ENEE 303H Fall 2009 HW5 YUH revised version 10/31/09 1. (a) example of using 2 power supplies VDD Rd 0 Cc2 Ri Cc1 M1 0 Vi 0 Rg Cc3 Rs CL RL 0 -VSS 0 0 vsig vi Rg Ri Rg , Rsig Ri // Rg (b)(d) (Vsig-Vg)/Rsig sCgs(Vg-Vo) Rsig Cgs G S Vo sCL*Vo Vo/RL' Cgd Vsig sCgd*Vg CL RL' gm*vgs gmb*vbs B D 0 0 RL' RL // ro Now the controlled current source gmb*vbs can be replaced by a resistor 1/gmb since the voltage across it is vbs (source absorption). (Vsig-Vg)/Rsig sCgs(Vg-Vo) Rsig Cgs G S Vo sCL*Vo Vo/RL' Cgd sCgd*Vg Vsig CL RL' gm*vgs B D 0 0 RL' RL // ro // at Gate: 1 1 , YL' ' gm b RL vsig vg Rsig sC gs vg vo sC gd vg , so vg vsig sC gs Rsig vo s C gs C gd Rsig 1 1 at Source: g m vg vo sC gs vg vo vo sCL ' RL gm sC gs vsig sC gs Rsig vo vo sCL YL' g m sC gs sCgs Cgd Rsig 1 g sC v v sC Y g sC sC C R 1 g sC sC R g m sC gs vg vo sCL YL' g m sC gs m gs sig o ' L L m gs gs Denominator = Y ' L gd sig m gs gs sig vo g m s CL Cgs Cgs Cgd RsigYL' g mCgd Rsig s 2 Rsig CL Cgs Cgd CgsCgd C gm 1 s gs gm gm vo ' vsig CL C gs C gs C gd Rsig YL g mC gd Rsig Rsig CL C gs C gd C gsC gd 1 s s2 ' YL g m YL' g m Mid-band gain AM YL' gm gm YL' < 1 If HP2 HP1 ( HP1 is dominant), then (see p.593 of Sedra/Smith) HP1 Y gm CL Cgs Cgs Cgd RsigYL' g mCgd Rsig ' L C C gs C gs C gd Rsig YL' g mC gd Rsig HP 2 L Rsig C L C gs C gd C gsC gd Solve with MATLAB: syms s Rsig Vsig gm Cgs Cgd CL YL % here YL=1/RL', RL'=RL//ro//(1/gmb) L=[s*(Cgs+Cgd)+1/Rsig -s*Cgs;… -gm-s*Cgs gm+s*Cgs+s*CL+YL] R=[Vsig/Rsig;0] H=L\R % solve [L]*[vg;vo]=[R] vo=H(2) vo = Vsig*(gm+s*Cgs)/(gm*s*Rsig*Cgd+gm+s^2*Cgs*Rsig*Cgd+s*Cgs+s^2*CL* Rsig*Cgs+s^2*CL*Rsig*Cgd+s*CL+YL*s*Rsig*Cgs+YL*s*Rsig*Cgd+YL) So vo/Vsig is obtained. Solve with Mathematica: To make vo in terms of Vsig and vbs, the process is similar, we just need to modify the node equation at source. (Vsig-Vg)/Rsig sCgs(Vg-Vo) Rsig Cgs G S Vo sCL*Vo Vo/RL' Cgd sCgd*Vg Vsig CL RL' gm*vgs gmb*vbs D 0 RL' RL // ro and YL' at Gate: vsig vg Rsig 1 RL' sC gs vg vo sC gd vg , so vg vsig sC gs Rsig vo s C gs C gd Rsig 1 1 at Source: g m vg vo g mbvbs sC gs vg vo vo sCL ' RL gm sC gs vsig sC gs Rsig vo g mbvbs vo sCL YL' gm sC gs sCgs Cgd Rsig 1 gm sCgs vsig gmbvbs vo sCL YL' gm sCgs sCgs Cgd Rsig 1 gm sCgs sCgsRsig vo g m sC gs vg g mbvbs vo sCL YL' g m sC gs we get the same denominator as that of the previous case (the only difference is now RL' RL // ro ) C gs gm g vsig ' m b vbs 1 s ' gm YL g m YL g m vo ' CL C gs C gs C gd Rsig YL g mC gd Rsig 2 Rsig C L C gs C gd C gsC gd 1 s s YL' g m YL' g m vo v bs 0 C gs gm vsig 1 s g m YL' g m CL C gs C gs C gd Rsig YL' g mC gd Rsig Rsig CL C gs C gd C gsC gd 1 s s2 ' YL g m YL' g m vo v sig 0 1 s CL C gs C gs C gd Y ' L gm b vbs gm Rsig YL' g mC gd Rsig gm YL' s2 Rsig CL C gs C gd C gsC gd Y ' L gm So there exists a zero in finite frequencies for vbs=0, Z gm , Cgs which can be found from setting the current which flow into the load as zero. The following s-plane plot is based on Exercise 6.34, p.639. Note that the zero and poles are all on the negative real axis. Im s plane Re -gm/Cgs -whp2 -whp1 (c) Because Cgs, Cgd and CL form a continuous loop, the voltages across these three capacitors are not totally independent; therefore, the transfer function is of order 2. (e) Example: Exercise 6.34, p.639 clear RL=10e3 ; CL=15e-15 Cgs=20e-15 ; Cgd=5e-15 Rsig=20e3 ro=20e3 gm=1.25e-3 ; gmb=0.2*gm YLp=1/RL+1/ro+gmb AM=gm/(YLp+gm) % mid-band gain wz=gm/Cgs s1=(CL+Cgs+(Cgs+Cgd)*Rsig*YLp+gm*Cgd*Rsig)/(YLp+gm) % coefficient of s^1 s2=Rsig*(CL*(Cgs+Cgd)+Cgs*Cgd)/(YLp+gm) % coefficient of s^2 wroot=-roots([s2,s1,1]) whp1=wroot(2) whp2=wroot(1) % dominant pole approximation wp1=1/s1 wp2=1/s2/wp1 % plot with the poles solved by s polynomial f=logspace(6,12,1e4); w=2*pi*f; H=AM*(1+i*w/wz)./(1+i*w/whp1)./(1+i*w/whp2); % H=gmb/(YLp+gm)./(1+i*w/whp1)./(1+i*w/whp2); Hmag=20*log10(abs(H)); semilogx(f,Hmag) title('Magnitude Response','fontsize',14) xlabel('Frequency, Hz','fontsize',14) ylabel('Magnitude, dB','fontsize',14) grid on we get fhp1=whp1/2/pi= 849 MHz (by solving s polynomial) fp1=wp1/2/pi= 729.5 MHz (by dominant pole approximation) fz=wz/2/pi= 9.95 GHz The following magnitude responses are plotted with the poles solved by s polynomial because the two poles are not too far away from each other (however, the numbers given by dominant pole approximation 729.5 MHz and method of open-circuit time constant 730 MHz are still not bad). vbs=0 Magnitude Response 0 -10 Magnitude, dB -20 -30 -40 -50 -60 -70 6 10 7 10 8 10 9 10 10 10 11 10 12 10 Frequency, Hz Vsig=0 Magnitude Response 0 -20 Magnitude, dB -40 -60 -80 -100 -120 -140 6 10 7 10 8 10 9 10 Frequency, Hz 10 10 11 10 12 10 2.(a) M4007P VDD 10Vdc M4007P M2 w M4 x y z Vw M1 M3 M4007N M4007N 0 10V 8V 6V 4V 2V 0V 0V 1V V(X) V(z) 2V 3V 4V 5V V(w) V_Vw A(0,0) B~=(VDD/2,VDD/2) C(VDD,VDD) 6V 7V 8V 9V 10V (b) VDD1 10Vdc M4007P M4007P M6 w2 M8 x2 z2 M5 M7 M4007N M4007N 0 8.0V 7.0V 6.0V 5.0V 4.0V 3.0V 2.0V 0s V(x2) 2us V(z2) 4us 6us 8us 10us 12us 14us 16us 18us 20us Time Without perturbation, the latch circuit stays at point B (but this can’t be seen in experiment). VDD1 10Vdc M4007P M4007P M6 w2 I1 = 0 I2 = 1u TD = 10n TR = 0 TF = 0 PW = 1n PER = 1m M8 x2 z2 I2 M5 M7 M4007N M4007N 0 10V 8V 6V 4V 2V 0V 0s 50ns V(x2) 100ns 150ns 200ns 250ns 300ns V(z2) Time With the current pulse perturbation, V(x) goes to high and V(z) goes to low very quickly via feedback loop. 10V 8V 6V 4V 2V 0V 0V V(x2) 0.5V V(z2) 1.0V 1.5V 2.0V 2.5V 3.0V V(W2) (c) VDD3 10Vdc M4007P M4007P M10 w3 M12 x3 z3 Cw 10n IC=0 M9 M11 M4007N M4007N 0 3.5V 4.0V 4.5V 5.0V 5.5V Initial condition: V(Cw) = 0 V 12V 8V 4V 0V -2V 0s 5us V(x3) 10us 15us 20us 25us 30us V(z3) Time 12V 8V 4V 0V -2V 0V 0.2nV V(x3) 0.4nV 0.6nV 0.8nV 1.0nV V(z3) V(w3) V(x) stays at high and V(z) stays at low. 1.2nV 1.4nV 1.6nV Initial condition: V(Cw) = 5 V 10V 8V 6V 4V 2V 0V 0s 5us V(x3) 10us 15us 20us 25us 30us V(z3) Time 10V 8V 6V 4V 2V 0V 0V V(x3) 0.5V V(z3) 1.0V 1.5V 2.0V 2.5V V(w3) 3.0V 3.5V 4.0V 4.5V 5.0V Initial condition: V(Cw) = 5.03 V (compare with initial condition V(Cw) = 5 V) 10V 8V 6V 4V 2V 0V 0s 5us V(x3) 10us 15us 20us 25us 30us V(z3) Time 10V 8V 6V 4V 2V 0V 5.0V V(x3) 5.5V V(z3) 6.0V 6.5V 7.0V 7.5V V(w3) 8.0V 8.5V 9.0V 9.5V 10.0V Initial condition of Cw = 10 V 12V 8V 4V 0V -2V 0s 5us V(x3) 10us 15us 20us 25us 30us V(z3) Time 12V 8V 4V 0V -2V 2.96nV V(x3) 2.98nV V(z3) 3.00nV 3.02nV 3.04nV 3.06nV 3.08nV V(x3) V(x) stays at low and V(z) stays at high. 3.10nV 3.12nV 3.14nV