is input b y p l a c i n g as m a n y as 8 F l O «2 values on a card u n t i l the array is i n p u t . This same format is used to input the E K 9 E Y 9 S P 9 A L B A E and BT B A B arrays in turn, 0. EXAMPLE NETWORK„ A n example is n o w p r esented to clarify the p r e p a r a t i o n of the input data and to d e monstrate the accomplishments wh i c h are possible w i t h the program= The t r a n s f e r function selected is the basic low-pass B u tterworth fourth-order transfer f u n ction T = ----------- ; -- ---- ----------------V ZL Z O 8^ + 2 .61328^+ 3 .4-14-3s^+2.6132s+!. (5, 11 ) It is k n o w n that the a p p l i c a t i o n of the synthesis procedure of S e c ti o n IV.D. will result in the 10 element network c o n ­ fi g u r a t i o n of Figure 5°2. The elements must be placed in the 'position s shown in order that the computer p r o g r a m operate properly. Node and element identifying numbers are properly- assigned a n d 9 since there are no elements connected across the input and output terminals of the network, dummy (aux­ iliary) elements are added to the network and assigned the i d e n t i fi c a t i o n numbers 11 and 12. This is done so that, the quantities IB and K B will have values other t h a n zero. An -84- DUMMY IO-L DUMMY Figure 5»2* 5.677F. C o n f i g u r a t io n of example n e t w o r k » 6 o 313F o - 7.324F.: VO L T A G E FOLLOWER 5.567Fo -p. 2.732F, Figure 5»3» rp _ v .363F, Optimized net w o r k for r e a l i z a t i o n of ____ » s 4 + 2 o6 1 3 2 s 5 + 3 o4-14-3s 2 + 2«6132 s +1° -85examinat i o n of Figure 5»2 enables the d e t e r m i n a t i o n of the followin g input variables. NOD=5 NELP=IO' N0=4 IB=Il IJ=5 KB= 12 NELM=IE NTYP(T)=O N T Y P (2)=0 N T Y P (3)=0 N T Y P (4)=0 NTYP(5)=1 NTYP(G)=I N TYP(7)=1 NTYP(S)=I NTYP(9)=1 NTYP(IO)=I NTYP(Il)=O NTYP(IE)=O IFB(I)=I I F E ( E ) =2 IPE(3)=3 IFE(4)=4 IFE(5)=1 . I F E ( G ) =2 IFE(7)=3 IFE(S)=3' IFE(9)=4 IFE(IO)=4 . IFE(Il)=I IFE(12)=5■ ITO( I ) =2 IT0(2)=3 I T O (3)=4 IT0(4)=0 ITO(5)=5 ITO(G)=O IT0(7)=5 ITO(S)=O ITO(9)=5 ITO(IO)=O ITO(Il)=O ITO(IE)=O The wei g h t i n g functions were i nitially set to I with the thought that these quantities could be altered after running t h r o u g h the c o m puter p r o g r a m once in order to obtain a feel for the magnitudes of the various •elements of the optimizet i o n parameter. It was found desirable to increase WEB and W E E to 100 and 10 respectively bec a u s e of the c h a r a c t e r i s t i c­ al l y low value of E D and E E in c o m p arison w i t h the other opti m i z a t l o n p a r a m e t e r elements. L i k e w i s e 9 the W K 1s and W T 1S were increased to 10 and 50 respectively. The rema i n i n g quantities w h i c h are to be input are set as f o l l o w s : -86- SP(I)=I0 . SP(2)=206132 SP(3)=3o4143 SP(4)=2o6132 SP(5)=10 ABBAR(I)=.0500 ALBAE(2)=o2500 ALBAR(3) = o3500 ALBAR(4)=4o0000 BTBAR (.1)= 01000 BTBAR(2) = 03000 These quantities are punched at their p r o p e r input card locations and are shown on lines 3 through 19 on page 122 of A p p e n d i x I I 0 below. 1. The results are interpreted and discussed See pages 123 and 124 for the printout. The results are printed out in two steps. Pirst9 the parameters of the initial net w o r k are printed 9 and the n the parameters of the final network. 2. The first item considered is the pole and zero l o ­ cations of tfte input admittance It is shown that the initial zero locations of 0 .059 0 .23 $ Oo 35 9, and 4.00 are shifted to the final zero lo ­ cations of 0.0644$ 0.4432$ I .04385 and 1.5467. The poles are shifted f r o m 0.1000$ 0.3000$ 2.2132 to 0 . 2 8 2 3 9 0.8960$ and 1.4348. and Note that these quantities are the distances fro m the origin to the l o c ation of the poles or zeros on the n e g a ­ tive real axis and not their actual location. 3. The next item considered is the coefficients of the input admittance The results show that the input admittance is shifted from 57 .143s^"+265.714s^ —87— +155<>286s 2 + 2 7 <.1 0 7 s + 1 /1 5 oO61 s 5 +39<»358s 2 + 13 o7 8 5 s + 1 o t o 2 1 „ 7 1 6 s Z|"+ 6 7 .2 7 7 s 5 +64.02 3 1 s 2 + 19 o3 9 9 s + 1 / 2 ,7 3 5 s 5 +7 o 1 9 9 s 2 +5 o 353s + 1 o 4». N e x t 9 the sensitivities of the coefficients of P(s) wit h respect to variations in the network element values are printed o u t « There are too m a n y of these quantities to be listed individually hereo As an example $ note that the optimization process changes the sensitivity of coefficient 3 wit h respect to element 10 from Q 0534 to 0 »309 = 5o Note that the capacitance d i s t ribution p arameter is reduced f rom 1160 to 33 °87 <> the r e s i stance d i s ­ t r i b u t i o n p a r a m e t e r f r o m 0*0861 to 0.00622, and the -total capac i t y of the net w o r k is reduced from 89°39 to 2 7 o9 8 o The capacitance and resistance ratios are reduced f rom 308.74 to 20.18 and 4*337 to 1*381 respectively* 6. The sensitivities of the coefficients of P(s) w i t h respect to the voltage follower gain are next c o n ­ sidered* Notice that the sensitivity of c o effi­ cient 2 to a one per c e n t vari a t i o n in the voltage follower gain is decreased from 0 *207% for the initial net w o r k to 0 *0926% for the final network* Finally, the s e n s i t i v i t y of the coefficients of P(s) w i t h respect to the voltage f o l lower input —88— conductance are c o n s i d e r e d o Hotice that the sen-n sitivity of c o e f f i c i e n t 3 to a voltage follower w i t h an input c o nductance equal to one percent of the r e c i procal of the total-resistance of the p a s 'sive p o r t i o n of the n e t w o r k is decreased from 0 .395 # to 0,0719#. 7. At.'this point, the sensitivities and other ele-ments of the opt i m i z a t i o n para m e t e r have "been .printed o u t . . The next item' to he considered is the element values of the associated n e t w o r k s . Notice that the 35*778 farad capacitor of the i n i ­ tial n e t w o r k has b e e n r educed to 6.312 farads in the final network. 8. The final item considered is the value of the o p ­ t i m i z a t i o n parameter. Notice that it has b e e n d e ­ creased f rom an initial value of 1884.5 to 126 .58 . This, of c o u r s e , m a y be considered a worthwhile i m ­ p rovement . The final optimized n e t w o r k is shown in -Figure 5»5» Several observations and recommendations w i t h regard to the optimi z a t i o n computer p r o g r a m w i l l n o w be made. First of all, it was observed that the optimization p a r a m e t e r c o n ­ verged to the s a m e .value for several starting points. This s u g g e s t s , a l t h o u g h it c e r t a i n l y does not p r o v e , that the —89— m i n i m u m obtained is a global m i n i m u m and not a local m i n i ­ mum. A l t h o u g h the opti m i z a t i o n p a r a m e t e r converged to the same value for each starting point, the rate of convergence is g r e at l y influenced b y the starting point. It is felt that it would be advantageous to ca r r y out an initial search technique w h i c h would r a n d o m l y generate a n u m b e r of start­ ing points. ■That n e t w o r k w h i c h results in the lowest value of optim i z a t i o n p a r a m e t e r would t h e n be selected as the i n i ­ tial n e t w o r k of the o p t i m i z a t i o n program. A f a ctor w h i c h was found to influence g r e a t l y the rate of convergence is the w e i g h t i n g factors applied to the g r a ­ dients . In the present case, it was found most advantageous to weight the gradients b y the magnitude squared of their associated independent parameters. tions, Z o r other c o n f i g u r a ­ it m a y be desirable to change this to a weig h t i n g of the gradients b y the m agnitude or some other f u n c t i o n of its associated independent parameter. A n y desired change is carried out b y altering statements 3 and 4 of subroutine EZGEAD. VI, P R A C T I C A L C O N S I D ERATIONS S e c t i o n A of this chapter will consider the effect of the departure of the voltage f o l lower from the ideally assumed conditions of zero input admittance and u n i t y v o l t ­ age gain. S e c t i o n B will he concerned w i t h the design of voltage follower networks and S e c t i o n C will p resent the experimental results of some example n e t w o r k s . A. T H E NON-IDEAL- V O L T A G E F O L L O W E R As s een in Eigure ! . I 9 if the voltage f o l lower p o s ­ sesses a finite input admittance Y ^ 9 this admittance m a y he considered to appear across the input terminals of the p a s sive RC network. If Y is assumed to consist of a con- ductance Ga in p a r allel w ith a susceptance Ca , t hen Y a =Ga +Ga s and the characteristic equation D(s) of the RG net w o r k is unchanged. T h u s 9 the q u a ntity N ^ - N ^ the quantity G D+C Ds:. is increased h y One w a y in which the effect of Y m a y he t a k e n into account is the following: I. Synthesize a p assive n e t w o r k for the transfer f u n ction T =A(s)/P(s) as t h ough Y & does not exist. One of the results of this step is the d e t e r m i n a ­ tio n of the admittance scaling factor K and the characteristic p o l y n o m i a l D(s). of the network. 2. N o w repeat the synthesis procedure for the trans- -91fer f u n c t i o n T = A ( s ) / P ' (s) where P 1(s)=P(s) V -G-a B(s)-C^sD(s) 0 If the K and the D(s) of the n e w n e t w o r k are the same as those of the p r e v i o u s ­ ly synthesized network, Ty=A(s)/P(s) t hen the t r a n s f e r function is o b t a i n e d .w h e n the second network is used in c o n j u n c t i o n w i t h a voltage follower whose input admittance is Y^=G^+C^s. To take account of,a v oltage follower g ain k other t h a n u n i t y reca l l that the v oltage trans f e r funct i o n T^ is from E q u a t i o n (1.5) 12 (6 .1 ) N 11 kH13 This equation m a y be interpreted in the t e r m i n o l o g y of the current dis t r i b u t i o n dia g r a m of Figure 4.1 As thou g h the quantity kE(s) is subtracted fro m the current of terminal I to determine P(s) instead of the quantity E(s). An alter­ nate current di a g r a m r e p r e s e n t a t i o n is shown in Figure 6.1. If the q u a ntity P'(s) is set equal to ( P ( s ) -(l-kE(s)) and the usua l synthesis procedures are applied, t h e n the r e s u l t ­ ing passive network, w h e n used in conjunction w i t h a vol t a g e follower whose gain is k, will p roduce a t r a n s f e r function Ty=A(s)/P(s). B. V O L T A G E F O L L O W E R CIRCUITS - 9 2 - P' (s) + (l-k)E(s) +kE(s) Figure 6.1. Current d i s t r i b u t i o n diagram for non- u n i t y voltage follower gain. +30 v 2N2222 Figure 6.2. 2N2222 Voltage follower of Reference $2 -93This s ection is concerned w ith the p r o b l e m of d e s i g n ­ ing p r ac t i c a l voltage f o l lower c i r c u i t s e mentioned 9 the cathode, As p r e v i o u s l y emitter or source f o l lower circuit offers a simple ap p r o x i m a t i o n to a voltage follower* Ex­ cept for the less critical applications, however, these circuits have b e e n found to depart too far from the ideal to p e r f o r m satisfactorily. It will be shown that voltage followers which v e r y clo s e l y approximate the ideal m a y be designed. The first thought that occurs is that a two stage am ­ p l i f i e r u s i n g the field effect t r a n sistor as an input stage would make an ideal network. Analysis of such a network r e ­ veals, h o w e v e r , that while extr e m e l y hig h input impedances are a t t a i n a b l e , a g a i n of the order of 0.98 is all that is possible. A r e v i e w of the literature does not immediately reveal a simple s a t i sfactory two stage voltage follower. The two stage voltage follower to be presented here, however, is one which is adopted from the literature. circuit (32) is shown in Figure 6.1. The original This circuit has an input impedance of ap p r o x i m a t e l y $ 0,000 ohms and a gain of a p p r oxi m a t e l y 0.998. The altered network shown in Figure 6.2 has a m uch h i g h e r input impedance and a g a i n m u c h closer to unity. This improvement is accomplished b y substituting a field effect t r a n sistor for a conventional transistor. -94— Notice also that the c i r c u i t r y of the altered net w o r k is g r e atly simplified and that the low frequency response has b e e n extended d o w n to d-c„ operati o n of the two networks, To gain insight into the consider the system d i a ­ gra m of Figure 6„3 w h i c h is valid for b o t h c i r c u i t s e and Ag r epresent the v oltage gains of the first and sec­ ond stages r e s p e c t i v e l y e Note that the first stage of each amplifier is operated in the common base c o n f i g u r a ­ t i o n but that the load impedance for the stage in Figure 6 ®I is low in co m p a r i s o n w i t h that for the corresponding stage in Figure S e 2» Since the voltage gai n of the c o m ­ m o n base config u r a t i o n is p r o p o r t i o n a l to load i m p e d a n c e 9 A ^ . for the first n e t w o r k is m u c h lower t han A^ for the second network. Straight forward analysis of Figure 6.3 show that A = ®out ' I CD ein A 1A2 A ^ A g +l—Ag zi (6e2) (6e3) ‘ 1 In where is the emitter-base impedance of the input stage. Note that Ag is essentially the voltage g a i n ■of an emitter follower and m a y be considered a p p r o x i m a t e l y 0.93= Typical values of A^ are 35 and 500 for the circuits of Figures 6.1 and 6.2 respectively. quations (6 .2 ) and S u b s t i t u t i n g typical values into e- (6 .3 ) r esults in A = O . 9985 and A ^ = 23 g300 IOM 2N4-04 6+ e X - 1004 X / - in o— 10K> _J= 'out — =0 +10 v, Figure S„3 o Altered network, out Figure 6.4 S y stem dia g r a m for networks of Figures 6.2 & 6.3° -96— ohms for Figure 6 d . The corresp o n d i n g results for the n e t w o r k of Figure 6.2 are A ~ 0.9999 and Ziri ^ $$0,000 ohms. T h u s , improvement has b e e n accomplished b y a l l owing the voltage g ain to a p p roach the m a x i m u m attainable from a conventional t r a n s i s t o r .■ A three stage Voltage follo w e r w h i c h follows n a t u r a l ­ ly fro m the net w o r k of Figure 6.2 is shown in Figure 6 .5 . Actually, the net w o r k contains fou r t r a n s i s t o r s . Since one of the transistors only b ehaves as a current source to p r o ­ p e r l y bias the first stage, the net w o r k is classified as a three stage network. The result accomplished b y this n e t ­ wor k is to make Ag of Figure 6.3 approach muc h closer to unity. Notice that the low f r e q u e n c y response of this n e t ­ wo r k also extends to'd-c. This net w o r k will be used in the experimental v e r i f i c a t i o n of S e c t i o n C. 0. EXPERIMENTAL R E S U L T S „ This section will present the experimental results of 'three transfer functions realized b y the methods of this report. The three transfer functions include a second-or­ der band-pass t r a nsfer function, a fourth-order low-pass tran s f e r f u n ction and a fou r t h - o r d e r band-pass transfer function. These transfer functions in normalized form are -97- -20 v« 15K< 2N3114 2N4-04 O+ X-1004 h °— I +o 'in 9 -20 Vc 'out IOK 8.2E 6+10 v, 2N3114 IOK 6 -20 Vc Figure 6.3« Three stage v oltage follower net w o r k -98respectivelj T = ■ - 2 °-^ -.s + O o l s +1 T = “TT v (GA) % ' : p ' s%2o6132s:)+3o4145s^+2e6132s+l . (6 o5) 2 V v - % ----------------- A s — :------------------- sAlo414-s^+3s^+lo4-14s+l C s - 6 ) The tran s f e r f u n c t i o n (6.4) corresponds to a band-pass f u n c ­ t i o n whose center frequency is I r a d i a n per second and whose pass band is 0.1 r a d i a n p e r second. E q u ation (6.5) is the low-pass f o u r t h-order m a x i m a l l y flat (Butterworth) charac- - teristic whose cutoff f r e q u e n c y is one r a dian p e r second. Equa t i o n (6 .6 ) is the fou r t h - o r d e r m a x i m a l l y flat band-pass transfer f u n c t i o n whose center freq u e n c y is one r a dian p e r second and whose pass— band is one r a d i a n p e r second. This means that the u p p e r cutoff f r e q u e n c y is 1.619 radians p e r second and the lower cutoff f r e q u e n c y is 0.619 radians p e r second. A n explicit synthesis p r o c e d u r e for the second-order b a nd-pas s t r a nsfer f u n ction was not presented in Chapter IV. One of the networks of S a l l e n and K e y m a y be utili z e d or % more simply, the current d i s t r i b u t i o n d iagram for the f a m i l ­ -99iar twin-T EC net w o r k m a y be computed as shown in Eigure 6 *6 , r If 0 9 kjj, and are chos e n 0.8, 1 . 6 7 9 and I respectively, a n e t w o r k w i t h p o s i t i v e elements results which can be c o n ­ verted to the form of Figure 6.7 for use in c o njunction w i t h the (ECYE) to realize the t r a n s f e r function of (6=4), This n e t w o r k is shown in Figure 6 .8 = The fou r t h - o r d e r low-pass f u n c t i o n of E q u a t i o n (6=5) is the same as that used in the c o m puter o p t i m ization example of Chapter Y= The n e t w o r k used in the present case is one that was obtained f r o m the c o m puter pro g r a m b e fore the c o m ­ p u t e r p r o g r a m was co m p l e t e l y debugged. The n e t w o r k is, not an opti m u m one although it turns out to be v e r y n e a r l y op ­ timum. The p r ocedure of IY E m a y be applied in the case of the transfe r f u n ction of E q u a t i o n (6 .6 ). A p p l i c a t i o n of this p r ocedur e results in the n e t w o r k of Figure 6 .6 . wor k realizes This n e t ­ (6 .6 ) w i t h i n a constant multiplier. The realized networks will be impedance and frequency scaled to typical values. The circuit for the second-order band-pas s transfer f u n c t i o n was impedance scaled to $ 00,000 ohms and freq u e n c y scaled to 1$0 cycles p e r second. sultant (ECYF) co n f i g u r a t i o n is shown in Figure 6.7« The r e ­ — 1 0 0 - Figure 6 .6 . Current d i st r i b u t i o n for tw i n - T network. R(s)=s +1 Figure 6.7 Network for r e a l i z a t i o n of T. 0 .1s s +.Is +1 -101- <66t>6K 600K ,00443 F. o-- W V - VOL T A G E 180 K F O L LOWER ydut .0296 F, ,0177 S’■6 0 K Figure 6.8./ (ROVF) for r e a l i z a t i o n of second-order bandpass t r a nsfer f u n c t i o n denormalized to 300)000 ohms and 150 cps. ,0104 F. - W + - ,0132 F. ,0104 F, 286E -AZW- m - "VOLTAGE" FOLLOWER ,012 F, -4,440pF. 780pF, "out. Figure 6.9« (EGVF) for r e a l i z a t i o n of fourth-order low-pass trans f e r f u n c t i o n denormalized to I m e gohm and 89 cps. - 1 0 2 - The circuit for the f o u r t h-order low-pass transfer func t i o n is impedance scaled to I m e g o h m and freq u e n c y scaled to 85 cps» This denormalized n etwork is shown in Figure 6 .8 . The circuit for the fou r t h - o r d e r hand-pass transfer func t i o n is impedance scaled to 500,000 ohms and f requency scaled to 5 9 »8 cps, The denormalized network is shown in Figure 6.9. The networks of Figures 6.7, 6 .8 , and 6.9 were built and tested us i n g the voltage f o l lower circuit of Figure 6 .4-0 A com p a r i s o n of theor e t i c a l and experimental results for these networks are shown in Tables 6.1 thr o u g h 6.5° - £ 0T114.3% ,1133 F. 2?. 6K' 0-- W v Hf ,0307 Fe if 292K — W\A- VOLTAGE .01362 F, FOLL O W E R <415% ,00884 Fe W W ,1359 %. % 85 .6E 17 .35% 'out .0862 F, Figure 6.10. (RCVF) for r e a l i z a t i o n of fourth-order band-pass transfer function denormalized to 500^000 ohms and 39=8 cps. 104 T A B L E 6.1 FREQUEinICY-ATTENUAT IOM TAu UL a T IOiQ OF EXPERIMENTAL AiMO THEORETICAL VALUES FOR NETWORK OF FIGURE 6. ti FREQUENCY (CPS) 17.0 25.0 38.0 55.0 75.0 96.0 109.0 119.1 125.1 130 =5 134.8 138.5 140.1 141.7 142.6 143.3 144.2 144.9 145.8 146.8 147.9 150.2 152.6 154.0. 154.9 155.8 156.6 157.5 158 =3 159.3 161.1 163.0 167.3 172.8 . 180.4 189.8 20 7.9 236.1 303.1 4u9.2 618.2 939.0 1476.0 DB ATTENUATION (EXPERIMENTAL) 40.00 36.00 32.00 28.00 24.00 20.00 .17.00 14.00 12.00 10.00 8.00 6.00 5.00 4.00 3.50 ' 3.00 2.50 2.00 1.50 I .00 .50 0.00 .50 I .00 1.50 2.00 2.50 3.00 3.50 4.00 5.00 6.00 8.00 10.00 12.00 14.0 17.00 2 0.00 24.00 28.00 32.00 36.00 4 0.00 DB ATTENUATION (THEORETICAL I 38.80 35.31 31.35 27.46 23.54 19.35 16.35 13.55 11.56 9.44 7.46 5.50 4.57 3.61 3.06 2.63 2.10 1.69 1.21 .74 .33 0.00 .48 1.06 1.50 1.97 2.41 . 2.90 DB ERROR 1.198 .680 .6 46 .531 .458 .649 •6 46 .446 .438 .551 .530 .497 .424 .387 .436 .361 .398 .300 .285 .259 .167 -.003 .015 - .0 6 2 -.002 .024 .089 .093 3.34 .152 3.89 .109 «163 4.83 5.76 7.62 7.57 I 1.69 13.72 16.54 19.49 23.63 27.47 31.77 35.70 39.77 .237 .3 77 .427 .303 .277 .452 .500 .311 .328 .223 .291 .229 105 TABLE 6.2 FREQUENCY-ATTENUATION TABULATION OF EXPERIMENTAL AND THEORETICAL VALUES FOR NETWORK OF FIGURE 6. 9 FREQUENCY (CPS) 50.0 65.7 71.8 75.9 79.5 82.4 84.9 87.2 89.7 93.8 97.5 105.0 113.0 120.0 128.0 140.0 153.0 172.0 195.0 222.0 267.0 DB ATTENUATION IEXPERIMENTAL) DB ATTENUAT ION ITHEORETICAL) ‘ DB ERROR 0.00 .50 a06 .52 1.00 1.47 -.062 -.021 1.00 1.50 2.00 2.50 3.00 3.50 4.00 5.00 6.00 8.00 10.00 12.00 14.00 17.00 20.00 24.00 28.00 32.00 3 6.00 2.00 2.50 2.99 — .001 .025 -.002 -.004 .009 3.47 .022 4.04 5.05 6.01 8.07 10.31 12.24 14.38 17.41 -.045 — .050 -.017 -.077 -.316 20.46 24.50 . 28.85 3 3.35 39.76 -.248 -.384 -.416 -.461 -.504 -.855 -1.356 -3.767 106 TABLE 6.3 FREQUENCY-ATTENUATION TABULATION OF EXPERIMENTAL AND THEORETICAL-VALUES FOR NETWORK OF FIGURE 6.10 FREQUENCY (CPS) 4.9 5.9 7.6 9.3 11.3 13.3 15.1 16.4 13.1 19.7 21.3 22.2 23.1 ' 23.8 24.3 24.9 25.6 ■ 26.5 27.5 29.0 39.8 50.8 54.6 56.9 59.6 61.5 63.0 65.0 66.6 69.8 72.7 79.1 86.1 94.1 102.5 118.0 136.0 166.0 206.0 255.0 319.0 401.0 -- DB ATTENUATION (EXPERIMENTAL) DB ATTENUATION ITHEORETICAL I DB ERROR 35.50 32 .00 28.00 24.00 20.00 17.00 14.00 12.00 10.00 8.00 6.00 5.00 4.00 3..50 3.00 2.50 2.00 1.50 I.00 .50 0.00 .50 1.00 I .50 2.00 2.50 3.00 3.50 4.00 5.00 6.00 8.00 10.00 12.00 14.00 17.00 20.00 24.00 28.00 32.00 36.00 40.00 36.12 32.77 28.12 24.29 20.45 17.06 14.30 12.42 10.10 8.06 6.19 5.21 4.31 3.67 3.24 2.77 2.2 7 1.72 1.22 .68 0.00 .24 .68 1.08 1.68 2.17 2.59 3.19 3.68 4.71 5.65 7.67 9.71 11.81 13.77 16.87 19.83 23.79 27.90. 31.84 35.88 39.95 -.623 -.777 -.124 -.296 -.451 — a069 — .300 -.424 -.107 -.069 -.190 -.218 -.318 -.175 -.249 -.275 -.278 -.226 -.228 — aI87 0.000 .251 .315 .414 .316 .325 .404 I «308 .310 .284 .345 «326 ; .282 .180 »220 ■,164: - m t m . .041: YIIo A, CONCLUSION AND RECOMMENDATIONS CON C L U S I O N This thesis has described and demonstrated the f e a s i ­ b i l i t y of u t i l i z i n g the grounded voltage f o l l o w e r as the active p o r t i o n of an active RC network. Chapter III es­ tablished the n e c e s s a r y and sufficient conditions for the r e a l i z a t i o n of the voltage t r a n s f e r function of such a d e ­ vice (the R C Y F ), cedures for the Chapter IY described some synthesis p r o ­ (RCYF) and Chapter V demonstrated that the digital computer m a y be u t i lized to optimize the synthe­ sized networks. Finally, Chapter YI gave a demons t r a t i o n of several transfer functions realized b y the (RCYF) active n e t w o r k wh i c h indicates the p r a c t i c a l value of the network b y showing excellent agreement b e t w e e n predicted and actual results. Indeed, the results of b o t h the theoretical and experimental investi g a t i o n lead one to conclude that the (RCVF) is a h i g h l y u s eful and p r a c t i c a l network, B. RECOMMENDATIONS. This study has resulted in a number of possibilities for future wor k b e c o m i n g apparent. Recommendations for future w ork are listed below. I. The first, and p r o b a b l y the most o b v i o u s , recom- -108m e n d a t i o n that comes to mind is the further i n v e s ­ t i g a t i o n of synthesis t e c h n i q u e s „ Although Chap­ ter III showed that a t r a nsfer f u n ction which meets specified n e c e s s a r y conditions m a y always he realized 9 the synthesis of optimum networks is another matter. Thus $ the further i n v estigation of networks for specific classes of t r a nsfer f u n c ­ tions is indicated. A n area wh i c h was not touched on in this i n v e s t i g a ­ t i o n but appears inter e s t i n g is that of the p o s s i ­ b i l i t y of v a r y i n g t r a n s f e r functions b y switching. The concept of the current dis t r i b u t i o n diagram offers a great deal of insight into the problem. A simple example of the possibilities for s w itch­ ing m a y be seen b y examining the n e t w o r k of Figure 6.7'. This n e t w o r k is used in conju n c t i o n wit h the voltage f o l l o w e r to realize the t r a n s f e r function o T^= a^s/pgS +a^s+pQ w h e r e a^= 0 »l and Pg=PQ=Io Notice that a^ m a y be halved -m e r e l y b y splitting the 2- ohm r e s istor c a r r y i n g the current 0 .1s into two 4 -ohm resistors and switching one of them over to the terminal associated w ith the current E(s). The result of this o p e r a t i o n is to cut the b a n d ­ w i d t h of the band - p a s s f u n ction in half. The above example represents onl y one p o s s i b i l i t y of a n u m b er -109w h i c h r e a d i l y come to m i n d „ 3<, The next r e c o m m e n d a t i o n concerns the o p t i mization of networks b y means of the digital computer. It is obvious that n e t w o r k optimization is an ex­ t r e m e l y c o m p l e x p r o b l e m and that it has only b e e n touched u p o n in this investigation. t i m i z a t i o n method O n l y one o p ­ (the method of steepest descent)' has b e e p attempted on a single class of networks. The opti m i z a t i o n att Ciipt was successful from the standpoint that significant improvement was a c c o m ­ plished, but the p r o b l e m is important enough to d e ­ serve a great d eal,of further study. It is s p e c i ­ f i c a l l y r e commended that the approach of continousIy equivalent networks (20 ,53 ) be investigated. This a p p r o a c h is not only concerned w i t h the o p t i ­ m i z a t i o n of a gi v e n n e t w o r k configu r a t i o n but wit h the ge n e r a t i o n of n e w configurations. 4* An o t h e r important area for further invest i g a t i o n is that concerned w i t h p r a c t i c a l considerations. Such considerations would include temperature effects and compensation, component selection, voltage follower circuitry, the max i m u m impedance level of a network, the m i n i m u m freq u e n c y a l l o w a b l e , e t c . A n o t h e r important ite m u n d e r this h e a d i n g would be the effect of joining two or more (EGVT) networks -IlOr- togetihero .These n e t works could be connected in cascade or. some other desirable configuration. 5. The final r e c o m m e n d a t i o n concerns the driving point impedance of the (ECVF)» This thesis has done n o t h i n g w i t h this p o s s i b l y important subject. P r e l i m i n a r y i n v e s t i g a t i o n has indicated that driving point f u n c t i o n synthesis u t i l i z i n g the v oltage f o l lower m a y be more difficult t h a n t r a n s ­ fer f u n c t i o n synthesis. It is felt that the p r o b ­ lem deserves a separate paper. \ PLEASE NOTE: This is not original copy. Appendix contains pages with extremely fine print. Filmed as received. University Microfilms, Inc. APPENDIX I O P T I M I Z A T I O N C O M PUTER P R O G R A M ' L I S T I N G n ZZJOS 5 ZZFOR 5 *LDISKCH0NE D I M E N S I O N L D O O J , N T Y P t 3 J I , V T R ( S O O ) , S E L M I 1 1 , 3 0 ) , S P ( 1 1 ) » W E L M ( 1 1 ,30 )» I E L E M t 3 0 ) , D T ( I l ) , W K ( I l ) , W Y t 1 1 ) , A L B A R ( I O ) , B T B A R t 1 0 ) , G R A D ( Z O ) , I F R (30) 2 , 1 T O ( 3 0 ) , I T R ( I O O O ) , F N C O ( I l ) , F D C O ( I l ) , B T ( I l ) ,A L P H ( 1 0 ) , B E T A ( 1 0 ) , F T C O 3( 11 ) C O M M O N N O D ,N TT »N E L P *N T S , N O »N S P ,L D »NT Y P »V T R »S E L M *S P , W E L M , W C D , W R D *W C 1 5 ' » W C R , W R R , E L E M , D T , W K ,W Y , N VB ,A L B A R , B T B A R ♦ V 0 3 A R , G R A D , G R M A X , N G R I , S T P , 2 N 0 M I , N O M T , N C O N , IJ , I B ,K B , I F R , IT O , N E L M , N IT R , IT R , N F A R ,F N C 0 , F D C 0 ^ B T ,A L 3 P H , B E T A , F T C O , C M A X ,CM IN ,R M A X ,R M IN , C D ,R D , C S U M , V O B J CALL RDIN DO I 1=1,NELM IFR(I)=IFR(I)+! 1 ITO(I)=ITO(I)+! CALL TRCOM D O 2 I = I ,NO 2 A L P H ( I ) = A L B A R t 11 DO 3 I=I,NOMT 3 BETA(I)=BTBARt I) C A L L L IN K ( C H T W O ) END IS »WCR»WRR»ELEM,DT,WK »WY,NVB,ALBAR,8T3A R ,VOBAR,GRAD,GRMAX,NGRI»STP, 2N0MI,NOMT,NCON,IJ ,IB,KB »IF R ,ITO»N£LM,NITR,ITR,NFAR,FNCO,FDCO,BT,AL 3PH,BETA1FTCO,CMAX,CMIN,RMAX1RM IN,CD,RD,C5UM,V0BJ I CALL DMPOT CALL LINK(CHFOUR) END ZZJ06 5 ZZFOR 5 ♦LDISKCHFOUR D I M E N S I O N L D (30) , N T Y P t 3 0 ) ♦ V P R (500) , S E L M ( 1 1 ,30 I , S P ( I l ) , W E L M ( 1 1 , 3 0 ) , 1 E L E M I 3 0 ) , D T t 1 1 ) , W K t 1 1 ) » W Y ( 1 1 ) , A L B A R ( I O ) , B T B A R ( I O ) , G R A D ( Z O ) »1 F R (30) 2 , 1 T O ( 3 0 ) , I T R ( I O O O ) , F N C O ( I l ) , F D C O I 1 1 ) , B T t 1 1 ) , A L P H ( I O ) , B E T A t 1 0 ) ,F T C O 3( 11 ) COMMON NO D , N T T, NE LP fN TS iNO , N S P f L D ,NTY P »VTR»SELM,SP»WELM»WCD*WRD»WC IS »W C R , W R R » E L E M » D T , W K , W Y , N V B ,A L B A R ,B T B A R , V O B A R ,G R A D , G R M A X , N G R I , S T P , 2 N 0 M I » N O M T , N C O N , IJ , I B , K B , !F R, I T O , N E L M , N I T R , I T R , N F A R , F N C O , F D C O f B T t A L 3 P H , B E T A ,F T C O f C M A X f C M I N ,R M A X ,RM I N ,C D ,R D , C S U M , V O B J 21 S T M U L = I , NY =O N f =o VORJ=VOBAR CALL EVGRAD CALL STCOMP ZZJOB 5 ZZFOR 5 * L D ISKCHTWO COMMON NOD,NTT,NELP,NTS,NO,NSP,LD,NTYP,VTR,SELM,SP,WELM,WCD,WRD,WC IS»WCR»WRR*ELEM»DT,W K »W Y *NVB»ALBAR,8TBAR,VOBAR»GRAD,GRMAX,NGRI»STP* 2N0MI,NOMTtNCON,IJ,IB,KB ,'IFR,I TO,NELM,N ITR,ITR,NFAR.FNCO,FDCO1BT,AL 3P H ,BETA,FTCO♦CM A X ,CM IN,RMAX,RMIN,CD,R D »CSUM»VOBJ CALL SYNTHZ CALL EVLVTR CALL SENSIT CALL EVOB VOBAR=VOBJ PRINT I I FORMATtlHl,6H*******26X,15HINITIAL NETWORK,27X,6H**»***) IF (SENSE SWITCH I) 17,18 17 PUNCH 19 19 FORMAT!6H******,26X,15HINITIAL NETWORK,26X,6H******) 10 CONTINUE CALL LINKtCHTRI) 22 201 76 202 2 3 I 104 41 106 END 12 ZZJOB 5 ZZFOR 5 ♦LDISKCHTRI D I M E N S I O N L D ( 3 0 ) , N T Y P t 3 0 ) * V T R ( 5 0 0 ) , S E L M t1 1 , 3 0 ) , S P ( 1 1 ) , W E L M t 1 1 , 3 0 ) , lELEM( -3 0) ,D T (1 1) , WKt 1 1 ) , W Y ( 1 1 1, A L B A R t 1 0 ) , B T B A R t 1 0 ) , GRAD(ZO)', IFRt 30) 2 , ITOt3 0 ) ,I TR t1 0 0 0 ) , F N C O t 1 1 ) ,F D C O t 1 1 ) , B Tt1 1 ) ,A L P H t 1 0 ) ,B ET At1 0 ) ,FTCO 3( 11 ) CO M M O N N O D, NT T,N E L P , N T S , N O ,N SP ,L D, NT YP ,V TR ,S EL M, SP ,W EL M, WC D,WRDeWC 13 27 35 36 VLAS=VOBJ STBAR=STP STBES=O, STP=STBAR*STMUL CALL TKSTP I F t S E N S E S W I T C H 2) 2 0 1 , 2 0 2 TYPE 76,VOBJ FORMAT(2HTO»E17.8»/) IF(NCON) 1,2,1 IF(VOBJ-VLAS) 3,1,1 NY=NY+! STBES=StP VLAS=VOBJ STMUL=Z.*STMUL I F (N F) 1 0 6 , 2 2 , 1 0 6 IF(NF-IO) 10 4, 10 4, 10 6 I F (N Y) 4 1 , 4 1 , 1 0 6 NF=NF+1 ST M U L = O . 5*STMUL G O T O 22 STP=STBES CALL TKSTP D O 12 1 = 1 , NO ALBARtn=ALPH(I) D O 13 1 = 1 , N O M I BTBARtn=BETA(I) VOBAR=VOBJ I F t S E N S E S W I T C H 3) 3 5 , 2 7 D E C S = ( tV O R U - V O B J )/ V O R J ) * 1 0 0 0 0 , D E C s = A B S F (D E C S ) IFtDECSr-S.) 3 5 , 3 5 , 2 1 P R I N T 36 FORMAT t7H1****** »27X.13HFINAL NETWORK,26X»6H******) I F t S E N S E S W I T C H I) 3 0 1 , 3 0 2 112 D I M E N S I O N L D (.30) ,NTY Pt 30) , V T R ( S O O ) ,S E L M (1 1 , 3 0 ) , S P ( I l ) , W E L M (1 1 ,30 ) * IEL EM(30),DT(Il),WKdll,WY(Il),ALBAR(IO),BTBAR(IO),GRAO(ZO),IFR(SO) 2 , I T 0 ( 3 0 ) * I T R ( 1 0 0 0 ) , F N C O I 1 1 ) , F D C O I 1 1 ) , B T ( 1 1 ) , A L P H l 1 0 ) , B E T A ( 1 0 ) ,FTCO 31 11 ) VOBJ=VOSAR 30 1 P U N C H 3 0 3 3 0 3 F O R M A T (6 H * * * * * * , 2 7 X , 1 3 H F I N A L N E T W O R K , 2 8 X , 6 H * * * * * * ) 302 CO N T I N U E CALL LINK(CHFIVE) END ZZJ06 5 ZZFOR 5 # L D IS K C H F IVE D I M E N S I O N L D ( 3 D ) , N T Y P t 3 J ) , V T R ( S O O ) , S E L M t 1 1 , 3 0 ) , S P ( I l ) »W E L M ( 1 1 * 3 0 ) # I E L E M ( 3 0 ) , D T ( I l ) , W K ( I l ) , W Y I 1 1 ) , A L B A R ( I O ) , B T B A R ( I O ) , G R A D ( 2 0 ) ,I F R (30) 2 , I T O ( 3 0 ) , I T R ( I C O O ) ,FNC.O(ll) , F D C O ( I l ) , B T ( I l ) , A L P H ( I O ) ,BET A! 10) , F T C O . 3( 1 1 ) C O M M O N N O D ,N T T , N E L P ,N T S , N O ,N S P , L D , N T Y P , V T R , S E L M ,S P ,W E L M , W C D » W R D » WC I S , W C R , W R R , E L E M , D T ,W K , W Y , N V U ,A L B A R , U T 3 A R ,V 0 3 A R , G R A D , G R M A X , N G R I »STP, 2 N 0 M I , N O M T , N C O N , IJ , 13 , K S ,I F R ,-I T O , N E L M , N IT R , I T R , N F A R , F N C O , F D C O , B T ,AL 3 P H , B ET A ,F T C O , C M A X ,CM IN , R M A X ,R M IN ,C D , R D ,C S U M , V O B J I CALL DMPOT C A L L EXIT END COMMON NOD,NTT,NELP,NTS,NO,NSP,LD,NTYP *VTR,SELM,SP,WELM,WCD,WRD,WC IS,WCR,WRRfELEM,DTiWK,W Y, NVBfALdAR,UTBAR,VOBAR,GRAD,GRMAXeNGR I,STP, 2NOMI ,NOMT,NCON,IJ,IB,KB,IFR,ITO,NELM,NITR,I TRfNFAR,FNCO,FDCO,BT,AL 3PH,BETA,FTCO,CMAXfCM IN,RMAX,R M INiCDfRD,CSUMtVOBJ B E T A (N O M I)=S P (N O ) DO I I=IfNOMT 1 BETA(N OM I) =B ET A( NO MI )- BE TA t I) I=I IF (A L P H I I )) 2 , 2 , 3 3 I Fl B E T A ! I )- A L P H I I ) ). 2 , 2 , 4 4 . J= I 1= 1+ 1 IF(ALPH(I)-BETAtJ)) 2,2,5 5 IF(I-NO) 3, 8, 3 8 NCON=O RETURN 2 NCON=I RETURN END ZZJOB 5 ZZFOR 5 # L D I SK SU BROUTINE DMPOT D I M E N S I O N L D ( 3 0 ) , N T Y P t 3 0 ) , V T R ( 5 0 0 ) , S E L M ( I I ♦ 3 0 ) , S R ( I I )• W E L M ( I I ,3 0) * 1 E L E M 1 3 0 ) , D T ( I l ) , W K ( I l ) » W Y ( 1 1 ) , A L B A R I 1 0 ) , B T B A R ( 1 0 ) , G R A D ( 2 0 ) , I F R I 30) 113 ZZJOB 5 ZZFOR 5 * L D IS K SUBROUTINE CONSTR D I M E N S I O N L D (30) , N T Y P I 3 0 ) , V T R ( S O O ) , S E L M l 1 1 , 3 0 ) , S P ( I l ) , W E L M I 1 1 , 3 0 ) » I E L E M ( 3 0 ) , D T ( I l ) , W K ( I l ) , W Y ( 1 1 ) , A L B A R t 1 0 ) , B T B A R I 1 0 ) , G R A D ( 2 0 ) , I F R (30) 2, ITOt 30) , I T R ( I O O O ) , F N C O ( H ) ,FDC OI 11) , G T ( I l ) , A L P H I 10) , B ET A! 10') , F T C O 3( 1 1 ) 2 , ( T O U T ) » IT R I I J O ) , r N C CI 11 I , FD C O f 11) , B T ( H ) ,A L P H ( 10 ), EETAI 10) , F T C O 3(11) C O M M O N N O D , N T T f N Z L P , N T S , NvO,.’-ISP,LI/,N T Y P , V T R t S E L M , S P , W E L M , W C C tW R D , W C IS, W C R ,WRR, E L E M tB T ,WK ,'VY,NVB , A L B A R , B T B A R , V O P A R , G R A D , G R M A X ,NGR I , S TP, 2 N 0 M I ,.NOuT , N CO N, IJ , IG , X B , I F P , I T O ,N E L M , N ITR , I T R , N F A R , F N C O , F D C O , B T , AL . 3 P H , 3 E T A , F T C 0 , C M A X , C-*. I N tK M A X , R M IN, C D , RB , C S U M tV O B J D I M E N S I O N M R E l 5 ) , J R C l D ) , V AL ID ) P R I N T 13 H F O R M A T Il H , / , I H , 1 3 X , 1 4 H Z E R 0 L O C A T I O N S ,2 6 X , 1 4 H P 0 L E L O C A T I O N S / ) I F IS E N S E S W I T C H I) 2 0 1 , 2 0 2 2T1 P U N C H 13 2 '2 CONTINUE D O 14 1 = 1 , N O N I I F IS E N S E S W I T C H I) 2 0 3 , 1 4 2 1 3 P U N C H I 5 ,ALPHI I ) ,B E T A I I ) 14 P R I N T 15 ,ALPl-H I ) ,BETA! I ) 15 F O R M A T I F2 fi,«20 , F 3 9 • 20,) PRINT 16,ALPH(NO) 16 F O K M A T ( F 2 8 , 2 0 » / ) - P R I N T 17 17 F O R M A T (2 0 X ,3 9 H C 0 E F F IC I E N T S OF IN PU T AD MI T T A N C E Y ( I , I ) , /9X» 2 2 H N U M E R 1 A T 0 R C O E F F I C I E N T S , 1 8 X , 2 4 H D E N 0 M IN A T O R C O E F F I C I E N T S / ) I F (S E N S E S W I T C H I) 2 0 4 , 2 0 5 2 0 4 P U N C H 1 6 , A L P r U NO). P U N C H 17 205 C O NT IN UE D O 21 I = 1 , N 0 IF ( S E N S E S W I T C H I) 2 0 6 , 2 1 2 0 6 P U N C H 1 5 , F T C O H ) ,FDCOI I) 21 P R I N T 1 5 , FTCOI I ) ,FDCOI I) PRINT 16 ,FTCO(NSP) P R I N T 24 24 F O R M A T I2tiX,23H O P T I M I Z h T ION P h R a M l T E R S / I N , 6 2 H S E N S IT IV I TY O F C O E F F I ICIENT M WITH RESPECT TO ELEMENT NUMBER J. /1 H ) M= I J= I IF (S E N S E S W I T C H I) 2 0 7 , 5 5 207 PUNCH 16,FTCO(NSP) P U N C H 24 55 D O 27 1 = 1 , 5 M R E ( I I=M J R E ( I I=J V A L ( I I = S E L M ( M tJ) IF(J-NELP) 2 b , 29,29 28 J = J + I G O T O 27 29 IF(M-NSP) 30,31,31 30 M = M + I J= I 27 CO NT IN UE 31 IF (I -S ) 3 2 , 3 3 , 3 3 32 D O 3 5 L= I ♦5 M R l (L ) = O JRE(L)=O 35 V A L ( L ) = O . 33 P R I N T 3 6 , 1 M R E ( I ) » J R C ( I ) , 1 =1 ,5 ) 36 F O R M A T ( 3 X » 5 ( 2 H M = , I 2 » 3 H » J = » I 2 » 7 X ) I P R I N T 3 7 , ( V A L I D , 1 = 1,5) 37 FO RM AT(5E16.5,/) I F t S E N S E S W I T C H I) 2 0 8 , 2 0 9 11 2 F O R M A T (2 3 H N E T W O R K E L E M E N T V A L U E S ) I F t S E N S E S W I T C H I) 2 1 6 , 2 1 7 2 1 6 P U N C H 112 217 CONTINUE D O 111 1 = 1 , N E L P I F t S E N S E S W I T C H I) 2 1 8 , 1 1 1 2 1 8 P U N C H 1 1 3 , I , E L E M t I) 111 P R I N T 1 1 3 , 1 ,ELE Mt 11 11 3 F O R M A T (9 H E L E M E N T ( , 1 2 , 2 H ) = , E l 8.8) 7 7 P R I N T 8 0 ,V O B J Bv F O R M A T t 2 8 X , 2 4 H 0 B J E C T I V £ F U N C T I O N V A L U E ,Z , 2 8 X , E 2 4 . 8 , Z I H » Z , 1 H ,1 QH * I * * * * * * * * * ,60 X ,9H**#-******) I F (S E N S E S W I T C H I) 2 1 9 , 2 2 0 219 PUNCH 80,VOBJ 220 CONTINUE RETURN END ZZJOB 5 ZZFOR 5 •LDISK SU BROUTINE EVGRAD D I M E N S I O N L D ( 30) , N T Y P t 3 0 ) , V T R ( S O O ) , S E L M t 1 1 , 3 0 ) , S P ( I l ) , W E L M (11 ,30 ) , I E L E M t 3 0 ) , D T t 1 1 ) , W K l 1 1 ) , W Y ( 1 1 ) , A L B A R ( I O ) , B T 3 A R ( 1 0 ) , G R A D ( 2 0 ) ,I FR (S O) 2 , 1 T O ( 3 0 ) , I T R ( I O O O ) , F N C O ( I l ) , F D C O t 1 1 ) » B T ( 11) , A L P H ( I O ) , B E T A ( I O ) , F T C O 3( 11 ) COMMON NOD,NTT,NE LP ,NT S,N O , NSP,LD,NTYP,VTR,SELM,SP,WELM,WCD,WRDtWC IS,WC Ri WR R,ELEM,DT,WK,WY ,N VB,ALBAR,BTBAR,VOBAR,GRAD,GRMAXiNGRI.STP, 2 N C M I , N C M T f N C C N f I J t I 3 , K B , IF R , I T O , N E L M f N I T R , I T R , N F A R , F N C O , F D C O , B T e A L 3PH,BETA,FTC0,CMAX,CMIN,RMAX.RMIN,CD,RD,CSUM,V03J DO I I=ItNO 1 ALPHI D = A L B A R l 1 1 00 2 1 = 1,NOMT 2 Ij ET A( I ) = B T B A R ( I ) 1 =I 5 A L P H l I )= 1 . 0 0 1 * A L B A R ( I I 7 CALL SYNTHZ CALL EVLVTR CALL SENSIT CALL EVOB IF(I-NO) 3,3,4 3 G R A D l D = ( ( V O B J - V O B A R ) Z t A L P H t I I - A L d A R l I ) ) ) * A L B A R ( I )**2 ALPH(I)=ALBARI I) 1=1 + 1 IF(I-NSP) 5,6,5 6 M=I-NO B E TA (M)= 1 . 0 0 1*BT8AR(M) GO TO 7 4 GR A D t 11 = ( ( V O B J - V O b A R ) Z t S E T A ( M ) - B T B A R ( M ) ) ) * B T B A R ( M ) * * 2 B E T A (M )= B T B A R (M ) IF(I-NVB) 10,33,10 10 1= 1+ 1 GO TO 6 33 R E T U R N END 114 20 8 P U N C H 3 6 » ( M R E ( I ) , J R E U )*1=1*5) P U N C H 3 7 , ( V A L ( I ) *1 =1 ,5 ) 209 CO NT IN UE IF(J-NELP) 55,53*53 53 IF(M-NSP) 56 ,5 7* 57 56 M = M + I J=I G O T O 55 57 CR =C MA X/ CM IN RR=RMAXZRMIN P R I N T 3 8 ,C D *RD, C S 1J M i C R »RR 3 8 F O R M A T ( 3 6 H C A P A C I T A N C E D I S T R I B U T I O N P A R A M E T E R = * E 1 7 . 8 , Z * 3 5 H R E S ISTA I N C E D I S T R I B U T I O N P A R A M E T E R = ,E l 7 . 8 * Z * 2 9 H T O T A L C A P A C I T A N C E P A R A M E T E 2 R = * E 1 7 . 8 , Z * 1 9 H C A P A C I T A N C E R A T I O = ,E l 7 * 8 » Z » 1 8 H R E S I S T A N C E R A T I O = , El 3 7 . 8 * Z * 7 0 H S E N S I T I V I T Y OF C O E F F I C I E N T M W I T H R E SP EC T TO V O LT AG E FOL 4L0WER GAIN K.*Z*1H ) M=I I F f S E N S E S W I T C H I) 2 1 0 , 4 5 2 1 0 P U N C H 3 8 , C D ,R D , C S U M , C R ,RR 4 5 D O 3 9 1= 1 , 5 M R E ( I )= M V A U II=FNCO(M)*.Ol IF(M-NSP) 39,41,41 39 M = M + I 4 1 IF ( I - S ) 4 2 , 4 3 , 4 3 42 D O 40 L = I ,5 ' MRE(L)=O 4 0 V A U I )=0. 4 3 P R I N T 4 4 , ( M R E ( I) ,1 = 1.5) 4 4 F O R M A T t 6 X , 5 ( 2 H M = , 1 2 , 12'X) ) P R I N T 37 , (V A L ( I) ,1 = 1,5) I F t S E N S E S W I T C H I) 2 1 1 , 2 1 2 21 1 P U N C H 44-,(MRE( I) ,1 = 1,5) P U N C H 3 7 , tV A Lt I ),1 = 1,5) 21 2 C O N T I N U E IF(M-NSP) 45 ,47,47 47 M=I P R I N T 18 9 1 8 9 F O R M A T ( 7 9 H S E N S I T I V I T Y OF C O E F F I C I E N T M W I T H R E S P E C T T O V O L T A G E FO I L L O W E R IN P U T A D M I T T A N C E , Z , I H ) I F t S E N S E S W I T C H I)' 2 1 3 , 7 5 2 1 3 . P U N C H 18 9 75 D O 69 1 = 1 , 5 MREt D = M V A U I I = F D C O ( M ) * . 01 IF(M-NO) 69,71,71 69 M= M+ 1 71 1= 1 + 1 I F t 1-5) 7 2 , 7 3 , 7 3 72 D O 7 0 L = I , 5 MRE(L)=O 70 V A U U = O . 73 P R I N T 4 4 , (MREt I ) ,1 =1,5) P R I N T 3 7 , (V A L ( I ) ,1 = 1,5) I F (S E N S E S W I T C H I) 2 1 4 . 2 1 5 2 1 4 P U N C H 4 4 , (MREt I ) ,1 = 1,5) P U N C H 3 7 , (V A L ( I) ,1 = 1,5) 215 CONTINUE IF(M-NO) 7 5 ,9 7, 97 9 7 P R I N T 112 ZZJ08 5 ZZFOR 5 *LDISK SUBROUTINE EVUVTR D I M E N S I O N L D { 3 0 ) »NTY.P(30) , V T R ( S O O ) »SEL M( 1 1 , 3 0 ) ,S P ( I l ) ♦ W E L M ( 11 ,30 I » I E L E M ( 3 0 ) , D T ( I l ) , W K ( I l ) , W Y ( I l ) , A L B A R t 1 0 ) , B T B A R ( 1 0 ) , G R A D (20 I , I F R ( 3 0 ) 2 , 1T O I 3 0 ) , I T R ( I O O O ) , F N C O ( I l ) , F D C O ( I l ) , B T ( I l ) , A L P H ( I O ) , B E T A t 1 0 ) , F T C O 3( 1 1 ) COMMON NOD,NTT,NELP,NTS,NO,NSP,LD,NTYP,VTR*SELM,SP,WELM,WCD,WRDfWC I S , W C R »W R R , E L E M » D T » W K » W Y , N V B , A L B A R , B T B A R , V O B A R , G R A D , G R M A X » N G R I , S T P , 2 N 0 M I , N O M T ,N C O N , IJ , I B ,K B , I F R , I T O , N E L M , N IT R , IT R * N F A R , F N C O * F D C O » B T ,A L 3 P H , B E T A , F T C O ’, C M A X ,CM IN , R M A X , R M I N , C D , R D , C S U M ,V O B J M=-I D O 18 1 = 1 , N T S DO" 2 N = 1 , N F A R , 2 M=M+2 K= 5 * ( N - I )+ 1 LD(K)=ITRtMIZlOOO L D ( K + 1 ) = ( I T R t M )- 1 0 0 0 * L D (K ) JZlO L D (K + 2 ) = ( IT R (M 1 * 1 0 0 0 0 ) Z 1 0 00 IF(N-NFAR) 4,5,5 4 L D (K + 2 I=L D t K + 2 I+ I T R ( M + 1 )Z 1 0 0 0 0 L D tK+ 3 ) = ( ITR(M+l')*.iO) Z l O O O 2 L D IK + 4 )= ( IT R (M + 1 ) * 1 0 0 0 I Z l O O O 5 VTR(I)=I. D O 18 L= Ii NT T J=LD(L) IF(J-NO) 21 ,2 1, 19 21 V T Rl I)= V T R ( I ) Z E L E M ( J ) G O T O 18 19 V T R ( I l = V T R t I ) * E L E M ( J ) 18 C O N T I N U E RETURN END ZZJOB 5 ZZFOR 5 * L D IS K SU BR OU TI NE EVOB D I M E N S I O N L D ( 3 0 ) , N T Y P ( 3 0 ) , V T R ( 5 0 0 ) , S E L M t 1 1 , 3 0 ) , S P I 1 1 ) , W E L M ( I I ,3 0) , 1 E L E M ( 30 I , D T ( I l ) » W K ( 1 1 ) , W Y ( I l ) , A L t i A R ( I O ) , B T B A R ( I O ) , G R A D t 2 0 ) , IFRt 30) 2 , 1 T O (30) , I T R ( I O O O ) , F N C O ( I l ) , F D C O t 1 1 ) , B T t 1 1 ) , A L P H ( I O ) , B E T A t 1 0 ) , F T C O 3( 1 1 ) C O M M O N N O D •N T T , N E L P , N T S i N O «N S P , L U , N T Y P » V T R , S E L M , S P , W E L M » W C D » W R D » W C 1S,WCR,WRRfELEM,DT,WK,WY,NVB,ALBAR,BTBAR,VOBARfGRAD,GRMAXiNGRIfSTPf 2 N 0 M I , N O M T , N C O N f I J , I B i K B , IF R , I T O , N E L M , N I T R f I T R , N F A R , F N C O f F D C O , B T , A L 3 P H, BE TA ,FT CO ,C M A X . C M I N , R M AX ,RM IN,CD,RD,C S UM ,VOBJ R M A X = EL E M t I ) RMIN=RMAx RSUm =RMAX DO I 1=2,NO IFIELEMt 11-RMAX) 2,2,3 3 RMAX=ELEM(I) 2 IFtELEMt D - R M I N I 4, 5, 5 4 RMIN=ELEMt I) 5 RSUM=RSUM+ELEMt11 I CONTINUE I =NO+1 8 7 9 10 6 11 CMAX=ELEMU) CMIN-CMAX CSUM=CMAx M = 1+ 1 DO 6 I=M,NELP I F ( EL E M I I J - C M A X l 7 , 7 , 8 CMAX=ELEM(I) IF(ELEM(I)-CMIN) 9,10,10 C M I n =ELEM(I) CSUM=CSUM+ELEM(I ) CONTINUE RBAR=NO RBAReRSUMZRBAR I=NELP-NO CBAR=I CB AR =C SU MZCBAR RD=O. D O 11 ! - ! , N O R D = R D + ( E L E M ( I )-RtiAR)**2 CD=O. I=NO 12 1= 1+1 C D = C D + (E L E M t I ) - C B A R ) * * 2 IF(I-NELP) 12,13,12 13 V O B J = W R D * R D + W C D * C D + W C S * C S U M + W C R * (C M A X Z C M I N ) + W R R * (R M A X Z R M I N I DO 14 J= 1 , N E L P D O 14 M = 2 , N SP 1 4 V O B J = V O B J + W E L M ( M , J ) * ( A B S F ( S E L M ( M , J ) ) ). D O 1 5 M = I 1NO V O B J = V O B J + W K (M ) * F N C O ( M ) * , O l 15 V O B J = V O B J + W Y ( M I* F D C O (M )* ,O l RETURN END ZZJO B- 5 ZZFOR 5 *LDISK S U B R O U T I N E P O A D (P O C , N O C , P O D , N O D , R P O 1N P O ) D I M E N S I O N L D I 3 0 ) , N T Y P t 3 0 ) , V T R ( 5 0 0 ) , S E L M I 1 1 , 3 0 ) , S P ( I l ) , W E L M t 11 , 3 0 ) , I E L E M (30) ,DT( 11) »WK( 11) »WY( 1 1 ) , A L B A R ( I O ) ,BTBA'RI 10 ) , G R A D I 20 ) , IFR( 30) 2 » I T O ( 30 ) ,ITRt 10 00 ) , F N C O (11 )•, FD CO l 1 1 ) ,BT (1 1 ) , A LP Ht 10 ) , B ET At 10 I , F T C O 3( 11 ) C O M M O N N O D , N T T , N E L P e N T S , N O , N S P , LD i N T Y P , V T R , S E L M , S P , W E L M , W C D , W R D , WC I S i W C R , W R R , E L E M , D T , W K , W Y , N V B , A L B A R , B T 8 A R , V O B A R , G R A D , G R M A X f N G R I,ST P, 2 N 0 M I , N O M T f N C O N , I J , I S , K B , I F R » I T O , N E L M f N I T R , IT R , N F A R , F N C O , F D C O , B T ,AL 3 P H , B E T A , F T C O , C M A X i C M I N , R M A X , RM I N , C D , R D , C S U M , V O B J D I M E N S I O N P O C ( 1 1 ) , P O D t 1 1 ) , R P O I 11) IF(NOC-NOD) 11,12,12 11 N = N O D G O T O 14 12 N = N O C 14 IF(N) 4 , 4 , 2 2 J=NOC+I K=NOD+! D O 8 I=Jfll8 POC(I)=O. DO 9 I=Kfll 9 P O D ( I )=0. DO 31 =1 ,N RPO< I )=PO C( I H-PO Dl I). NPO=N RETURN END ZZJOB 5 ZZFOR 5 •Ld i s k S U B R O U T I N E P O M U L l A N t I A N iP O D >N O D ♦P O C iN O C iKERR) D I M E N S I O N LD l 3.>) t N T Y P O V ) t V T R I 5 0 0 > t S E L M U I #3 0 > t S P (11 1 > W E L M 111 >30) > I E L E M l 3 0 > tD T I 1 1 ) tWKl 1 1 ) tWYI 1 1 ) t A L B A R l 1 0 ) t B T B A R l 1 0 ) » G R A D ( 2 0 ) 11 FR I 30) 2» ITOl 30.) i ITRl 10 00 ) t F N C O l 11) t F D C O ( I l H B T l l l ) , A LP Hl 10) ,BET A! 10) t F T C O 3( 11 ) COMMON NODtNTT,NELP,NTS,NO,NSPtLDtNTYP,VTRtSELM,SR,WELM,WCD,WRD,WC IStWCRtWRRtCLEMtDT tWKtW Y tNVBtALGARtBTBARtVOBAR,GRADtGRMAXtNGRItSTPt 2N0MI,NOMTtNCONt IJ 11G ,KU,IF R ,ITOtNELM,NIT R 11T R tNFAR,FNCOtFDCOtBT,AL SPHtBETA,FTCC,CMAXtCMINtRMAXtRM IN,CD,RDtCSUMtVOGJ DIMENSION ANl 1 1 ) ,PODl 1 1 1 ,POCI 1 1 ),DPC(Il),DPD(Il),DRP(Il) ZZJOB 5 Z7 F O R 5 *LDI5K S U BR OU TI NE RDIN D I M E N S I O N LU I 30) , N T Y P l 3 v ) , V T R ( S O O ) tS E L M I 1 1 , 3 0 ) , S P ( I l ) i W E L M ( l l , 3 0 J t I E L E M t 3 0 ) t D T ( 1 1 ) t W K ( I D t W Y ( 1 1 J ,A L U A R ( 1 0 ) , B T d A R t l O J , G R A D ( Z O ) t J F R ( 3 0 ) 2 t I T O l 3 0 ) , I T R l U O O ) , F NC Ol I D , F D C O ( I l ) , J T C 1 1 > , A L P H I 1 0 ) , B E T A l 1 0 ) , F T C O 3( 11 ) C O M M O N NOD >N T T , N E L P , N T S , N O , N S P t L D i N T Y P , V T R , S E L ‘I,SP,V.'ELM,W C D t W R D , W C I S , W C R tW R R t E L E M t D T , K K ,W Y , N V B , A L G A R , U T t i A R ,V O U A R , G R A D , G R M A X t N G R I t ST P, Z N O M I , N O M T t N C O N t I J t I t i t K B , I F R , I T C t N E L M t N I T R t I T R t N F A R , F N C 0 , F D C 0 , 3 T ,AL 3 P H tB E T A , F T C 0 1C M A X ,C M IN ,R M A X ,R M IN ,C D ,R D i C S U M ,V O B J READ 11,HOD,NCLP,N O 1 IiitU,KB ,NELM 11 F O R M A T I 712) R E A D 1 2 , IN T Y P ( I ) t I= I , 3 0 R E A D 12, I I F R l D ,1 = 1,30) READ 13,WCOIWRDtWCStWCR,WRR 13 F O R M A T 1 9 F 1 0 « 2 ) NSP=NOtl NCARD=I I INELP*NSPl-l)/8)tl DI ME NS IO N WTPEtai MV=O D O 31 I = H N C A R D R E A D 1 7 , I W T P E l M ) ,M = l » 8 ) 1 7 F O R M A T I 8 F 1 0 e2 ) D O 31 J = I ,8 MV=MVtl MI=(MV-I)ZNELPtl MJ=MV-IMi-IHNELP 31 W E L M ( M H M J ) = W T P E ( J ) R E A D 1 7 , ( W K ( I ) , 1=1,8) IF(NSP-B) 41 ,41,42 4 2 R E A D i 8 , W K ( 9 ) ,W K( IO ) 4 1 R E A D 1 7 , "IWYl I ) ,1 = 1,8) IF IN S P - 8 ) 4 3 , 4 3 , 4 4 4 4 R E A D 1 8 , W Y l 9 ) ,W Y(IO) ' 43 NCARD=INOZBHl R E AD 1 7 , I S P l 11,1=1,8) IF(NCARD-I) 26,26,27 2 7 R E A D 1 8 , S P ( 9 ) ,S P( IO ) 18 FO RMAT(2F10,2) 2 6 R E A D 1 7 , IA LB AR I I ) , 1 = 1,8) I F IN O - 8 ) 3 3 , 3 3 , 3 4 3 4 R E A D 1 8 , A L B A R I 9 ) , A L B A R l 10) 3 3 R E A D 1 7 , IB TB AR l 1 1 ,1 = 1,8) NTT=NOD-I NVB=2*NO-2 NOMI=NO-I N0MT=N0-2 N F A R = I I I4 * N T T ) - l )Z l O l t l RETURN END ZZJOB 5 ZZFOR 5 #LDISK SU BR OU TI NE SENSIT D I M E N S I O N L D I 3 0 ) , N T Y P I 3 0 ) , V T R I 50 0> t S E L M l l l , 3 0 ) ,SP 11 1 ) tW E L M l 1 1 , 3 0 ) , I E L E M I 3 0 ) ,DT I 1 1 ) tW K l 1 1 ) tW Y ( I l ) , A L G A R l 1 0 ) tB T B A R ( 1 0 ) , G R A D I2 0 ) 11 F R l 30) 2 t I T O l 3 0 ) , I T R I 1 0 0 0 ) , F N C O ( I l ) , F D C O ( I l ) , B T ( I l ) , A L P H 11 0 ) tBETAl 1 0 ) , F T C O 3( 1 1 ) C O M M O N N O D ,N TT , N E L P , N T S ,N O , N S P , L D t N T YPt V T R, SELIlt SPt W E L M , W C D , W R D t W C IS,WC R, WR R,E L E M tD Tt WK ,W Yt NV Bt AL dA Rt oT BA R, VO BA R,G RA D,GRMAXtNGRItSTPt ZNOMltNOMTtNCONtIJtIBtKBtIFR1ITOtNELMtNITRtITRtNFARtFNCOtFDCOtBTtAL S P H t B E T A t F T C O t C M A X t C M I N ,R MA Xt RM IN,C D , RDtC SU Mt VO BJ D I M E N S I O N S T O T l 11) ST0T(1)=ELEM(1) D O 21 I = Z t N O 21 S T O T ( I ) = S T O T ( I ) K E L E M l I) STOT(I)=I.ZSTOT(I) O O 22 I = 2 , N S P 22 S T O T ( I ) = O . 116 IFl IA N * N O D ) 1 , 1 , 2 1 NOC=O RETURN 2 N=IANtNOD-I I F IN - i I ) 3 , 3 , 4 4 KERR=3 RETURN 3 DO 8 I=ItN 8 D R P { I )=0. D O 9 I = I , IAN 9 D P Cl I )= A N I I I O O 10 I = I i N O D 10 DP D l D = P O D l I ) DO 5 I=ItIAN DO 5 J=ItNOD K= I+J-l 5 D R P ( K ) = D R P I K ) + D P C i n * D P D I J) NOC =N DO 6 I=ItNOC 6 P O C l I ) = DR P( 11 RETURN END RE AD 1 2 , U T O U ) , 1*1,30) 12 F O R M A T I 3 0 1 2 1 ZZJOB 5 ZZFOR 5 *LDISK SUBROUTINE SEOGEN(NEL) D I M E N S I O N L U (30) tN T Y P I 3 0 ) t V T R I 5 0 0 ) tS E L M ( 1 1 1 3 0 ) t S P ( 1 1 ) » W E L M I 1 1 1 3 0 ) t I E L E M O O ) t D T( ll ) t'.VKUl) t W Y ( l l ) tALI3AR (10) tIiTBAR ( 10 ) t G R A D (20 ) 11 FR (30) 2 t I T O ( 3 0 ) t l T R ( l O O O ) . F N C O ( I l ) tFD C O f 1 1 ) t B T ( 1 1 ) tA L P H I 1 0 ) t B E T A ( 1 0 ) tF f CO 3( 11 ) C O M M O N N O D t N T T t N E L P t MT S . N O . N S P t L D t NT Y P t V T R t S E L M t S P t W E L M t W C D t W R D t W C I S . W C R . W R R . E L C M t D T tW K iW Y t N V B . A U i A R v i T ^ A R t V O B A R . G R A D t G R M A X t N G R I .STPt 2 N 0 M I t N O M T t H C O N t I J t I U t K J 1 1F R . I T O t H E L M t N I T R t I T R t N F A R . F N C O t F D C O . t i T t A L 3 P H t B E T At F T C O i C M A X t C M I N t R M A X t R M I N . C D t R D t C S U M t V O B J I F ( N E L - I ) I t I t2 1 NEL=2 IF(LD(I)-NELM) 3t2t2 3 LD(I)=LD(I)+! RETURN 2 J=NELM+1-NEL 10 IF ( LDI N E D -J ) 6 t 5 1 5 5 IF(NEL-NOD) 7t6t7 7 NEL=NEL+! J=J-I G O T O 10 6 J=LD(NEL)+NEL L=O 15 I F ( L - N E L ) 1 3 » 1 2 t l 2 13 L = L + I LD(L)=J J=J-I G O T O 15 8 NEL=O 12 R E T U R N END ZZJOB 5 ZZFOR 5 ♦LDISK SUBROUTINE STCOMP D I M E N S I O N L D ( 3 0 ) t N T Y P U G ) . V T R ( S O O ) . S E L M C 1 1 1 3 0 ) t S P (1 1 ) t W E L M ( 1 1 . 3 0 ) • I E L E M O O ) t D T U D .W K ( I l ) . W Y d D t A L B A R I 10 T t B T B A R I 10) t G R A D (20 I 11 FR (30) 2 t I T 0 ( 3 0 ) t ITR ( 1 0 0 0 I I F N C O I 11 Jt F D C O f 11 ) , B T ( l l ) t A L P H d O J t B E T A d O J t F T C O 3( 1 1 ) COMMON NODtNTTiNELP.NTStNOtNSPtLDtNTYPtVTRtSELMtSPtWELMtWCDtWRDtWC I S t W C R t W R R t E L E M t D T t W K tW Y t N V B t A L B A R t B T B A R t V O B A R t G R A D t G R M A X t N G R I t S T P t 2N O M I tNOMTtNCONtIJt IB tK Bt IF Rt IT Ot NE LM tN IT Rt IT Ri NF AR tF NC Ot FD COi BT iA L SP Ht BE TA tF TC Ot CM AX tCM INtRMAXtRMINtCDtRDtCSUMtVOBJ GRMAX=O. D O 4 -I= I . N VB IF ( I - N O ) 2 12 16' 2 COMP=ABSF(GRAD(I)ZALBARd)) GO TO 5 6 J=I-NO C O M P = A B S F (G R A D ( I I Z ti TB AR (J )) 5 IF (GRMAX-COMP) 3i4t4 3 GRMAX=COMP 4 CONTINUE STP=*01/GRMAX RETURN END ZZJOB 5 ZZFOR 5 ♦LDISK SUBROUTINE SYNTHZ D I M E N S I O N L D I 30 I t N T Y P O O " ) . V T R ( S O O ) t S E L M I l I .30) t S P d l ) t W E L M I 1 1 . 3 0 ) . I E L E M O O ) . D T d D .WK( I D t W Y d l ) . A L B A R I 10 ) t B T B A R (10 ) t G R A D ( 20 ) 11 FR! 30) 2 t ITOC 30) 1 1 T R (100 0) t F N C O d D . F D C O d I ) tBT (I U . A L P H I 10) .BET A! 10) t F T C O 3dl) COMMON N O D iNTT.NELPtNTSiNO.NSPtLDtNTYPtVTRtSELMtSPtWELMtWCDtWRDtWC I S tW C R t W R R t E L E M t D T tW K t W Y t N V t i tA L B A R i B T B A R . V O B A R . G R A D i G R M A X i N G R I . S T P t 2N 0M IiNOMTtNCON.IJtIBtKBtIFRtITOtNELMtNITRtITRiNFARtFNCOtFDCOtBTtAL 3 P H. BE TA .FTCOtCMAXtCM IN.R M A X t R M INtCDiRDtCSUM.VOBJ D I M E N S I O N C A P d O ) t C A P D d O ) . R E S d O I . A N d D tP O D d D tPOC( 11) . Q T d D NRTN=NOD BETA(NOMI)=SP(NO) DO 3 I=ItNOMT 3 B E T A (N O M I )= B E T A l N O M I )- B E T A l I ) AN(2)=1. AN(D=ALPHd) P0D(2)=1. 11? DO I MsltMOD D O I J = I l MELP. 1 SELM(MtJ)=O. M=-I DO 8 I=ItNTS D O 2 N = I t N F A R 12 M=M+2 • K= 5 * ( N - I )+ 1 LD(K)=ITR(M)ZlOOO L D (K + l ) = ( I TR (M )- 1 0 0 0 * L D (K ) )/10. L D (K + 2 ) = ( I T R ( M m O O D O ) / 1 3 0 0 I F ( N - N F A R ) 4 15 t 5 4 L D ( K + 2 ) = L D ( K + 2 ) + I T R ( M + l 1/10000 L D (K + 3 ) = ( I T R t M + I ) # 1 0 ) / 1 00 3 2 L D (K + 4 ) = ( I T R ( M + 1 ) * 1 3 0 D ) / 1 0 0 0 5 K= I DO 6 N=ItNTT L=LD(N) 6 K=K+NTYP(L) STOT(K)=STOTJK)+VT R(I) DO 7 L=ItNTT J=LD(L) 7 S E L M ( K t J ) = S C L M I K tJ ).+VTR ( I ) 8 CONTINUE DO 9 J=ItNELP D O 9 M = 2 t N SP IF(J-NO) 27t2 7t 2U 2 7 S E L M (MtJ) = I S E L M (.Mt J ) # ( - 3 . 9 9 0 0 9 9 0 0 9 9 ) ! / S T C T ( M ) GO TO 9 28 SC LM (M tJ)=SELM(MtJ)ZSTOT(M) 9 CONTINUE RETURN END 9 6 4 5 7 8 10 928:1 I=I P O D( U = A L P H l I+ D C A L L P O M U L I A N i I A N t P O D iN O D tP O C tN O C tKERR) L= I AN(U=POClL) I F ( L - N O C ) 4 t 5, 4 L=L+I GO TO 6 IAN=NOC IFl I- N O M I) 7 , 8 , 7 1= 1+ 1 GO TO 9 D O 10 I = 1 , I A N DTl I )= A N 11 I NDT=IAN AN(Z)=I. ANl U = B E T A ( I ) IAN=Z I=I C A P D ( i S ) = sJ T( NQT) ZJiT CL. T ) A N ( Z ) = - I . S C A P C - 1) IAN = Z C A L L P O 1VJLlAXt IAN t PT , N B T , r o C , I1IOC, K C R fU CALL- P O A D l O O C t N O C t J T ,N D T ,A N , IAN) I J DO I V I = I t I A N DT(I ) -AN( I I NDT=IAN-; AN(A)=-CAPDl N ZZJOB 5 ZZFOR 5 SLDISK SU BR OU TI NE TKSTP D I M E N S I O N L D (30 I , N T Y P ( S O ) , V T R ( S O C ) , S E L M I 1 1 , 3 0 ) ,S P (11) , W E U 1U 11 ,30 ) , I E L E M 1I 30) ,D T ( I l ) ,WKl 11) , W Y ( I l ) ,A L B A R t 10) ,U T B A R l 10 ) , G R A D (20) ,I FR ('30) 2 , 1 T O (30) , I T R ( I O O O ) , F N C O I 11) , FD CO l I U ,BT 111)., A L P h (10 ), B E T A I 10) ,F TC O 3(11) C O M M O N N O D , N T T , N E L P , N T S , N O t N S P f L D , N T Y P , V T R , S E L M t SP,WELM',WCDtWRDtWC 1 $ , W C R » W R R , E L E M , D T , W K , W Y , N V B f A L B A R , U T B A R t V O r i M R , G R A D , G R M A X , N G R J ,STP, 2 N 0 M I * N O M T » N C O N , I J , I 6 » K B , I F R , I T O , N E L M , N I T R , I T R i N F A R , F N C O , F D C O » U T ,AL 3 P H , B E T A f F T C O , C M A X t C M I N , R M A X i R M I N , C D , R D fC S U M , V O B J DO I I=ItNO 1 ALPH(I)=ALBAR(I)-STPSGRADm DO 2 - I=I,NOMT M = I + NO 2 BETA(I)=BTBAR(I)-SfPSGRAD(M) 3 CALL CONSTR CALL SYNTHZ CALL EVLVTR CALL SENSIT C A LL EVOB RETURN END ZZJOB 5 ZZFOR 5 SLDISK SUBROUTINE TRCUICLS) 118 13 P O D ( I ) = B E T A l I+ U C A L L P O M U L (A N ♦ I A N , P O D , N O D ,P O C t N O C ,K E R R ) D O il L = I t N O C 11 A N ( L ) = P O C ( L ) IAN=NOC I F ( I - N O M T ) 1 2 , 1 8 , IZ IZ 1= 1+ 1 G O T O 13 18 D O 14 1 = 1 , IAN U 3 T I I I=ANl I ) NRT=IAN D O 15 I = U N O L = N D T + I-I J = L- I D T ( L ) = I D T ( L ) Z D T (I I ) UT(J)=BT(J)ZoT(I) FDvU(J)=OT(J) 15 F T C O ( L ) = D T ( L ) UT(I)=I. FTCO(I)=-I. DT(I)=I. F D C O t I )= 1. D O 16 I = Z t N D T 16 IT (I I=DTt U - G P t I ) V< U=D. NOT=NDT X O = CiiT D O 17 X = I f 1IO CAPm=DT(NUT)ZBT(NJT) AN(I)=O. IAN»2 CALL POMUL(AN,IANiBTtNBT,POCtNOCtKERR) CALL POADlPOC,NOC,QT,NQT,AN,IAN) DO 20 I-IiIAN IF(M-I) 20,317,20 317 FNCO(I)=AN(I) 20 OT(I)=AN(I) NQT=IAN-I RES(M)=BTlNBT>ZDT(NDT) DO 21 I=IiNDT 21 BT(I)=BT(I)-RES(M)SDTin CAP(M)=CAP(M)-CAPDlM) 17 NBT=NBT-I DO 22 I=I,NO 22 ELEM(I)=RES(I) I=NO ELEMl1+1)=CAP(I) ELEM(1+2)=CAPDl2) 1=1+3 DO 23 J=3,N0 ELEM(I)=CAP(J) ELEMl !+U=CAPD(J) 23 1=1+2 NOD=NRTN RETURN END ZZJOB 5 ZZFOR 5 * L D ISK SU BR OU TI NE TRCOM D I M E N S I O N L D ( 3 0 ) , N TY P( 3 0 ) , V T R ( 5 0 0 ) ,S EL Ml1 1 , 3 0 ) , S P ( I l ) , W E L M l 11,301» I E L E M (.30) ,D T ( I l ) ,W K ( I l ) ,WY( 11) »A L B A R (IsO ) , B T B A R d O ) , G R A D ( Z O ) , I F R (.30) 2 , if 0 ( 3 0 ) , I T R ( I O O O ) , F N C O d i ) , F D C O l 1 1 ) , B T ( 1 1 ) , A L P H I 1 0 ) , B E T A l 1 0 ) ,F T C O 3(11) C O M M O N N O D » N T T , N E L P , N T S , N O , N S P , L D ,N T Y P ,V T R »S E L M ,S P » W E L M » W C D , W R D » W C I S t W C R t W R R t E L E M t D T ♦W K »W Y tN VI l, A L B A R iJ T B A R , V O B A R , G R A D , G R M A x , N GR I ,STPt 2 N O M I , N O M T t N C O N t I J i I B t K B , IF R , I f 0 , N E L M , N I T R , I T R t N F A R t F N C O t F D C O t B T t A L ’ 3 P H , B E T A ,F T C O , C M A X t C M I N ,R M A X ,R M I N , C D ,R D , C S U M , V 0 3 J NITR=O KCR =O NEL=C DO 3 I=ItNOD L = N O D + I-I 3 LD(I)=L 4 C A L L T R D E T ( N E L j KE R) IF(KER) 5,6,5 5 C A L L E X IT 6 IF(NEL) 40 2, 7, 40 2 7 CALL TRCLIICLSl IF ( I C L S - 6 ) 4 0 2 , 8 , 4 0 2 8 DO 9 I=ItNOD IF(LD(I)-NELP) 9,9,408 408 IF(LD(I)-IB) 402,9,402 9 CONTINUE NITR=NfTR+! I=I 32 L= I 31 I F ( L D ( I ) - I B ) 3 3 , 3 4 , 3 3 34 L=L-I G O T O 19 33 IF(L-Z) 10,11,12 11 I T R ( N I T R ) = I T R t N I T R I+ 1 0 * L D ( I ) G O T O 19 10 I T R ( N I T R ) = L D t I ) * 1 0 0 0 G O T O 19 12 I F (L - 4 ) 1 3 , 1 4 , 1 5 14 I T R ( N I T R ) = I T R l N I T R ) + L D ( 1 1 * 1 0 0 G O T O 19 1 5 I T R ( N I T R ) = I T R ( N I T R ) + L D ( I) G O T O 19 13 I T R ( N I T R ) = I T R l N I T R ) + L D ( D Z l O NITR=NITR+1 IT R( N I T R ) = I L D ( I ) - I O * ( L D ( I ) Z l O ) 1*10000 19 I F ( I - N O D ) 2 6 , 4 0 2 , 4 0 2 2 6 i= I + l L=L+1 IF (N IT R - 9 9 8 ) 5 1 , 5 2 , 5 2 52 P R I N T 53 53 FO RM A T ( 2 6 H N O . OF TREES EX CE ED S LIMIT) CA LL EXIT 51 IF ( L - S ) 3 1 , 3 1 , 3 2 402 CA LL SEQGEN(NEL) IF ( N E L ) 4 , 5 0 1 , 4 50 1 N T S = N I T R Z N F A R RETURN END ZZJOB 5 ZZFOR 5 * L D ISK SUBROUTINE TRDET(NELtKER) D I M E N S I O N L D (30) , N T Y P ( 3 0 ),, V T R ( 5 0 0 ) ,SELMI 1 1 »30 ) *S P (1 1 ) »W E L M I 11 , 3 0 ) , I E L E M (30 I ,DT( 11) ,WK( 1 1 1 ,WY( 1 1 ) , A L B A R CIO ) , B T B A R (10 ) , G R A D ( 20 ) 11 FR (.30) 2 » I T0( 3 0 ) , I T R I 10 00 ) , F N C O ( I l ) ,FDCOI I D t B T ( H ) , A L P H (10 ) ,BET A! 10) ,F T C O 3( 11 ) C O M M O N N O D t N T T , N E L P ,N T S , N O , N S P , L D , N T Y P , V T R , S E L M , S P t W b L M t W C D , W R D t W C I S , W C R t W R R t E L E M . D T , W K , W Y , N V B t A L B A R , B T B A R , V O B A R , G R A D , G R M A X t N G R I , ST P, 2 N 0 M I , N O M T , N C O N , I J , I B , K B , I F R , I T O , N E L M , N O T R d T R , N F A R , F N C O t F O C O t B T , AL 3PH,BETA»FTC0,CMAX»CMIN,RMAX,RMIN»CD»RD,CSUM,V0BJ D I M E N S I O N N D E ( 30) MT=NOD+! DO I 1=1,MT I NDE(I)=O NSET=I M=NOD 601 J=LD(M) K=IFR(J) L=ITO(J) IF(NDEtK)-NDE(L)) 4,2,3 4 IF lN DE (K )) 676,646,650 119 2 t I T 0 ( 3 0 ) W T R ( I C O O ) , F N C O ( I l ) , F O C O ( I l ) , B T ( I l ) , A L P H ( I O ) »B E T A t 1 0 ) ,F T C O 3( 11 ) C O M M O N N O D ,TjTT , N E L P ,((T S ,NO , N S P , L O ,NT Y P , V T R , S E L M , S P , W E L M , W C D , W R D ♦ WC I S , W C R , ‘, / R R » E L E M , D T , W K »V,'Y»NVS,A L BA R, .i T B A R * V O B A R , G R A D t G R M A X , N G R I »STP» 2.N0MI ,N O M T , N C O N , IJ , IQ ,KB , I F R , I T O , N E L M , N I T R ♦ I T R t N F A R , F N C O t F D C O ,BT ,AL 3 P H , B E T A , F T C O , C M A X , C M I N , RM AX tR M IN,C D , N D , CS UM tV OB J ICLS=I DO I I=ItNOD IF(LD(I)-IB) 2,3,2 3 IK=I ICLS=ICLS+! 2 I F ( L D l I ) -KB) '1 ,1 7, 1 17 I C L S = I C L S + 2 I CONTINUE I F ( IC L S - 2 ) 5 , 6 , 5 C KSV=LD(IK) L D ( IKI=KB CALL TRDET(NELtKER) L D ( IK l = K S V IF(NEL) 5,7,5 7 K S V = L D l IK) LD(IK)=IJ CALL TRDET(NbLtKER) L D ( IKI=KSV IF(NEL) 51,52,51 52 I C L S = I C L S - 1 51 I C L S = I C L S + A 5 RETURN END 120 2 IF(NDElL)) 676,644,653 3 I F t N D E l U ) 676,645,647 676 KER=6 RETURN 644 NDE(K)=NSET •ND E(L)=NSET NSET=NSET+! 60 TO 637 645 NDE(L)=NOE(K) GO T O 637 646 NDE(K)=NDE(L) GO TO 637 647 NEL=NDE(K) IF=O 25 IF(IF-MT) 2 6 ,6 37 ,6 37 •26 I F = I F + ! I F t N D E t IFl-NEL) 25 ,27,25 27 NDElIFl=NDE(L) G O T O 25 650 NEL=NDE(L) IF =O 35 I F ( I F - M T ) 3 6 , 6 3 7 , 6 3 7 36 IF=IF+! IF(NDE(IR)-NEL) 35,37,35 37 N D E ( IFJ=NDE(K) G O T O 35 653 NEL=N RETURN6 3 7 IF ( M - I ) 4 1 , 4 1 , 4 2 42 M= M- I GO T O 601 41 NE L= O RETURN ' END A P P E N D I X . II O P T I M I Z A T I O N COMPUTER P R O G R A M INPUT A N D OUTPUT DATA P O R E X A M P L E P R O B L E M 122 ZZJOB 5 Z Z XEQSCHONE 5100411051212 101010101010000 1 0 2030401020303 0'4040 105 20304000500050005000000 100. Io Ic IO I. Io I. I. Io Io Io I. Io Io Io Io IO Io IO I. I. Io IO 10. 10. IOo 50. 50. 50 O 3.4143 2.6132 Io .35 o05 .25 .3 ol I. I. IO I. IO I. IO 10. I. I. Io I. I. I. 10. 50. 2.6132 4. 10. 50 . I. Ie I. IO I. I. I. I. IO I. Ie I. I. Ie i. I. I. Ie I. INITIAL NETWORK SENSITIVITY OF COEFFICIENT M WITH RESPECT TO VOLTAGE FOLLOWER INPUT ADMITTANCE POLE LOCATIONS •05000000 ,25000000 •35000000 4,00000000 ,30000000 2,21320000 C O E F F I C I E N T S O F IN P U T A D M I T T A N C E Y l l , l ) NUMERATOR COEFFICIENTS DENOMINATOR COEFFICIENTS 1,00000000 1,00000000 27,10714200 155,28571000 265,71428000 57,14285700 13.78516700 39,35779200 15,06114800 OPTIMIZATION PARAMETERS S E N S I T I V I T Y O F C O E F F I C I E N T M WITH. R E S P E C T T O E L E M E N T N U M B E R J, M= I,J= 2 000.OOOOOE-99 M = I ,J= 3 000,OOOOOE-99 M = I ,J = 4 00 0 . OOOOOE-99 M= I,J= 5 00 0.OOOOOE-99 M= I,J= 6 ,OOOOOE-99 M = I ,J= 7 000.OOOOOE-99 M= I,J= 8 00 0.OOOOOE-99 M= I,J= 9 0 0 0 . OOOOOE-99 M= I , J = IO O O O oO O O O O E - 9 9 M= 2 , J= I -986.843656-03 M= 2, J= 2 - 9 7 8 . 1 5 7 3 4 E - 03 M= 2 , J= 3 -807.85073E-O3 M= 2,J= 4 -197.445356-03 M = 2 , J= 5 254.07932E-04 M= 2 , J= 6 .OOOOOE-99 M= 2, J= 7 244.34303E-03 M= 2 , J= 8 000.OOOOOE-99 M= 2,J= 9 730.249246^03 M= 2,J=IO 00 0.OOOOOE-99 M= 3,J= I -932.013Y3E-03 M = 3» J = 2 - 8 0 7 . 3 1 1 1 4 E - 03 M= 3 , J= 3 - 114.477646-03 M= 3 , J= 4 -126.39550E-03 M= 3 , J= 5 2 6 8 . 0 7 2 6 2 E — 03 M= 3,J= 6 979.53544E-05 M= 3,J= 7 7 3 9 . 4 2 8 8 1 E-03 M = 3, J= 8 I57.09939E-03 M= 3 , J= 9 291.261546-03 M= 3,J=IO 534.343T3E-03 M= 4 , J= I -654.85183E-03 M= 4, J= 2 - 2 0 1 . 0 6 6 7 2 E - 03 M= 4 » J= 3 — 6 4 1 . 3 6 3 6 8 6 —04 M= 4 , J= 4 -700.44388E-04 M= 4,J= 5 100.00000E-02 M= 4,J= 6 I58•30040E-03 M= 4»J= 7 42 6, 13 20 2 E-04 M= 4 , J= 8 896.66212E-03 M= 4 , J= 9 318.29200E-03 M= 4,J=IO 583.93274E-03 M= 5,J= I .OOOOOE-99 M = 5, J = 2 000,OOOOOE-99 M = 5 iJ = 3 000.OOOOOE-99 M= 5,J= 4 000.OOOOOE-99 M= 5,J= 5 100.00000E-02 M= 5,J= 6 100.00000E-02 M = 5, J = 7 4 5 5 . 7 1389E-04 M= 5,J= 8 954.426616-03 M= 5,J= 9 352.78571Er03 M= 5,J=IO 6 4 7 . 2 1 4 2 8 E — 03 CAPACITANCE DISTRIBUTION PARAMETER= .11601307E+04 RESISTANCE DISTRIBUTION PARAMETER= .8 6115 8 75 E- 01 ' TOTAL CAPACITANCE PARAMETER= ,89351085E+02 CAPACITANCE RATIO= ,50874323E+03 RESISTANCE RATIO= .43371771E+01 S E N S I T I V I T Y O F C O E F F I C I E N T M W I T H R E S P E C T T O V O L T A G E F O L L O W E R G A I N K, M= 2 2 0 7 , 6 6 2 8 1 E-03 M= 3 100.484^85-02 M= 4 I 16.38857E-02 M= 2 137.85167E-03 M= 3 393,57?92E-03 M= 4 150,6U48E-03 M= 5 20 0.OOpOOE-IO NETWORK ELEMENT VALUES ,12940401E+00 ELEMENT ( D = E L E M E N T I 21 = ,47469814E+00 •28644922E+00 E L E M E N T ! 31 = EL EM EN T! 4)= ,10944864E+00 •66396000E-01 E L E M E N T ! 51 = .44711095E+01 E L E M E N T ! 61 = .16128330E+01 E L E M E N T ! 7) = ELEMENT! 8)= •33778516E+02 E L E M E N T ! 9) = •17435457E+02 .31986774E+02 E L E M E N T (1 0 ) = OBJECTIVE FUNCTION VALUE ,18845128E+04 ********** M= I,J= I .OOOOOE-99 M= I .OOOOOE-99 M= I 100.00000E-04 ,10000000 M= 5 200.00000E-10 GZT ZERO LOCATIONS FINAL NETWORK SENSITIVITY OF COEFFICIENT M WITH R ES PE C T TO VOLTAGE FOLLOWER INPUT ADMITTANCE M= I 100.00000E-04 .28234783 .89603857 1.43481370 .06435969 .44318273 1.04377740 1.54673050 C O E F F I C I E N T S O F IN PU T A D M I T T A N C E Y ( I f l ) DENOMINATOR COEFFICIENTS NUMERATOR COEFFICIENTS 1.00000000 1.00000000 5.35470860 7.19889770 2.75482070 19.39866400 64.23079800 67.27739100 21.71604200 OPTIMIZATION PARAMETERS S E N S I T I V I T Y O F C O E F F I C I E N T M W I T H R E S P E C T T O C L E M E N T N U M B E R J. M= IfJ= 2 COO.OOOOOE-99 M= IfJ= 3 00 0 . OO uO OE-99 M= IfJ= 4 OOO.OOOOOE-99 M= ItJ= 5 OOO.OOOOOE-99 M = I 1J = 6 .OOOOOE-99 M= IfJ= 7 000.OOOOOE-99 M= IfJ= 8 000.OOOOOE-99 M= IfJ= 9 OOO.OOOOOE-99 M= IfJ=IO OOO.OOOOOE-99 M = 2 , J= I - 9 5 4 . 1 1 4 3 IE-03 M= 2♦ J= 2 -9 49 . 4 6 9 9 2 E-03 M = 2 fJ = 3 -689.555396-03 M = 2 fJ= 4 -377.13733E-O3 M= 2f J= 5 138.91015E-03 M= 2,J= 6 .OOOOOE-99 M = 2 fJ = 7 463.20888E-03 M= 2 f J= 8 OOO.OOOOOE-99 M= 2fJ= 9 397.88097E-03 M= 2fJ=IO OOO.OOOOOE-99 M = 3 fJ = I -672.93738E-03 M= SfJ= 2 - 7 4 0 . 7 0 0 9 0 E- 0 3 M = 3 fJ = 3 -319.39494E-O3 M = 3 fJ = 4 - 2 4 7 . 1 6 4 7 5 E - 03 M = 3 fJ= 5 569•29967E-03 M= 3»J= 6 150.43165E-03 M = 3 . J= 7 502.36999E-r03„ M = 3 tJ = 8 165.61801E-O3 M = 3 fJ = 9 3 0 3 • 12235E-03 M= 3tJ=IO 309.15830E-03 M= 4 . J= I -197.71502E-03 M = 4 f J= 2 - 3 1 O 7 4 2 O E -0 3 M = 4 1 J= 3 -294.9052 5E- O3 M = 4 fJ = 4 -180.40453E-03 ,M= 4f J = 5 100.00000E-02 M = 4 fJ = 6 623.355806-03 M= 4«J= 7 2 0 8 . 6 9 9 9 1 E-03 M = 4 tJ = 8 462.73674E-03 M = 4 t J= 9 3 3 9 . 12733E-03 M= 4fJ=IO 345.88022E-03 M= SfJ= I .OOOOOE-99 M= SfJ= 2 OOO.OOOOOE-99 M= SfJ= 3 000.00000E-99 M= SfJ= 4 OOO.OOOOOE-99 M= 5.J= 5 100.00000E-02 M= SfJ= 6 100.00000E-02 M= SfJ= 7 302.03707E-03 M= SfJ= B 697.96294E-03 M= SfJ= 9 495.07091E-03 M= SfJ=IO 504.92908E-03 CAPACITANCE DISTRIBUTION PARAMETER= .33868742E+02 RESISTANCE DISTRIBUTION PARAMETER= .62202557E-02 TOTAL CAPACITANCE PARAMETER= .27975746E+02 CAPACITANCE RATIO= .20176902E+02 RESISTANCE RATIO= .15807672E+dl S E N S I T I V I T Y O F C O E F F I C I E N T M W I T H R E S P E C T T O V O L T A G E F O L L O W E R G A I N K. M= I •OOOOOE-99 M= 2 926.55410E-04 M= 3 205.49498E-03 M= 4 105.29029E-03 M= 5 100.OOOOOE-IO M= 3 719.88977E-04 M= 4 275.48207E-04 M= 5 100.OOOOOE-IO NETWORK ELEMENT VALUES ELEMENT ( D = •26164052E+00 E L E M E N T ( 21= .29526399E+00 E L E M E N T ( 31 = .25631027E+00 • 18678524E+00 E L E M E N T ( 4) = E L E M E N T ( 51 = .36300010E+00 . 7 3 2 4 2 1 7 8 E + 31 E L E M E N T ( 61 = •27318214E+01 E L E M E N T I 71 = E L E M E N T ( 8) = .6 3 128343E+Q1 •55665150E+01 E L E M E N T t 9) = ELEMENTdOI = .56773586E+01 OBJECTIVE FUNCTION VALUE •12658011E+03 ********** M = I 1J = I .OOOOOE-99 M= 2 535.47086E-04 ********* irST POLE LOCATIONS ZERO LOCATIONS A P P E N D I X III NIMERIQAL COMPUTATIONS -126Thls a p p endix presents the numerical computations for the networks of Figures 1.2 and 13» A. F U M E R I G A L COMPUTATIONS F O R N E T W O R K OF F I G U R E 1.2. Guil l e m i n presents a meth o d for Synthesizing the t r a n s ­ fer fun c t i o n (s) =Eou^.(s)/I^n (s) j where is m i n i m u m phase and stable, "by means of a lossless n e t w o r k terminated b y resistors (see pages 535-537 of Reference 25)» In the p r e sent case, we desire to synthesize Z1 2 (S)=E ^ + 2 - 8 2 8 8 + 4 (III.I). s +1.414-s+l Sep a r a t i n g the n u m e r a t o r into its even and odd parts, we have K(s *4) Z1 2 =Z1 2 a +Z1 2 b = . + K (2.828s) s +1.414s+l (m.2) s +1.414s+l Guillem i n points out that since the numerator of Z-^ is a H u r witz polynomial, its even and odd parts hav e simple zeros restricted to the j-axis. Chapter Il of Refe r e n c e 25 shows that tr a n s f e r functions of the for m of Z ^ a and Z^g^ m a y be realized b y a lossless n e t w o r k terminated b y a. single r e ­ sistor. T h u s , the two t r a n s f e r functions are realized s e p a ­ r a t e l y and connected in series to f orm the total transfer function. Util i z i n g K = X and a p p lying G u i l l e m i n 8s procedure to Z^ 2 a and Z-^-J3 separately. -127- s 2 +4 J12 a „ 1.414s (III.3) 12a" 1+z22a 1+ s2 +l 1.414s ' A n e t w o r k p o s s e s s i n g the above z parameters consists of a two section ladder n e t w or k whose input shunt arm is the impedance Z^pa and whose output series arm is the impedance z2 2 a _ z 1 2 a ‘ Thus 9 2 '12a= ^ I .414s = 0<>177s+ T T T O s (III.4) wh i c h is a 0.177 h e n r y coil in series with a 1.414 farad capacitor. The series arm is 22a 12a s2 +l 1.414s 1 A s 2 +! As' 1.414s “ 1.414s 0.530s (III.5) w h i c h is a 0.530 h e n r y inductor. Si m i l a r l y is realized as z12b V 2.828s 1 A — o---s^+1 (III.6) 12b= l+sgzb = 1+ s^+1 and o7 0 7 s . '12b" s 2 +l (III.7) 1.414s+ ,707s w h i c h is a 1.414 farad capa c i t o r in parallel w i t h a 0.707 h e n r y coil. The series arm is -128- = .707s = .707s 1.414s g2 +1- z22b z 12Td = s2 +l s 2 +l I_____ 1.4 1 4 s + I .707s (III.8) w h i c h is a repeat of ^12a an ^ ^12h are n o w r e a l i z e d b y connecting I ohm r e ­ sistors across the output t e rminals of the above ladders. These two networks are t h e n connected in series as shown in Figure 1.2 to realize Z - ^ C s ) « B. M J M E R I O A L COMPUTATIONS F O R N E T W O R K OF F I G U R E 1.3« Kartell's method (41) is to be applied to the transfer function 8^+2.828s+4 (III.9) s2 + 1 . 4 1 4 s +1 Pantell shows that a m i n i m u m ph a s e and stable T V m a y al- ways be realized w i t h resistive terminations at each end provide d T j axis. ( )=0 or T ( )=0 and no poles are located on the o U t i l i z i n g the surplus factor s +8.4 8 s +4 and let t ing K = K ^ k c,, Tv m a y be put into the form T_(s)= k1 s +1.414s+l k2 S2 ^ a8f8^/4 = fIof2 s +8.48s+4 (III.10) -129where b o t h and. fg are p o s i t i v e real. Set t i n g k^=l and k 2=0.2 and f o l l o w i n g P a n t e l l 1s procedure Z-] = __ 1 _ _k ' f = s 2 -i-1.414s-fl _ 0 . 2 s 2 + 5656 s +Q.S ^2 2 I" ^ f 1 s +8.48s+4 s +8.48s+4 " 0.8s +0.848s+0.2 s2 +8.48s+4 (III.11) and a + 8 .48s +4 "I" kgf g (III.12) s +2.8 2 8 s +4 P a n t e l l 's method relies on the fact that the surplus factor, k^, and kg m a y always be chos e n such that f-^, fg, Z^, and are pos i t i v e real. ladder n e t w o r k and en across Y^. in Figure 1.3. b ecomes the input series arm of a its f o l l o w i n g shunt arm w i t h tak­ Z^ and Y^ are realized w i t h the results shown L I T E R A T U R E CONSULTED 1. J 0 L. M e r i l l 5 "Theory of the negative impedance c o n v e r t ­ er „" Bell S y stem Tech. J . „ vol. 30, pp. 88-109; Jannary- 1951. 2. J. Go L i n v i l l 5 "Transistor negative impedance c o n v e r t ­ ers," P r o c 0 I R E , vol, 41, pp. 725-729; J u n e , 1953» 3» A. I, L a r k y , "Negative impedance c o n v e r t e r s ," IRE Trans, on Circuit Theory, vol. CT-4, pp, 124-131; September, 1957. 4. J. Go L i n v i l l , "RC active filters," P r o c , I R E , vol. 42, ppo 555-564; March, 1954. 5. To T a n a g i s awa, "RC active networks us i n g current in­ v e r s i o n type n e g ative impedance c o n v e r t e r s ," IRE T r a n s . on Circuit Theory, vol. CT-4, pp. 140-144; September, 1957» 6. I. W. Sandberg, "Active R C n e t w o r k s ," M icrowave R e s . Inst., Polytechnic Inst, of Brooklyn, N. T., R e p t . No. R-662-58, PIB-590; 1958. 7o B. R 0 Myers, "Contribution to transistor RC network s y n t h e s i s ," Ph.D. T h e s i s , Un i v e r s i t y of I l l i n o i s , U r b a n a ; 1959» 8, J. Mo S i p r e s s , "Synthesis of active RC n e t w o r k s ," IRE Trans, on Circuit T h e o r y , vol, CT-8, pp, 260-269; S e p ­ tember, 1961. 9. R. L. D i e t z o l d , "Frequency discriminative electrical t r a n s d u c e r , " U.S. Patent No, 2 5 4 9 0 6 5 1 1951» 10. I. Mo Horowitz, "Exact d e s i g n of t r a n sistor RC b a n d ­ pass filters w i t h pr e s c r i b e d active p a r a m e t e r ins e n s i ­ tivity," IRE Trans, on Circuit T h e o r y , vol. CT-7, pp, 313-520; September, I960. 11. Ie Mo H o r o w i t z , "Optimum d e s i g n of single-stage gyrator-RC filters wit h pr e s c r i b e d s e n s i t i v i t y , " IRE T r a n s . on Circuit T h e o r y , vol. CT-8, pp. 88^94; June, 1961. 12 Be Ko K i n a r i w a l a , "Necessary and sufficient conditions for the existence of ±R, C networks," I R E T r a n s . on Circuit T h e o r y , vol. CT-7, P P » 330-335; September, I960. -13113» C= L 6 Phillips and K 6 L 6 S u 9 "Synthesis of three t e r m i ­ nal I E 9 C n e t w o r k s 9" I E E T r a n s 0 on Circuit T h e o r y 6 vol„ C T - I l 9 p p 0 80-82; M a r c h 9 1964-„ 14-6 E 6 P 0 Sall e n and E 6 L 6 K e y 9 "A prac t i c a l method of d e ­ signing EC active f i l t e r s 9” IBE Trans, on Circuit T h e o ­ r y , v o l 0 C T - P 9 p p 0 74— 83; M a r c h 9 1933° 13° N 6 B a l a hanian and B 6 P a t e l 9 "Active r e a l i z a t i o n of co m p l e x zeros," IBE T r a n s 6 on Circuit T h e o r y , (Corre­ spondence) 9 v o l 0 C T - I O 9 p p „ 299-300; J u n e 9 1963° 16, J 0 L 6 H o g i n 9 "An active BC low-pass filter c o n f i g u r a ­ ti o n u t i l i z i n g the v oltage f o l l o w e r 9" Douglas Aircraft Co. Engin e e r i n g Paper N o 6 2072; Presented at Seventh Midwest Symp o s i u m on Circuit T h e o r y 9 U n i v 6 of M i e h 6 9 M a y 3, 1964. 17« D 6 A 6 C a l a h a n 9 "Sensitivity m i n i m i z a t i o n in active BC s y n t h e s i s , " IBE Trans, on Circuit T h e o r y 9 v o l 6 CT-9? pp-0 38-4-2; M a r c h 9 1962. 18. I, M 0 H o r o w i t z 9 "Optimization of negative-impedance' c o n v e r s i o n methods of active BC s y n t h e s i s 9" IBE Trans. on Circuit T h e o r y 9 vol. C T - G 9 pp. 296-303; S e p t e m h e r 9 193^° 19» A 0G 0J 0 Holt and P 6 W 0 S t e p h e n s o n 9 "The effects of er­ ror in the element values and the converter transfer characteristic on the responses of active BC filters," J 6 Brit. I B E 9 vol. 26, No. 6, pp. 4-4-9-4-57; December, 1963° 20. J 6 D 0 S c h o e f f l e r 9 "The synthesis of m i n i m u m s ensitivi­ t y networks," IBE Trans, on Circuit T h e o r y , vol. C T - I l 9 p p 6 271-276; J u n e 9 1984. 21. N 6 M. H e r h s t 9 "Optimization of pole s e nsitivity in a c ­ tive BC networks," P h 0D 0 thesis, Cornell University; September, 1963° 22. E 0 S. K u h 9 "Transfer f u n c t i o n synthesis of active BC n e t w o r k s ," IBE National C o n v ention B e c o r d 9 Part 2 (i 960 ), p p . 154- 158 . 23. J 6 J 6 B o n g i o r n o , "Synthesis of active BC single-tuned bandpass f i l t e r s ," IBE National Convention B e c o r d . Part 2 (1958), p p ° 30-41. -13224-, B„ K» K i n a r i w a l a 9 "Synthesis of active BG n e t w o r k s 9" The Bell Syst e m Technical J o u r n a l , pp« 1 2 6 9 - 1 3 1 6 9 S e p t e m b e r 9 1939« ; 23» E, A 0 G u i l l e m i n 9 "Synthesis of passive networks," John W i l e y & S o n s , I n c 09 N e w Y o r k 9 1937® 26. B. J 0 D a s h e r 9 "Synthesis of passive EG networks as u n ­ balanced two-terminal p a i r networks," M 0I 0T 0 Research La b o r a t o r y Electronics Tech, B e p t „ No. 215 (1952)« 27® A. F i a l k o w and I 0 G e r s t 9 "The transfer f u n c t i o n of an BC ladder network," J „ M a t h . and P h y s „ , vol« 30, pp. 49-72, 1951® 28. A. F i a l k O w and I 0 G e r s t 9 "The transfer f u n c t i o n of general two-terminal p a i r RC networks," Q u a r t . A p p l „ M a t h . , v o l 0 10, p p «. 1 1 3 - 1 2 7 5 1952. 29» N 0 B a l a b a n i a n 9 "Network synthesis," P r e n t i c e - H a l l , Inc., Englewood C l i f f s , N 0 J 09 1958. 300 H. B o d e , "Network analysis and feedback amplifier d e ­ s i g n , ” McGraw-Hill, Inc., N e w York, 1945® 31. H 0 Hr, "Boot locus p r o p erties and s e nsitivity relations in control systems," IBE TranSAG5 (January, I960), p® 57® 32. D. K« Phillips, "Unity-gain b u f f e r acquires precision, b y feedback," E l e c t r o n i c s , pp. 36-37, December 20, 1963® 33 . J. D. S c h o e f f l e r , "C o n t i n o u x l y equivalent networks and th e i r a p p l i c a t i o n , " IEE E Trans, on Commun i c a t i o n and Electronics, vol. 83, No. 75, P P ® 763-767, November, 1964. 34. "High-Input- I m p e dance U N I F E T A m p l i f i e r s ," File No. 606 Ap p l i c a t i o n N o t e ; S i l c o n i x Incorporated, Sunnyvale, C a l i f , , February, 1963® 35 . G 0 L e i t m a n n , "Optimization t e c h n i q u e s ," Chapter 6 , Academic Press, N e w York, 1962. 36. S 0 S e s h u and M. B. B e e d , "Linear graph and electrical networks," A d d i s o n - W e s l e y P u b l ishing Company, I n c . , Reading, M a s s a c h u s e t t s , 1961. -13537» L. Weinberg, "Net w o r k analysis and s y n t h e s i s ," Chapter 11, McGraw-Hill- Book Company, Inc,, N e w York, 1962. 38. W. M a y e d a and M. E. V a n V a l k e n h u r g 9 "Network analysis and synthesis b y digital c o m p u t e r ," IRE W e s c o n C o nven­ ti o n R e c o r d , Pt. 2, pp. 137-14-4-; 1957» 39» "Digital computer p r o g r a m P O 14-A p r o g r a m to obtain the d r i v i n g point and t r a n s f e r immittance functions of a fo u r-terminal R L C passive n e t w o r k , " Douglas Aircraft Com p a n y M e m o r a n d u m Numb e r A-260-C222-144. June 18, 1964 b y J. L» H o gin, 40. D » .H a z o n y , "Elements of n e t w o r k s y n t h e s i s , " Reinhold P u b l i s h i n g Corporation, N e w York, N . Y . , pp. 235-237; 196$. 41 R. H. P a n t e l l , "Minimum ph a s e transfer f u n c t i o n syn­ thesis," IRE Trans, on Circuit T h e o r y , vol. CT-2, pp. 133-157, 1955»' MONTANA STATE UNIVERSITY LIBRARIES I i i i i r 762 10005632 III D378 Hogin, J. L. Active RC networks utilizing the voltage follower__________ IM A M K A N D /Z-V- / . PvtnC A P D ftg a a _____T / 2)378 /iG78