U niversity of S outhern C alifornia School Of Engineering Department Of Electrical Engineering EE 402: Homework Assignment #06 (Due 11/19/2003) Fall, 2003 Choma Problem #25: By inspection: (a) Z1Z2 +Zo Z2 Z1 Zo 2 Z1 2 +Zo Z2 2 Z Z Zin = 1 +Z2 || 1 +Zo & Zin =Zo (25.1) 2 2 (25.2) Z1 Z1 Z1Z2 +Zo Z2 (25.3) Zo +Zo Z2 2 2 2 Z12 Z Zo Z1Z2 Zo Z1Z2 1 1 (25.4) {ANS} 4 4Z2 2 (b) When a cascade structure is in need, one can independently design each block of the whole network; hence, one can simply cascade them to each other to meet given requirement because each block from input will see the characteristic impedance. (c) Z1 =sL Z2 = 1 sC Z1Z2 = L C & Z1 Z2 =s 2 LC (25.5) 2 L L 2 s LC Zo = 1+ Ro = ωc = (25.6) C 2 C LC Zo =R o s 2 1+ (25.7){ANS} ωc In (25.7), one can scrutinize that if the network operates much below the cut off ( ωc )frequency of the characteristic impedance, the input impedance will always be resistive and equal to load resistance which is Ro. This means our lumped circuit tries to be like a match terminated transmission line. (d) Since the load termination is not exactly as the characteristic impedance Zo, we must derive input impedance when output is terminated with low frequency characteristic impedance. Thus: sL sL 1 sL + Zin = + || +R o (25.8) Zin 2 2 sC 2 sL +R o 2 (25.9) s 2 LC 1 +sR o C 2 sL s 2 LC s3L2 C s3L2 C s 2 R o LC sL+R o + + 1+ R + 2 + 4R o o 4 2 Zin = = Ro (25.10) s 2 LC s 2 LC 1+ +sR o C 1+ 2 +sR o C 2 2 3 s s s 1+2 +2 +2 ω c ωc L 2 ωc (25.11) =R o C= LC= Zin =R o 2 Ro ωc s s 1+2 +2 ωc ω c ω 2 ω ω 3 1-2 +j2 - ω ω c c ωc Zin (jω) = (25.12) 2 Ro ω ω 1-2 +j2 ωc ωc Zin 1 -2jy3 ω Ro where y= (25.14) 11 (25.13) ρ11 = 2 3 Zin 2(1-2y )+2j(2y-y ) ωc 1 Ro One can straightforwardly use matlab or a similar program to plot the imaginary and real parts of this reflection coefficient. Hence, the result will be as in figure (P.25d). (e) In the network of fig.3.17b, every element has c (transmission)-parameters. Also, the overall c-parameters of the network will be tantamount to multiplication of these c parameter matrices. Hence, s 2 LC s 2 LC sL sL 1+ + 1+ 2 2 1 sL 1 0 1 sL 2 2 (25.16) 2 2 (25.15) C C s 2 LC 0 1 sC 1 0 1 sC 1+ 2 Figure 25d: The reflection coefficient vs. normalized frequency V Vo Ro = (25.17) o = Vi c11R o +c12 Vi Vo = Vi 1 L 2 and =R o C= LC= (25.18) 2 L 2 LC 3 CL Ro ωc 1+s +s +s Ro 2 4R o 1 2 s s s 1+2 +2 +2 ωc ωc ωc 3 (25.19) (f) Similarly, by using c matrix and the c-parameters based circuit analysis, one can arrive at the following results: V Vo Ro = (25.20) o = Vs Vs c11R o +c12 c21R o +c22 R o Vo = Vs 12 (25.21) L R o C 2 LC 3 CL2 1+s + +s +s 2 2 8R o 2R o 1 2 s s s 1+2 +2 + ωc ωc ωc 3 (25.22) (g) The delay is negative derivative of phase of the transfer function with respect to ω. Thus, one can do the subsequent mathematical manipulation: 2y-2y3 d 2 1-2y 3 Vo dφ dy -1 2y-2y D(y)= =φ=-tan D(y)=2 2 Vs dy 2y-2y3 1-2y 1 2 1-2y D(y)= 4 y4 2 y2 2 1-2y 2y-2y 2 2 3 2 (h) We have two specs, namely, D(0) =20psec and Ro=50. Thus, there are two equations that relate these two specifications. Hence, D(0)= 2 2 L ωc 1011 rad sec ωc 1011 & R o 50 ωc C LC C=.1pF&L=1nH And the simulation result will be as in the succeeding lines: Phase Response Magnitude Response The simulation results agree with the analysis done sections (e) and (g). Also, one will view in the magnitude response there is peeking because the relation from Vi to Vo is not exactly a Butterworth relation. Nevertheless, it gives a flat delay within bandwidth of interest. Netlist The simulation of the third order Bessel filter * Lossless .opt post Vi 1 0 AC 1 L1 1 2 .5n C1 2 0 .4p L2 2 3 .5n Rl 3 0 50 ** use the following function find the phase and delay of specific node .plot AC VP(3) VT(3) .op .ac dec 100 0 100g .end Delay Response Problem #26: When we replace the impedance elements of expression derived in section (a) previous question, we will get the characteristic impedance of the system. Hence, (a) 4+ 1-m 2 s 2 LC L Z1Z2 = 4 1-m 2 C 1 Z1 =smL Z2 = + sL smC 4m Z1 s 2 m 2 LC = 4Z 2 2 2 4+ 1-m s LC Based on the previous derivation: Zo =R o 2 s s 2 ω 2 1-m 2 2m 1+ 1+ 2 2 Zo (0)=R o ω1 & ω2 ω1 LC LC 1+ s ω1 From the previous equation it follows that: s s Zo =R o 1+ + ω1 ω2 2 2 Zo =R o ωc s2 1+ 2 ω1ω2 ω 2 ω 2 1 2 ω1ω2 ω ω2 2 1 2 = 2 LC for ωc >>ω Zo (s) R o Thus, for the cut-off frequency there will not be any improvement; however, after the cut-off frequency one will observe better and sharper attenuation owing to the resonance of L1 and mC elements. (b) If one makes the similar analysis done in the section (e) and (g) of the previous question, He will find out that : D(0)=m LC=20pSec R o L 50 Let m=.5 Hence, C mL=1nH mC=.4pF L1 =.177nH . Using these values, the following netlist is set up and simulated, and delay phase and magnitude response graphs are obtained. Netlist Improved k-constant T network .opt post Vi in 0 AC 1 L1 in 2 .5n C1 2 3 .4p L3 3 0 .177n L2 2 out .5n Rl out 0 1e14 ** Used to find input reflection coefficient ** use the following function find the phase and delay of specific node .net V(out) Vi Rout=50 Rin=50 ** Used to find input reflection coefficient .plot AC S11(I) S11(R) .plot AC VP(out) VT(out) .op .ac dec 100 0 100g .end Magnitude Response Delay Response Phase Response Input Reflection Coefficient (c) Using the same netlist, the preceding graph is obtained. The important point is about this graph that it is has zero reflection within a bandwidth. This simply implies that our input port see the characteristic impedance for some frequencies. Problem #27: In order to find, s-parameters of the network we need to terminate both input and output ports with characteristic impedances. However, before doing so, we can simply unearth z-parameters of the network and then convert z-parameters to sparameters. Since this is already done in the previous homework, conversion can simply be referred. Thus, Io: Output current into the network. Vo: Output voltage with respect to ground Ii: Input current into the network Vi: Input voltage with respect to ground. Vi =z11Ii +z12 Io Vo =z 21Ii +z 22 Io Let set Io to zero! Ii =V(g m +g h ) Vi 1 rg (g m +g h ) Vi V 1 rg (g m +g h ) z11 V=Vi Ii rg Ii (g m +g h ) sCg g m Vo z 21 Ii sCg (g m +g h ) sCg Vo V sCg g m Vg m sCg V-Vo Let set Ii to zero! sCg +g h Vo sCg Vo V sCg +g h z 22 Vg h sCg Vo V Io sCg (g m +g h ) Io =V(g m +g h ) Io =V(g m +g h ) Vi =V z12 Vi 1 Io (g m +g h ) From the previous homework we have: S11 z11 R 0 R 0 +z 22 z12 z 21 z11 +R 0 R 0 +z 22 z12 z 21 S21 2R 0 z 21 2R 0 z12 S12 z11 +R 0 R 0 +z22 z12z21 z11 +R 0 R 0 +z22 z12z21 Problem #28: S22 z 22 R 0 R 0 +z11 z12 z 21 z11 +R 0 R 0 +z 22 z12 z 21 2 ρ +G T 1 2 4R SR L R +R 4R V G T = S o = S 6L R L Vs 1+ω 2 2 R L RS 6 ω R +R 2 ρ S L 6 1+ω R L RS 3 s R S +R L ω=-js ρ 2s3 +2s 2 +2s+ 4 3 2 3 1+2s+2s +s ZINnormalized = 2s 2 +2s+ 2 3 1 ρ ZINnormalized 1 ρ Applying the long division: Zα =50Ω ZINnormalized = s+ 1 3 1 s+ 2 2s+2 Also we have Cα = Lα = 1 =1.33pF Zα ωα Zα =3.32nH ωα Thus, the network will be as in the following figure: Netlist Butterworth Network .opt post Vi 1 0 AC 1 Rs 1 2 50 L1 2 3 3.32n C 3 0 1.99p L2 3 4 6.63n Rl 4 0 100 .op .ac lin 100 0 100g .end Magnitude Response