ProblemSchematicTransient responseDesign procedureComment on the two-pole approximationGain and Phase Margin Problem Problem Select the frequency f1 in the gain expression of EQ. 1 below to obtain a two-pole Butterworth step response for a voltage feedback amplifier with βFB = 10 mV/V. EQ. 1 +++=υυ3210ffj1ffj1ffj1A)f(A. The frequencies f2 = 106Hz and f3 = 107 Hz, and the low-frequency gain is Aυ0 = 105 V/V. Also, determine the gain and phase margins of this amplifier. Schematic The PSPICE circuit for this amplifier is shown in Figure 1. -++-E1GAIN = 100{B_FB}PARAMETERS:f_1 = 500f_2 = 1Megf_3 = 10MegB_FB = 10mV/VR_1 = {1/(2*pi*C*f_1)}R_2 = {1/(2*pi*C*f_2)}R_3 = {1/(2*pi*C*f_3)}C = 1nFpi = 3.1415926Sweep+-ACV_AC1VVPWL_ENHFIRST_NPAIRS = 0,0, 0.5,0, 0.501,1, 100,1TSF = 1us/sVSF = 1V/V0-++-E4GAIN = 1V0-++-E3GAIN = 1000-++-E5GAIN = 1+R2{R_2}+C1{C}+C2{C}+C3{C}+R3{R_3}-++-E2GAIN = 100+R1{R_1}0 INOUTFIGURE 1 PSPICE representation of three-pole amplifier Using the circuit of Figure 1 we can find the gain and phase plots, as shown in Figure 2 and Figure 3 below. Frequency1.0Hz10Hz100Hz1.0KHz10KHz100KHz1.0MHz10MHzVDB(OUT)050100(Closed Loop f_3dB=739.8381443K,36.9897000)(Closed Loop,1.00,39.99)(Open Loop,1.00,100)(f_3dB=500.00,96.9896550)(Open Loop f_1/Bfb=454.6919600K,40.0002854)(Open Loop f_180=3.16315256M,13.1467863) FIGURE 2 PSPICE gain plots for open lop and closed loop amplifier; this plot determines the frequency f1/β_FB where the open-loop gain is 1/βFB; gain margin (26.8dB) is labeled with double arrow Create Date 4/22/06 by J R Brews Page 1 4/22/2006Frequency1.0Hz 10Hz 100Hz 1.0KHz 10KHz 100KHz 1.0MHz 10MHz 100MHz 1.0GHzP(V(OUT))-300d-200d-100d-0dClosed Loop(Open Loop f_3dB=500.0000000,-45.0305268)( f_3dB=739.8381443K,-98.2611966) (Open Loop f_1/Bfb=454.6919600K,-116.9909543)(Open Loop f_180=3.16315256M,-179.9999511)FIGURE 3 PSPICE phase plots for open lop and closed loop amplifier; this plot determines the frequency f180 where the phase is –180°; phase margin (63°) is labeled with double arrow Transient response Time0s 0.5us 1.0us 1.5us 2.0us 2.5us 3.0us 3.5us 4.0us 4.5us 5.0usV(OUT) V(V_IN/B_FB)0V40V80V120V(1.47594102u,105.7474442) FIGURE 4 Step response of closed loop amplifier; time to first maximum is 0.98 µs and overshoot is 5.7% The step response in Figure 4 can be compared with the two-pole estimates for a Butterworth design of tMAX = 1/(f1 + f2) = 1 µs and overshoot of 4.3%. Design procedure For a Butterworth response we require a time constant separation factor of 2βFBAυ0, so EQ. 2 0FB1221A2ffυβ≈=ττ. That is, f1 = f2/(2 × 10–2× 105) = 500 Hz. We then draw the Bode plots for EQ. 1 with the poles f1, f2, f3 that approximate Figure 2 and Figure 3, and determine the frequencies f1/β_FB and f180. The gain and phase margins are then EQ. 3 phase margin = [])180()f(AFB/1o−−βυarg = 63° EQ. 4 gain margin = ()(18010FB10fAog201og20υ−βll) = 26.9 dB The numerical work can be done very conveniently using a spreadsheet, as shown in Figure 5 below. Create Date 4/22/06 by J R Brews Page 2 4/22/2006Open-Loop AmplifierInput pi 3.1415926 Frequency Phase Gain Gain(dB)A_v0 1.00E+05 f_1/Bfb 4.55E+05 -116.9913 1.00E+02 39.99993f_1 500 f_180 3.16E+06 -180.0001 4.54E+00 13.14674f_2 1.00E+06 f_3dB 5.00E+02 -45.03151 7.07E+04 96.9897f_3 1.00E+07C_1 1.00E-09B_FB 1.00E-02Feedback AmplifierCalculated R_1 318309.89 Phase margin 63.00875R_2 159.15495 Gain Margin (dB) 26.85318R_3 15.915495A_v0 (dB) 100A_v (f_3dB) 96.9897A_vFB (f_3dB))36.9897Phase=-(ATAN2(1,Frequency/f_1)+ATAN2(1,Frequency/f_2)+ATAN2(1,Frequency/f_3))*180/piGain=A_v0/(SQRT(1+(Frequency/f_1)^2)*SQRT(1+(Frequency/f_2)^2)*SQRT(1+(Frequency/f_3)^2)) FIGURE 5 Spreadsheet for gain and phase margin calculations With the spreadsheet of Figure 5 the frequencies f1/β_FB and f180 are readily found using GOAL SEEK to set the magnitude to 1/βFB and the phase to –180° by varying the frequency. Comment on the two-pole approximation Figure 4 shows that the two-pole approximation to design for a Butterworth amplifier provides a good approximation for setting the lowest pole at f1 provided the higher poles are not too close to f2. In this example the overshoot in step response (Figure 4) is a bit larger than Butterworth because the third pole makes the gain margin a little lower than for the two-pole system comprised of only of poles at f1 and f2. The time to maximum overshoot is very nearly as expected. If we move f3 to very high frequency, the modified system approaches a two-pole system with phase margin 65.6°, a bit larger than our original system with 63° margin. So a two-pole estimate of phase margin is not a bad approximation, and accounts for the success of the two-pole Butterworth design. However, the two-pole estimate of gain margin is terrible, as explained next. You may recall that a two-pole system is always stable, with f180 = ∞. Of course, no real amplifier has f180 = ∞, so this two-pole estimate of f180 is hopelessly inaccurate. Because the gain of any amplifier tends to zero at very high frequencies, the gain → 0 as f180 → ∞. That is, the poor estimate of f180 using a two-pole system makes the two-pole estimate of gain margin hopelessly inaccurate for any real amplifier (log10(0) = – infinity), even if the two-pole system approximates the gain curve quite well over a range of frequencies from low values to somewhere above the second pole. For these reasons, it is more useful to focus on phase margin as a stability estimate when using a two-pole approximation, not gain margin. Create Date 4/22/06 by J R Brews Page 3