Control_Systems_Engineering_-_Solution_Manual_(Nise)

Control_Systems_Engineering_-_Solution_Manual_(Nise) - ONE...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ONE Introduction ANSWERS TO REVIEW QUESTIONS 1. Guided missiles, automatic gain control in radio receivers, satellite tracking antenna 2. Yes - power gain, remote control, parameter conversion; No - Expense, complexity 3. Motor, low pass filter, inertia supported between two bearings 4. Closed-loop systems compensate for disturbances by measuring the response, comparing it to the input response (the desired output), and then correcting the output response. 5. Under the condition that the feedback element is other than unity 6. Actuating signal 7. Multiple subsystems can time share the controller. Any adjustments to the controller can be implemented with simply software changes. 8. Stability, transient response, and steady-state error 9. Steady-state, transient 10. It follows a growing transient response until the steady-state response is no longer visible. The system will either destroy itself, reach an equilibrium state because of saturation in driving amplifiers, or hit limit stops. 11. Transient response 12. True 13. Transfer function, state-space, differential equations 14. Transfer function - the Laplace transform of the differential equation State-space - representation of an nth order differential equation as n simultaneous first-order differential equations Differential equation - Modeling a system with its differential equation SOLUTIONS TO PROBLEMS 1. Five turns yields 50 v. Therefore K = 50 volts = 1.59 5 x 2π rad 2 Chapter 1: Introduction 2. Desired temperature Voltage difference Temperature difference Actual temperature Fuel flow + Amplifier and valves Thermostat Heater - 3. - Roll angle Roll rate Aileron position control + Pilot controls Aileron position Error voltage Input voltage Desired roll angle Aircraft dynamics Integrate Gyro Gyro voltage 4. Input voltage Desired speed Speed Error voltage + transducer Motor and drive system Amplifier Voltage proportional to actual speed Dancer position sensor Actual speed Dancer dynamics 5. Input voltage Desired power Transducer Rod position Power Error voltage + Amplifier - Voltage proportional to actual power Motor and drive system Sensor & transducer Actual power Reactor Solutions to Problems 3 6. Desired student population Desired student rate Population error + Graduating and drop-out rate Administration Actual student rate + Actual student population Net rate of influx Integrate Admissions - 7. Voltage proportional to desired volume Desired volume + Transducer Volume error Voltage representing actual volume Volume control circuit Radio Effective volume + - Transducer Speed Voltage proportional to speed Actual volume 4 Chapter 1: Introduction 8. a. Fluid input Valve Actuator Power amplifier +V Differential amplifier + - R Desired level -V +V R Float -V Tank Drain b. Desired level Potentiometer voltage in + Amplifiers - Actuator and valve Actual level Flow rate in + Integrate - Drain Flow rate out voltage out Displacement Potentiometer Float Solutions to Problems 5 9. Desired force Current + Transducer Displacement Amplifier Actual force Displacement Actuator and load Valve Tire - Load cell 10. Commanded blood pressure + Actual blood pressure Isoflurane concentration Vaporizer Patient - 11. Desired depth + Controller & motor - Force Feed rate Grinder Depth Integrator 12. Coil voltage + Desired position Coil circuit Transducer Coil current Solenoid coil & actuator - LVDT 13. a. L di + Ri = u(t) dt Force Armature & spool dynamics Depth 6 Chapter 1: Introduction b. Assume a steady-state solution iss = B. Substituting this into the differential equation yields RB = 1, 1 R . The characteristic equation is LM + R = 0, from which M = - . Thus, the total R L 1 1 solution is i(t) = Ae-(R/L)t + . Solving for the arbitrary constants, i(0) = A + = 0. Thus, A = R R 1 1 1 -(R/L)t 1 −( R / L) t . The final solution is i(t) = -= (1 − e e ). R RR R from which B = c. 14. di 1 idt + vC (0 ) = v(t ) + dt C ∫ d 2i di b. Differentiating and substituting values, + 30i = 0 2 +2 dt dt a. Writing the loop equation, Ri + L Writing the characteristic equation and factoring, 2 M + 2 M + 30 = M + 1 + 29 i M + 1 - 29 i . The general form of the solution and its derivative is -t i = e cos 29 t A + B sin 29 t e -t di = - A + 29 B e -t cos 29 t - 29 A + B e- t sin dt v (0 ) 1 di Using i (0) = 0; (0) = L = =2 L L dt i 0 = A =0 di (0) = − A + 29 B =2 dt 2 . Thus, A = 0 and B = 29 The solution is 29 t Solutions to Problems 7 i= 2 29 -t 29 t 29 e sin c. i t 15. a. Assume a particular solution of Substitute into the differential equation and obtain Equating like coefficients, 35 From which, C = 53 10 and D = 53 . The characteristic polynomial is Thus, the total solution is 35 35 Solving for the arbitrary constants, x(0) = A +53 = 0. Therefore, A = - 53 . The final solution is b. Assume a particular solution of xp = Asin3t + Bcos3t 8 Chapter 1: Introduction Substitute into the differential equation and obtain (18A − B)cos(3t) − (A + 18B)sin(3t) = 5sin(3t) Therefore, 18A – B = 0 and –(A + 18B) = 5. Solving for A and B we obtain xp = (-1/65)sin3t + (-18/65)cos3t The characteristic polynomial is 2 M +6 M+ 8 = M+ 4 M+ 2 Thus, the total solution is 18 1 cos 3 t sin 3 t 65 65 18 Solving for the arbitrary constants, x (0) = C + D − =0. 65 x =C e -4t + De -2t +- Also, the derivative of the solution is dx = - 3 cos 3 t + 54 sin 3 t -4t -2t - 4 C e -2 D e dt 65 65 . 15 3 3 . Solving for the arbitrary constants, x(0) − − 4C − 2 D = 0 , or C = − and D = 10 65 26 The final solution is x =- 18 3 - 4 t 15 - 2 t 1 e e+ sin 3 t cos 3 t 26 65 10 65 c. Assume a particular solution of xp = A Substitute into the differential equation and obtain 25A = 10, or A = 2/5. The characteristic polynomial is 2 M + 8 M + 25 = M + 4 + 3 i M + 4 - 3 i Thus, the total solution is x= 2 -4t +e B sin 3 t + C cos 3 t 5 Solving for the arbitrary constants, x(0) = C + 2/5 = 0. Therefore, C = -2/5. Also, the derivative of the solution is dx -4t dt = 3 B - 4 C cos 3 t - 4 B + 3 C sin 3 t e Solutions to Problems 9 . Solving for the arbitrary constants, x(0) = 3B – 4C = 0. Therefore, B = -8/15. The final solution is x (t ) = 2 2 ⎛8 − e −4 t sin(3t ) + cos(3t )⎞ ⎝ 15 ⎠ 5 5 16. a. Assume a particular solution of Substitute into the differential equation and obtain Equating like coefficients, 1 From which, C = - 5 1 and D = - 10 . The characteristic polynomial is Thus, the total solution is 11 1 Solving for the arbitrary constants, x(0) = A - 5 = 2. Therefore, A = . Also, the derivative of the 5 solution is dx dt . 3 Solving for the arbitrary constants, x(0) = - A + B - 0.2 = -3. Therefore, B = − . The final solution 5 is 1 1 3 −t ⎛ 11 x (t ) = − cos( 2t ) − sin(2 t ) + e cos(t ) − sin(t )⎞ ⎝5 ⎠ 5 10 5 b. Assume a particular solution of xp = Ce-2t + Dt + E Substitute into the differential equation and obtain 10 Chapter 1: Introduction Equating like coefficients, C = 5, D = 1, and 2D + E = 0. From which, C = 5, D = 1, and E = - 2. The characteristic polynomial is Thus, the total solution is Solving for the arbitrary constants, x(0) = A + 5 - 2 = 2 Therefore, A = -1. Also, the derivative of the solution is dx = ( − A + B )e − t − Bte −t − 10e −2 t + 1 dt . Solving for the arbitrary constants, x(0) = B - 8 = 1. Therefore, B = 9. The final solution is c. Assume a particular solution of xp = Ct2 + Dt + E Substitute into the differential equation and obtain 1 Equating like coefficients, C = 4 , D = 0, and 2C + 4E = 0. 1 1 From which, C = 4 , D = 0, and E = - 8 . The characteristic polynomial is Thus, the total solution is 9 1 Solving for the arbitrary constants, x(0) = A - 8 = 1 Therefore, A = 8 . Also, the derivative of the solution is dx dt . Solving for the arbitrary constants, x(0) = 2B = 2. Therefore, B = 1. The final solution is Solutions to Problems 11 17. Spring displacement Desired force Input voltage + Input transducer F up Controller Actuator - Sensor Pantograph dynamics F out Spring TWO Modeling in the Frequency Domain SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Transfer Functions Finding each transfer function: Vi(s) 10 = ; θi(s) π Vp(s) Pre-Amp: V (s) = K; i Ea(s) 150 Power Amp: V (s) = s+150 p Pot: 2 ( ) = 0.25 50 2 Dm =0.01 + 3(250 ) = 0.13 50 Motor: Jm = 0.05 + 5 250 Kt 1 Ra = 5 K tK b 1 Ra = 5 θm(s) Therefore: E (s) = a Kt RaJm 0.8 KtKb = s(s+1.32) 1 s(s+J (Dm+ R )) m a θo(s) 1 θm(s) 0.16 And: E (s) = 5 E (s) = s(s+1.32) a a Transfer Function of a Nonlinear Electrical Network Writing the differential equation, 2 2 0 d(i 0 + δi) 2 + 2(i0 + δi) − 5 = v(t ) . Linearizing i2 about i0, dt 2 2 0 . (i + δi) - i = 2i ⎮ δi = 2i δi. Thus, (i + δi) = i + 2i δi. 0 i=i 0 0 0 0 Solutions to Problems 13 dδi Substituting into the differential equation yields, dt + 2i02 + 4i0δi - 5 = v(t). But, the resistor voltage equals the battery voltage at equilibrium when the supply voltage is zero since the voltage across the inductor is zero at dc. Hence, 2i02 = 5, or i0 = 1.58. Substituting into the linearized dδi δi(s) 1 differential equation, dt + 6.32δi = v(t). Converting to a transfer function, V(s) = s+6.32 . Using the linearized i about i0, and the fact that vr(t) is 5 volts at equilibrium, the linearized vr(t) is vr(t) = 2i2 = 2(i0+δi)2 = 2(i02+2i0δi) = 5+6.32δi. For excursions away from equilibrium, vr(t) - 5 = 6.32δi = δvr(t). δVr(s) 6.32 Therefore, multiplying the transfer function by 6.32, yields, V(s) = s+6.32 as the transfer function about v(t) = 0. ANSWERS TO REVIEW QUESTIONS 1. Transfer function 2. Linear time-invariant 3. Laplace 4. G(s) = C(s)/R(s), where c(t) is the output and r(t) is the input. 5. Initial conditions are zero 6. Equations of motion 7. Free body diagram 8. There are direct analogies between the electrical variables and components and the mechanical variables and components. 9. Mechanical advantage for rotating systems 10. Armature inertia, armature damping, load inertia, load damping 11. Multiply the transfer function by the gear ratio relating armature position to load position. 12. (1) Recognize the nonlinear component, (2) Write the nonlinear differential equation, (3) Select the equilibrium solution, (4) Linearize the nonlinear differential equation, (5) Take the Laplace transform of the linearized differential equation, (6) Find the transfer function. SOLUTIONS TO PROBLEMS 1. ∞ a. ∞ 1 1 −st − st = F( s) = ∫ e dt = − e s s 0 0 ∞ e −st − (st + 1) ∞ ∞ b. F( s) = ∫ te dt = 2 ( − st − 1) 0 = s s2 e st 0 0 − st 14 Chapter 2: Modeling in the Frequency Domain Using L'Hopital's Rule F( s) t → ∞ = −s s 3e st = 0. Therefore, F( s) = t →∞ 1 . s2 ∞ ∞ e − st ω c. F( s ) = ∫ sin ωt e dt = 2 =2 2 ( − s sin ωt − ω cos ωt ) 2 s +ω s +ω 0 0 − st ∞ ∞ s e − st d. F( s ) = ∫ cos ωt e dt = 2 =2 2 ( − s cosω t + ω sin ωt ) s +ω2 s +ω 0 0 − st 2. ω . (s+a)2+ω2 (s+a) b. Using the frequency shift theorem and the Laplace transform of cos ωt, F(s) = . (s+a)2+ω2 a. Using the frequency shift theorem and the Laplace transform of sin ωt, F(s) = t2 c. Using the integration theorem, and successively integrating u(t) three times, ⌠dt = t; ⌠tdt = 2 ; ⌡ ⌡ 2 3 ⌠t dt = t , the Laplace transform of t3u(t), F(s) = 6 . 6 ⌡2 s4 3. a. The Laplace transform of the differential equation, assuming zero initial conditions, 5s is, (s+7)X(s) = 2 2 . Solving for X(s) and expanding by partial fractions, s +2 Or, 35 10 35 Taking the inverse Laplace transform, x(t) = - 53 e-7t + (53 cos 2t + 53 sin 2t). b. The Laplace transform of the differential equation, assuming zero initial conditions, is, (s2+6s+8)X(s) = 15 . s +9 2 Solving for X(s) X(s) = 15 (s + 9)(s2 + 6s + 8) 2 and expanding by partial fractions, 3 X(s) = − 65 1 9 15 1 31 9 + − 2 10 s + 4 26 s + 2 s +9 6s + Solutions to Problems 15 Taking the inverse Laplace transform, x(t) = − 18 1 3 15 cos(3t) − sin(3t) − e −4t + e −2t 65 65 10 26 c. The Laplace transform of the differential equation is, assuming zero initial conditions, (s2+8s+25)x(s) = 10 . Solving for X(s) s Xs= 10 s s 2 + 8 s + 25 and expanding by partial fractions, 21 2 Xs= 5s 5 4 9 2 s+4 +9 1 s+4 + 9 Taking the inverse Laplace transform, x (t ) = 2 2 ⎛8 − e −4 t sin(3t ) + cos(3t )⎞ ⎝ 15 ⎠ 5 5 4. 2 a. Taking the Laplace transform with initial conditions, s2X(s)-2s+3+2sX(s)-4+2X(s) = 2 2 . s +2 Solving for X(s), 2s 3 + s2 + 8s + 6 . X(s) = 2 2 (s + 4)( s + 2 s + 2) Expanding by partial fractions ⎛1 X(s) = − ⎝ ⎞ 5⎠ 1 3 4 11(s + 1) − 1 ⎛ 1⎞ 4 1 +⎝ ⎠ 2 2 s +4 5 (s + 1) + 1 s+ Therefore, x(t) = -0.2 cos2t - 0.1 sin2t +e-t (2.2 cost - 0.6 sint). 5 1 b. Taking the Laplace transform with initial conditions, s2X(s)-2s-1+2sX(s)-4+X(s) = s+2 + 2 . s Solving for X(s), Therefore, x(t) = 5e-2t - e-t + 9te-t - 2 + t. 2 c. Taking the Laplace transform with initial conditions, s2X(s)-s-2+4X(s) = 3 . Solving for X(s), s 16 Chapter 2: Modeling in the Frequency Domain 11 9 Therefore, x(t) = 8 cos2t + sin2t - 8 + 4 t2. 5. Program: syms t f=5*t^2*cos(3*t+45); pretty(f) F=laplace(f); F=simple(F); pretty(F) 'b' f=5*t*exp(-2*t)*sin(4*t+60); pretty(f) F=laplace(f); F=simple(F); pretty(F) Computer response: ans = a 2 5 t cos(3 t + 45) 3 2 s cos(45) - 27 s cos(45) - 9 s sin(45) + 27 sin(45) 10 ----------------------------------------------------2 3 (s + 9) ans = b 5 t exp(-2 t) sin(4 t + 60) sin(60) ((s + 2) sin(60) + 4 cos(60)) (s + 2) -5 ------------- + 10 ------------------------------------2 2 2 (s + 2) + 16 ((s + 2) + 16) 6. Program: syms s 'a' G=(s^2+3*s+7)*(s+2)/[(s+3)*(s+4)*(s^2+2*s+100)]; pretty(G) g=ilaplace(G); pretty(g) 'b' G=(s^3+4*s^2+6*s+5)/[(s+8)*(s^2+8*s+3)*(s^2+5*s+7)]; pretty(G) g=ilaplace(G); pretty(g) Computer response: ans = a 2 Solutions to Problems 17 (s + 3 s + 7) (s + 2) -------------------------------2 (s + 3) (s + 4) (s + 2 s + 100) 11 4681 1/2 1/2 - 7/103 exp(-3 t) + -- exp(-4 t) - ----- exp(-t) 11 sin(3 11 t) 54 61182 4807 1/2 + ---- exp(-t) cos(3 11 t) 5562 ans = b 3 2 s +4s +6s+5 ------------------------------------2 2 (s + 8) (s + 8 s + 3) (s + 5 s + 7) 299 1367 1/2 - --- exp(-8 t) + ---- exp(-4 t) cosh(13 t) 93 417 4895 1/2 1/2 - ---- exp(-4 t) 13 sinh(13 t) 5421 232 1/2 1/2 - ----- exp(- 5/2 t) 3 sin(1/2 3 t) 12927 272 1/2 - ---- exp(- 5/2 t) cos(1/2 3 t) 4309 7. The Laplace transform of the differential equation, assuming zero initial conditions, is, (s3+3s2+5s+1)Y(s) = (s3+4s2+6s+8)X(s). Y( s) s3 + 4s2 + 6 s + 8 Solving for the transfer function, = . X (s) s3 + 3s 2 + 5s + 1 8. a. Cross multiplying, (s2+2s+7)X(s) = F(s). Taking the inverse Laplace transform, d 2 x dx + 7x = f(t). 2 +2 dt dt b. Cross multiplying after expanding the denominator, (s2+15s+56)X(s) = 10F(s). d2 x dx Taking the inverse Laplace transform, + 56x =10f(t). 2 + 15 dt dt c. Cross multiplying, (s3+8s2+9s+15)X(s) = (s+2)F(s). Taking the inverse Laplace transform, 9. d 3 x d 2 x dx df (t ) + 15x = +2f(t). 2 +9 3 +8 dt dt dt dt s5 + 2s 4 + 4 s3 + s2 + 3 C( s ) The transfer function is =6 . 5 4 3 2 R( s ) s + 7s + 3s + 2s + s + 3 18 Chapter 2: Modeling in the Frequency Domain Cross multiplying, (s6+7s5+3s4+2s3+s2+3)C(s) = (s5+2s4+4s3+s2+3)R(s). Taking the inverse Laplace transform assuming zero initial conditions, d 6 c d 5 c d 4c d 3 c d 2 c d 5r d 4r d 3r d 2r +7 5 +3 4 +2 3 + + 3c = +2 4 +4 3 + 6 2 5 2 + 3r . dt dt dt dt dt dt dt dt dt 10. The transfer function is s 4 + 2s3 + 5s2 + s + 1 C( s ) =5 . R( s ) s + 3s 4 + 2s3 + 4s2 + 5s + 2 Cross multiplying, (s5+3s4+2s3+4s2+5s+2)C(s) = (s4+2s3+5s2+s+1)R(s). Taking the inverse Laplace transform assuming zero initial conditions, d 4r d 3r d 2 r dr d 5 c d 4c d 3 c d 2 c dc + r. + 2c = +2 3 +5 2 + +3 4 +2 3 +4 2 +5 dt 4 dt dt dt dt dc 5 dt dt dt d 5 c d 4c d 3 c d 2 c dc +3 4 +2 3 +4 2 +5 Substituting r(t) = t , + 2c dt dc 5 dt dt dt 3 = 18δ(t) + (36 + 90t + 9t2 + 3t3) u(t). 11. Taking the Laplace transform of the differential equation, s2X(s)-s+1+2sX(s)-2+3x(s)=R(s). Collecting terms, (s2+2s+3)X(s) = R(s)+s+1. Solving for X(s), X(s) = R( s) s +1 +2 . s + 2s + 3 s + 2s + 3 2 The block diagram is then, 12. Program: 'Factored' Gzpk=zpk([-15 -26 -72],[0 -55 roots([1 5 30])' roots([1 27 52])'],5) 'Polynomial' Gp=tf(Gzpk) Computer response: ans = Factored Zero/pole/gain: 5 (s+15) (s+26) (s+72) -------------------------------------------s (s+55) (s+24.91) (s+2.087) (s^2 + 5s + 30) ans = Solutions to Problems 19 Polynomial Transfer function: 5 s^3 + 565 s^2 + 16710 s + 140400 -------------------------------------------------------------------s^6 + 87 s^5 + 1977 s^4 + 1.301e004 s^3 + 6.041e004 s^2 + 8.58e004 s 13. Program: 'Polynomial' Gtf=tf([1 25 20 15 42],[1 13 9 37 35 50]) 'Factored' Gzpk=zpk(Gtf) Computer response: ans = Polynomial Transfer function: s^4 + 25 s^3 + 20 s^2 + 15 s + 42 ----------------------------------------s^5 + 13 s^4 + 9 s^3 + 37 s^2 + 35 s + 50 ans = Factored Zero/pole/gain: (s+24.2) (s+1.35) (s^2 - 0.5462s + 1.286) -----------------------------------------------------(s+12.5) (s^2 + 1.463s + 1.493) (s^2 - 0.964s + 2.679) 14. Program: numg=[-10 -60]; deng=[0 -40 -30 (roots([1 7 100]))' (roots([1 6 90]))']; [numg,deng]=zp2tf(numg',deng',1e4); Gtf=tf(numg,deng) G=zpk(Gtf) [r,p,k]=residue(numg,deng) Computer response: Transfer function: 10000 s^2 + 700000 s + 6e006 ----------------------------------------------------------------------------s^7 + 83 s^6 + 2342 s^5 + 33070 s^4 + 3.735e005 s^3 + 2.106e006 s^2 + 1.08e007 s Zero/pole/gain: 10000 (s+60) (s+10) -----------------------------------------------s (s+40) (s+30) (s^2 + 6s + 90) (s^2 + 7s + 100) r= -0.0073 0.0313 2.0431 2.0431 -2.3329 -2.3329 0.5556 p= + + - 2.0385i 2.0385i 2.0690i 2.0690i 20 Chapter 2: Modeling in the Frequency Domain -40.0000 -30.0000 -3.5000 -3.5000 -3.0000 -3.0000 0 + + - 9.3675i 9.3675i 9.0000i 9.0000i k= 15. Program: syms s '(a)' Ga=45*[(s^2+37*s+74)*(s^3+28*s^2+32*s+16)]... /[(s+39)*(s+47)*(s^2+2*s+100)*(s^3+27*s^2+18*s+15)]; 'Ga symbolic' pretty(Ga) [numga,denga]=numden(Ga); numga=sym2poly(numga); denga=sym2poly(denga); 'Ga polynimial' Ga=tf(numga,denga) 'Ga factored' Ga=zpk(Ga) '(b)' Ga=56*[(s+14)*(s^3+49*s^2+62*s+53)]... /[(s^2+88*s+33)*(s^2+56*s+77)*(s^3+81*s^2+76*s+65)]; 'Ga symbolic' pretty(Ga) [numga,denga]=numden(Ga); numga=sym2poly(numga); denga=sym2poly(denga); 'Ga polynimial' Ga=tf(numga,denga) 'Ga factored' Ga=zpk(Ga) Computer response: ans = (a) ans = Ga symbolic 2 3 2 (s + 37 s + 74) (s + 28 s + 32 s + 16) 45 ----------------------------------------------------------2 3 2 (s + 39) (s + 47) (s + 2 s + 100) (s + 27 s + 18 s + 15) ans = Ga polynimial Transfer function: 45 s^5 + 2925 s^4 + 51390 s^3 + 147240 s^2 + 133200 s + 53280 -------------------------------------------------------------------------------- Solutions to Problems 21 s^7 + 115 s^6 + 4499 s^5 + 70700 s^4 + 553692 s^3 + 5.201e006 s^2 + 3.483e006 s + 2.75e006 ans = Ga factored Zero/pole/gain: 45 (s+34.88) (s+26.83) (s+2.122) (s^2 + 1.17s + 0.5964) ----------------------------------------------------------------(s+47) (s+39) (s+26.34) (s^2 + 0.6618s + 0.5695) (s^2 + 2s + 100) ans = (b) ans = Ga symbolic 3 2 (s + 14) (s + 49 s + 62 s + 53) 56 ---------------------------------------------------------2 2 3 2 (s + 88 s + 33) (s + 56 s + 77) (s + 81 s + 76 s + 65) ans = Ga polynimial Transfer function: 56 s^4 + 3528 s^3 + 41888 s^2 + 51576 s + 41552 -------------------------------------------------------------------------------s^7 + 225 s^6 + 16778 s^5 + 427711 s^4 + 1.093e006 s^3 + 1.189e006 s^2 + 753676 s + 165165 ans = Ga factored Zero/pole/gain: 56 (s+47.72) (s+14) (s^2 + 1.276s + 1.111) --------------------------------------------------------------------------(s+87.62) (s+80.06) (s+54.59) (s+1.411) (s+0.3766) (s^2 + 0.9391s + 0.8119) 16. a. Writing the node equations, b. Thevenizing, Vo − Vi Vo 1 V . + + Vo = 0 . Solve for o = s s Vi s + 2 22 Chapter 2: Modeling in the Frequency Domain 1 Vi (s) 1 V (s ) s Using voltage division, Vo (s ) = . Thus, o =2 2 1 +s+ 1 Vi (s) 2s + s + 2 2 s 17. a. Writing mesh equations (s+1)I1(s) – I2(s) = Vi(s) -I1(s) + (s+2)I2(s) = 0 But, I1(s) = (s+2)I2(s). Substituting this in the first equation yields, or (s+1)(s+2)I2(s) – I2(s) = Vi(s) I2(s)/Vi(s) = 1/(s2 + 3s + 1) But, VL(s) = sI2(s). Therefore, VL(s)/Vi(s) = s/(s2 + 3s + 1). b. Solutions to Problems 23 i1(t) i2(t) 2 1 (2 + )I1 ( s) − (1 + ) I2 ( s) = V (s) s s 1 1 −(1 + ) I1 ( s) + (2 + + 2s )I2 (s) = 0 s s Solving for I2(s): 2(s + 1) V (s ) s s +1 − 0 V (s)s s I2 (s ) = =2 2(s + 1) s +1 4 s + 3s + 1 − s s s + 1 2 s2 + 2s + 1 − s s I2(s) VL(s) 2s2 Therefore, V(s) = 2s V(s) = 2 4s +3s+1 18. a. Writing mesh equations, (2s + 1)I1(s) – I2(s) = Vi(s) -I1(s) + (3s + 1 + 2/s)I2(s) = 0 Solving for I2(s), 24 Chapter 2: Modeling in the Frequency Domain 2 s + 1 Vi (s) −1 0 I2 (s ) = 2s + 1 −1 3s 2 + s + 2 −1 s Solving for I2(s)/Vi(s), I2 (s ) s =3 2 Vi ( s) 6 s + 5s + 4 s + 2 But Vo(s) = I2(s)3s. Therefore , G(s) = 3s2/(6s3 + 5s2 +4s + 2). b. Transforming the network yields, Writing the loop equations, s s )I1 (s ) − 2 I (s) − sI 3 (s) = Vi (s ) s +1 s +1 2 s s 1 −2 I1 (s ) + ( 2 + 1 + )I2 ( s ) − I3 (s ) = 0 s +1 s +1 s (s + 2 − sI1 (s) − I2 (s) + (2s + 1)I3 (s) = 0 Solving for I2(s), I2 (s ) = s(s2 + 2s + 2) V ( s) s 4 + 2s 3 + 3s2 + 3s + 2 i I2(s) (s 2 + 2s + 2) But, Vo(s) = s = 4 V ( s) . Therefore, s + 2s3 + 3s2 + 3s + 2 i s2 + 2 s + 2 Vo (s) = Vi (s) s4 + 2 s3 + 3s2 + 3s + 2 19. a. Writing the nodal equations yields, Solutions to Problems 25 VR (s) − Vi (s) VR ( s) VR (s) − VC ( s) + + =0 2s 1 3s 1 1 ⎛1 − VR (s) + s + ⎞ VC (s) = 0 ⎝2 3s 3s ⎠ Rewriting and simplifying, 6s + 5 1 1 VR ( s) − VC (s) = Vi ( s) 6s 3s 2s ⎛ 3s2 + 2 ⎞ 1 V (s ) = 0 − VR (s) + ⎜ ⎝ 6s ⎠ C 3s Solving for VR(s) and VC(s), 1 Vi (s) 2s 0 VR (s ) = 6s + 5 6s 1 − 3s 1 3s 3s 2 + 2 6s ; VC ( s ) = 1 − 3s 3s 2 + 2 6s − 6s + 5 6s 1 − 3s 6s + 5 6s 1 − 3s 1 Vi ( s ) 2s 0 1 3s 3s 2 + 2 6s − Solving for Vo(s)/Vi(s) Vo (s) VR ( s) − VC (s) 3s2 = =3 Vi (s) Vi ( s) 6s + 5s2 + 4s + 2 b. Writing the nodal equations yields, (V1 (s) − Vi (s)) (s 2 + 1) + V1 (s) + ( V1 ( s) − Vo (s)) = 0 s s (V (s) − Vi (s)) (Vo (s) − V1 (s)) + sVo ( s) + o =0 s Rewriting and simplifying, 2 1 + 1)V1 ( s) − Vo (s) = Vi (s) s s 1 1 V1 (s) + ( s + + 1)Vo (s) = Vi (s) s s (s + 26 Chapter 2: Modeling in the Frequency Domain Solving for Vo(s) Vo(s) = (s 2 + 2s + 2) V ( s) . s 4 + 2s3 + 3s2 + 3s + 2 i Hence, Vo (s) ( s2 + 2 s + 2) =4 Vi (s) s + 2 s3 + 3s2 + 3s + 2 20. a. Mesh: (2+2s)I1(s) - (1+2s)I2(s) - I3(s) = V(s) - (1+2s)I1(s) + (7+5s)I2(s) - (2+3s)I3(s) = 0 5 -I1(s) - (2+3s)I2(s) + (3+3s+ s )I3(s) = 0 Nodal: V1(s) (V1(s) - Vo(s)) V1(s) - V(s) + (1+2s) + =0 2+3s (Vo(s) - V1(s)) Vo(s) (Vo(s) - V(s)) +4 + =0 2+3s 5 s or 6s 2 + 12s + 5 1 V1(s) Vo(s) = V(s) 2 6s + 7s + 2 3s + 2 1 12s2 + 23s + 30 s 1 V1(s) + Vo(s) = V(s) − 20 3s + 2 5 3s + 2 b. Program: syms s V %Construct symbolic object for frequency Solutions to Problems 27 %variable 's' and V. 'Mesh Equations' A2=[(2+2*s) V -1 -(1+2*s) 0 -(2+3*s) -1 0 (3+3*s+(5/s))] A=[(2+2*s) -(1+2*s) -1 -(1+2*s) (7+5*s) -(2+3*s) -1 -(2+3*s) (3+3*s+(5/s))] I2=det(A2)/det(A); G1=I2/V; G=4*G1; 'G(s) via Mesh Equations' pretty(G) %Form Ak = A2. %Form A. %Use Cramer's Rule to solve for I2. %Form transfer function, G1(s) = I2(s)/V(s). %Form transfer function, G(s) = V4(s)/V(s). %Display label. %Pretty print G(s) 'Nodal Equations' A2=[(6*s^2+12*s+5)/(6*s^2+7*s+2) V -1/(3*s+2) s*(V/5)] %Form Ak = A2. A=[(6*s^2+12*s+5)/(6*s^2+7*s+2) -1/(3*s+2) -1/(3*s+2) (1/20)*(12*s^2+23*s+30)/(3*s+2)] %Form A. I2=simple(det(A2))/simple(det(A)); %Use Cramer's Rule to solve for I2. G1=I2/V; %Form transfer function, G1(s) = I2(s)/V(s). 'G(s) via Nodal Equations' %Display label. pretty(G) %Pretty print G(s) Computer response: ans = Mesh Equations A2 = [ [ [ 2+2*s, -1-2*s, -1, V, -1] 0, -2-3*s] 0, 3+3*s+5/s] 2+2*s, -1-2*s, -1, -1-2*s, -1] 7+5*s, -2-3*s] -2-3*s, 3+3*s+5/s] A= [ [ [ ans = G(s) via Mesh Equations 2 3 15 s + 12 s + 5 + 6 s 4 -------------------------2 3 120 s + 78 s + 65 + 24 s ans = Nodal Equations A2 = [ (6*s^2+12*s+5)/(2+7*s+6*s^2), [ -1/(2+3*s), A= V] 1/5*s*V] 28 Chapter 2: Modeling in the Frequency Domain [ [ (6*s^2+12*s+5)/(2+7*s+6*s^2), -1/(2+3*s)] -1/(2+3*s), (3/5*s^2+23/20*s+3/2)/(2+3*s)] ans = G(s) via Nodal Equations 2 3 15 s + 12 s + 5 + 6 s 4 -------------------------3 2 24 s + 78 s + 120 s + 65 21. a. Z1 (s ) = 5x10 5 + 106 s 10 6 Z2 ( s) = 10 + s 5 Therefore, - 1 s + 10 Z2 s =Z1 s 5 s+2 b. 1 = 100000 s + 10 −6 s 1×10 s 1 Z2 s = 100000 + = 100000 s + 20 −6 1 s + 10 1×10 s + 100000 Therefore, Z1 s = 100000 + G s =− Z2 = − s + 20 s 2 Z1 s + 10 22. a. Z1 s = 200000 + 1 1×10 −6 s 1 Z2 s = 100000 + 1×10 −6 s Therefore, s + 20 Z +Z 3. G s = 1 2=3 2 s+5 Z1 Solutions to Problems 29 b. 5x1011 s Z1 (s) = 2 x105 + 10 6 5 x105 + s 11 10 s Z2 ( s) = 5x10 5 + 106 105 + s Therefore, Z1 (s) + Z 2 (s) 7 (s + 3.18)(s + 11.68) = Z1 ( s) 2 ( s + 7)(s + 10) 23. Writing the equations of motion, where x2(t) is the displacement of the right member of springr, (s2+s+1)X1(s) -X2(s) = 0 -X1(s) +X2(s) = F(s) Adding the equations, (s2+s)X1(s) = F(s) From which, X 1 (s) 1 = . F(s) s(s + 1) 24. Writing the equations of motion, (s 2 + s + 1) X1 (s) − (s + 1) X2 (s) = F (s) − ( s + 1) X1 (s) + (s2 + s + 1) X 2 ( s) = 0 Solving for X2(s), From which, ⎡2 ⎤ ⎢(s + s + 1) F (s) ⎥ ⎢ ⎥ ⎣ − (s + 1) 0⎦ ( s + 1)F (s) X2 (s) = ⎡ 2 =22 (s + s + 1) − (s + 1) ⎤ s ( s + 2 s + 2) ⎥ ⎢ ⎥ ⎢ ⎣ − (s + 1) ( s2 + s + 1)⎦ (s + 1) X2 (s) =22 . F (s) s (s + 2s + 2) 25. Let X1(s) be the displacement of the left member of the spring and X3(s) be the displacement of the mass. Writing the equations of motion 30 Chapter 2: Modeling in the Frequency Domain 2 x1 ( s) − 2 x2 (s) = F (s) −2 X1 (s) + (5s + 2) X 2 ( s) − 5sX3 ( s) = 0 −5sX2 (s) + (10 s2 + 7s)X 3 ( s) = 0 Solving for X2(s), X (s) = 2 Thus, ⎪5s2 +10 ⎪ ⎪ -10 F(s)⎪ ⎪ 0⎪ ⎪ ⎪ 1 s + 10 5 ⎪ ⎪5s2 +10 ⎪ ⎪ -10 -10 = 10F(s) s(s2 +50s+2) X2 (s) 1 (10 s + 7) = F (s) 10 s(5s + 1) 26. 2 s + 3 s + 2 X1 s − s + 1 X2 s = 0 2 − s + 1 X1 s + s + 2 s + 1 X2 s = F s 0 s2 + 2 s + 1 s2 + 3 s + 2 − s+ 1 − s+ 1 Solving for X1(s); X1 = − s+ 1 F s2 + 2 s + 1 = Fs 1 . Thus, X1 = 3 2 Fs s +4 s +4 s+1 s +4 s +4 s+1 3 2 27. Writing the equations of motion, (s 2 + s + 1) X1 (s) − sX2 (s) = 0 − sX1 ( s) + (s2 + 2s + 1) X2 (s) − X3 (s) = F (s) − X 2 ( s) + (s2 + s + 1)X 3 ( s) = 0 Solving for X3(s), ( s2 + s + 1) −s −s X3 (s ) = (s + 2s + 1) F (s) −1 0 2 (s + s + 1) −s 0 0 2 −s 2 0 0 ( s + 2 s + 1) −1 2 −1 (s + s + 1) = F (s ) s(s + 3s2 + 3s + 3) 3 Solutions to Problems 31 From which, X3 (s ) 1 = . 3 2 F (s) s( s + 3s + 3s + 3) 28. a. (s 2 + 2s + 1)X 1 (s) − 2sX 2 (s) − X 3 (s) = F(s) 2 − 2sX1 (s) + (s + 4s)X 2 (s) − sX 3 (s) = 0 − X1 (s) − sX 2 (s) + (s + 1)X 3 (s) = 0 Solving for X2(s), X 2 (s) = (s2 + 2s + 1) F(s) − 1 − 2s 0 −s −1 0 s+1 ∆ − F(s) = − 2s − s −1 s + 1 ∆ or, X 2 (s) 2s + 3 =3 F(s) s(s + 6s2 + 9s + 3) b. (4 s2 + s+ 4) X1 (s) − (s+ 1) X2 (s) − 3X 3 (s) = 0 2 − (s + 1) X1 (s) + (2 s + 5s+ 1) X 2 (s) − 4sX 3 (s) = F(s) − 3X 1 (s) − 4sX 2 (s) + (4 s+ 3) X 3 (s) = 0 Solving for X3(s), X 3 (s ) = (4 s2 + s+ 4) − (s+ 1) 0 2 − (s+ 1) (2 s + 5s + 1) F(s) −3 − 4s 0 ∆ − F(s) = (4s2 + s+ 4) − (s+ 1) −3 −4s ∆ or X3(s) 16 s3 + 4s2 + 19s + 3 = F(s) 32 s5 + 48s 4 + 114 s3 + 18 s 2 29. Writing the equations of motion, (s 2 + 2s + 2) X1 (s) − X 2 ( s) − sX3 (s ) = 0 − X1 ( s) + (s2 + s + 1) X 2 ( s ) − sX3 (s) = F( s ) − sX1 ( s) − sX2 (s) + ( s 2 + 2 s + 1) = 0 32 Chapter 2: Modeling in the Frequency Domain 30. a. Writing the equations of motion, (s 2 + 9s + 8)θ1 (s) − (2 s + 8)θ 2 ( s) = 0 − (2 s + 8)θ1 ( s ) + (s 2 + 2s + 11)θ 2 (s) = T (s) b. Defining θ 1 (s) = rotation of J1 θ 2 (s) = rotation between K1 and D1 θ 3 ( s) = rotation of J 3 θ 4 (s) = rotation of right - hand side of K2 the equations of motion are ( J1s 2 + K1 )θ1 (s) − K1θ 2 (s) = T (s) − K1θ1 ( s) + ( D1 s + K1 )θ2 ( s) − D1sθ 3 (s) = 0 − D1 sθ2 (s) + ( J2 s2 + D1s + K2 )θ 3 ( s) − K2θ 4 ( s) = 0 − K2θ 3 ( s) + ( D2 s + ( K2 + K3 ))θ 4 (s) = 0 31. Writing the equations of motion, (s 2 + 2s + 1)θ1 (s) − (s + 1)θ 2 (s) = T (s) − ( s + 1)θ 1 (s) + (2s + 1)θ 2 ( s ) = 0 Solving for θ 2 (s) 2 ( s + 2 s + 1) T (s) θ 2 (s) = − (s + 1) 0 T (s) = (s + 2s + 1) − (s + 1) 2s(s + 1) 2 −( s + 1) (2 s + 1) Hence, θ 2 (s ) T( s) 32. Reflecting impedances to θ3, = 1 2s(s + 1) Solutions to Problems 33 (Jeqs2+Deqs)θ3(s) = T(s) ( N4 N2 ) N3 N1 Thus, N 4N 2 θ 3 ( s) N3 N1 = T (s) Jeq s2 + Deq s where ⎛N ⎞2 ⎛N N ⎞2 4⎟ 42 ⎟ + J1 ⎜ N N ⎟ , and ⎜ ⎟ ⎝ N3 ⎠ ⎝ 3 1⎠ Jeq = J4+J5+(J2+J3) ⎜ ⎜ D eq = (D 4 + D 5 ) + (D 2 + D 3 )( 33. N4 2 NN ) + D1 ( 4 2 ) 2 N3 N 3 N1 Reflecting all impedances to θ2(s), 2 2 2 2 2 N N {[J2+J1(N2 ) +J3 (N3 ) ]s2 + [f2+f1(N2 ) +f3(N43 ) ]s + [K(N3 ) ]}θ2(s) = T(s)N2 N4 N1 N1 N4 1 Substituting values, 2 2 2 {[1+2(3)2+16(1 ) ]s2 + [2+1(3)2+32(1 ) ]s + 64(1 ) }θ2(s) = T(s)(3) 4 4 4 Thus, θ2(s) 3 T(s) = 20s2+13s+4 34. Reflecting impedances to θ2, ⎡ ⎡ ⎞2⎤ ⎞2⎤ ⎡ ⎛ ⎞2 ⎛ ⎛ ⎛ ⎞2⎤ ⎛ ⎞ ⎢200 + 3⎜ 50 ⎟ + 200⎜ 5 x 50 ⎟ ⎥ s 2 + ⎢1000⎜ 5 x 50 ⎟ ⎥ s + ⎢250 + 3⎜ 50 ⎟ ⎥ = ⎜ 50 ⎟ T (s) ⎝ 5⎠ ⎝ 25 5 ⎠ ⎦ ⎝ 25 5 ⎠ ⎦ ⎣ ⎝ 5 ⎠ ⎦ ⎝ 5⎠ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎣ ⎣ Thus, θ 2 (s) T( s) = 10 1300s + 4000s + 550 2 35. Reflecting impedances and applied torque to respective sides of the spring yields the following equivalent circuit: 34 Chapter 2: Modeling in the Frequency Domain Writing the equations of motion, θ2(s) - θ3(s) = 4T(s) -θ2(s) + (s+1)θ3(s) = 0 Solving for θ3(s), 1 θ (s ) = 3 -1 4 T (s ) 0 1 -1 -1 = 4 T (s ) s (s + 1 ) θ3(s) 4 θ4(s) 4/5 1 Hence, T(s) = s . But, θ4(s) = 5 θ3(s). Thus, T(s) = s . 36. Reflecting impedances and applied torque to respective sides of the viscous damper yields the following equivalent circuit: Writing the equations of motion, (s 2 + s)θ 2 (s) − sθ 3 ( s) = 10T (s) − sθ 2 (s) + ( s + 1)θ 3 (s) − θ 4 (s) = 0 −θ3 (s) + ( s + 1)θ 4 (s) = 0 Solving for θ 4 (s) , Solutions to Problems 35 −s 10T (s ) s(s + 1) −s (s + 1) 0 θ 4 (s) = 0 −1 s (s + 1) −s −s (s + 1) −1 0 0 0 −1 = s10T (s) s( s + 1) −s −s ( s + 1) (s + 1) 0 −1 0 −1 ( s + 1) Thus, θ 4 (s) T( s) = 10 s(s + 1)2 = 50 s(s + 1)2 But, θ L (s) = 5θ 4 (s) . Hence, θ 4 (s) T( s) 37. Reflect all impedances on the right to the viscous damper and reflect all impedances and torques on the left to the spring and obtain the following equivalent circuit: Writing the equations of motion, (J1eqs2+K)θ2(s) -Kθ3(s) = Teq(s) -Kθ2(s)+(Ds+K)θ3(s) -Dsθ4(s) = 0 -Dsθ3(s) +[J2eqs2 +(D+Deq)s]θ4(s) = 0 N2 where: J1eq = J2+(Ja+J1) N 1 () N1 N2 . 2 N3 ; J2eq = J3+(JL+J4) N 4 () 2 N3 ; Deq = DL N 4 () 2 ; θ2(s) = θ1(s) 36 Chapter 2: Modeling in the Frequency Domain 38. Reflect impedances to the left of J5 to J5 and obtain the following equivalent circuit: Writing the equations of motion, [Jeqs2+(Deq+D)s+(K2+Keq)]θ5(s) -[Ds+K2]θ6(s) = 0 -[K2+Ds]θ5(s) + [J6s2+2Ds+K2]θ6(s) = T(s) From the first equation, θ6(s) Jeqs2+(Deq+D)s+ (K2+Keq) θ5(s) N1N3 = . But, = . Therefore, Ds+K2 θ5(s) θ1(s) N2N4 θ6(s) N1N3 ⎛ Jeqs2+(Deq+D)s+ (K2+Keq) ⎞ = ⎜ ⎟, Ds+K2 ⎠ θ1(s) N2N4 ⎝ N4N2 where Jeq = J1 N N 31 [( [(N4N2 ) N3N1 Deq = D 2 ) 2 N4 + (J2+J3) N 3 N4 +N 3 () 39. Draw the freebody diagrams, () 2 +1 . 2 N4 + (J4+J5) , Keq = K1 N 3 () 2 , and Solutions to Problems 37 Write the equations of motion from the translational and rotational freebody diagrams, -fvrsθ(s) = F(s) (Ms2+2fv s+K2)X(s) -fvrsX(s) +(Js2+fvr2s)θ(s) = 0 Solve for θ(s), 2 θ (s ) = Ms +2fvs+K 2 -fvrs 2 F(s ) 0 Ms +2fvs+ K 2 -fvrs -fvrs fvrF (s ) = 3 22 22 JMs +(2J fv+Mfvr )s +(JK +fvr )s + K fvr 2 2 J s +fvr2s θ(s) From which, F(s) = 2 fvr 3+(2Jf +Mf r2)s2+(JK +f 2r2)s+K f r2 . JMs v v 2v 2v 40. Draw a freebody diagram of the translational system and the rotating member connected to the translational system. From the freebody diagram of the mass, F(s) = (s2+s+1)X(s). Summing torques on the rotating member, (Jeqs2 +Deqs)θ(s) + F(s)2 = Teq(s). Substituting F(s) above, (Jeqs2 +Deqs)θ(s) + (2s2+2s+2)X(s) = X(s) Teq(s). However, θ(s) = 2 . Substituting and simplifying, Jeq Deq +2 s2 + 2 +2 s+2 X(s) Teq = 2 [( )( )] 2 38 Chapter 2: Modeling in the Frequency Domain But, Jeq = 1+1(4)2 = 17, Deq = 1(2)2 +1 = 5, and Teq(s) = 4T(s). Therefore, [ 221 s2 +9 s+2]X(s) = 2 8 21 X(s) 4T(s). Finally, T(s) = 9 4. s2 + 21 s + 21 41. Writing the equations of motion, (J1s2+K1)θ1(s) - K1θ2(s) = T(s) 2+D s+K )θ (s) +F(s)r -D sθ (s) = 0 -K1θ1(s) + (J2s 3 12 33 -D3sθ2(s) + (J2s2+D3s)θ3(s) = 0 where F(s) is the opposing force on J2 due to the translational member and r is the radius of J2. But, for the translational member, F(s) = (Ms2+fvs+K2)X(s) = (Ms2+fvs+K2)rθ(s) Substituting F(s) back into the second equation of motion, (J1s2+K1)θ1(s) - K1θ2(s) = T(s) -K1θ1(s) + [(J2 + Mr2)s2+(D3 + fvr2)s+(K1 + K2r2)]θ2(s) -D3sθ3(s) = 0 -D3sθ2(s) + (J2s2+D3s)θ3(s) = 0 Notice that the translational components were reflected as equivalent rotational components by the square of the radius. Solving for θ2(s), θ 2 (s ) = K1 ( J3 s 2 + D3 s)T( s ) , where ∆ is the ∆ determinant formed from the coefficients of the three equations of motion. Hence, θ 2 (s) T( s) = Since X ( s) rK1 ( J 3 s2 + D3 s) = ∆ T (s) X (s) = rθ 2 (s), 42. K1 ( J3 s2 + D3 s) ∆ Kt Tstall 100 Ea 50 1 = = = 2 ; Kb = = = Ra Ea 50 ω no− load 150 3 Also, Jm = 2+18 Thus, θ m (s ) Ea (s) = (1) 3 2 = 4; Dm = 2+36 2 ( 1 ) = 6. 3 2/4 1/ 2 = 1 5 2 s (s + (6 + )) s (s + ) 4 3 3 Solutions to Problems 39 Since θL(s) = 1 θm(s), 3 θ L (s ) Ea (s) = 1 6 5. s (s + ) 3 43. The parameters are: 5 E Kt Ts 5 = = 1; Kb = a = = = ω 600 2π 1 Ra Ea 5 π 60 2 1⎞ 2 1⎞ 2 1 ⎛1 ⎞ ⎛ ⎛ ; Jm = 16⎝ ⎠ + 4⎝ ⎠ + 1 = 3; Dm = 32 ⎝ ⎠ = 2 4 2 4 4 Thus, 1 θ m (s ) 3 = = Ea (s) s(s + 1 (2 + (1)( 1 ))) s(s + 0.75) 4 3 1 3 Since θ2(s) = 1 θm(s), 4 1 θ 2 (s ) 12 . = Ea (s) s(s + 0.75) 44. The following torque-speed curve can be drawn from the data given: T 100 50 500 Therefore, 1000 v Kt Tstall 100 Ea 10 1 12 = = = 10 ; Kb = = = . Also, Jm = 5+100( ) = 9; Ra Ea 10 ω no− load 1000 100 5 Dm = 1. Thus, 40 Chapter 2: Modeling in the Frequency Domain 10 10 10 1 θ L (s ) 9 9 45 = = . Since θL(s) = θm(s), = = 1 Ea (s) s (s + (1 + 0.1)) s(s + 0.122) 5 Ea (s) s(s + 0.122) 9 0.222 . s (s + 0.122) θ m (s ) 45. From Eqs. (2.45) and (2.46), RaIa(s) + Kbsθ(s) = Ea(s) (1) Also, Tm(s) = KtIa(s) = (Jms2+Dms)θ(s). Solving for θ(s) and substituting into Eq. (1), and simplifying yields Dm ) Jm Ia (s ) 1 = Ea (s) Ra s + Ra Dm + K b Kt Ra J m (s + (2) Using Tm(s) = KtIa(s) in Eq. (2), Dm ) Jm Tm ( s) Kt = Ea (s) Ra s + Ra Dm + K b Kt Ra J m (s + 46. For the rotating load, assuming all inertia and damping has been reflected to the load, (JeqLs2+DeqLs)θL(s) + F(s)r = Teq(s), where F(s) is the force from the translational system, r=2 is the radius of the rotational member, JeqL is the equivalent inertia at the load of the rotational load and the armature, and DeqL is the equivalent damping at the load of the rotational load and the armature. Since JeqL = 1(2)2 +1 = 5, and DeqL = 1(2)2 +1 = 5, the equation of motion becomes, (5s2+5s)θL(s) + F(s)r = Teq(s). For the translational system, (s2+s)X(s) = F(s). Since X(s) = 2θL(s), F(s) = (s2+s)2θL(s). Substituting F(s) into the rotational equation, (9s2+9s) θL(s) = Teq(s). Thus, the equivalent inertia at the load is 9, and the equivalent damping at the load is 9. Reflecting these back Kt 9 9 to the armature, yields an equivalent inertia of 4 and an equivalent damping of 4 . Finally, R = 1; a Solutions to Problems 41 4 4 2 θL(s) 9 9 9 1 = 49 13 . Since θL(s) = 2 θm(s), Ea(s) = 13 . But s(s+9(4+1)) s(s+ 9 ) s(s+ 9 ) 4 X(s) 9 X(s) = rθL(s) = 2θL(s). therefore, = . E (s) s(s+13) a 9 θm(s) Kb = 1. Hence, E (s) = a 47. The equations of motion in terms of velocity are: K1 K2 K + ]V1 ( s) − 2 V2 ( s) − fv 3V3 (s) = 0 s s s K K − 2 V1 (s) + [ M2 s + ( fv 2 + f v 4 ) + 2 ]V2 (s) − f v4 V3 (s) = F (s) s s − f v3 V1 (s) − f v4 V2 (s) + [ M3 s + fV 3 + fv 4 ]V3 ( S) = 0 [ M1s + ( fv1 + fv 3 ) + For the series analogy, treating the equations of motion as mesh equations yields In the circuit, resistors are in ohms, capacitors are in farads, and inductors are in henries. For the parallel analogy, treating the equations of motion as nodal equations yields In the circuit, resistors are in ohms, capacitors are in farads, and inductors are in henries. 42 Chapter 2: Modeling in the Frequency Domain 48. Writing the equations of motion in terms of angular velocity, Ω(s) yields K1 K )Ω1 (s) − (D1 + 1 )Ω2 (s) = T ( s ) s s K1 (K + K2 ) − ( D1 + )Ω 1 (s) + ( J 2 s + D1 + 1 )Ω 2 (s ) = 0 s s K K − 2 Ω 2 ( s) − D2Ω 3 ( s) + ( D2 + 2 )Ω 4 (s) = 0 s s K3 ( J3 s + D2 + )Ω 3 ( s) − D2 Ω 4 (s) = 0 s ( J1s + D1 + For the series analogy, treating the equations of motion as mesh equations yields In the circuit, resistors are in ohms, capacitors are in farads, and inductors are in henries. For the parallel analogy, treating the equations of motion as nodal equations yields In the circuit, resistors are in ohms, capacitors are in farads, and inductors are in henries. 49. An input r1 yields c1 = 5r1+7. An input r2 yields c2 = 5r2 +7. An input r1 +r2 yields, 5(r1+r2)+7 = 5r1+7+5r2 = c1+c2-7. Therefore, not additive. What about homogeneity? An input of Kr1 yields c = 5Kr1+7 ≠ Kc1. Therefore, not homogeneous. The system is not linear. 50. a. Let x = δx+0. Therefore, Solutions to Problems 43 .. . δ x + 3 δ x + 2 δ x =s in (0 + δ x) d sinx δ x = 0 + co sx ⎮ δ x = δ x dx ⎮ Bu t , s i n (0 + δ x) = s i n 0 + x =0 .. x =0 .. . . T h erefo re, δ x + 3 δ x + 2 δ x = δ x . C o ll ectin g t erm s , δ x + 3 δ x + δ x = 0 . b. Let x = δx+π. Therefore, .. . δx+3δx+2δx =sin (π +δx) But, sin (π + δx) = sin π + .. d sinx ⎮ δx = 0+cosx ⎮ δx = − δx dx x=π x=π . .. . Therefore, δx+3δx+2δx = -δx. Collecting terms, δx+3δx+3δx = 0 . 51. If x = 0 + δx, ... .. . -(δ x) δx + 10δx + 31δx + 30δx = e -(δ x) = e-0 + But e ... -x de δx = 1 - e-x ⎮ δx = 1 - δx dx ⎮ .. x=0 . x=0 ... .. . Therefore, δx + 10δx + 31δx + 30δx =1 - δx, or, δx + 10δx + 31δx + 31δx =1. 52. The given curve can be described as follows: f(x) = -4 ; -∞<x<-2; f(x) = 2x; -2<x<2; f(x) = 4; 2<x<+∞ Thus, .. . .. . .. . a. x + 15x +50x = -4 .. . b. x + 15x + 50x = 2x, or x + 15x +48x = 0 c. x + 15x + 50x = 4 53. The relationship between the nonlinear spring’s displacement, xs(t) and its force, fs(t) is − fs ( t) xs ( t ) = 1 − e Solving for the force, f s (t ) = − ln(1 − x s (t )) (1) Writing the differential equation for the system by summing forces, d 2 x (t ) dx(t ) + − ln(1 − x (t )) = f (t ) dt 2 dt (2) 44 Chapter 2: Modeling in the Frequency Domain Letting x(t) = x0 + δx and f(t) = 1 + δf, linearize ln(1 – x(t)). ln(1 − x ) − ln(1 − x0 ) = d ln(1 − x ) δx dx x =x 0 Solving for ln(1 – x), ln(1 − x ) = ln(1 − x0 ) − 1 1− x x = x0 δx = ln(1 − x 0 ) − 1 δx 1 − x0 (3) When f = 1, δx = 0. Thus from Eq. (1), 1 = -ln(1 – x0 ). Solving for x0, 1 – x0 = e-1 , or x0 = 0.6321. Substituting x0 = 0.6321 into Eq. (3), ln (1 - x ) = ln(1 – 0.6321) - 1 δx = -1 - 2.718δx 1 - 0.6321 Placing this value into Eq. (2) along with x(t) = x0 + δx and f(t) = 1 + δf, yields the linearized differential equation, d 2 δx dδx + + 1 + 2.718δx = 1 + δf dt 2 dt or d 2 δ x dδ x + + 2.718δx = δf dt 2 dt Taking the Laplace transform and rearranging yields the transfer function, δx( s) 1 =2 δf (s ) s + s + 2.718 54. First assume there are n plates without the top plate positioned at a displacement of y2(t) where y2(t) = 0 is the position of the unstretched spring. Assume the system consists of mass M, where M is the mass of the dispensing system and the n plates, viscous friction, fv, where the viscous friction originates where the piston meets the sides of the cylinder, and of course the spring with spring constant, K. Now, draw the freebody diagram shown in Figure (b) where Wn is the total weight of the n dishes and the piston. If we now consider the current position, y2(0), Solutions to Problems 45 Restaurant Plate Dispenser the equilibrium point and define a new displacement, y1(t), which is measured from equilibrium, we can write the force in the spring as Ky2(t) = Ky2(0) + Ky1(t). Changing variables from y2(t) to y1(t), we sum forces and get, where d2y2 d2y1 2 = dt2 dt dy1 d2y1 (1) M 2 + fv dt + Ky1 + Ky2(0) + Wn = 0 dt dy2 dy1 and dt = dt . But, Ky2(0) = -Wn , since it is the component of the spring force that balances the weight at equilibrium when y1 = 0. Thus, the differential equation becomes, dy1 d2y1 M 2 + fv dt + Ky1 = 0 dt (2) WD When the top plate is added, the spring is further compressed by an amount, K , where WD is the weight of the single dish, and K is the spring constant. We can think of this displacement as an initial WD dy1 condition. Thus, y1(0-) = - K and dt (0-) =0, and y1(t) = 0 is the equilibrium position of the spring with n plates rather than the unstretched position. Taking the Laplace transform of equation (2), using the initial conditions, 46 Chapter 2: Modeling in the Frequency Domain WD WD M(s2Y1(s) + s K ) + fv(sY1(s) + K ) + KY1(s) = 0 or (Ms2 + fvs + K)Y1(s) = - WD (Ms + fv ) K (3) (4) Now define a new position reference, Y(s), which is zero when the spring is compressed with the initial condition, WD Y(s) = Y1(s) + Ks (5) WD Y1(s) = Y(s) - Ks (6) or Substituting Y1(s) in Equation (4), we obtain, WD (Ms2 + fvs + K)Y(s) = s = F(s) (7) a differential equation that has an input and zero initial conditions. The schematic is shown in Figure Y(s) (c). Forming the transfer function, F(s) , we show the final result in Figure (d), where for the WD removal of the top plate, F(s) is always a step, F(s) = s . 55. Writing the equations of motion, (17.2s2 + 160s + 7000)Yf(s) – (130s + 7000)Yh(s) - (130s+7000)Yf(s) + (9.1ss + 130s + 89300)Yh(s) - 0Yf(s) - 82300Yh(s) – 0Ycat(s) = Fup(s) - 82300Ycat(s) = 0 + 1.6173 x106 Ycat(s) = 0 These equations are in the form AY=F, where det(A) = 2.5314 x 108 (s2 + 15.47 s + 9283) (s2 + 8.119 s + 376.3) Using Cramer’s rule: Ycat ( s) 0.04227(s + 53.85) =2 2 Fup (s) (s + 15.47s + 9283)(s + 8.119s + 376.3) 0.8306(s + 53.85) Yh ( s) =2 Fup ( s) (s + 15.47s + 9283)(s2 + 8.119s + 376.3) 0.7883( s + 53.85) Yh ( s) − Ycat (s) =2 (s + 15.47s + 9283)(s2 + 8.119 s + 376.3) Fup ( s) THREE Modeling in the Time Domain SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: State-Space Representation . Ea(s) 150 For the power amplifier, V (s) = s+150 . Taking the inverse Laplace transform, ea +150ea = p 150vp. Thus, the state equation is • e a = − 150e a + 150vp . For the motor and load, define the state variables as x1 = θm and x2 = θ m. Therefore, . x 1 = x2 (1) Using the transfer function of the motor, cross multiplying, taking the inverse Laplace transform, and using the definitions for the state variables, . K tK a Kt 1 x 2 = - J (Dm+ R ) x2 + R J m a am ea (2) Using the gear ratio, the output equation is y = 0.2x1 (3) Kt 1 1 1 Also, Jm = Ja+5(5 )2 = 0.05+0.2 = 0.25, Dm = Da+3(5 )2 = 0.01+0.12 = 0.13, R J = (5)(0.25) am K tK a 1 = 0.8, and J (Dm+ R ) = 1.32. Using Eqs. (1), (2), and (3) along with the previous values, the m a state and output equations are, . x= 0 0 0 1 e ; y = 0. 2 x+ 0. 8 a -1. 32 0x 48 Chapter 3: Modeling in the Time Domain Aquifer: State-Space Representation dh1 C1 dt = qi1-qo1+q2-q1+q21 = qi1-0+G2(h2-h1)-G1h1+G21(H1-h1) = -(G2+G1+G21)h1+G2h2+qi1+G21H1 dh2 C2 dt = qi2-q02+q3-q2+q32 = qi2-qo2+G3(h3-h2)-G2(h2-h1)+0 = G2h1-[G2+G3]h2+G3h3+qi2-qo2 dh3 C3 dt = qi3-qo3+q4-q3+q43 = qi3-qo3+0-G3(h3-h2)+0 = G3h2-G3h3+qi3-qo3 Dividing each equation by Ci and defining the state vector as x = [h1 h2 h3]T ⎡ − (G1 + G2 + G3 ) ⎢ C1 . G2 ⎢ x= C2 ⎢ ⎢ 0 ⎣ G2 C1 − (G2 + G3 ) C2 G3 C3 ⎡ qi1 + G21 H1 ⎤ ⎤ ⎢ ⎥ ⎥ C1 G3 ⎥ q − qo 2 ⎥ x + ⎢ i2 u(t ) C2 ⎥ C2 ⎢ ⎥ − G3 ⎥ ⎢ qi3 − qo 3 ⎥ C3 ⎦ C3 ⎣ ⎦ 0 ⎡ 1 0 0⎤ y = ⎢ 0 1 0⎥ x ⎢ 0 0 1⎥ ⎣ ⎦ where u(t) = unit step function. ANSWERS TO REVIEW QUESTIONS 1. (1) Can model systems other than linear, constant coefficients; (2) Used for digital simulation 2. Yields qualitative insight 3. That smallest set of variables that completely describe the system 4. The value of the state variables 5. The vector whose components are the state variables 6. The n-dimensional space whose bases are the state variables 7. State equations, an output equation, and an initial state vector (initial conditions) 8. Eight 9. Forms linear combinations of the state variables and the input to form the desired output 10. No variable in the set can be written as a linear sum of the other variables in the set. 11. (1) They must be linearly independent; (2) The number of state variables must agree with the order of the differential equation describing the system; (3) The degree of difficulty in obtaining the state equations for a given set of state variables. 12. The variables that are being differentiated in each of the linearly independent energy storage elements Solutions to Problems 49 13. Yes, depending upon the choice of circuit variables and technique used to write the system equations. For example, a three -loop problem with three energy storage elements could yield three simultaneous second-order differential equations which would then be described by six, first-order differential equations. This exact situation arose when we wrote the differential equations for mechanical systems and then proceeded to find the state equations. 14. The state variables are successive derivatives. SOLUTIONS TO PROBLEMS 1. Add the branch currents and node voltages to the network. Write the differential equation for each energy storage element. di2 = v1 dt di4 = v2 dt dv o = i5 dt ⎡i ⎤ ⎢ 2⎥ Therefore, the state vector is x = ⎢ i4 ⎥ ⎢⎥ ⎣vo ⎦ Now obtain v1, v2, and i5 in terms of the state variables. First find i1 in terms of the state variables. − v i + i1 + i3 + i5 + vo = 0 But i3 = i1 − i2 and i5 = i3 − i4 . Thus, − v i + i1 + (i1 − i2 ) + (i3 − i4 ) + vo = 0 Making the substitution for i3 yields − v i + i1 + (i1 − i2 ) + ((i1 − i2 ) − i4 ) + vo = 0 Solving for i1 50 Chapter 3: Modeling in the Time Domain i1 = 1 1 1 2 i2 + i4 − vo + vi 3 3 3 3 Thus, 2 1 1 2 v1 = vi − i1 = − i2 − i4 + vo + vi 3 3 3 3 Also, 1 1 1 1 i3 = i1 − i2 = − i2 + i4 − vo + vi 3 3 3 3 and 1 2 1 1 i 5 = i3 − i4 = − i2 − i4 − vo + vi 3 3 3 3 Finally, 1 2 2 1 v 2 = i5 + vo = − i2 − i 4 + vo + vi 3 3 3 3 Using v1, v2, and i5, the state equation is ⎡2 1 − ⎢− 3 ⎢3 • ⎢− 1 − 2 x=⎢ 3 3 ⎢1 2 ⎢− − ⎣3 3 ⎡2 ⎤ 1⎤ ⎥ ⎢⎥ 3⎥ ⎢ 3⎥ 2⎥ ⎢1 ⎥ ⎥ x + ⎢ 3 ⎥ vi 3⎥ ⎢1 ⎥ 1 ⎢⎥ −⎥ ⎣ 3⎦ 3⎦ y = [0 0 1]x 2. Add branch currents and node voltages to the schematic and obtain, Write the differential equation for each energy storage element. Solutions to Problems 51 dv1 = i2 dt di3 = vL dt ⎡v1 ⎤ ⎢⎥ ⎣ i3 ⎦ Therefore the state vector is x = ⎢ ⎥ Now obtain v L and i2 , in terms of the state variables, v L = v1 − v2 = v1 − iR = v1 − (i3 + 4v1 ) = − 3v1 − i3 i2 = i1 − i3 = ( vi − v1 ) − i3 = − v1 − i3 + vi Also, the output is y = iR = 4v1 + i3 Hence, • ⎡ −1 − 1⎤ ⎡ 1⎤ x=⎢ ⎥ x + ⎢ ⎥ vi ⎣− 3 − 1⎦ ⎣ 0⎦ y = [4 1]x 3. Let C1 be the grounded capacitor and C 2 be the other. Now, writing the equations for the energy storage components yields, di L = v i − vC1 dt dv C1 = i1 − i2 (1) dt dv C2 = i2 − i3 dt ⎡⎤ ⎢ iL ⎥ Thus the state vector is x = ⎢ vC1 ⎥ . Now, find the three loop currents in terms of the state variables ⎢⎥ ⎢vC2 ⎦ ⎣⎥ and the input. Writing KVL around Loop 2 yields vC 1 = vC 2 + i2 .Or, i2 = vC 1 − vC 2 Writing KVL around the outer loop yields i3 + i2 = vi Or, i3 = vi − i2 = vi − vC1 + vC 2 52 Chapter 3: Modeling in the Time Domain Also, i1 − i3 = iL . Hence, i1 = iL + i3 = i L + vi − v C1 + vC 2 Substituting the loop currents in equations (1) yields the results in vector-matrix form, ⎡ di ⎤ ⎢ L⎥ ⎢ dt ⎥ ⎡ 0 −1 0 ⎤ ⎡ i L ⎤ ⎡ 1 ⎤ ⎥⎢ ⎥ ⎢ ⎥ ⎢ dvC1 ⎥ ⎢ = ⎢ 1 − 2 2 ⎥ ⎢ vC1 ⎥ + ⎢ 1 ⎥ vi ⎢ ⎥ ⎢ dt ⎥ ⎢ 0 2 − 2⎥ ⎢v ⎥ ⎢ − 1⎥ ⎦ ⎢ C2 ⎥ ⎣ ⎦ ⎣⎦ ⎢ dv C2 ⎥ ⎣ ⎢ dt ⎥ ⎣ ⎦ Since vo = i2 = vC1 − vC 2 , the output equation is ⎡⎤ ⎢ iL ⎥ y = [0 1 1]⎢ v C1 ⎥ ⎢⎥ ⎢ vC2 ⎦ ⎣⎥ 4. Equations of motion in Laplace: (s 2 + 3s + 1) x1 ( s) − (s + 1) x2 ( s) − sx3 (s) = 0 −( s + 1) x1 ( s) + (s 2 + 2s + 1) x2 ( s) − sx3 (s) = F (s) − sx1 ( s) − sx 2 ( s) + (s 2 + 3s)x 3 ( s) = 0 Equations of motion in the time domain: d 2 x1 dx dx dx + 3 1 + x1 − 2 − x 2 − 3 = 0 2 dt dt dt dt d 2 x2 dx 2 dx dx + x2 − 1 − x1 − 3 = f (t ) 2 +2 dt dt dt dt 2 d x3 dx 3 dx dx + x1 − 2 − 1 = 0 2 +3 dt dt dt dt Define state variables: z1 = x1 or dx z2 = 1 or dt z3 = x2 or x1 = z1 dx1 = z2 dt x2 = z3 dx2 or dt z5 = x3 or dx z6 = 3 or dt dx 2 = z4 dt x 3 = z5 dx3 = z5 dt z4 = (1) (2) (3) (4) (5) (6) Solutions to Problems 53 Substituting Eq. (1) in (2), (3) in (4), and (5) in (6), we obtain, respectively: dz1 = z2 dt dz3 = z4 dt dz5 = z6 dt (7) (8) (9) Substituting Eqs. (1) through (6) into the equations of motion in the time domain and solving for the derivatives of the state variables and using Eqs. (7) through (9) yields the state equations: dz1 dt dz 2 dt dz 3 dt dz 4 dt dz 5 dt dz 6 dt = z2 = − z1 − 3z2 + z3 + z 4 + z6 = z4 = z1 + z 2 − z3 − 2z4 + z6 + f (t ) = z6 = z2 + z4 − 3z6 The output is x3 = z5. In vector-matrix form: 0 ⎡0 1 0 ⎢ −1 − 3 1 1 ⎢0 0 0 . 1 z=⎢ 1 1 −1 −2 ⎢ 000 0 ⎢ 1 ⎣0 1 0 y = [0 0 5. Writing the equations of motion, 0 0 0 0 0 0 0 0 1 0]z 0 ⎤ ⎡0 ⎤ 1 ⎥ ⎢0 ⎥ 0 ⎥ ⎢0 ⎥ z+ f ( t) 1 ⎥ ⎢1 ⎥ ⎥ ⎢⎥ 1 0 ⎥ ⎢⎥ − 3⎦ ⎣0 ⎦ 54 Chapter 3: Modeling in the Time Domain (s 2 + 2s + 1) X1 (s) − sX2 ( s) − (s + 1) X3 ( s) = 0 − sX1 ( s) + (s2 + 2s + 1) X2 (s) − (s + 1) X3 (s) = 0 − ( s + 1) X1 (s) − (s + 1) X2 (s) + (s 2 + 2s + 2) X3 (s) = F (s) Taking the inverse Laplace transform, •• • • • x1 + 2 x1 + x1 − x 2 − x3 − x 3 = 0 • •• • • − x1 + x 2 + 2 x 2 + x2 − x 3 − x3 = 0 • • •• • − x1 − x1 − x2 − x 2 + x3 + 2 x3 + 2 x 3 = f (t ) Simplifying, •• • • • x1 = −2 x1 − x1 + x2 + x 3 + x 3 •• • •• • • • x2 = x1 − 2 x2 − x 2 + x3 + x3 • • x3 = x1 + x1 + x2 + x 2 − 2 x3 − 2 x3 + f (t ) Defining the state variables, • • • z1 = x1 ; z2 = x1 ; z3 = x2 ; z4 = x2 ; z5 = x 3 ; z6 = x3 Writing the state equations using the simplified equations above yields, • • • •• • • • •• • • z1 = x1 = z 2 z2 = x1 = − 2z2 − z1 + z4 + z6 + z5 z3 = x2 = z4 z4 = x 2 = z2 − 2 z4 − z3 + z6 + z5 z5 = x3 = z6 • •• z6 = x3 = z 2 + z1 + z4 + z3 − 2 z6 − 2 z5 + f (t ) Converting to vector-matrix form, Solutions to Problems 55 ⎡0 1 0 0 0 ⎢ − 1 −2 0 1 1 ⎢0 0 0 1 0 • z =⎢ 0 1 −1 −2 1 ⎢0 0 0 0 0 ⎢ ⎣ 1 1 1 1 −2 y = [1 0 0 0 ⎤ ⎡ 0⎤ 1 ⎥ ⎢ 0⎥ 0 ⎥ ⎢ 0⎥ z+ f (t ) 1 ⎥ ⎢ 0⎥ 1 ⎥ ⎢ 0⎥ ⎥ ⎢⎥ − 2⎦ ⎣ 1⎦ 0 0 0 ]z 6. Drawing the equivalent network, Writing the equations of motion, (100 s 2 + 100)θ2 − 100θ 3 = 10T − 100θ 2 + (100s2 + 100s + 100 )θ 3 = 0 Taking the inverse Laplace transform and simplifying, •• θ 2 + 2 − θ3 = •• 1 T 10 • − θ2 + θ 3 + θ3 + θ 3 = 0 Defining the state variables as • • x1 = θ2 , x 2 = θ 2 , x3 = θ3 , x 4 = θ 3 Writing the state equations using the equations of motion and the definitions of the state variables 56 Chapter 3: Modeling in the Time Domain • x1 = x 2 • •• x2 = θ 2 = −θ 2 + θ3 + 1 1 T = − x1 + x3 + T 10 10 • x3 = x 4 • , •• • x4 = θ 3 = θ 2 − θ 3 − θ3 = x1 − x 3 − x 4 y = 10θ 2 = 10 x1 In vector-matrix form, ⎡0 ⎢ • ⎢−1 x=⎢ ⎢0 ⎢1 ⎣ ⎡0⎤ 1 0 0⎤ ⎢1⎥ ⎥ 0 1 0⎥ ⎢⎥ ⎥ x + ⎢ 10 ⎥T 0 0 1⎥ ⎢0⎥ ⎢⎥ ⎥ 0 − 1 −1⎦ ⎣0⎦ y = [10 0 0 0]x 7. Drawing the equivalent circuit, Writing the equations of motion, 42θ 2 ( s) − 2θ 3 (s) = 20 T(s) −2θ 2 (s) + (3s + 2)θ 3 ( s) − 3sθ 4 (s) = 0 −3sθ3 (s) + 5sθ 4 (s) = 0 Taking the inverse Laplace transform, 42 θ2 (t ) − 2θ 3 (t ) = 20T (t ) • • −2θ 2 (t ) + 3 θ 3 (t ) + 2θ 3 − 3 θ 4 (t ) = 0 • (2) • −3 θ3 (t ) + 5 θ 4 (t ) = 0 From (3), (1) (3) Solutions to Problems 57 5 3 • • θ 3 (t ) = θ4 (t ) and 5 3 θ 3 (t ) = θ4 (t ) (4) assuming zero initial conditions. From (1) θ 2 (t ) = 1 10 5 10 θ 3 (t ) + T (t ) = θ4 (t ) + T (t ) 21 21 63 21 (5) Substituting (4) and (5) into (2) yields the state equation (notice there is only one equation), • θ 4 (t ) = − 100 10 θ 4 ( t ) + T (t ) 63 21 The output equation is given by, θL ( t) = 1 θ 4 (t ) 10 8. Solving Eqs. (3.44) and (3.45) in the text for the transfer functions 0 X 1 s F = −K s2 + K M2 −K M 1 s2 + D s + K −K M 1 s2 + D s + K M2 X1 (s) X (s) and 2 : F( s) F (s) and X 2 F M 1 s2 + D s + K = s 0 −K −K M 2 s2 + K −K s2 + K Thus, X1 s Fs = X2 s Fs = K 4 3 2 2 M 2 M 1 s +D M 2 s +K M 2 s +K M 1 s +D K s and 2 M 1 s +D s +K 4 3 2 2 M 2 M 1 s +D M 2 s +K M 2 s +K M 1 s +D K s Multiplying each of the above transfer functions by s to find velocity yields pole/zero cancellation at the origin and a resulting transfer function that is third order. 9. a. . Using the standard form derived in the textbook, 1 0 0⎤ ⎡0 ⎡ 0⎤ • ⎢0 ⎢ 0⎥ 0 1 0⎥ r( t ) x= x+ ⎢0 ⎢ 0⎥ 0 0 1⎥ ⎢− 100 −7 −10 −20 ⎥ ⎢ 1⎥ ⎦ ⎣ ⎣⎦ c = [100 0 0 0]x b. Using the standard form derived in the textbook, 58 Chapter 3: Modeling in the Time Domain 1 0 0 0⎤ ⎡0 ⎡0 ⎤ ⎢0 ⎢0 ⎥ 0 1 0 0⎥ • ⎥ x + ⎢0 ⎥ r(t ) ⎢0 x= 0010 ⎢0 ⎢0 ⎥ 0 0 0 1⎥ ⎢− 30 −1 −6 −9 −8⎥ ⎢1 ⎥ ⎦ ⎣ ⎣⎦ c = [30 0 0 0 0]x 10. Program: 'a' num=100; den=[1 20 10 7 100]; G=tf(num,den) [Acc,Bcc,Ccc,Dcc]=tf2ss(num,den); Af=flipud(Acc); A=fliplr(Af) B=flipud(Bcc) C=fliplr(Ccc) 'b' num=30; den=[1 8 9 6 1 30]; G=tf(num,den) [Acc,Bcc,Ccc,Dcc]=tf2ss(num,den); Af=flipud(Acc); A=fliplr(Af) B=flipud(Bcc) C=fliplr(Ccc) Computer response: ans = a Transfer function: 100 --------------------------------s^4 + 20 s^3 + 10 s^2 + 7 s + 100 A= 0 0 0 -100 1 0 0 -7 0 1 0 -10 0 0 1 -20 0 0 0 B= 0 0 0 1 C= 100 ans = b Solutions to Problems 59 Transfer function: 30 -----------------------------------s^5 + 8 s^4 + 9 s^3 + 6 s^2 + s + 30 A= 0 0 0 0 -30 1 0 0 0 -1 0 1 0 0 -6 0 0 1 0 -9 0 0 0 1 -8 0 0 0 0 B= 0 0 0 0 1 C= 30 11. a. Using the standard form derived in the textbook, ⎡0 • ⎢0 x= ⎢0 ⎢− 10 ⎣ 0⎤ ⎡ 0⎤ 0⎥ ⎢ 0⎥ x+ r (t ) ⎢ 0⎥ 00 1⎥ ⎢ 1⎥ −5 − 1 − 2⎥ ⎣⎦ ⎦ 1 0 0 1 c = [10 5 0 0 0]x b. Using the standard form derived in the textbook, ⎡0 ⎢0 • x = ⎢0 ⎢0 ⎢0 ⎣ 10 0 0⎤ ⎡ 0⎤ ⎢ 0⎥ 01 0 0⎥ 00 1 0 ⎥ x + ⎢ 0⎥ r(t ) ⎢ 0⎥ 00 0 1⎥ ⎢ 1⎥ 0 − 8 − 10 −9⎥ ⎦ ⎣⎦ c = [3 7 12 2 1]x 12. Program: 'a' num=[5 10]; den=[1 2 1 5 10] G=tf(num,den) [Acc,Bcc,Ccc,Dcc]=tf2ss(num,den); Af=flipud(Acc); A=fliplr(Af) B=flipud(Bcc) 60 Chapter 3: Modeling in the Time Domain C=fliplr(Ccc) 'b' num=[1 2 12 7 3]; den=[1 9 10 8 0 0] G=tf(num,den) [Acc,Bcc,Ccc,Dcc]=tf2ss(num,den); Af=flipud(Acc); A=fliplr(Af) B=flipud(Bcc) C=fliplr(Ccc) Computer response: ans = a den = 1 2 1 5 10 Transfer function: 5 s + 10 ---------------------------s^4 + 2 s^3 + s^2 + 5 s + 10 A= 0 0 0 -10 1 0 0 -5 0 1 0 -1 0 0 1 -2 5 0 0 9 10 8 B= 0 0 0 1 C= 10 ans = b den = 1 0 Transfer function: s^4 + 2 s^3 + 12 s^2 + 7 s + 3 -----------------------------s^5 + 9 s^4 + 10 s^3 + 8 s^2 A= 0 0 0 0 0 1 0 0 0 0 0 1 0 0 -8 0 0 1 0 -10 0 0 0 1 -9 0 Solutions to Problems 61 B= 0 0 0 0 1 C= 3 7 12 2 1 13. The transfer function can be represented as a block diagram as follows: R(s) X(s) 1 3 2 s + 6s + 9s + 4 Y(s) s + 3s + 7 2 Writing the differential equation for the first box: ••• •• • x + 6 x + 9 x + 4 x = r (t ) Defining the state variables: x1 = x • x2 = x •• x3 = x Thus, • x1 = x2 • x2 = x3 • • •• x3 = −4 x − 9 x − 6 x + r (t ) = −4 x1 − 9 x2 − 6 x3 + r (t ) From the second box, •• • y = x + 3 x + 7 x = 7 x1 + 3 x2 + x3 In vector-matrix form: 0⎤ ⎡0 1 ⎡0 ⎤ ⎢ 0 0 1 ⎥ x + ⎢ 0 ⎥ r (t ) x=⎢ ⎥ ⎢⎥ ⎢ −4 −9 −6 ⎦ ⎥ ⎢1 ⎥ ⎣ ⎣⎦ • y = [ 7 3 1] x 14. 62 Chapter 3: Modeling in the Time Domain a. G(s)=C(sI-A)-1B 0 0 1 0 0 1 -3 A= -2 0 0 -5 ;B= ;C= 1 -1 0 10 2 (sI - A) = 0 1 2 s3 + 5s + 2s +3 s +5s+2 s+5 -3 s(s+5) -3s -2s-3 1 s 2 s 10 Therefore, G(s) = 3 2 . s +5s +2s+3 b. G(s)=C(sI-A)-1B A= s I− A −1 2 3 −8 053 −3 −5 −4 1 ;B= 4 6 ; C = 1, 3, 6 s2 − s − 5 1 = 3 − 3 s 2 − 27 s + 157 s 3 s + 52 −9 s2 + 2 s − 32 3 s− 6 − 5 s+1 s2 − 7 s + 10 − 3 s + 15 − 8 s + 49 2 Therefore, G s = 49 s − 349 s + 452 3 2 s − 3 s − 27 s + 157 . c. G(s)=C(sI-A)-1B ⎡ 3 −5 2 ⎤ ⎡5⎤ A = ⎢ 1 −8 7 ⎥; B = ⎢ − 3⎥ ; C = [1 − 4 3] ⎢ −3 − 6 2 ⎥ ⎢2⎥ ⎦ ⎣ ⎣⎦ (s I − A ) −1 ⎡ (s 2 + 6s + 26) − (5s + 2) (2s − 19) ⎤ 1 2 ⎢ (s − 23) =3 (s − 5s + 12) (7s − 19) ⎥ s + 3s 2 + 19s − 133 ⎢ 2 ⎥ − (6s − 33) (s + 5s − 19) ⎦ ⎣ − (3s + 30) Therefore, G(s) = 23s2 − 48s − 7 . s3 + 3s2 + 19s − 133 15. Program: 'a' A=[0 1 3 0;0 0 1 0;0 0 0 1;-7 -9 -2 -3]; B=[0;5;8;2]; C=[1 3 4 6]; D=0; statespace=ss(A,B,C,D) Solutions to Problems 63 [num,den]=ss2tf(A,B,C,D); G=tf(num,den) 'b' A=[3 1 0 4 -2;-3 5 -5 2 -1;0 1 -1 2 8;-7 6 -3 -4 0;-6 0 4 -3 1]; B=[2;7;6;5;4]; C=[1 -2 -9 7 6]; D=0; statespace=ss(A,B,C,D) [num,den]=ss2tf(A,B,C,D); G=tf(num,den) Computer response: ans = a a= x1 x2 x3 x4 x1 0 0 0 -7 x2 1 0 0 -9 x3 3 1 0 -2 x4 0 0 1 -3 x2 3 x3 4 x4 6 b= x1 x2 x3 x4 u1 0 5 8 2 c= y1 x1 1 d= y1 u1 0 Continuous-time model. Transfer function: 59 s^3 - 164 s^2 - 1621 s - 260 ------------------------------s^4 + 3 s^3 + 2 s^2 + 30 s + 7 ans = b a= x1 x2 x3 x4 x5 x1 3 -3 0 -7 -6 b= x1 u1 2 x2 1 5 1 6 0 x3 0 -5 -1 -3 4 x4 4 2 2 -4 -3 x5 -2 -1 8 0 1 64 Chapter 3: Modeling in the Time Domain x2 x3 x4 x5 7 6 5 4 c= y1 x1 1 x2 -2 x3 -9 x4 7 x5 6 d= y1 u1 0 Continuous-time model. Transfer function: -7 s^4 - 408 s^3 + 1708 s^2 + 1.458e004 s + 2.766e004 ----------------------------------------------------s^5 - 4 s^4 - 32 s^3 + 148 s^2 - 1153 s - 4480 16. Program: syms s 'a' A=[0 1 3 0 0010 0001 -7 -9 -2 -3]; B=[0;5;8;2]; C=[1 3 4 6]; D=0; I=[1 0 0 0 0100 0010 0 0 0 1]; 'T(s)' T=C*((s*I-A)^-1)*B+D; T=simple(T); pretty(T) 'b' A=[3 1 0 4 -2 -3 5 -5 2 -1 0 1 -1 2 8 -7 6 -3 -4 0 -6 0 4 -3 1]; B=[2;7;6;5;4]; C=[1 -2 -9 7 6]; D=0; I=[1 0 0 0 0 01000 00100 00010 0 0 0 0 1]; 'T(s)' T=C*((s*I-A)^-1)*B+D; T=simple(T); pretty(T) Computer response: ans = a Solutions to Problems 65 ans = T(s) 2 3 -164 s - 1621 s - 260 + 59 s -----------------------------4 3 2 s + 3 s + 2 s + 30 s + 7 ans = b ans = T(s) 2 3 4 14582 s + 1708 s - 408 s - 7 s + 27665 -----------------------------------------5 4 3 2 s - 4 s - 32 s + 148 s - 1153 s - 4480 17. . d θz Let the input be dt =ωz, x1=θx , x2=θ x . Therefore, . x 1 = x2 . Kx Dx x 2 = - J x1 - J x2 + Jωωz x x The output is θx. In vector-matrix form, θx = x1 . Therefore, y = x1. . x= y= - 0 Kx - Jx 1 0 1 Dx x+ Jx 0 Jω ωz x 18. The equivalent cascade transfer function is as shown below. Ka K d (s) X(s) 3 3 s+ K K 22 s+ 3 K K 1 3 K s+ K K s+ 0 3 b Ka F (s) 66 Chapter 3: Modeling in the Time Domain .. . For the first box, x + K 2 .. K x+ 3 K K 1. K 3 x+ 0 K 3 x= Ka K δ(t). 3 . .. Selecting the phase variables as the state variables: x =x, x =x, x =x. 1 2 3 Writing the state and output equations: . x 1 = x2 . x 2 = x3 . K1 K0 K2 Ka x 3 = - K x1- K x2- K x3+ K δ(t) 3 3 3 3 . Kb Kb y = φ(t) = x + K x = K x1+x2 a a In vector-matrix form, 0 1 0 0 K 0 0 1 0 b . x + K δ(t) ; y = x= K K K Ka a 2 1 0 K K K K 3 3 19. 3 1 0x 3 dωm Since Tm = Jeq dt + Deqωm, and Tm = Kt ia, dωm Jeq dt + Deqωm = Kt ia Or, Kt dωm Deq =- J ωm + J ia dt eq eq N2 But, ωm = N ωL. 1 Substituting in (1) and simplifying yields the first state equation, Kt N1 Deq dωL =-J ωL + J N ia dt eq eq 2 The second state equation is: dθL dt = ωL Since N2 dia dia ea = Raia+La dt +Kbωm = Raia+La dt +KbN ωL, 1 dia the third state equation is found by solving for dt . Hence, dia Kb N2 Ra 1 = - L N ωL - L ia+ L ea dt a1 a a (1) Solutions to Problems 67 Thus the state variables are: x1 = ωL, x2 = θL , and x3 = ia. N2 Finally, the output is y = θm = N θL . 1 In vector-matrix form, D eq - J eq . x= K b 0 N La N 0 2 1 - Kt N1 J eq N 0 2 0 - 1 0 x+ 0 1 La Ra N ea ; y = 0 N 2 1 La where, ω x= θ L L ia 20. Writing the differential equations, d2x1 dx1 dx2 2 2 + dt + 2x1 - dt = 0 dt d2x2 dx2 dx1 + dt - dt = f(t) dt2 Defining the state variables to be x1, v1, x2, v2, where v's are velocity, . x 1 = v1 . x 2 = v2 . v 1 = -v1-2x12+v2 . v 2 = v1-v2+f(t) . . Around x1 = 1, x1 = 1+δx1, and x 1 = δ x 1 . Also, 2 1 x =x 2 1 dx ⎮ x =1 + 1 dt ⎮ δ x 1 = 1 + 2x 1 ⎮ δ x 1 = 1 + 2 δ x 1 x =1 Therefore, the state and output equations are, . δx 1 = v1 . x 2 = v2 . v 1 = -v1-2(1+2δx1)+v2 x =1 0x 68 Chapter 3: Modeling in the Time Domain . v 2 = v1-v2+f(t) y = x2 In vector-matrix form, . δx 1 δx v 0 0 = -4 0 0 0 1 0 -1 . 0 0 1 1 . 2 1 v 0 0 + -2 0 0 0 v -1 x δx v 0 1 1 . 0 1 1 x 2 1 1 f(t) x ;y= 0 0 0 2 v 1 v 2 2 2 1 where f(t) = 2 + δf(t), since force in nonlinear spring is 2 N and must be balanced by 2 N force on damper. 21. Controller: The transfer function can be represented as a block diagram as follows: Rc (s) Xc (s) K1 s+b s+ a Writing the differential equation for the first box, K1 s+b • and solving for xc , • xc = −bxc + K1rc (t ) From the second box, • yc = xc + axc = −bxc + K1rc (t ) + axc = (a − b) xc + K1rc (t ) Wheels: The transfer function can be represented as a block diagram as follows: Rw (s) c s+c Xw(s) Yc (s) Solutions to Problems 69 Writing the differential equation for the block of the form, c s+c • and solving for xw , • xw = −cxw + crw (t ) The output equation is, yw = xw Vehicle: The transfer function can be represented as a block diagram as follows: Rv (s) Xv(s) 1 s Writing the differential equation for the block, 1 s • and solving for xv , • xv = rv (t ) The output equation is yv = xv 22. A= − 1.702 50.72 263.38 0.22 − 1.418 − 31.99 0 0 − 14 ;B = − 272.06 0 14 For G1(s), C 1 = 1, 0, 0 , and G1(s) = C1(sI-A)-1B Thus, s 2 + 15.418 s + 19.852 G 1s 1 =C1 s 3 + 17.12 s 2 + 34.935 s − 122.43 0.22 s + 3.08 50.72 s + 710.08 2 + 15.702 s + 23.828 s 0 0 263.38 s − 1249.1 − 31.99 s + 3.4966 s 2 + 3.12 s − 8.745 Or G1 s = − 272.06 s 2 − 507.3 s − 22888 2 = − 272.06 s + 1.8647 s + 84.128 s + 14 s − 1.7834 s + 4.9034 s 3 + 17.12 s 2 + 34.935 s − 122.43 For G2(s), C2 = (0,1,0), and B 70 Chapter 3: Modeling in the Time Domain G2(s) = C2(sI-A)-1B Thus, s 2 + 15.418 s + 19.852 G2 s 0.22 s + 3.08 Or G2 s = 50.72 s + 710.08 2 + 15.702 s + 23.828 s 0 1 =C 2 s 3 + 17.12 s 2 + 34.935 s − 122.43 0 − 507.71 s 788.99 s 3 + 17.12 s 2 + 34.935 s − 122.43 = 263.38 s − 1249.1 − 31.99 s + 3.4966 s 2 + 3.12 s − 8.745 − 507.71 s + 1.554 s + 14 s − 1.7834 s + 4.9034 23. Adding displacements to the figure, xs xr xe z Writing the differential equations for noncontact, d 2 xr dxr dx + 2 x r − x s − s = u( t ) 2 +2 dt dt dt 2 dx d x dx − r − xr + 2 s + s + xs = 0 dt dt dt Define the state variables as, • • x1 = x r ; x 2 = xr ; x 3 = x s ; x 4 = xs Writing the state equations, using the differential equations and the definition of the state variables, we get, • • • •• • • • •• x1 = x r = x 2 x2 = x r = −2 x1 − 2 x2 + x 3 + x 4 + u(t ) x3 = x s = x4 x4 = x s = x1 + x2 − x3 − x 4 B Solutions to Problems 71 Assuming the output to be xs, the output equation is, y = x3 In vector-matrix form, ⎡0 1 0 0⎤ ⎡ 0⎤ • ⎢−2 −2 1 1 ⎥ ⎢ 1⎥ u(t ) x= x+ ⎢0 0 0 1⎥ ⎢ 0⎥ ⎢ 1 1 −1 −1⎥ ⎢ 0⎥ ⎦ ⎣ ⎣⎦ y = [0 0 1 0]x Writing the differential equations for contact, 2 d xr dx dx + 2 r + 2 x r − x s − s = u( t ) 2 dt dt dt dxr d 2 x dx − xr + 2 s + s + xs − z − x e = 0 dt dt dt dz dxe − xs + +z − =0 dt dt dz d 2 xe dx − xs − + 2 + 2 e + 2 xe = 0 dt dt dt − Defining the state variables, • • • • x1 = x r ; x 2 = xr ; x 3 = x s ; x 4 = xs ; x 5 = z; x 6 = z; x7 = xe ; x8 = xe Using the differential equations and the definitions of the state variables, we write the state equations. • x1 = x 2 • x2 = − x1 − 2 x 2 + x3 + x4 + u (t ) • x3 = x 4 • x4 = x1 + x2 − x 3 − x 4 + x5 + x 7 • x5 = x 6 Differentiating the third differential equation and solving for d2z/dt2 we obtain, • x6 = d 2 z dxs dz d 2 xe = − + dt dt dt 2 dt 2 But, from the fourth differential equation, dz d 2 xe dx − 2 e − 2 xe = x3 + x6 − 2 x 8 − 2 x7 2 = xs + dt dt dt 72 Chapter 3: Modeling in the Time Domain • Substituting this expression back into x6 along with the other definitions and then simplifying yields, • x6 = x 4 + x3 − 2 x8 − 2x 7 Continuing, • x7 = x 8 • x8 = x3 + x 6 − 2 x7 − 2 x 8 Assuming the output is xs, y = xs Hence, the solution in vector-matrix form is ⎡0 1 0 0 ⎢− 1 −2 1 1 ⎢0 0 0 1 ⎢ 1 1 −1 −1 • x=⎢ 0000 ⎢ 0011 ⎢ ⎢0 0 0 0 ⎣0 0 1 0 0 0 0 1 0 0 0 0 0 0 0⎤ ⎡ 0⎤ ⎢ 1⎥ 0 0 0⎥ ⎢ 0⎥ 0 0 0⎥ ⎢ 0⎥ 0 1 0⎥ x + ⎢ ⎥ u(t ) 0 1 0 0⎥ ⎥ ⎢⎥ 0 0 −2 − 2 ⎥ ⎢⎥ 0 0 1⎥ ⎢ 0⎥ 1 −2 − 2⎦ ⎣ 0⎦ y = [0 0 1 0 0 0 0 0]x 24. Writing the equations of motion, d 2y f dy dy Mf + ( fvf + fvh ) f + Kh y f − f vh h − Kh yh = fup (t ) 2 dt dt dt 2 dy dy dy − f vh f − Kh y f + Mh 2h + fvh h + ( Kh + Ks )yh − Ks ycat = 0 dt dt dt − Ks yh + ( Ks + K ave )ycat = 0 The last equation says that ycat = Ks y ( Ks + K ave ) h Defining state variables for the first two equations of motion, • • x1 = yh ; x 2 = yh ; x3 = y f ; x4 = y f Solving for the highest derivative terms in the first two equations of motion yields, Solutions to Problems 73 d 2 yf ( f vf + f vh ) dy f Kh f dy K 1 − y f + vh h + h yh + f (t ) 2 =− dt Mf dt M f M f dt M f M f up d 2 yh f dy K f dy ( K + Ks ) K = vh f + h y f − vh h − h yh + s ycat 2 dt Mh dt Mh Mh dt Mh Mh Writing the state equations, • x1 = x 2 • x2 = f vh K f ( K + Ks ) K Ks x 4 + h x3 − vh x2 − h x1 + s x Mh Mh Mh Mh Mh ( Ks + K ave ) 1 • x3 = x 4 • x4 = − ( f vf + f vh ) K f K 1 x4 − h x 3 + vh x2 + h x1 + f (t ) Mf Mf Mf Mf M f up The output is yh - ycat. Therefore, y = y h − ycat = yh − K ave Ks x yh = ( Ks + K ave ) 1 ( Ks + Kave ) Simplifying, rearranging, and putting the state equations in vector-matrix form yields, 0 ⎡ ⎞ Ks2 1⎛ ⎢ ⎜ − ( Kh + Ks )⎟ • M ⎝ ( Ks + K ave ) ⎠ x=⎢ h 0 ⎢ Kh ⎢ Mf ⎣ 1 f − vh Mh 0 f vh Mf 0 Kh Mh 0 K −h Mf 0 ⎤ ⎡0 f vh ⎥ ⎢0 Mh ⎥x + ⎢ 0 1 ( f vf + f vh ) ⎥ ⎢1 − ⎥ ⎢ Mf ⎣ Mf ⎦ ⎡ ⎤ Kave y=⎢ 0 0 0⎥ x ⎣ (K s + Kave ) ⎦ Substituting numerical values, 1 0 0⎤ ⎡0 ⎡0⎤ • ⎢− 9353 −14.29 769.2 14.29 ⎥ ⎢0⎥ f (t ) x= x+ ⎢0 ⎢ 0 ⎥ up 0 0 1⎥ ⎢ 406 ⎢0.0581⎥ 7.558 − 406 −9.302⎥ ⎦ ⎦ ⎣ ⎣ y = [0.9491 0 0 0]x ⎤ ⎥ ⎥ f up (t ) ⎥ ⎥ ⎦ FOUR Time Response SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Open-Loop Response The forward transfer function for angular velocity is, ω0(s) 24 G(s) = V (s) = (s+150)(s+1.32) P a. ω0(t) = A + Be-150t + Ce-1.32t 24 b. G(s) = 2 . Therefore, 2ζωn =151.32, ωn = 14.07, and ζ = 5.38. s +151.32s+198 24 c. ω0(s) = 2+151.32s+198) = s(s Therefore, ω0(t) = 0.12121 + .0010761 e-150t - 0.12229e-1.32t. d. Using G(s), •• • ω 0 + 151.32 ω 0 + 198ω 0 = 24v p (t ) Defining, x1 = ω 0 • x2 = ω 0 Thus, the state equations are, • x1 = x 2 • x2 = −198 x1 − 151.32 x2 + 24 v p (t ) y = x1 In vector-matrix form, 1⎤ • ⎡0⎤ ⎡0 x=⎢ x + ⎢ ⎥ v p (t ); y = [1 0 ]x ⎣ 24⎦ ⎣− 198 − 151.32⎥ ⎦ Solutions to Case Studies Challenges 75 e. Program: 'Case Study 1 Challenge (e)' num=24; den=poly([-150 -1.32]); G=tf(num,den) step(G) Computer response: ans = Case Study 1 Challenge (e) Transfer function: 24 ------------------s^2 + 151.3 s + 198 Ship at Sea: Open-Loop Response a. Assuming a second-order approximation: ωn2 = 2.25, 2ζωn = 0.5. Therefore ζ = 0.167, ωn = 1.5. Ts = π 4 = 16; TP = = 2.12 ; ζωn ωn 1-ζ2 %OS = e-ζπ / b. θ s = 1− s 1-ζ 2 x 100 = 58.8%; ωnTr = 1.169 therefore, Tr = 0.77. s + 0.5 2.25 = = 1− 2 + 0.5 s + 2.25 2 + 0.5 s + 2.25 ss ss 0.25 s + 0.25 + 2.1875 2.1875 s + 0.25 2 + 2.1875 76 Chapter 4: Time Response = 1 − s + 0.25 + 0.16903 1.479 s s + 0.25 2 + 2.1875 Taking the inverse Laplace transform, θ(t) = 1 - e-0.25t ( cos1.479t +0.16903 sin1.479t) c. Program: 'Case Study 2 Challenge (C)' '(a)' numg=2.25; deng=[1 0.5 2.25]; G=tf(numg,deng) omegan=sqrt(deng(3)) zeta=deng(2)/(2*omegan) Ts=4/(zeta*omegan) Tp=pi/(omegan*sqrt(1-zeta^2)) pos=exp(-zeta*pi/sqrt(1-zeta^2))*100 t=0:.1:2; [y,t]=step(G,t); Tlow=interp1(y,t,.1); Thi=interp1(y,t,.9); Tr=Thi-Tlow '(b)' numc=2.25*[1 2]; denc=conv(poly([0 -3.57]),[1 2 2.25]); [K,p,k]=residue(numc,denc) '(c)' [y,t]=step(G); plot(t,y) title('Roll Angle Response') xlabel('Time(seconds)') ylabel('Roll Angle(radians)') Computer response: ans = Case Study 2 Challenge (C) ans = (a) Transfer function: 2.25 -----------------s^2 + 0.5 s + 2.25 omegan = 1.5000 zeta = 0.1667 Ts = 16 Solutions to Case Studies Challenges 77 Tp = 2.1241 pos = 58.8001 Tr = 0.7801 ans = (b) K= 0.1260 -0.3431 + 0.1058i -0.3431 - 0.1058i 0.5602 p= -3.5700 -1.0000 + 1.1180i -1.0000 - 1.1180i 0 k= ans = (c) 78 Chapter 4: Time Response ANSWERS TO REVIEW QUESTIONS 1.Time constant 2. The time for the step response to reach 67% of its final value 3. The input pole 4. The system poles 5. The radian frequency of a sinusoidal response 6. The time constant of an exponential response 7. Natural frequency is the frequency of the system with all damping removed; the damped frequency of oscillation is the frequency of oscillation with damping in the system. 8. Their damped frequency of oscillation will be the same. 9. They will all exist under the same exponential decay envelop. 10. They will all have the same percent overshoot and the same shape although differently scaled in time. 11. ζ, ωn, TP, %OS, Ts 12. Only two since a second-order system is completely defined by two component parameters 13. (1) Complex, (2) Real, (3) Multiple real 14. Pole's real part is large compared to the dominant poles, (2) Pole is near a zero 15. If the residue at that pole is much smaller than the residues at other poles 16. No; one must then use the output equation 17. The Laplace transform of the state transition matrix is (sI -A)-1 18. Computer simulation 19. Pole-zero concepts give one an intuitive feel for the problem. 20. State equations, output equations, and initial value for the state-vector 21. Det(sI-A) = 0 SOLUTIONS TO PROBLEMS 1. a. Overdamped Case: C(s) = 9 s(s2 + 9s + 9) Expanding into partial fractions, Taking the inverse Laplace transform, c(t) = 1 + 0.171 e-7.854t - 1.171 e-1.146t Solutions to Problems 79 b. Underdamped Case: K2 and K3 can be found by clearing fractions with K1 replaced by its value. Thus, 9 = (s2 + 3s + 9) + (K2s + K3)s or 9 = s2 + 3s +9 + K2s2 + K3s Hence K2 = -1 and K3 = -3. Thus, c(t) = 1 - 2 -3t/2 e cos( 3 27 4 t - φ) = 1 - 1.155 e -1.5t cos (2.598t - φ) where φ = arctan ( c. Oscillatory Case: 3 ) = 30 o 27 80 Chapter 4: Time Response The evaluation of the constants in the numerator are found the same way as they were for the underdamped case. The results are K2 = -1 and K3 = 0. Hence, Therefore, c(t) = 1 - cos 3t d. Critically Damped The constants are then evaluated as Now, the transform of the response is c(t) = 1 - 3t e-3t - e-3t 2. 5 1 1 a. C(s) = s(s+5) = s - s+5 . Therefore, c(t) = 1 - e-5t. 1 2.2 2.2 4 4 Also, T = 5 , Tr = a = 5 = 0.44, Ts = a = 5 = 0.8. 20 1 1 1 b. C(s) = s(s+20) = s - s+20 . Therefore, c(t) = 1 - e-20t. Also, T = 20 , 2.2 2.2 4 4 Tr = a = 20 = 0.11, Ts = a = 20 = 0.2. Solutions to Problems 81 3. Program: '(a)' num=5; den=[1 5]; Ga=tf(num,den) subplot(1,2,1) step(Ga) title('(a)') '(b)' num=20; den=[1 20]; Gb=tf(num,den) subplot(1,2,2) step(Gb) title('(b)') Computer response: ans = (a) Transfer function: 5 ----s+5 ans = (b) Transfer function: 20 -----s + 20 82 Chapter 4: Time Response 4. 1 Cs 2 5 10 5 5 1 = (s+2) . Since Vi(s) = s , VC(s) = s(s+2) = s - s+2 . (R+Cs) 1 2.2 2.2 4 4 Therefore vC(t) = 5 - 5e-2t. Also, T = 2 , Tr = a = 2 = 1.1, Ts = a = 2 = 2. VC(s) Using voltage division, V (s) = i 5. Program: clf num=2; den=[1 2]; G=tf(num,den) step(5*G) Computer response: Transfer function: 2 ----s+2 6. Writing the equation of motion, ( Ms + 8s ) X (s ) = F (s ) 2 Thus, the transfer function is, Solutions to Problems 83 X (s ) 1 = 2 F (s) Ms + 8s Differentiating to yield the transfer function in terms of velocity, sX (s ) 1 1/ M = = F (s) Ms + 8 s + 8 M Thus, the settling time, Ts, and the rise time, Tr, are given by Ts = 4 1 2.2 = M ; Tr = = 0.275M 8/ M 2 8/M 7. Program: Clf M=1 num=1/M; den=[1 8/M]; G=tf(num,den) step(G) pause M=2 num=1/M; den=[1 8/M]; G=tf(num,den) step(G) Computer response: Transfer function: M= 1 Transfer function: 1 ----s+8 M= 2 Transfer function: 0.5 ----s+4 84 Chapter 4: Time Response From plot, time constant = 0.125 s. From plot, time constant = 0.25 s. 8. a. Pole: -2; c(t) = A + Be-2t ; first-order response. b. Poles: -3, -6; c(t) = A + Be-3t + Ce-6t; overdamped response. c. Poles: -10, -20; Zero: -7; c(t) = A + Be-10t + Ce-20t; overdamped response. Solutions to Problems 85 d. Poles: (-3+j3 15 ), (-3-j3 15 ) ; c(t) = A + Be-3t cos (3 15 t + φ); underdamped. e. Poles: j3, -j3; Zero: -2; c(t) = A + B cos (3t + φ); undamped. f. Poles: -10, -10; Zero: -5; c(t) = A + Be-10t + Cte-10t; critically damped. 9. Program: p=roots([1 6 4 7 2]) Computer response: p= -5.4917 -0.0955 + 1.0671i -0.0955 - 1.0671i -0.3173 10. G(s) = C (sI-A)-1 B ⎡ 8 −4 1 ⎤ ⎡1 ⎤ 0 ⎥ ; B = ⎢ 3⎥; C = [2 8 − 3] A = ⎢ −3 2 ⎢5 ⎢ 7⎥ 7 − 9⎥ ⎣ ⎦ ⎣⎦ (s I − A ) −1 ⎡ (s2 + 7s − 18) − (4 s + 29) (s − 2) ⎤ 1 ⎢ − (3s + 27) ⎥ (s 2 + s − 77) −3 =3 s − s2 − 91s + 67 ⎢ 2 7s − 76 (s − 10s + 4)⎥ ⎣ 5s − 31 ⎦ Therefore, G(s ) = 5s2 + 136s − 1777 . s 3 − s 2 − 91s + 67 Factoring the denominator, or using det(sI-A), we find the poles to be 9.683, 0.7347, -9.4179. 11. Program: A=[8 -4 1;-3 2 0;5 7 -9] B=[1;3;7] C=[2 8 -3] D=0 [numg,deng]=ss2tf(A,B,C,D,1); G=tf(numg,deng) poles=roots(deng) Computer response: A= 8 -3 5 B= 1 -4 2 7 1 0 -9 86 Chapter 4: Time Response 3 7 C= 2 8 -3 D= 0 Transfer function: 5 s^2 + 136 s - 1777 --------------------s^3 - s^2 - 91 s + 67 poles = -9.4179 9.6832 0.7347 12. VC(s) - V(s) 1 1 = 0. Writing the node equation at the capacitor, VC(s) (R + Ls + Cs) + R1 2 1 R1 VC(s) 10s 10 Hence, V(s) = 1 =2 . The step response is 2 .The poles 1 1 s +20s+500 s +20s+500 + R + Ls + Cs R1 2 are at -10 ± j20. Therefore, vC(t) = Ae-10t cos (20t + φ). 13. Program: num=[10 0]; den=[1 20 500]; G=tf(num,den) step(G) Computer response: Transfer function: 10 s ---------------s^2 + 20 s + 500 Solutions to Problems 87 14. X(s) 1 1 The equation of motion is: (Ms2+fvs+Ks)X(s) = F(s). Hence, F(s) = =2 . Ms2+fvs+Ks s +s+5 1 1 5s+5 1 1/5 The step response is now evaluated: X(s) = =s 1 19 = s(s2+s+5) (s+2)2+ 4 1 19 11 5(s+2) + 5 19 2 . 1 19 (s+2)2 + 4 11 19 1 Taking the inverse Laplace transform, x(t) = 5 - 5 e-0.5t ( cos 2 t + sin 19 1⎡ 5 19 ⎤ = 5 1 - 2 19 e-0.5t cos ( 2 t - 12.92o) . ⎣ ⎦ 19 2 t) 15. C(s) = s + 2ζωn 1 2+2ζω s+ω 2) = s - s2+2ζω s+ω 2 s(s n n n n ωn2 (s + ζωn) + ζωn (s+ζωn) + 1 1 =s=s 2 + (ω 1 - ζ2)2 (s+ζωn) n Hence, c(t) = 1 - e -ζω n t s + 2ζωn 1 = s(s+ζωn)2 + ωn2 - ζ2ωn2 ζωn ωn 1 - ζ2 ωn 1 - ζ2 (s+ζωn)2 + (ωn 1 - ζ2)2 ⎛ ⎜ cos ω n 1 - ζ 2 t + ⎝ ⎞ sin ω n 1 - ζ 2 t⎟ 1 - ζ2 ⎠ ζ 88 Chapter 4: Time Response = 1 - e-ζωnt where φ = tan-1 1+ ζ2 cos (ωn 1 - ζ2 t - φ) = 1 - e-ζωnt 1-ζ2 1 1-ζ2 cos (ωn 1 - ζ2 t - φ), ζ 1 - ζ2 16. %OS = e-ζπ / 1-ζ 2 x 100. Dividing by 100 and taking the natural log of both sides, %OS ln2 ( 100 ) ζπ %OS ln ( 100 ) = . Squaring both sides and solving for ζ2, ζ2 = %OS . Taking the 1 - ζ2 π2 + ln2 ( 100 ) %OS - ln ( 100 ) negative square root, ζ = . %OS π2 + ln2 ( 100 ) 17. a. b. c. d. Solutions to Problems 89 e. f. 18. a. N/A b. s2+9s+18, ωn2 = 18, 2ζωn = 9, Therefore ζ = 1.06, ωn = 4.24, overdamped. c. s2+30s+200, ωn2 = 200, 2ζωn = 30, Therefore ζ = 1.06, ωn = 14.14, overdamped. d. s2+6s+144, ωn2 = 144, 2ζωn = 6, Therefore ζ = 0.25, ωn = 12, underdamped. e. s2+9, ωn2 = 9, 2ζωn = 0, Therefore ζ = 0, ωn = 3, undamped. f. s2+20s+100, ωn2 = 100, 2ζωn = 20, Therefore ζ = 1, ωn = 10, critically damped. 19. 1002 1 s+100 1 (s+50) + 50 1 =s =s =s(s+50)2+7500 (s+50)2+7500 50 sin 7500 t) Therefore, x(t) = 1 - e-50t (cos 7500 t + 7500 2 -50t 1 =1e cos (50 3 t - tan-1 ) 3 3 X(s) = s(s2 +100s+1002) 50 7500 7500 (s+50)2+7500 (s+50) + 20. a. ωn2 = 16 r/s, 2ζωn = 3. Therefore ζ = 0.375, ωn = 4. Ts = 0.8472 s; %OS = e-ζπ / therefore, Tr = 0.356 s. 1-ζ 2 4 = 2.667 s; TP = ζωn ω π 2 n 1-ζ = x 100 = 28.06 %; ωnTr = (1.76ζ3 - 0.417ζ2 + 1.039ζ + 1) = 1.4238; 90 Chapter 4: Time Response b. ωn2 = 0.04 r/s, 2ζωn = 0.02. Therefore ζ = 0.05, ωn = 0.2. Ts = 15.73 s; %OS = e-ζπ / 1-ζ 2 4 = 400 s; TP = ζωn ω π 2 n 1-ζ = x 100 = 85.45 %; ωnTr = (1.76ζ3 - 0.417ζ2 + 1.039ζ + 1); therefore, Tr = 5.26 s. c. ωn2 = 1.05 x 107 r/s, 2ζωn = 1.6 x 103. Therefore ζ = 0.247, ωn = 3240. Ts = π ωn 1-ζ2 = 0.001 s; %OS = e-ζπ / 1-ζ 2 4 = 0.005 s; TP = ζωn x 100 = 44.92 %; ωnTr = (1.76ζ3 - 0.417ζ2 + 1.039ζ + 1); therefore, Tr = 3.88x10-4 s. 21. Program: '(a)' clf numa=16; dena=[1 3 16]; Ta=tf(numa,dena) omegana=sqrt(dena(3)) zetaa=dena(2)/(2*omegana) Tsa=4/(zetaa*omegana) Tpa=pi/(omegana*sqrt(1-zetaa^2)) Tra=(1.76*zetaa^3 - 0.417*zetaa^2 + 1.039*zetaa + 1)/omegana percenta=exp(-zetaa*pi/sqrt(1-zetaa^2))*100 subplot(221) step(Ta) title('(a)') '(b)' numb=0.04; denb=[1 0.02 0.04]; Tb=tf(numb,denb) omeganb=sqrt(denb(3)) zetab=denb(2)/(2*omeganb) Tsb=4/(zetab*omeganb) Tpb=pi/(omeganb*sqrt(1-zetab^2)) Trb=(1.76*zetab^3 - 0.417*zetab^2 + 1.039*zetab + 1)/omeganb percentb=exp(-zetab*pi/sqrt(1-zetab^2))*100 subplot(222) step(Tb) title('(b)') '(c)' numc=1.05E7; denc=[1 1.6E3 1.05E7]; Tc=tf(numc,denc) omeganc=sqrt(denc(3)) zetac=denc(2)/(2*omeganc) Tsc=4/(zetac*omeganc) Tpc=pi/(omeganc*sqrt(1-zetac^2)) Trc=(1.76*zetac^3 - 0.417*zetac^2 + 1.039*zetac + 1)/omeganc percentc=exp(-zetac*pi/sqrt(1-zetac^2))*100 subplot(223) step(Tc) title('(c)') Computer response: ans = (a) Solutions to Problems 91 Transfer function: 16 -------------s^2 + 3 s + 16 omegana = 4 zetaa = 0.3750 Tsa = 2.6667 Tpa = 0.8472 Tra = 0.3559 percenta = 28.0597 ans = (b) Transfer function: 0.04 ------------------s^2 + 0.02 s + 0.04 omeganb = 0.2000 zetab = 0.0500 Tsb = 400 Tpb = 92 Chapter 4: Time Response 15.7276 Trb = 5.2556 percentb = 85.4468 ans = (c) Transfer function: 1.05e007 ----------------------s^2 + 1600 s + 1.05e007 omeganc = 3.2404e+003 zetac = 0.2469 Tsc = 0.0050 Tpc = 0.0010 Trc = 3.8810e-004 percentc = 44.9154 Solutions to Problems 93 22. Program: T1=tf(16,[1 3 16]) T2=tf(0.04,[1 0.02 0.04]) T3=tf(1.05e7,[1 1.6e3 1.05e7]) ltiview Computer response: Transfer function: 16 -------------s^2 + 3 s + 16 Transfer function: 0.04 ------------------s^2 + 0.02 s + 0.04 Transfer function: 1.05e007 ----------------------s^2 + 1600 s + 1.05e007 94 Chapter 4: Time Response Solutions to Problems 95 23. a. ζ = %OS - ln ( 100 ) %OS π2 + ln2 ( 100 ) = 0.56, ωn = 4 = 11.92. Therefore, poles = -ζωn ± jωn ζTs = -6.67 ± j9.88. b. ζ = %OS - ln ( 100 ) %OS π2 + ln2 ( 100 ) = 0.591, ωn = Therefore, poles = -ζωn ± jωn π TP 1-ζ2 = 0.779. 1-ζ2 = -0.4605 ± j0.6283. π 4 c. ζωn = T = 0.571, ωn 1-ζ2 = T = 1.047. Therefore, poles = -0.571 ± j1.047. s p 24. Re = 4 -ln(12.3/100) = 4; ζ = = 0.5549 2 Ts π + ln 2 (12.3/100) Re = ζω n = 0.5549ω n = 4; ∴ ω n = 7.21 Im = ω n 1 − ζ = 6 2 51.96 ω n2 ∴ G (s ) = 2 2= 2 s + 2ζω n s + ω n s + 8s + 51.96 25. a. Writing the equation of motion yields, (3s + 15s + 33) X ( s ) = F( s ) 2 1-ζ 2 96 Chapter 4: Time Response Solving for the transfer function, X (s ) 1/ 3 =2 F (s) s + 5s + 11 b. ωn2 = 11 r/s, 2ζωn = 5. Therefore ζ = 0.754, ωn = 3.32. Ts = s; %OS = e-ζπ / s. 1-ζ 2 4 = 1.6 s; TP = ζωn ω π 2 n 1-ζ = 1.44 x 100 = 2.7 %; ωnTr = (1.76ζ3 - 0.417ζ2 + 1.039ζ + 1); therefore, Tr = 0.69 26. Writing the loop equations, 2 s +s θ 1 s −s θ 2 s =T s −s θ 1 s + s +1 θ 2 s = 0 Solving for θ2(s), s 2 + s T (s θ2 s = −s 0 −s −s Forming the transfer function, s2 + s ) =Ts 2 s +s +1 s+1 θ2 s Ts = 1 2 s +s +1 Thus ωn = 1, 2ζωn = 1. Thus, ζ = 0.5. From Eq. (4.38), %OS = 16.3%. From Eq. (4.42), Ts = 8 seconds. From Eq. (4.34), Tp = 3.63 seconds. 27. a. 24.542 s+4 1 1 ==2 s(s + 4s + 24.542) s (s + 2) + 20.542 s 2 2 4.532 4.532 . 2 (s + 2) + 20.542 (s + 2) + Thus c(t) = 1 - e-2t (cos4.532t+0.441 sin 4.532t) = 1-1.09e-2t cos(4.532t -23.80). b. Solutions to Problems 97 Therefore, c(t) = 1 - 0.29e-10t - e-2t(0.71 cos 4.532t + 0.954 sin 4.532t) = 1 - 0.29e-10t - 1.189 cos(4.532t - 53.34o). c. Therefore, c(t) = 1 - 1.14e-3t + e-2t (0.14 cos 4.532t - 0.69 sin 4.532t) = 1 - 1.14e-3t + 0.704 cos(4.532t +78.53o). 28. Since the third pole is more than five times the real part of the dominant pole, s2+1.204s+2.829 determines the transient response. Since 2ζωn = 1.204, and ωn = 2.829 = ωn = 1.682, ζ = 0.358, %OS = e −ζπ / 1 −ζ 2 π 4 x100 = 30% , Ts = ζω = 6.64 sec, Tp = n ωn 1-ζ2 = 2 sec; ωnTr = 1.4, therefore, Tr = 0.832. 29. a. Measuring the time constant from the graph, T = 0.0244 seconds. Response 3 2 1 0 0 0.05 0.1 0.15 Time(seconds) T = 0.0244 seconds 0.2 0.25 Chapter 4: Time Response K K Estimating a first-order system, G(s) = s+a . But, a = 1/T = 40.984, and a = 2. Hence, K = 81.967. Thus, 81.967 G(s) = s+40.984 b. Measuring the percent overshoot and settling time from the graph: %OS = (13.82-11.03)/11.03 = 25.3%, 25 20 Response 98 15 cmax = 13.82 cfinal = 11.03 10 5 0 0 1 2 3 4 5 Time(seconds) Ts = 2.62 seconds and Ts = 2.62 seconds. Estimating a second-order system, we use Eq. (4.39) to find ζ = 0.4 , and Eq. K K . Since Cfinal = 11.03, 2 = 11.03. Hence, (4.42) to find ωn = 3.82. Thus, G(s) = 2 s +2ζωns +ωn2 ωn K = 160.95. Substituting all values, 160.95 G(s) = 2 s +3.056s+14.59 c. From the graph, %OS = 40%. Using Eq. (4.39), ζ = 0.28. Also from the graph, Tp = π ωn 1 − ζ 2 = 4. Substituting ζ = 0.28, we find ωn = 0.818. Thus, 0.669 K . G(s) = 2 2=2 s +2ζωns +ωn s + 0.458s + 0.669 Solutions to Problems 99 30. a. Since the amplitude of the sinusoids are of the same order of magnitude as the residue of the pole at 2, pole-zero cancellation cannot be assumed. b. Since the amplitude of the sinusoids are of the same order of magnitude as the residue of the pole at 2, pole-zero cancellation cannot be assumed. c. Since the amplitude of the sinusoids are of two orders of magnitude larger than the residue of the pole at -2, pole-zero cancellation can be assumed. Since 2ζωn = 1, and ωn = 5 = 2.236, ζ = 0.224, %OS = e −ζπ / 1 −ζ 2 4 x100 = 48.64% , Ts = ζω = 8 sec, Tp = n π ωn 1-ζ2 = 1.44 sec; ωnTr = 1.23, therefore, Tr = 0.55. d. Since the amplitude of the sinusoids are of two orders of magnitude larger than the residue of the pole at -2, pole-zero cancellation can be assumed. Since 2ζωn = 5, and ωn = 20 = 4.472, ζ = 0.559, 100 Chapter 4: Time Response %OS = e −ζπ / 1 −ζ 2 4 x100 = 12.03% , Ts = ζω = 1.6 sec, Tp = n π ωn 1-ζ2 = 0.847 sec; ωnTr = 1.852, therefore, Tr = 0.414. 31. Program: %Form sC(s) to get transfer function clf num=[1 3]; den=conv([1 3 10],[1 2]); T=tf(num,den) step(T) Computer response: Transfer function: s+3 ----------------------s^3 + 5 s^2 + 16 s + 20 %OS = (0.163 - 0.15) = 8.67% 0.15 32. Only part c can be approximated as a second-order system. From the exponentially decaying cosine the poles are located at s1,2 = − 2 ± j 9.796 . Thus, Solutions to Problems 101 Ts = Also, ω n = overshoot. π π 4 4 = 0.3207 s = = = 2 s; Tp = Im 9.796 Re 2 2 2 2 + 9.796 = 10 and ζω n = Re = 2 . Hence, ζ = 0.2 , yielding 52.66 percent 33. a. (1) 1 33.75 33.75 1 C a1 s = = = 0.17213 33.75 = 0.17213 5.8095 2 + 33.75 s + 1.5 s + 1.5 2 + 33.75 s + 1.5 2 + 33.75 s 2 + 3 s + 36 Taking the inverse Laplace transform Ca1(t) = 0.17213 e-1.5t sin 5.8095t (2) C a2 s s = 1 s +1 6= = 1 1 − 18 18 s s 2 + 3 s + 36 2 + 3 s + 36 ss 1 1− 18 s 2 1 s + 3 + 0.083333 2 18 33.75 s +3 2 = 0.055556 1 − s 2 33.75 + 33.75 0.055556 s + 3 + 0.014344 33.75 2 s +3 2 2 + 33.75 Taking the inverse Laplace transform Ca2(t) = 0.055556 - e-1.5t (0.055556 cos 5.809t + 0.014344 sin 5.809t) The total response is found as follows: Cat(t) = Ca1(t) + Ca2(t) = 0.055556 - e-1.5t (0.055556 cos 5.809t - 0.157786 sin 5.809t) Plotting the total response: b. (1) Same as (1) from part (a), or Cb1(t) = Ca1(t) (2) Same as the negative of (2) of part (a), or Cb2 (t) = - Ca2(t) 102 Chapter 4: Time Response The total response is Cbt(t) = Cb1(t) + Cb2(t) = Ca1(t)- Ca2(t) = -0.055556 + e-1.5t (0.055556 cos 5.809t + 0.186474 sin 5.809t) Notice the nonminimum phase behavior for Cbt(t). Solutions to Problems 103 34. 10 Gain Unit Step 1 s 2 +3s+10 Transfer Fcn 1 10 Gain1 Unit Step1 s 2 +3s+10 Saturation 1 0.25 volts Gain2 Transfer Fcn1 Scope1 1 2 +3s+10 s 10 Unit Step2 Scope Saturation 2 0.25 volts Transfer Fcn2 Backlash Deadzone 0.02 Scope2 104 Chapter 4: Time Response 35. 1⎤ ⎡ 1 0 ⎤ ⎡ − 2 −1⎤ ⎡ (s + 2) sI − A = s⎢ −⎢ =⎢ (s + 5)⎥ ⎣ 0 1 ⎥ ⎣ − 3 − 5⎥ ⎣ 3 ⎦ ⎦ ⎦ sI − A = s + 7s + 7 2 Factoring yields poles at –5.7913 and –1.2087. Solutions to Problems 105 36. a. ⎡1 0 sI − A = s⎢0 1 ⎢0 0 ⎣ 0⎤ ⎡0 0⎥ − ⎢0 1⎥ ⎢1 ⎦⎣ 2 3⎤ ⎡ s −2 −3 ⎤ 6 5 ⎥ = ⎢ 0 (s − 6) −5 ⎥ 4 2 ⎥ ⎢−1 −4 (s − 2)⎥ ⎦⎣ ⎦ sI − A = s 3 − 8 s 2 − 11s + 8 b. Factoring yields poles at 9.111, 0.5338, and –1.6448. 37. x = (sI - A ) -1 (x0 + B u ) 38. x = (sI - A ) -1 (x0 + B u ) 39. x = (sI - A ) -1 (x0 + B u ) 106 Chapter 4: Time Response 40. −1 x = (sI − A) (x 0 + Bu) ⎛ ⎡ 1 0 0⎤ ⎡ −3 1 0 ⎤ ⎞ x = ⎜ s⎢ 0 1 0⎥ − ⎢ 0 − 6 1 ⎥ ⎟ ⎜ ⎟ ⎝ ⎢ 0 0 1⎥ ⎢ 0 0 − 5⎥ ⎠ ⎣ ⎦⎣ ⎦ 1 ⎡ ⎤ ⎢ s (s + 3)( s + 5) ⎥ 1 ⎥ x=⎢ ⎢ s( s + 5) ⎥ 1 ⎥ ⎢ ⎣ s( s + 5) ⎦ ⎡ 1 − 1 e −3 t + 1 e −5 t ⎤ 10 ⎢ 15 6 ⎥ 1 1 −5 t ⎥ x (t ) = ⎢ −e 55 ⎢ ⎥ 1 1 −5 t −e ⎢ ⎥ ⎦ ⎣ 55 y(t ) = [0 1 1]x = 41. Program: A=[-3 1 0;0 -6 1;0 0 -5]; B=[0;1;1]; C=[0 1 1]; D=0; S=ss(A,B,C,D) step(S) Computer response: a= x1 x2 x3 x1 -3 0 0 x2 1 -6 0 x3 0 1 -5 2 2 −5t −e 55 −1 ⎛ ⎡ 0⎤ ⎡ 0⎤ ⎞ ⎜ ⎢ 0⎥ + ⎢ 1⎥ 1 ⎟ ⎜ s⎟ ⎝ ⎢ 0⎥ ⎢ 1⎥ ⎠ ⎣⎦ ⎣⎦ Solutions to Problems 107 b= x1 x2 x3 u1 0 1 1 c= y1 x1 0 x2 1 x3 1 d= y1 u1 0 Continuous-time model. 42. Program: syms s 'a' A=[-3 1 0;0 -6 1;0 0 -5] B=[0;1;1]; C=[0 1 1]; X0=[1;1;0] U=1/s; I=[1 0 0;0 1 0;0 0 1]; X=((s*I-A)^-1)*(X0+B*U); x1=ilaplace(X(1)) x2=ilaplace(X(2)) x3=ilaplace(X(3)) y=C*[x1;x2;x3] y=simplify(y) 'y(t)' pretty(y) Computer response: ans = a %Construct symbolic object for %frequency variable 's'. %Display label %Create matrix A. %Create vector B. %Create C vector %Create initial condition vector,X(0). %Create U(s). %Create identity matrix. %Find Laplace transform of state vector. %Solve for X1(t). %Solve for X2(t). %Solve for X3(t). %Solve for output, y(t). %Simplify y(t). %Display label. %Pretty print y(t). 108 Chapter 4: Time Response A= -3 0 0 1 -6 0 0 1 -5 X0 = 1 1 0 x1 = 7/6*exp(-3*t)-1/3*exp(-6*t)+1/15+1/10*exp(-5*t) x2 = exp(-6*t)+1/5-1/5*exp(-5*t) x3 = 1/5-1/5*exp(-5*t) y= 2/5+exp(-6*t)-2/5*exp(-5*t) y= 2/5+exp(-6*t)-2/5*exp(-5*t) ans = y(t) 2/5 + exp(-6 t) - 2/5 exp(-5 t) 43. |λI - A | = λ2 + 5λ +1 |λI - A | = (λ + 0.20871) (λ + 4.7913) Therefore, Solutions to Problems 109 Solving for Ai's two at a time, and substituting into the state-transition matrix To find x(t), To find the output, 44. |λI - A | = λ2 + 1 Solving for the Ai's and substituting into the state-transition matrix, To find the state vector, 110 Chapter 4: Time Response 45. |λI - A | = (λ + 2) (λ + 0.5 - 2.3979i) (λ + 0.5 + 2.3979i) Let the state-transition matrix be . .. Since φ(0) = I, Φ(0) = A, and φ(0) = A2, we can evaluate the coefficients, Ai's. Thus, Solving for the Ai's taking three equations at a time, Solutions to Problems 111 U s i n g x (t ) = φ (t ) x (0 ) + t ∫ φ (t - τ ) B u ( τ )d τ , a n d y = 1 0 0 x (t ), 0 11 = 2 - 2 e-2t 46. Program: syms s t tau 'a' A=[-2 1 0;0 0 1;0 -6 -1] B=[1;0;0] C=[1 0 0] X0=[1;1;0] I=[1 0 0;0 1 0;0 0 1]; 'E=(s*I-A)^-1' E=((s*I-A)^-1) %Construct symbolic object for %frequency variable 's', 't', and 'tau. %Display label. %Create matrix A. %Create vector B. %Create vector C. %Create initial condition vector,X(0). %Create identity matrix. %Display label. %Find Laplace transform of state %transition matrix, (sI-A)^-1. %Take inverse Laplace transform %of each element Fi11=ilaplace(E(1,1)); Fi12=ilaplace(E(1,2)); Fi13=ilaplace(E(1,3)); Fi21=ilaplace(E(2,1)); Fi22=ilaplace(E(2,2)); Fi23=ilaplace(E(2,3)); Fi31=ilaplace(E(3,1)); Fi32=ilaplace(E(3,2)); %to find state transition matrix. Fi33=ilaplace(E(3,3)); %of (sI-A)^-1. 'Fi(t)' %Display label. Fi=[Fi11 Fi12 Fi13 %Form Fi(t). Fi21 Fi22 Fi23 Fi31 Fi32 Fi33]; pretty(Fi) %Pretty print state transition matrix, Fi. Fitmtau=subs(Fi,t,t-tau); %Form Fi(t-tau). 'Fi(t-tau)' %Display label. pretty(Fitmtau) %Pretty print Fi(t-tau). x=Fi*X0+int(Fitmtau*B*1,tau,0,t); %Solve for x(t). x=simple(x); %Collect terms. x=simplify(x); %Simplify x(t). x=vpa(x,3); 'x(t)' %Display label. pretty(x) %Pretty print x(t). y=C*x; %Find y(t) y=simplify(y); y=vpa(simple(y),3); y=collect(y); 'y(t)' pretty(y) %Pretty print y(t). 112 Chapter 4: Time Response Computer response: ans = a A= -2 0 0 1 0 -6 0 1 -1 0 0 B= 1 0 0 C= 1 X0 = 1 1 0 ans = E=(s*I-A)^-1 E= [ [ [ 1/(s+2), (s+1)/(s+2)/(s^2+s+6), 0, (s+1)/(s^2+s+6), 0, -6/(s^2+s+6), 1/(s+2)/(s^2+s+6)] 1/(s^2+s+6)] s/(s^2+s+6)] ans = Fi(t) [ 13 [exp(-2 t) , - 1/8 exp(-2 t) + 1/8 %1 + --- %2 , [ 184 1/8 exp(-2 t) - 1/8 %1 + 3/184 %2] [ [0 , 1/23 %2 + %1 , - 1/23 1/2 1/2 1/2 (-23) (exp((-1/2 + 1/2 (-23) ) t) - exp((-1/2 - 1/2 (-23) ) t)) [ [0 , 6/23 1/2 1/2 1/2 (-23) (exp((-1/2 + 1/2 (-23) ) t) - exp((-1/2 - 1/2 (-23) ) t)) Solutions to Problems 113 , - 1/23 %2 + %1] 1/2 %1 := exp(- 1/2 t) cos(1/2 23 t) 1/2 1/2 %2 := exp(- 1/2 t) 23 sin(1/2 23 t) ans = Fi(t-tau) [ [exp(-2 t + 2 tau) , [ 13 1/2 - 1/8 exp(-2 t + 2 tau) + 1/8 %2 cos(%1) + --- %2 23 sin(%1) , 184 1/2 1/8 exp(-2 t + 2 tau) - 1/8 %2 cos(%1) + 3/184 %2 23 sin(%1)] [ 1/2 1/2 [0 , 1/23 %2 23 sin(%1) + %2 cos(%1) , - 1/23 (-23) ( 1/2 exp((-1/2 + 1/2 (-23) ) (t - tau)) 1/2 - exp((-1/2 - 1/2 (-23) ) (t - tau)))] [ 1/2 1/2 [0 , 6/23 (-23) (exp((-1/2 + 1/2 (-23) ) (t - tau)) 1/2 - exp((-1/2 - 1/2 (-23) ) (t - tau))) , 1/2 - 1/23 %2 23 sin(%1) + %2 cos(%1)] 1/2 %1 := 1/2 23 (t - tau) %2 := exp(- 1/2 t + 1/2 tau) ans = x(t) [.375 exp(-2. t) + .125 exp(-.500 t) cos(2.40 t) + .339 exp(-.500 t) sin(2.40 t) + .500] [.209 exp(-.500 t) sin(2.40 t) + exp(-.500 t) cos(2.40 t)] [1.25 i (exp((-.500 + 2.40 i) t) - 1. exp((-.500 - 2.40 i) t))] ans = y(t) .375 exp(-2. t) + .125 exp(-.500 t) cos(2.40 t) 114 Chapter 4: Time Response + .339 exp(-.500 t) sin(2.40 t) + .500 47. The state-space representation used to obtain the plot is, . 0 -1 x= 1 -0 . 8 x+ 0 1 u (t); y (t) = 1 0 x Using the Step Response software, Calculating % overshoot, settling time, and peak time, 2ζωn = 0.8, ωn = 1, ζ = 0.4. Therefore, Tp = π ωn 1-ζ2 = 3.43 sec. %OS = e −ζπ / 1 −ζ 2 4 x100 = 25.38% , Ts = ζω = 10 sec, n Solutions to Problems 115 48. 49. s+0.5 1/20 1/4 3/10 1 1 3 a. P(s) = s(s+2)(s+5) = s + s+2 - s+5 . Therefore, p(t) = 20 + 4 e-2t - 10 e-5t. b. To represent the system in state space, draw the following block diagram. 116 Chapter 4: Time Response V(s) Y(s) 1 s +7s+ 10 For the first block, .. P(s) s+ 0. 5 2 . y + 7 y + 1 0 y = v (t ) . Let x1 = y, and x2 = y . Therefore, . x 1 = x2 . x 2 = -10x1 - 7x2 + v(t) Also, . p(t) = 0.5y + y = 0.5x1 + x2 Thus, . x= 0 1 -1 0 -7 c. Program: A=[0 1;-10 -7]; B=[0;1]; C=[.5 1]; D=0; S=ss(A,B,C,D) step(S) Computer response: a= x1 x2 x1 0 -10 x2 1 -7 b= x1 x2 u1 0 1 c= y1 x1 0.5 x2 1 d= y1 u1 0 Continuous-time model. x+ 0 1 1 ; p (t ) = 0.5 1 x Solutions to Problems 117 50. a. ωn = 10 = 3.16; 2ζωn = 4. Therefore ζ = 0.632. Ts = 4 ξω n = 2 seconds. T p = π ω n 1− ξ 2 %OS = e − ξπ / 1 −ξ 2 * 100 = 7.69%. = 1.28 seconds. From Figure 4.16, Trωn = 1.93. Thus, Tr = 0.611 second. To justify second-order assumption, we see that the dominant poles are at – 2 ± j2.449. The third pole is at -10, or 5 times further. The second-order approximation is valid. b. Ge(s) = K K = . Representing the system in phase-variable form: (s+10)(s2+4s+10) s3+14s2+50s+100 ⎡0 A=⎢ 0 ⎢ −100 ⎣ c. Program: numg=100; deng=conv([1 10],[1 4 10]); G=tf(numg,deng) step(G) Computer response: Transfer function: 100 ------------------------s^3 + 14 s^2 + 50 s + 100 1 0 −50 0⎤ ⎡0⎤ 1 ⎥; B = ⎢ 0 ⎥; C = [1 0 0 ] ⎢K ⎥ − 14⎥ ⎦ ⎣⎦ 118 Chapter 4: Time Response %OS = (1.08-1) * 100 = 8% 1 51. a. ωn = 0.28 = 0.529; 2ζωn = 1.15. Therefore ζ = 1.087. 7.63x10-2 2 0.545 b. P(s) = U(s) s2+1.15s+0.28 , where U(s) = s . Expanding by partial fractions, P(s) = s + natural response terms. Thus percent paralysis = 54.5%. 7.63x10-2 0.2725 0.48444 0.21194 c. P(s) = - s+0.35 + s+0.8 . 2+1.15s+0.28) = s s(s Hence, p(t) = 0.2725 - 0.48444e-0.35t + 0.21194e-0.8t. Plotting, for 1% isoflurane Fractional paralysis Solutions to Problems 119 K 7.63x10-2 1 7.63x10-2 K d. P(s) = s * 2 = s + natural response terms. Therefore, = 1. Solving 0.28 s +1.15s+0.28 for K, K = 3.67%. 52. a. Writing the differential equation, dc(t) i(t) dt = -k10c(t) + Vd Taking the Laplace transform and rearranging, I(s) (s+k10)C(s) = V d from which the transfer function is found to be 1 Vd C(s) I(s) = s+k10 I0 For a step input, I(s) = s . Thus the response is I0 Vd I0 1 1 C(s) = s(s+k ) = k V ( s - s+k ) 10 10 d 10 Taking the inverse Laplace transform, c (t) = I0 (1 − e −k 10 t ) k 10V d where the steady-state value, CD, is I0 CD = k V 10 d Solving for IR = I0, IR = CDk10Vd 2.2 4 b. Tr = k ; Ts = k 10 10 µg mg c. IR = CDk10Vd = 12 ml x 0.07 hr-1 x 0.6 liters = 0.504 h d. Using the equations of part b, where k10 = 0.07, Tr = 31.43 hrs, and Ts = 57.14 hrs. 120 Chapter 4: Time Response SOLUTIONS TO DESIGN PROBLEMS 53. Writing the equation of motion, ( f vs + 2) X (s ) = F (s ) . Thus, the transfer function is X (s) 1 / fv 44 Ts = . Hence, Ts = = 2 2 = 2 fv , or f v = 2 . F (s) s + a fv fv 54. The transfer function is, M= F( s) = 1/ M 4 4 1 K . Now, Ts = 2 = Re = 1 = 8M . Thus, s2 + s + M M 2M 1 . Substituting the value of M in the denominator of the transfer function yields, 4 s 2 + 4s + 4 K . Identify the roots s1,2 = −2 ± j 2 K − 1 . Using the imaginary part and π π substituting into the peak time equation yields Tp = 1 = , from which = Im 2 K − 1 K = 3.467 . 55. ( Ms2 + f vs + 1) X (s) = F (s) . Thus, the transfer function is X (s ) 1/ M fv 4 = fv 1 . Since Ts = 10 = ζω n , ζω n = 0.4 . But, M = 2ζω n = 0.8. Also, F (s ) s 2 + s + M M Writing the equation of motion, from Eq. (4.39) 30% overshoot implies ζ = 0.358. Hence, ωn = 1.117. Now, 1/M = ωn2 = 1.248. Therefore, M = 0.801. Since fv = 2ζω n = 0.8, fv = 0.641 . M 56. Writing the equation of motion: (Js2+s+K)θ(s) = T(s). Therefore the transfer function is 1 J θ(s) T(s) = 2 1 K . s + J s+ J ζ= %OS - ln ( 100 ) %OS π2 + ln2 ( 100 ) Ts = = 0.358. 4 4 = 1 = 8J = 4. ζωn 2J 4 4 K 1 = . Hence, ωn = 2.793. Now, J = ωn2 = 7.803. Therefore J = 2 . Also, Ts = 4 = ζωn (0.358)ωn Finally, K = 3.901. 57. Writing the equation of motion Solutions to Design Problems 121 1 [s2+D(5)2s+4 (10) 2]θ(s) = T(s) The transfer function is θ (s) 1 T(s) = s2+25Ds+25 Also, ζ= %OS - ln ( 100 ) %OS π2 + ln2 ( 100 ) = 0.358 and 2ζωn = 2(0.358)(5) = 25D Therefore D = 0.14. 58. The equivalent circuit is: N1 N1 N1 where Jeq = 1+(N )2 ; Deq = (N )2; Keq = (N )2. Thus, 2 2 2 θ1(s) N1 1 T(s) = Jeqs2+Deqs+Keq . Letting N2 = n and substituting the above values into the transfer function, θ1(s) T(s) = 1 1+n2 s2 + n2 1+n2 s+ n2 . Therefore, ζωn = n2 4 8(1+n2) = 16. Thus 2) . Finally, Ts = ζωn = n2 2(1+n 1+n2 n = 1. 59. Let the rotation of the shaft with gear N2 be θL(s). Assuming that all rotating load has been reflected to the N2 shaft, (J eqL s 2 + DeqL s + K ) L (s) + F (s)r = Teq (s) , where F(s) is the force from the θ translational system, r = 2 is the radius of the rotational member, JeqL is the equivalent inertia at the N2 shaft, and DeqL is the equivalent damping at the N2 shaft. Since JeqL = 1(2)2 + 1 = 5 and DeqL = 122 Chapter 4: Time Response ( 1(2)2 = 4, the equation of motion becomes, 5s + 4s + K 2 )θ L ( s ) + 2 F (s ) = Teq (s ) . For the translational system ( Ms + s ) X ( s ) = F ( s) . Substituting F(s) into the rotational equation of 2 ( motion, 5s + 4s + K 2 )θ ( s ) + (Ms + s )2 X ( s ) = Teq ( s ) . 2 L X ( s ) X (s ) = and Teq ( s) = 2T (s ). Substituting these quantities in the equation r 2 X ( s) 2 above yields ((5 + 4M )s + 8s + K ) = T (s ) . Thus, the transfer function is 4 X (s ) 4 /( 5 + 4 M ) 4 4 = = = (5 + 4 M ) . . Now, Ts = 10 = 8 K 8 T( s) s2 + Re s+ (5 + 4 M) (5 + 4M ) 2(5 + 4 M ) But,θ L (s ) = Hence, M = 5/4. For 10% overshoot, ζ = 0.5912 from Eq. (4.39). Hence, 2ζω n = ωn = 8 = 0.8 . Solving for ωn yields ωn = 0.6766. But, (5 + 4 M ) K K = = 0.6766. Thus, K = 4.578. (5 + 4M ) 10 60. VC(s) The transfer function for the capacitor voltage is V(s) = For 20% overshoot, ζ = %OS - ln ( 100 ) %OS π2 + ln2 ( ) 1 Cs 106 =2 6. 1 R+Ls+Cs s +Rs+10 = 0.456. Therefore, 2ζωn = R = 2(0.456)(103) = 100 912Ω. 61. Solving for the capacitor voltage using voltage division, VC (s ) = Vi (s ) 1/(CS ) 1 . Thus, the CS 1/( LC) VC (s) 4 R transfer function is = . Since Ts = = 10 −3 , Re = = 4000 . Thus Vi (s ) s2 + R s + 1 Re 2L L LC R = 8 KΩ . Also, since 20% overshoot implies a damping ratio of 0.46 and 1 2ζω n = 8000, ω n = 8695.65 = . Hence, C = 0.013 µF . LC 62. Using voltage division the transfer function is, 1 VC (s) Cs = = Vi (s) R + Ls + 1 s2 + Cs 1 LC 1 R s+ LC L R + LS + Solutions to Design Problems 123 4 4 8L R =R= . Thus, = 4000 . Using Eq. (4.39) with 15% overshoot, Re R L 2L 1 1 ζ = 0.5169. But, 2ζωn = R/L. Thus, ω n = 3869 = . Therefore, L = 6.7 mH and = LC L (10 −5 ) Also, Ts = 2 x10 −3 = R = 26.72 Ω. 63. For the circuit shown below L= R1 = o i2(t) i1(t) write the loop equations as R 1+L s I 1 s −R 1 I 2 s −R 1 I 1 s + R 1+R 2+ 1 Cs =V i s I2 s =0 Solving for I2(s) I2 s = R 1 +L s −R 1 R 1 +L s −R 1 But, V o s = V i (s ) 0 −R 1 R 1 +R 2 + 1 Cs 1 I 2 s . Thus, Cs Vo s R1 = 2+ C R R +L s +R Vi s R 2+R 1 C L s 21 1 Substituting component values, 1 Vo (s) ( R2 + 1000000)C = 1000000 (1000000CR2 + 1) 1 Vi (s ) s2 + s + 1000000 ( R2 + 1000000)C ( R2 + 1000000)C 4 For 15% overshoot, ζ = 0.517. For Ts = 0.001, ζωn = 0.001 = 4000. Hence, ωn = 7736.9. Thus, 1000000 or, 1 = 7736.92 R 2 + 1000000 C 124 Chapter 4: Time Response 1 R 2 + 1000000 (1) 1000000 C R 2 + 1 = 8000 R 2 + 1000000 C (2) C = 0.016706 Also, Solving (1) and (2) simultaneously, R 2 = 8003.7 Ω, and C = 1.6573 x 10-2 µF. 64. ⎡ s 0 ⎤ ⎡ (3.45 − 14000 Kc ) sI − A = ⎢ − 11 ⎣ 0 s ⎥ ⎢ 0.499 x10 ⎦⎣ −0.255 x10−9 ⎤ − 3.68 ⎥ ⎦ ⎡ s − (3.45 − 14000 Kc ) 0.255x10 −9 ⎤ =⎢ −0.499 x1011 s + 3.68 ⎥ ⎣ ⎦ 2 5 sI − A = s + (0.23 + 0.14x10 Kc )s + (51520 Kc + 0.0285) (2ζω n )2 = [2 * 0.9]2 * ( 51520K c + 0.0285) = (0.23 + 0.14 x105 Kc )2 or −4 Kc − 8.187x10 Kc − 2.0122 x10 2 −10 =0 Solving for Kc, Kc = 8.189 x10 −4 65. a. The transfer function from Chapter 2 is, 0.7883( s + 53.85) Yh ( s) − Ycat (s) =2 (s + 15.47s + 9283)(s2 + 8.119 s + 376.3) Fup ( s) The dominant poles come from s 2 + 8.119s + 376.3 . Using this polynomial, 2 2ζω n = 8.119, and ω n = 376.3 . Thus, ω n = 19.4 and ζ = 0.209 . Using Eq. (4.38), %OS = 51.05%. Also, Ts = 4 ζω n = 0.985 s, and Tp = Figure 4.16. Thus, ω n Tr = 1.2136 or π ωn 1− ζ 2 = 0.166 s . To find rise time, use Tr = 0.0626 s. b. The other poles have a real part of 15.47/2 = 7.735. Dominant poles have a real part of 8.119/2 = 4.06. Thus, 7.735/4.06 = 1.91. This is not at least 5 times. c. Program: syms s numg=0.7883*(s+53.85); deng=(s^2+15.47*s+9283)*(s^2+8.119*s+376.3); 'G(s) transfer function' G=vpa(numg/deng,3); pretty(G) numg=sym2poly(numg); deng=sym2poly(deng); G=tf(numg,deng) Solutions to Design Problems 125 step(G) Computer response: ans = G(s) transfer function .788 s + 42.4 -----------------------------------------2 2 (s + 15.5 s + 9280.) (s + 8.12 s + 376.) Transfer function: 0.7883 s + 42.45 ---------------------------------------------------s^4 + 23.59 s^3 + 9785 s^2 + 8.119e004 s + 3.493e006 The time response shows 58 percent overshoot, Ts = 0.86 s, Tp = 0.13 s, Tr = 0.05 s. FIVE Reduction of Multiple Subsystems SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Designing a Closed-Loop Response a. Drawing the block diagram of the system: Pots ui + Pre amp 10 Π Motor, load and gears Power amp 1 50 s+150 K 0.16 s (s+1.32) - Thus, T(s) = uo 76.39K s3+151.32s2+198s+76.39K b. Drawing the signal flow-diagram for each subsystem and then interconnecting them yields: ui 10 Π K pot pre amp 1 s x3 150 0.8 1 s x2 1 s x1 0.2 gears -1.32 -150 power amp - 10 Π pot motor and load uo Solutions to Case Studies Challenges 127 . x 1 = x2 . x 2 = - 1.32x2 + 0.8x3 . 10 x 3 = -150x3 +150K( (q i − 0.2x1 )) = -95.49Kx1 - 150x3 + 477.46Kθi π θo = 0.2x1 In vector-matrix notation, . x= 0 1 0 -95. 49K θo = 0 . 2 0 0 -1. 32 0 0 0. 8 -150 0 x+ 0 477. 46K θ i x c. T1 = 76.39 ⎛ 10 ⎞ ⎛ 1⎞ ⎛ 1⎞ ⎛ 1⎞ ⎝ π ⎠ ( K )(150)⎝ s ⎠ (0.8)⎝ s⎠ ⎝ s ⎠ (0.2) = s 3 G L1 = −150 − 1.32 ⎛1 ⎛1 ⎛1 ⎛ −10 ⎞ −76.39 K ; GL 2 = ; GL 3 = (K )(150) ⎞ (0.8) ⎞ ⎞ (0.2 ) ⎝ s⎠ ⎝ s ⎠ ⎝ s⎠ ⎝ π ⎠= s s s3 Nontouching loops: 198 GL1GL2 = 2 s ∆ = 1 - [GL1 + GL2 + GL3] + [GL1GL2] = 1 + 150 1.32 76.39K 198 +2 s+s+ s3 s ∆1 = 1 T(s) = T1∆1 76.39K =3 ∆ s +151.32s2 +198s+76.39K 10 π 0.16K d. The equivalent forward path transfer function is G(s) = s(s+1.32) . Therefore, 2.55 T(s) = 2 s +1.32s+2.55 The poles are located at -0.66 ± j1.454. ωn = 2.55 = 1.597 rad/s; 2ζωn = 1.32, therefore, ζ = 0.413. 128 Chapter 5: Reduction of Multiple Subsystems %OS = e −ζπ / 1 −ζ 2 4 x100 = 24% ; Ts = ζω n 4 = 0.66 = 6.06 seconds; Tp = π π = 1.454 = ωn 1-ζ2 2.16 seconds; Using Figure 4.16, the normalized rise time is 1.486. Dividing by the natural frequency, 1.486 Tr = = 0.93 seconds. 2.55 e. %OS - ln ( 100 ) = 0.517 for 15% 2 + ln2 (%OS) π 100 1.32 1.32 overshoot; ωn = 0.51K ; and 2ζωn = 1.32. Therefore, ωn = = 2(0.5147) = 1.277 = 0.51K . 2ζ 0.51K 0.51K f. Since G(s) = s(s+1.32) , T(s) = 2 . Also, ζ = s +1.32s+0.51K Solving for K, K=3.2. UFSS Vehicle: Pitch-Angle Control Representation a. Use the observer canonical form for the vehicle dynamics so that the output yaw rate is a state variable. u 1 1 -1 2 1 s -0.125 1 s 0.437 x4 1 1 s x3 x2 -1.483 -2 -0.24897 1 -1 b. Using the signal flow graph to write the state equations: 1 s 1 y x1 Solutions to Case Studies Challenges 129 x1 = x2 Ý Ý x2 = − 1. 483x 2 + x 3 − 0.125 x4 x3 = − 0. 24897 x2 − (0.125 * 0. 437) x4 Ý Ý x4 = 2 x1 + 2 x 2 − 2 x4 − 2u In vector-matrix form: 1 ⎡0 ⎢ 0 −1. 483 Ý x= ⎢ 0 − 0. 24897 ⎢2 2 ⎣ 0 0 ⎡0⎤ ⎤ 1 −0.125 ⎥ ⎢0⎥ x+ u ⎢0⎥ 0 − 0. 054625⎥ ⎢ −2 ⎥ ⎥ 0 −2 ⎣⎦ ⎦ y = [1 0 0 0]x c. Program: numg1=-0.25*[1 0.437]; deng1=poly([-2 -1.29 -0.193 0]); 'G(s)' G=tf(numg1,deng1) numh1=[-1 0]; denh1=[0 1]; 'H(s)' H=tf(numh1,denh1) 'Ge(s)' Ge=feedback(G,H) 'T(s)' T=feedback(-1*Ge,1) [numt,dent]=tfdata(T,'V'); [Acc,Bcc,Ccc,Dcc]=tf2ss(numt,dent) Computer response: ans = G(s) Transfer function: -0.25 s - 0.1093 -------------------------------------s^4 + 3.483 s^3 + 3.215 s^2 + 0.4979 s ans = H(s) Transfer function: -s ans = Ge(s) Transfer function: -0.25 s - 0.1093 -------------------------------------s^4 + 3.483 s^3 + 3.465 s^2 + 0.6072 s 130 Chapter 5: Reduction of Multiple Subsystems ans = T(s) Transfer function: 0.25 s + 0.1093 ----------------------------------------------s^4 + 3.483 s^3 + 3.465 s^2 + 0.8572 s + 0.1093 Acc = -3.4830 1.0000 0 0 -3.4650 0 1.0000 0 -0.8572 0 0 1.0000 -0.1093 0 0 0 0 0.2500 0.1093 Bcc = 1 0 0 0 Ccc = 0 Dcc = 0 ANSWERS TO REVIEW QUESTIONS 1. Signals, systems, summing junctions, pickoff points 2. Cascade, parallel, feedback 3. Product of individual transfer functions, sum of individual transfer functions, forward gain divided by one plus the product of the forward gain times the feedback gain 4. Equivalent forms for moving blocks across summing junctions and pickoff points 5. As K is varied from 0 to ∞, the system goes from overdamped to critically damped to underdamped. When the system is underdamped, the settling time remains constant. 6. Since the real part remains constant and the imaginary part increases, the radial distance from the origin is increasing. Thus the angle θ is increasing. Since ζ= cos θ the damping ratio is decreasing. 7. Nodes (signals), branches (systems) 8. Signals flowing into a node are added together. Signals flowing out of a node are the sum of signals flowing into a node. 9. One 10. Phase-variable form, cascaded form, parallel form, Jordan canonical form, observer canonical form 11. The Jordan canonical form and the parallel form result from a partial fraction expansion. 12. Parallel form Answers to Review Questions 131 13. The system poles, or eigenvalues 14. The system poles including all repetitions of the repeated roots 15. Solution of the state variables are achieved through decoupled equations. i.e. the equations are solvable individually and not simultaneously. 16. State variables can be identified with physical parameters; ease of solution of some representations 17. Systems with zeros 18. State-vector transformations are the transformation of the state vector from one basis system to another. i.e. the same vector represented in another basis. 19. A vector which under a matrix transformation is collinear with the original. In other words, the length of the vector has changed, but not its angle. 20. An eigenvalue is that multiple of the original vector that is the transformed vector. 21. Resulting system matrix is diagonal. SOLUTIONS TO PROBLEMS 1. a. Combine the inner feedback and the parallel pair. Multiply the blocks in the forward path and apply the feedback formula to get, 50(s-2) T(s) = 3 2 . s +s +150s-100 b. Program: 'G1(s)' G1=tf(1,[1 0 0]) 'G2(s)' G2=tf(50,[1 1]) 'G3(s)' G3=tf(2,[1 0]) 'G4(s)' G4=tf([1 0],1) 'G5(s)' G5=2 'Ge1(s)=G2(s)/(1+G2(s)G3(s))' Ge1=G2/(1+G2*G3) 'Ge2(s)=G4(s)-G5(s)' Ge2=G4-G5 'Ge3(s)=G1(s)Ge1(s)Ge2(s)' Ge3=G1*Ge1*Ge2 132 Chapter 5: Reduction of Multiple Subsystems 'T(s)=Ge3(s)/(1+Ge3(s))' T=feedback(Ge3,1); T=minreal(T) Computer response: ans = G1(s) Transfer function: 1 --s^2 ans = G2(s) Transfer function: 50 ----s+1 ans = G3(s) Transfer function: 2 s ans = G4(s) Transfer function: s ans = G5(s) G5 = 2 ans = Ge1(s)=G2(s)/(1+G2(s)G3(s)) Transfer function: 50 s^2 + 50 s ------------------------s^3 + 2 s^2 + 101 s + 100 ans = Ge2(s)=G4(s)-G5(s) Transfer function: s-2 Solutions to Problems ans = Ge3(s)=G1(s)Ge1(s)Ge2(s) Transfer function: 50 s^3 - 50 s^2 - 100 s ------------------------------s^5 + 2 s^4 + 101 s^3 + 100 s^2 ans = T(s)=Ge3(s)/(1+Ge3(s)) Transfer function: 50 s - 100 ----------------------s^3 + s^2 + 150 s - 100 2. Push G1(s) to the left past the pickoff point. H1 + G1 + G2 G3 + 1 G1 ⎛ G1 ⎞ ⎛ 1⎞ (G G + 1)G3 ⎟ ⎜ G2 + ⎟ G3 = 1 2 ⎟⎜ ⎟ G1 ⎠ (1 + G1 H1 ) ⎝ 1 + G1 H1 ⎠ ⎝ Thus, T (s ) = ⎜ ⎜ 3. a. Split G3 and combine with G2 and G4. Also use feedback formula on G6 loop. 133 134 Chapter 5: Reduction of Multiple Subsystems Push G2 +G3 to the left past the pickoff point. Using the feedback formula and combining parallel blocks, Multiplying the blocks of the forward path and applying the feedback formula, Solutions to Problems 4. Push G2(s) to the left past the summing junction. Collapse the summing junctions and add the parallel transfer functions. Push G1(s)G2(s) + G5(s) to the right past the summing junction. 135 136 Chapter 5: Reduction of Multiple Subsystems Collapse summing junctions and add feedback paths. Applying the feedback formula, T ( s) = = G3 ( s ) + G1 ( s )G2 ( s ) ⎡ ⎤ G2 ( s )G4 ( s ) 1 + [G3 ( s ) + G1 ( s )G2 ( s )] ⎢ H + G3 ( s ) + G1 ( s )G2 ( s ) ⎥ ⎣ ⎦ G3 ( s ) + G1 ( s )G2 ( s ) 1 + H [G3 ( s ) + G1 ( s )G2 ( s )] + G2 ( s )G4 ( s ) 5. a. Push G7 to the left past the pickoff point. Add the parallel blocks, G3+G4. Solutions to Problems Push G3+G4 to the right past the summing junction. Collapse the minor loop feedback. 137 138 Chapter 5: Reduction of Multiple Subsystems G7(G3+G4) Push 1+G G to the left past the pickoff point. 67 Push G1 to the right past the summing junction. Add the parallel feedback paths to get the single negative feedback, G5 G2(1+G6G7) G8 H(s) = G + G (G +G ) - G . Thus, 7 734 1 Solutions to Problems T(s) = b. Program: G1=tf([0 1],[1 7]); %G1=1/s+7 input transducer G2=tf([0 0 1],[1 2 3]); %G2=1/s^2+2s+3 G3=tf([0 1],[1 4]); %G3=1/s+4 G4=tf([0 1],[1 0]); %G4=1/s G5=tf([0 5],[1 7]); %G5=5/s+7 G6=tf([0 0 1],[1 5 10]); %G6=1/s^2+5s+10 G7=tf([0 3],[1 2]); %G7=3/s+2 G8=tf([0 1],[1 6]); %G8=1/s+6 G9=tf([1],[1]); %Add G9=1 transducer at the input T1=append(G1,G2,G3,G4,G5,G6,G7,G8,G9); Q=[1 -2 -5 9 2180 3180 4180 5 3 4 -6 6700 7 3 4 -6 8 7 0 0]; inputs=9; outputs=7; Ts=connect(T1,Q,inputs,outputs); T=tf(Ts) 139 140 Chapter 5: Reduction of Multiple Subsystems Computer response: Transfer function: 6 s^7 + 132 s^6 + 1176 s^5 + 5640 s^4 + 1.624e004 s^3 + 2.857e004 s^2 + 2.988e004 s + 1.512e004 ----------------------------------------------------------s^10 + 33 s^9 + 466 s^8 + 3720 s^7 + 1.867e004 s^6 + 6.182e004 s^5 + 1.369e005 s^4 + 1.981e005 s^3 + 1.729e005 s^2 + 6.737e004 s - 1.044e004 6. Combine G6 and G7 yielding G6G7. Add G4 and obtain the following diagram: Next combine G3 and G4+G6G7. Push G5 to the left past the pickoff point. Solutions to Problems 141 G2 Notice that the feedback is in parallel form. Thus the equivalent feedback, Heq(s) = G + 5 G3(G4+G6G7) + G8. Since the forward path transfer function is G(s) = Geq(s) = G1G5, the closedloop transfer function is Geq(s) T(s) = 1+G (s)H (s) . eq eq Hence, 7. Push 2s to the right past the pickoff point. 142 Chapter 5: Reduction of Multiple Subsystems Combine summing junctions. Combine parallel 2s and s. Apply feedback formula to unity feedback with G(s) = s. Combine cascade pair and add feedback around 1/(s+1). Solutions to Problems Combine parallel pair and feedback in forward path. 5s2 + 2s Combine cascade pair and apply final feedback formula yielding T (s ) = . 6 s2 + 9s + 6 8. Push G3 to the left past the pickoff point. Push G6 to the left past the pickoff point. 143 144 Chapter 5: Reduction of Multiple Subsystems Hence, Thus the transfer function is the product of the functions, or θ22(s) G1G2G4G5G6G7 =1-G G +G G G +G G G -G G G G G +G G G G G G θ11(s) 45 456 123 12345 123456 9. Combine the feedback with G6 and combine the parallel G2 and G3. Move G2+G3 to the left past the pickoff point. Solutions to Problems 145 Combine feedback and parallel pair in the forward path yielding an equivalent forward-path transfer function of G4 ⎞ ⎛ G6 ⎞ ⎛ G2+G3 ⎞ ⎛ Ge(s) =⎜1+G (G +G )⎟ ⎜G5+G +G ⎟ ⎜1+G ⎟ 2 3⎠ ⎝ 6⎠ ⎝ 1 2 3 ⎠⎝ Ge(s) But, T(s) = 1+G (s)G (s) . Thus, e 7 10. Push G3(s) to the left past the pickoff point. 146 Chapter 5: Reduction of Multiple Subsystems Push G2(s)G3(s) to the left past the pickoff point. Push G1(s) to the right past the summing junction. Solutions to Problems Collapsing the summing junctions and adding the feedback transfer functions, T (s ) = G1 ( s)G2 ( s)G3 ( s ) 1 + G1 (s )G2 (s )G 3 (s) Heq ( s ) where Heq ( s) = 11. T (s ) = H (s ) H2 ( s) H1 (s) H3 (s ) +1 +4 + + G3 (s ) G2 ( s)G 3 (s) G1 ( s )G3 ( s ) G1 (s) 225 . Therefore, 2ζωn = 12, and ωn = 15. Hence, ζ = 0.4. s + 12 s + 225 2 %OS = e −ζπ / 1 −ζ 2 π 4 x100 = 16.3% ; Ts = ζω =0.667; Tp = n 12. C( s ) = ωn 1-ζ2 =0.229. 5 A Bs + C = +2 s(s + 3s + 5) s s + 3s + 5 2 A=1 5 = s2 + 3s + 5 + Bs 2 + Cs ∴ B = −1, C = −3 1 s +3 1 s+3 C( s ) = − 2 =− s s + 3s + 5 s (s + 1.5) 2 + 2.75 1 (s + 1.5) + 1.5 1 =− =− 2 s (s + 1.5) + 2.75 s c (t ) = 1 − e −1. 5t (cos 2.75t + 1.5 2.75 2.75 (s + 1.5)2 + 2.75 ( s + 1.5) + 1.5 sin 2.75t ) 2.75 13. Push 2s to the left past the pickoff point and combine the parallel combination of 2 and 1/s. 147 148 Chapter 5: Reduction of Multiple Subsystems Push (2s+1)/s to the right past the summing junction and combine summing junctions. Hence, T (s) = 14. Since G (s ) = 2(2s + 1) s 5 , where Heq ( s ) = 1 + +. 1 + 2(2 s + 1)Heq (s) 2s + 1 2 s K K G(s ) , T (s ) = =2 . Therefore, 2ζωn = 30. Thus, ζ = s(s + 30) 1 + G(s) s + 30 s + K 15/ωn = 0.456 (i.e. 20% overshoot). Hence, ωn = 32.89 = K . Therefore K = 1082. 15. %OS ) 4 K 100 = 0.358 ; Ts = ;ζ = = 0.2 . Therefore, ωn = T (s ) = 2 ζω n s + αs + K 2 2 %OS π + ln ( ) 100 − ln( 55.89. K = ωn2 = 3124. α = 2ζωn = 40. 16. The equivalent forward-path transfer function is G (s ) = T (s ) = 10 K1 G (s ) . Since =2 1 + G (s ) s + (10 K2 + 2)s + 10K1 10 K1 . Hence, s[s + (10 K2 + 2)] π 4 = 1, ∴ Im = π . The poles are thus at –2+jπ. Hence, = 2, ∴ Re = 2; and Tp = Im Re ω n = 2 2 + π 2 = 10 K1 . Thus, K1 = 1.387. Also, (10K2 + 2)/2 = Re = 2. Hence, K2 = 1/5. Ts = 17. a. For the inner loop, Ge(s) = Ge(s) 20 , and He(s) = 0.2s. Therefore, Te(s) = 1 + G (s)H (s) = e e s(s + 12) 20 s(s+16) . Combining with the equivalent transfer function of the parallel pair, Gp(s) = 20, the system Solutions to Problems 149 400 is reduced to an equivalent unity feedback system with G(s) = Gp(s) Te(s) = s(s+16) . Hence, T(s) = G(s) 400 1+G(s) = s2+16s+400 . b. ωn2 = 400; 2ζωn = 16. Therefore, ωn = 20, and ζ = 0.4. Ts = 4 =0.5; Tp = ζωn ω π %OS = e −ζπ / 1 −ζ 2 x100 = 25.38 ; =0.171. From Figure 4.16, ωnTr = 1.463. Hence, Tr = 0.0732. 2 n 1-ζ ωd = Im = ωn 1 - ζ2 = 18.33. 18. T (s ) = 28900 ; from which, 2ζωn = 200 and ωn = 28900 = 170. Hence, s + 200 s + 28900 2 ζ = 0.588. %OS = e -ζπ/ Also, Ts = 19. 4 ζω n 1 -ζ 2 x100 = 10.18% ; Tp = π ωn 1 − ζ 2 = 0.0229 s. = 0.04 s. Ei(s) Eg(s) 2 For the generator, Eg(s) = KfIf(s). But, If(s) = R +L s . Therefore, E (s) = s+1 . For the motor, ff i 1 1 1 consider Ra = 2 Ω, the sum of both resistors. Also, Je = Ja+JL(2 )2 = 0.75+4 = 1; De = DL(2 )2 = 1. Therefore, Kt RaJe θm(s) 0.5 KtKa = s(s+1.5) . Eg(s) = 1 s(s+J (De+ R )) e a But, θo(s) θo (s) 1 0.25 = 2 . Thus, E (s) = s(s+1.5) . Finally, θm(s) g Eg(s) θo (s) θo(s) 0.5 Ei(s) = Ei(s) Eg(s) = s(s+1)(s+1.5) . 20. 2 2(s) ( ) s2θ2(s) = T(N2 ) . For the potentiometer, Ei (s)= 10 θ2π , or N1 N2 For the mechanical system, J N 1 π θ2(s) = 5 Ei(s). For the network, Eo(s) = Ei(s) R s 1 = Ei(s) 1 , or Ei(s) = Eo(s) R+Cs s+RC 1 s+RC s . 150 Chapter 5: Reduction of Multiple Subsystems Therefore, θ 2 (s) = π 5 s+ E o (s) 5N1 E o (s) Jπ N 2 . = T(s) s⎛ s + 1 ⎞ ⎝ RC ⎠ 1 RC . Substitute into mechanical equation and obtain, s 21. The equivalent mechanical system is found by reflecting mechanical impedances to the spring. Writing the equations of motion: (4 s 2 + 2 s + 5) 1 ( s) − 5θ 2 ( s) = 4 T( s) θ θ −5θ1 ( s) + (2s 2 + 5) 2 ( s) = 0 Solving for θ2(s), (4 s + 2 s + 5) 4T (s ) −5 0 20T (s ) θ 2 (s ) = =4 2 8s + 4s 3 + 30s 2 + 10s (4 s + 2s + 5) −5 −5 (2s2 + 5) 2 The angular rotation of the pot is 0.2 that of θ2, or θ p( s) T (s) For the pot: = 2 s(4 s + 2s 2 + 15s + 5) 3 Ep (s) 50 5 = = θ p ( s) 5( 2π ) π For the electrical network: Using voltage division, Solutions to Problems Eo (s) = Ep (s) 200, 000 s = 1 1 + 200, 000 s + −5 10 s 2 Substituting the previously obtained values, 10 s ⎛ θ p (s ) ⎞ ⎛ E p (s ) ⎞ ⎛ E (s ) ⎞ Eo ( s) π ⎟⎜ o ⎟ = ⎟⎜ ⎜ = T (s) ⎝ T ( s ) ⎠ ⎜ θ p (s) ⎟ ⎜ E p ( s )⎟ s⎛ s + 1 ⎞ 4 s 3 + 2s 2 + 15s + 5 ⎝ ⎠⎝ ⎠ ⎜ ⎟( ) ⎝ 2⎠ 22. a. r 1 x4 1 s2 x3 50 s +1 2 − s −1 x2 s 2 x1 151 152 Chapter 5: Reduction of Multiple Subsystems b. x x5 x x1 xx 2 x3 1 2 x 4 4 5 - G1 r 1 x G2 5 G3 -1 x 1 x 3 1 4 G4 -G 7 Gx 5 2 G6 x 1 Solutions to Problems c. x 5 x4 x3 G8 1 x5 G x 14 G3 - G2 r x3 G4 1 -G 5 23. a. • x1 = x 2 • x2 = x 3 • x3 = − 2 x1 − 4 x2 − 6 x 3 + r y = x1 + x2 x2 - G6 x G 2 x1 7 x 1 153 154 Chapter 5: Reduction of Multiple Subsystems r 1 s 1 x 1 s 3 x 2 1 s -6 -4 -2 b. • x1 = x 2 • x2 = −3 x2 + x 3 + r • x3 = − 3x1 − 4 x 2 − 5x 3 + r y = x1 + 2 x 2 1 c. • x1 = 7x1 + x2 + r • x2 = −3 x1 + 2 x2 − x3 + 2 r • x3 = − x1 + 2 x3 + r y = x1 + 3x 2 + 2 x 3 1 x 1 1 y Solutions to Problems 2 -3 24. a. Since G(s) = 10 C(s) = R(s) , s + 24 s + 191s + 504 3 2 • •• •• • c + 24 c + 191c + 504c = 10r Let, c = x1 • c = x2 •• c = x3 Therefore, • x1 = x 2 • x2 = x 3 • x3 = − 504 x1 − 191x2 − 24 x 3 + 10r y = x1 r 10 1 s x3 1 s -24 -191 -504 x 2 1 s x 1 1 y 155 156 Chapter 5: Reduction of Multiple Subsystems b. G(s) = ( r 1 1 10 ) )( )( s + 7 s +8 s + 9 1 s 10 x 3 1 s 1 x 2 1 -8 -7 Therefore, 1 s -9 • x1 = −9 x1 + x 2 • x2 = −8 x2 + x 3 • x3 = − 7 x3 + 10r y = x1 25. a. Since G(s) = 20 C(s) = R(s) , 2 s + 15s + 66s + 80s 4 3 • •• • • •• •• • c + 15 c + 66 c + 80 c = 20r Let, c = x1 • c = x2 •• c = x3 • •• c = x4 Therefore, • x1 = x 2 • x2 = x 3 • x3 = x 4 • x4 = − 80x 2 − 66 x3 − 15 x 4 + 20r y = x1 x 1 1 y Solutions to Problems r 1 s 20 1 s x 4 x3 1 s 1 s x 2 x 11 157 y -15 -66 -80 b. G(s) = ( r 20 20 1 1 1 ). Hence, )( )( )( s s + 2 s + 5 s +8 1 s x 4 1 1 s x 3 1 -2 From which, 1 s x 2 -5 1 1 s x 1 1 -8 • x1 = −8 x1 + x 2 • x2 = −5 x2 + x 3 • x3 = − 2 x3 + x 4 • x4 = 20r y = x1 26. ∆ = 1 + [G2G3G4 + G3G4 + G4 + 1] + [G3G4 + G4]; T1 = G1G2G3G4; ∆1 = 1. Therefore, T(s) = T1∆1 G1G2G3G4 = 2 + G G G + 2G G + 2G ∆ 234 34 4 27. Closed-loop gains: G2G4G6G7H3; G2G5G6G7H3; G3G4G6G7H3; G3G5G6G7H3; G6H1; G7H2 Forward-path gains: T1 = G1G2G4G6G7; T2 = G1G2G5G6G7; T3 = G1G3G4G6G7; T4 = G1G3G5G6G7 Nontouching loops 2 at a time: G6H1G7H2 ∆ = 1 - [H3G6G7(G2G4 + G2G5 + G3G4 + G3G5) + G6H1 + G7H2] + [G6H1G7H2] ∆1 = ∆2 = ∆3 = ∆4 = 1 y 158 Chapter 5: Reduction of Multiple Subsystems T(s) = T1∆1 + T2∆2 + T3∆3 + T4∆4 ∆ G1G2G4G6G7 + G1G2G5G6G7 + G1G3G4G6G7 + G1G3G5G6G7 = 1 - H G G (G G + G G + G G + G G ) - G H - G H + G H G H 367 24 25 34 35 61 72 6172 28. 1 1 Closed-loop gains: -s2; - s ; - s ; -s2 1 Forward-path gains: T1 = s; T2 = 2 s Nontouching loops: None 11 ∆ = 1 - (-s2 - s - s - s2) ∆1 = ∆2 = 1 T1∆1 + T2∆2 = G(s) = ∆ 29. 1 s+ 2 s s3+1 1 1 2 = 2s4+s2+2s 1 + ( s2 + s + s + s ) ⎛ G2G3G4G5 ⎞ G1⎜(1-G H )(1-G H )⎟ 21 4 2⎠ ⎝ T(s) = = G2G3G4G5G6G7G8 1 - (1-G H )(1-G H )(1-G H ) 21 42 74 G1G2G3G4G5(1-G7H4) 1-G2H1-G4H2+G2G4H1H2-G7H4+G2G7H1H4+G4G7H2H4-G2G4G7H1H2H4-G2G3G4G5G6G7G8 30. a. G (s ) = (s + 1)(s + 2) 2 5 6 = + 2 2− ( s + 3) (s + 4) (s + 3) s+3 s+4 Solutions to Problems Writing the state and output equations, . x 1 = -3x1 + x2 . x 2 = -3x2 + r . x 3 = -4x3 + r y = 2x1 - 5x2 + 6x3 In vector-matrix form, 0⎤ ⎡− 3 1 ⎡ 0⎤ x = ⎢ 0 − 3 0 ⎥ x + ⎢ 1⎥ r ⎢0 ⎢ 1⎥ 0 −4 ⎥ ⎦ ⎣ ⎣⎦ • y = [2 b. G(s) = G(s) = (s + 2) 3/ 4 1 5/4 1 − 2 2 =− 2+ 2− ( s + 5) ( s + 5) (s + 7) s + 5 (s + 7) s + 7 1 s r −5 6] 1 x 1 s 1 2 x 1 -3 4 -5 -5 y 1 1 1 s -7 x 4 1 s 1 -7 -1 Writing the state and output equations, . x 1 = -5x1 + x2 . x 2 = -5x2 + r . x 3 = -7x3 + x4 . x 4 = -7x4 + r 5 3 y = - 4 x1 + x2 - 4 x3 - x4 In vector matrix form, x 3 -5 4 159 160 Chapter 5: Reduction of Multiple Subsystems . x= y= - -5 0 0 0 3 4 1 -5 0 0 1 0 0 -7 0 - 5 4 0 1 x+ 0 1 0 0 1 -7 -1 x c. Writing the state and output equations, . x 1 = - 2x1 + x2 . x 2 = - 2x2 + r . x 3 = - 4x 3 + r . x 4 = - 5x4 + r 1 1 2 1 y = 6 x1 + 36 x2 - 4 x3 + 9 x4 In vector-matrix form, r Solutions to Problems 31. a. Writing the state equations, . x 1 = x2 . x 2 = - 7x1 - 2x2 + r y = 3x1 + x2 In vector matrix form, b. 161 162 Chapter 5: Reduction of Multiple Subsystems Writing the state equations, • x1 = x 2 • x2 = x 3 • x3 = − x1 − 2 x2 − 5 x3 + r y = 6 x1 + 2 x 2 + x3 In vector matrix form, 1 ⎡0 x = ⎢0 0 ⎢− 1 −2 ⎣ • 0⎤ ⎡ 0⎤ 1 ⎥ X + ⎢ 0⎥ r ⎢ 1⎥ −5⎥ ⎦ ⎣⎦ y = [6 2 1]x c. . x 1 = x2 . x 2 = x3 . x 3 = x4 . x 4 = - 4x1 - 6x2 - 5x3 - 3x4 + r y = x1 + 7x2 + 2x3 + x4 In vector matrix form, Solutions to Problems 163 32. a. Controller canonical form: From the phase-variable form in Problem 5.31(a), reverse the order of the state variables and obtain, . x 2 = x1 . x 1 = - 7x2 - 2x1 + r y = 3x2 + x1 Putting the equations in order, . x 1 = - 2x1 - 7x2 + r . x 2 = x1 y = x1 + 3x2 In vector-matrix form, • x= ⎡− 2 −7⎤ ⎡1 ⎤ x+ ⎢1 0⎥ ⎢0 ⎥ r ⎣ ⎦ ⎣⎦ y = [1 3]x Observer canonical form: s+3 1 G(s) = 2 . Divide each term by 2 s s +2s+7 G( s ) = and get 1 3 + s 2 s 2 7 1+ + s 2 s = C( s ) R( s ) Cross multiplying, 13 2 7 ( s + 2 ) R(s) = (1 + s + 2 ) C(s) s s Thus, 1 1 s (R(s) - 2C(s)) + s2 (3R(s) - 7C(s)) = C(s) Drawing the signal-flow graph, 164 Chapter 5: Reduction of Multiple Subsystems R(s) 1 3 -2 -7 Writing the state and output equations, . x 1 = - 2x1 + x2 + r . x 2 = - 7x1 + 3r y = x1 In vector matrix form, • x= ⎡− 2 1 ⎤ ⎡1⎤ x+ r ⎢ −7 0 ⎥ ⎢ 3⎥ ⎣ ⎦ ⎣⎦ y = [1 0]x b. Controller canonical form: From the phase-variable form in Problem 5.31(b), reverse the order of the state variables and obtain, • x3 = x2 • x2 = x1 • x1 = − x 3 − 2 x2 − 5 x1 y = 6 x3 + 2 x 2 + x1 Putting the equations in order, • x1 = −5 x1 − 2 x 2 − x3 • x2 = x1 • x3 = x2 y = x1 + 2 x 2 + 6 x3 In vector-matrix form, Solutions to Problems 165 ⎡− 5 − 2 − 1⎤ ⎡1 ⎤ 0 0 ⎥ x + ⎢0 ⎥ r x = ⎢1 ⎢0 ⎢0 ⎥ 1 0⎥ ⎦ ⎣ ⎣⎦ • y = [1 2 6]x Observer canonical form: G(s) = s2 + 2s + 6 1 . Divide each term by 3 and get 3 2 s + 5s + 2s + 1 s 126 ++ s s2 s3 = C(s) G(s) = 521 1 + + 2 + 3 R(s) ss s Cross-multiplying, Thus, ⎛1 2 6 ⎞ ⎛521 + 2 + 3 ⎠ R(s) = ⎝1 + + 2 + 3 ⎞ C(s) ⎝s s s s⎠ ss 1 1 1 (R(s) − 5c(s)) + 2 (2R(s) − 2C(s)) + 3 (6R(s) − C(s)) = C(s) s s s Drawing the signal-flow graph, 1 2 R(s) 6 1 s X3(s) 1 s 1 X2(s) 1 s 1 X1(s) -5 -2 -1 Writing the state and output equations, • x1 = −5 x1 + x 2 + r • x2 = −2 x1 + x3 + 2 r • x3 = − x1 + 6r y = [1 0 0]x 1 C(s) 166 Chapter 5: Reduction of Multiple Subsystems In vector-matrix form, ⎡ −5 1 x = ⎢− 2 0 ⎢ −1 0 ⎣ 0⎤ ⎡1 ⎤ 1⎥ x + ⎢2 ⎥ r ⎢6 ⎥ 0⎥ ⎦ ⎣⎦ • y = [1 0 0]x c. Controller canonical form: From the phase-variable form in Problem 5.31(c), reverse the order of the state variables and obtain, . x 4 = x3 . x 3 = x2 . x 2 = x1 . x 1 = - 4x4 - 6x3 - 5x2 - 3x1 + r y = x4 + 7x3 + 2x2 + x1 Putting the equations in order, . x 1 = - 3x1 - 5x2 - 6x3 - 4x4 + r . x 2 = x1 . x 3 = x2 . x 4 = x3 y = x1 + 2x2 +7x3 + x4 In vector-matrix form, ⎡ 1⎤ ⎡− 3 − 5 −6 −4 ⎤ • ⎢ 0⎥ ⎢1 0 0 0 ⎥ x= r X+ ⎢ 0⎥ ⎢0 1 0 0 ⎥ ⎢ 0⎥ ⎢0 0 1 0 ⎥ ⎣⎦ ⎣ ⎦ y = [1 2 7 1]x Observer canonical form: s3+2s2+7s+1 1 G(s) = 4 3 2 . Divide each term by 2 s s +3s +5s +6s+4 1 + s G( s ) = 2 s 3 + 1+ s 7 + 2 s 5 s 2 + and get + 3 1 s 6 s 3 + 4 = 4 s 4 C( s ) R( s ) Solutions to Problems Cross multiplying, 7 1 3 5 6 4 12 ( s + 2 + 3 + 4 ) R(s) = (1 + s + 2 + 3 + 4 ) C(s) s s s s s s Thus, 1 1 1 1 s (R(s) - 3C(s)) + s2 (2R(s) - 5C(s)) + s3 (7R(s) - 6C(s)) + s4 (R(s) - 4C(s)) = C(s) Drawing the signal-flow graph, 1 R(s) 2 1 7 -3 -5 -6 -4 Writing the state and output equations, . x 1 = - 3x1 + x2 + r . x 2 = - 5x1 + x3 + 2r . x 3 = - 6x1 + x4 +7r . x 4 = - 4x1 + r y = x1 In vector matrix form, 167 168 Chapter 5: Reduction of Multiple Subsystems 33. a. 1 r 1 s x3 1 50 1 s -5 -2 -1 Writing the state equations, • x1 = −7 x1 + x 2 • x2 = −5 x2 + x 3 • x3 = − 50 x1 − 2 x3 + 50r y = x1 In vector-matrix form, ⎡ −7 ⎢ x=⎢ 0 ⎢ ⎣− 50 • ⎡0 ⎤ 0⎤ ⎥ ⎢⎥ −5 1 ⎥ x + ⎢ 0 ⎥ r ⎥ ⎢⎥ 0 − 2⎦ ⎣50 ⎦ 1 y = [1 0 0]x b. x 2 1 1 s x1 c= y -7 Solutions to Problems 1 s 1 r x3 1x 2 s 10 1 s x1 c= y -8 -25 -1 Writing the state equations, • x1 = x 2 • x2 = −25 x1 − 8 x2 + 10 x3 • x3 = − x1 + r y = x1 In vector-matrix form, ⎡0 ⎡ 0⎤ 1 0⎤ ⎢ ⎥ ⎢⎥ x = ⎢− 25 − 8 10⎥ x + ⎢ 0⎥ r ⎢ ⎥ ⎢⎥ ⎣ −1 0 0 ⎦ ⎣ 1⎦ • y = [1 0 0]x c. 1 1 100 1 s x2 1 1 s x1 r -1 -1 Tach feedback before integrator -1 . x 1 = x2 . x 2 = -x2 - x2 + 100(r-x1) = -100x1 -2x2 +100r y = x1 c= y 169 170 Chapter 5: Reduction of Multiple Subsystems In vector-matrix form, 0 1 - 100 - 2 0 x= y= d. Since x+ 0 r 100 10x 1 1 =2 , we draw the signal-flow as follows: (s+1)2 s +2s+1 1 1 1 s 10 1 x2 s r x1 2 c= y -2 -1 -1 Writing the state equations, . x 1 = x2 . x 2 = -x1 - 2x2 + 10(r-c) = -x1 - 2x2 + 10(r - (2x1+x2) = -21x1 - 12x2 + 10r y = 2x1 + x2 In vector-matrix form, x= 0 -21 1 -12 x+ 0 r 10 y= 2 1 x 34. a. Phase-variable form: 10 T(s) = 3 2 s +3s +2s+10 10 1 s x 3 r= u 1 s -3 -2 -10 Writing the state equations, x 2 1 s x1 c= y Solutions to Problems . x 1 = x2 . x 2 = x3 . x 3 = -10x1 -2x2 -3x3 + 10u y = x1 In vector-matrix form, x= y= 0 0 -10 1 0 -2 0 1 -3 0 0 10 x+ u 100x b. Parallel form: -10 5 5 G(s) = s + s+1 + s+2 1 s x1 1 5 r=u -10 1 r=u 1 s x2 1 c=y -1 5 1 s x3 1 -2 -1 Writing the state equations, . x 1 = 5(u - x1 - x2 - x3) = -5x1 -5x2 -5x3 +5u . x 2 = -10(u - x1 - x2 - x3) - x2 = 10x1 + 9x2 + 10x3 - 10u . x 3 = 5(u - x1 - x2 - x3) - 2x3 = -5x1 -5x2 -7x3 +5u y = x 1 + x 2 + x3 In vector-matrix form, 171 172 Chapter 5: Reduction of Multiple Subsystems ⎡− 5 − 5 −5⎤ ⎡5⎤ x = ⎢10 9 10 ⎥ x + ⎢ −10 ⎥u ⎢− 5 − 5 −7⎥ ⎢5⎥ ⎣ ⎦ ⎣ ⎦ • y = [1 1 1]x 35. a . T (s ) = 10(s2 + 5s + 6) s4 + 16 s3 + 99s2 + 244 s + 180 Drawing the signal-flow diagram, 1 5 r 10 1 s x4 16 1 s 1 s 1 s x3 x2 99 244 180 Writing the state and output equations, • x1 = x 2 • x2 = x 3 • x3 = x 4 • x1 = −180 x1 − 244 x2 − 99x 3 − 16 x4 + 10r y = 6 x1 + 5 x2 + x 3 In vector-matrix form, ⎡0 • ⎢0 x= ⎢0 ⎢− 180 ⎣ 1 0 0 1 0 0 −244 −99 y = [6 5 1 0]x 0⎤ ⎡0⎤ ⎢0⎥ 0⎥ x+ r ⎢0⎥ 1⎥ ⎢10 ⎥ −16⎥ ⎦ ⎣⎦ 6 y x1 Solutions to Problems b. G ( s ) = 10( s + 2)( s + 3) 1/ 3 10 / 3 15 12 = − + − ( s + 1)(s + 4)(s + 5)( s + 6) s + 1 s + 4 s + 5 s + 6 Drawing the signal-flow diagram and including the unity-feedback path, 1 s 1 3 r=u 1 -1 − 10 3 1 s -4 15 1 1 x3 1 1 s -6 y 1 x2 1 s -5 -12 1 x1 x4 -1 Writing the state and output equations, • 1 x1 = (u − x1 − x2 − x3 − x 4 ) − x1 3 • −10 x2 = (u − x1 − x 2 − x3 − x4 ) − 4 x2 3 • x3 = 15(u − x1 − x2 − x3 − x 4 ) − 5x 3 • x4 = − 12(u − x1 − x 2 − x3 − x4 ) − 12 x4 y = x1 + x2 + x 3 + x 4 In vector-matrix form, 173 174 Chapter 5: Reduction of Multiple Subsystems ⎡− 4 − 1 − 1 − 1 ⎤ ⎡1⎤ 3 3 3⎥ ⎢3 ⎢3⎥ • 10 2 10 10 10 − ⎥ x + ⎢− ⎥u x=⎢ 3 3 3 ⎢− 15 − 15 −3 −3 ⎥ ⎢ 15 ⎥ 20 15 ⎢ 12 12 12 ⎢ −12 ⎥ 0⎥ ⎣ ⎣ ⎦ ⎦ y = [1 1 1 1]x 36. Program: '(a)' 'G(s)' G=zpk([-2 -3],[-1 -4 -5 -6],10) 'T(s)' T=feedback(G,1,-1) [numt,dent]=tfdata(T,'v'); 'Find controller canonical form' [Acc,Bcc,Ccc,Dcc]=tf2ss(numt,dent) A1=flipud(Acc); 'Transform to phase-variable form' Apv=fliplr(A1) Bpv=flipud(Bcc) Cpv=fliplr(Ccc) '(b)' 'G(s)' G=zpk([-2 -3],[-1 -4 -5 -6],10) 'T(s)' T=feedback(G,1,-1) [numt,dent]=tfdata(T,'v'); 'Find controller canonical form' [Acc,Bcc,Ccc,Dcc]=tf2ss(numt,dent) 'Transform to modal form' [A,B,C,D]=canon(Acc,Bcc,Ccc,Dcc,'modal') Computer response: ans = (a) ans = G(s) Zero/pole/gain: 10 (s+2) (s+3) ----------------------(s+1) (s+4) (s+5) (s+6) ans = T(s) Zero/pole/gain: 10 (s+2) (s+3) -----------------------------------------(s+1.264) (s+3.412) (s^2 + 11.32s + 41.73) ans = Solutions to Problems Find controller canonical form Acc = -16.0000 1.0000 0 0 Bcc = -99.0000 -244.0000 -180.0000 0 0 0 1.0000 0 0 0 1.0000 0 1 0 0 0 Ccc = 0 10.0000 50.0000 60.0000 Dcc = 0 ans = Transform to phase-variable form Apv = 0 1.0000 0 0 0 0 -180.0000 -244.0000 0 1.0000 0 -99.0000 0 0 1.0000 -16.0000 10.0000 0 Bpv = 0 0 0 1 Cpv = 60.0000 50.0000 ans = (b) ans = G(s) Zero/pole/gain: 10 (s+2) (s+3) ----------------------(s+1) (s+4) (s+5) (s+6) ans = T(s) Zero/pole/gain: 175 176 Chapter 5: Reduction of Multiple Subsystems 10 (s+2) (s+3) -----------------------------------------(s+1.264) (s+3.412) (s^2 + 11.32s + 41.73) ans = Find controller canonical form Acc = -16.0000 1.0000 0 0 -99.0000 -244.0000 -180.0000 0 0 0 1.0000 0 0 0 1.0000 0 Bcc = 1 0 0 0 Ccc = 0 10.0000 50.0000 60.0000 Dcc = 0 ans = Transform to modal form A= -5.6618 -3.1109 0 0 3.1109 -5.6618 0 0 0 0 -3.4124 0 0 0 0 -1.2639 0.6973 -0.1401 4.2067 B= -4.1108 1.0468 1.3125 0.0487 C= 0.1827 D= 0 37. Solutions to Problems 1 1 1x 4 s 1 s x3 1 s 1 r x2 1 s -1 -1 Writing the state equations, . x 1 = x2 . x 2 = - x1 + x 3 . x 3 = x4 . x 4 = x1 - x 2 + r y = -x 1 + x2 In vector-matrix form, 0 1 x = -0 1 1 0 0 -1 0 1 0 0 0 0 0x+0r 1 0 0 1 y = c = [-1 1 0 0] x 38. a. .. . . θ 1 + 5θ 1 + 6θ1 - 3θ 2 - 4θ2 = 0 . .. . -3θ 1 - 4θ1 + θ 2 + 5θ 2 + 5θ2 = T or .. . . θ 1 = - 5θ 1 - 6θ1 + 3θ 2 + 4θ2 .. . . θ 2 = 3θ 1 + 4θ1 - 5θ 2 - 5θ2 + T . . Letting, θ1 = x1 ; θ 1 = x2 ; θ2 = x3 ; θ 2 = x4 , x1 -1 c= y 177 178 Chapter 5: Reduction of Multiple Subsystems 1 x2 s 1 s 4 3 x1 -5 -6 4 3 1 s T 1 x4 s x3 1 -5 -5 where x = θ. b. Using the signal-flow diagram, . x 1 = x2 . x 2 = -6x1 - 5x2 + 4x3 + 3x4 . x 3 = x4 . x 4 = 4x1 + 3x2 - 5x3 - 5x4 + T y = x3 In vector-matrix form, 0 0100 0 -6 -5 4 3 . T x+ x= 0 0001 1 4 3 -5 -5 y= 0 0 1 0 x 39. Program: numg=7; deng=poly([0 -9 -12]); G=tf(numg,deng); T=feedback(G,1) [numt,dent]=tfdata(T,'v') [A,B,C,D]=tf2ss(numt,dent); '(a)' A=flipud(A); A=fliplr(A) B=flipud(B) C=fliplr(C) %Obtain controller canonical form %Display label %Convert to phase-variable form %Convert to phase-variable form %Convert to phase-variable form %Convert to phase-variable form Solutions to Problems '(b)' [a,b,c,d]=canon(A,B,C,D) Computer response: Transfer function: 7 -----------------------s^3 + 21 s^2 + 108 s + 7 numt = 0 0 0 7 21 108 7 1 0 -108 0 1 -21 0 0 dent = 1 ans = (a) A= 0 0 -7 B= 0 0 1 C= 7 ans = %Display label %Convert to parallel form 179 180 Chapter 5: Reduction of Multiple Subsystems (b) a= -0.0657 0 0 0 -12.1807 0 0 0 -8.7537 -0.0470 -0.0908 b= -0.0095 -3.5857 2.5906 c= -6.9849 d= 0 40. . x 1 = A1x1 + B1r y1 = C1x1 . x 2 = A2x2 + B2y1 y2 = C2x2 (1) (2) (3) (4) Substituting Eq. (2) into Eq. (3), . x 1 = A1x1 + B1r . x 2 = B2C1x1 + A2x2 y2 = C2x2 In vector-matrix notation, x1 A 1 O x1 = + B1 r x2 B2 C 1 A 2 x 2 O y 2 = O C2 x1 x2 41. . x 1 = A1x1 + B1r y1 = C1x1 (1) (2) Solutions to Problems . x 2 = A2x2 + B2r y2 = C2x2 181 (3) (4) In vector-matrix form, x1 = x2 A 1 -O x 1 + B1 r B2 O A2 x 2 x1 y = y1 + y 2 = C1 C2 x2 42. . x 1 = A1x1 + B1e y = C1x1 . x 2 = A2x2 + B2y p = C2x2 Substituting e = r - p into Eq. (1) and substituting Eq. (2) into (3), we obtain, . x 1 = A1x1 + B1(r - p) y = C1x1 . x 2 = A2x2 + B2C1x1 p = C2x2 Substituting Eq. (8) into Eq. (5), . x 1 = A1x1 - B1C2x2 + B1r . x 2 = B2C1x1 + A2x2 In vector-matrix form, y = C1x1 x1 = A 1 - B1 C2 B 2 C1 A2 x2 x1 y = C1 0 x2 43. • z = P −1 APz + P −1 Bu y = CPz x1 + B 1 r x2 0 (1) (2) (3) (4) (5) (6) (7) (8) 182 Chapter 5: Reduction of Multiple Subsystems ⎡2 P = ⎢1 ⎢4 ⎣ −1 1 −2 6 ⎡ −1.6 P AP = ⎢ 3.33 ⎢1.63 ⎣ −1 −4 ⎤ ⎡ 0.0606 0.3939 0.1212 ⎤ 0 ⎥; ∴ P = ⎢ 0.0303 -0.3030 0.0606 ⎥ ⎥ ⎢ -0.2121 0.1212 0.0758 ⎦ 2⎥ ⎣ ⎦ 3.81 ⎤ ⎡ −3⎤ -1 1.33 − 2.33⎥; P B = ⎢ 4 ⎥ ; CP = [-0.544 -0.0702 ⎢4⎥ −1.79 1.26 ⎥ ⎣⎦ ⎦ 1.23 0.912] 44. 45. Eigenvalues are -1, -2, and -3 since, |λΙ - A | = (λ + 3) (λ + 2) (λ + 1) Solving for the eigenvectors, Ax = λx or, x3 x2 For λ = -1, x2 = 0, x1 = x3 . For λ = -2, x1 = x2 = 2 . For λ = -3, x1 = - 2 , x2 = x3 . Thus, . z = P-1APz + P-1Bu ; y = CPz, where 46. Eigenvalues are 1, -2, and 3 since, Solutions to Problems |λI - A | = (λ - 3) (λ + 2) (λ - 1) Solving for the eigenvectors, Ax = λx or, x3 For λ = 1, x1 = x2 = 2 . For λ = -2, x1 = 2x3, x2 = -3x3. For λ = 3, x1 = x3 , x2 = -2x3 . Thus, . z = P-1APz + P-1Bu ; y = CPz, where 47. Program: A=[-10 -3 7;18.25 6.25 -11.75;-7.25 -2.25 5.75]; B=[1;3;2]; C=[1 -2 4]; [P,d]=eig(A); Ad=inv(P)*A*P Bd=inv(P)*B Cd=C*P Computer response: Ad = -2.0000 -0.0000 0.0000 0.0000 3.0000 0.0000 0.0000 -0.0000 1.0000 3.6742 2.8577 Bd = 1.8708 -3.6742 3.6742 Cd = 3.2071 48. a. Combine G1(s) and G2(s). Then push K1 to the right past the summing junction: 183 184 Chapter 5: Reduction of Multiple Subsystems R(s) + + K K 1 - C(s) + G (s) 1 2 - - G (s) 2 s 2 s K 1 1 Push K1K2 to the right past the summing junction: R(s) + + C(s) + K - - - K 1 G (s) G (s) 21 2 s 2 K K 1 2 s K 1 1 Hence, T(s) = K1K2G1(s)G2(s) s s2 ⎞ ⎛ 1 + K1K2G1(s)G2(s) ⎜1 + K + K K ⎟ 1 1 2⎠ ⎝ b. Rearranging the block diagram to show commanded pitch rate as the input and actual pitch rate as the output: Solutions to Problems 185 K 1 Commanded pitch rate + K Actual pitch rate + G (s) 1 2 - - G (s) 2 s s 2 1 Pushing K2 to the right past the summing junction; and pushing s to the left past the pick-off point yields, K 1 Commanded pitch + rate Actual pitch rate s + K - - 2 s G (s) G (s) 2 1 s K 2 1 Finding the closed-loop transfer function: T(s) = K2sG1(s)G2(s) K1⎞ s ⎛ 1 + K2sG1(s)G2(s)⎜1 + K + s ⎟ 2 ⎝ ⎠ = K2sG1(s)G2(s) 1 + G1(s)G2(s)(s2 + K2s + K1K2) 186 Chapter 5: Reduction of Multiple Subsystems c. Rearranging the block diagram to show commanded pitch acceleration as the input and actual pitch acceleration as the output: KK 12 K Commanded pitch acceleration - 2 s Actual pitch acceleration - + + G (s) 1 - G (s) 2 s 2 1 Pushing s2 to the left past the pick-off point yields, KK 12 s 2 K Commanded pitch acceleration 2 s - - + + - 2 s G (s) G (s) 2 1 Actual pitch acceleration 1 Finding the closed-loop transfer function: T(s) = 49. s2G1(s)G2(s) s2G1(s)G2(s) = ⎛ K1K2 K2⎞ 1 + G1(s)G2(s)(s2 + K2s + K1K2) 1 + s2G1(s)G2(s)⎜1 + 2 + s ⎟ s ⎝ ⎠ K1 Establish a sinusoidal model for the carrier: T(s) = 2 2 s +a Solutions to Problems r 1 K1 1 s 1 x2 s 187 x1 -a K2 Establish a sinusoidal model for the message: T(s) = 2 2 s +b r 2 K2 1 1 s x4 s x3 -b Writing the state equations, . x 1 = x2 . x 2 = - a2x1 + K1r . x 3 = x4 . x 4 = - b2x3 + K2r y = x1x3 50. K1K2 The equivalent forward transfer function is G(s) = s(s+a ) . The equivalent feedback transfer function 1 is K4s H(s) = K3 + s+a . Hence, the closed-loop transfer function is 2 K1K2(s+a2) G(s) T(s) = 1 + G(s)H(s) = 3 2 + (a a +K K K +K K K )s + K K K a s + (a1+a2)s 12 1 2 3 1 2 4 1 2 32 51. a. The equivalent forward transfer function is 5 s s +2 5K 1 = Ge s =K s s +3 5 s s +3 s 2+2 s +5 1+ s s +2 Ge 5K Ts= = 4 + 5 s 3 + 11 s 2 + 15 s + 5 K 1+ G e s 188 Chapter 5: Reduction of Multiple Subsystems b. Draw the signal-flow diagram: 1 K 1 s 5 u 1 s x4 1 -1 -1 Writing the state and output equations from the signal-flow diagram: . x1 = x2 . x2 = − 3x 2 + x 3 . x3 = x 4 . x4 = − 5Kx1 − 5x 3 − 2 x 4 + 5 Ku y = x1 1 0 0⎤ ⎡ 0⎤ ⎡0 . ⎢0 −3 1 0 ⎥ ⎢ 0 ⎥ u x= x+ ⎢0 0 0 1⎥ ⎢ 0⎥ ⎢ − 5K 0 −5 −2 ⎥ ⎢ 5K⎥ ⎣ ⎦⎣⎦ c. Program: for K=1:1:5 numt=5*K; dent=[1 5 11 15 5*K]; T=tf(numt,dent); hold on; subplot(2,3,K); step(T,0:0.01:20) title(['K=',int2str(K)]) end x2 -3 y = [1 0 0 0]x 1 s 1 y x3 -2 In vector-matrix form: 1 s x1 Solutions to Problems 189 Computer response: 52. a. Draw the signal-flow diagram: 1 1666.67 0.06 1 s u 15x10 6 1 s x4 x3 -82 -720 1 s 1 s x2 -4x10 6 -2x10 7 -1 Write state and output equations from the signal-flow diagram: 1 x1 y 190 Chapter 5: Reduction of Multiple Subsystems . x1 = x2 . x2 = x3 . x3 = − 2*107 x1 − 4 * 106 x2 − 82 x3 + 15 *10 6 x4 . x4 = − 100 x1 − 720 x 4 + 100u y = x1 In vector-matrix form: ⎡0 . ⎢0 x= ⎢ − 2*107 ⎢ −100 ⎣ 1 0 − 4 *10 6 0 0 0⎤ ⎡0⎤ ⎢0⎥ 1 0⎥ u x+ 6⎥ ⎢0⎥ −82 15 * 10 ⎢100⎥ 0 −720 ⎥ ⎦ ⎣⎦ y = [1 0 0 0]x b. Program: numg=1666.67*0.06*15e6; deng=conv([1 720],[1 82 4e6 2e7]); 'G(s)' G=tf(numg,deng) 'T(s)' T=feedback(G,1) step(T) Computer response: ans = G(s) Transfer function: 1.5e009 ---------------------------------------------------s^4 + 802 s^3 + 4.059e006 s^2 + 2.9e009 s + 1.44e010 ans = T(s) Transfer function: 1.5e009 ---------------------------------------------------s^4 + 802 s^3 + 4.059e006 s^2 + 2.9e009 s + 1.59e010 Solutions to Problems 53. -272(s2+1.9s+84) a. Phase-variable from: G(s) = 3 s +17.1s2+34.58s-123.48 Drawing the signal-flow diagram: 1 1.9 1 s -272 u -17.1 1 s x3 1 s x2 84 x1 -34.58 123.48 Writing the state and output equations: . x1 = x2 . x2 = x3 . x3 = 123. 48 x1 − 34.58 x 2 − 17.1x3 − 272u y = 84 x1 + 1. 9 x2 + x3 In vector-matrix form: y 191 192 Chapter 5: Reduction of Multiple Subsystems 1 0⎤ ⎡0 ⎡0⎤ x=⎢ 0 0 1 ⎥x + ⎢ 0 ⎥u ⎢123. 48 − 34. 58 −17.1⎥ ⎢ − 272⎥ ⎣ ⎦ ⎣ ⎦ . y = [84 1. 9 1]x -272(s2+1.9s+84) b. Controller canonical form: G(s) = 3 s +17.1s2+34.58s-123.48 Drawing the signal-flow diagram: 1 1.9 1 s -272 1 s 1 s 84 y u -17.1 x1 x2 x3 -34.58 123.48 Writing the state and output equations: . . x1 = − 17.1x1 − 34. 58 x2 + 123. 48 x3 − 272u . x2 = x1 . x3 = x 2 y = x1 + 1. 9 x2 + 84 x3 In vector-matrix form: ⎡ − 17.1 −34. 58 123. 48⎤ ⎡ − 272⎤ x=⎢ 1 0 0 ⎥x + ⎢ 0 ⎥u ⎢0 ⎢0⎥ 1 0⎥ ⎣ ⎦ ⎣ ⎦ . y = [1 1. 9 84]x c. Observer canonical form: Divide by highest power of s and obtain -272 516.8 22848 s - s 2 - s3 G(s) = 17.1 34.58 123.48 1+ s + 2 s s3 Cross multiplying, Solutions to Problems 193 17.1 34.58 123.48 -272 516.8 22848C(s) [ s - 2 - 3 ]R(s) = [ 1 + s + 2 s s s s3 Rearranging, 1 1 1 C(s) = s [ -272R(s) - 17.1C(s)] + 2 [ -516.8R(s) - 34.58C(s)] + 3 [ -22848R(s) + 123.48C(s)] s s Drawing the signal-flow diagram, where r = u and y = c: -272 -516.8 u 22848 1 s 1 s 1 1 s 1 1 y x3 x2 x1 -17.1 -34.58 123.48 d. Draw signal-flow ignoring the polynomial in the numerator: u -272 1 s 1 s 1 x3 1 s x2 1.8 -14 Write the state equations: 1 . . x1 = − 4.9 x1 + x 2 . x2 = 1. 8x 2 + x 3 . x3 = − 14 x3 − 272 u x1 -4.9 194 Chapter 5: Reduction of Multiple Subsystems The output equation is .. . y = x1 +1. 9 x1 + 84 x1 But, (1) . x1 = −4.9 x1 + x 2 and .. . (2) . x1 = − 4.9 x1 + x 2 = − 4. 9( −4. 9 x1 + x2 ) + 1.8 x 2 + x 3 Substituting Eqs. (2) and (3) into (1) yields, y = 98. 7x1 − 1.2 x 2 + x 3 In vector-matrix form: 0⎤ ⎡ − 4. 9 1 ⎡0⎤ x=⎢ 0 1. 8 1 ⎥ x + ⎢ 0 ⎥ u ⎢0 ⎢ −272 ⎥ 0 − 14⎥ ⎣ ⎦ ⎣ ⎦ . y = [98. 7 −1. 2 1]x e. Expand as partial fractions: G s = − 479.38 1 − 232.94 1 + 440.32 1 s + 14 s − 1.8 s + 4.9 Draw signal-flow diagram: 1 s -14 -479.38 u -232.94 1 x1 1.8 440.32 1 y x 2 1 s -4.9 Write state and output equations: x . x1 = − 14 x1 + − 479. 38u . . x2 = 1. 8x 2 − 232. 94u . x3 = − 4. 9 x3 + 440. 32u y = x1 + x2 + x3 In vector-matrix form: 1 1 s 3 1 (3) Solutions to Problems ⎡ − 14 0 x = ⎢ 0 1.8 ⎢0 0 ⎣ . 0⎤ ⎡ − 479. 38 ⎤ 0 ⎥ x + ⎢ −232. 94 ⎥u ⎢ 440. 32 ⎥ −4. 9⎥ ⎦ ⎣ ⎦ y = [1 1 1]x 54. Push Pitch Gain to the right past the pickoff point. Collapse the summing junctions and add the feedback transfer functions. Apply the feedback formula and obtain, T (s ) = G(s) 0.25(s + 0.435) =4 3 1 + G(s) H( s) s + 3.4586s + 3.4569s2 + 0.9693s + 0.15032 55. Program: numg1=-0.125*[1 0.435] deng1=conv([1 1.23],[1 0.226 0.0169]) 'G1' G1=tf(numg1,deng1) 'G2' G2=tf(2,[1 2]) G3=-1 'H1' H1=tf([-1 0],1) 'Inner Loop' Ge=feedback(G1*G2,H1) 'Closed-Loop' T=feedback(G3*Ge,1) 195 196 Chapter 5: Reduction of Multiple Subsystems Computer response: numg1 = -0.1250 -0.0544 deng1 = 1.0000 1.4560 0.2949 0.0208 ans = G1 Transfer function: -0.125 s - 0.05438 -----------------------------------s^3 + 1.456 s^2 + 0.2949 s + 0.02079 ans = G2 Transfer function: 2 ----s+2 G3 = -1 ans = H1 Transfer function: -s ans = Inner Loop Transfer function: -0.25 s - 0.1088 -----------------------------------------------s^4 + 3.456 s^3 + 3.457 s^2 + 0.7193 s + 0.04157 ans = Closed-Loop Solutions to Problems Transfer function: 0.25 s + 0.1088 ----------------------------------------------s^4 + 3.456 s^3 + 3.457 s^2 + 0.9693 s + 0.1503 56. 197 198 Chapter 5: Reduction of Multiple Subsystems Linear Deadzone Backlash Linear 57. a. Since VL(s) = Vg(s) – VR(s), the summing junction has Vg(s) as the positive input and VR(s) as the negative input, and VL(s) as the error. Since I(s) = VL(s) (1/(Ls)), G(s) = 1/(Ls). Also, since VR(s) = I(s)R, the feedback is H(s) = R. Summarizing, the circuit can be modeled as a negative feedback system, where G(s) = 1/(Ls), H(s) = R, input = Vg(s), output = I(s), and error = VL(s), where the negative input to the summing junction is VR(s). 1 I( s ) G( s) Ls = 1 . Hence, I( s ) = V (s ) 1 . = = b. T (s) = g Vg ( s) 1 + G(s)H (s) 1 + 1 R Ls + R Ls + R Ls Vg (s) c. Using circuit analysis, I( s) = . Ls + R SOLUTIONS TO DESIGN PROBLEMS 58. 1 1 1 Je = Ja+JL(20 )2 = 2+2 = 4; De = Da+DL(20 )2 = 2+DL(20 )2. Therefore, the forward-path transfer function is, Answers to Design Problems 25 2 1 ⎛ 4 ⎞⎛1⎞ G G(s) = (1000) ⎜ ⎟ ⎝20⎠ . Thus, T(s) = 1+G 1 ⎝s(s+4(De+2))⎠ Hence, ζ = %OS - ln ( 100 ) = 0.456; ωn = 2 + ln2 (%OS) π 100 = 199 1 25 . s2+4(De+2)s+ 2 De+2 25 . Therefore De = 10.9; from 2 ; 2ζωn = 4 which DL = 3560. 59. 25 ; from which, 2ζωn = 1 and ωn = 5. Hence, ζ = 0.1. Therefore, a. T(s) = 2 s +s+25 %OS = e −ζπ / 1−ζ 2 4 x100 = 72.92% ; Ts = ζω = 8. n 25K1 b. T(s) = 2 ; from which, 2ζωn = 1+25K2 and ωn = 5 K1 . Hence, s +(1+25K2)s+25K1 %OS - ln ( 100 ) 4 39 ζ= = 0.404. Also, Ts = = 0.2, Thus, ζωn = 20; from which K2 = 25 and ζωn %OS π2 + ln2 ( 100 ) ωn = 49.5. Hence, K1 = 98.01. 60. Ge(s) K The equivalent forward path transfer function is Ge(s) = s(1+(1+K )) . Thus, T(s) = 1+G (s) = 2 e K K . Prior to tachometer compensation (K2 = 0), T(s) = 2 . Therefore K = ωn2 = s +s+K s2+(1+K2)s+K 100 100. Thus, after tachometer compensation, T(s) = 2 . Hence, ωn = 10; 2ζωn = 1+K2. s +(1+K2)s+100 Therefore, K2 = 2ζωn - 1 = 2(0.5)(10) - 1 = 9. 61. At the N2 shaft, with rotation,θ L (s) ( Jeq s2 + Deqs)θ L ( s) + F( s)r = Teq (s) 2 F( s) = ( Ms + f vs) X (s) Thus, ( Jeq s2 + Deq s)θ L ( s) + ( Ms2 + f vs) X (s)r = Teq ( s) But, X (s) = rθ L (s) . Hence, [( J where + Mr )s + ( Deq + fv r )s θ L ( s) = Teq( s) 2 eq 2 2 200 Chapter 5: Reduction of Multiple Subsystems Jeq = Ja (2) 2 + J = 5 Deq = Da (2)2 + D = 4 + D r=2 Thus, the total load inertia and load damping is J L = J eq + Mr 2 = 5 + 4 M DL = Deq + fv r 2 = 4 + D + (1)( 2)2 = 8 + D Reflecting JL and DL to the motor yields, Jm = (8 + D ) (5 + 4 M) ; Dm = 4 4 Thus, the motor transfer function is 1 Jm Kt Ra J m θ m (s ) = Ea (s) s(s + 1 ( D + K t Ka )) s( s + 1 ( D + 1)) m m Jm Ra Jm The gears are (10/20)(1) = 1/2. Thus, the forward-path transfer function is 1 ⎛ ⎞ Jm ⎜ ⎟1 Ge (s) = (500 ) ⎜ s(s + 1 (D + 1))⎟ 2 m Jm ⎝ ⎠ Finding the closed-loop transfer function yields, T (s ) = For Ts = 2, 250 / J m Ge (s) = Dm + 1 250 1 + Ge (s) s 2 + s+ Jm Jm Dm + 1 = 4 . For 20% overshoot, ζ = 0.456. Thus, Jm 2ζω n = 2(0.456)ω n = Or, ω n = 4.386 = Jm = Dm + 1 =4 Jm 250 ; from which J m = 13 and hence, Dm = 51. But, Jm (8 + D ) (5 + 4 M) . Thus, M = 11.75 and D = 196. ; Dm = 4 4 Solutions to Design Problems 62. a. Desired force Yh-Y cat Spring displacement Input voltage+ 1 100 K 1 1000 Controller 0. 7883( s + 53 .85 ) (s 2 + 15. 47s + 9283 )( s 2 + 8. 119 s + 376 .3 ) Actuator - Input transducer F up Pantograph dynamics 1 100 Sensor b. G(s) = Yh ( s) − Ycat (s) 0.7883( s + 53.85) =2 2 Fup ( s) (s + 15.47s + 9283)(s + 8.119 s + 376.3) 648.7709 (s+53.85) Ge(s) = (K/100)*(1/1000)*G(s)*82.3e3 = (s2 + 8.119s + 376.3) (s^2 + 15.47s + 9283) 648.7709 (s+53.85) T(s) = Ge/(1+Ge) = (s^2 + 8.189s + 380.2) (s2 + 15.4s + 9279) 648.8 s + 3.494e04 = s4 + 23.59 s3 + 9785 s2 + 8.184e04 s + 3.528e06 c. For G(s) = (yh-ycat)/Fup Phase-variable form Ap = 0 0 0 -3.493e6 Bp = 0 0 0 1 Cp = 42.45 1 0 0 -81190 0.7883 0 1 0 - 9785 0 0 0 1 -23.59 0 Using this result to draw the signal-flow diagram, 82300 Spring Fout 201 202 Chapter 5: Reduction of Multiple Subsystems 0.7883 fdesired K =1 v 0.01 up 1000 1 s 1 s x4 1 s x 3 1 s x 2 42.45 82300 fout x1 -23.59 -9785 -81190 -3.493x106 -0.01 Writing the state and output equations • x1 = x 2 • x2 = x 3 • x3 = x 4 • x4 = − 23.59x 4 − 9785 x3 − 81190 x2 − 3493000 x1 + 0.01 f desired − 0.01 fout But, fout = 42.45 * 82300 x1 + 0.7883 * 82300 x 2 Substituting fout into the state equations yields • x4 = − 3527936.35x1 − 81838.7709 x 2 − 9785x 3 − 23.59 x4 + 0.01 fdesired Putting the state and output equations into vector-matrix form. 0 ⎡ • ⎢ 0 x= ⎢ 0 ⎢− 3.528 x106 ⎣ 1 0 0 ⎤ ⎡0⎤ 0 1 0 ⎥ ⎢0⎥ f + 0 0 1 ⎥ ⎢ 0 ⎥ desired − 81840 − 9785 − 23.59⎥ ⎢ 0.01⎥ ⎦ ⎦⎣ y = fout = [3494000 64880 0 0]x SIX Stability SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Stability Design via Gain From the antenna control challenge of Chapter 5, 76.39K T(s) = 3 s +151.32s2+198s+76.39K Make a Routh table: s3 1 198 s2 76.39K s1 151.32 29961.36-76.39K 151.32 s0 76.39K 0 0 From the s1 row, K<392.2. From the s0 row, 0<K. Therefore, 0<K<392.2. UFSS Vehicle: Stability Design via Gain G3 = − K1 G2 G3 = T (s ) = (0.25s + 0.10925)K1 s + 3. 483s3 + 3. 465s2 + 0. 60719s 4 G3 (s) (0.25s + 0.10925) K1 =4 3 1 + G3 (s ) s + 3.483s + 3.465s2 + 0.25( K1 + 2.4288)s + 0.10925K1 204 Chapter 6: Stability s4 1 3.465 0.10925K1 s3 3.483 0.25(K1+2.4288) 0 s2 1 - 4 (K1 - 45.84) 3.483 0.10925K1 0 0 0 0 0 s1 (K1 + 4.2141)(K1 - 26.42) 0.25 K1 - 45.84 s0 0.10925K1 For stability : 0 < K1 < 26.42 ANSWERS TO REVIEW QUESTIONS 1. Natural response 2. It grows without bound 3. It would destroy itself or hit limit stops 4. Sinusoidal inputs of the same frequency as the natural response yield unbounded responses even though the sinusoidal input is bounded. 5. Poles must be in the left-half-plane or on the jω axis. 6. The number of poles of the closed-loop transfer function that are in the left-half-plane, the right-halfplane, and on the jω axis. 7. If there is an even polynomial of second order and the original polynomial is of fourth order, the original polynomial can be easily factored. 8. Just the way the arithmetic works out 9. The presence of an even polynomial that is a factor of the original polynomial 10. For the ease of finding coefficients below that row 11. It would affect the number of sign changes 12. Seven 13. No; it could have quadrantal poles. 14. None; the even polynomial has 2 right-half-plane poles and two left-half-plane poles. 15. Yes 16. Det (sI-A) = 0 Solutions to Design Problems 205 SOLUTIONS TO PROBLEMS 1. s5 1 5 1 s4 3 4 3 s3 3.667 0 0 s2 4 3 0 s1 -2.75 0 0 s0 3 0 0 2 rhp; 3 lhp 2. s5 1 4 3 s4 -1 -4 -2 ε 1 0 ε 2 ε + 1 − 4ε 1 − 4ε -2 0 0 0 -2 0 0 s3 s2 1 − 4ε 2 s1 s0 3 rhp, 2 lhp 3. s5 s4 s3 s2 s1 s0 1 -1 -2 -3 -1/3 -4 3 -3 -3 -4 2 -2 ROZ 206 Chapter 6: Stability Even (4): 4 jω; Rest(1): 1 rhp; Total (5): 1 rhp; 4 jω 4. s4 1 8 15 s3 4 20 0 s2 3 15 0 s1 6 0 0 s0 15 0 0 ROZ Even (2): 2 jω; Rest (2): 2 lhp; Total: 2 jω; 2 lhp 5. s6 1 -6 1 s5 1 0 1 s4 -6 0 -6 s3 -24 0 0 s2 ε -6 s1 -144/ε 0 s0 -6 -6 ROZ Even (4): 2 rhp; 2 lhp; Rest (2): 1 rhp; 1 lhp; Total: 3 rhp; 3 lhp 6. Program: den=[1 1 -6 0 1 1 -6] A=roots(den) Computer response: den = 1 1 -6 A= -3.0000 2.0000 -0.7071 -0.7071 0.7071 0.7071 + + - 7. Program: 0.7071i 0.7071i 0.7071i 0.7071i 0 1 1 -6 Solutions to Design Problems 207 %-det([si() si();sj() sj()])/sj() %Template for use in each cell. syms e %Construct a symbolic object for %epsilon. %%%%%%%%%%%%%%%%%%%%%%%%%%$$$$$$$$$%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% s5=[1 4 3 0 0] %Create s^5 row of Routh table. %%%%%%%%%%%%%%%%%%%%%%%%%%%$$$$$$$$$%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% s4=[-1 -4 -2 0 0] %Create s^4 row of Routh table. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% if -det([s5(1) s5(2);s4(1) s4(2)])/s4(1)==0 s3=[e... -det([s5(1) s5(3);s4(1) s4(3)])/s4(1) 0 0]; %Create s^3 row of Routh table %if 1st element is 0. else s3=[-det([s5(1) s5(2);s4(1) s4(2)])/s4(1)... -det([s5(1) s5(3);s4(1) s4(3)])/s4(1) 0 0]; %Create s^3 row of Routh table %if 1st element is not zero. end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% if -det([s4(1) s4(2);s3(1) s3(2)])/s3(1)==0 s2=[e ... -det([s4(1) s4(3);s3(1) s3(3)])/s3(1) 0 0]; %Create s^2 row of Routh table %if 1st element is 0. else s2=[-det([s4(1) s4(2);s3(1) s3(2)])/s3(1) ... -det([s4(1) s4(3);s3(1) s3(3)])/s3(1) 0 0]; %Create s^2 row of Routh table %if 1st element is not zero. end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% if -det([s3(1) s3(2);s2(1) s2(2)])/s2(1)==0 s1=[e ... -det([s3(1) s3(3);s2(1) s2(3)])/s2(1) 0 0]; %Create s^1 row of Routh table %if 1st element is 0. else s1=[-det([s3(1) s3(2);s2(1) s2(2)])/s2(1) ... -det([s3(1) s3(3);s2(1) s2(3)])/s2(1) 0 0]; %Create s^1 row of Routh table %if 1st element is not zero end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% s0=[-det([s2(1) s2(2);s1(1) s1(2)])/s1(1) ... -det([s2(1) s2(3);s1(1) s1(3)])/s1(1) 0 0]; %Create s^0 row of Routh table. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 's3' s3=simplify(s3); pretty(s3) 's2' s2=simplify(s2); pretty(s2) 's1' s1=simplify(s1); pretty(s1) 's0' s0=simplify(s0); pretty(s0) %Display label. %Simplify terms in s^3 %Pretty print s^3 row. %Display label. %Simplify terms in s^2 %Pretty print s^2 row. %Display label. %Simplify terms in s^1 %Pretty print s^1 row. %Display label. %Simplify terms in s^0 %Pretty print s^0 row. row. row. row. row. 208 Chapter 6: Stability Computer response: s5 = 1 4 3 0 0 -1 -4 -2 0 0 s4 = ans = s3 [e 1 0 0] -2 0 ans = s2 [ -1 + 4 e [- -------[ e 0] ans = s1 [ 2 [ 2e +1-4e [- -------------[ -1 + 4 e 0 0 ans = s0 [-2 8. 240 T(s) = 4 s + 10s3 + 35s2 + 50s + 264 2 rhp, 2 lhp 9. Program: numg=240; deng=poly([-1 -2 -3 -4]); 'G(s)' G=tf(numg,deng) 'Poles of G(s)' pole(G) 'T(s)' 0 0 0] 0] Solutions to Design Problems 209 T=feedback(G,1) 'Poles of T(s)' pole(T) Computer response: ans = G(s) Transfer function: 240 --------------------------------s^4 + 10 s^3 + 35 s^2 + 50 s + 24 ans = Poles of G(s) ans = -4.0000 -3.0000 -2.0000 -1.0000 ans = T(s) Transfer function: 240 ---------------------------------s^4 + 10 s^3 + 35 s^2 + 50 s + 264 ans = Poles of T(s) ans = -5.3948 -5.3948 0.3948 0.3948 + + - 2.6702i 2.6702i 2.6702i 2.6702i System is unstable, since two closed-loop poles are in the right half-plane. 10. T(s) = 1 4s4 + 4s2 + 1 Even (4): 4 jω 11. T (s ) = 84 s + 5s + 12 s + 25s + 45s 4 + 50s 3 + 82 s 2 + 60 s + 84 8 7 6 5 210 Chapter 6: Stability s8 1 12 45 82 s7 1 5 10 12 s6 1 5 10 12 s5 3 10 10 s4 5 20 36 s3 -5 -29 s2 -1 4 s1 -49 s0 4 84 ROZ Even (6): 2 rhp, 2 lhp, 2 jω; Rest (2): 0 rhp, 2 lhp, 0 jω ; Total: 2 rhp, 4 lhp, 2 jω 12. T (s ) = 1 2 s + 5s + s 2 + 2 s + 1 4 3 s4 2 1 1 s3 5 2 0 s2 1 5 s1 -23 0 s0 5 Total: 2 lhp, 2 rhp 13. T (s ) = 8 s − 2 s − s + 2s + 4s 3 − 8s 2 − 4s + 8 7 6 5 4 Solutions to Design Problems 211 s7 1 -1 4 -4 s6 -2 2 -8 8 s5 -12 8 -16 0 s4 0.6667 -5.333 8 0 s3 -88 128 0 0 s2 -4.364 8 0 0 s1 -33.33 0 0 0 s0 8 0 0 0 ROZ Even (6): 3 rhp, 3 lhp; Rest (1): 1 rhp; Total: 4 rhp, 3 lhp 14. Program: numg=8; deng=[1 -2 -1 2 4 -8 -4 0]; 'G(s)' G=tf(numg,deng) 'T(s)' T=feedback(G,1) 'Poles of T(s)' pole(T) Computer response: ans = G(s) Transfer function: 8 ----------------------------------------------s^7 - 2 s^6 - s^5 + 2 s^4 + 4 s^3 - 8 s^2 - 4 s ans = T(s) Transfer function: 8 --------------------------------------------------s^7 - 2 s^6 - s^5 + 2 s^4 + 4 s^3 - 8 s^2 - 4 s + 8 ans = 212 Chapter 6: Stability Poles of T(s) ans = -1.0000 -1.0000 -1.0000 2.0000 1.0000 1.0000 1.0000 + 1.0000i - 1.0000i + 1.0000i - 1.0000i Thus, there are 4 rhp poles and 3 lhp poles. 15. Even (6): 1 rhp, 1 lhp, 4 jω; Rest (1): 1 lhp; Total: 1 rhp, 2 lhp, 4 jω 16. T (s ) = 18 s + s − 7s − 7s 2 − 18s − 18 5 4 3 s5 1 -7 -18 s4 1 -7 -18 s3 4 -14 0 s2 -3.5 -18 0 s1 -34.57 0 0 s0 -18 0 0 ROZ Even (4): 1 rhp, 1 lhp, 2 jω; Rest (1): 1 lhp; Total: 1 rhp, 2 lhp, 2 jω 17. Gs = Ts = 507 s 4 + 3 s 3 + 10 s 2 + 30 s + 169 ;H s = 1 . Therefore, s G 507 s = 5 + 3 s 4 + 10 s 3 + 30 s 2 + 169 s + 507 1+G H s s5 1 10 169 s4 3 30 507 s3 12 60 0 s2 15 507 0 s1 -345.6 0 0 ROZ Solutions to Design Problems 213 s0 507 0 0 Even (4): 2 rhp, 2 lhp, 0 jω; Rest (1): 0 rhp, 1 lhp, 0 jω; Total (5): 2 rhp, 3 lhp, 0 jω 18. T(s) = K(s2+1) . For a second-order system, if all coefficients are positive, the roots (1+K)s2 + 3s + (2+K) will be in the lhp. Thus, K > -1. 19. K(s+6) T(s) = 3 s + 4s2 + (K+3)s + 6K s3 1 3+K s2 6K s1 4 1 3-2 K s0 6K 0 0 Stable for 0 < K < 6 20. T (s ) = K( s + 3)(s + 5) (1 + K )s + (8K − 6)s + (8 + 15 K ) 2 For 1st column negative For 1st column positive s2 1+K 8+15K K < -1 K > -1 s1 8K-6 0 K < 6/8 K > 6/8 s0 8+15K 0 K < -8/15 K > -8/15 Stable for K > 6/8 21. Program: K=[-6:0.00005:0]; for i=1:length(K); dent=[(1+K(i)) (8*K(i)-6) (8+15*K(i))]; R=roots(dent); A=real(R); B=max(A); if B>0 R K=K(i) break end end K=[6:-0.00005:0]; for i=1:length(K); dent=[(1+K(i)) (8*K(i)-6) (8+15*K(i))]; R=roots(dent); 214 Chapter 6: Stability A=real(R); B=max(A); if B>0 R K=K(i) break end end Computer response: R= 1.0e+005 * 2.7999 -0.0000 K= -1.0000 R= 0.0001 + 3.3166i 0.0001 - 3.3166i K= 0.7500 22. Program: %-det([si() si();sj() sj()])/sj() %Template for use in each cell. syms K %Construct a symbolic object for %gain, K. s2=[(1+K) (8+15*K) 0]; %Create s^2 row of Routh table. s1=[(8*K-6) 0 0]; %Create s^1 row of Routh table. s0=[-det([s2(1) s2(2);s1(1) s1(2)])/s1(1)... -det([s2(1) s2(3);s1(1) s1(3)])/s1(1) 0 0]; %Create s^0 row of Routh table. 's2' %Display label. s2=simplify(s2); %Simplify terms in s^1 row. pretty(s2) %Pretty print s^1 row. 's1' %Display label. s1=simplify(s1); %Simplify terms in s^1 row. pretty(s1) %Pretty print s^1 row. 's0' %Display label. s0=simplify(s0); %Simplify terms in s^0 row. pretty(s0) %Pretty print s^0 row. Computer response: ans = s2 [1 + K ans = 8 + 15 K 0] Solutions to Design Problems 215 s1 [8 K - 6 0 0] ans = s0 [8 + 15 K 23. T (s ) = 0 0 0] K (s + 2)(s − 2) 3 . For positive coefficients in the denominator, − 1 < K < . Hence 2 ( K + 1)s + (3 − 4K ) 4 marginal stability only for this range of K. 24. K(s+1) T(s) = 5 . Always unstable since s3 and s2 terms are missing. s + 2s4 + Ks + K 25. T (s ) = K (s − 2)(s + 4)( s + 5) Ks + (7 K + 1) s2 + 2 Ks + (3 − 40 K ) 3 s3 K s1 54 K − K 7K + 1 s0 2K 3-40K 2 For stability, 0 0 1 3 <K< 54 40 26. K(s+2) T(s) = 4 s + 3s3 - 3s2 + (K+3)s + (2K-4) s4 1 -3 2K - 4 s3 K+ 3 0 2K - 4 0 s1 3 - (K+12) 3 K(K+33) K+12 0 0 s0 2K - 4 0 0 s2 Conditions state that K < -12, K > 2, and K > -33. These conditions cannot be met simultaneously. System is not stable for any value of K. 216 Chapter 6: Stability 27. T (s ) = K s + 80 s + 2001s + ( K + 15390) 3 2 s3 s2 1 2001 80 K+15390 s1 1 14469 − K+ 80 8 s0 K+15390 0 0 There will be a row of zeros at s1 row if K = 144690. The previous row, s2, yields the auxiliary 2 equation, 80s + (144690+ 15390 ) = 0 . Thus, s = ± j44.73. Hence, K = 144690 yields an oscillation of 44.73 rad/s. 28. T(s) = Ks4 - Ks2 + 2Ks + 2K (K+1)s2 + 2(1-K)s + (2K+1) Since all coefficients must be positive for stability in a second-order polynomial, -1 < K < ∞; 1 - ∞< K < 1; -1 < 2K < ∞. Hence, - 2 < K < 1. 29. T (s ) = (s + 2)(s + 7) s + 11s + (K + 31)s2 + (8 K + 21)s + 12 K 4 3 Making a Routh table, s4 1 K + 31 12 K s3 11 8K + 21 0 s2 s1 s0 3K + 320 12 K 11 2 24 K + 1171K + 6720 0 3K + 320 12 K 0 0 0 0 s2 row says –106.7 < K. s1 row says K < -42.15 and –6.64 < K. s0 row says 0 < K. Solutions to Design Problems 217 30. T (s) = K (s + 4) s + 3s + (2 + K )s + 4 K 3 2 Making a Routh table, s3 1 2+K s2 3 4K s1 6-K 0 s0 4K 0 a. For stability, 0 < K < 6. b. Oscillation for K = 6. c. From previous row with K = 6, 3s2 + 24 = 0. Thus s = ± j 8 , or 31. a. G s = K s −1 s +2 s −2 . Therefore, T s s 2+2 s +2 = ω= 8 s −2 rad/s. s −1 K s 3 + K + 4 s 2 + 6− 3 K s +2 K +2 Making a Routh table, s3 1 6-3K s2 4+2K s1 4+K - (3K2+8K-20) K+4 0 s0 4+2K 0 From s1 row: K = 1.57, -4.24; From s2 row: - 4 < K; From s0 row: - 2 < K. Therefore, - 2 < K < 1.57. b. If K = 1.57, the previous row is 5.57s2 + 7.14. Thus, s = ± j1.13. c. From part b, ω = 1.13 rad/s. 32. Applying the feedback formula on the inner loop and multiplying by K yields Ge s s Thus, Ts Making a Routh table: K = = s 2+5 s +7 K s 3+5 s 2+7 s +K . 218 Chapter 6: Stability s3 1 7 s2 K s1 5 35 - K 5 s0 K 0 0 For oscillation, the s1 row must be a row of zeros. Thus, K = 35 will make the system oscillate. The previous row now becomes, 5s2 + 35. Thus, s2 + 7 = 0, or s = ± j 7 . Hence, the frequency of oscillation is 7 rad/s. 33. Ks2 + 2Ks T(s) = 3 s + (K-1)s2 + (2K-4)s + 24 s3 s2 1 2K-4 K-1 24 s1 2K2 − 6K − 20 K −1 s0 24 0 0 For stability, K > 5; Row of zeros if K = 5. Therefore, 4s2 + 24 = 0. Hence, ω = 6 for oscillation. 34. Program: K=[0:0.001:200]; for i=1:length(K); deng=conv([1 -4 8],[1 3]); numg=[0 K(i) 2*K(i) 0]; dent=numg+deng; R=roots(dent); A=real(R); B=max(A); if B<0 R K=K(i) break end end Computer response: R= -4.0000 -0.0000 + 2.4495i Solutions to Design Problems 219 -0.0000 - 2.4495i K= 5 a. From the computer response, (a) the range of K for stability is 0 < K < 5. b. The system oscillates at K = 5 at a frequency of 2.4494 rad/s as seen from R, the poles of the closed-loop system. 35. K(s+2) T(s) = 4 3 - 3s2 + (K+3)s + (2K-4) s + 3s s4 1 -3 2K-4 s3 K+3 0 2K-4 0 s1 3 K+12 -3 K(K+33) K+12 0 0 s0 2K-4 0 0 s2 For K < -33: 1 sign change; For –33 < K < -12: 1 sign change; For –12 < K < 0: 1 sign change; For 0 < K < 2: 3 sign changes; For K > 2: 2 sign changes. Therefore, K > 2 yields two right-half-plane poles. 36. K T(s) = 4 s + 7s3 + 15s2 + 13s + (4+K) s4 1 15 K+4 s3 13 0 K+4 0 s1 7 92 7 1000-49K 92 0 0 s0 K+4 0 0 s2 a. System is stable for - 4 < K < 20.41. 92 b. Row of zeros when K = 20.41. Therefore, 7 s2 + 24.41. Thus, s = ± j1.3628, or ω = 1.3628 rad/s. 37. K T(s) = 3 s + 14s2 + 45s + (K+50) 220 Chapter 6: Stability s3 1 45 s2 K+50 s1 14 580-K 14 s0 K+50 0 0 a. System is stable for -50 < K < 580. b. Row of zeros when K = 580. Therefore, 14s2 + 630. Thus, s = ± j 45 , or ω = 6.71 rad/s. 38. K T(s) = 4 s + 8s3 + 17s2 + 10s + K s4 1 17 K s3 10 0 K 0 s1 8 126 8 32 - 63 K + 10 0 0 s0 K 0 0 s2 a. For stability 0 < K < 19.69. 126 b. Row of zeros when K = 19.69. Therefore, 8 s2 + 19.69. Thus, s = ± j 1.25 , or ω = 1.118 rad/s. c. Denominator of closed-loop transfer function is s4 + 8s3 + 17s2 + 10s + K. Substituting K = 19.69 and solving for the roots yield s = ± j1.118, -4.5, and -3.5. 39. K(s2 + 2s + 1) T(s) = 3 s + 2s2 + (K+1)s - K s3 K+1 s2 2 3K+2 2 -K s1 s0 2 Stability if - 3 < K < 0. 1 -K 0 0 Solutions to Design Problems 221 40. T(s) = 2s4 + (K+2)s3 + Ks2 s3 + s2 + 2s + K s3 1 2 s2 1 K s1 2-K 0 s0 K 0 Row of zeros when K = 2. Therefore s2 + 2 and s = ± j 2 , or ω = 1.414 rad/s. Thus K = 2 will yield the even polynomial with 2 jω roots and no sign changes. 41. 1 s3 s2 s1 K2 1 K1 5 0 1 0 0 0 0 0 K1 K2 − 5 K1 2 K 1 − 5K 1K 2 + 25 5 − K1K 2 s0 1 For stability, K1K2 > 5; K12 + 25 < 5K1K2 ; and K1 > 0 . Thus 0 < K12 < 5K1K2 - 25, or 0 < K 1 < 5K1 K2 − 25 . 42. s4 1 1 1 s3 K2 0 s2 K1 K1-K2 K1 1 0 s1 K12 - K1K2 + K22 K2 - K1 0 0 s0 1 0 0 For two jω poles, K12 - K1K2 + K22 = 0. However, there are no real roots. Therefore, there is no relationship between K1 and K2 that will yield just two jω poles. 222 Chapter 6: Stability 43. s8 1 1.18E+03 2.15E+03 -1.06E+04 -415 s7 103 4.04E+03 -8.96E+03 -1.55E+03 0 s6 1140.7767 2236.99029 -10584.951 -415 0 s5 3838.02357 -8004.2915 -1512.5299 0 0 s4 4616.10784 -10135.382 -415 0 0 s3 422.685462 -1167.4817 0 0 0 s2 2614.57505 -415 0 0 0 s1 -1100.3907 0 0 0 0 s0 -415 0 0 0 0 a. From the first column, 1 rhp, 7 lhp, 0 jω. b. G(s) is not stable because of 1 rhp. 44. Eigenvalues are the roots of the following equation: ⎡ s 0 0⎤ ⎡ 0 1 3⎤ ⎡s −1 −3 ⎤ ⎢ ⎥⎢ ⎥⎢ ⎥ sI − A = ⎢ 0 s 0⎥ − ⎢ 2 2 − 4⎥ = ⎢ −2 s − 2 4 ⎥ = s3 − 5s 2 − 15s + 40 ⎢ ⎥⎢ ⎥⎢ ⎥ ⎣0 0 s⎦ ⎣1 −4 3 ⎦ ⎣−1 4 s − 3⎦ Hence, eigenvalues are -3.2824, 1.9133, 6.3691. Therefore, 1 rhp, 2 lhp, 0 jω. 45. Program: A=[0 1 0;0 1 -4;-1 1 3]; eig(A) Computer response: ans = 1.0000 1.5000 + 1.3229i 1.5000 - 1.3229I 46. Writing the open-loop state and output equations we get, Solutions to Design Problems 223 • x1 = x 2 • x2 = x 2 + 3x 3 • x3 = − 3x1 − 4 x 2 − 5x 3 + u y = x2 + x3 Drawing the signal-flow diagram and including the unity feedback path yields, 1 1 1 1 3 s 1 r x 3 -5 1 1 s 1 x 1 s 2 -4 -3 -1 Writing the closed-loop state and output equations from the signal-flow diagram, • x1 = x 2 • x2 = x 2 + 3x 3 • x3 = − 3x1 − 4 x 2 − 5x 3 + r − c = −3 x1 − 4x 2 − 5 x3 + r − ( x2 + x3 ) = −3 x1 − 5 x2 − 6x 3 + r y = x2 + x3 In vector-matrix form, 1 0⎤ ⎡0 ⎡ 0⎤ x = ⎢0 1 3 ⎥ X + ⎢ 0⎥ r ⎢− 3 − 5 −6 ⎥ ⎢ 1⎥ ⎦ ⎣ ⎣⎦ • y = [0 1 1]x Now, find the characteristic equation. x c=y 1 224 Chapter 6: Stability −1 0⎤ ⎡ s 0 0⎤ ⎡ 0 1 0 ⎤ ⎡ s −3 ⎥ sI − A = ⎢0 s 0⎥ − ⎢ 0 1 3 ⎥ = ⎢ 0 (s − 1) ⎢0 0 s ⎥ ⎢ − 3 −5 − 6⎥ ⎢ 3 5` (s + 6)⎥ ⎦⎣ ⎦ ⎦⎣ ⎣ = s3 + 5s2 + 9s + 9 Forming a Routh table to determine stability s3 s2 s1 s0 1 5 36 5 9 9 9 0 0 Since there are no sign changes, the closed-loop system is stable. 47. Program: A=[0,1,0;0,1,3;-3,-4,-5]; B=[0;0;1]; C=[0,1,1]; D=0; 'G' G=ss(A,B,C,D) 'T' T=feedback(G,1) 'Eigenvalues of T' ssdata(T); eig(T) Computer response: ans = G a= x1 x2 x3 x1 0 0 -3 x2 1 1 -4 x3 0 3 -5 x2 1 x3 1 b= x1 x2 x3 u1 0 0 1 c= y1 x1 0 d= u1 Solutions to Design Problems 225 y1 0 Continuous-time model. ans = T a= x1 x2 x3 x1 0 0 -3 x2 1 1 -5 x3 0 3 -6 x2 1 x3 1 b= x1 x2 x3 u1 0 0 1 c= y1 x1 0 d= y1 u1 0 Continuous-time model. ans = Eigenvalues of T ans = -1.0000 + 1.4142i -1.0000 - 1.4142i -3.0000 SOLUTIONS TO DESIGN PROBLEMS 48. K(s+1)(s+10) T(s) = 3 s + (5.45+K)s2 + (11.91+11K)s + (43.65+10K) s3 1 11.91+11K s2 5.45+K 43.65+10K s1 s0 0 43.65+10K 0 226 Chapter 6: Stability For stability, - 0.36772 < K < ∞. Stable for all positive K. 49. 0.7K(s+0.1) T(s) = 4 3 + 1.14s2 + 0.193s + (0.07K+0.01) s + 2.2s s4 1 1.14 0.07K+0.01 s3 2.2 0.193 0 s2 1.0523 0.07K+0.01 0 s1 0.17209 - 0.14635K 0 0 s0 0.07K+0.01 0 0 For stability, - 0.1429 < K < 1.1759 50. Ts = 0.6 K + 10 K s 2 + 60.1 K s s 5 + 130 s 4 + 3229 s 3 + 10 K + 2348 s 2 + 60.1 K + 58000 s + 0.6 K s5 1 3229 60.1K+58000 s4 130 10K+23480 0.6K -10K+396290 7812.4K+7540000 0 s3 -100K2+2712488K+8.3247E9 -10K+396290 0.6K 0 s1 7813E3K4-5.1401E11K3+7.2469E15K2+3.3213E19K+2.4874E22 1000K3-66753880K2+9.9168E11K+3.299E15 0 0 s0 0.6K 0 0 s2 Note: s3 row was multiplied by 130 From s1 row after canceling common roots: 7813000 K − 39629 K + 967.31586571671 K + 2776.9294183336 K − 29908.070615165 1000 K − 39629 K + 2783.405672635 K − 29908.285672635 From s0 row: K>0 From s3 row: K < 39629 From s2 row: K < 29908.29; 39629 < K From s1 row: 29908.29 < K, or K < 29908.07; Solutions to Design Problems 227 Therefore, for stability, 0 < K < 29908.07 51. s5 1 1311.2 1000(100K+1) s4 112.1 10130 60000K s3 1220.8 99465K+1000 0 s2 10038-9133.4K 60000K 0 s1 99465 (K+0.010841)(K-1.0192) (K-1.0991) 0 0 s0 60000K 0 0 From s2 row: K < 1.099 From s1 row: -0.010841 < K < 1.0192; K > 1.0991 From s0 row: 0 < K Therefore, 0 < K < 1.0192 52. Find the closed-loop transfer function. G(s) = 63x106 K ( s + 30)( s + 140)( s + 2.5) T (s ) = G(s) 63x10 6 K =3 1 + G(s) H( s) s + 172.5s2 + 4625s + (10500 + 63x10 6 K ) Make a Routh table. s3 1 4625 s2 172.5 1 0 5 0 0 + 6 3 x 1 0 6K s1 4564.13-365217.39K 0 s0 10500+63x106K 0 The s1 line says K < 1.25x10-2 for stability. The s0 line says K > -1.67x10-4 for stability. 228 Chapter 6: Stability Hence, -1.67x10-4 < K < 1.25x10-2 for stability. 53. Find the closed-loop transfer function. 7570 Kp(s + 103)(s + 0.8) s( s + 62.61)( s − 62.61) G (s ) 7570 Kp(s + 103)(s + 0.8) T (s ) = =3 2 1 + G( s)H( s) s + 7570K p s + (785766 Kp − 3918.76 )s + 623768K p G (s ) = Make a Routh table. s3 1 785766Kp – 3918.76 s2 7570 623768Kp s1 785766Kp – 4001.16 0 s0 623768Kp 0 The s1 line says Kp > 5.09x10-3 for stability. The s0 line says Kp > 0 for stability. Hence, Kp > 5.09x10-3 for stability. 54. Yh-Ycat Spring displacement Desired force Input voltage+ 1 100 Input transducer K Controller - 1 1000 F up 0 .7883 ( s + 53 .85 ) ( s2 + 15 .47 s + 9283 )( s 2 + 8 .119 s + 376 .3 ) Actuator 82300 Pantograph dynamics F out Spring 1 100 Sensor Y h-Ycat Spring displacement Desired force + 100 1 1000 Controller Actuator K - G(s) = Fup 0 .7883 ( s + 53 .85 ) (s 2 + 15 .47 s + 9283 )( s 2 + 8 .119 s + 376 .3 ) Pantograph dynamics 0.7883( s + 53.85) Yh ( s) − Ycat (s) =2 (s + 15.47s + 9283)(s2 + 8.119 s + 376.3) Fup ( s) Ge(s)=(K/100)*(1/1000)*G(s)*82.3e3 82300 Spring Fout Solutions to Design Problems 229 0.6488K (s+53.85) Ge(s) = (s2 + 8.119s + 376.3) (s2 + 15.47s + 9283) 0.6488K (s+53.85) T(s) = ___________________________________________________________________________________________________________________________ 4 3 2 7 s + 23.589 s + 9784.90093 s + (0.6488 K + 81190.038 )s + (34.94 K + 0.34931929 10 ) s4 1 9785 (0.3493e7+34.94K) + s3 23.59 (0.6488K+81190) 0 + s2 (-0.0275K+6343) (0.3493e7+34.94K 0 K<230654 s1 −0.0178 2 + 10587K + 432.59e6 K . K −.0275 + 6343 0 -128966<K<188444 s0 (0.3493e7+34.94K) 0 -99971<K ) The last column evaluates the range of K for stability for each row. Therefore –99971 < K < 188444. SEVEN Steady-State Errors SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Steady-State Error Design via Gain 76.39K a. G(s) = s(s+150)(s+1.32) . System is Type 1. Step input: e(∞) = 0; Ramp input: 1 1 2.59 = 76.39K = K ; Parabolic input: e(∞) = ∞. 150 x 1.32 1 2.59 b. K = K = 0.2. Therefore, K = 12.95. Now test the closed-loop transfer function, v 989.25 T(s) = 3 , for stability. Using Routh-Hurwitz, the system is stable. s +151.32s2+198s+989.25 e( ∞ ) = K v s3 1 198 s2 151.32 989.25 s1 191.46253 0 s0 989.25 0 Video Laser Disc Recorder: Steady-State Error Design via Gain a. The input, 15t2 , transforms into 30/s3. e(∞) = 30/Ka = 0.005. 30 0.2*600 Ka = 20000 * K1K2K3 = 6x10-3 K1K2K3. Therefore: e(∞) = 30/Ka = −3 6x10 K 1 K 2 K 3 = 5x10-3. Therefore K1K2K3 = 106. b. Using K1K2K3 = 106, G(s) = 2x10 5 (s + 600) . Therefore, T(s) = 2 4 s (s + 2x10 ) 2x10 5 (s + 600) . s 3 + 2x104 s 2 + 2x10 5 s + 1.2x10 8 Making a Routh table, Answers to Review Questions 231 s3 1 2x105 s2 2x104 1.2x108 s1 194000 0 s0 120000000 0 we see that the system is stable. c. Program: numg=200000*[1 600]; deng=poly([0 0 -20000]); G=tf(numg,deng); 'T(s)' T=feedback(G,1) poles=pole(T) Computer response: ans = T(s) Transfer function: 200000 s + 1.2e008 -----------------------------------s^3 + 20000 s^2 + 200000 s + 1.2e008 poles = 1.0e+004 * -1.9990 -0.0005 + 0.0077i -0.0005 - 0.0077I ANSWERS TO REVIEW QUESTIONS 1. Nonlinear, system configuration 2. Infinite 3. Step(position), ramp(velocity), parabola(acceleration) 4. Step(position)-1, ramp(velocity)-2, parabola(acceleration)-3 5. Decreases the steady-state error 6. Static error coefficient is much greater than unity. 7. They are exact reciprocals. 8. A test input of a step is used; the system has no integrations in the forward path; the error for a step input is 1/10001. 9. The number of pure integrations in the forward path 10. Type 0 since there are no poles at the origin 232 Chapter 7: Steady-State Errors 11. Minimizes their effect 12. If each transfer function has no pure integrations, then the disturbance is minimized by decreasing the plant gain and increasing the controller gain. If any function has an integration then there is no control over its effect through gain adjustment. 13. No 14. A unity feedback is created by subtracting one from H(s). G(s) with H(s)-1 as feedback form an equivalent forward path transfer function with unity feedback. 15. The fractional change in a function caused by a fractional change in a parameter 16. Final value theorem and input substitution methods SOLUTIONS TO PROBLEMS 1. e (∞ ) = s R(s ) lim s E(s ) = lim s →0 s →0 1+G(s ) where 450(s + 12)(s + 8)(s + 15) . s(s + 38)(s2 + 2s + 28) 37 For step, e (∞) = 0. For 37tu(t) , R(s) = 2 . Thus, e (∞) = 6.075x10-2. For parabolic input, s G(s) = e(∞) = ∞ 2. s R(s ) lim s E(s ) = lim 1+G(s ) s →0 s →0 s(60/s 3 ) = lim = 0.9375 20(s + 3)(s + 4)(s + 8) s→ 0 1+ 2 s (s + 2)(s + 15) e (∞ ) = 3. Reduce the system to an equivalent unity feedback system by first moving 1/s to the left past the ⎛1 ⎞ + 1⎟ in cascade ⎝s ⎠ summing junction. This move creates a forward path consisting of a parallel pair, ⎜ with a feedback loop consisting of G (s ) = 2 and H(s) = 7 . Thus, s+3 ⎛ (s + 1⎞ ⎛ 2 /( s + 3) ⎞ 2( s + 1) ⎟= ⎟⎜ Ge (s) = ⎜ ⎝ s ⎠ ⎝ 1 + 14 /(s + 3) ⎠ s( s + 17) Hence, the system is Type 1 and the steady-state errors are as follows: Steady-state error for 15u(t ) = Steady-state error for 15tu(t ) = 0. 15 15 = = 127.5 . K v 2 / 17 Solutions to Design Problems 233 2 Steady-state error for 15t u(t ) = ∞ 4. System is type 0. Kp = 5 . 2 40 80 = = 11.43 1 + Kp 7 For 70tu(t), e(∞) = ∞ For 40u(t), e(∞) = For 80t2u(t), e(∞) = ∞ 5. R( s) 72 / S4 E( s) = = 1 + G (s) 1 + 200( S + 2)(S + 5)(S + 7)(S + 9) 3 S ( S + 3)(S + 10)( S + 15) Thus, 72 e( ∞ ) = lim sE(s) = ( 200)(2)( 5)( 7)(9) = 0.2571 s→ 0 (3)(10)(15) 6. de = s E (s ) dt 6 s4 9 100(s+1)(s+2) = 10 . s∅0 1 + s2(s+10)(s+3) . R(s) Therefore, e(∞) = lim s2E(s) = lim s2 1+G(s) = lim s∅ 0 7. e( ∞ ) = s∅ 0 s2 1000(12 )(25)(32) 15 = 24.78 . Therefore, e(∞) = 0.582. ; Kp = (61)(73)(87) 1 + Kp 8. 8 For 8u(t), ess = 1+K p = 2; For 8tu(t), ess = ∞, since the system is Type 0. 9. a. The closed-loop transfer function is, T (s ) = 5000 s + 75s + 5000 2 5000 and 2ζωn = 75. Thus, ζ = 0.53 and from which, ωn = 2 %OS = e −ζπ / 1 −ζ x100 = 14.01%. 4 4 b . Ts = = = 0.107 second. ζω n 75 / 2 c. Since system is Type 1, ess for 5u(t) is zero. 234 Chapter 7: Steady-State Errors d. Since Kv is 5 5000 = 66.67, ess = = 0.075. Kv 75 e. ess = ∞, since system is Type 1. 10. Kv = lim sG(s) = s→ 0 105 (3)(10)(20) 4 = 10 (25)(α )(30) Thus, α = 8. 11. K a = lim s 2 G ( s ) = s →0 Kx 2 x 4 x 6 = 10,000 . Therefore, K = 7291.667. 5 x7 12. 5 5 s(s + 1)(s + 2) a. Ge(s) = 5(s + 3) = s 3 + 3s 2 + + 7s + 15 1+ s(s + 1)(s + 2) Therefore, Kp = 1/3; Kv = 0; and Ka = 0. 50 b. For 50u(t), e(∞) = 1 + K = 37.5; For 50tu(t), e(∞) = ∞; For 50t2u(t), e(∞) = ∞ p c. Type 0 13. R( s) E( s) = . Thus, e( ∞ ) = lim sE(s ) = lim s→ 0 s→ 0 1 + G(s) 6 s4 2 1000(s + 4s + 20 )(s2 + 20s + 15) 1+ s 3 (s + 2)(s + 10 ) s = 4x10-4. 14. Collapsing the inner loop and multiplying by 1000/s yields the equivalent forward-path transfer function as, 10 5 (s + 2) Ge (s) = 2 s( s + 1005s + 2000 ) Hence, the system is Type 1. 15. • e(∞)= lim s 2 E(s) = lim s 2 s →0 s →0 R( s) . 1 + G(s) Solutions to Design Problems 235 • s 1 For Type 0, step input: R(s) = s , and e(∞) = lim =0 s →0 1 + G ( s ) 1 For Type 0, ramp input: R(s) = 2 , and s • 1 = s →0 1 + G ( s ) e(∞)= lim 1 1 = 1 + lim G ( s ) 1 + K p s →0 • 1 1 For Type 0, parabolic input: R(s) = 3 , and e(∞)= lim =∞ s s → 0 s + sG ( s ) • s 1 For Type 1, step input: R(s) = s , and e(∞)= lim =0 s →0 1 + G ( s ) • 1 1 For Type 1, ramp input: R(s) = 2 , and e(∞) = lim =0 s s →0 1 + G ( s ) • 1 1 1 For Type 1, parabolic input: R(s) = 3 , and e(∞) = lim =K v s s → 0 s + sG ( s ) • s 1 For Type 2, step input: R(s) = s , and e(∞)= lim =0 s →0 1 + G ( s ) • 1 1 For Type 2, ramp input: R(s) = 2 , and e(∞) = lim =0 s s →0 1 + G ( s ) • 1 1 For Type 2, parabolic input: R(s) = 3 , and e(∞) = lim =0 s s → 0 s + sG ( s ) 16. a. e(∞ )= 1 /10 7K = 0.01; where K v = = 10. Thus, K = 685.71 . 5 x8 x12 Kv b. Kv = 10. 236 Chapter 7: Steady-State Errors c. The minimum error will occur for the maximum gain before instability. Using the Routh-Hurwitz Criterion along with K ( s + 7) : s + 25s + 196s 2 + ( 480 + K ) s + 7 K T ( s) = 4 3 s4 1 196 s3 25 480+K s2 4420-K 175K s1 − K 2 − 435 K + 2121600 7K For Stability K < 4420 -1690.2 < K < 1255.2 s0 175K K>0 Thus, for stability and minimum error K = 1255.2. Thus, K v = e(∞ )= 7K = 18.3 and 5 x8 x12 1 / 10 1 / 10 = = 0.0055 . 18.3 Kv 17. 15 e( ∞ ) = K v 15 150 = Ka/10 = Ka = 0.003. Hence, Ka = 50,000. 18. K s(s+1) Find the equivalent G(s) for a unity feedback system. G(s) = 10 1 + s+1 100 100 Kv = K/11 = 0.01; from which K = 110,000. K = s(s+11) . Thus, e(∞) = 19. 20 2K Ka = 4 . e(∞) = K = 0.01. Hence, K = 4000. a 20. 10 1 30K a. e(∞) = K = 6000 . But, Kv = 5 = 60,000. Hence, K = 10,000. For finite error for a ramp v input, n = 1. b. Kp = lim G(s) = lim s∅ 0 10000(s2+3s+30) Kv = lim sG(s) = lim s s∅ 0 s(s+5) s∅ 0 10000(s2+3s+30) s∅ 0 Ka = lim s2G(s) = lim s2 s∅ 0 21. s∅ 0 =∞ s(s+5) 10000(s2+3s+30) s(s+5) = 60,000 =0 Solutions to Design Problems 237 a. Type 0 R(s) b. E(s) = 1 + G(s) . Thus, e(∞) = lim sE(s) = lim s s∅ 0 s∅ 0 10/s 120 2+2s+5) = 12+5K . K(s 1 + (s+2)2(s+3) c. e(∞) = ∞, since the system is Type 0. 22. e(∞ ) = 25 25 = = 0.1. Thus, K = 700 . K v 150 K / 420 23. 50 e( ∞ ) = 1 + K = p 50 = 0.05. Thus, K = 6493.5. 4K 1+ 26 24. The system is stable for 0 < K < 2000. Since the maximum Kv is Kv = 1 minimum steady-state error is K v = 1 = 0.16. 6.25 K 2000 = 6.25, the = 320 320 25. To meet steady-state error characteristics: Therefore, Kα = 9β2. K(s+α) , To meet the transient requirement: Since T(s) = 2 s + (K+2β)s + (β2 + Kα) ωn2 = 10 = β2 + Kα ; 2ζωn = 10 = K+2β. Solving for β, β = ±1. For β = +1, K = 1.16 and α = 7.76. An alternate solution is β = -1, K = 5.16, and α = 1.74. 26. a. System Type = 1 K 1 1 K . Therefore, e(∞) = K = = 0.01, or = 100. s(s+α) K/α α v G(s) K But, T(s) = 1 + G(s) = 2 . s +αs+K 100 Since ωn = 10, K = 100, and α = 1. Hence, G(s) = s(s+1) . 1 c. 2ζωn = α = 1. Thus, ζ = 20 . b. Assume G(s) = 27. 238 Chapter 7: Steady-State Errors G(s) K(s+α) T(s) = 1 + G(s) = 2 . Hence, K+β = 2, Kα = ωn2 = (12+12) = 2. s +(K+β) s+αK 1 β Also, e(∞) = K = = 0.1. Therefore, β = 0.1Kα = 0.2, K = 1.8, and α = 1.111. Kα v 28. G(s) K K System Type = 1. T(s) = 1 + G(s) = 2 . From G(s), Kv = a = 100. For 10% overshoot, ζ = s +as+K 0.6. Therefore, 2ζωn = a , and ωn2 = K. Hence, a = 1.2 K . K Also, a = 100 Solving simultaneously, K = 1.44 x 104, and a = 1.44 x 102. 29. K K a. For 20% overshoot, ζ = 0.456. Also, Kv = 1000 = a . Since T(s) = 2 , 2ζωn = a, and s +as+K ωn = K . Hence, a = 0.912 K . Solving for a and K, K = 831,744, and a = 831.744. 1 K K b. For 10% overshoot, ζ = 0.591. Also, K = 0.01. Thus, Kv = 100 = a . Since T(s) = 2 , v s +as+K 2ζωn = a, and ωn = K . Hence, a = 1.182 K . Solving for a and K, K = 13971 and a = 139.71. 30. a. For the inner loop: 1 s2(s+1) s G1(s) = =4 3 1 s +s +1 1+ 3 s (s+1) 1 1 Ge(s) = 2 G (s) = 5 4 3 s (s+3) 1 s(s +4s +3s +s+3) Ge(s) 1 T(s) = 1+G (s) = 6 5 4 2 e s +4s +3s +s +3s+1 b. From Ge(s), system is Type 1. c. Since system is Type 1, ess = 0 1 5 d. ; From Ge(s), Kv = lim sGe (s ) = 3 . Therefore, ess = K = 15. v s→ 0 e. Poles of T(s) = -3.0190, -1.3166, 0.3426 ± j0.7762, -0.3495. Therefore, system is unstable and results of (c) and (d) are meaningless 31. a. For the inner loop: 10 s(s+1)(s+3)(s+4) 10 G1(s) = =32 20 s(s +8s +19s+32) 1 + (s+1)(s+3)(s+4) Solutions to Design Problems 239 Ge(s) = 20 s(s3+8s2+19s+32) Ge(s) 20 T(s) = 1+G (s) = 4 3 e s +8s +19s2+32s+20 b. From Ge(s), system is Type 1. c. Since system is Type 1, ess = 0 20 5 5 d. From Ge(s), Kv = lim sGe (s ) = 32 = 8 . Therefore, ess = K = 8. v s→ 0 e. Poles of T(s) = -5.4755, -0.7622 ± j1.7526, -1. Therefore, system is stable and results of parts c and d are valid. 32. Program: numg1=[1 7];deng1=poly([0 -4 -8 -12]); 'G1(s)=' G1=tf(numg1,deng1) numg2=5*poly([-9 -13]);deng2=poly([-10 -32 -64]); 'G2(s)=' G2=tf(numg2,deng2) numh1=10;denh1=1; 'H1(s)=' H1=tf(numh1,denh1) numh2=1;denh2=[1 3]; 'H2(s)=' H2=tf(numh2,denh2) %Close loop with H1 and form G3 'G3(s)=G2(s)/(1+G2(s)H1(s)' G3=feedback(G2,H1) %Form G4=G1G3 'G4(s)=G1(s)G3(s)' G4=series(G1,G3) %Form Ge=G4/1+G4H2 'Ge(s)=G4(s)/(1+G4(s)H2(s))' Ge=feedback(G4,H2) %Form T(s)=Ge(s)/(1+Ge(s)) to test stability 'T(s)=Ge(s)/(1+Ge(s))' T=feedback(Ge,1) 'Poles of T(s)' pole(T) %Computer response shows that system is stable. Now find error specs. Kp=dcgain(Ge) 'sGe(s)=' sGe=tf([1 0],1)*Ge; 'sGe(s)' sGe=minreal(sGe) Kv=dcgain(sGe) 's^2Ge(s)=' s2Ge=tf([1 0],1)*sGe; 's^2Ge(s)' s2Ge=minreal(s2Ge) Ka=dcgain(s2Ge) essstep=30/(1+Kp) essramp=30/Kv essparabola=60/Ka Computer response: ans = 240 Chapter 7: Steady-State Errors G1(s)= Transfer function: s+7 -----------------------------s^4 + 24 s^3 + 176 s^2 + 384 s ans = G2(s)= Transfer function: 5 s^2 + 110 s + 585 -----------------------------s^3 + 106 s^2 + 3008 s + 20480 ans = H1(s)= Transfer function: 10 ans = H2(s)= Transfer function: 1 ----s+3 ans = G3(s)=G2(s)/(1+G2(s)H1(s) Transfer function: 5 s^2 + 110 s + 585 Solutions to Design Problems 241 -----------------------------s^3 + 156 s^2 + 4108 s + 26330 ans = G4(s)=G1(s)G3(s) Transfer function: 5 s^3 + 145 s^2 + 1355 s + 4095 ------------------------------------------------------s^7 + 180 s^6 + 8028 s^5 + 152762 s^4 + 1.415e006 s^3 + 6.212e006 s^2 + 1.011e007 s ans = Ge(s)=G4(s)/(1+G4(s)H2(s)) Transfer function: 5 s^4 + 160 s^3 + 1790 s^2 + 8160 s + 12285 ------------------------------------------------------s^8 + 183 s^7 + 8568 s^6 + 176846 s^5 + 1.873e006 s^4 + 1.046e007 s^3 + 2.875e007 s^2 + 3.033e007 s + 4095 ans = T(s)=Ge(s)/(1+Ge(s)) Transfer function: 5 s^4 + 160 s^3 + 1790 s^2 + 8160 s + 12285 ------------------------------------------------------s^8 + 183 s^7 + 8568 s^6 + 176846 s^5 + 1.873e006 s^4 + 1.046e007 s^3 + 2.875e007 s^2 + 3.034e007 s + 16380 242 Chapter 7: Steady-State Errors ans = Poles of T(s) ans = -124.7657 -21.3495 -12.0001 -9.8847 -7.9999 -4.0000 -2.9994 -0.0005 Kp = 3 ans = sGe(s)= ans = sGe(s) Transfer function: 5 s^5 + 160 s^4 + 1790 s^3 + 8160 s^2 + 1.229e004 s --------------------------------------------------------s^8 + 183 s^7 + 8568 s^6 + 1.768e005 s^5 + 1.873e006 s^4 + 1.046e007 s^3 + 2.875e007 s^2 + 3.033e007 s + 4095 Kv = 0 ans = Solutions to Design Problems 243 s^2Ge(s)= ans = s^2Ge(s) Transfer function: 5 s^6 + 160 s^5 + 1790 s^4 + 8160 s^3 + 1.229e004 s^2 --------------------------------------------------------s^8 + 183 s^7 + 8568 s^6 + 1.768e005 s^5 + 1.873e006 s^4 + 1.046e007 s^3 + 2.875e007 s^2 + 3.033e007 s + 4095 Ka = 0 essstep = 7.5000 Warning: Divide by zero. (Type "warning off MATLAB:divideByZero" to suppress this warning.) > In D:\My Documents\Control Systems Engineering Book\CSE 4th ed\Solutions Manual\Chap 7 References\p7_32.m at line 40 essramp = Inf Warning: Divide by zero. (Type "warning off MATLAB:divideByZero" to suppress this warning.) > In D:\My Documents\Control Systems Engineering Book\CSE 4th ed\Solutions Manual\Chap 7 References\p7_32.m at line 41 essparabola = Inf 244 Chapter 7: Steady-State Errors 33. 10K1 The equivalent forward transfer function is, G(s) = s(s+1+10K ) . f 10K1 10K1 G(s) Also, T(s) = 1 + G(s) = s2+(10K +1)s+10K . From the problem statement, Kv = 1+10K = 10. f 1 f Also, 2ζωn = 10Kf+1 = 2(0.5) 10K1 = 10K1 . Solving for K1 and Kf simultaneously, K1 = 10 and Kf = 0.9. 34. e(∞) = lim s∅ 0 sR(s) - sD(s)G2(s) 1 + G1(s)G2(s) 1 100 , where G1(s) = s+5 and G2 = s+2 . From the problem statement, 1 R(s) = D(s) = s . Hence, e(∞) = lim 100 1 - s+2 49 100 = - 11 . s∅ 0 1 + 1 s+5 s+2 35. Error due only to disturbance: Rearranging the block diagram to show D(s) as the input, Therefore, K2 K2(s+3) s(s+4) -E(s) = D(s) K1K2(s+2) = D(s) s(s+3)(s+4) + K1K2(s+2) 1 + s(s+3)(s+4) 3 1 For D(s) = s , eD(∞) = lim sE(s) = - 2K . 1 s∅ 0 1 1 6 Error due only to input: eR(∞) = K = K K = K K . v 12 12 6 Design: 3 eD(∞) = - 0.000012 = - 2K , or K1 = 125,000. 1 Solutions to Design Problems 245 6 eR(∞) = 0.003 = K K , or K2 = 0.016 12 36. C( s ) G1 ( s)G2 ( s ) E (s ) G1 (s) = ; ∴ a1 = R( s) 1 + G2 (s ) H1 (s) R(s) 1 + G2 (s) H1 ( s) ea 1 (∞ ) = lim s →0 sR(s)G1 (s) 1 + G2 (s ) H1 (s) 37. System 1: Forming a unity-feedback path, the equivalent unity feedback system has a forward transfer function of 10( s + 10) 10( s + 10) s(s + 2) Ge (s ) = = 2 10(s + 10)( s + 3) 11s + 132s + 300 1+ s (s + 2) a. Type 0 System; b. Kp = K p = lim Ge (s ) = 1/ 3 ; c. step input; d. e(∞) = s→ 0 1 = 3/4; 1 + Kp ⎛ 1⎞ ⎝ s⎠ sR( s) e. ea −step (∞ ) = lim = lim = 0. 10( s + 10 )(s + 4) s → 0 1 + G (s ) H ( s ) s→ 0 1+ s( s + 2) s System 2: Forming a unity-feedback path, the equivalent unity feedback system has a forward transfer function of 10(s + 10) 10(s + 10) s( s + 2) Ge (s) = 10(s + 10)s = s(11s + 102) 1+ s(s + 2) a. Type 1 System; b. Kv = lim sGe (s ) = 0.98 ; c. ramp input; d. e( ∞ ) = s→ 0 ⎛1 s⎝ 2 ⎞ ⎠ sR(s) 1 e. ea −ramp (∞ ) = lim = lim 10(s +s10)(s + 1) = . s →0 1 + G (s ) H (s ) s →0 50 1+ s(s + 2) 38. System 1. Push 5 to the right past the summing junction: 1 = 1.02 ; Kv 246 Chapter 7: Steady-State Errors + R(s) 5(s+4) C(s) (s+ 5)(s+ 8) - 2 Produce a unity-feedback system: 5(s+4) R(s) + - - C(s) (s + 5) (s + 8) 1 5(s + 4) 5(s + 4) 1 (s + 5)( s + 8) 1 =2 . Kp = . estep = 1+K = 0.75, eramp = ∞, Thus, Ge (s ) = 5(s + 4) p s + 18s + 60 3 1+ (s + 5)( s + 8) eparabola = ∞. 5( s + 4) Checking for stability, from first block diagram above, T(s) = 2 . The system is stable. s + 23s + 80 System 2. Push 20 to the right past the summing junction and push 10 to the left past the pickoff point: R(s) 200(s+4) + (s+5)(s+8) - 1 40 C(s) Solutions to Design Problems 247 Produce a unity-feedback system: + R(s) - 200(s+4) C(s) (s+5)(s+8) - -39 40 200( s + 4) 200( s + 4) 200( 4) ( s + 5)(s + 8) Thus, Ge (s ) = .K = = = −1.08 . 200(s + 4) ⎛ 39 ⎞ s 2 − 182s − 740 p − 740 1− (s + 5)( s + 8) ⎝ 40 ⎠ 1 estep = 1+K = -12.5, eramp = ∞, eparabola = ∞. p Checking for stability, from first block diagram above, T (s ) = 200( s + 4) Ge (s) . =2 1 + Ge (s ) s + 18s + 60 Therefore, system is stable and steady-state error calculations are valid. 39. Produce a unity-feedback system: + R(s) - - (s+1) s2(s+2) K-1 C(s) 248 Chapter 7: Steady-State Errors (s+1) s2(s+2) s+1 1 . Error = 0.001 = 1+K . Thus, Ge(s) = (s+1)(K-1) = 3 2 p s +2s +(K-1)s+(K-1) 1+ 2 s (s+2) 1 Therefore, Kp = 999 = K-1 . Hence, K = 1.001001. (s+1) s2(s+2) s+1 Check stability: Using original block diagram, T(s) = K(s+1) = 3 2 . s +2s +Ks+K 1+ 2 s (s+2) Making a Routh table: s3 1 K s2 K s1 2 K 2 s0 K 0 0 Therefore, system is stable and steady-state error calculations are valid. 40. a. Produce a unity-feedback system: s+4 1 H1(s) = s+3 - 1 = s+3 + R(s) - - K(s+1) s2(s+2) 1 s+3 K(s+1) s2(s+2) K(s+1)(s+3) Thus, Ge(s) = =4 3 2 . System is Type 0. 1 s +5s +6s +Ks+K K(s+1) s+3 1+ 2 s (s+2) C(s) Solutions to Design Problems 249 b. Since Type 0, appropriate static error constant is Kp. 3K c. K p = K = 3 1 1 d. estep = 1+K = 4 p K(s+1) s2(s+2) K(s+1)(s+3) . Check stability: Using original block diagram, T(s) = (s+4) K(s+1) = 4 s + 5s3+(K+6)s2+5Ks+4K 1+(s+3) 2 s (s+2) Making a Routh table: s4 1 K+6 4K s3 5 5K 0 s2 4K 0 s1 6 5 3K 0 0 s0 4K 0 0 Therefore, system is stable for 0 < K and steady-state error calculations are valid. 41. Program: K=10 numg1=K*poly([-1 -2]);deng1=poly([0 0 -3 -4 -5]); 'G1(s)=' G1=tf(numg1,deng1) numh1=[1 6];denh1=poly([-7 -8]); 'H1(s)=' H1=tf(numh1,denh1) 'H2(s)=H1-1' H2=H1-1 %Form Ge(s)=G1(s)/(1+G1(s)H2(s) 'Ge(s)=G1(s)/(1+G1(s)H2(s))' Ge=feedback(G1,H2) %Test system stability 'T(s)=Ge(s)/(1+Ge(s))' T=feedback(Ge,1) pole(T) Kp=dcgain(Ge) 'sGe(s)' sGe=tf([1 0],1)*Ge; sGe=minreal(sGe) Kv=dcgain(sGe) 's^2Ge(s)' s2Ge=tf([1 0],1)*sGe; s2Ge=minreal(s2Ge) Ka=dcgain(s2Ge) essstep=30/(1+Kp) essramp=30/Kv essparabola=60/Ka K=1E6 numg1=K*poly([-1 -2]);deng1=poly([0 0 -3 -4 -5]); 'G1(s)=' G1=tf(numg1,deng1) numh1=[1 6];denh1=poly([-7 -8]); 250 Chapter 7: Steady-State Errors 'H1(s)=' H1=tf(numh1,denh1) 'H2(s)=H1-1' H2=H1-1 %Form Ge(s)=G1(s)/(1+G1(s)H2(s) 'Ge(s)=G1(s)/(1+G1(s)H2(s))' Ge=feedback(G1,H2) %Test system stability 'T(s)=Ge(s)/(1+Ge(s))' T=feedback(Ge,1) pole(T) Kp=dcgain(Ge) 'sGe(s)' sGe=tf([1 0],1)*Ge; sGe=minreal(sGe) Kv=dcgain(sGe) 's^2Ge(s)' s2Ge=tf([1 0],1)*sGe; s2Ge=minreal(s2Ge) Ka=dcgain(s2Ge) essstep=30/(1+Kp) essramp=30/Kv essparabola=60/Ka Computer response: K= 10 ans = G1(s)= Transfer function: 10 s^2 + 30 s + 20 -----------------------------s^5 + 12 s^4 + 47 s^3 + 60 s^2 ans = H1(s)= Transfer function: s+6 --------------s^2 + 15 s + 56 ans = H2(s)=H1-1 Transfer function: -s^2 - 14 s - 50 ---------------s^2 + 15 s + 56 ans = Ge(s)=G1(s)/(1+G1(s)H2(s)) Solutions to Design Problems 251 Transfer function: 10 s^4 + 180 s^3 + 1030 s^2 + 1980 s + 1120 --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 + 1427 s^4 + 3362 s^3 + 2420 s^2 - 1780 s - 1000 ans = T(s)=Ge(s)/(1+Ge(s)) Transfer function: 10 s^4 + 180 s^3 + 1030 s^2 + 1980 s + 1120 --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 + 1437 s^4 + 3542 s^3 + 3450 s^2 + 200 s + 120 ans = -7.6131 -7.4291 -5.2697 -3.3330 -3.3330 -0.0111 -0.0111 + + - 0.1827i 0.1827i 0.1898i 0.1898i Kp = -1.1200 ans = sGe(s) Transfer function: 10 s^5 + 180 s^4 + 1030 s^3 + 1980 s^2 + 1120 s --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 + 1427 s^4 + 3362 s^3 + 2420 s^2 - 1780 s - 1000 Kv = 0 ans = s^2Ge(s) Transfer function: 10 s^6 + 180 s^5 + 1030 s^4 + 1980 s^3 + 1120 s^2 --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 + 1427 s^4 + 3362 s^3 + 2420 s^2 - 1780 s - 1000 252 Chapter 7: Steady-State Errors Ka = 0 essstep = -250.0000 Warning: Divide by zero. (Type "warning off MATLAB:divideByZero" to suppress this warning.) > In D:\My Documents\Control Systems Engineering Book\CSE 4th ed\Solutions Manual\Chap 7 References\p7_41.m at line 27 essramp = Inf Warning: Divide by zero. (Type "warning off MATLAB:divideByZero" to suppress this warning.) > In D:\My Documents\Control Systems Engineering Book\CSE 4th ed\Solutions Manual\Chap 7 References\p7_41.m at line 28 essparabola = Inf K= 1000000 ans = G1(s)= Transfer function: 1e006 s^2 + 3e006 s + 2e006 -----------------------------s^5 + 12 s^4 + 47 s^3 + 60 s^2 ans = H1(s)= Transfer function: s+6 --------------s^2 + 15 s + 56 ans = H2(s)=H1-1 Transfer function: -s^2 - 14 s - 50 ---------------s^2 + 15 s + 56 Solutions to Design Problems 253 ans = Ge(s)=G1(s)/(1+G1(s)H2(s)) Transfer function: 1e006 s^4 + 1.8e007 s^3 + 1.03e008 s^2 + 1.98e008 s + 1.12e008 --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 - 998563 s^4 - 1.7e007 s^3 - 9.4e007 s^2 - 1.78e008 s - 1e008 ans = T(s)=Ge(s)/(1+Ge(s)) Transfer function: 1e006 s^4 + 1.8e007 s^3 + 1.03e008 s^2 + 1.98e008 s + 1.12e008 --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 + 1437 s^4 + 1.004e006 s^3 + 9.003e006 s^2 + 2e007 s + 1.2e007 ans = -26.9750 -26.9750 17.9750 17.9750 -6.0000 -1.9998 -1.0002 +22.2518i -22.2518i +22.2398i -22.2398i Kp = -1.1200 ans = sGe(s) Transfer function: 1e006 s^5 + 1.8e007 s^4 + 1.03e008 s^3 + 1.98e008 s^2 + 1.12e008 s --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 - 9.986e005 s^4 - 1.7e007 s^3 - 9.4e007 s^2 - 1.78e008 s - 1e008 Kv = 0 ans = s^2Ge(s) 254 Chapter 7: Steady-State Errors Transfer function: 1e006 s^6 + 1.8e007 s^5 + 1.03e008 s^4 + 1.98e008 s^3 + 1.12e008 s^2 --------------------------------------------------------------------s^7 + 27 s^6 + 283 s^5 - 9.986e005 s^4 - 1.7e007 s^3 - 9.4e007 s^2 - 1.78e008 s - 1e008 Ka = 0 essstep = -250.0000 Warning: Divide by zero. (Type "warning off MATLAB:divideByZero" to suppress this warning.) > In D:\My Documents\Control Systems Engineering Book\CSE 4th ed\Solutions Manual\Chap 7 References\p7_41.m at line 56 essramp = Inf Warning: Divide by zero. (Type "warning off MATLAB:divideByZero" to suppress this warning.) > In D:\My Documents\Control Systems Engineering Book\CSE 4th ed\Solutions Manual\Chap 7 References\p7_41.m at line 57 essparabola = Inf 42. D(s)G2(s) G1(s)G2(s) Y(s) = R(s) 1 + G (s)G (s)H(s) + 1 + G (s)G (s)H(s) 1 2 1 2 D(s)G2(s) G1(s)G2(s) E(s) = R(s) - Y(s) = R(s) - 1 + G (s)G (s)H(s) R(s) - 1 + G (s)G (s)H(s) 1 2 1 2 G1(s)G2(s) G2(s) = ⎡1 - 1 + G (s)G (s)H(s)⎤ R(s) - 1 + G (s)G (s)H(s) D(s) 1 2 1 2 ⎣ ⎦ Thus, G1(s)G2(s) G2(s) ⎧ ⎫ e(∞) = lim sE(s) = lim ⎨ ⎡1 - 1 + G (s)G (s)H(s)⎤ R(s) - 1 + G (s)G (s)H(s) D(s)⎬ 1 2 1 2 ⎦ ⎭ s∅ 0 s∅ 0 ⎩ ⎣ 43. a. E(s) = R(s) - C(s). But, C(s) = [R(s) - C(s)H(s)]G1(s)G2(s) + D(s). Solving for C(s), R(s)G1(s)G2(s) D(s) C(s) = 1 + G (s)G (s)H(s) + 1 + G (s)G (s)H(s) 1 2 1 2 Substituting into E(s), G1(s)G2(s) 1 E(s) = ⎡1 - 1 + G (s)G (s)H(s)⎤ R(s) - 1 + G (s)G (s)H(s) D(s) 1 2 1 2 ⎣ ⎦ Solutions to Design Problems 255 1 b. For R(s) = D(s) = s , lim G1(s)G2(s) s∅ 0 1 e(∞) = lim sE(s) = 1 1 + lim G1(s)G2(s)H(s) 1 + lim G1(s)G2(s)H(s) s∅ 0 s∅ 0 s∅ 0 c. Zero error if G1(s) and/or G2(s) is Type 1. Also, H(s) is Type 0 with unity dc gain. 44. First find the forward transfer function of an equivalent unity feedback system. K K s(s + 1)(s + 4) Ge (s) = K (s + a − 1) = s3 + 5s 2 + ( K + 4)s + K( a − 1) 1+ s(s + 1)(s + 4) Thus, e(∞) = e( ∞ ) = 1 = 1 + Kp 1 + Finding the sensitivity of e(∞), 1 = K K( a − 1) Se:a = a δe = e δa a −1 a a ⎛ a − (a − 1) ⎞ a − 1 . = a⎝ ⎠ a2 a2 a −1 45. From Eq. (7.70), KK 1 ⎛ (s+2)2 ⎞ e(∞) = 1 - lim ⎜ K K (s+1)⎟ 2 s∅0 1 + 1(s+2) ⎝ ⎠ Sensitivity to K1: K 2 ⎛ (s+2) ⎞ - lim ⎜ K K2(s+1)⎟ s∅0 1 + 1(s+2) ⎝ ⎠ 2-K2 = 2+K K 12 256 Chapter 7: Steady-State Errors K1 δe K1K2 (100)(0.1) Se:K1 = e δK = - 2+K K = - 2+(100)(0.1) = - 0.833 12 1 Sensitivity to K2: K2 δe 2K2(1+K1) 2(0.1)(1+100) Se:K2 = e δK = (K -2)(2+K K ) = (0.1-2)(2+(100)(0.1)) = - 0.89 2 12 2 46. a. Using Eq. (7.89) with s2 + 15s+ 50 - (4 s + 2 2 ) - (2 s + 2 0 ) - (3 s + 1 5 ) s2 + 10s+ 23 6 - (s + 1 3 ) (s I - A ) -1 s+ 9 s2 + 15s+ 38 1 =3 s + 2 0 s 2 + 1 11 s+ 16 4 yields e(∞) = 1.09756 for a step input and e(∞) = ∞ for a ramp input. The same results are obtained using and Eq. (7.96) for a step input and Eq. (7.103) for a ramp input. b. Using Eq. (7.89) with 2 s + 9s (s I - A ) -1 = 1 s 2 2 - (5 s + 7 ) s - (s + 9 ) 3 s + 9s + 5s+ 7 -1 7 7s s2 + 9s+ 5 5 yields e(∞) = 0 for a step input and e(∞) = 7 for a ramp input. The same results are obtained using and Eq. (12.123) for a step input and Eq. (12.130) for a ramp input. c. Using Eq. (7.89) with yields e(∞) = 6 for a step input and e(∞) = ∞ for a ramp input. The same results are obtained using Solutions to Design Problems 257 and Eq. (7.96) for a step input and Eq. (7.103) for a ramp input. 47. Find G4(s): Since 100 mi/hr = 146.67 ft/sec, the velocity response of G4(s) to a step displacement of the accelerator is v(t) = 146.67(1 - e-αt). Since 60 mi/hr = 88 ft/sec, the velocity equation at 10 seconds K1 becomes 88 = 146.67 (1 - e-α10). Solving for α yields α = 0.092. Thus, G4(s) = s+0.092 . But, from K1 13.49 the velocity equation, the dc value of G4(s) is 0.092 = 146.67. Solving for K1, G4(s) = s+0.092 . Find error: The forward transfer function of the velocity control loop is G3 (s)G 4 (s) = 13.49K 13.49K 1 . Therefore, Kv = 0.092 . e(∞) = K = 6.82 x 10-3K. v s(s + 1)(s + 0.092) 48. First, reduce the system to an equivalent unity feedback system. Push K1 to the right past the summing junction. K1 Convert to a unity feedback system by adding a unity feedback path and subtracting unity from K . 3 The equivalent forward transfer function is, K1K2 Ge(s) = K1K2 Js2+Ds K1K2 ⎛K3 ⎞ = Js2 +Ds+K2(K3-K1) 1+ 2 -1 Js +Ds ⎝K1 ⎠ The system is Type 0 with K p = Ro e(∞) = 1+K p = K1 . Assuming the input concentration is Ro, K3 − K1 R o (K 3 − K 1 ) . The error can be reduced if K3 = K1. K3 258 Chapter 7: Steady-State Errors 49. (s+0.01) Kc s2 K (s+0.01) a. For the inner loop, G1e(s) = (s+0.01) = s2+Ks+0.01K , where K = J . 1+K s2 K (s+0.01) (s+0.01)2 =K 2 2 . s (s +Ks+0.01K) s2 System is Type 2. Therefore, estep = 0, b. eramp = 0, 1 1 c. eparabola = K = 0.01 = 100 a Ge(s) K(s+0.01)2 d. T(s) = 1+G (s) = 4 e s +Ks3+1.01Ks2+0.02Ks+10-4K Form Ge(s) = G1e(s) s4 1 1.01K 10-4K s3 K 0.02K 0 0<K s2 10-4K 0 0.0198 < K s1 1.01K-0.02 0.0201 K2 − 0.0004 K 1.01 K − 0.02 0 0 0.0199 < K s0 10-4K 0<K Kc Thus, for stability K = J > 0.0199 SOLUTIONS TO DESIGN PROBLEMS 50. Pot gains: K1 = = 2. Also, 3π π = 3; Amplifier gain: K2 ; Motor transfer function: Since time constant = 0.5, α 100 C(s) K = 10 = 10. Hence, K = 20. The motor transfer function is now computed as E (s) = α a 20 s(s+2) . The following block diagram results after pushing the potentiometers to the right past the summing junction: Solutions to Design Problems 259 Finally, since Kv = 10 = 51. 60K2 1 2 , from which K2 = 3 . First find Kv: Circumference = 2π nautical miles. Therefore, boat makes 1 revolution 2π in 20 = 0.314 hr. 2π 1 rev rad rad Angular velocity is thus, 0.314 hr = 3600 x 0.314 sec = 5.56 x 10-3 sec . 1/10o 5.56 x 10-3 K . Thus Kv = 3.19 = 4 ; from which, For 0.1o error, e(∞) = o x 2π rad = Kv 360 K = 12.76. 52. a. Performing block diagram reduction: R(s) + K1 - + - − s −13 s + 13 + 100 s + 14 s +100 2 - 3s s+ 0. 2 -s 2 (s + 0. 5)(s2 + 9. 5s + 78) C(s) 260 Chapter 7: Steady-State Errors R(s) + + K1 - - 100 ( s + 0. 2 ) − s − 13 s + 13 3 2 s + 14. 2 s + 402 . 8 s + 20 2 C(s) (s + 0. 5 )(s + 9. 5 s + 78) 2 −s R(s) + K1 - + Ge(s) C(s) - −s Ge(s) = − 200 s 2 − 12.8 s − 2.6 s 7 + 37.2 s 6 + 942.15 s 5 + 13420 s 4 + 1.0249 ×10 5 s 3 + 4.6048 ×10 5 s 2 + 2.2651 ×10 5 s + 10140 System is unity feedback with a forward transfer function, Gt(s), where Gt(s) = − 200 K 1 s 2 − 12.8 s − 2.6 s 7 + 37.2 s 6 + 942.15 s 5 + 13420 s 4 + 1.0269 ×10 5 s 3 + 4.5792 ×10 5 s 2 + 2.2599 ×10 5 s + 10140 Thus, system is Type 0. 520K1 b. From Gt(s), Kp = 10140 = 700. Thus, K1 = 13650. Gt c. T s = 1+ G t For K1 = 13650, Ts = − 2730000 s 2 − 12.8 s − 2.6 s 7 + 37.2 s 6 + 942.15 s 5 + 13420 s 4 + 1.0269 ×10 5 s 3 − 2.2721 ×10 6 s 2 + 3.517 ×10 7 s + 7108140 Because of the negative coefficient in the denominator the system is unstable and the pilot would not be hired. Solutions to Design Problems 261 53. The force error is the actuating signal. The equivalent forward-path transfer function is Ge (s) = K1 . The feedback is H (s ) = De s + Ke . Using Eq. (7.72) s ( s + K1 K2 ) Ea (s) = R(s) . Applying the final value theorem, 1 + Ge ( s )H (s ) ⎛1 s 2⎞ ⎝s ⎠ K ea _ ramp (∞ ) = lim = 2 < 0.1. Thus, K2 < 0.1Ke. Since the closed-loop system K ( D s + Ke ) K s →0 e 1+ 1 e s( s + K1 K2 ) is second-order with positive coefficients, the system is always stable. 54. a. The minimum steady-state error occurs for a maximum setting of gain, K. The maximum K possible is determined by the maximum gain for stability. The block diagram for the system is shown below. ωi _ desired ( s) V i ( s) + 3 - K 2 (s + 10)(s + 4 s + 10) ωo ( s) 3 Pushing the input transducer to the right past the summing junction and finding the closed-loop transfer function, we get 3K 3K (s + 10)( s2 + 4 s + 10) T (s ) = =3 2 3K s + 14s + 50s + (3 K + 100) 1+ (s + 10)(s 2 + 4 s + 10) Forming a Routh table, s3 1 50 s2 14 3K+100 − 3K + 600 14 0 3K+100 0 s 1 s0 262 Chapter 7: Steady-State Errors The s1 row says -∞ < K < 200. The s0 row says − 100 < K. Thus for stability, 3 100 < K < 200. Hence, the maximum value of K is 200. 3 1 1 3K =. b. K p = = 6 . Hence, estep (∞ ) = 1 + Kp 7 100 − c. Step input 55. a. Yh-Ycat Spring displacement Desired force Input voltage+ 1 100 Input transducer K Controller - 1 1000 F up 0 .7883 ( s + 53 .85 ) ( s2 + 15 .47 s + 9283 )( s 2 + 8 .119 s + 376 .3 ) Actuator F out 82300 Pantograph dynamics Spring 1 100 Sensor Y h-Ycat Spring displacement Desired force + 100 1 1000 Controller Actuator K - b. G(s) = Fup 0 .7883 ( s + 53 .85 ) (s 2 + 15 .47 s + 9283 )( s 2 + 8 .119 s + 376 .3 ) Pantograph dynamics 82300 Fout Spring Yh ( s) − Ycat (s) 0.7883( s + 53.85) =2 2 Fup ( s) (s + 15.47s + 9283)(s + 8.119 s + 376.3) Ge(s) = (K/100)*(1/1000)*G(s)*82.3e3 0.6488K (s+53.85) Ge(s) = (s2 + 8.119s + 376.3) (s2 + 15.47s + 9283) Kp = 0.6488K*53.85/[(376.3)(9283)] = K*1.0002E-5 Maximum K minimizes the steady-state error. Maximum K possible is that which yields stability. From Chapter 6 maximum K for stability is K = 1.88444 x 105. Therefore, Kp = 1.8848. c. ess = 1/(1+Kp) = 0.348. EIGHT Root Locus Techniques SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Transient Design via Gain a. From the Chapter 5 Case Study Challenge: 76.39K G(s) = s(s+150)(s+1.32) 1 Since Ts = 8 seconds, we search along - 2 , the real part of poles with this settling time, for 180o. We find the point to be - 0.5+j6.9 with 76.39K = 7194.23, or K = 94.18. Second-order approximation is OK since third pole is much more than 5 times further from the imaginary axis than the dominant second-order pair. b. Program: numg= 1; deng=poly([0 -150 -1.32]); 'G(s)' G=tf(numg,deng) rlocus(G) axis([-2,0,-10,10]); title(['Root Locus']) grid on [K1,p]=rlocfind(G) K=K1/76.39 Computer response: ans = G(s) Transfer function: 1 ----------------------s^3 + 151.3 s^2 + 198 s Select a point in the graphics window selected_point = -0.5034 + 6.3325i 264 Chapter 8: Root Locus Techniques K1 = 6.0690e+003 p= 1.0e+002 * -1.5027 -0.0052 + 0.0633i -0.0052 - 0.0633i K= 79.4469 >> ans = G(s) Transfer function: 1 ----------------------s^3 + 151.3 s^2 + 198 s Select a point in the graphics window selected_point = -0.5000 + 6.2269i K1 = 5.8707e+003 p= 1.0e+002 * -1.5026 -0.0053 + 0.0623i -0.0053 - 0.0623i K= 76.8521 Solutions to Case Studies Challenges 265 UFSS Vehicle: Transient Design via Gain K(s+0.437) a. Push -K1 to the right past the summing junction yielding G(s) = s(s+2)(s+1.29)(s+0.193) , where K = 0.25K1. Combine the parallel feedback paths and obtain H(s) = (s+1). Hence, G(s)H(s) = K(s+0.437)(s+1) s(s+2)(s+1.29)(s+0.193) . The root locus is shown below in (b). Searching the 10% overshoot line (ζ = 0.591; θ = 126.24ο), we find the operating point to be -1.07 ± j1.46 where K = 3.389, or K1 = 13.556. b. Program: numg= [1 0.437]; deng=poly([0 -2 -1.29 -0.193]); G=tf(numg,deng); numh=[1 1]; denh=1; H=tf(numh,denh); GH=G*H; rlocus(GH) pos=(10); z=-log(pos/100)/sqrt(pi^2+[log(pos/100)]^2); sgrid(z,0) title(['Root Locus with ' , num2str(pos), ' Percent Overshoot Line']) [K,p]=rlocfind(GH); pause K1=K/0.25 T=feedback(K*G,H) T=minreal(T) step(T) 266 Chapter 8: Root Locus Techniques title(['Step Response for Design of ' , num2str(pos), ' Percent']) Computer response: Select a point in the graphics window selected_point = -1.0704 + 1.4565i K1 = 13.5093 Transfer function: 3.377 s + 1.476 --------------------------------------------s^4 + 3.483 s^3 + 6.592 s^2 + 5.351 s + 1.476 Transfer function: 3.377 s + 1.476 --------------------------------------------s^4 + 3.483 s^3 + 6.592 s^2 + 5.351 s + 1.476 Solutions to Problems 267 ANSWERS TO REVIEW QUESTIONS 1. The plot of a system's closed-loop poles as a function of gain 2. (1) Finding the closed-loop transfer function, substituting a range of gains into the denominator, and factoring the denominator for each value of gain. (2) Search on the s-plane for points that yield 180 degrees when using the open-loop poles and zeros. 3. K = 1/5 4 . No 5. At the zeros of G(s) and the poles of H(s) 6. (1) Apply Routh-Hurwitz to the closed-loop transfer function's denominator. (2) Search along the imaginary axis for 180 degrees. 7. If any branch of the root locus is in the rhp, the system is unstable. 8.If the branch of the root locus is vertical, the settling time remains constant for that range of gain on the vertical section. 9. If the root locus is circular with origin at the center 10. Determine if there are any break-in or breakaway points 11. (1) Poles must be at least five times further from the imaginary axis than the dominant second order pair, (2) Zeros must be nearly canceled by higher order poles. 12. Number of branches, symmetry, starting and ending points 13. The zeros of the open loop system help determine the root locus. The root locus ends at the zeros. Thus, the zeros are the closed-loop poles for high gain. SOLUTIONS TO PROBLEMS 1. a. No: Not symmetric; On real axis to left of an even number of poles and zeros 268 Chapter 8: Root Locus Techniques b. No: On real axis to left of an even number of poles and zeros c. No: On real axis to left of an even number of poles and zeros d. Yes e. No: Not symmetric; Not on real axis to left of odd number of poles and/or zeros f. Yes g. No: Not symmetric; real axis segment is not to the left of an odd number of poles h. Yes 2. jω jω s-plane X s-plane X O X σ σ X X (a) (b) jω X s-plane O jω X s-plane σ σ O X X (c) (d) Solutions to Problems 269 jω X s-plane O jω X s-plane σ σ O X X (c) (d) jω jω s-plane s-plane X X σ XX O (e) O X (f) 3. a. b. c. X σ 270 Chapter 8: Root Locus Techniques -4 -1 d. 4. Breakaway: σ = -2.43 for K = 52.1 Solutions to Problems 271 5. 2 1.5 1 Imag Axis 0.5 0 -0.5 -1 -1.5 -2 -7 -6 -5 -4 -3 Real Axis -2 -1 0 1 Break-in: σ = -1.5608 for K = 61.986; Breakaway: σ = -5.437 for K = 1.613. 6. Convert the denominator to the following form: D ( s ) = 1 + G ( s) = 20 K ( s + 5) 20 K ( s + 5) . = 3 2 s + 2 s + 7 s s ( s 2 + 2 s + 7) Plotting the root locus yields 20 K ( s + 5) and thus identify s3 + 2s 2 + 7 s 272 Chapter 8: Root Locus Techniques 7. Im Im s-plane Re Re 8. 2.5 2 1.5 1 Imag Axis 0.5 0 -0.5 -1 -1.5 -2 -2.5 -8 -6 -4 -2 Real Axis 0 2 4 Closed-loop poles will be in the left-half-plane when rhp pole reaches the origin, or K > ( (3)(3) 9 =. 2 )( 2 )(1) 2 s-plane Solutions to Problems 273 9. Closed-loop poles will be in the right-half-plane for K > (2)(2) 4 = (gain at the origin). (3)(3) 9 Therefore, stable for K < 4/9; unstable for K > 4/9 . 10. 2.5 2 1.5 1 Imag Axis 0.5 0 -0.5 -1 -1.5 -2 -2.5 -5 -4 -3 -2 Real Axis -1 0 1 Breakaway: σ = -3.436 for K = 1.781. System is never unstable. System is marginally stable for K = ∞. 274 Chapter 8: Root Locus Techniques 11. System 1: (a) a. Breakaway: σ = 1.41 for K = 0.03; Break-in: σ = -1.41 for K = 33.97. b. Imaginary axis crossing at j1.41 for K = 1. Thus stable for K > 1. c. At break-in point, poles are multiple. Thus, K = 33.97. d. Searching along 1350 line for 1800, K = 5 at 1.414 ∠ 1350. System 2: (b) a. Break-in: σ = -1.41 for K = 28.14. Solutions to Problems 275 b. Imaginary axis crossing at j1.41 for K = 0.67. Thus stable for K > 0.67. c. At break-in point, poles are multiple. Thus, K = 28.14. d. Searching along 1350 line for 1800, K = 4 at 1.414 ∠ 1350. 12. a. Root locus crosses the imaginary axis at the origin for K = 6. Thus the system is stable for K > 6. b. Root locus crosses the imaginary axis at j0.65 for K = 0.79. Thus, the system is stable for K < 0.79. 13. 6 4 Imag Axis 2 0 -2 -4 -6 -7 -6 -5 -4 -3 -2 -1 Real Axis 0 1 2 3 4 276 Chapter 8: Root Locus Techniques There will be only two right-half-plane poles when pole at +2 moves into the left-half-plane at the origin. Thus K = (5)( 2 )( 2 )(2 ) = 6.67 . 3 14. 15 10 Imag Axis 5 0 -5 -10 -15 -12 -10 -8 -6 -4 Real Axis -2 0 2 Root locus crosses the imaginary axis at j7.348 with a gain of 810. Real axis breakaway is at –2.333 at a gain of 57.04. Real axis intercept for the asymptotes is − 15 = − 5 . The angle of the asymptotes 3 5π π is = 3 , π, 3 . Some other points on the root locus are: ζ = 0.4: -1.606 + j3.68, K = 190.1 ζ = 0.6: -1.956 + j2.6075, K = 117.8 ζ = 0.8: -2.189 + j1.642, K = 79.55 Solutions to Problems 277 15. a. θ θ1 3 θ2 90 Imaginary axis crossing: j1.41 at K = 1.5. Stability: K < 1.5. Breakaway: -1.41 at K = 0.04. Points on root locus: -1.5 ± j0, K = 0.0345; -0.75 ± j1.199, K = 0.429; 0 ± j1.4142, K = 1.5; 0.75 ± j1.1989, K = 9. Finding angle of arrival: 90 - θ1 - θ2 + θ3 = 90o - tan-1(1/3) - tan-1(1/2) + θ3 = 180o. Thus, θ3 = 135o. b. Imaginary axis crossing: j1.41 at K = 1. Stability: K < 1. Breakaway: -1.41 at K = 0.03. Break-in: 1.41 at K = 33.97. Points on root locus: -1.5 ± j0, K = 0.02857; -0.75 ± j1.199, K = 0.33; 0 ± j1.4142, K = 1; 0.75 ± j1.1989, K = 3. 16. a. Root locus crosses the imaginary axis at ± j 3.162 at K = 52. b. Since the gain is the product of pole lengths to -5, K = (1) ( 42 + 12 )( ) 42 + 12 = 17 . 278 Chapter 8: Root Locus Techniques 17. a. b. σ a = 8 (2k+1)π 5π (0 - 2 - 3 - 4) - (-1) π = - 3 ; Angle = = 3 , π, 3 3 3 c. Root locus crosses imaginary axis at j4.28 with K = 140.8. d. K = 13.125 18. Assume that root locus is epsilon away from the asymptotes. Thus, σa = Angle = (0 - 3 - 6) - (-α) ≈ -1 ; 2 (2k+1)π π 3π = 2 , 2 . Hence α = 7. Checking assumption at –1 ± j100 yields -180o with K = 2 9997.02. 19. a. Breakaway: -0.37 for K = 0.07. Break-in: 1.37 for K = 13.93 b. Imaginary axis crossing: ±j0.71 for K = 0.33 c. System stable for K < 0.33 Solutions to Problems 279 d. Searching 120o find point on root locus at 0.5∠120ο = - 0.25 ± j0.433 for K = 0.1429 20. a. 20 15 10 Imag Axis 5 0 -5 -10 -15 -20 -40 -30 -20 -10 Real Axis 0 10 20 b. 0 < K < 23.93 c. K = 81.83 @ -13.04 ± j13.04 d. At the break-in point, s = -14.965, K = 434.98. 21. a. Asymptotes: σint = (2k+1)π π 3π 5π 7π (-1 -2 -3 -4) - (0) 5 = - 2 ; Angle = =4 , 4 , 4 , 4 4 4 b. Breakaway: -1.38 for K = 1 and -3.62 for K = 1 280 Chapter 8: Root Locus Techniques c. Root locus crosses the imaginary axis at ±j2.24 for K = 126. Thus, stability for K < 126. d. Search 0.7 damping ratio line (134.427 degrees) for 1800. Point is 1.4171∠134.427ο = - 0.992 ± j1.012 for K = 10.32. e. Without the zero, the angles to the point ±j5.5 add up to -265.074o. Therefore the contribution of 5.5 the zero must be 265.074 - 180 = 85.074o. Hence, tan 85.074o = z , where - zc is the location of the c zero. Thus, zc = 0.474. f. After adding the zero, the root locus crosses the imaginary axis at ±j5.5 for K = 252.5. Thus, the system is stable for K < 252.5. g. The new root locus crosses the 0.7 damping ratio line at 2.7318∠134.427o for K = 11.075 compared to 1.4171∠134.427o for K = 10.32 for the old root locus. Thus, the new system's settling time is shorter, but with the same percent overshoot. Solutions to Problems 281 22. 23. 1 1 T(s) = 2 = s + αs + 1 s2 + 1 αs 1+ 2 s +1 1 . Thus an equivalent system has G(s) = 2 s +1 αs Plotting a root locus for G(s)H(s) = 2 , we obtain, s +1 and H(s) = αs. 282 Chapter 8: Root Locus Techniques 24. a. b. Root locus crosses 20% overshoot line at 1.8994 ∠ 117.126o = - 0.866 ± j1.69 for K = 9.398. π 4 c. Ts = 0.866 = 4.62 seconds; Tp = 1.69 = 1.859 seconds d. Root locus crosses imaginary axis at ±j3.32 for K = 60. Therefore stability for K < 60. e. Other poles with same gain as dominant poles: σ = -4.27 25. a. 20 15 10 Imag Axis 5 0 -5 -10 -15 -20 -8 b. -7 -6 -5 -4 -3 -2 Real Axis -1 0 1 2 3 Solutions to Problems 283 ( −6 − 5 − 4 − 2) − (2 ) = − 9.5 4−2 (2k + 1)π π 3π θa = =, 4−2 22 σa = c. At the jω axis crossing, K = 115.6. Thus for stability, 0 < K < 115.6. d. Breakaway points at σ = -2.524 @ K = 0.496 and σ = -5.576 @ K = 0.031. e. For 25% overshoot, Eq. (4.39) yields ζ = 0.404. Searching along this damping ratio line, we find the 1800 point at –0.6608 + j1.496 where K = 35.98. f. –7.839 ± j7.425 g. Second-order approximation not valid because of the existence of closed-loop zeros in the rhp. h. Program: numg=35.98*[1 -2 2]; deng=poly([-2 -4 -5 -6]); G=tf(numg,deng); T=feedback(G,1) step(T) Computer response: Transfer function: 35.98 s^2 - 71.96 s + 71.96 -----------------------------------s^4 + 17 s^3 + 140 s^2 + 196 s + 312 Simulation shows over 30% overshoot and nonminimum-phase behavior. Second-order approximation not valid. 26. a. Draw root locus and minimum damping ratio line. 284 Chapter 8: Root Locus Techniques Minimum damping ratio 145.55 o Minimum damping ratio is ζ = cos (180 - 145.55) = cos 34.45o = 0.825. Coordinates at tangent point of ζ = 0.825 line with the root locus is approximately –1 + j0.686. The gain at this point is 0.32. b. Percent overshoot for ζ = 0.825 is 1.019%. π 4 c. Ts = 1 = 4 seconds; Tp = 0.6875 = 4.57 seconds d. Second-order approximation is not valid because of the two zeros and no pole-zero cancellation. 27. The root locus intersects the 0.55 damping ratio line at –7.217 + j10.959 with K = 134.8. A justification of a second-order approximation is not required. The problem stated the requirements in terms of damping ratio and not percent overshoot, settling time, or peak time. A second-order approximation is required to draw the equivalency between percent overshoot, settling time, and peak time and damping ratio and natural frequency. 28. Since the problem stated the settling time at large values of K, assume that the root locus is 4 -11 + α = - T . Since Ts is given approximately close to the vertical asymptotes. Hence, σint = 2 s as 4 seconds, σint = -1 and α = 9. The root locus is shown below. Solutions to Problems 285 29. The design point is - 0.506 ± j1.0047. Excluding the pole at -α , the sum of angles to the design point is -141.37o. Thus, the contribution of the pole at -α is 141.37 - 180 = - 38.63o. The following geometry applies: Hence, tan θ = 1.0047 = tan 38.63 = 0.799. Thus α = 1.763. Adding this pole at -1.763 yields α - 0.506 180o at - 0.506 ± j1.0047 with K = 7.987. 30. a. 286 Chapter 8: Root Locus Techniques 15 10 Imag Axis 5 0 -5 -10 -15 -10 -8 -6 -4 Real Axis -2 0 2 b. Searching along the 10% overshoot line (angle = 126.239o), the point - 0.7989 + j1.0898 yields 180o for K = 81.74. c. Higher-order poles are located at approximately –6.318 and –7.084. Since these poles are more than 5 times further from the imaginary axis than the dominant pole found in (b), the second-order approximation is valid. d. Searching along the imaginary axis yields 180o at j2.53, with K = 394.2. Hence, for stability, 0 < K < 394.2. 31. Program: pos=10; z=-log(pos/100)/sqrt(pi^2+[log(pos/100)]^2) numg=1; deng=poly([0 -3 -4 -8]); G=tf(numg,deng) Gzpk=zpk(G) rlocus(G,0:1:100) pause axis([-2,0,-2,2]) sgrid(z,0) pause [K,P]=rlocfind(G) T=feedback(K*G,1) pause step(T) Computer response: z= 0.5912 Transfer function: 1 Solutions to Problems 287 ---------------------------s^4 + 15 s^3 + 68 s^2 + 96 s Zero/pole/gain: 1 ------------------s (s+8) (s+4) (s+3) Select a point in the graphics window selected_point = -0.7994 + 1.0802i K= 81.0240 P= -7.1058 -6.2895 -0.8023 + 1.0813i -0.8023 - 1.0813i Transfer function: 81.02 -----------------------------------s^4 + 15 s^3 + 68 s^2 + 96 s + 81.02 288 Chapter 8: Root Locus Techniques 32. a. For a peak time of 1s, search along the horizontal line, Im = π/ Tp= π, to find the point of intersection with the root locus. The intersection occurs at –2 ± jπ at a gain of 11. Solutions to Problems 289 10 8 6 4 Imag Axis 2 0 -2 -4 -6 -8 -10 -5 -4 -3 -2 -1 Real Axis b. Program: numg=11*[1 4 5]; deng=conv([1 2 5],poly([-3 -4])); G=tf(numg,deng); T=feedback(G,1); step(T) 0 1 2 290 Chapter 8: Root Locus Techniques 33. a. b. Searching the jω axis for 180o, we locate the point j6.29 at a gain of 447.83. c. Searching for maximum gain between -4 and -5 yields the breakaway point, -4.36. Searching for minimum gain between -2 and -3 yields the break-in point, -2.56. d. jω j2 θ6 x x -6 θ1 x -5 θ2 x -4 θ3 O -3 θ4 O -2 j1 θ5 σ -1 x 90o -j1 Solutions to Problems 291 To find the angle of departure from the poles at -1±j1: -θ1 - θ2 - θ3 + θ4 + θ5 - θ6 - 900 = - tan-1(1/5) - tan-1(1/4) - tan-1(1/3) + tan-1(1/2) + tan-1(1/1) - θ6 - 90o = 1800 . Thus, θ6 = 242.22o e. Searching along the ζ = 0.3 line (θ = 180 - cos-1(ζ) = 107.458o) for 180o we locate the point 3.96 ∠ 107.458o = -1.188±j3.777. The gain is 127.133. 34. a. b. Searching the jω axis for 180o, we locate the point j2.56 at a gain of 30.686. c. Searching for maximum gain between 0 and -2 yields the breakaway point, -0.823. Searching for maximum gain between -4 and -6 yields the breakaway point, -5.37. Searching for minimum gain beyond -8 yields the break-in point, -9.39. e. Searching along the ζ = 0.3 line (θ = 180 - cos-1(ζ) = 107.458o) for 180o we locate the point 1.6 ∠ 107.458o = -0.48 ± j1.53. The gain is 9.866. 292 Chapter 8: Root Locus Techniques 35. a. Searching the 15% overshoot line (ζ = 0.517; θ = 121.131ο) for 180o, we find the point 2.404 ∠ 121.131ο = -1.243 + j2.058. b. K = 11.09. c. Another pole is located left of -3. Searching for a gain of 11.09 in that region, we find the third pole at -4.514. d. The third pole is not 5 times farther than the dominant pair from the jω axis. the second-order approximation is estimated to be invalid. Solutions to Problems 293 36. a. 15 10 Imag Axis 5 0 -5 -10 -15 -5 -4 -3 -2 -1 0 Real Axis 1 2 3 4 b. Searching the jω axis for 180o, we locate the point j1.69 at a gain of 4.249. c. Searching between -2 and -3 for maximum gain, the breakaway is found at -2.512. d. jω s-plane X X -3 θ6 X -2 θ3 θ1 j2 θ5 X σ 2 -1 θ4 θ2 -j2 To find the angle of arrival to the zero at 2 + j2: ⎛4 ⎛2 ⎛2 θ 1 + θ 2 − θ3 − θ4 − θ 5 − θ 6 = θ1 + 90 − 0 − tan−1 ⎝ ⎞ − tan−1 ⎝ ⎞ − tan −1 ⎝ ⎞ = 180 ⎠ ⎠ ⎠ Solving for θ1, the angle of arrival is θ1 = –191.5 . 0 3 4 5 294 Chapter 8: Root Locus Techniques e. The closed-loop zeros are the poles of H(s), or –1 ± j2. f. Searching the ζ = 0.358; (θ = 110.97ο) for 180o, we find the point = -0.6537+j1.705. The gain, K = 0.8764. g. Higher-order poles are at –2.846 ± j1.731. These are not 5 times further than the dominant poles. Further, there are closed-loop zeros at –1 ± j2 that are not cancelled any higher-order poles. Thus, the second-order approximation is not valid. 37. 50 40 30 20 Imag Axis 10 0 -10 -20 -30 -40 -50 -40 -35 -30 -25 -20 -15 Real Axis -10 -5 0 5 a. The root locus crosses the imaginary axis at j2.621 with K = 4365. Therefore, the system is stable for 0 < K < 4365. b. Search the 0.707 damping ratio line for 180o and find –0.949 + j0.949 with K = 827.2. c. Assume critical damping where root locus breaks away from the real axis. Locus breaks away at – 1.104 with K = 527.6. 38. Program: numg=1; deng=poly([0 -3 -7 -8]); numh=[1 30]; denh=[1 20 200]; G=tf(numg,deng) Gzpk=zpk(G) H=tf(numh,denh) Solutions to Problems 295 rlocus(G*H) pause K=0:10:1e4; rlocus(G*H,K) sgrid(0.707,0) axis([-2,2,-5,5]); pause for i=1:1:3; [K,P]=rlocfind(G*H) end T=feedback(K*G,H) step(T) Computer response: Transfer function: 1 -----------------------------s^4 + 18 s^3 + 101 s^2 + 168 s Zero/pole/gain: 1 ------------------s (s+8) (s+7) (s+3) Transfer function: s + 30 ---------------s^2 + 20 s + 200 Select a point in the graphics window selected_point = -0.9450 + 0.9499i K= 828.1474 P= -9.9500 -9.9500 -8.1007 -8.1007 -0.9492 -0.9492 +10.0085i -10.0085i + 1.8579i - 1.8579i + 0.9512i - 0.9512i Select a point in the graphics window selected_point = 0.0103 + 2.6385i K= 4.4369e+003 P= -9.7320 +10.0691i 296 Chapter 8: Root Locus Techniques -9.7320 -9.2805 -9.2805 0.0126 0.0126 -10.0691i + 3.3915i - 3.3915i + 2.6367i - 2.6367i Select a point in the graphics window selected_point = -1.0962 - 0.0000i K= 527.5969 P= -9.9682 -9.9682 -7.9286 -7.9286 -1.1101 -1.0962 +10.0052i -10.0052i + 1.5303i - 1.5303i Transfer function: 527.6 s^2 + 1.055e004 s + 1.055e005 ------------------------------------------------s^6 + 38 s^5 + 661 s^4 + 5788 s^3 + 23560 s^2 + 3.413e004 s + 1.583e004 Solutions to Problems 297 39. a. Search jω = j10 line for 180o and find -4.533 + j10 with K = 219.676. b. K a = 219.676 x 6 20 c. A settling time of 0.4 seconds yields a real part of -10. Thus if the zero is at the origin, G(s) K s(s+20) , which yields complex poles with -10 as the real part. At the design point, -10 + j10, K = 200. 298 Chapter 8: Root Locus Techniques 40. a. Searching along ζωn = -1 for 180o, find –1 + j2.04 with K = 170.13. b. Assume critical damping when root locus breaks away form the real axis. Searching for maximum gain, the breakaway point is at -1.78 with K = 16.946. 41. K 3 2 T(s) = 3 2 + 5s + K . Differentiating the characteristic equation, s + 6s + 5s + K = 0, yields, s + 6s δs δs δs + 12s +5 + 1 = 0. 3 s2 δK δK δK δs , Solving for δK -1 δs = δK 3s2 + 12s + 5 The sensitivity of s to K is K δs K -1 =s Ss:K = s δK 3s2 + 12s + 5 a. Search along the ζ = 0.591 line and find the root locus intersects at s = 0.7353∠126.228ο = - 0.435 + j0.593 with K = 2.7741. Substituting s and K into Ss:K yields Ss:K = 0.487 - j0.463 = 0.672∠-43.553o b. Search along the ζ = 0.456 line and find the root locus intersects at s = 0.8894∠117.129ο = - 0.406 + j0.792 with K = 4.105. Substituting s and K into Ss:K yields Ss:K = 0.482-j0.358 = 0.6∠-36.603o c. Least sensitive: ζ = 0.456. 42. 0.00076s3 The sum of the feedback paths is He(s) = 1 + 0.02s + s+0.06 . Thus, and Solutions to Problems 299 Plotting the root locus, Searching vertical lines to calibrate the root locus, we find that 0.00076K is approximately 49.03 at -10 ± j41.085. Searching the real axis for 0.00076K = 49.03, we find the third pole at -36.09. 41.085 a. ζ = cos (tan-1 ( 10 )) = 0.236 b. %OS = e −ζπ / 1−ζ 2 x100 = 46.63% 102 + 41.0852 = 42.28 rad/s 4 4 d. Ts = = 10 = 0.4 seconds ζωn c. ωn = e. Tp = π ωn 1-ζ2 π = 41.085 = 0.076 seconds 43. K 1s K2 Push K2 to the right past the summing junction and find, T(s) = (1 + K ) ( 2 ) 2 s + K3s + K2 K2 K1(s+K ) 1 K2 K 1 (s + K ) s2 + K 2 K 3s 1 . Changing form, T(s) = . Thus, G(s)H(s) = 2 . Sketching the =2 K3s s + K2 s + K 3s + K 2 1+ 2 s + K2 root locus, 300 Chapter 8: Root Locus Techniques a. - K2 K1(s+K ) 1 s2 + K b. T(s) = 2 K3s 1+ 2 s + K2 K2 K1(s+K ) K2 1 =2 . Therefore closed-loop zero at - K . Notice that the zero 1 s + K3s + K2 at the origin of the root locus is not a closed-loop zero. K1s K2 c. Push K2 to the right past the summing junction and find, T(s) = (1 + K ) ( 2 ) 2 s + K3s + K2 K2 K1(s+K ) 1 K2 K 1 (s + K ) 2+K s s K2 1 3 . Changing form, T(s) = . Thus, G(s)H(s) = 2 . Sketching the =2 K2 s + K3s s + K3s + K2 1+ 2 s + K3s root locus, K2 The closed-loop zero is at - K . 1 Solutions to Problems 301 44. a b c d 45. a. Using Figure P8.15(a), [Ms2+(D+Dc)s+(K+Kc)]X(s) - [Dcs+Kc]Xa(s) = 0 Rearranging, [Ms2+Ds+K]X(s) = -[Dcs+Kc](X(s)-Xa(s)) where [Dcs+Kc](X(s)-Xa(s)) can be thought of as the input to the plant. For the active absorber, (Mcs2+Dcs+Kc)Xa(s) - (Dcs+Kc)X(s) = 0 or Mcs2Xa(s)+Dcs(Xa(s)-X(s))+Kc(Xa(s)-X(s)) = 0 Adding -Mcs2X(s) to both sides, Mcs2(Xa(s)-X(s))+Dcs(Xa(s)-X(s))+Kc(Xa(s)-X(s)) = -Mcs2X(s) Let Xa(s)-X(s) = Xc(s) and s2X(s) = C(s) = plant output acceleration. Therefore, (1) 302 Chapter 8: Root Locus Techniques Mcs2Xc(s)+DcsXc(s)+KcXc(s) = -McC(s) or (Mcs2+Dcs+Kc)Xc(s) = -McC(s) (2) Using Eqs. (1) and (2), and Xa(s)-X(s) = Xc(s), -Mc Dcs+Kc Xc(s) X(s) C(s) = Mcs2+Dcs+Kc ; Xc(s) = Ms2+Ds+K which suggests the following block diagram: Structure Input force F(s) + 1 Ms + Ds + K X(s) s2 2 Output structure acceleration C(s) force feedback from absorber -Fc(s) Dc s+ Kc -Xc(s) Mc M cs + Dcs + K c 2 Active vibration absorber b. Substituting M = D = K = Dc = Kc =1 and redrawing the block diagram above to show X(s) as the output yields a block diagram with G s = Mc s 2 s + 1 1 . To study the steadyand H s = s2 + s + 1 Mc s2 + s + 1 state error, we create a unity-feedback system by subtracting unity from H(s). Thus He(s) = H(s)-1 = G 1 . The equivalent G(s) for this unity-feedback system is Ge s = Mc s3 − s − 1 2 +s+1 1 + G He Mc s = Mc s 2 + s + 1 Mc s4 + 2 Mc s3 + s 3 + Mc s 2 + 2 s 2 + s . Hence the equivalent unity-feedback system is Type 1 and will respond with zero steady-state error for a step force input. Solutions to Design Problems 303 c. Using Ge(s) in part b, we find T s = Ge 1 + Ge = Mc s 2 + s + 1 s 2 + 2 s + 2 Mc s2 + s 3 + 2 s 2 + 2 s + 1 M c s2 + s + 1 numerator and denominator by s3 + 2 s2 + 2 s + 1, T s = s3 + 2 s2 + 2 s + 1 s2 + 2 s + 2 M c s 2 . Dividing . Thus, the system +1 s3 + 2 s2 + 2 s + 1 s2 + 2 s + 2 M c s 2 s2 + 2 s + 2 M c s 2 has the same root locus as a system with G(s)H(s) = = . s3 + 2 s2 + 2 s + 1 s + 1 s2 + s + 1 Sketching the root locus, jω o x x oo σ x o SOLUTIONS TO DESIGN PROBLEMS 46. 4 a. For a settling time of 0.1 seconds, the real part of the dominant pole is - 0.1 = - 40. Searching along the σ = - 40 line for 180o, we find the point –40 + j57.25 with 20,000K = 2.046 x 109, or K = 102,300. 57.25 b. Since, for the dominant pole, tan-1 ( 40 ) = 55.058o, ζ = cos (55.058o) = 0.573. Thus, %OS = e −ζπ / 1−ζ 2 x100 = 11.14%. c. Searching the imaginary axis for 180o, we find ω = 169.03 rad/s for 20,000K = 1.43 x 1010. Hence, K = 715,000. Therefore, for stability, K < 715,000. 47. G(s) = 61.73K (s+10)3 (s2 + 11.11s + 61.73) 304 Chapter 8: Root Locus Techniques a. Root locus crosses the imaginary axis at ±j6.755 with 61.73K equal to 134892.8. Thus for oscillations, K = 2185.21. b. From (a) the frequency of the oscillations is 6.755 rad/s. c. The root locus crosses the 20% overshoot line at 6∠117.126o = - 2.736 + j5.34 with 61.73K = 23323.61. Thus, K = 377.83 and Ts = 4 4 = 2.736 = 1.462 seconds. ζωn 48. a. Finding the transfer function with Ca as a parameter, 2 .. Y m (s ) Y (s ) G Plotting the root locus, 2 s (s + 1 ) 2 = s (2 s + 2 ) 2 (C a + 1 ) s + 4 s + 2 = s2 + 4s+ 2 1+ C as 2 2 s + 4s+ 2 Solutions to Design Problems 305 4 2 2 b. Since 2ζωn = C +1 ; ωn2 = C +1 , ζ2 = C +1 = 0.692. Hence, Ca = 3.2. a a a 49. a. b. The pole at 1.8 moves left and crosses the origin at a gain of 77.18. Hence, the system is stable for K > 77.18, where K = -508K2. Hence, K2 < -0.152. c. Search the ζ = 0.5 (θ = 120o ) damping ratio line for 180o and find the point, -8.044 + j13.932 = 16.087 ∠ 120ο with a K = -508K2 = 240.497. Thus, K2 = -0.473. d. Search the real axis between 1.8 and -1.6 for K = 240.497 and find the point -1.01. 240.497K1(s+1.6) Thus Ge(s) = s(s+1.01)(s+8.044+j13.932)(s+8.044-j13.932) = 240.497K1(s+1.6) s(s+1.01)(s2+16.088s+258.8066) . Plotting the root locus and searching the jω axis for 180o we find j15.792 with 240.497K1 = 4002.6, or K1 = 16.643. e. Root Locus with 0.45 Damping Ratio Line 30 20 Imag Axis 10 0 -10 -20 -30 -30 -20 -10 0 Real Axis 10 20 30 306 Chapter 8: Root Locus Techniques Search the ζ = 0.45 (θ = 116.744o ) damping ratio line for 180o and find the point, -6.685 + j13.267 = 14.856 ∠ 116.744o with a K = 240.497K1 = 621.546. Thus, K1 = 2.584. 50. a. Update the block diagram to show the signals that form Hsys(s). Perform block diagram reduction of the parallel paths from TW(s). Reduce the momentum wheel assembly to a single block. Solutions to Design Problems 307 Substitute values and find Ge(s) = 4.514x10 −6 K(s + 0.01) . Plotting the root locus yields 2 s (s + 0.043) Root Locus with 0.404 Damping ratio Line 0.06 0.04 Imag Axis 0.02 0 -0.02 -0.04 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 Real Axis 0.01 0.02 0.03 0.04 0.05 b. Searching the 25% overshoot line (ζ = 0.404; θ = 113.8o) for 180o yields -0.0153 + j0.0355 with a gain = 4.514E-6K = 0.0019. Thus, K = 420.9. c. Searching the real axis between –0.025 and –0.043 for a gain of 0.0019. we find the third pole at - 0.0125. Simulate the system. There is no pole-zero cancellation. A simulation shows approximately 95% overshoot. Thus, even though the compensator yields zero steady-state error, a system redesign for transient response is necessary using methods discussed in Chapter 9. 308 Chapter 8: Root Locus Techniques 51. a. 1 x 104 Configuration A 0.8 0.6 0.4 Imag Axis 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 1 -2000 -1500 x 104 -1000 -500 Real Axis 0 500 1000 Configuration B 0.8 0.6 0.4 Imag Axis 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -2000 -1500 -1000 Real Axis -500 0 500 Solutions to Design Problems 309 1 x 104 Configuration C 0.8 0.6 0.4 Imag Axis 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -2000 -1500 -1000 -500 0 500 Real Axis 1000 1500 2000 2500 b. Configuration A: System is always unstable. Configuration B: root locus crosses jω axis at j2897 with a gain of 3.22 x 106. Thus, for stability, K < 3.22 x 106. Configuration C: root locus crosses jω axis at j1531 with a gain of 9.56 x 105. System is unstable at high gains. Thus, for stability, 9.56 x 105 > K. 52. a. Using MATLAB and the Symbolic Math Toolbox, the open-loop expression that yields a root locus as a function of N2 is 0.2284x107N2 (s2 + 3.772e-05s + 66.27) (s2 + 49.99s + 8789) Gdt(s) = ________________ s(s+45.12) (s2 + 4.893s + 8.777e04) 310 Chapter 8: Root Locus Techniques Program: syms s N KLSS KHSS KG JR JG tel s numGdt=3.92*N^2*KLSS*KHSS*KG*s; denGdt=(N^2*KHSS*(JR*s^2+KLSS)*(JG*s^2*[tel*s+1]+KG*s)+JR*s^2*KLSS*[(JG*s^2 +KHSS)*(tel*s+1)+KG*s]); Gdt=numGdt/denGdt; 'Gdt in General Terms' pretty(Gdt) 'Values to Substitute' KLSS=12.6e6 KHSS=301e3 KG=668 JR=190120 JG=3.8 tel=20e-3 numGdt=3.92*N^2*KLSS*KHSS*KG*s; Solutions to Design Problems 311 numGdt=vpa(numGdt,4); denGdt=(N^2*KHSS*(JR*s^2+KLSS)*(JG*s^2*[tel*s+1]+KG*s)+JR*s^2*KLSS*[(JG*s^2 +KHSS)*(tel*s+1)+KG*s]); denGdt=vpa(denGdt,4); 'Gdt with Values Substituted' Gdt=numGdt/denGdt; pretty(Gdt) Gdt=expand(Gdt); Gdt=vpa(Gdt,4); 'Gdt Different Form 1' pretty(Gdt); denGdt=collect(denGdt,N^2); 'Gdt Different Form 2' Gdt=collect(Gdt,N^2); pretty(Gdt) [numGdt,denGdt]=numden(Gdt); numGdt=numGdt/0.4349e10; denGdt=denGdt/0.4349e10; denGdt=expand(denGdt); denGdt=collect(denGdt,N^2); Gdt=vpa(numGdt/denGdt,4); 'Gdt Different Form 3' pretty(Gdt) 'Putting into Form for RL as a Function of N^2 using previous results' numGH=[1 49.99 8855 3313 582400]; denGH=[41.87 2094 0.3684e7 0.1658e9 0]; denGH=denGH/denGH(1) GH=tf(numGH,denGH) GHzpk=zpk(GH) 'Zeros of GH' rootsnumGH=roots(numGH) 'Poles of GH' rootsdenGH=roots(denGH) K=0:1:10000; rlocus(GH,K) sgrid(0.5,0) pause axis([-10,0,-20,20]) [K,P]=rlocfind(GH) Computer response: ans = Gdt in General Terms 98 2 / 2 2 2 -- N KLSS KHSS KG s / (N KHSS (JR s + KLSS) (JG s (tel s + 1) + KG s) 25 / 2 2 + JR s KLSS ((JG s + KHSS) (tel s + 1) + KG s)) ans = Values to Substitute KLSS = 12600000 KHSS = 301000 312 Chapter 8: Root Locus Techniques KG = 668 JR = 190120 JG = 3.8000 tel = 0.0200 ans = Gdt with Values Substituted 16 2 .9931 10 Ns / / (301000. / 2 2 8 2 N (190100. s + .1260 10 ) (3.800 s (.02000 s + 1.) + 668. s) 13 2 2 + .2396 10 s ((3.800 s + 301000.) (.02000 s + 1.) + 668. s)) ans = Gdt Different Form 1 16 2 .9931 10 Ns / 10 2 5 12 2 4 14 2 3 / (.4349 10 N s + .2174 10 N s + .3851 10 Ns / 14 2 2 16 2 12 5 13 4 + .1441 10 N s + .2533 10 N s + .1821 10 s + .9105 10 s 17 3 18 2 + .1602 10 s + .7212 10 s) ans = Gdt Different Form 2 16 2 .9931 10 Ns / 10 5 12 4 14 3 / ((.4349 10 s + .2174 10 s + .3851 10 s / 14 2 16 2 18 2 12 5 + .1441 10 s + .2533 10 s) N + .7212 10 s + .1821 10 s 13 4 17 3 + .9105 10 s + .1602 10 s) ans = Solutions to Design Problems 313 Gdt Different Form 3 72 .2284 10 N s / /( / 5 4 3 2 2 (1.000 s + 49.99 s + 8855. s + 3313. s + 582400. s) N 92 5 4 73 + .1658 10 s + 41.87 s + 2094. s + .3684 10 s ) ans = Putting into Form for RL as a Function of N^2 using previous results denGH = 1.0e+006 * Columns 1 through 4 0.0000 0.0001 0.0880 3.9599 Column 5 0 Transfer function: s^4 + 49.99 s^3 + 8855 s^2 + 3313 s + 582400 -------------------------------------------s^4 + 50.01 s^3 + 8.799e004 s^2 + 3.96e006 s Zero/pole/gain: (s^2 + 66.27) (s^2 + 49.99s + 8789) --------------------------------------s (s+45.12) (s^2 + 4.893s + 8.777e004) ans = Zeros of GH rootsnumGH = -24.9950 -24.9950 -0.0000 -0.0000 +90.3548i -90.3548i + 8.1404i - 8.1404i ans = Poles of GH rootsdenGH = 1.0e+002 * 0 314 Chapter 8: Root Locus Techniques -0.0245 + 2.9624i -0.0245 - 2.9624i -0.4512 Select a point in the graphics window selected_point = -3.8230 + 6.5435i K= 51.5672 P= -21.1798 -21.1798 -3.8154 -3.8154 +97.6282i -97.6282i + 6.5338i - 6.5338i b. From the computer response, K = 0.2284x107N2 = 49.6. Therefore, N is approximately 5/1000. 53. a. Yh-Ycat Spring displacement Desired force 1 100 Input voltage+ 1 1000 Controller - Input transducer F up K 0 .7883 ( s + 53 .85 ) ( s2 + 15 .47 s + 9283 )( s 2 + 8 .119 s + 376 .3 ) Actuator 82300 Pantograph dynamics F out Spring 1 100 Sensor Y h-Ycat Spring displacement Desired force + Fup 100 1 1000 Controller - G(s) = 0 .7883 ( s + 53 .85 ) (s 2 + 15 .47 s + 9283 )( s 2 + 8 .119 s + 376 .3 ) Actuator K Pantograph dynamics Yh ( s) − Ycat (s) 0.7883( s + 53.85) =2 2 Fup ( s) (s + 15.47s + 9283)(s + 8.119 s + 376.3) Ge(s)=(K/100)*(1/1000)*G(s)*82.3e3 0.6488K (s+53.85) Ge(s) = 2 __________________________ (s + 8.119s + 376.3) (s^2 + 15.47s + 9283) 82300 Spring Fout Solutions to Design Problems 315 100 80 60 40 Imag Axis 20 0 -20 -40 -60 -80 -100 -100 -80 -60 -40 Real Axis -20 0 b. 38% overshoot yields ζ = 0.294. The ζ = 0.294 line intersects the root locus at –9 + j27.16. Here, Ke = 7.179 x 104. Thus K = Ke/0.6488, or K = 1.107 x 105. c. Ts = 4/Re = 4/9 = 0.44 s; Tp = π/Im = π/27.16 = 0.116 s d. Nondominant closed-loop poles are located at –3.4 ± j93.94. Thus poles are closer to the imaginary axis than the dominant poles. Second order approximation not valid. e. Program: syms s numg=(s+53.85); deng=(s^2+15.47*s+9283)*(s^2+8.119*s+376.3); numg=sym2poly(numg); deng=sym2poly(deng); G=tf(numg,deng) K=7.179e4 Ke=0.6488*K T=feedback(Ke*G,1) step(T) Computer response: Transfer function: s + 53.85 ------------------------------------------------s^4 + 23.59 s^3 + 9785 s^2 + 8.119e004 s + 3.493e006 K= 71790 Ke = 316 Chapter 8: Root Locus Techniques 4.6577e+004 Transfer function: 4.658e004 s + 2.508e006 ------------------------------------------------s^4 + 23.59 s^3 + 9785 s^2 + 1.278e005 s + 6.001e006 Tp = 0.12 s, Ts = 0.6 s, %OS = 0.66 − 0.42 = 57.1% . 0.42 NINE Design via Root Locus SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Lag-Lead Compensation 76.39K a. Uncompensated: From the Chapter 8 Case Study Challenge, G(s) = s(s+150)(s+1.32) = 7194.23 -1 6.9 s(s+150)(s+1.32) with the dominant poles at - 0.5 ± j6.9. Hence, ζ = cos (tan 0.5 ) = 0.0723, or %OS = 79.63% and Ts = 4 4 7194.23 = 0.5 = 8 seconds. Also, Kv = 150 x 1.32 = 36.33. ζωn b. Lead-Compensated: Reducing the percent overshoot by a factor of 4 yields, %OS = 79.63 = 4 8 19.91%, or ζ = 0.457. Reducing the settling time by a factor of 2 yields, Ts = 2 = 4. Improving Kv by 2 yields Kv = 72.66. Using Ts = 4 = 4, ζωn = 1, from which ωn = 2.188 rad/s. Thus, the ζωn design point equals -ζωn + j ωn 1-ζ2 = -1 + j1.946. Using the system's original poles and assuming a lead compensator zero at -1.5, the summation of the system's poles and the lead compensator zero to the design point is -123.017o . Thus, the compensator pole must contribute 1.946 123.017o-180o = -56.98o. Using the geometry below, p - 1 = tan 56.98o, or pc = 2.26. c Adding this pole to the system poles and the compensator zero yields 76.39K = 741.88 at -1+j1.946. Hence the lead-compensated open-loop transfer function is GLead-comp(s) = 318 Chapter 9: Design Via Root Locus 741.88(s + 1.5) . Searching the real axis segments of the root locus yields higher-order s(s + 150)(s + 1.32)(s + 2.26) poles at greater than -150 and at -1.55. The response should be simulated since there may not be pole/zero cancellation. The lead-compensated step response is shown below. Since the settling time and percent overshoot meet the transient requirements, proceed with the lag 741.88 x 1.5 compensator. The lead-compensated system has Kv = 150 x 1.32 x 2.26 = 2.487. Since we want Kv s+0.002922 72.66 = 72.66, an improvement of 2.487 = 29.22 is required. Select G(s)Lag = s+0.0001 to improve the steady-state error by 29.22. A simulation of the lag-lead compensated system, 741.88(s+1.5)(s+0.002922) GLag-lead-comp(s) = s(s+150)(s+1.32)(s+2.26)(s+0.0001) is shown below. Answers to Review Questions 319 UFSS Vehicle: Lead and Feedback Compensation 0.25K2(s+0.437) Minor loop: Open-loop transfer function G(s)H(s) = (s+2)(s+1.29)(s+0.193) ; Closed-loop transfer function: TML (s) = 0.25K 2 (s + 0.437) . Searching along the 126.87o line (ζ = 0.6), find the s(s3 + ...) dominant second-order poles at -1.554 ± j2.072 with 0.25K2 = 4.7. Thus K2 = 18.8. Searching the real axis segment of the root locus for a gain of 4.7 yields a 3rd pole at -0.379. Major loop: The unity feedback, open-loop transfer function found by using the minor-loop closed-0.25K1(s+0.437) loop poles is GML(s) = s(s+0.379)(s+1.554+j2.072)(s+1.554-j2.072) . Searching along the 120o line (ζ = 0.5), find the dominant second-order poles at -1.069±j1.85 with 0.25K1 = 4.55. Thus K1 = 18.2. Searching the real axis segment of the root locus for a gain of 4.55 yields a 3rd pole at -0.53 and a 4th pole at -0.815. ANSWERS TO REVIEW QUESTIONS 1. Chapter 8: Design via gain adjustment. Chapter 9: Design via cascaded or feedback filters 2. A. Permits design for transient responses not on original root locus and unattainable through simple gain adjustments. B. Transient response and steady-state error specifications can be met separately and independently without the need for tradeoffs 3. PI or lag compensation 4. PD or lead compensation 5. PID or lag-lead compensation 6. A pole is placed on or near the origin to increase or nearly increase the system type, and the zero is placed near the pole in order not to change the transient response. 7. The zero is placed closer to the imaginary axis than the pole. The total contribution of the pole and zero along with the previous poles and zeros must yield 1800 at the design point. Placing the zero closer to the imaginary axis tends to speed up a slow response. 8. A PD controller yields a single zero, while a lead network yields a zero and a pole. The zero is closer to the imaginary axis. 9. Further out along the same radial line drawn from the origin to the uncompensated poles 10. The PI controller places a pole right at the origin, thus increasing the system type and driving the error to zero. A lag network places the pole only close to the origin yielding improvement but no zero error. 11. The transient response is approximately the same as the uncompensated system, except after the original settling time has passed. A slow movement toward the new final value is noticed. 12. 25 times; the improvement equals the ratio of the zero location to the pole location. 320 Chapter 9: Design Via Root Locus 13. No; the feedback compensator's zero is not a zero of the closed-loop system. 14. A. Response of inner loops can be separately designed; B. Faster responses possible; C. Amplification may not be necessary since signal goes from high amplitude to low. SOLUTIONS TO PROBLEMS 1. Uncompensated system: Search along the ζ = 0.5 line and find the operating point is at -1.5356 ± j2.6598 with K = 73.09. Hence, = %OS = e −ζπ / 1 −ζ 2 x100 = 16.3%; Ts = 4 = 2.6 seconds; Kp 1.5356 73.09 =2.44. A higher-order pole is located at -10.9285. 30 Compensated: Add a pole at the origin and a zero at -0.1 to form a PI controller. Search along the ζ = 0.5 line and find the operating point is at -1.5072 ± j2.6106 with K = 72.23. Hence, the estimated performance specifications for the compensated system are: %OS = e −ζπ / 1 −ζ 2 x100 = 16.3%; Ts = 4 = 2.65 seconds; Kp = ∞. Higher-order poles are located at -0.0728 and -10.9125. The 1.5072 compensated system should be simulated to ensure effective pole/zero cancellation. 2. a. Insert a cascade compensator, such as Gc (s ) = s + 0.01 . s b. Program: K=1 G1=zpk(,[0,-2,-5],K) Gc=zpk([-0.01],[0],1) G=G1*Gc rlocus(G) T=feedback(G,1) T1=tf(1,[1,0]) T2=T*T1 t=0:0.1:200; step(T1,T2,t) Computer response: K= 1 Zero/pole/gain: 1 ------------s (s+2) (s+5) %G1=1/s(s+2)(s+5) %Gc=(s+0.01)/s %Form 1/s to integrate step input %Show input ramp and ramp response Solutions to Problems 321 Zero/pole/gain: (s+0.01) -------s Zero/pole/gain: (s+0.01) --------------s^2 (s+2) (s+5) Zero/pole/gain: (s+0.01) -----------------------------------------(s+5.064) (s+1.829) (s+0.09593) (s+0.01126) Transfer function: 1 s Zero/pole/gain: (s+0.01) -----------------------------------------s (s+5.064) (s+1.829) (s+0.09593) (s+0.01126) 3. a. Searching along the 126.16o line (10% overshoot, ζ = 0.59), find the operating point at 322 Chapter 9: Design Via Root Locus 20 -1.4 + j1.92 with K = 20. Hence, Kp = 1 x 5 x 3 = 1.333. s+0.3 b. A 3x improvement will yield Kp = 4. Use a lag compensator, Gc(s) = s+0.1 . c. 4. a. Searching along the 126.16o line (10% overshoot, ζ = 0.59), find the operating point at -1.009 + j1.381 with K = 17.5. Hence, Kv = 17.5 = 1.1667. 5x3 b. A 3.429x improvement will yield Kv = 4. Use a lag compensator, Gc(s) = s + 0.3429 . s + 0.1 c. Program: K=17.5 G=zpk(,[0,-3,-5],K) Gc=zpk([-0.3429],[-0.1],1) Ge=G*Gc; T1=feedback(G,1); T2=feedback(Ge,1); T3=tf(1,[1,0]); %Form 1/s to integrate step input T4=T1*T3; T5=T2*T3; t=0:0.1:20; step(T3,T4,T5,t) %Show input ramp and ramp responses Solutions to Problems 323 Computer response: K= 17.5000 Zero/pole/gain: 17.5 ------------s (s+3) (s+5) Zero/pole/gain: (s+0.3429) ---------(s+0.1) 5. a. Uncompensated: Searching along the 126.16o line (10% overshoot, ζ = 0.59), find the operating 45.72 20 point at -2.03 + j2.77 with K = 45.72. Hence, Kp = 2 x 4 x 6 = 0.9525. An improvement of 0.9525 0.201 = 20.1 is required. Let Gc(s) = 0.01 . Compensated: Searching along the 126.16o line (10% overshoot, ζ = 0.59), find the operating point at - 1.99+j2.72 with K = 46.05. Hence, Kp = 46.05 x 0.201 2 x 4 x 6 x 0.01 = 19.28. 324 Chapter 9: Design Via Root Locus b. c. From (b), about 28 seconds 6. Uncompensated: Searching along the 135o line (ζ = 0.707), find the operating point at -2.32 + j2.32 with K = 4.6045. Hence, Kp = π 2.32 ωn = = 1.354 seconds; %OS = e −ζπ / 1−ζ 2 4.6045 4 = 0.153; Ts = = 1.724 seconds; Tp = 30 2.32 x100 = 4.33%; 2.322 + 2.322 = 3.28 rad/s; higher-order pole at -5.366. Compensated: To reduce the settling time by a factor of 2, the closed-loop poles should be – 4.64 ± j4.64. The summation of angles to this point is 119o . Hence, the contribution of the compensating zero should be 180o -119o =61o . Using the geometry shown below, 4.64 = tan (61o). Or, zc = 7.21. zc − 4.64 Solutions to Problems 325 After adding the compensator zero, the gain at -4.64+j4.64 is K = 4.77. Hence, Kp = 4.77 x 6 x 7.21 4 π = 6.88 . Ts = = 0.86 second; Tp = = 0.677 second; 4.64 2 x3 x5 4.64 %OS = e −ζπ / 1−ζ 2 x100 = 4.33%; ωn = 4.642 + 4.642 = 6.56 rad/s; higher-order pole at -5.49. The problem with the design is that there is steady-state error, and no effective pole/zero cancellation. The design should be simulated to be sure the transient requirements are met. 7. Program: clf 'Uncompensated System' numg=[1 6]; deng=poly([-2 -3 -5]); 'G(s)' G=tf(numg,deng); Gzpk=zpk(G) rlocus(G,0:1:100) z=0.707; pos=exp(-pi*z/sqrt(1-z^2))*100; sgrid(z,0) title(['Uncompensated Root Locus with ' , num2str(z), ' Damping Ratio Line']) [K,p]=rlocfind(G); %Allows input by selecting point on graphic 'Closed-loop poles = ' p i=input('Give pole number that is operating point '); 'Summary of estimated specifications' operatingpoint=p(i) gain=K estimated_settling_time=4/abs(real(p(i))) estimated_peak_time=pi/abs(imag(p(i))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(i))^2+imag(p(i))^2) Kp=dcgain(K*G) 'T(s)' T=feedback(K*G,1) 'Press any key to continue and obtain the step response' pause step(T) title(['Step Response for Uncompensated System with ' , num2str(z),... ' Damping Ratio']) 'Press any key to go to PD compensation' pause 'Compensated system' 326 Chapter 9: Design Via Root Locus done=1; while done>0 a=input('Enter a Test PD Compensator, (s+a). a = ') numc=[1 a]; 'Gc(s)' GGc=tf(conv(numg,numc),deng); GGczpk=zpk(GGc) wn=4/[(estimated_settling_time/2)*z]; rlocus(GGc) sgrid(z,wn) title(['PD Compensated Root Locus with ' , num2str(z),... ' Damping Ratio Line', 'PD Zero at ', num2str(a), ', and Required Wn']) done=input('Are you done? (y=0,n=1) '); end [K,p]=rlocfind(GGc); %Allows input by selecting point on graphic 'Closed-loop poles = ' p i=input('Give pole number that is operating point '); 'Summary of estimated specifications' operatingpoint=p(i) gain=K estimated_settling_time=4/abs(real(p(i))) estimated_peak_time=pi/abs(imag(p(i))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(i))^2+imag(p(i))^2) Kp=dcgain(K*GGc) 'T(s)' T=feedback(K*GGc,1) 'Press any key to continue and obtain the step response' pause step(T) title(['Step Response for Compensated System with ' , num2str(z),... ' Damping Ratio']) Computer response: ans = Uncompensated System ans = G(s) Zero/pole/gain: (s+6) ----------------(s+5) (s+3) (s+2) Select a point in the graphics window selected_point = -2.3104 + 2.2826i ans = Closed-loop poles = p= -5.3603 -2.3199 + 2.2835i Solutions to Problems 327 -2.3199 - 2.2835i Give pole number that is operating point 2 ans = Summary of estimated specifications operatingpoint = -2.3199 + 2.2835i gain = 4.4662 estimated_settling_time = 1.7242 estimated_peak_time = 1.3758 estimated_percent_overshoot = 4.3255 estimated_damping_ratio = 0.7070 estimated_natural_frequency = 3.2552 Kp = 0.8932 ans = T(s) Transfer function: 4.466 s + 26.8 ----------------------------s^3 + 10 s^2 + 35.47 s + 56.8 ans = Press any key to continue and obtain the step response ans = 328 Chapter 9: Design Via Root Locus Press any key to go to PD compensation ans = Compensated system Enter a Test PD Compensator, (s+a). a = 6 a= 6 ans = Gc(s) Zero/pole/gain: (s+6)^2 ----------------(s+5) (s+3) (s+2) Are you done? (y=0,n=1) 1 Enter a Test PD Compensator, (s+a). a = 7.1 a= 7.1000 ans = Gc(s) Zero/pole/gain: (s+7.1) (s+6) ----------------(s+5) (s+3) (s+2) Are you done? (y=0,n=1) 0 Select a point in the graphics window selected_point = -4.6607 + 4.5423i ans = Closed-loop poles = p= -4.6381 + 4.5755i -4.6381 - 4.5755i -5.4735 Give pole number that is operating point ans = Summary of estimated specifications 1 Solutions to Problems 329 operatingpoint = -4.6381 + 4.5755i gain = 4.7496 estimated_settling_time = 0.8624 estimated_peak_time = 0.6866 estimated_percent_overshoot = 4.3255 estimated_damping_ratio = 0.7070 estimated_natural_frequency = 6.5151 Kp = 6.7444 ans = T(s) Transfer function: 4.75 s^2 + 62.22 s + 202.3 --------------------------------s^3 + 14.75 s^2 + 93.22 s + 232.3 ans = Press any key to continue and obtain the step response 330 Chapter 9: Design Via Root Locus Solutions to Problems 331 332 Chapter 9: Design Via Root Locus 8. The uncompensated system performance is summarized in Table 9.8 in the text. To improve settling time by 4, the dominant poles need to be at -7.236 ± j14.123. Summing the angles from the open-loop poles to the design point yields -277.326o. Thus, the zero must contribute 277.326o - 180o = 97.326o. Using the geometry below, jω j14.123 s-plane 97.236o -7.236 -zc σ 14.123 7.236 - zc = tan(180-97.326). Thus, zc = 5.42. Adding the zero and evaluating the gain at the design point yields K = 256.819. Summarizing results: Solutions to Problems 333 9. 4 a. ζωn = T = 2.5; ζ = s %OS - ln ( 100 ) = 0.404. Thus, ωn = 6.188 rad/s and the operating %OS π2 + ln2 ( 100 ) point is - 2.5 ± j5.67. b. Summation of angles including the compensating zero is -150.06o. Therefore, the compensator pole must contribute 150.06o - 180o = -29.94o. 5.67 c. Using the geometry shown below, p - 2.5 c = tan 29.94o. Thus, pc = 12.34. 334 Chapter 9: Design Via Root Locus d. Adding the compensator pole and using -2.5 + j5.67 as the test point, K = 357.09. e. Searching the real axis segments for K = 1049.41, we find higher-order poles at -15.15, and -1.186. f. Pole at -15.15 is more than 5 times further from the imaginary axis than the dominant poles. Pole at -1.186 may not cancel the zero at -1 g. A simulation of the system shows a percent overshoot of 37.5% and a settling time of 2.12 seconds. Thus, the specifications were not met because pole-zero cancellation was not achieved. A redesign is required. 10. 4 a. ζωn = T = 2.4; ζ = s %OS - ln ( 100 ) = 0.5. Thus, ωn = 4.799 rad/s and the operating point is 2 + ln2 (%OS) π 100 -2.4 ± j4.16. b. Summation of angles including the compensating zero is -131.36o. Therefore, the compensator 4.16 pole must contribute 180o - 131.36o = -48.64o. Using the geometry shown below, p - 2.4 c = Solutions to Problems 335 tan 48.64o. Thus, pc = 6.06. c. Adding the compensator pole and using -2.4 + j4.16 as the test point, K = 29.117. d. Searching the real axis segments for K = 29.117, we find a higher-order pole at -1.263. e. Pole at -1.263 is near the zero at -1. Simulate to ensure accuracy of results. 29.117 f. Ka = 6.06 = 4.8 g. From the plot, Ts = 1.4 seconds; Tp = 0.68 seconds; %OS = 35%. 336 Chapter 9: Design Via Root Locus 11. a. Uncompensated Root Locus with 0.8 Damping Line 10 8 6 4 Imag Axis 2 0 -2 -4 -6 -8 -10 -10 -8 -6 -4 Real Axis -2 0 2 b. and c. Searching along the ζ = 0.8 line (143.13o), find the operating point at –2.682 + j2.012 with K = 35.66. 4 d. Since ζωn = T , the real part of the compensated dominant pole is -4. The imaginary part is s 4 tan (180o-143.13o) = 3. Using the uncompensated system's poles and zeros along with the compensator zero at - 4.5, the summation of angles to the design point, -4 + j3 is –158.71o. Thus, the contribution of the compensator pole must be 158.71o - 180o = -21.290. Using the following geometry, 3 = tan 21.290, or pc = 11.7. pc − 4 3 21.29 4 Adding the compensator pole and using – 4 + j3 as the test point, K = 172.92. e. Compensated: Searching the real axis segments for K = 172.92, we find higher-order poles at 14.19, and approximately at –5.26 ± j0.553. Since there is no pole/zero cancellation with the zeros at -6 and –4.5, the system should be simulated to check the settling time. Solutions to Problems 337 f. Step Response 0.45 0.4 0.35 Amplitude 0.3 0.25 0.2 0.15 0.1 0.05 0 0 0.5 1 Time (sec.) The graph shows about 2% overshoot and a 0.8 second settling time compared to a desired 1.52% overshoot and a settling time of 1 second. 12. Program: clf numg=[1 6]; deng=poly([-2 -4 -7 -8]); 'G(s)' G=tf(numg,deng); Gzpk=zpk(G) rlocus(G) z=0.8; pos=exp(-pi*z/sqrt(1-z^2))*100; sgrid(z,0) title(['Uncompensated Root Locus with ' , num2str(z), ' Damping Ratio Line']) [K,p]=rlocfind(G); 'Closed-loop poles = ' p i=input('Give pole number that is operating point '); 'Summary of estimated specifications' operatingpoint=p(i) gain=K estimated_settling_time=4/abs(real(p(i))) estimated_peak_time=pi/abs(imag(p(i))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(i))^2+imag(p(i))^2) Kp=K*numg(max(size(numg)))/deng(max(size(deng))) 'T(s)' T=feedback(K*G,1) 'Press any key to continue and obtain the step response' pause step(T) title(['Step Response for Uncompensated System with ' , num2str(z),... ' Damping Ratio']) 'Press any key to go to Lead compensation' pause 338 Chapter 9: Design Via Root Locus 'Compensated system' b=4.5; 'Lead Zero at -4.5 ' done=1; while done>0 a=input('Enter a Test Lead Compensator Pole, (s+a). a = '); 'Gc(s)' Gc=tf([1 b],[1 a]) GGc=G*Gc; [numggc,denggc]=tfdata(GGc,'v'); 'G(s)Gc(s)' GGczpk=zpk(GGc) wn=4/((1)*z); rlocus(GGc); sgrid(z,wn) title(['Lead Compensated Root Locus with ' , num2str(z),... ' Damping Ratio Line, Lead Pole at ', num2str(-a), ', and Required Wn']) done=input('Are you done? (y=0,n=1) '); end [K,p]=rlocfind(GGc); %Allows input by selecting point on graphic 'Closed-loop poles = ' p i=input('Give pole number that is operating point '); 'Summary of estimated specifications' operatingpoint=p(i) gain=K estimated_settling_time=4/abs(real(p(i))) estimated_peak_time=pi/abs(imag(p(i))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(i))^2+imag(p(i))^2) Kp=dcgain(K*GGc) 'T(s)' T=feedback(K*GGc,1) 'Press any key to continue and obtain the step response' pause step(T) title(['Step Response for Compensated System with ' , num2str(z),... ' Damping Ratio']) Computer response: ans = G(s) Zero/pole/gain: (s+6) ----------------------(s+8) (s+7) (s+4) (s+2) Select a point in the graphics window selected_point = -2.7062 + 2.0053i ans = Closed-loop poles = p= -9.3056 -6.3230 Solutions to Problems 339 -2.6857 + 2.0000i -2.6857 - 2.0000i Give pole number that is operating point 3 ans = Summary of estimated specifications operatingpoint = -2.6857 + 2.0000i gain = 35.2956 estimated_settling_time = 1.4894 estimated_peak_time = 1.5708 estimated_percent_overshoot = 1.5165 estimated_damping_ratio = 0.8000 estimated_natural_frequency = 3.3486 Kp = 0.4727 ans = T(s) Transfer function: 35.3 s + 211.8 ---------------------------------------s^4 + 21 s^3 + 154 s^2 + 491.3 s + 659.8 ans = Press any key to continue and obtain the step response ans = 340 Chapter 9: Design Via Root Locus Press any key to go to Lead compensation ans = Compensated system ans = Lead Zero at -4.5 Enter a Test Lead Compensator Pole, (s+a). a = 10 ans = Gc(s) Transfer function: s + 4.5 ------s + 10 ans = G(s)Gc(s) Zero/pole/gain: (s+6) (s+4.5) -----------------------------(s+10) (s+8) (s+7) (s+4) (s+2) Are you done? (y=0,n=1) 1 Enter a Test Lead Compensator Pole, (s+a). a = ans = Gc(s) Transfer function: s + 4.5 -------s + 11.7 ans = G(s)Gc(s) Zero/pole/gain: (s+6) (s+4.5) -------------------------------(s+11.7) (s+8) (s+7) (s+4) (s+2) Are you done? (y=0,n=1) 0 Select a point in the graphics window selected_point = -3.9885 + 3.0882i 11.7 Solutions to Problems 341 ans = Closed-loop poles = p= -14.2326 -3.9797 -3.9797 -5.2540 -5.2540 + + - 3.0860i 3.0860i 0.5076i 0.5076i Give pole number that is operating point ans = Summary of estimated specifications operatingpoint = -3.9797 + 3.0860i gain = 178.3530 estimated_settling_time = 1.0051 estimated_peak_time = 1.0180 estimated_percent_overshoot = 1.5165 estimated_damping_ratio = 0.8000 estimated_natural_frequency = 5.0360 Kp = 0.9187 ans = T(s) Transfer function: 2 342 Chapter 9: Design Via Root Locus 178.4 s^2 + 1873 s + 4816 -----------------------------------------s^5 + 32.7 s^4 + 399.7 s^3 + 2436 s^2 + 7656 s + 1.006e004 ans = Press any key to continue and obtain the step response Solutions to Problems 343 344 Chapter 9: Design Via Root Locus 13. a. Searching along the 117.13o line (%OS = 20%; ζ = 0.456), find the operating point at Solutions to Problems 345 -6.39 + j12.47 with K = 9273. Searching along the real axis for K = 9273, we find a higher-order pole at –47.22. Thus, Ts = 4 ζω n = 4 = 0.626 second. 6.39 b. For the settling time to decrease by a factor of 2, Re = -ζωn = -6.39 x 2 = -12.78. The imaginary part is Im = -12.78 tan 117.13o = 24.94. Hence, the compensated closed-loop poles are -12.78 ± j24.94. A settling time of 0.313 second would result. c. Assume a compensator zero at -20. Using the uncompensated system's poles along with the compensator zero, the summation of angles to the design point, -12.78 ± j24.94 is –159.63o. Thus, the contribution of the compensator pole must be 159.63o-180o = -20.37o. Using the following geometry, 24.94 = tan 20.37o, or pc = 79.95. pc − 12.78 24.94 20.37 12.78 Adding the compensator pole and using --12.78 ± j24.94 as the test point, K = 74130. d. Step Response Uncompensated & Lead-Compensated System 1 Amplitude 0.8 0.6 0.4 0.2 0 0 0.15 0.3 0.45 Time (sec.) 0.6 0.75 346 Chapter 9: Design Via Root Locus 14. a. Searching along the 110.97o line (%OS = 30%; ζ= 0.358), find the operating point at -2.065 + j5.388 with K = 366.8. Searching along the real axis for K = 366.8, we find a higher-order pole at –16.87. Thus, Ts = 4 ζω n = 4 = 1.937 seconds. For the settling time to decrease by a 2.065 factor of 2, Re = -ζωn = -2.065 x 2 = - 4.13. The imaginary part is – 4.13 tan 110.970 = 10.77. Hence, the compensated dominant poles are – 4.13 ± j10.77. The compensator zero is at -7. Using the uncompensated system's poles along with the compensator zero, the summation of angles to the design point, – 4.13 ± j10.77 is –162.06o. Thus, the contribution of the compensator pole must be – 162.06o - 180o = -17.94o. Using the following geometry, 10.77 = tan 17.94o, or pc = 37.4. pc − 4.13 10.77 17.94 -4.13 Adding the compensator pole and using – 4.13 ± j10.77 as the test point, K = 5443. b. Searching the real axis segments for K = 5443 yields higher-order poles at approximately –8.12 and –42.02. The pole at –42.02 can be neglected since it is more than five times further from the imaginary axis than the dominant pair. The pole at –8.12 may not be canceling the zero at -7. Hence, simulate to be sure the requirements are met. c. Program: 'Uncompensated System G1(s)' numg1=1; deng1=poly([-15 (-3+2*j) (-3-2*j)]); G1=tf(numg1,deng1) G1zpk=zpk(G1) K1=366.8 'T1(s)' T1=feedback(K1*G1,1); T1zpk=zpk(T1) 'Compensator Gc(s)' numc=[1 7]; denc=[1 37.4]; Gc=tf(numc,denc) 'Compensated System G2(s) = G1(s)Gc(s)' K2=5443 G2=G1*Gc; G2zpk=zpk(G2) 'T2(s)' T2=feedback(K2*G2,1); T2zpk=zpk(T2) step(T1,T2) title(['Uncompensated and Lead Compensated Systems']) Solutions to Problems 347 Computer response: ans = Uncompensated System G1(s) Transfer function: 1 -------------------------s^3 + 21 s^2 + 103 s + 195 Zero/pole/gain: 1 ----------------------(s+15) (s^2 + 6s + 13) K1 = 366.8000 ans = T1(s) Zero/pole/gain: 366.8 --------------------------------(s+16.87) (s^2 + 4.132s + 33.31) ans = Compensator Gc(s) Transfer function: s+7 -------s + 37.4 ans = Compensated System G2(s) = G1(s)Gc(s) K2 = 5443 Zero/pole/gain: (s+7) -------------------------------(s+37.4) (s+15) (s^2 + 6s + 13) ans = T2(s) Zero/pole/gain: 348 Chapter 9: Design Via Root Locus 5443 (s+7) ------------------------------------------(s+42.02) (s+8.118) (s^2 + 8.261s + 133.1) 15. a. Searching the 15% overshoot line (121.127o) for 180o yields -0.372 + j0.615. Hence, Ts = 4 = σd 4 0.372 = 10.75 seconds. 4 4 b. For 7 seconds settling time, σd = T = 7 = 0.571. ωd = 0.571 tan (180o - 121.127o) = 0.946. s Therefore, the design point is -0.571 + j0.946. Summing the angles of the uncompensated system's poles as well as the compensator pole at -15 yields -213.493o. Therefore, the compensator zero must contribute (213.493o - 180o) = 33.493o. Using the geometry below, jω s-plane j0.946 33.493o -zc -0.571 σ 0.946 o zc - 0.571 = tan (33.493 ) . Hence, zc = 2. The compensated open-loop transfer function is Solutions to Problems 349 K(s+2) . Evaluating the gain for this function at the point, -0.571 + j0.946 yields K s(s+1)(s2+10s+26)(s+15) = 207.512. c. Program: numg= 207.512*[1 2]; r=roots([1,10,26]); deng=poly([0 ,-1, r(1),r(2),-15]); 'G(s)' G=tf(numg,deng); Gzpk=zpk(G) T=feedback(G,1); step(T) title(['Step Response for Design of Ts = 7, %OS = 15']) Computer response: ans = G(s) Zero/pole/gain: 207.512 (s+2) ------------------------------s (s+15) (s+1) (s^2 + 10s + 26) 16. a. From 20.5% overshoot evaluate ζ = 0.45 . Also, since ζωn = 44 = , ωn = 2.963 . The Ts 3 compensated dominant poles are located at -ζωn ± jωn 1-ζ2 = - 1.3333 ± j2.6432. Assuming the compensator zero at -0.02, the contribution of open-loop poles and the compensator zero to the design point, - 1.3333 ± j2.6432 is -175.78o. Hence, the compensator pole must contribute 350 Chapter 9: Design Via Root Locus 175.78o - 180o = -4.22o. Using the following geometry, 2.6432 = tan 4.22o , or pc = 37.16 pc − 1.3333 Adding the pole to the system, K = 4401.52 at the design point.. b. Searching along the real axis segments of the root locus for K = 4401.52, we find higher-order poles at -0.0202, -13.46, and -37.02. There is pole/zero cancellation at -0.02. Also, the poles at , -13.46, and -37.02 are at least 5 times the design point’s real part. Thus, the second-order approximation is valid. c. Solutions to Problems 351 From the plot, Ts = 2.81 seconds, and %OS = 20.8%. Thus, the requirements are met. 17. 4 4 a. ζωn = T = 0.5 = 8. Since ζ = 0.4, ωn = 20. Therefore the compensated closed-loop poles are s located at - ζωn ± jωn 1-ζ2 = -8 ± j18.33. b. Using the system's poles along with the compensator's pole at -15, the sum of angles to the test point –8 ± j18.33 is -293.4o . Therefore, the compensator's zero must contribute 293.4o - 180o = 18.33 113.4o . Using the following geometry, 8 - z = tan 66.6o, or zc = 0.0679. c c. Adding the compensator zero and using –8 ± j18.33 as the test point, K = 7297. d. Making a second-order assumption, the predicted performance is as follows: Uncompensated: Searching along the 133.58o line (ζ = 0.4), find the uncompensated closed-loop pole at -5.43 + j12.45 with K = 3353. Hence, Ts = 25.38%; Kp = 4 = 0.74 seconds; ζωn %OS = e −ζπ / 1−ζ 2 x100 = 3353 = 1.66. Checking the second-order assumption by searching the real axis 101x20 segments of the root locus for K = 3353, we find a higher-order pole at -29.13. Since this pole is more than five times further from the imaginary axis than the dominant pair, the second order assumption is reasonable. Compensated: Using the compensated dominant pole location, - 8 ± j18.33, Ts = seconds; %OS = e −ζπ / 1−ζ 2 x100 = 25.38%; Kp = 4 = 0.5 ζωn 7297x 0.0679 = 0.016. Checking the second101x 20x15 order assumption by searching the real axis segments of the root locus for K = 7297, we find higherorder poles at -2.086 and -36.91. The poles are not five times further from the imaginary axis nor do they yield pole/zero cancellation. The second-order assumption is not valid. 352 Chapter 9: Design Via Root Locus e. The uncompensated system exhibits a steady-state error of 0.38, a percent overshoot of 22.5%, and a settling time of 0.78 seconds. Since there is no pole/zero cancellation the closed-loop zero near the origin produces a large steadystate error. The student should be asked to find a viable design solution to this problem by choosing the compensator zero further from the origin. For example, placing the compensator zero at -20 yields a compensator pole at -90.75 and a gain of 28730. This design yields a valid second-order approximation. 18. a. Since %OS = 1.5%, ζ = %OS - ln ( 100 ) 4 2 = 0.8. Since Ts = = 3 second, ζωn %OS π2 + ln2 ( 100 ) ωn = 7.49 rad/s. Hence, the location of the closed-loop poles must be -6±j4.49. The summation of angles from open-loop poles to -6±j4.49 is -226.3o. Therefore, the design point is not on the root locus. Solutions to Problems 353 b. A compensator whose angular contribution is 226.3o-180o = 46.3o is required. Assume a compensator zero at -5 canceling the pole. Thus, the breakaway from the real axis will be at the required -6 if the compensator pole is at -9 as shown below. Adding the compensator pole and zero to the system poles, the gain at the design point is found to be s+5 29.16. Summarizing the results: Gc(s) = s+9 with K = 29.16. 19. Lead compensator design: Searching along the 120o line (ζ = 0.5), find the operating point at -1.531 + j2.652 with K = 354.5. Thus, Ts = 4 4 = 1.531 = 2.61 seconds. For the settling time to ζωn 4 decrease by 0.5 second, Ts = 2.11 seconds, or Re = -ζωn = - 2.11 = -1.9. The imaginary part is -1.9 tan 60o = 3.29. Hence, the compensated dominant poles are -1.9 ± j3.29. The compensator zero is at -5. Using the uncompensated system's poles along with the compensator zero, the summation of angles to the design point, -1.9 ± j3.29 is -166.09o. Thus, the contribution of the compensator pole 3.29 must be 166.09o - 180o = -13.91o. Using the following geometry, p - 1.9 = tan 13.91o, or pc = c 15.18. Adding the compensator pole and using -1.9 ± j3.29 as the test point, K = 1417. Computer simulations yield the following: Uncompensated: Ts = 3 seconds, %OS = 14.6%. 354 Chapter 9: Design Via Root Locus Compensated: Ts = 2.3 seconds, %OS = 15.3%. Lag compensator design: The lead compensated open-loop transfer function is GLC(s) = 1417(s + 5) . The uncompensated (s + 2)(s + 4)(s + 6)(s + 8)(s + 15.18) 1 Kp = 354.5/(2 x 4 x 6 x 8) = 0.923. Hence, the uncompensated steady-state error is 1+K p = 0.52. Since we want 30 times improvement, the lag-lead compensated system must have a steady-state error of 0.52/30 = 0.017. The lead compensated Kp = 1417*5/(2*4*6*8*15.18) = 1.215. Hence, the 1 lead-compensated error is 1+K = 0.451. Thus, the lag compensator must improve the leadp 1 compensated error by 0.451/0.017 = 26.529 times. Thus 0.451/ ( 1+K pllc ) = 26.529, where Kpllc = 57.823 is the lead-lag compensated system's position constant. Thus, the improvement in Kp from the lead to the lag-lead compensated system is 57.823/1.215 = 47.59. Use a lag compensator, whose zero is 47.59 times farther than its pole, or Glag = loop transfer function is GLLC(s) = (s + 0.04759) . Thus, the lead-lag compensated open(s + 0.001) 1417(s + 5)(s + 0.04759) . (s + 2)(s + 4)(s + 6)(s + 8)(s + 15.18)(s + 0.001) 20. Program: numg=1; deng=poly([-2 -4 -6 -8]); 'G(s)' G=tf(numg,deng); Gzpk=zpk(G) rlocus(G,0:5:500) z=0.5; pos=exp(-pi*z/sqrt(1-z^2))*100; sgrid(z,0) title(['Uncompensated Root Locus with ' , num2str(z), ' Damping Ratio Line']) [K,p]=rlocfind(G); %Allows input by selecting point on graphic 'Closed-loop poles = ' p i=input('Give pole number that is operating point '); 'Summary of estimated specifications for uncompensated system' operatingpoint=p(i) gain=K estimated_settling_time=4/abs(real(p(i))) estimated_peak_time=pi/abs(imag(p(i))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(i))^2+imag(p(i))^2) Kpo=dcgain(K*G) T=feedback(K*G,1); 'Press any key to continue and obtain the step response' pause step(T) whitebg('w') title(['Step Response for Uncompensated System with ' Damping Ratio'],'color','black') 'Press any key to go to Lead compensation' ' , num2str(z),... Solutions to Problems 355 pause 'Compensated system' b=5; 'Lead Zero at -b ' done=1; while done>0 a=input('Enter a Test Lead Compensator Pole, (s+a). a = '); numgglead=[1 b]; dengglead=conv([1 a],poly([-2 -4 -6 -8])); 'G(s)Glead(s)' GGlead=tf(numgglead,dengglead); GGleadzpk=zpk(GGlead) wn=4/((estimated_settling_time-0.5)*z); clf rlocus(GGlead,0:10:2000) sgrid(z,wn) axis([-10 0 -5 5]) title(['Lead Compensated Root Locus with ' , num2str(z),... ' Damping Ratio Line, Lead Pole at ', num2str(-a), ', and Required Wn']) done=input('Are you done? (y=0,n=1) '); end [K,p]=rlocfind(GGlead); %Allows input by selecting point on graphic 'Closed-loop poles = ' p i=input('Give pole number that is operating point '); 'Summary of estimated specifications for lead-compensated system' operatingpoint=p(i) gain=K estimated_settling_time=4/abs(real(p(i))) estimated_peak_time=pi/abs(imag(p(i))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(i))^2+imag(p(i))^2) Kplead=dcgain(K*GGlead) T=feedback(K*GGlead,1); 'Press any key to continue and obtain the step response' pause step(T) whitebg('w') title(['Step Response for Lead Compensated System with ' , num2str(z),... ' Damping Ratio'],'color','black') 'Press any key to continue and design lag compensation' pause 'Improvement in steady-state error with lead compensator is' error_improvement=(1+Kplead)/(1+Kpo) additional_error_improvement=30/error_improvement Kpnn=additional_error_improvement*(1+Kplead)-1 pc=0.001 zc=pc*(Kpnn/Kplead) numggleadlag=conv(numgglead,[1 zc]); denggleadlag=conv(dengglead,[1 pc]); 'G(s)Glead(s)Glag(s)' GGleadGlag=tf(numggleadlag,denggleadlag); GGleadGlagzpk=zpk(GGleadGlag) rlocus(GGleadGlag,0:10:2000) z=0.5; pos=exp(-pi*z/sqrt(1-z^2))*100; sgrid(z,0) title(['Lag-Lead Compensated Root Locus with ' , num2str(z), ... ' Damping Ratio Line and Lag Pole at ',num2str(-pc)]) [K,p]=rlocfind(GGleadGlag); %Allows input by selecting point on graphic 'Closed-loop poles = ' p i=input('Give pole number that is operating point '); 'Summary of estimated specifications for lag-lead compensated system' operatingpoint=p(i) 356 Chapter 9: Design Via Root Locus gain=K estimated_settling_time=4/abs(real(p(i))) estimated_peak_time=pi/abs(imag(p(i))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(i))^2+imag(p(i))^2) Kpleadlag=dcgain(K*GGleadGlag) T=feedback(K*GGleadGlag,1); 'Press any key to continue and obtain the step response' pause step(T) whitebg('w') title(['Step Response for Lag-Lead Compensated System with ', num2str(z),... ' Damping Ratio and Lag Pole at ',num2str(-pc)],'color','black') Computer response: ans = G(s) Zero/pole/gain: 1 ----------------------(s+8) (s+6) (s+4) (s+2) Select a point in the graphics window selected_point = -1.5036 + 2.6553i ans = Closed-loop poles = p= -8.4807 + 2.6674i -8.4807 - 2.6674i -1.5193 + 2.6674i -1.5193 - 2.6674i Give pole number that is operating point ans = 3 Solutions to Problems 357 Summary of estimated specifications for uncompensated system operatingpoint = -1.5193 + 2.6674i gain = 360.8014 estimated_settling_time = 2.6328 estimated_peak_time = 1.1778 estimated_percent_overshoot = 16.3034 estimated_damping_ratio = 0.5000 estimated_natural_frequency = 3.0698 Kpo = 0.9396 ans = 358 Chapter 9: Design Via Root Locus Press any key to continue and obtain the step response ans = Press any key to go to Lead compensation ans = Compensated system ans = Lead Zero at -b Enter a Test Lead Compensator Pole, (s+a). a = 10 ans = G(s)Glead(s) Zero/pole/gain: (s+5) -----------------------------(s+10) (s+8) (s+6) (s+4) (s+2) Are you done? (y=0,n=1) 1 Enter a Test Lead Compensator Pole, (s+a). a = ans = G(s)Glead(s) Zero/pole/gain: (s+5) -----------------------------(s+15) (s+8) (s+6) (s+4) (s+2) Are you done? (y=0,n=1) 0 Select a point in the graphics window 15 Solutions to Problems 359 selected_point = -1.9076 + 3.2453i ans = Closed-loop poles = p= -13.0497 + 1.9313i -13.0497 - 1.9313i -5.0654 -1.9176 + 3.2514i -1.9176 - 3.2514i Give pole number that is operating point 4 ans = Summary of estimated specifications for lead-compensated system operatingpoint = -1.9176 + 3.2514i gain = 1.3601e+003 estimated_settling_time = 2.0860 estimated_peak_time = 0.9662 360 Chapter 9: Design Via Root Locus estimated_percent_overshoot = 16.3034 estimated_damping_ratio = 0.5000 estimated_natural_frequency = 3.7747 Kplead = 1.1806 ans = Press any key to continue and obtain the step response ans = Press any key to continue and design lag compensation ans = Improvement in steady-state error with lead compensator is error_improvement = 1.1243 additional_error_improvement = 26.6842 Solutions to Problems 361 Kpnn = 57.1876 pc = 0.0010 zc = 0.0484 ans = G(s)Glead(s)Glag(s) Zero/pole/gain: (s+5) (s+0.04844) ---------------------------------------(s+15) (s+8) (s+6) (s+4) (s+2) (s+0.001) Select a point in the graphics window selected_point = -1.8306 + 3.2919i ans = Closed-loop poles = p= -13.0938 + 2.0650i -13.0938 - 2.0650i -5.0623 -1.8617 + 3.3112i -1.8617 - 3.3112i -0.0277 362 Chapter 9: Design Via Root Locus Give pole number that is operating point 4 ans = Summary of estimated specifications for lag-lead compensated system operatingpoint = -1.8617 + 3.3112i gain = 1.4428e+003 estimated_settling_time = 2.1486 estimated_peak_time = 0.9488 estimated_percent_overshoot = 16.3034 estimated_damping_ratio = 0.5000 estimated_natural_frequency = 3.7987 Kpleadlag = Solutions to Problems 363 60.6673 ans = Press any key to continue and obtain the step response 364 Chapter 9: Design Via Root Locus Solutions to Problems 365 366 Chapter 9: Design Via Root Locus 21. a. For the settling time to be 2.86 seconds with 4.32% overshoot, the real part of the compensated 4 4 dominant poles must be T = 2.86 = 1.4. Hence the compensated dominant poles are -1.4 ± j1.4. s Assume the compensator zero to be at -1 canceling the system pole at -1. The summation of angles to the design point at -1.4 ± j1.4 is -176.19o. Thus the contribution of the compensator pole is 1.4 176.19o - 180o = 3.81o. Using the geometry below, p - 1.4 = tan 3.81o, or pc = 22.42. c Adding the compensator pole and using -1.4 ± j1.4 as the test point, K = 88.68. b. Uncompensated: Search the 135o line (4.32% overshoot) and find the uncompensated dominant 1.11 4 4 pole at - 0.419 + j0.419 with K = 1.11. Thus Kv = 3 = 0.37. Hence, Ts = = 0.419 = 9.55 ζωn Solutions to Problems 367 88.68 seconds and %OS = 4.32%. Compensated: Kv = 22.42 x 3 = 1.32 (Note: steady-state error improvement is greater than 2). Ts = 4 4 = 1.4 = 2.86 seconds and %OS = 4.32%. ζωn c. Uncompensated: K = 1.11; Compensated: K = 88.68. d. Uncompensated: Searching the real axis segments for K = 1.11 yields a higher-order pole at -3.16 which is more than five times the real part of the uncompensated dominant poles. Thus the secondorder approximation for the uncompensated system is valid. Compensated: Searching the real axis segments for K = 88.68 yields a higher-order pole at -22.62 which is more than five times the real part of the compensated dominant poles' real part. Thus the second order approximation is valid. e. Step Response Uncompensated 1 0.9 0.8 Amplitude 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 5 10 Time (sec.) 368 Chapter 9: Design Via Root Locus Ramp Response Uncompensated 80 70 60 Amplitude 50 40 30 20 10 0 0 10 20 30 40 50 60 70 3 3.5 Time (sec.) Step Response Compensated 1 0.9 0.8 Amplitude 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 Time (sec.) 2.5 80 Solutions to Problems 369 Ramp Response Compensated 80 70 60 Amplitude 50 40 30 20 10 0 0 10 20 30 40 50 60 70 80 Time (sec.) 22. a. Searching the 30% overshoot line (ζ = 0.358; 110.97o) for 180o yields -1.464 + j3.818 with a gain, K = 218.6. b . Tp = π π = = 0.823 second. Kv = ωd 3.818 218.6 = 3.975. (5)(11) c. Lead design: From the requirements, the percent overshoot is 15% and the peak time is 0.4115 second. Thus, ζ = -ln(%/100) π2+ln2(%/100) π = 0.517; ωd = T = 7.634 = ωn 1-ζ2 . Hence, ωn = 8.919. The p design point is located at -ζωn + jωn 1-ζ2 = -4.61 + j7.634. Assume a lead compensator zero at -5. Summing the angles of the uncompensated system's poles as well as the compensator zero at -5 yields –171.2o. Therefore, the compensator pole must contribute (171.2o - 180o) = -8.8o. Using the geometry below, jω s-plane X -pc 8.8o -4.61 j7.634 σ 370 Chapter 9: Design Via Root Locus 7.634 = tan (8.8o) . Hence, pc = 53.92. The compensated open-loop transfer function is pc − 4.61 K . Evaluating the gain for this function at the point, -4.61 + j7.634 yields s(s + 11)( s + 53.92) K = 4430. Lag design: The uncompensated The lead compensated Kv = Kv = 218.6 = 3.975 . The required Kv is 30*3.975 = 119.25. (5)(11) 4430 = 7.469. Thus, we need an improvement over the lead (11)(53.92) compensated system of 119.25/7.469 = 15.97. Thus, use a lag compensator Glag(s) = 23. 4430(s + 0.01597) s + 0.01597 . The final open-loop function is . s + 0.001 s(s + 11)( s + 53.92)(s + 0.001) a. Searching along the 10% overshoot line (ζ = 0.591) the operating point is found to be –1.85 + j2.53 with K = 21.27. A third pole is at –10.29. Thus, the estimated performance before compensation is: 10% overshoot, Ts = 21.27 4 = 0.266 . = 2.16 seconds, and K p = (8)(10) 1.85 b. Lead design: Place compensator zero at –3. The desired operating point is found from the desired specifications. ζω n = 4 4 44 = 6.768 . Thus, = = 4 and ω n = = Ts 1 ζ 0.591 Im = ω n 1 − ζ = 6.768 1 − 0.591 = 5.46 . Hence the design point is –4 +j5.46. The angular 2 2 contribution of the system poles and compensator zero at the design point is –166.960. Thus, the compensator pole must contribute –1800 + 166.960 = -13.040. Using the geometry below, jω s-plane X -pc j5.46 13.04o -4 σ 5.46 = tan (13.04o) . Hence, pc = 27.57. The compensated open-loop transfer function is pc − 4 K (s + 3) . Evaluating the gain for this function at the point 2 (s + 4s + 8)( s + 10)( s + 27.57) -4 + j5.46 yields K = 1092 with higher-order poles at –4.055 and –29.52. Solutions to Problems 371 Lag design: For the lead-compensated system, Kp = 1.485. Thus, we need an improvement of (s + 0.06734) 10 . Finally, the equivalent forward-path = 6.734 times. Hence, Glag ( s) = 1.485 ( s + 0.01) 1092( s + 3)(s + 0.06734 ) . transfer function is Ge (s ) = 2 (s + 4 s + 8)(s + 10 )(s + 27.57)( s + 0.01) c. Step Response Lag-Lead Compensated 0.9 0.8 0.7 Amplitude 0.6 0.5 0.4 0.3 0.2 0.1 0 0 20 40 60 80 100 120 Time (sec.) Step Response Lag-Lead Compensated 0.7 0.6 Amplitude 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Time (sec.) 1.2 1.4 1.6 1.8 2 372 Chapter 9: Design Via Root Locus 24. a. Uncompensated: Search the 135o line (4.32% overshoot) for 180o and find the dominant pole at –3 + j3 with K = 10. Lag Compensated: Search the 135o line (4.32% overshoot) for 180o and find the dominant pole at 2.88 + j2.88 with K = 9.95. 10 b. Uncompensated: Kp = 2 x 4 = 1.25 9.95 x 0.5 Lag compensated: Kp = 2 x 4 x 0.1 = 6.22 c. %OS = 4.32% both cases; Uncompensated Ts = 4 4 4 = 3 1.33 seconds; Compensated Ts = 2.88 = 1.39 seconds ζωn d. Uncompensated: Exact second-order system; approximation OK Compensated: Search real axis segments of the root locus and find a higher-order pole at -0.3. System should be simulated to see if there is effective pole/zero cancellation with zero at - 0.5. e. Solutions to Problems 373 The compensated system's response takes a while to approach the final value. f. We will design a lead compensator to speed up the system by a factor of 5. The lead-compensated dominant poles will thus be placed at –15 ± j15. Assume a compensator zero at - 4 that cancels the open-loop pole at - 4. Using the system's poles and the compensator's zero, the sum of angles to the design point, -15±j15 is 131.69o. Thus, the angular contribution of the compensator pole must be 131.69o - 180o = - 48.31o. Using the geometry below, pc = 28.36. K(s+0.5)(s+4) Using the compensated open-loop transfer function, Ge(s) = (s+2)(s+4)(s+0.1)(s+28.36) and using the design point –15 ± j15, K = 404.1.The time response of the lag-lead compensated system is shown below. 25. π Since Tp = 1.047, the imaginary part of the compensated closed-loop poles will be 1.047 = 3. Im Since Re = tan (cos-1ζ), the magnitude of the real part will be point is – 4 + j3. Assume an PI controller, Gc(s) = Im = 4. Hence, the design tan(cos-1ζ) s+0.1 s , to reduce the steady-state error to zero. 374 Chapter 9: Design Via Root Locus Using the system's poles and the pole and zero of the ideal integral compensator, the summation of angles to the design point is -225.7o. Hence, the ideal derivative compensator must contribute 225.7o180o = 45.7o . Using the geometry below, zc = 6.93. The PID controller is thus (s+6.93)(s+0.1) . Using all poles and zeros of the system and PID s controller, the gain at the design point is K = 3.08. Searching the real axis segment, a higher-order pole is found at - 0.085. A simulation of the system shows the requirements are met. 26. a. The desired operating point is found from the desired specifications. ζω n = ωn = 2 ζ = 44 = = 2 and Ts 2 2 = 4.954 . Thus, Im = ωn 1 − ζ 2 = 4.954 1 − 0.4037 2 = 4.5324 . Hence 0.4037 the design point is –2 +j4.5324. Now, add a pole at the origin to increase system type and drive error to zero for step inputs. Now design a PD controller. The angular contribution to the design point of the system poles and pole at the origin is 101.90. Thus, the compensator zero must contribute 1800 – 101.90 =78.10. Using the geometry below, 4.5324 = tan(78.10 ) . Hence, zc = 2.955. The compensated open-loop transfer function with PD zc − 2 compensation is K ( s + 2.955) . Adding the compensator zero to the system and s ( s + 4)( s + 6)( s + 10) Solutions to Problems 375 evaluating the gain for this at the point –2 + j4.5324 yields K = 294.51 with a higher-order pole at -2.66 and -13.34. PI design: Use G PI (s ) = Ge ( s ) = ( s + 0.01) . Hence, the equivalent open-loop transfer function is s K ( s + 2.955)( s + 0.01) with K = 294.75. s 2 ( s + 4)( s + 6)( s + 10) b. Program (Step Response): numg=[-2.995 -0.01]; deng=[0 0 -4 -6 -10]; K=294.75; G=zpk(numg,deng,K) T=feedback(G,1); step(T) Computer response: Zero/pole/gain: 294.75 (s+2.995) (s+0.01) ------------------------s^2 (s+4) (s+6) (s+10) Program (Ramp Response): numg=[-2.995 -0.01]; deng=[0 0 -4 -6 -10]; K=294.75; G=zpk(numg,deng,K) T=feedback(G,1); Ta=tf([1],[1 0]); step(T*Ta) 376 Chapter 9: Design Via Root Locus Computer response: Zero/pole/gain: 294.75 (s+2.995) (s+0.01) ------------------------s^2 (s+4) (s+6) (s+10) 27. Program: numg= deng=[-4 -6 -10] 'G(s)' G=zpk(numg,deng,1) pos=input('Type desired percent overshoot '); z=-log(pos/100)/sqrt(pi^2+[log(pos/100)]^2); Ts=input('Type desired settling time '); zci=input(... 'Type desired position of integral controller zero (absolute value) '); wn=4/(Ts*z); desired_pole=(-z*wn)+(wn*sqrt(1-z^2)*i) angle_at_desired_pole=(180/pi)*angle(evalfr(G,desired_pole)) PD_angle=180-angle_at_desired_pole; zcpd=((imag(desired_pole)/tan(PD_angle*pi/180))-real(desired_pole)); 'PD Compensator' numcpd=[1 zcpd]; dencpd=[0 1]; 'Gcpd(s)' Gcpd=tf(numcpd,dencpd) Gcpi=zpk([-zci],[0],1) Ge=G*Gcpd*Gcpi rlocus(Ge) sgrid(z,0) title(['PID Compensated Root Locus with ' ,... num2str(pos), '% Damping Ratio Line']) [K,p]=rlocfind(Ge); 'Closed-loop poles = ' p f=input('Give pole number that is operating point '); 'Summary of estimated specifications for selected point' Solutions to Problems 377 'on PID compensated root locus' operatingpoint=p(f) gain=K estimated_settling_time=4/abs(real(p(f))) estimated_peak_time=pi/abs(imag(p(f))) estimated_percent_overshoot=pos estimated_damping_ratio=z estimated_natural_frequency=sqrt(real(p(f))^2+imag(p(f))^2) T=feedback(K*Ge,1); step(T) title(['Step Response for PID Compensated System with ' ,... num2str(pos),'% Damping Ratio Line']) pause one_over_s=tf(1,[1 0]); Tr=T*one_over_s; t=0:0.01:10; step(one_over_s,Tr) title('Ramp Response for PID Compensated System') Computer response: numg = deng = 0 -4 -6 -10 ans = G(s) Zero/pole/gain: 1 -------------------s (s+4) (s+6) (s+10) Type desired percent overshoot 25 Type desired settling time 2 Type desired position of integral controller zero (absolute value) 0.01 desired_pole = -2.0000 + 4.5324i angle_at_desired_pole = 101.8963 ans = PD Compensator ans = 378 Chapter 9: Design Via Root Locus Gcpd(s) Transfer function: s + 2.955 Zero/pole/gain: (s+0.01) -------s Zero/pole/gain: (s+2.955) (s+0.01) ---------------------s^2 (s+4) (s+6) (s+10) Select a point in the graphics window selected_point = -1.9931 + 4.5383i ans = Closed-loop poles = p= -13.3485 -1.9920 + 4.5377i -1.9920 - 4.5377i -2.6575 -0.0100 Give pole number that is operating point 2 ans = Summary of estimated specifications for selected point ans = on PID compensated root locus operatingpoint = -1.9920 + 4.5377i gain = 295.6542 estimated_settling_time = 2.0081 estimated_peak_time = Solutions to Problems 379 0.6923 estimated_percent_overshoot = 25 estimated_damping_ratio = 0.4037 estimated_natural_frequency = 4.9557 380 Chapter 9: Design Via Root Locus 28. Open-loop poles are at -2, - 0.134, and -1.87. An open-loop zero is at -3. Searching the 121.13o line (ζ = 0.517), find the closed-loop dominant poles at -0.747 + j1.237 with K = 1.58. Searching the real axis segments locates a higher-order pole at -2.51. Since the open-loop zero is a zero of H(s), it is not a closed-loop zero. Thus, there are no closed-loop zeros. 29. a. The damping ratio for 15% overshoot is 0.517. The desired operating point is found from the desired specifications. ζω n = 1.333 1.333 44 = = 2.578 . Thus, = = 1.333 and ω n = Ts 3 ζ 0.517 Solutions to Problems 381 Im = ω n 1 − ζ = 2.578 1 − 0.517 = 2.207 . Hence the design point is –1.333 + j2.207. The 2 2 angular contribution of the system poles and compensator zero at the design point is 100.80. Thus, the compensator zero must contribute 1800 – 100.80 = 79.20. Using the geometry below, jω s-plane j2.207 X -zc 79.2o -1.333 σ 2.207 = tan (79.2o) . Hence, zc = 1.754. The compensated open-loop transfer function with PD zc − 1.333 compensation is K ( s + 1.754) . Evaluating the gain for this function at the point s (s + 2)(s + 4)( s + 6) –1.333 + j2.207 yields K = 47.28 with higher-order poles at –1.617 and –7.718. Following Figure 9.49(c) in the text, 1 = 1.754 . Therefore, K f = 0.5701 . Also, using the notation of Kf Figure 9.49(c), K1 K f = 47.28 , from which K1 = 82.93 . b. Program: K1=82.93; numg=K1; deng=poly([0 -2 -4 -6]); 'G(s)' G=tf(numg,deng); Gzpk=zpk(G) Kf=0.5701 numh=Kf*[1 1.754]; denh=1 'H(s)' H=tf(numh,denh); Hzpk=zpk(H) 'T(s)' T=feedback(G,H); T=minreal(T) step(T) title('Step Response for Feedback Compensated System') Computer response: ans = G(s) Zero/pole/gain: 82.93 ------------------s (s+6) (s+4) (s+2) 382 Chapter 9: Design Via Root Locus Kf = 0.5701 denh = 1 ans = H(s) Zero/pole/gain: 0.5701 (s+1.754) ans = T(s) Transfer function: 82.93 --------------------------------------s^4 + 12 s^3 + 44 s^2 + 95.28 s + 82.93 30. a. σd = ζωn = 4/Ts = 4/1 = 4. 5% overshoot -> ζ = 0.69. Since ζωn = 4, ωn = 5.8. ωd = ωn 1-ζ2 = 4.195. Thus, the design point is –1 + j4.195. The sum of angles from the minor- loop's open-loop poles to the design point is -263.634o. Thus, the minor-loop's open-loop zero must Solutions to Problems 383 4.195 contribute 83.634o to yield 180o at the design point. Hence, z - 4 = tan 83.634o, or zc = a = 4.468 c from the geometry below. jω j4.195 83.634o o 83.634 -zc s-plane σ -4 Adding the zero and calculating the gain at the design point yields K1 = 38.33. Therefore, the minorloop open-loop transfer function is K1G(s)H(s) = 38.33(s+4.468) s(s+4)(s+9) . The equivalent minor-loop K1G(s) 38.33 closed-loop transfer function is Gml(s) = 1+K G(s)H(s) = 3 . A simulation 1 s +13s2+74.33s+171.258 of the step response of the minor loop is shown below. Computer response: Minor-loop Closed-Loop Response 0.25 Amplitude 0.2 0.15 0.1 0.05 0 0 0.5 Time (secs) b. The major-loop open-loop transfer function is G e (s) = 1 1.5 38.33K . s + 13s + 74.33s + 171.258 3 2 Drawing the root locus using Ge(s) and searching along the 10% overshoot line (ζ = 0.591) for 180o yields the point -3.349 + j4.572 with a gain 38.33K = 31.131, or K = 0.812. 384 Chapter 9: Design Via Root Locus c. Program: numg=31.131; deng=[1 13 74.33 171.258]; 'G(s)' G=tf(numg,deng) T=feedback(G,1); step(T) title('Major-loop Closed-Loop Response') Computer response: G(s) Transfer function: 31.13 -----------------------------s^3 + 13 s^2 + 74.33 s + 171.3 d. Adding the PI compensator, Ge(s) = 31.131(s+0.1) . s(s3+13s2+74.33s+171.258) Program: numge=31.131*[1 0.1]; denge=[1 13 74.33 171.258 0]; 'Ge(s)' Ge=tf(numge,denge) T=feedback(Ge,1); t=0:0.1:10; step(T,t) title('Major-loop Closed-Loop Response with PI Compensator') pause step(T) Solutions to Problems 385 title('Major-loop Closed-Loop Response with PI Compensator') Computer response: ans = Ge(s) Transfer function: 31.13 s + 3.113 ---------------------------------s^4 + 13 s^3 + 74.33 s^2 + 171.3 s 386 Chapter 9: Design Via Root Locus 31. R2 a. PI controller: Using Table 9.10, R 1 1 s+R C 2 s+0.01 =s , R2C = 100. Let C = 25 µF. Therefore, s R2 = 4 MΩ. For unity gain, R1 = 4 MΩ. Compensate elsewhere in the loop for the compensator negative sign. 1 b. PD controller: Using Table 9.10, R2C(s+R C ) = s+2. Hence, R1C = 0.5. Let C = 1 µF. 1 Therefore, R1 = 500 KΩ. For unity gain, R2C = 1, or R2 = 1 MΩ. Compensate elsewhere in the loop for the compensator negative sign. 32. 1 R2 C s + 0.1 . Thus, R2C = 10, and a. Lag compensator: See Table 9.11. = 1 s + 0.01 s+ (R1 + R 2 )C s+ (R1 + R2)C = 100. Letting C = 10 µ F, we find R2 = 1 MΩ. Also R1C = 100 - R2C = 90, which yields R1 = 9 MΩ. The loop gain also must be multiplied by R1 + R 2 . R2 1 s+R C 1 s+2 b. Lead compensator: See Table 9.11. 1 1 = s+5 . Thus, R1C = 0.5, and s+R C+R C 1 2 1 1 R1C + R2C = 5. Letting C = 1 µF, R2 = 333 KΩ, and R1 = 500 KΩ. c. Lag-lead compensation: See Table 9.11. 1 1 (s + R C )(s + R C ) 11 22 (s+0.1)(s+1) =2 . Thus, R1C1 = 1, and 1 1 1 1 s + 10.01s + 0.1 s2 + (R C + R C + R C )s + R R C C 11 22 21 1212 1 1 1 1 R2C2 = 10. Also, R C + R C + R C = 1 + 0.1 + R C = 10.01, or R2C1 = 0.112. Letting C1 = 11 22 21 21 10 µF, we find R1 = 10 MΩ , R2 = 1.12 MΩ, and C2 = 8.9 µF. 33. C1 s+0.1 a. Lag compensator: See Table 9.10 and Figure 9.58. s+0.01 = C 2 1 (s+R C ) 11 . Therefore, 1 (s+R C ) 22 R1C1 = 10; R2C2 = 100. Letting C1 = C2 = 20 µF, we find R1 = 500 KΩ and R2 = 5 MΩ. Compensate elsewhere in the loop for the compensator negative sign. Solutions to Design Problems 387 C1 s+2 b. Lead compensator: See Table 9.10 and Figure 9.58. s+5 = C 2 1 (s+R C ) 11 . Therefore, 1 (s+R C ) 22 R1C1 = 0.5 and R2C2 = 0.2. Letting C1 = C2 = 20 µF, we find R1 = 25 KΩ and R2 = 10 MΩ. Compensate elsewhere in the loop for the compensator negative sign. c. Lag-lead compensator: See Table 9.10 and Figure 9.58. For lag portion, use (a). For lead: C1 s+1 s+10 = C2 1 (s+R C ) 11 . Therefore, R1C1 = 1 and R2C2 = 0.1. Letting C1 = C2 = 10 µF, we find 1 (s+R C ) 22 R1 = 100 KΩ and R2 = 10 KΩ. The following circuit can be used to implement the design. SOLUTIONS TO DESIGN PROBLEMS 34. θm(s) a. E (s) = a Kb = Kt RaJ KtKb 1 s(s+ J (D + R )) a Ea 5 4 1 1 = 60000 1 = 0.005; Jeq = 5 (10 x 4 )2 = 0.05; Deq = 1 (10 )2 = 0.01; ω x 60 x 2π 2π θm(s) Ts Kt 0.5 2 Ra = Ea = 5 = 0.1. Therefore, Ea(s) = s(s+0.21) . b. The block diagram of the system is shown below. 388 Chapter 9: Design Via Root Locus Forming an equivalent unity feedback system, 1000 Now, T(s) = 2 . Thus, ωn = 1000 ; 2ζωn = 0.21 + 0.2Kt. Since ζ = 0.5, s + (0.21 + 0.2Kt)s + 1000 Kt = 157.06. 1000 c. Uncompensated: Kt = 0; T(s) = 2 ; ωn = 31.62 rad/s; ζ = 3.32 x 10-3; s + 0.21s + 1000 %OS = e −ζπ / Tp = π ωn 1-ζ2 1−ζ 2 4 x100 = 98.96%; Ts = ζω = 38.09 seconds; n 1000 = 9.93 x 10-2 second; Kv = 0.21 = 4761.9. 1000 Compensated: Kt = 157.06; T(s) = 2 ; ωn = 31.62 rad/s; ζ = 0.5; s + 31.62s + 1000 %OS = e −ζπ / 1−ζ 2 4 x100 = 16.3%; Ts = ζω = 0.253 second; Tp = n 1000 Kv = 31.62 = 31.63. 35. 25 a. T(s) = 2 ; Therefore, ωn = 5; 2ζωn = 1; ζ = 0.1; s + s + 25 %OS = e −ζπ / 1−ζ 2 4 x100 = 73%; Ts = ζω = 8 seconds. n π ωn 1-ζ2 = 0.115 second; Solutions to Design Problems 389 b. From Figure P9.6(b), T(s) = ωn = 25K 1 . Thus, s + (1 + 25K f )s + 25K 1 2 25K1 ; 2ζωn = 1 + 25Kf. For 25% overshoot, ζ = 0.404. For Ts = 0.2 = Therefore 1 + 25Kf = 2ζωn = 40, or Kf = 1.56. Also, ωn = 4 , ζωn = 20. ζωn 20 = 49.5. ζ ωn2 49.52 Hence K1 = 25 = 25 = 98.01. 25 1 c. Uncompensated: G(s) = s(s+1) ; Therefore, Kv = 25, and e(∞) = K v Compensated: G(s) = 1 e(∞) = K v = 0.04. 25K 1 25 x 98.01 ; Therefore, Kv = 1+25 x 1.56 = 61.26, and s(s + 1 + 25K f ) = 0.0163. 36. a. The transfer functions of the subsystems are as follows: Pot: Gp(s) = K1 5π 1 = 2 ; Amplifier: Ga(s) = s+20 ; Motor and load: Since the time to rise to 63% of 10π the final value is 0.5 second, the pole is at -2. Thus, the motor transfer function is of the form, Gm(s) K K 100 = s(s+2) . But, from the problem statement, 2 = 10 , or K = 20. The block diagram of the system is shown below. 390 Chapter 9: Design Via Root Locus Using the equivalent system, search along the 117.126o line (20% overshoot) and find the dominant second-order pole at - 0.89 + j1.74 with K = 10K1 = 77.4. Hence, K1 = 7.74. 77.4 1 b. Kv = 2 x 20 = 1.935. Therefore, e(∞) = K v c. %OS = 20%; ζ = Ts = %OS - ln ( 100 ) %OS π2 + ln2 ( 100 ) 4 = 4.49 seconds; Tp = ζωn ω π 2 n 1-ζ = 0.517. = 0.456; ωn = 0.892 + 1.742 = 1.95 rad/s; = 1.81 seconds. d. The block diagram of the minor loop is shown below. 20 The transfer function of the minor loop is GML(s) = s(s+2+20K ) . Hence, the block diagram of the f equivalent system is where a = 2 + 20Kf. The design point is now found. Since %OS = 20%, ζ = 0.456. Also, since Ts = %OS - ln ( 100 ) = %OS π2 + ln2 ( 100 ) 4 = 2 seconds, ωn = 4.386 rad/s. Hence, the design point is –2 + j3.9. ζωn Solutions to Design Problems 391 Using just the open-loop poles at the origin and at -20, the summation of angles to the design point is -129.37o. The pole at -a must then be contributing 129.37o - 180o = -50.63o. Using the geometry below, a = 5.2, or Kf = 0.16. Adding the pole at -5.2 and using the design point, we find 10K1 = 407.23, or K1 = 40.723. Summarizing the compensated transient characteristics: ζ = 0.456; ωn = 4.386; %OS = 20%; Ts = 4 = 2 seconds; Tp = ζωn ω π 1-ζ2 n 407.23 = 0.81 seconds; Kv = 20 x 5.2 = 3.92. 37. Block diagram K1 20π volts = 2. Preamplifier/Power amplifier: (s+40) ; Pots: 5(2π) rad. Torque-speed curve: T (N-m) 50 v 75 25 50 150 ω (rad/sec) rad rev 1 min rev 1 min rad where 1432.35 min x 60 sec x 2π rev = 150 rad/sec; 477.45 min x 60 sec x 2π rev = 50 rad/sec. 50 The slope of the line is - 100 = - 0.5. Thus, its equation is y = -0.5x + b. Substituting one of the Kt Tstall 100 ea points, find b = 100. Thus Tstall = 100, and ωno load = 200. R = e = 50 = 2; Kb = = ωno load a a 50 200 = 0.25. 392 Chapter 9: Design Via Root Locus θm(s) Motor: E (s) = a Kt/(RaJ) 0.02 KtKb = s(s+0.505) , where J = 100, D = 50. 1 s(s+ J (D+ R )) a Gears: 0.1 Drawing block diagram: θc (s) + 0.02 K 40 s +140 (s + 40 ) 2 s( s + 0. 505) θL (s) 0.1 - 2 θc(s) + 0. 004 K1 θL(s) s( s + 0. 505)(s + 40 ) - b. Compensator design - Lead 10% overshoot and Ts = 1 sec yield a design point of - 4 + j5.458. Sum of angles of uncompensated system poles to this point is -257.491o. If we place the lead compensator zero over the uncompensated system pole at -0.505, the angle at the design point is -134.858o. Thus, the lead compensator pole must contribute 134.858o - 180o = -45.142o. Using the geometry below 5.458 o pc - 4 = tan(45.142 ), or pc = 9.431. X j5.458 45.142o -p c -4 Using the uncompensated poles and the lead compensator, the gain at the design point is 0.004K1 = 1897.125. Solutions to Design Problems 393 Compensator design - Lag zlag 1000 1897.125 With lead compensation, Kv = (40)(9.431) = 5.0295.029. Since we want Kv = 1000, p = = lag 5.029 198.85. Use plag = 0.001. Hence zlag = 0.1988. The lag compensated 1897.125(s+0.1988) Ge(s) = s(s+40)(s+9.431)(s+0.001) . c. Compensator schematic 1 1 lag: R C = 0.1988. Let C = 100 µF. Then R2 = 50.3 kΩ. Now, (R +R )C = 0.001. 2 12 R2 Thus, R1 = 9.95 MΩ. Buffer gain = reciprocal of lag compensator's R + R . Hence buffer 1 2 gain = R1 + R2 = 198.8. R2 1 1 1 lead: R C = 0.505. Let C = 10 µF. Then R1 = 198 kΩ. Now, R C + R C = 9.431. 1 1 2 Thus, R2 = 11.2 k Ω. d. Program: numg= 1897.125*[1 0.1988]; deng=poly([0 -40 -9.431 -.001]); 'G(s)' G=tf(numg,deng); Gzpk=zpk(G) rlocus(G) pos=10 z=-log(pos/100)/sqrt(pi^2+[log(pos/100)]^2) sgrid(z,0) title(['Root Locus with ' , num2str(pos), ' Percent Overshoot Line']) [K,p]=rlocfind(G) %Allows input by selecting point on graphic pause T=feedback(K*G,1); step(T) title(['Step Response for Design of ' , num2str(pos), ' Percent']) 394 Chapter 9: Design Via Root Locus Computer response: ans = G(s) Zero/pole/gain: 1897.125 (s+0.1988) ---------------------------s (s+40) (s+9.431) (s+0.001) pos = 10 z= 0.5912 Select a point in the graphics window selected_point = -3.3649 + 4.8447i K= 0.9090 p= -41.3037 -3.9602 + 4.9225i -3.9602 - 4.9225i -0.2080 Solutions to Design Problems 395 396 Chapter 9: Design Via Root Locus 38. Consider only the minor loop. Searching along the 143.13o line (ζ = 0.8), locate the minor-loop dominant poles at -3.36 ± j2.52 with Kf = 8.53. Searching the real axis segments for Kf = 8.53 locates a higher-order pole at - 0.28. Using the minor-loop poles as the open-loop poles for the entire system, search along the 120o line (ζ = 0.5) and find the dominant second-order poles at -1.39 + j2.41 with K = 27.79. Searching the real axis segment locates a higher-order pole at - 4.2. 39. Consider only the minor loop. Searching along the 143.13o line (ζ = 0.8), locate the minor-loop dominant poles at -7.74 ± j5.8 with Kf = 36.71. Searching the real axis segments for Kf = 36.71 locates a higher-order pole at - 0.535. Using the minor-loop poles at -7.74 ± j5.8 and - 0.535 as the open-loop poles (the open-loop zero at the origin is not a closed-loop zero) for the entire system, search along the 135o line (ζ = 0.707; 4.32% overshoot) and find the dominant second-order poles at - 4.38 + j4 .38 with K = 227.91. Searching the real axis segment locates a higher-order pole at -7.26. Uncompensated system performance: Setting Kf = 0 and searching along the 135o line (4.32% overshoot) yields -2.39 + j2.39 as the point on the root locus with K = 78.05. Searching the real axis segments of the root locus for K = 78.05 locates a higher-order pole at -11.2. The following table compares the predicted uncompensated characteristics with the predicted compensated characteristics. Uncompensated 78.05 G(s) = (s+1)(s+5)(s+10) Dominant poles: -2.39 + j2.39 Compensated 227.91 G(s) = 3 2+101.71s+50 s +16s Dominant poles: - 4.38 + j4 .38 ζ = 0.707 ζ = 0.707 %OS = e −ζπ / ωn = Ts = Tp = 1−ζ 2 x100 = 4.32% 2.392+2.392 = 3.38 rad/s 4 = 1.67 seconds ζωn π ωn 1-ζ2 = 1.31 seconds %OS = e −ζπ / ωn = Ts = Tp = 1−ζ 2 x100 = 4.32% 4.382+4.382 = 6.19 rad/s 4 = 0.91 second ζωn π ωn 1-ζ2 = 0.72 second 78.05 Kp = 1 x 5 x 10 = 1.56 Kp = 227.91 = 4.56 50 Higher-order pole: -11.22 Higher-order pole: -7.26 Second-order approximation OK Higher-order pole not 5x further from imaginary axis than dominant poles. Simulate to be sure of the performance. 40. In Problem 46, Chapter 8 , the dominant poles, - 40 ± j57.25, yielded Ts = 0.1 second and 11.14% overshoot. The unity feedback system consisted of a gain adjusted forward transfer function of Solutions to Design Problems 397 20000K G(s) = s(s+100)(s+500)(s+800) , where K = 102,300. To reduce the settling time by a factor of 2 to 0.05 seconds and keep the percent overshoot the same, we double the coordinates of the dominant poles to –80 ± j114.5. Assume a lead compensator with a zero at -100 that cancels the plant's pole at -100. The summation of angles of the remaining plant poles to the design point is 149.23o. Thus, the angular contribution of the compensator pole must be 149.23o - 180o = 30.77o. Using the 114.5 geometry below, p - 80 = tan 30.77o, or pc = 272.3. c Adding this pole to the poles at the origin, -500, and -800 yields K = 9.92 x 109 at the design point, -80 ± j114.5. Any higher-order poles will have a real part greater than 5 times that of the dominant pair. Thus, the second-order approximation is OK. 41. 0.35K Uncompensated: G(s)H(s) = (s+0.4)(s+0.5)(s+0.163)(s+1.537) . Searching the 133.639o line (%OS = 5%), find the dominant poles at - 0.187 ± j0.196 with gain, 0.35K = 2.88 x 10-2. Hence, the estimated values are: %OS = 5%; Ts = 4 4 = 0.187 = 21.39 seconds; Tp = ζωn ω n π 1-ζ2 π = 0.196 = 16.03 seconds; Kp = 0.575. PD compensated: Design for 8 seconds peak time and 5% overshoot. ζ= %OS - ln ( 100 ) π = 0.69. Since Tp = = 8 seconds and ωn 1-ζ2 = 0.393, %OS ωn 1-ζ2 π2 + ln2 ( 100 ) ωn = 0.5426. Hence, ζωn = 0.374. Thus, the design point is - 0.374 + j0.393. The summation of angles from the system's poles to the design point is -295.34o. Thus, the angular contribution of the controller zero must be 295.34o-180o = 115.34o. Using the geometry below, 398 Chapter 9: Design Via Root Locus 0.393 o o 0.374 - zc = tan (180 - 115.34 ), from which zc = 0.19. Adding this zero to the system's poles and using the design point, - 0.374 + j0.393, the gain, 0.35K = 0.205. PID compensated: Assume the integral controller, Gc(s) = s+0.01 . The total open-loop transfer s 0.35K(s+0.19)(s+0.01) function is GPID(s)H(s) = s(s+0.4)(s+0.5)(s+0.163)(s+1.537) . Check: The PID compensated system yields a very slow rise time due to the lag zero at 0.01. The rise time can be sped up by moving the zero further from the imaginary axis with resultant changes in the transient response. The plots below show the step response with the PI zero at - 0.24. The response compares favorably with a two-pole system step response that yields 5% overshoot and a peak time of 8 seconds as shown below. Solutions to Design Problems 399 42. a. PD compensator design: Pushing the gain, 10, to the right past the summing junction, the system can be represented as an equivalent unity feedback system with G e (s) = 10 6 . 2 (s − 4551)(s + 286) This system is unstable at any gain. For 1% overshoot and Ts = 0.1, the design point is –40 + j27.29. The summation of angles from the poles of Ge(s) to this point is -216.903o. Therefore, the compensator zero must contribute 216.903o - 180o = 36.903o. Using the following geometry: X j27.29 36.903 o -zc -40 27.29 zc - 40 = tan (36.903). Thus, zc = 76.34. Adding this zero to the poles of Ge(s), the gain at the design point is 106K = 23377. The PD compensated response is shown below. 400 Chapter 9: Design Via Root Locus Step Response for zero at -76.34 and 1% overshoot 4 3.5 3 Amplitude 2.5 2 1.5 1 0.5 0 0 0.02 0.04 0.06 0.08 Time (secs) 0.1 0.12 0.14 b. PI compensator design: To reduce the steady-state error to zero, we add a PI controller of the s+1 form s . The PID compensated step response is shown below. PID Compensated Step Response for zero at -76.34 and 1% overshoot 4 3.5 3 Amplitude 2.5 2 1.5 1 0.5 0 0 0.2 0.4 0.6 0.8 1 1.2 Time (secs) 1.4 1.6 1.8 2 Solutions to Design Problems 401 We can see the 1% overshoot at about 0.1 second as in the PD compensated system above. But the system now corrects to zero error. 43. a. Root locus sketch yields; 25 20 15 10 Imag Axis 5 0 -5 -10 -15 -20 -25 -50 -40 -30 -20 -10 0 Real Axis Root locus sketch near imaginary axis yields; Close-Up Root Locus to Determine Stability 0.25 0.2 0.15 0.1 Imag Axis 0.05 0 -0.05 -0.1 -0.15 -0.2 -0.25 -0.25 -0.2 -0.15 -0.1 -0.05 0 Real Axis 0.05 0.1 0.15 0.2 0.25 402 Chapter 9: Design Via Root Locus Searching imaginary axis for 180o yields: j0.083 at a gain of 0.072K = 0.0528 and j0.188 at a gain of 0.072K = 0.081. Also, the gain at the origin is 0.0517. Thus, the system is stable for 0.0517 < 0.072K < 0.0528; 0.072K > 0.081. Equivalently, for 0.7181 < K < 0.7333; 0.072K > 1.125. b. See (a) c. Uncompensated system: Searching the 20% overshoot line, we find the operating point at -8.987 + j17.4542 = 19.71∠117.126o at 0.072K = 16.94 for the uncompensated system. Simulating the response at this gain yields, Step Response for Uncompensated System 1.4 1.2 Amplitude 1 0.8 0.6 0.4 0.2 0 0 0.05 0.1 0.15 0.2 0.25 0.3 Time (secs) 0.35 0.4 0.45 0.5 For 20% overshoot and Ts = 0.05 s, a design point of –80 + j156.159 is required. The sum of angles to the design point is -123.897o. To meet the requirements at the design point, a zero would have to contribute +303.897o, which is too high for a single zero. Let us first add the pole at the origin to drive the steady-state error to zero to reduce the angle required from the zero. Summing angles with this pole at the origin yields -241.023. Thus a zero contributing 61.023o is required. Using the 156.159 geometry below with z - 80 = tan (61.023), zc = 166.478. c X 61.023 -zc o -80 The gain at the design point is 0.072K = 181.55. j156.159 Solutions to Design Problems 403 d. Step Response for Ccompensated System 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 Time (secs) The settling time requirement has been met, but the percent overshoot has not. Repeating the design for 1% overshoot and a Ts = 0.05 s yields a design point of –80 + j54.575. The compensator zero is found to be at -47.855 at a gain 0.072K = 180.107. Step Response for Redesigned System 1.4 1.2 1 Amplitude 0.8 0.6 0.4 0.2 0 0 0.01 0.02 0.03 0.04 0.05 Time (secs) 0.06 0.07 0.08 0.09 0.1 404 Chapter 9: Design Via Root Locus 44. 4 ζωn = T = 2.667; ζ = s %OS - ln ( 100 ) = 0.591. Thus, ωn = 4.512 rad/s. 2 + ln2 (%OS) π 100 Im = ω n 1 − ζ = 4.512 1 − 0.591 = 3.64 . Thus, and the operating point is -2.667 ± j3.64. 2 2 Summation of angles, assuming the compensating zero is at –5 (to cancel the open-loop pole at –5, is –170.88o. Therefore, the compensator pole must contribute 180o – 170.88o = -9.12o. Using the geometry shown below, 3.64 9.12o -2.667 3.64 = tan 9.12o. Thus, pc = 25.34. Adding the compensator pole and using -2.667 ± j3.64 as pc − 2.667 the test point, 50K = 2504, or K = 50.08. Thus the compensated open-loop transfer function is G e (s) = 2504(s + 5) . Higher-order pole are at –25.12, -5, and-4.898. The s(s + 5)(s + 10s + 50)(s + 25.34) 2 pole at –5 is cancelled by the closed-loop zero at -5. The pole at –4.898 is not far enough away from the dominant second-order pair. Thus, the system should be simulated to determine if the response meets the requirements. Program: syms s numg=2504; deng=expand(s*(s^2+10*s+50)*(s+25.34)); deng=sym2poly(deng); G=tf(numg,deng); Gzpk=zpk(G) T=feedback(G,1); step(T) Computer response: Zero/pole/gain: 2504 ---------------------------s (s+25.34) (s^2 + 10s + 50) Solutions to Design Problems 405 45. a. From Chapter 8, 0.6488K (s+53.85) Ge(s) = 2 ______ 2 (s + 8.119s + 376.3) (s + 15.47s + 9283) Cascading the notch filter, 0.6488K (s+53.85)(s2 + 16s + 9200) Get(s) = (s2 + 8.119s + 376.3) (s2 + 15.47s + 9283)(s+60)2 Arbitrarily design for %OS = 30% (ζ = 0.358) and Ts = 0.3 s. This places desired poles at –13.33 ± j34.79. At the design point, the sum of the angles without the PD controller is 107.190. Thus, 34.79 = tan 72.81 zc − 13.33 406 Chapter 9: Design Via Root Locus j34.79 72.810 -13.33 -zc From which, zc = 24.09. Putting this into the forward path, 0.6488K (s+53.85)(s2 + 16s + 9200)(s+24.09) Get(s) = (s2 + 8.119s + 376.3) (s^2 + 15.47s + 9283)(s+60)2 Using root locus, the gain 0.6488K = 1637, or K = 2523. b. Add a PI controller G PI (s ) = Thus, ( s + 0.1) s 0.6488K (s+53.85)(s2 + 16s + 9200)(s+24.09)(s+0.1) Get(s) = s (s2 + 8.119s + 376.3) (s^2 + 15.47s + 9283)(s+60)2 Using root locus, the gain 0.6488K = 1740, or K = 2682. c. Program: syms s numg=1637*(s+53.85)*(s^2+16*s+9200)*(s+24.09)*(s+0.1); deng=s*(s^2+15.47*s+9283)*(s^2+8.119*s+376.3)*(s+60)^2; numg=sym2poly(numg); deng=sym2poly(deng); G=tf(numg,deng); Gzpk=zpk(G) T=feedback(G,1); step(T,0:0.01:1) title(['With PD, Notch, and PI']) pause step(T) title(['With PD, Notch, and PI']) Computer response: Zero/pole/gain: 1637 (s+53.85) (s+24.09) (s+0.1) (s^2 + 16s + 9200) ------------------------------------------------------s (s+60)^2 (s^2 + 8.119s + 376.3) (s^2 + 15.47s + 9283) _____ Solutions to Design Problems 407 TEN Frequency Response Techniques SOLUTION TO CASE STUDY CHALLENGE Antenna Control: Stability Design and Transient Performance First find the forward transfer function, G(s). Pot: K1 = 10 = 3.18 π Preamp: K Power amp: 100 G1(s) = s(s+100) Motor and load: Kt 1 1 1 J = 0.05 + 5 (5 )2 = 0.25 ; D = 0.01 + 3 (5 )2 = 0.13; Ra = 5 ; Kb = 1. Therefore, θm(s) Gm(s) = E (s) = a Kt RaJ KtKb 1 s(s+ J (D + R )) a 0.8 = s(s+1.32) . Gears: 50 1 K2 = 250 = 5 Therefore, 50.88K G(s) = K1KG1(s)Gm(s)K2 = s(s+1.32)(s+100) Plotting the Bode plots for K = 1, Solution to Case Study Challenge 409 a. Phase is 180o at ω = 11.5 rad/s. At this frequency the gain is - 48.41 dB, or K = 263.36. Therefore, for stability, 0 < K < 263.36. b. If K = 3, the magnitude curve will be 9.54 dB higher and go through zero dB at ω = 0.94 rad/s. At this frequency, the phase response is -125.99o. Thus, the phase margin is 180o - 125.99o = 54.01o. Using Eq. (10.73), ζ = 0.528. Eq. (4.38) yields %OS = 14.18%. c. Program: numga=50.88; denga=poly([0 -1.32 -100]); 'Ga(s)' Ga=tf(numga,denga); Gazpk=zpk(Ga) '(a)' bode(Ga) title('Bode Plot at Gain of 50.88') pause [Gm,Pm,Wcp,Wcg]=margin(Ga); 'Gain for Stability' Gm pause '(b)' numgb=50.88*3; dengb=denga; 'Gb(s)' Gb=tf(numgb,dengb); Gbzpk=zpk(Gb) bode(Gb) title('Bode Plot at Gain of 3*50.88') [Gm,Pm,Wcp,Wcg]=margin(Gb); 'Phase Margin' Pm for z=0:.01:1 Pme=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi); if Pm-Pme<=0; break end end z percent=exp(-z*pi/sqrt(1-z^2))*100 Computer response: ans = Ga(s) 410 Chapter 10: Frequency Response Methods Zero/pole/gain: 50.88 -----------------s (s+100) (s+1.32) ans = (a) ans = Gain for Stability Gm = 262.8585 ans = (b) ans = Gb(s) Zero/pole/gain: 152.64 -----------------s (s+100) (s+1.32) ans = Phase Margin Pm = 53.9644 z= 0.5300 percent = 14.0366 Answers to Review Questions 411 ANSWERS TO REVIEW QUESTIONS 1. a. Transfer functions can be modeled easily from physical data; b. Steady-state error requirements can be considered easily along with the design for transient response; c. Settles ambiguities when sketching root locus; (d) Valuable tool for analysis and design of nonlinear systems. 2. A sinusoidal input is applied to a system. The sinusoidal output's magnitude and phase angle is measured in the steady-state. The ratio of the output magnitude divided by the input magnitude is the magnitude response at the applied frequency. The difference between the output phase angle and the input phase angle is 412 Chapter 10: Frequency Response Methods the phase response at the applied frequency. If the magnitude and phase response are plotted over a range of different frequencies, the result would be the frequency response for the system. 3. Separate magnitude and phase curves; polar plot 4. If the transfer function of the system is G(s), let s=jω. The resulting complex number's magnitude is the magnitude response, while the resulting complex number's angle is the phase response. 5. Bode plots are asymptotic approximations to the frequency response displayed as separate magnitude and phase plots, where the magnitude and frequency are plotted in dB. 6. Negative 6 dB/octave which is the same as 20 dB/decade 7. Negative 24 dB/octave or 80 dB/decade 8. Negative 12 dB/octave or 40 dB/decade 9. Zero degrees until 0.2; a negative slope of 45o/decade from a frequency of 0.2 until 20; a constant -90o phase from a frequency of 20 until ∞ 10. Second-order systems require a correction near the natural frequency due to the peaking of the curve for different values of damping ratio. Without the correction the accuracy is in question. 11. Each pole yields a maximum difference of 3.01 dB at the break frequency. Thus for a pole of multiplicity three, the difference would be 3x3.01 or 9.03 dB at the break frequency, - 4. 12. Z = P - N, where Z = # of closed-loop poles in the right-half plane, P = # of open-loop poles in the righthalf plane, and N = # of counter-clockwise encirclements of -1 made by the mapping. 13. Whether a system is stable or not since the Nyquist criterion tells us how many rhp the system has 14. A Nyquist diagram, typically, is a mapping, through a function, of a semicircle that encloses the right half plane. 15. Part of the Nyquist diagram is a polar frequency response plot since the mapping includes the positive jω axis. 16. The contour must bypass them with a small semicircle. 17. We need only map the positive imaginary axis and then determine that the gain is less than unity when the phase angle is 180o. 18. We need only map the positive imaginary axis and then determine that the gain is greater than unity when the phase angle is 180o. 19. The amount of additional open-loop gain, expressed in dB and measured at 180o of phase shift, required to make a closed-loop system unstable. 20. The phase margin is the amount of additional open-loop phase shift, ΦM, required at unity gain to make the closed-loop system unstable. 21. Transient response can be obtained from (1) the closed-loop frequency response peak, (2) phase margin 22. a. Find T(jω)=G(jω)/[1+G(jω)H(jω)] and plot in polar form or separate magnitude and phase plots. b. Superimpose G(jω)H(jω) over the M and N circles and plot. c. Superimpose G(jω)H(jω) over the Nichols chart and plot. Solutions to Problems 413 23. For Type zero: Kp = low frequency gain; For Type 1: Kv = frequency value at the intersection of the initial slope with the frequency axis; For Type 2: Ka = square root of the frequency value at the intersection of the initial slope with the frequency axis. 24. No change at all 25. A straight line of negative slope, ωT, where T is the time delay 26. When the magnitude response is flat and the phase response is flat at 0o. SOLUTIONS TO PROBLEMS 1. a. ; ; b. ; ; c. ; ; 2. a. 414 Chapter 10: Frequency Response Methods b. c. 3. a. 90° 120° 60° 150° 30° X X X0 X X X 180° 0° 0.5 1 210° 1.5 330° X 240° 300° 270° Solutions to Problems 415 b. 90° 120° 60° 150° 30° X X 0 X 0.2 X X X XX X 180° 0.4 0° X X X X X X X X 0.6 X X 210° 330° 240° 300° 270° c. 90° 120° 60° 150° 30° X X X X 0 X X X X 180° 0° 5 10 15 210° 330° 240° X 270° 4. a. 20 300° 416 Chapter 10: Frequency Response Methods b. c. 30 -90 20 -20 dB/dec -100 0 -40 dB/dec -20 dB/dec -40 dB/dec -10 -20 .1 1 5. a. System 1 v -45 deg/dec -110 +45 deg/dec -120 -130 -140 -20 dB/dec -30 Phase dB 10 10 100 -150 .1 1 v 10 100 Solutions to Problems 417 b. System 2 c. System 3 d. 418 Chapter 10: Frequency Response Methods 6. 0.5 0.4 0.3 0.2 Imag Axis 0.1 0 -0.1 -0.2 -0.3 -0.4 -0.5 -0.1 0 0.1 0.2 0.3 0.4 0.5 Real Axis 7. Gain dB 0 -50 -100 -150 -1 10 10 0 10 1 10 2 10 1 10 2 Frequency (rad/sec) Phase deg 0 -90 -180 -270 10 -1 10 0 Frequency (rad/sec) 8. Program: numg=[1 5]; deng=conv([1 6 100],[1 4 25]); G=tf(numg,deng); 'G(s)' Gzpk=zpk(G) nyquist(G) axis([-3e-3,4e-3,-5e-3,5e-3]) w=0:0.1:100; [re,im]=nyquist(G,w); for i=1:1:length(w) 0.6 Solutions to Problems 419 M(i)=abs(re(i)+j*im(i)); A(i)=atan2(im(i),re(i))*(180/pi); if 180-abs(A(i))<=1; re(i); im(i); K=1/abs(re(i)); fprintf('\nw = %g',w(i)) fprintf(', Re = %g',re(i)) fprintf(', Im = %g',im(i)) fprintf(', M = %g',M(i)) fprintf(', Angle = %g',A(i)) fprintf(', K = %g',K) Gm=20*log10(1/M(i)); fprintf(', Gm = %g',Gm) break end end Computer response: ans = G(s) Zero/pole/gain: (s+5) ---------------------------------(s^2 + 4s + 25) (s^2 + 6s + 100) w = 10.1, Re = -0.00213722, Im = 2.07242e-005, M = 0.00213732, Angle = 179.444, K = 467.898, Gm = 53.4026 ans = G(s) Zero/pole/gain: (s+5) ---------------------------------(s^2 + 4s + 25) (s^2 + 6s + 100) w = 10.1, Re = -0.00213722, Im = 2.07242e-005, M = 0.00213732, Angle = 179.444, K = 467.898, Gm = 53.4026 420 Chapter 10: Frequency Response Methods 9. a. Since the real-axis crossing is at -0.3086, P = 0, N = 0. Therefore Z = P - N = 0. System is stable. Derivation of real-axis crossing: ( ) 50 ⎡ −9ω 2 − jω 18 − ω 2 ⎤ 50 ⎦. G ( jω ) = =⎣ 4 3 s ( s + 3)( s + 6) s = jω 81ω + 18ω − ω ( Thus, the imaginary part = 0 at ω = 18 . ) Substituting this frequency into evaluated to be -0.3086. b. P = 0, N = -2. Therefore Z = P - N = 2. System is unstable. c. P = 0, N = 0. Therefore Z = P - N = 0. System is stable d. P = 0, N = -2. Therefore Z = P - N = 2. System is unstable. 10. System 1: For K = 1, G ( jω ) , the real part is Solutions to Problems 421 The Nyquist diagram intersects the real axis at -0.0021. Thus K can be increased to 478.63 before there are encirclements of -1. There are no poles encircles by the contour. Thus P = 0. Hence, Z = P N, Z = 0 + 0 if K <478.63; Z = 0 –(-2) if K > 478.63. Therefore stability if 0 < K < 478.63. System 2: For K = 1, The Nyquist diagram intersects the real axis at -0.720. Thus K can be increased to 1.39 before there are encirclements of -1. There are no poles encircles by the contour. Thus P = 0. Hence, Z = P - N, Z = 0 + 0 if K <1.39; Z = 0 – (-2) if K > 1.39. Therefore stability if 0 < K < 1.39. System 3: For K = 1, 422 Chapter 10: Frequency Response Methods Stable if 0<K<1. 11. Note: All results for this problem are based upon a non-asymptotic frequency response. System 1: Plotting Bode plots for K = 1 yields the following Bode plot, K = 1000: For K = 1, phase response is 180o at ω = 6.63 rad/s. Magnitude response is -53.6 dB at this frequency. For K = 1000, magnitude curve is raised by 60 dB yielding + 6.4 dB at 6.63 rad/s. Thus, the gain margin is - 6.4 dB. Solutions to Problems 423 Phase margin: Raising the magnitude curve by 60 dB yields 0 dB at 9.07 rad/s, where the phase curve is 200.3o. Hence, the phase margin is 180o-200.3o = - 20.3o. K = 100: For K = 1, phase response is 180o at ω = 6.63 rad/s. Magnitude response is -53.6 dB at this frequency. For K = 100, magnitude curve is raised by 40 dB yielding – 13.6 dB at 6.63 rad/s. Thus, the gain margin is 13.6 dB. Phase margin: Raising the magnitude curve by 40 dB yields 0 dB at 2.54 rad/s, where the phase curve is 107.3o. Hence, the phase margin is 180o-107.3o = 72.7o. K = 0.1: For K = 1, phase response is 180o at ω = 6.63 rad/s. Magnitude response is -53.6 dB at this frequency. For K = 0.1, magnitude curve is lowered by 20 dB yielding – 73.6 dB at 6.63 rad/s. Thus, the gain margin is 73.6 dB.. System 2: Plotting Bode plots for K = 1 yields K = 1000: For K = 1, phase response is 180o at ω = 1.56 rad/s. Magnitude response is -2.85 dB at this frequency. For K = 1000, magnitude curve is raised by 60 dB yielding + 57.15 dB at 1.56 rad/s. Thus, the gain 424 Chapter 10: Frequency Response Methods margin is – 57.15 dB. Phase margin: Raising the magnitude curve by 54 dB yields 0 dB at 500 rad/s, where the phase curve is -91.03o. Hence, the phase margin is 180o-91.03o = 88.97o. K = 100: For K = 1, phase response is 180o at ω = 1.56 rad/s. Magnitude response is -2.85 dB at this frequency. For K = 100, magnitude curve is raised by 40 dB yielding + 37.15 dB at 1.56 rad/s. Thus, the gain margin is – 37.15 dB. Phase margin: Raising the magnitude curve by 40 dB yields 0 dB at 99.8 rad/s, where the phase curve is -84.3o. Hence, the phase margin is 180o-84.3o = 95.7o. K = 0.1: For K = 1, phase response is 180o at ω = 1.56 rad/s. Magnitude response is -2.85 dB at this frequency. For K = 0.1, magnitude curve is lowered by 20 dB yielding – 22.85 dB at 1.56 rad/s. Thus, the gain margin is – 22.85 dB. Phase margin: Lowering the magnitude curve by 20 dB yields 0 dB at 0.162 rad/s, where the phase curve is -99.8o. Hence, the phase margin is 180o-99.86o = 80.2o. System 3: Plotting Bode plots for K = 1 yields Solutions to Problems 425 K = 1000: For K = 1, phase response is 180o at ω = 1.41 rad/s. Magnitude response is 0 dB at this frequency. For K = 1000, magnitude curve is raised by 60 dB yielding 60 dB at 1.41 rad/s. Thus, the gain margin is - 60 dB. Phase margin: Raising the magnitude curve by 60 dB yields no frequency where the magnitude curve is 0 dB. Hence, the phase margin is infinite. K = 100: For K = 1, phase response is 180o at ω = 1.41 rad/s. Magnitude response is 0 dB at this frequency. For K = 100, magnitude curve is raised by 40 dB yielding 40 dB at 1.41 rad/s. Thus, the gain margin is - 40 dB. Phase margin: Raising the magnitude curve by 40 dB yields no frequency where the magnitude curve is 0 dB. Hence, the phase margin is infinite. K = 0.1: For K = 1, phase response is 180o at ω = 1.41 rad/s. Magnitude response is 0 dB at this frequency. For K = 0.1, magnitude curve is lowered by 20 dB yielding -20 dB at 1.41 rad/s. Thus, the gain margin is 20 dB. Phase margin: Lowering the magnitude curve by 20 dB yields no frequency where the magnitude curve is 0 dB. Hence, the phase margin is infinite. 426 Chapter 10: Frequency Response Methods 12. Program: %Enter G(s)************************ numg=1; deng=poly([0 -3 -12]); 'G(s)' G=tf(numg,deng) w=0.01:0.1:100; %Enter K ************************** K=input('Type gain, K '); bode(K*G,w) pause [M,P]=bode(K*G,w); %Calculate Gain Margin************** for i=1:1:length(P); if P(i)<=-180; fprintf('\nGain K = %g',K) fprintf(', Frequency(180 deg) = %g',w(i)) fprintf(', Magnitude = %g',M(i)) fprintf(', Magnitude (dB) = %g',20*log10(M(i))) fprintf(', Phase = %g',P(i)) Gm=20*log10(1/M(i)); fprintf(', Gain Margin (dB) = %g',Gm) break end end %Calculate Phase Margin************** for i=1:1:length(M); if M(i)<=1; fprintf('\nGain K = %g',K) fprintf(', Frequency (0 dB) = %g',w(i)) fprintf(', Magnitude = %g',M(i)) fprintf(', Magnitude (dB) = %g',20*log10(M(i))) fprintf(', Phase = %g',P(i)) Pm=180+P(i);; fprintf(', Phase Margin = %g',Pm) break end end 'Alternate program using MATLAB margin function:' clear clf %Bode Plot and Find Points %Enter G(s)************************ numg=1; deng=poly([0 -3 -12]); 'G(s)' G=tf(numg,deng) w=0.01:0.1:100; %Enter K ************************** K=input('Type gain, K '); bode(K*G,w) [Gm,Pm,Wcp,Wcg]=margin(K*G) 'Gm(dB)' 20*log10(Gm) Computer response: Solutions to Problems 427 ans = G(s) Transfer function: 1 ------------------s^3 + 15 s^2 + 36 s Type gain, K 40 Gain K = 40, Frequency(180 deg) = 6.01, Magnitude = 0.0738277, Magnitude (dB) = -22.6356, Phase = -180.076, Gain Margin (dB) = 22.6356 Gain K = 40, Frequency (0 dB) = 1.11, Magnitude = 0.93481, Magnitude (dB) = -0.585534, Phase = -115.589, Phase Margin = 64.4107 Alternate program using MATLAB margin function: ans = G(s) Transfer function: 1 ------------------s^3 + 15 s^2 + 36 s Type gain, K 40 Gm = 428 Chapter 10: Frequency Response Methods 13.5000 Pm = 65.8119 Wcp = 6 Wcg = 1.0453 ans = Gm(dB) ans = 22.6067 13. Program: numg=10000; deng=poly([-5 -18 -30]); G=tf(numg,deng) Ltiview Solutions to Problems 429 Computer response: ans = 10.4N Transfer function: 10000 --------------------------s^3 + 53 s^2 + 780 s + 2700 430 Chapter 10: Frequency Response Methods 14. 2 ⎛1⎞ Squaring Eq. (10.51) and setting it equal to yields ⎝ 2⎠ ωn4 1 =2 (ωn2 - ω2)2 + 4ζ2ωn2ω2 Simplifying, ω 4 + 2ωn2(2ζ2 - 1)ω2 - ωn4 = 0 Solving for ω2 using the quadratic formula and simplifying yields, ω2 = ωn2 [ - (2ζ2 - 1) ± 4ζ 4 - 4ζ 2 + 2 ] Taking the square root and selecting the positive term, ω = ωn 15. (1 - 2ζ2) + a. Using Eq. (10.55), ωBW = 10.06 rad/s. b. Using Eq. (10.56), ωBW = 1.613 rad/s. 4ζ 4 - 4ζ 2 + 2 Solutions to Problems 431 c. First find ζ. Since Ts = π Tp Tp ζπ ,T = . Solving for ζ with T = 2 s s 4 1 - ζ2 n 1-ζ 4 and Tp = ζωn ω 0.5 yields ζ = 0.537. Using either Eq. (10.55) or (10.56) yields ωBW = 2.29 rad/s. d. Using ζ = 0.3, ω n Tr = 1.76ζ − 0.417ζ + 1.039ζ + 1 = 1.3217 . Hence, 3 ωn = 2 1.3217 1.3217 = = 0.3304 rad/s. Using Eq. (10.54) yields ωBW = 0.4803 rad/s. Tr 4 16. Im 3 G-plane v=1 o F = 20 2 M = 1.3 M = 1.0 1.4 1.5 1 1.6 1.8 70 40 50 o o 9.0 v=7 -70 v=3 v=5 0.4 Re v=0.07 v=0.4 o -40 v=2 v=4 -2 (b) v=4 o -50 v=0.9 0.5 v=5 v=4 v=6 -3 M = 0.7 0.6 v=3 v=2 v=10 (a) o 2.0 3.0 5.0 -1 30 o 25 o o o -30 o -25 v=0.9 -20 o v=3 -4 (c) -3 -2 -1 1 2 432 Chapter 10: Frequency Response Methods a. b. c. Solutions to Problems 433 17. (c) (a) v=1 v=0.3 o -355 2 0.8 1.0 o -340 o -10 o -20 4 o v=2 -40 o -320 o 3 1.4 7 4 o -300 3 6 o o 10 5 o o o o o -200 -270 -260 -240 -220 (b) 3 2 -280 -5 0.5 -180 o-160 o -60 o -140 o -120 -100 o o -90 o -80 18. a. The polar plot is approximately tangent to M = 5. Using Figure 10.40, the student would estimate 72% overshoot. However, notice that the polar plot intersects the negative real axis at a magnitude greater than unity. Hence, the system is actually unstable and the estimated percent overshoot is not correct. b. The polar plot is approximately tangent to M = 3. Using Figure 10.40, we estimate 58% overshoot. c. The polar plot is approximately tangent to M = 2.5. Using Figure 10.40, we estimate 52% overshoot. 19. Raise each curve in Problem 17 by (a) 9.54 dB, (b) 7.96 dB, and (c) 3.52 dB, respectively. 434 Chapter 10: Frequency Response Methods (a) v=0.3 (c) 0.5 v=1 0.8 -355 o 1.0 -5 o 2 (b) v=2 o -340 o -10 3 1.4 o -20 5 -320 4 o 7 5 3 6 10 o o -60 -300 -280 o o o o o -200 -270 -260 -240 -220 o -40 o -180 o-160 o o -140 o -120 -100 o o -90 o -80 Systems (a) and (b) are both unstable since the open-loop magnitude is greater than unity when the open-loop phase is 180o. System (c) is tangent to approximately M = 3. Using Figure 10.40, we estimate 58% overshoot. 20. Program: %Enter G(s)*********************** numg=[1 5]; deng=[1 4 25 0]; 'G(s)' G=tf(numg,deng) %Enter K ************************* K=input('Type gain, K '); 'T(s)' T=feedback(K*G,1) bode(T) title('Closed-loop Frequency Response') [M,P,w]=bode(T); [Mp i]=max(M); Mp MpdB=20*log10(Mp) wp=w(i) for i=1:1:length(M); if M(i)<=0.707; fprintf('Bandwidth = %g',w(i)) break end end Solutions to Problems 435 Computer response: ans = G(s) Transfer function: s+5 -----------------s^3 + 4 s^2 + 25 s Type gain, K 40 ans = T(s) Transfer function: 40 s + 200 -----------------------s^3 + 4 s^2 + 65 s + 200 Mp = 6.9745 MpdB = 16.8702 wp = 7.8822 Bandwidth = 11.4655 436 Chapter 10: Frequency Response Methods 21. Program: numg=[7 35]; deng=[1 4 10 0]; G=tf(numg,deng) bode(G) %Make a Bode plot. title('Open-Loop Frequency Response') %Add a title to the Bode plot. [Gm,Pm,Wcp,Wcg]=margin(G); %Find margins and margin %frequencies. 'Gain margin(dB); Phase margin(deg.); 0 dB freq. (r/s);' '180 deg. freq. (r/s)' %Display label. margins=[20*log10(Gm),Pm,Wcg,Wcp] %Display margin data. Ltiview Computer response: Transfer function: 7 s + 35 -----------------s^3 + 4 s^2 + 10 s ans = Gain margin(dB); Phase margin(deg.); 0 dB freq. (r/s); ans = 180 deg. freq. (r/s) Solutions to Problems 437 margins = 15.1403 31.0397 3.2252 7.0715 22. Program: %Enter G(s)************************************ numg=7*[1 5]; deng=[1 4 10 0]; 'Open-Loop System' 'G(s)' G=tf(numg,deng) clf w=.10:1:10; nichols(G,w) ngrid title('Nichols Plot') [M,P]=nichols(G,w); for i=1:1:length(M); if M(i)<=0.45; BW=w(i); break end end pause MpdB=input('Enter Mp in dB from Nichols Plot '); Mp=10^(MpdB/20); z2=roots([4,-4,(1/Mp^2)]);%Since Mp=1/sqrt(4z^2(1-z^2)) z1=sqrt(z2); z=min(z1); Pos=exp(-z*pi/(sqrt(1-z^2))); 438 Chapter 10: Frequency Response Methods Ts=(4/(BW*z))*sqrt((1-z^2)+sqrt(4*z^4-4*z^2+2)); Tp=(pi/(BW*sqrt(1-z^2)))*sqrt((1-z^2)+sqrt(4*z^4-4*z^2+2)); 'Closed-Loop System' 'T(s)' T=feedback(G,1) bode(T) title('Closed-Loop Frequency Resposne Plots') fprintf('\nDamping Ratio = %g',z) fprintf(', Percent Overshoot = %g',Pos*100) fprintf(', Bandwidth = %g',BW) fprintf(', Mp (dB) = %g',MpdB) fprintf(', Mp = %g',Mp) fprintf(', Settling Time = %g',Ts) fprintf(', Peak Time = %g',Tp) pause step(T) title('Closed-Loop Step Response') Computer response: ans = Open-Loop System ans = G(s) Transfer function: 7 s + 35 -----------------s^3 + 4 s^2 + 10 s Enter Mp in dB from Nichols Plot ans = Closed-Loop System ans = T(s) Transfer function: 7 s + 35 ----------------------s^3 + 4 s^2 + 17 s + 35 6 Solutions to Problems 439 Damping Ratio = 0.259481, Percent Overshoot = 42.9946, Bandwidth = 5.1, Mp 0.957852 440 Chapter 10: Frequency Response Methods Solutions to Problems 441 23. System 1: Using non-asymptotic frequency response plots, the zero dB crossing is at 9.7 rad/s at a phase of –163.2o. Therefore the phase margin is 180o – 163.2o = 16.8o. |G(jω)| is down 7 dB at 14.75 rad/s. Therefore the bandwidth is 14.75 rad/s. Using Eq. (10.73), ζ = 0.15. Using Eq. (4.38), %OS = 62.09%. Eq. (10.55) yields Ts = 2.76 s, and Eq. (10.56) yields Tp = 0.329 s. System 2: Using non-asymptotic frequency response plots, the zero dB crossing is at 6.44 rad/s at a phase of -150.73o. Therefore the phase margin is 180o - 150.73o = 29.27o. |G(jω)| is down 7 dB at 10.1 rad/s. Therefore the bandwidth is 10.1 rad/s. Using Eq. (10.73), ζ = 0.262. Using Eq. (4.38), %OS = 42.62%. Eq. (10.55) yields Ts = 2.23 s, and Eq. (10.56) yields Tp = 0.476 s. 24. a. Gain dB 50 0 -50 -1 10 10 1 10 2 10 10 0 1 10 Frequency (rad/sec) Phase deg -60 -90 -120 -150 -180 10 -1 10 0 2 Frequency (rad/sec) b. Zero dB frequency = 7.8023; Looking at the phase diagram at this frequency, the phase margin is 8.777 degrees. Using Eq. (10.73) or Figure 10.48, ζ = 0.08. Thus, %OS = 77.7. c. Step Response for K = 40 1.6 1.4 1.2 Amplitude 1 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 6 Time (secs) 7 8 9 10 442 Chapter 10: Frequency Response Methods 25. From the Bode plots: Gain margin = 14.96 dB; phase margin = 49.570; 0 dB frequency = 2.152 rad/s; 1800 frequency = 6.325 rad/s; bandwidth(@-7 dB point) = 3.8 rad/s. From Eq. (10.73) ζ = 0.48; from Eq. (4.38) %OS = 17.93; from Eq. (10.55) Ts = 2.84 s; from Eq. (10.56) Tp = 1.22 s. 26. Program: G=zpk([-2],[0 -1 -4],100) %G=zpk([-3 -5],[0 -2 -4 -6],50) G=tf(G) bode(G) title('System 1') %title('System 2') pause %Find Phase Margin [Gm,Pm,Wcg,Wcp]=margin(G); w=1:.01:20; [M,P,w]=bode(G,w); %Find Bandwidth for k=1:1:length(M); if 20*log10(M(k))+7<=0; 'Mag' 20*log10(M(k)) 'BW' wBW=w(k) break end end %Find Damping Ratio,Percent Overshoot, Settling Time, and Peak Time for z= 0:.01:10 Pt=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi); if (Pm-Pt)<=0 z; Po=exp(-z*pi/sqrt(1-z^2)); Ts=(4/(wBW*z))*sqrt((1-2*z^2)+sqrt(4*z^4-4*z^2+2)); Tp=(pi/(wBW*sqrt(1-z^2)))*sqrt((1-2*z^2)+sqrt(4*z^4-4*z^2+2)); fprintf('Bandwidth = %g ',wBW) fprintf('Phase Margin = %g',Pm) fprintf(', Damping Ratio = %g',z) fprintf(', Percent Overshoot = %g',Po*100) fprintf(',Settling Time = %g',Ts) fprintf(', Peak Time = %g',Tp) break end end T=feedback(G,1); step(T) title('Step Response System 1') %title('Step Response System 2') Computer response: Zero/pole/gain: 100 (s+2) ------------s (s+1) (s+4) Transfer function: 100 s + 200 ----------------s^3 + 5 s^2 + 4 s ans = Solutions to Problems 443 Mag ans = -7.0007 ans = BW wBW = 14.7500 Bandwidth = 14.75 Phase Margin = 16.6617, Damping Ratio = 0.15, Percent Overshoot = 62.0871,Settling Time = 2.76425, Peak Time = 0.329382 444 Chapter 10: Frequency Response Methods Zero/pole/gain: 50 (s+3) (s+5) ------------------s (s+2) (s+4) (s+6) Transfer function: 50 s^2 + 400 s + 750 ---------------------------s^4 + 12 s^3 + 44 s^2 + 48 s ans = Mag ans = -7.0026 ans = Solutions to Problems 445 BW wBW = 10.1100 Bandwidth = 10.11 Phase Margin = 29.2756, Damping Ratio = 0.27, Percent Overshoot = 41.439,Settling Time = 2.1583, Peak Time = 0.475337 446 Chapter 10: Frequency Response Methods 27. The phase margin of the given system is 20o. Using Eq. (10.73), ζ = 0.176. Eq. (4.38) yields 57% overshoot. The system is Type 1 since the initial slope is - 20 dB/dec. Continuing the initial slope down to the 0 dB line yields Kv = 4. Thus, steady-state error for a unit step input is zero; steady state error for a unit ramp input is 1 = 0.25; steady-state error for a parabolic input is infinite. Kv 28. The magnitude response is the same for all time delays and crosses zero dB at 0.5 rad/s. The following is a plot of the magnitude and phase responses for the given time delays: Solutions to Problems 447 a. For T = 0, ΦM = 93.3o; System is stable. For T = 0.1, ΦM = 55.1o; System is stable. 448 Chapter 10: Frequency Response Methods For T = 0.2, ΦM = 17o; System is stable. For T = 0.5, ΦM = -97o; System is unstable. Solutions to Problems 449 For T = 1, ΦM = 72.2o; System is unstable because the gain margin is -4.84 dB. b. For T = 0, the phase response reaches 180o at infinite frequency. Therefore the gain margin is infinite. The system is stable. For T = 0.1, the phase response is -180o at 11.4 rad/s. The magnitude response is -5.48 dB at 11.4 rad/s. Therefore, the gain margin is 5.48 dB. The system is stable. For T = 0.2, the phase response is -180o at 7.55 rad/s. The magnitude response is -1.09 dB at 7.55 rad/s. Therefore, the gain margin is 1.09 dB and the system is stable. For T = .5, the phase response is -180o at 4.12 rad/s. The magnitude response is +3.09 dB at 4.12 rad/s. Therefore, the gain margin is – 3.09 dB and the system is unstable. For T = 1, the phase response is -180o at 2.45 rad/s. The magnitude response is +4.84 dB at 2.45 rad/s. Therefore, the gain margin is -4.84 dB and the system is unstable. c. T = 0; T = 0.1; T = 0.2 d. T = 0.5, -3.09 dB; T = 1, - 4.84 dB; 29. The Bode plots for K = 1 and 0.5 second delay is: 450 Chapter 10: Frequency Response Methods Gain dB 0 -50 -100 -150 -1 10 10 0 10 1 10 2 Frequency (rad/sec) Phase deg 0 -360 -720 -1080 -1440 -1 10 10 0 10 1 10 2 Frequency (rad/sec) The phase is -180o at 2.12 rad/s. At this frequency, the gain is -34.76 dB. Thus the gain can be raised by 34.76 dB = 54.71. Hence for stability, 0<K<54.71. 30. The Bode plots for K = 40 and a delay of 0.5 second is shown below. Gain dB 50 0 -50 -100 -1 10 10 0 10 1 10 2 Frequency (rad/sec) 0 Phase deg -360 -720 -1080 -1440 -1 10 10 0 10 1 10 2 Frequency (rad/sec) The magnitude curve crosses zero dB at a frequency of 1.0447 rad/s. At this frequency, the phase plot shows a phase margin of 35.74 degrees. Using Eq. (10.73) or Figure 10.48, ζ = 0.33. Thus, %OS = 33.3. 31. Program: %Enter G(s)************************ numg1=1; deng1=poly([0 -3 -12]); 'G1(s)' G1=tf(numg1,deng1) [numg2,deng2]=pade(0.5,5); 'G2(s) (delay)' Solutions to Problems 451 G2=tf(numg2,deng2) 'G(s)=G1(s)G2(s)' G=G1*G2 %Enter K ************************** K=input('Type gain, K '); T=feedback(K*G,1); step(T) title(['Step Response for K = ',num2str(K)]) Computer response: ans = G1(s) Transfer function: 1 ------------------s^3 + 15 s^2 + 36 s ans = G2(s) (delay) Transfer function: -s^5 + 60 s^4 - 1680 s^3 + 2.688e004 s^2 - 2.419e005 s + 9.677e005 -----------------------------------------------------------------s^5 + 60 s^4 + 1680 s^3 + 2.688e004 s^2 + 2.419e005 s + 9.677e005 ans = G(s)=G1(s)G2(s) Transfer function: -s^5 + 60 s^4 - 1680 s^3 + 2.688e004 s^2 - 2.419e005 s + 9.677e005 -------------------------------------------------------------------------s^8 + 75 s^7 + 2616 s^6 + 5.424e004 s^5 + 7.056e005 s^4 + 5.564e006 s^3 + 2.322e007 s^2 + 3.484e007 s 452 Chapter 10: Frequency Response Methods Type gain, K 40 32. Bode Diagrams Original Problem showing 1/s estimate 80 60 40 Phase (deg); Magnitude (dB) 20 0 -20 -40 -60 -80 -100 -120 -140 -160 -180 10 -2 10 -1 10 0 Frequency (rad/sec) 10 1 10 2 Solutions to Problems 453 Original-1/s Magnitude 30 Magnitude (dB) 25 20 15 10 5 0 -5 10 -1 10 0 10 1 10 2 10 1 10 2 Frequency (rad/s) Original-1/s 0 Phase (deg) -20 -40 -60 -80 -100 10 -1 10 0 Frequency (rad/s) Original-1/s(s+5.3) Magnitude 60 40 20 0 -20 10 -1 10 0 10 1 10 2 10 1 10 2 Frequency (rad/s) Original-1/s(s+5.3) 8 6 4 2 0 -2 -4 10 -1 10 0 Frequency (rad/s) Estimated K = 41 dB = 112. Therefore, final estimate is G (s ) = 112 . s(s + 5.3) 33. Program: %Generate total system Bode plots - numg0,deng0 - M0,P0 clf numg0=12*poly([-1 -20]); deng0=conv([1 7],[1 4 100]); G0=tf(numg0,deng0); w=0.1:0.1:100; 454 Chapter 10: Frequency Response Methods [M0,P0]=bode(G0,w); M0=M0(:,:); P0=P0(:,:); [20*log10(M0),P0,w]; bode(G0,w) pause %Subtract (s+1) [numg1,deng1] and generate Bode plot-M2,P2 numg1=[1 1]; deng1=1; G1=tf(numg1,deng1); [M1,P1]=bode(G1,w); M1=M1(:,:); P1=P1(:,:); M2=20*log10(M0)-20*log10(M1); P2=P0-P1; clf subplot(2,1,1) semilogx(w,M2) grid subplot(2,1,2) semilogx(w,P2) grid pause %Subtract10^2/(s^2+2*0.3*10s+10^2) [numg2,deng2] and generate Bode plotM4,P4 numg2=100; deng2=[1 2*0.3*10 10^2]; G2=tf(numg2,deng2); [M3,P3]=bode(G2,w); M3=M3(:,:); P3=P3(:,:); M4=M2-20*log10(M3); P4=P2-P3; clf subplot(2,1,1) semilogx(w,M4) grid subplot(2,1,2) semilogx(w,P4) grid pause %Subtract(8.5/23)(s+23)/(s+8.5) [numg3,deng3] and generate Bode plot-M6,P6 numg3=(8.5/23)*[1 23]; deng3=[1 8.5]; G3=tf(numg3,deng3); [M5,P5]=bode(G3,w); M5=M5(:,:); P5=P5(:,:); M6=M4-20*log10(M5); P6=P4-P5; clf subplot(211) semilogx(w,M6) grid subplot(212) semilogx(w,P6) grid Solutions to Problems 455 Computer responses and analysis: 15 10 Gain dB 5 0 -5 -10 -15 -20 10-1 100 101 102 101 102 Frequency (rad/sec) 50 Phase deg 0 -50 -100 -150 -200 10-1 100 Frequency (rad/sec) Original data showing estimate of a component, (s+1) 0 -10 -20 -30 -40 -50 -60 10-1 100 101 102 100 101 102 0 -50 -100 -150 -200 10-1 Original data minus (s+1) showing estimate of (102/(s2+2*0.3*10s+102)) 456 Chapter 10: Frequency Response Methods -8 -10 -12 -14 -16 -18 -20 10-1 100 101 102 100 101 102 0 -10 -20 -30 -40 -50 10-1 Original data minus (s+1)(102/(s2+2*0.3*10s+102)) showing estimate of (8.5/23)(s+23)/(s+8.5) -6.5 -7 -7.5 -8 -8.5 -9 -9.5 -10 10-1 100 101 102 100 101 102 10 5 0 -5 -10 -15 10-1 Original data minus final estimate of G(s) = (s + 1) * 100 8.5 s + 23 * s + 6s + 100 23 s + 8.5 2 Solutions to Problems 457 Thus the final estimate is G(s) = (s + 1) * 100 8.5 s + 23 * * K. Since the original plot starts s + 6s + 100 23 s + 8.5 2 from -10 dB, 20 log K = -10, or K = 0.32. 34. a. Open-Loop Frequency Response 40 20 0 Phase (deg); Magnitude (dB) -20 -40 -60 -80 -100 0 -50 -100 -150 -200 -250 10 -4 10 -2 10 0 10 2 10 3 Frequency (rad/sec) From the Bode plot: Gain margin = 29 52 dB; phase margin = 157.50; 0 dB frequency = 1.63 rad/s; 1800 frequency = 49.8 rad/s. b. System is stable since it has 1800 of phase with a magnitude less than 0 dB. 458 Chapter 10: Frequency Response Methods 35. a. Open-Loop Frequency Response 20 0 -20 -40 Phase (deg); Magnitude (dB) -60 -80 -100 -120 -100 -150 -200 -250 -300 -350 10 0 10 1 10 2 10 3 10 4 Frequency (rad/sec) From the Bode plot: Gain margin = 17.1 dB; phase margin = 57.220; 0 dB frequency = 45.29 rad/s; 1800 frequency = 169.03 rad/s; bandwidth(@-7 dB open-loop) = 85.32 rad/s. b. From Eq. (10.73) ζ = 0.58; from Eq. (4.38) %OS = 10.68; from Eq. (10.55) Ts = 0.0949 s; from Eq. (10.56) Tp = 0.0531 s. c. Step Response 1 Amplitude 0.8 0.6 0.4 0.2 0 0 0.05 0.1 Time (sec.) Solutions to Problems 459 36. Resonance at 70 rad/s. 37. 10 G(s) = s(s+2)(s+10) . Plotting the Bode plots, The gain is zero dB at 0.486 rad/s and the phase angle is -106.44. Thus, the phase margin is 180o 106.44o = 73.56o . Using Eq. (10.73), ζ = 0.9. Using Eq. (4.38), %OS = 0.15%. 38. G(s) = 22.5 . Plotting the Bode plots, (s+4)(s2+0.9s+9) The phase response is 180o at ω = 3.55 rad/s, where the gain is -1.17 dB. Thus, the gain margin is 1.17 dB. Unity gain is at ω = 2.094 rad/s, where the phase is - 49.85o and at ω = 3.452 rad/s, where 460 Chapter 10: Frequency Response Methods the phase is -173.99o. Hence the phase margin is measured at ω = 3.452 rad/s and is 180o- 173.99o = 6.01o. Using Eq. (10.73), ζ = 0.0525. Eq. (4.38) yields %OS = 84.78%. 39. a. 40 20 0 -20 -2 10 10 -1 10 0 10 1 10 0 10 1 Frequency (rad/sec) 180 90 0 -90 10 -2 10 -1 Frequency (rad/sec) The frequencies that will be reduced occur at the peaks of the magnitude plot. The frequencies at the peaks are 4.14 rad/s and 0.754 rad/s. b. Consider a system with a disturbance, Rd at the output of a system: R (s) d R(s) + + G(s) + C(s) - The transfer function relating C(s) to Rd(s) is C(s) = C(s) 1 = . Therefore, R d (s 1 + G(s) N Rd DG N Rd 1 * = * N 1 + G D Rd DG + N G D Rd DG Thus, if the poles of G(s) match the poles of Rd (DG = DRd) there will be cancellation and the dynamics of the disturbance will be reduced. Thus, if the dynamics of Rd is oscillation, add poles in cascade with G(s) that have the same dynamics. Since the poles yield large gain at these bending frequencies a zero is placed near the poles so that the filter will have minimal effect on the transient Solutions to Problems 461 response (similar to placing a zero near a pole for a lag compensator). This arrangement of poles and zeros is called a dipole. Also note that a high gain at the bending frequency yields negative feedback for the output to subtract from Rd. Care should be exercised through analysis and simulation to be sure that the system's response to an input, other than the disturbance, is not adversely affected by the additional poles. 40. a. From Chapter 8, 0.6488K (s+53.85) Ge(s) = (s + 8.119s + 376.3) (s^2 + 15.47s + 9283) Cascading the notch filter, Get(s) = _________________________ 2 0.6488K (s+53.85)(s2 + 16s + 9200) _________________ 2 (s + 8.119s + 376.3) (s^2 + 15.47s + 9283)(s+60)2 Plotting the Bode plot, From the Bode plot: Gain margin = 96.74 dB; phase margin = ∞; 0 dB frequency = N/A; 1800 frequency = 30.44 rad/s. b. K = 96.74 dB = 68732 c. In Chapter 6 K = 188444. The difference is due to the notch filter. ELEVEN Design via Frequency Response SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Gain Design a. The required phase margin for 25% overshoot (ζ = 0.404), found from Eq. (10.73), is 43.49o. 50.88K From the solution to the Case Study Challenge problem of Chapter 10, G(s) = s(s+1.32)(s+100) . Using the Bode plots for K = 1 from the solution to the Case Study Challenge problem of Chapter 10, we find the required phase margin at ω = 1.35 rad/s, where the magnitude response is -14 dB. Hence, K = 5.01 (14 dB). b. Program: %Input system numg=50.88; deng=poly([0 -1.32 -100]); G=tf(numg,deng); %Percent Overshoot to Damping Ratio to Phase Margin Po=input('Type %OS '); z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); Pm=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi); fprintf('\nPercent Overshoot = %g',Po) fprintf(', Damping Ratio = %g',z) fprintf(', Phase Margin = %g',Pm) %Get Bode data bode(G) pause w=0.01:0.05:1000;%Step size can be increased if memory low. [M,P]=bode(G,w); M=M(:,:); P=P(:,:); Ph=-180+Pm; for i=1:1:length(P); if P(i)-Ph<=0; M=M(i); K=1/M; fprintf(', Frequency = %g',w(i)) fprintf(', Phase = %g',P(i)) fprintf(', Magnitude = %g',M) fprintf(', Magnitude (dB) = %g',20*log10(M)) fprintf(', K = %g',K) break end end T=feedback(K*G,1); step(T) Solutions to Case Studies Challenges 463 Computer response: Type %OS 25 Percent Overshoot = 25, Damping Ratio = 0.403713, Phase Margin = 43.463, Frequency = 1.36, Phase = -136.634, Magnitude = 0.197379, Magnitude (dB) = -14.094, K = 5.06641 464 Chapter 11: Design via Frequency Response Antenna Control: Cascade Compensation Design 50.88K a. From the solution to the previous Case Study Challenge in this chapter, G(s) = s(s+1.32)(s+100) . For Kv = 20, K = 51.89. Hence, the gain compensated system is 2640.16 G(s) = s(s+1.32)(s+100) Using Eq. (10.73), 15% overshoot (i.e. ζ = 0.517) requires a phase margin of 53.18o. Using the Bode plots for K = 1 from the solution to the Case Study Challenge problem of Chapter 10, we find the required phase margin at ω = 0.97 rad/s where the phase is -126.82o. To speed up the system, we choose the compensated phase margin frequency to be 4.6 * 0.97 = 4.46 rad/s. Choose the lag compensator break a decade below this frequency, or ω = 0.446 rad/s. At the phase margin frequency, the phase angle is -166.067o, or a phase margin of 13.93o. Using 5o leeway, we need to add 53.18o - 13.93o + 5o = 44.25o. From Figure 11.8, β = 0.15, or γ = 1 = β 6.667. Using Eq. (11.15), the lag portion of the compensator is GLag (s) = Using Eqs. (11.9) and (11.15), T2 = (s+0.446) s+0.446 0.446 = s+0.0669 . (s+6.667) 1 ωmax β = 0.579. From Eq. (11.15), the lead portion of the compensator is s+1.727 GLead (s) = s+11.51 The final forward path transfer function is 2640.16(s+0.446)(s+1.727) G(s)GLag (s)GLead(s) = s(s+1.32)(s+100)(s+0.0669)(s+11.51) b. Program: %Input system ***************************** K=51.89; numg=50.88*K; deng=poly([0 -1.32 -100]); G=tf(numg,deng); Po=15; z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); %Determine required phase margin************** Pmreq=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi) phreq=Pmreq-(180)%required phase w=0.1:0.01:10; [M,P]=bode(G,w); for i=1:1:length(P);%search for phase angle if P(i)-phreq<=0; ph=P(i) w(i) break end end wpm=4.6*w(i) Solutions to Case Studies Challenges 465 [M,P]=bode(G,wpm);%Find phase at wpm Pmreqc=Pmreq-(180+P)+5%Find contribution required from compensator+5 beta=(1-sin(Pmreqc*pi/180))/(1+sin(Pmreqc*pi/180)) %Design lag compensator*************** zclag=wpm/10; pclag=zclag*beta; Kclag=beta; %Design lead compensator********** zclead=wpm*sqrt(beta); pclead=zclead/beta; Kclead=1/beta; %Create compensated forward path********* numgclag=Kclag*[1 zclag]; dengclag=[1 pclag]; 'Gclag(s)' Gclag=tf(numgclag,dengclag); Gclagzpk=zpk(Gclag) numgclead=Kclead*[1 zclead]; dengclead=[1 pclead]; 'Gclead(s)' Gclead=tf(numgclead,dengclead); Gcleadzpk=zpk(Gclead) Gc=Gclag*Gclead; 'Ge(s)=G(s)*Gclag(s)*Gclead(s)' Ge=Gc*G; Gezpk=zpk(Ge) %Test lag-lead compensator**************** T=feedback(Ge,1); bode(Ge) title('Lag-lead Compensated') [Gm,Pm,wcp,wcg]=margin(Ge); 'Compensated System Results' fprintf('\nResulting Phase Margin = %g',Pm) fprintf(', Resulting Phase Margin Frequency = %g',wcg) pause step(T) title('Lag-lead Compensated') Computer response: Pmreq = 53.1718 phreq = -126.8282 ph = -126.8660 ans = 0.9700 wpm = 4.4620 Pmreqc = 44.2468 beta = 0.1780 ans = 466 Chapter 11: Design via Frequency Response Gclag(s) Zero/pole/gain: 0.17803 (s+0.4462) -----------------(s+0.07944) ans = Gclead(s) Zero/pole/gain: 5.617 (s+1.883) --------------(s+10.58) ans = Ge(s)=G(s)*Gclag(s)*Gclead(s) Zero/pole/gain: 2640.1632 (s+1.883) (s+0.4462) ---------------------------------------s (s+100) (s+10.58) (s+1.32) (s+0.07944) ans = Compensated System Results Resulting Phase Margin = 57.6157, Resulting Phase Margin Frequency = 2.68618» Answers to Review Questions 467 Answers to Review Questions 1. Steady-state error requirements can be designed simultaneously with transient response requirements. 2. Via the phase margin 3. The lag compensator is a low pass filter. Thus, while the low frequency gain is increased, the highfrequency gain at 180 o is decreased to make the system stable. 4. The lag network affects the phase angle at low frequencies, but not at high frequencies. For the compensated system, the phase plot is about the same as that of the uncompensated system around and above the phase margin frequency yielding the same transient response. 5. To compensate for the slight negative angle that the lag compensator has near the phase margin frequency 6. Compensated system has higher low-frequency gain than the uncompensated system designed to yield the same transient response; compensated and uncompensated system have the same phase margin frequency; the compensated system has lower gain around the phase margin frequency; the compensated and uncompensated system's have approximately the same phase values around the phase margin frequency. 7. The lead network is a high pass filter. It raises the gain at high frequencies. The phase margin frequency is increased. 468 Chapter 11: Design via Frequency Response 8. Not only is the magnitude curve increased at higher frequencies, but so is the phase curve. Thus the 180o point moves up in frequency with the increase in gain. 9. To correct for the negative phase angle of the uncompensated system 10. When designing the lag portion of a lag-lead compensator, we do not worry about the transient design. The transient response will be considered when designing the lead portion of a lag-lead compensator. SOLUTIONS TO PROBLEMS 1. a. Plot Bode plots for K = 1; angle is 180o at ω = 14.39 rad/s where the magnitude is –74.29 dB. Therefore a 64.29 dB ( or K = 1639) increase will yield a 10 dB gain margin. b. Plot Bode plots for K = 1; angle is 180o at ω = 5.196 rad/s where the magnitude is –50.21 dB. Therefore a 40.21 dB ( or K = 102.4) increase will yield a 10 dB gain margin. Solutions to Problems 469 c. Plot Bode plots for K = 1; angle is 180o at ω = 5.233 rad/s where the magnitude is –48.58 dB. Therefore a 38.58 dB ( or K = 84.92) increase will yield a 10 dB gain margin. 2. a. For a 40o phase margin, the phase must be -140o when the magnitude plot is zero dB. The phase is -140o at ω = 8.097 rad/s. At this frequency, the magnitude curve is –65.02 dB. Thus a 65.02 dB increase (K = 1782) will yield a 40o phase margin. b. For a 40o phase margin, the phase must be -140o when the magnitude plot is zero dB. The phase is -140o at ω = 2.201 rad/s. At this frequency, the magnitude curve is – 37.60 dB. Thus a 37.60 dB increase (K = 75.86) will yield a 40o phase margin. 470 Chapter 11: Design via Frequency Response c. For a 40o phase margin, the phase must be -140o when the magnitude plot is zero dB. The phase is -140o at ω = 2.653 rad/s. At this frequency, the magnitude curve is – 38.78 dB. Thus a 38.78 dB increase (K = 86.9) will yield a 40o phase margin. 3. 20% overshoot => ζ = 0.456 => φM = 48.15o. a. Looking at the phase diagram, where φM = 48.15o (i.e. φ = -131.85o), the phase margin frequency = 3.105 rad/s. At this frequency, the magnitude curve is -48.3 dB. Thus the magnitude curve has to be raised by 48.3 dB (K = 260). b. Looking at the phase diagram, where φM = 48.15o (i.e. φ = -131.85o), the phase margin frequency = 6.462 rad/s. At this frequency, the magnitude curve is – 63.04 dB. Thus the magnitude curve has to be raised by 63.04 dB (K = 1419). c. Looking at the phase diagram, where φM = 48.15o (i.e. φ = -131.85o), the phase margin frequency = 6.939 rad/s. At this frequency, the magnitude curve is – 64.42 dB. Thus the magnitude curve has to be raised by 64.42 dB (K = 1663). 4. a. Bode plots for K = 1: Using Eqs. (4.39) and (10.73) a percent overshoot = 15 is equivalent to a ζ = 0.517 and φM = 53.170. The phase-margin frequency = 2.2 rad/s where the phase is 53.170 – 1800 = -126.830. The magnitude = -8.966 dB, or 0.0.3562. Hence K = 1/ 0.3562 = 2.807. b. Program: G=zpk([-20 -25],[0 -5 -8 -14],1) K=2.807 T=feedback(K*G,1); Solutions to Problems 471 step(T) Computer response: Zero/pole/gain: (s+20) (s+25) -------------------s (s+5) (s+8) (s+14) K= 2.8070 5. For Kv = 50, K = 350. Plot the Bode plots for this gain. 472 Chapter 11: Design via Frequency Response Also, since %OS = 15%, ζ = 0.517. Using Eq. (10.73), φM = 53.17o. Increasing φM by 10o we will design for a phase margin of 63.17o. The phase margin frequency is where the phase angle is 63.17 - 180o = -116.83o, or ωφM = 3.54 rad/s. At this frequency, the magnitude is 22 dB. Start the magnitude of the compensator at - 22 dB and draw it to 1 decade below ωφM. 30 20 dB 10 0 -10 -20 -30 .001 .01 v .1 1 Then begin +20 dB/dec until zero dB is reached. Read the break frequencies as 0.028 rad/s and 0.354 rad/s from the Bode plot and form a lag transfer function that has unity dc gain: Gc(s) =0.0791 s + 0.354 s + 0.028 The compensated forward path is G(s) = 350 * 0.0791(s + 0.354) 27.69(s + 0.354) = s(s + 7)(s + 0.028) s(s + 7)(s + 0.028) Solutions to Problems 473 6. a. For Kv = 1000, K = 1473. Plotting the Bode for this value of K: Using Eqs. (4.39) and (10.73) a percent overshoot = 15 is equivalent to a ζ = 0.517 and φM = 53.17. Using an extra 10o, the phase margin is 63.17o. The phase-margin frequency = 1.21 rad/s. At this frequency, the magnitude = 57.55 dB = 754.2. Hence the lag compensator K = 1/754.2 = 0.001326. Following Steps 3 and 4 of the lag compensator design procedure in Section 11.3, s + 0.121 Glag(s) = 0.001326 s + 0.0001604 b. Program: %Input system numg=1473*poly([-10 -11]); deng=poly([0 -3 -6 -9]); G=tf(numg,deng); numc=0.001326*[1 0.121]; denc=[1 0.0001604]; Gc=tf(numc,denc); Ge=G*Gc; T=feedback(Ge,1); step(T) Computer response: 474 Chapter 11: Design via Frequency Response 7. Uncompensated system: Searching along the 121.1o line (15% overshoot), find the dominant pole at -2.15 ± j3.56 with K = 97.7. Therefore, the uncompensated static error constant is Kvo = 97.7 = 1.396. On the frequency 70 response curves, plotted for K = 97.7, unity gain occurs at ω = 1.64 rad/s with a phase angle of -71o. Therefore the uncompensated phase margin is 180o - 71o = 109o. Compensated system: 1 1 = = 0.4174 . For a 5 times improvement 1 + K po 1 + 97.7 70 1 K = 0.0835 , yielding , K pn = 10.98 = . Thus in steady-state error, estep (∞) = 1 + K pn 70 K = 768.6 . Plotting the Bode plots at this gain, The old steady-state error, estep (∞ ) = Solutions to Problems 475 Adding 5o, the desired phase margin for 15% overshoot is 58.17o, or a phase angle of -121.83o. This phase angle occurs at ω = 3.505 rad/s. At this frequency the magnitude plot is +12 dB. Start the magnitude of the compensator at - 12 dB and draw it to 1 decade below ωΦM. 476 Chapter 11: Design via Frequency Response Then, begin +20 dB/dec until zero dB is reached. Read the break frequencies as 0.08797 rad/s and 0.3505 rad/s from the Bode plot and form a lag transfer function that has unity dc gain, Gc ( s ) = 0.251 s + 0.3505 s + 0.08797 The compensated forward path is G ( s ) = 0.251* 768.6 ( s + 0.3505) 192.91( s + 0.3505) . = ( s + 2)( s + 5)( s + 7)( s + 0.08797) ( s + 2)( s + 5)( s + 7)( s + 0.08797) 8. For K p = 100 = K (4 ) , K = 2400. Plotting the Bode plot for this gain, (2 )(6)(8) Uncompensated system 40 20 log M 30 20 10 0 -10 -20 0.1 1 10 100 10 100 Frequency (rad/s) Phase (degrees) 0 -50 -100 -150 -200 0.1 1 Frequency (rad/s) We will design the system for a phase margin 100 larger than the specification. Thus φm = 550. The phase margin frequency is where the phase angle is –1800 + 550 = -1250. From the Bode plot this frequency is ωφ = 11 rad/s. At this frequency the magnitude is 23.37 dB. Start the magnitude of the m lag compensator at –23.37 dB and draw it to 1 decade below ωφ = 11 , or 1.1 rad/s. Then begin a m +20 dB/dec climb until 0 dB is reached. Read the break frequencies as 0.0746 rad/s and 1.1 rad/s from the Bode plot as shown below. Solutions to Problems 477 Lag compensator 0 20 log M -5 -10 -15 -20 -25 0.01 0.1 1 10 1 10 Frequency (rad/s) 0 Phase (degrees) -10 -20 -30 -40 -50 -60 -70 0.01 0.1 Frequency (rad/s) Ensuring unity dc gain, the transfer function of the lag is Glag ( s ) = 0.06782 ( s + 1.1) . The ( s + 0.0746) compensated forward-path transfer function is thus the product of the plant and the compensator, or Ge (s) = 162.8( s + 4)(s + 1.1) (s + 2)(s + 6)(s + 8)( s + 0.0746) 9. From Example 11.1, K = 58251 yields 9.48% overshoot or a phase margin of 59.19o. Also, 58251 G(s) = s(s+36)(s+100) Allowing for a 10o contribution from the PI controller, we want a phase margin of 69.19o, or a phase angle of -180o + 69.19o = -110.81o. This phase angle occurs at ω = 9.8 rad/s where the magnitude is 4 dB. Thus, the PI controller should contribute - 4 dB at ω = 9.8 rad/s. Selecting a break frequency a decade below the phase margin frequency, Gc(s) = s+0.98 s This function has a high-frequency gain of zero dB. Since we want a high-frequency gain of -4 dB (a gain of 0.631), 478 Chapter 11: Design via Frequency Response Gc(s) = 0.631 s+0.98 s The compensated forward path is G(s) = 58251*0.631(s+0.98) s(s+36)(s+100) = 36756.38(s+0.98) s(s+36)(s+100) 10. Bode plots for K = 1: Using Eqs. (4.39) and (10.73) a percent overshoot = 15 is equivalent to a ζ = 0.517 and φM = 53.17o. The phase-margin frequency = 1.66 rad/s. The magnitude = -9.174 dB = 0.3478. Hence K = 1/ 0.3478 = 2.876. b. Bode plots for K = 2.876. Solutions to Problems 479 Adding 10o to the phase margin yields 63.17. Thus, the required phase is –1800 + 63.170 = -116.830, which occurs at a frequency of 1.21 rad/s. The magnitude = 3.366 dB = 1.473. Hence, the lag compensator K = 1/ 1.473 = =0.6787. Selecting the break a decade below the phase-margin frequency, Gc(s) = 0.6787 c. Program: %Input system numg=2.876*poly([-10 -11]); deng=poly([0 -3 -6 -9]); G=tf(numg,deng); numc=0.6787*[1 0.121]; denc=[1 0]; Gc=tf(numc,denc); Ge=G*Gc; T=feedback(Ge,1); step(T) s+0.121 s 480 Chapter 11: Design via Frequency Response Computer response: 11. Program: %PI Compensator Design via Frequency Response %Input system G=zpk(,[-5 -10],1); G=tf(G); %Percent Overshoot to Damping Ratio to Phase Margin Po=input('Type %OS '); z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); Pm=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi)+10; fprintf('\nPercent Overshoot = %g',Po) fprintf(', Damping Ratio = %g',z) fprintf(', Phase Margin = %g',Pm) %Get Bode data bode(G) title('Uncompensated') pause %Find frequency at desired phase margin and the gain at this frequency w=logspace(-1,2,10000); %w=.1:0.1:100; [M,P,w]=bode(G,w); Ph=-180+Pm for i=1:1:length(P); if P(i)-Ph<=0 Mag=M(i) wf=w(i); fprintf(', Frequency = %g',wf) fprintf(', Phase = %g',P(i)) fprintf(', Magnitude = %g',Mag) fprintf(', Magnitude (dB) = %g',20*log10(Mag)) break end end Solutions to Problems 481 %Design PI compensator %Break frequency is a decade below phase margin frequency wh=wf/10; %Magnitude is reciprocal of magnitude of G at the phase margin frequency %so net magnitude is 0 dB at the phase margin frequency Kc=1/Mag 'PI Compensator' Gpi=tf(Kc*[1 wh],[1 0]) bode(Gpi) title(['PI compensator']) pause 'G(s)Gpi(s)' Ge=series(G,Gpi); Ge=zpk(Ge) bode(Ge) title('PI Compensated') [Gm,Pm,Wcp,Wcg]=margin(Ge); 'Gain margin(dB); Phase margin(deg.); 0 dB freq. (r/s);' '180 deg. freq. (r/s)' margins=[20*log10(Gm),Pm,Wcg,Wcp] pause T=feedback(Ge,1); step(T) title('PI Compensated') Computer response: Type %OS 25 Percent Overshoot = 25, Damping Ratio = 0.403713, Phase Margin = 53.463 Ph = -126.5370 Mag = 0.0037 , Frequency = 14.5518, Phase (dB) = -48.6811 Kc = 271.6786 ans = PI Compensator Transfer function: 271.7 s + 395.3 --------------s ans = G(s)Gpi(s) Zero/pole/gain: 271.6786 (s+1.455) -----------------s (s+10) (s+5) = -126.54, Magnitude = 0.00368082, Magnitude 482 Chapter 11: Design via Frequency Response ans = Gain margin(dB); Phase margin(deg.); 0 dB freq. (r/s); ans = 180 deg. freq. (r/s) margins = Inf 47.6277 14.5975 Inf Solutions to Problems 483 484 Chapter 11: Design via Frequency Response 12. For Kv = 4, K = 4 , or K = 3600. Plot the Bode diagrams. 900 The magnitude curve crosses zero dB at ω = 2.83 rad/s. with a phase angle of 152.1o, which yields an Solutions to Problems 485 uncompensated phase margin of 27.9o. Thus, we need an additional 12.1o plus an additional amount to compensate for the fact that the phase margin frequency will increase. Assume a lead network with a phase contribution of 22.1o. Using Eqs. (11.11), and (11.12), The value of beta is: The |G(jwmax)| for the compensator is: or in db: 0.453 1.485 3.44 The magnitude curve has a gain of -3.44 dB at ω = 3.625 rad/s. Therefore, choose this frequency as the new phase margin frequency. Using Eqs. (11.9) and (11.6), the compensator transfer function has the following specifications: T zero pole gain 0.41 -2.44 -5.38 2.21 The compensated forward path is G ( s) = 3600 * 2.21( s + 2.44) 7956( s + 2.44) = s ( s + 3)( s + 15)( s + 20)( s + 5.38) s ( s + 3)( s + 15)( s + 20)( s + 5.38) A Bode plot of G(s) shows a phase margin of 37.8o. Thus, a redesign is necessary to meet the exact requirement. This redesign can be done by adding a larger correction factor to the phase required from the lead compensator, See Control Solutions for the redesign. 13. a. Gain-compensated time response: 486 Chapter 11: Design via Frequency Response Bode plots for K = 1000 (Kv = 10): The specifications for the gain compensated system are: K = 1000, percent overshoot = 10, ζ = 0.591155, peak time = 0.5 s, current phase margin = 22.5362o. To meet the requirements: required phase margin (Eq. 10.73) = 58.5931o, required phase margin with correction factor of 20o = 78.5931, required bandwidth (Eq. 10.56) = 9.03591, required phase contribution from compensator = 78.5931o - 22.5362o = 56.0569o, compensator beta (Eq. 11.11) = 0.0931398, new phase margin frequency (Eq. 11.12) = 11.51. Now design the compensator: Compensator gain Kc = 1/β = 10.7366, compensator zero (Eq. 11.12) = -3.51272, compensator pole = zc/β = -37.7144. Lead-compensated Bode plots: Solutions to Problems 487 Gain dB 50 0 -50 -100 -1 10 100 10 1 Frequency (rad/sec) 10 2 10 3 10 2 10 3 Lead-Compensated System Phase deg 0 -90 -180 -270 10 -1 100 10 1 Frequency (rad/sec) Lead-compensated phase margin = 50.2352. b. Program: numg=1000; deng=poly([0 -5 -20]); G=tf(numg,deng); numc=[1 3.51272]; denc=[1 37.7144]; Gc=tf(numc,denc); Ge=G*10.7366*Gc; T=feedback(Ge,1); step(T) Computer response: 488 Chapter 11: Design via Frequency Response 14. Uncompensated system: Searching the ζ = 0.456 line (20% overshoot), find the dominant poles Q = -6.544 ± j12.771 with a gain of 178.21. Hence, Ts = Kp = 4 = 0.611 second, ζωn 178.21* 5 = 7.425 . The Bode plot for the uncompensated system is: 2 * 6 *10 The uncompensated system has a phase margin of 55.8o and a phase margin frequency of 11.3 rad/s. Compensated system: For a threefold improvement in Kp, Kpn = 22.28. Therefore, K = 3*178.21 = 534.63. For a twofold reduction in settling time, the new dominant poles are Qn = 2Q = - 13.09 ± j25.54. The gain adjusted system is Gc ( s ) = 534.63( s + 5) ( s + 2)( s + 6)( s + 10) Plotting the Bode diagrams for the gain compensated system, Solutions to Problems 489 At unity gain the phase is -147.8o at ω = 21.9 rad/s. Thus, the gain compensated phase margin is 180o – 147.8o = 32.2o. Using Eq. (10.73) with ζ = 0.456 (i.e. 20% overshoot), the required ΦM = 48.15o. We add 15.95o plus a correction factor of 5o to the phase margin of the gain compensated system for a total additional phase of 20.95o. Using Eqs. (11.11), and (11.12), The value of beta is: The |G(jwmax)| for the compensator is: or in db: 0.473 1.45 3.25 The magnitude curve has a gain of -3.25 dB at ω = 26.9 rad/s. Therefore, choose this frequency as the new phase margin frequency. Using Eqs. (11.9) and (11.6), the compensator transfer function has the following specifications: 490 Chapter 11: Design via Frequency Response T zero pole gain 0.054 -18.51 -39.1 2.11 The compensated forward path is G ( s) = 534.63 * 2.11( s + 5)( s + 18.51) 1128.1( s + 5)( s + 18.51) = ( s + 2)( s + 6)( s + 10)( s + 39.1) ( s + 2)( s + 6)( s + 10)( s + 39.1) A simulation of the system shows percent overshoot = 23.2%, settling time = 0.263, phase margin = 48.4o, phase margin frequency = 26.7 r/s. 15. If G(s) = 144000 , Kv = 40. Also, for a 0.1 second peak time, and ζ = 0.456 (20% s(s + 36)(s + 100) overshoot), Eq. (10.56) yields a required bandwidth of 46.59 rad/s. Using Eq. (10.73), the required phase margin is 48.15o. Let us assume that we raise the phase margin frequency to 39 rad/s. The phase angle of the uncompensated system at this frequency is -158.6o. To obtain the required phase margin, the compensator must contribute 26.75o more at 39 rad/s. Assume the following form for the 1 compensator: Gc(s) = K'KD(s+K ). The angle contributed by the compensator is D φc = tan-1 ω 1/ K D = 26.75o. Letting ω = 39 rad/s, KD = 0.0129. Hence, the compensator is Gc(s) = 0.0129 (s+77.37). The compensated forward path is G(s) = 144000 * 0.0129(s + 77.37) 1857.6(s + 77.37) = s(s + 36)(s + 100) s(s + 36)(s + 100) The closed-loop bandwidth is approximately 50 rad/s, which meets the requirements. The lag compensated forward path is G(s) = 7.759 (s+0.058) s(s2+2s+5)(s+3)(s+0.0015) 16. a. Bode plots and specifications for gain compensated system are the same as Problem 13. Required phase margin and required bandwidth is the same as Problem 13. Select a phase margin frequency 7 rad/s higher than the bandwidth = 9 + 7 = 16 rad/s. The phase angle at the new phase-margin frequency is -201.60. The phase contribution required from the compensator is –1800 + 201.60 + 58.590 = 80.20 at the phase-margin frequency. Using the geometry below: Solutions to Problems 491 phase-margin frequency=j16 80.2 o -z c 16 1 tan (80.2) = z . Therefore, zc = 2.76. Thus, Gc (s) = 2.76 (s + 2.76) . c The PD compensated Bode plots: Gain dB 50 0 -50 -1 10 10 0 10 1 10 2 10 1 10 2 Frequency (rad/sec) PD Compensated Phase deg -60 -90 -120 -150 -180 10-1 10 0 Frequency (rad/sec) Compensated phase margin is 62.942o. b. Program: numg=1000; deng=poly([0 -5 -20]); G=tf(numg,deng); numc=(1/2.76)*[1 2.76]; denc=1; Gc=tf(numc,denc); Ge=G*Gc; T=feedback(Ge,1); step(T) title('PD Compensated') 492 Chapter 11: Design via Frequency Response Computer response: 17. Program: %Lead Compensator Design via Frequency Response %Input system K=input('Type K to meet steady-state error '); numg=K*[1 1]; deng=poly([0 -2 -6]); 'Open-loop system' 'G(s)' G=tf(numg,deng) %Generate uncompensated step response T=feedback(G,1); step(T) title('Gain Compensated') %Input transient response specifications Po=input('Type %OS '); %Ts=input('Type settling time '); Tp=input('Type peak time '); %Determine required bandwidth z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); %wn=4/(z*Ts); wn=pi/(Tp*sqrt(1-z^2)); wBW=wn*sqrt((1-2*z^2)+sqrt(4*z^4-4*z^2+2)); %Make a Bode plot and get Bode data %Get Bode data bode(G) title('Gain Compensated') w=0.01:0.1:100; [M,P]=bode(numg,deng,w); %Find current phase margin [Gm,Pm,wcp,wcg]=margin(M,P,w); %Calculate required phase margin z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); Pmreq=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi); Solutions to Problems 493 %Add a correction factor of 10 degrees Pmreqc=Pmreq+10; %Calculate phase required from compensator Pc=Pmreqc-Pm; %Design lead compensator %Find compensator beta and peak compensator magnitude beta=(1-sin(Pc*pi/180))/(1+sin(Pc*pi/180)); magpc=1/sqrt(beta); %Find frequency at which uncompensated system has a magnitude of 1/magpc %This frequency will be the new phase margin frequency for i=1:1:length(M); if M(i)-(1/magpc)<=0; wmax=w(i); break end end %Calculate the lead compensator's break frequencies zc=wmax*sqrt(beta); pc=zc/beta; Kc=1/beta; numc=[1 zc]; denc=[1 pc]; 'Gc(s)' Gc=tf(numc,denc) %Display data fprintf('\nK = %g',K) fprintf(' Percent Overshoot = %g',Po) fprintf(', Damping Ratio = %g',z) %fprintf(', Settling Time = %g',Ts) fprintf(', Peak Time = %g',Tp) fprintf(', Current Phase Margin = %g',Pm) fprintf(', Required Phase Margin = %g',Pmreq) fprintf(', Required Phase Margin with Correction Factor = %g',Pmreqc) fprintf(', Required Bandwidth = %g',wBW) fprintf(', Required Phase Contribution from Compensator = %g',Pc) fprintf(', Compensator Beta = %g',beta) fprintf(', New phase margin frequency = %g',wmax) fprintf(' Compensator gain, Kc = %g',Kc) fprintf(' Compensator zero,= %g',-zc) fprintf(' Compensator pole,= %g',-pc) 'G(s)Gc(s)' Ge=G*Kc*Gc pause %Generate compensated Bode plots %Make a Bode plot and get Bode data %Get Bode data bode(Ge) title('Lead Compensated') w=0.01:0.1:1000; [M,P]=bode(Ge,w); %Find compensated phase margin [Gm,Pm,wcp,wcg]=margin(M,P,w); fprintf('\nCompensated Phase Margin,= %g',Pm) pause %Generate step response T=feedback(Ge,1); step(T) title('Lead Compensated') Computer response: Type K to meet steady-state error ans = 360 494 Chapter 11: Design via Frequency Response Open-loop system ans = G(s) Transfer function: 360 s + 360 -----------------s^3 + 8 s^2 + 12 s Type %OS 10 Type peak time 0.1 ans = Gc(s) Transfer function: s + 11.71 --------s + 77.44 K = 360 Percent Overshoot = 10, Damping Ratio = 0.591155, Peak Time = 0.1, Current Phase Margin = 21.0851, Required Phase Margin = 58.5931, Required Phase Margin with Correction Factor = 68.5931, Required Bandwidth = 45.1795, Required Phase Contribution from Compensator = 47.508, Compensator Beta = 0.151164, New phase margin frequency = 30.11 Compensator gain, Kc = 6.61532 Compensator zero,= -11.7067 Compensator pole,= -77.4437 ans = G(s)Gc(s) Transfer function: 2382 s^2 + 3.026e004 s + 2.788e004 ------------------------------------s^4 + 85.44 s^3 + 631.5 s^2 + 929.3 s Compensated Phase Margin,= 60.676» Solutions to Problems 495 496 Chapter 11: Design via Frequency Response Solutions to Problems 497 18. Program: %PD Compensator Design via Frequency Response %Input system %Uncompensated system K=input('Type K to meet steady-state error '); numg=K*[1 1]; deng=poly([0 -2 -6]); G=tf(numg,deng); T=feedback(G,1); step(T) title('Gain Compensated') 'Open-loop system' 'G(s)' Gzpk=zpk(G) %Input transient response specifications Po=input('Type %OS '); %Ts=input('Type settling time '); Tp=input('Type peak time '); %Determine required bandwidth z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); %wn=4/(z*Ts); wn=pi/(Tp*sqrt(1-z^2)); wBW=wn*sqrt((1-2*z^2)+sqrt(4*z^4-4*z^2+2)); %Make a Bode plot and get Bode data %Get Bode data bode(G) title('Gain Compensated') w=0.01:0.1:100; [M,P]=bode(G,w); %Find current phase margin [Gm,Pm,wcp,wcg]=margin(M,P,w); %Calculate required phase margin z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); Pmreq=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi)+20; %Determine a phase margin frequency wpm=wBW+7; %Find phase angle at new phase margin frequency and %calculate phase required from the compensator for i=1:1:length(w); if w(i)-wpm>=0; wpm=w(i); Pwpm=P(i); break end end %Design PD compensator Pc=Pmreq-(180+Pwpm); zc=wpm/tan(Pc*pi/180); Kc=1/zc; numc=Kc*[1 zc]; denc=1; 'Gc(s)' Gc=tf(numc,denc); Gczpk=zpk(Gc) %Display data fprintf('\nK = %g',K) fprintf(' Percent Overshoot = %g',Po) fprintf(', Damping Ratio = %g',z) %fprintf(', Settling Time = %g',Ts) 498 Chapter 11: Design via Frequency Response fprintf(', fprintf(', fprintf(', fprintf(', fprintf(', fprintf(', fprintf(', fprintf(' fprintf(' Peak Time = %g',Tp) Current Phase Margin = %g',Pm) Required Phase Margin = %g',Pmreq) Required Bandwidth = %g',wBW) New phase margin frequency = %g',wpm) Phase angle at new phase margin frequency = %g',Pwpm) Required Phase Contribution from Compensator = %g',Pc) Compensator gain, Kc = %g',Kc) Compensator zero,= %g',-zc) pause %Generate compensated Bode plots %Make a Bode plot and get Bode data %Get Bode data 'G(s)Gc(s)' Ge=G*Gc; Gezpk=zpk(Ge) bode(Ge) title('PD Compensated') w=0.01:0.1:100; [M,P]=bode(Ge,w); %Find compensated phase margin [Gm,Pm,wcp,wcg]=margin(M,P,w); fprintf('\nCompensated Phase Margin,= %g',Pm) pause %Generate step response T=feedback(Ge,1); step(T) title('PD Compensated') Computer response: Type K to meet steady-state error 360 ans = Open-loop system ans = G(s) Zero/pole/gain: 360 (s+1) ------------s (s+6) (s+2) Type %OS 10 Type peak time 0.1 ans = Gc(s) Zero/pole/gain: 0.05544 (s+18.04) K = 360 Percent Overshoot = 10, Damping Ratio = 0.591155, Peak Time = 0.1, Current Phase Margin = 21.0851, Required Phase Margin = 78.5931, Required Bandwidth = 45.1795, New phase margin frequency = 52.21, Phase angle at new phase margin frequency = -172.348, Required Phase Contribution from Compensator = 70.9409 Compensator gain, Kc = 0.0554397 Compensator zero,= -18.0376 ans = Solutions to Problems 499 G(s)Gc(s) Zero/pole/gain: 19.9583 (s+18.04) (s+1) ----------------------s (s+6) (s+2) Compensated Phase Margin,= 69.546 500 Chapter 11: Design via Frequency Response 19. K = 10714 for Kv = 1000. ζ = 0.517 for 15% overshoot using Eq. (4.39). Using Eq. (4.42), ωn = 77.37. Using Eq. (10.54) the required bandwidth, ωBW = 96.91. Using Eq. (10.73) with 5o additional, Φm = 58.17o. Choose the new phase-margin frequency ωPm= 0.8 wBW= 77.53. Plotting the Bode plots for K = 10714, Solutions to Problems 501 80 Gain dB 60 40 20 0 -1 10 10 0 10 1 10 2 10 1 10 2 Frequency (rad/sec) Gain compensated Phase deg -60 -90 -120 -150 -180 10 -1 10 0 Frequency (rad/sec) At the new phase-margin frequency, the phase angle is -170.52. Thus, the contribution required from the lead is 58.17 - (180 -170.52) = 48.69o. Using Eq. (11.11), β = 0.142. Lag compensator design: zclag= ωPm/10 = 77.53/10 = 7.753. pclag = zclag*β = 1.102. Kclag = s+7.753 pclag/ zclag = 0.1421. Thus, Glag(s) = 0.1421 s+1.102 . Lead compensator design: Using Eqs. (11.6), (11.9), and (11.12) zlead = 1/T = ωPm* β = 29.22. s+29.22 plead = zlead /β = 205.74. Klead = plead /zlead = 7.04. Thus, Glead(s) = 7.04 s+205.74 . 20. Program: %Lag-Lead Compensator Design via Frequency Response %Input system K=input('Type K to meet steady-state error '); numg=K*[1 7]; deng=poly([0 -5 -15]); G=tf(numg,deng); 'G(s)' Gzpk=zpk(G) %Input transient response specifications Po=input('Type %OS '); Ts=input('Type settling time '); %Tp=input('Type peak time '); T=feedback(G,1); step(T) title('Gain Compensated') pause %Determine required bandwidth z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); wn=4/(z*Ts); %wn=pi/(Tp*sqrt(1-z^2)); 502 Chapter 11: Design via Frequency Response wBW=wn*sqrt((1-2*z^2)+sqrt(4*z^4-4*z^2+2)); %wBW=(4/(Ts*z))*sqrt((1-2*z^2)+sqrt(4*z^4-4*z^2+2)); %wBW=(pi/(Tp*sqrt(1-z^2)))*sqrt((1-2*z^2)+sqrt(4*z^4-4*z^2+2)); %Determine required phase margin Pmreq=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi)+5; %Choose new phase margin frequency wpm=0.8*wBW; %Determine additional phase lead required at the %new phase margin frequency from the lead compensator [M,P]=bode(G,wpm); Pmreqc=Pmreq-(180+P); beta=(1-sin(Pmreqc*pi/180))/(1+sin(Pmreqc*pi/180)); %Display data fprintf('\nPercent Overshoot = %g',Po) fprintf(', Settling Time = %g',Ts) %fprintf(', Peak Time = %g',Tp) fprintf(', Damping Ratio = %g',z) fprintf(', Required Phase Margin = %g',Pmreq) fprintf(', Required Bandwidth = %g',wBW) fprintf(', New Phase Margin Frequency = %g',wpm) fprintf(', Required Phase from Lead Compensator = %g',Pmreqc) fprintf(', Beta = %g',beta) bode(numg,deng) title('Gain compensated') pause %Design lag compensator zclag=wpm/10; pclag=zclag*beta; Kclag=beta; 'Lag compensator' 'Gclag' Gclag=tf(Kclag*[1 zclag],[1 pclag]); Gclagzpk=zpk(Gclag) %Design lead compensator zclead=wpm*sqrt(beta); pclead=zclead/beta; Kclead=1/beta; 'Lead compensator' 'Gclead' Gclead=tf(Kclead*[1 zclead],[1 pclead]); Gcleadzpk=zpk(Gclead) %Create compensated forward path 'Gclag(s)Gclead(s)G(s)' Ge=G*Gclag*Gclead; Gezpk=zpk(Ge) %Test lag-lead compensator T=feedback(Ge,1); bode(Ge) title('Lag-lead Compensated') [M,P,w]=bode(Ge); [Gm,Pm,wcp,wcg]=margin(M,P,w); 'Compensated System Results' fprintf('\nResulting Phase Margin = %g',Pm) fprintf(', Resulting Phase Margin Frequency = %g',wcg) pause step(T) title('Lag-lead Compensated') Computer response: Type K to meet steady-state error ans = G(s) 10714.29 Solutions to Problems 503 Zero/pole/gain: 10714.29 (s+7) -------------s (s+15) (s+5) Type %OS 15 Type settling time 0.1 Percent Overshoot = 15, Settling Time = 0.1, Damping Ratio = 0.516931, Required Phase Margin = 58.1718, Required Bandwidth = 96.9143, New Phase Margin Frequency = 77.5314, Required Phase from Lead Compensator = 48.6912, Beta = 0.142098 ans = Lag compensator ans = Gclag Zero/pole/gain: 0.1421 (s+7.753) ---------------(s+1.102) ans = Lead compensator ans = Gclead Zero/pole/gain: 7.0374 (s+29.23) ---------------(s+205.7) ans = Gclag(s)Gclead(s)G(s) Zero/pole/gain: 10714.29 (s+29.23) (s+7.753) (s+7) ---------------------------------s (s+205.7) (s+15) (s+5) (s+1.102) ans = Compensated System Results Resulting Phase Margin = 53.3994, Resulting Phase Margin Frequency = 55.5874 504 Chapter 11: Design via Frequency Response Solutions to Problems 505 Percent overshoot exceeds requirements. Redesign if required. 21. The required bandwidth for a peak time of 2 seconds and ζ = 0.456 (i.e. 20% overshoot) is 2.3297 rad/s. Plotting the Bode diagrams for K = 1, 506 Chapter 11: Design via Frequency Response For 20% overshoot, ΦM = 48.15o, or a phase angle of -180o + 48.15o = -131.85o. This angle occurs at 1.12 rad/s. If K = 13.1, the magnitude curve will intersect zero dB at 1.12 rad/s. Thus, the following function yields 20% overshoot: G(s) = 13.1 . s ( s + 2)( s + 5) PI controller design: Allowing for a 5o margin, we want ΦM = 48.15o + 5o = 53.15o, or a phase angle of -180o + 53.15o = -126.85o. This angle occurs at ω = 0.97 rad/s where the magnitude curve is 1.5321 dB. The controller should contribute - 1.5321 dB so that the magnitude curve passes through 0 dB at ω = 0.97 rad/s. Choosing the break frequency one decade below the phase margin frequency of 0.97 rad/s, and adjusting the controller's gain to yield -1.5321 dB at high frequencies, the ideal integral controller is G cPI (s) = 1.198( s + 0.097) s and the PI compensated forward path is G PI (s) = G(s)G cPI (s) = 15.694(s+0.097) s 2 ( s + 2)( s + 5) Plotting the Bode diagram for the PI compensated system yields, Solutions to Problems 507 This function is zero dB at ω = 1.28 rad/s. The phase at this frequency is – 141.4o. Thus, we have a phase margin of 38.6o. PID controller design: Let us increase the phase margin frequency to 4 rad/s. At this frequency the phase is -193.48o. To obtain the required phase margin of 48.15o the phase curve must be raised an 1 additional 61.63o. Assume the following form for the compensator : GcPD(s) = K'KD(s+K ). The D ω angle contributed by the compensator is φc = tan-1 1/K D = 61.63o. Letting ω = 4 rad/s, KD = 0.463. Hence, the compensator is GcPD (s) = 0.463K' (s+2.16). The final PID compensated forward path is G PID (s)=G PI ( s )GcPD ( s ) = 15.694( s + 0.097) 2.266 K ' ( s + 0.097)( s + 2.16) * 0.463K ' ( s + 2.16) = s 2 ( s + 2)( s + 5) s 2 ( s + 2)( s + 5) Letting K' = 1 the magnitude of this function at 4 rad/s is -20.92 dB. Thus, K' must be adjusted to bring the magnitude to zero dB. Hence, K' = 11.12 (20.92 dB). 508 Chapter 11: Design via Frequency Response Thus, G PID (s)= 25.2( s + 0.097)( s + 2.16) s 2 ( s + 2)( s + 5) The PID compensated Bode plot follows: The PID compensated time response is shown below: Solutions to Problems 509 22. Program: %Input system numg1=1; deng1=poly([0 -3 -6]); G1=tf(numg1,deng1); [numg2,deng2]=pade(0.5,5); G2=tf(numg2,deng2); 'G(s)=G1(s)G2(s)' G=G1*G2; Gzpk=zpk(G) Tu=feedback(G,1); step(Tu) title('K = 1') %Percent Overshoot to Damping Ratio to Phase Margin Po=input('Type %OS '); z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); Pm=atan(2*z/(sqrt(-2*z^2+sqrt(1+4*z^4))))*(180/pi); fprintf('\nPercent Overshoot = %g',Po) fprintf(', Damping Ratio = %g',z) fprintf(', Phase Margin = %g',Pm) %Get Bode data bode(G) title('K = 1') pause w=0.1:0.01:100; [M,P]=bode(G,w); Ph=-180+Pm; for i=1:1:length(P); if P(i)-Ph<=0; M=M(i); K=1/M; 510 Chapter 11: Design via Frequency Response fprintf(', Frequency = %g',w(i)) fprintf(', Phase = %g',P(i)) fprintf(', Magnitude = %g',M) fprintf(', Magnitude (dB) = %g',20*log10(M)) fprintf(', K = %g',K) break end end T=feedback(K*G,1); step(T) title('Gain Compensated') Computer response: ans = G(s)=G1(s)G2(s) Zero/pole/gain: - (s-14.59) (s^2 - 26.82s + 228.4) (s^2 - 18.6s + 290.5) -------------------------------------------------------------------s (s+14.59) (s+6) (s+3) (s^2 + 26.82s + 228.4) (s^2 + 18.6s + 290.5) Type %OS 20 Percent Overshoot = 20, Damping Ratio = 0.45595, Phase Margin = 48.1477, Frequency = 0.74, Phase = -132.087, Magnitude = 0.0723422, Magnitude (dB) = -22.8122, K = 13.8232» Solutions to Problems 511 Second-order approximation not valid. 512 Chapter 11: Design via Frequency Response SOLUTIONS TO DESIGN PROBLEMS 23. a. Plot the Bode plot for K = 1. Using Eqs. (4.39) and (10.73) a percent overshoot = 10 is equivalent to a ζ = 0.591 and φM = 58.590. The phase-margin frequency = 1.933 rad/s where the phase is 58.590 – 1800 = -121.410. The magnitude = 38.37 dB, or 82.85. Hence K = 1/ 82.85 = 0.01207. b. Plot the gain-compensated Bode plot (K = 0.01207). Solutions to Design Problems 513 The bandwidth, ωBW, is the frequency at which the magnitude is –7dB. From the compensated plots, this frequency is 3.9 rad/s. Eq. (10.55), Ts = 2.01 s. Using Eq. (10.57), Tp = 1.16 s. c. Step Response 1 0.9 0.8 Amplitude 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 2.5 Time (sec.) 24. 10K G(s) = s(s+1)(s+5) . For Kv = 5, K = 2.5. Plot the Bode diagrams for this value of gain. 514 Chapter 11: Design via Frequency Response The uncompensated system has unity gain at ω = 2.04 rad/s. The phase is - 176.08o at this frequency yielding a phase margin of 3.92o. We want a 60o phase margin plus, after trial and error, a correction factor of 20o, or a total of 80o. Thus, the lead compensator must contribute 80o - 3.92o = 76.08o. Using Eqs. (11.11), and (11.12), The value of beta is: The |G(jwmax)| for the compensator is: or in db: 0.01490254 8.1916207 18.2673967 The magnitude curve has a gain of - 18.27 dB at ω = 5.27 rad/s. Therefore, choose this frequency as the new phase margin frequency. Using Eqs. (11.9) and (11.6), the compensator transfer function has the following specifications: T zero pole gain 1.55438723 -0.6433403 -43.169841 67.1026497 The compensated forward path is 25*67.1(s+0.64) 1677.5(s+0.64) G(s) = s(s+1)(s+5)(s+43.17) = s(s+1)(s+5)(s+43.17) 25. G(s) = 10 2 . Therefore, Kvo = 3 . We want Kvn = 30Kvo = 20. Increasing K by 30 s(s2+2s+5)(s+3) times yields G(s) = Plotting the Bode diagrams, 300 s(s2+2s+5)(s+3) Solutions to Design Problems 515 For 11% overshoot, the phase margin should be 57.48o. Adding a correction, we will use a 65o phase margin, or a phase angle of 115o, which occurs at ω = 0.58 rad/s. The magnitude curve is 30.93 dB. Thus the high-frequency asymptote of the lag compensator is - 30.93 dB. Drawing the lagcompensator curve as shown on the magnitude curve, the break frequencies are found and the compensator's transfer function is evaluated as s+0.058 Gc(s) = 25.86 x 10-3 s+0.0015 26. 4.514e-06K a. The equivalent forward transfer function is Ge(s) = s(s+0.04348) . Kv = 200 = 4.514e-06K 0.04348 or K = 1926500. Using Eq. (4.39), ζ = 0.456. Using Eq. (10.55), ωBW = 1.16. Using Eq. (10.73) with 15o additional, the required phase margin, φreq = 63.15o. Select a new phase-margin frequency, ωPm = 0.8 ωBW = 0.93. Plot the Bode plots for K = 1926500. 516 Chapter 11: Design via Frequency Response 150 Gain dB 100 50 0 -50 -3 10 10 -2 -1 10 Frequency (rad/sec) 10 0 10 1 Gain compensated Phase deg -60 -90 -120 -150 -180 10-3 10 -2 10-1 Frequency (rad/sec) 10 0 10 1 At ωPm = 0.93, the phase, φ = -177.3o. Hence, the phase required from the compensator, φC = φreq (180+φ) = 63.15 - (180 - 177.3) = 60.45o. Using Eq. (11.11), β = 0.07. Design lag: zclag = ωpm/10 = 0.093; pclag = zclag* β = 0.0065; Kclag = β = 0.07. Thus, s+0.093 Gclag(s) = 0.07 s+0.0065 . Design lead compensator: zclead=ωPm* β = 0.25; pclead=zclead/β = 3.57; Kclead=1/β = 14.29. Thus, s+0.25 Gclead(s) = 14.29 s+3.57 . Solutions to Design Problems 517 The lag-lead compensated Bode plot: Gain dB 200 100 0 -100 10-4 10 -3 10 -2 10 -1 Frequency (rad/sec) 10 0 10 1 102 10 0 10 1 102 Lag-lead Compensated Phase deg -60 -90 -120 -150 -180 10-4 10 -3 10 -2 10 -1 Frequency (rad/sec) b. Program: K=1926500; numg=4.514e-6; deng=[1 0.04348 0]; G=tf(numg,deng); numgclag=0.07*[1 0.093]; dengclag=[1 0.0065]; Gclag=tf(numgclag,dengclag); numgclead=14.29*[1 0.25]; dengclead=[1 3.57]; Gclead=tf(numgclead,dengclead); Ge=K*G*Gclag*Gclead; T=feedback(Ge,1); step(T) title('Lag-lead Compensated') 518 Chapter 11: Design via Frequency Response Computer response: 27. From Chapter 8, 0.6488K (s+53.85) _____ Ge(s) = 2 (s + 8.119s + 376.3) (s^2 + 15.47s + 9283) Cascading the notch filter, 0.6488K (s+53.85)(s2 + 16s + 9200) Get(s) = Since estep (∞ ) = (s2 + 8.119s + 376.3) (s^2 + 15.47s + 9283)(s+60)2 1 K * 53.85 * 9200 , Kp = 9 yields 10% error. Thus, K p = e = 9 . Thus, 376.3 * 9283 * 60 2 1 + Kp Ke = 0.6488K = 228452. Let us use Ke = 30,000 in designing the lead portion and we’ll make up the rest with the lag. Plotting the Bode plot for Ke = 30,000, Solutions to Design Problems 519 Design lead: The uncompensated phase margin = 10.290. Assume a required phase margin of 450. The required phase margin, assuming a 300 correction is 750. The phase contribution required from the compensator is 750 – 10.290 = 64.710. Using the inverse of Eq. (11.11), the compensator’s β = 0.05033. Using Eq. (11.12), Gc ( jω max ) = 1 β = 4.457 = 12.98 dB . The new phase margin frequency is where the uncompensated system has a magnitude of –12.98 dB, or ωmax = 44.65 rad/s. Using Eqs. (11.6) and (11.9), the compensator is Glead (s ) = 19.87( s + 10.02) . The plant is (s + 199) 228452(s + 53.85)( s2 + 16 s + 9200) G(s) = . Draw the lead-compensated 22 2 ( s + 60) ( s + 8.119s + 376.3)(s + 15.47s + 9283) Bode plot. 520 Chapter 11: Design via Frequency Response Design lag: The phase-margin frequency occurs where the phase is –1350, or at the required 450 phase margin. From the lead-compensated Bode plots, this phase margin occurs at 43.64 rad/s. Let the upper break of the lag compensator be one decade lower, or 4.364. Since the magnitude is 17.97 dB at the new phase-margin frequency, set the high-frequency asymptote of the lag compensator at – 17.97 dB. Draw a -20 dB/dec slope starting at 0.4364 rad/s and –17.96 dB and moving toward 0 dB. At 0 dB locate the lag compensator’s low-frequency break, or 0.551. Thus, Glag ( s) = 0.551 ( s + 4.364 ) (s + 4.364) = 0.126 (s + 0.551) 4.364 (s + 0.551) Check bandwidth: Draw the lag-lead compensated Bode diagram for G(s)Glag(s)Glead(s). Solutions to Design Problems 521 From the open-loop plot, the magnitude is at –7 dB at 70 rad/s. Hence, the bandwidth is sufficient. Also, the lag-lead compensated Bode plot shows a phase margin of 400 . The transfer function, G(s) = G(s)Glag(s)Glead(s) shows Kp = 9, or an error of 0.1. Thus all requirements have been met. TWELVE Design via State Space SOLUTION TO CASE STUDY CHALLENGE Antenna Control: Design of Controller and Observer a. We first draw the signal-flow diagram of the plant using the physical variables of the system as state variables. Writing the state equations for the physical variables shown in the signal-flow diagram, we obtain . z= 0 1 0 0 0 -1 . 3 2 0 0.8 -1 0 0 0 z+ 0 2000 u; y= 0.2 0 0 z The characteristic polynomial for this system is s3 + 101.32s2 + 132s + 0. Hence, the A and B matrices of the phase-variable form are Ax 0 0 0 1 0 -132 0 1 -101.32 Bx 0 0 1 Writing the controllability matrices and their determinants for both systems yields Solution to Case Study Challenge 523 CMz 0 0 2000 Controllability Matrix of z 0 1600 1600 -162112 -200000 20000000 Det(CMz) -5.12E+09 CMx 0 0 1 Controllability Matrix of x 0 1 1 -101.32 -101.32 10133.7424 Det(CMx) -1 where the system is controllable. Using Eq. (12.39), we find the transformation matrix and its inverse to be P 1600 0 0 Transformation Matrix z=Px 0 0 1600 0 2640 2000 PINV 0.000625 0 0 0 0.000625 -0.000825 0 0 0.0005 The characteristic polynomial of the phase-variable system with state feedback is s3 + (k3 + 101.32)s2 + (k2 + 132)s + (k1 + 0) For 15% overshoot, Ts = 2 seconds, and a third pole 10 times further from the imaginary axis than the dominant poles, the characteristic polynomial is (s + 20)(s2 + 4s + 14.969) = s3 + 24s2 + 94.969s + 299.38 Equating coefficients, the controller for the phase-variable system is Kx 299.38 Controller for x -37.031 -77.32 Using Eq. (12.42), the controller for the original system is Kz 0.1871125 Controller for z 0.04064463 -0.03866 b. Using Kz, gain from θm = - 0.1871125 (including gear train, pot, and operational amplifier); gain from tachometer = - 0.04064463; and gain from power amplifier output = 0.03866. 524 Chapter 12: Design via State Space c. Using the original system from part (a) and its characteristic polynomial, we find the observer canonical form which has the following A and C matrices: Ax -101.32 -132 0 Cx 1 1 0 0 0 1 0 0 0 Writing the observability matrices and their determinants for both systems yields OMz 0.2 0 0 Det(OMz) Observability Matrix of z 0 0 0.2 0 -0.264 0.16 0.0064 OMx 1 -101.32 10133.7424 Det(OMx) Observability Matrix of x 0 0 1 0 -101.32 1 1 where the system is observable. Using Eq. (12.89), we find the transformation matrix and its inverse to be P 5 -506.6 62500 Transformation Matrix z=Px 0 0 5 0 -625 6.25 PINV 0.20 20.26 26.40 0.00 0.20 20.00 The characteristic polynomial of the dual phase-variable system with state feedback is 0.00 0.00 0.16 Solution to Case Study Challenge 525 s3 + (l1 + 101.32)s2 + (l2 + 132)s + (l3 + 0) For 10% overshoot, ωn = 10 14.969 = 38.69 rad/s, and a third pole 10 times further from the imaginary axis than the dominant observer poles, the characteristic polynomial is (s + 228.72)(s2 + 45.743s + 1496.916) = s3 + 274.46s2 + 11959s + 3.4237x105 Equating coefficients, the observer for the observer canonical system is Lx 173.14 11827 342370 Observer for x Using Eq. (12.92), the observer for the original system is Lz 865.7 -28577.724 5569187.5 d. Observer for z 526 Chapter 12: Design via State Space e. Program: 'Controller' A=[0 1 0;0 -1.32 0.8;0 0 -100]; B=[0;0;2000]; C=[0.2 0 0]; D=0; pos=input('Type desired %OS '); Ts=input('Type desired settling time '); z=(-log(pos/100))/(sqrt(pi^2+log(pos/100)^2)); wn=4/(z*Ts); %Calculate required natural %frequency. [num,den]=ord2(wn,z); %Produce a second-order system that %meets the transient response %requirements. r=roots(den); %Use denominator to specify dominant %poles. poles=[r(1) r(2) 10*real(r(1))]; %Specify pole placement for all %poles. K=acker(A,B,poles) 'Observer' pos=input('Type desired %OS '); z=(-log(pos/100))/(sqrt(pi^2+log(pos/100)^2)); wn=10*wn %Calculate required natural %frequency. [num,den]=ord2(wn,z); %Produce a second-order system that %meets the transient response %requirements. r=roots(den); %Use denominator to specify dominant %poles. poles=[r(1) r(2) 10*real(r(1))];%Specify pole placement for all %poles. l=acker(A',C',poles)' Computer response: ans = Controller Type desired %OS 15 Type desired settling time K= 0.1871 0.0406 ans = Observer Type desired %OS wn = 38.6899 l= 1.0e+006 * 0.0009 -0.0286 5.5691 10 -0.0387 2 Answers to Review Questions 527 ANSWERS TO REVIEW QUESTIONS 1. Both dominant and non-dominant poles can be specified with state-space design techniques. 2. Feedback all state variables to the plant's input through a variable gain for each. Decide upon a closedloop characteristic equation that has a pole configuration to yield a desired response. Write the characteristic equation of the actual system. Match coefficients and solve for the values of the variable gains. 3. Phase-variable form 4. The control signal developed by the controller must be able to affect every state variable. 5. If the signal-flow diagram is in the parallel form, which leads to a diagonal system matrix, controllability can be determined by inspection by seeing that all state variables are fed by the control signal. 6. The system is controllable if the determinant of the controllability matrix is non-zero. 7. An observer is a system that estimates the state variables using information from the output of the actual plant. 8. If the plant's state-variables are not accessible, or too expensive to monitor 9. An observer is a copy of the plant. The difference between the plant's output and the observer's output is fed back to each of the derivatives of the observer's state variables through separate variable gains. 10. Dual phase-variable 11. The characteristic equation of the observer is derived and compared to a desired characteristic equation whose roots are poles that represent the desired transient response. The variable gains of each feedback path are evaluated to make the coefficients of the observer's characteristic equation equal the coefficients of the desired characteristic equation. 12. Typically, the transient response of the observer is designed to be much faster than that of the controller. Since the observer emulates the plant, we want the observer to estimate the plant's states rapidly. 13. Det[A-BK], where A is the system matrix, B is the input coupling matrix, and K is the controller. 14. Det[A-LC], where A is the system matrix, C is the output coupling matrix, and L is the observer. 15. The output signal of the system must be controlled by every state variable. 16. If the signal-flow diagram is in the parallel form, which leads to a diagonal system matrix, observability can be determined by inspection by seeing that all state variables feed the output. 17. The system is observable if the determinant of the observability matrix is non-zero. 528 Chapter 12: Design via State Space SOLUTIONS TO PROBLEMS 1. i. G ( s ) = ( s + 3) 1 =2 * ( s + 3) 2 ( s + 4) s + 8 s + 16 a. b. c. • 0 1 ⎡ ⎤ ⎡0 ⎤ x=⎢ ⎥ x + ⎢1 ⎥ r ; y = [3 1] x ⎣⎦ ⎣ −(k1 + 16) −(k2 + 8) ⎦ d. T ( s) = s+3 s + (k2 + 8) s + (k1 + 16) 2 Solutions to Problems 529 • x1 = (−20 − 71.25k1 ) x1 − 71.25k2 x2 − 71.25k3 x3 + 71.25r • x 2 = 27.5k1 x1 + (−10 x2 + 27.5k2 ) x2 + 27.5k3 x3 − 27.5r • x3 = −6.25k1 x1 − 6.25k2 x2 − 6.25k3 x3 + 6.25r −71.25k2 −71.25k3 ⎤ ⎡(−20 − 71.25k1 ) ⎡ 71.25 ⎤ ⎢ ⎥ ; B = ⎢ −27.5⎥ ; C = 1 1 1 A=⎢ 27.5k1 (−10 x2 + 27.5k2 ) 27.5k3 ⎥ [ ⎢ ⎥ ⎢ ⎢ 6.25 ⎥ −6.25k1 −6.25k2 −6.25k3 ⎥ ⎣ ⎦ ⎣ ⎦ 2 200( s + 7 s + 25) T ( s) = 3 4 s + (120 + 285k1 − 110k2 + 25k3 ) s 2 + (800 + 2850k1 − 2200k2 + 750k3 ) s + 5000k3 e. Part d. yields same result as i(d). ii. G ( s ) = a. b. s 1 =2 *s ( s + 5)( s + 7) s + 12s + 35 530 Chapter 12: Design via State Space c. • 0 1 ⎡ ⎤ ⎡0⎤ x=⎢ ⎥ x + ⎢1 ⎥ r ; y = [ 0 1] x ⎣⎦ ⎣ −(k1 + 35) −(k2 + 12) ⎦ d. T (s) = s s + (k2 + 12) s + (k1 + 35) 2 e. 0 1 ⎡ ⎤ ⎡0⎤ T ( s ) = C(sI − A) −1 B; A = ⎢ ; B = ⎢ ⎥ ; C = [ 0 1] ⎥ ⎣1 ⎦ ⎣ −(k1 + 35) −(k2 + 12) ⎦ which yields the same result as ii(d). iii. G ( s ) = a. b. 20s ( s + 7) 1 =3 * (20s 2 +140s) 2 ( s + 3)( s + 7)( s + 9) s + 19s + 111s + 189 Solutions to Problems 531 c. 0 1 0 ⎡ ⎤ ⎡0 ⎤ ⎢ ⎥ x + ⎢0 ⎥ r ; y = 0 140 20 x x=⎢ 0 0 1 [ ⎥ ⎢⎥ ⎢ −(k1 + 189) −(k2 + 111) −(k3 + 19) ⎥ ⎢1 ⎥ ⎣ ⎦ ⎣⎦ • d. T ( s) = 20 s ( s + 7) s + (k3 + 19) s + (k2 + 111) s + (k1 + 189) 3 2 e. 0 1 0 ⎡ ⎤ ⎡0⎤ ⎢ ⎥ ; B = ⎢0⎥ ; C = 0 140 20 T ( s) = C(sI − A) B; A = ⎢ 0 0 1 [ ⎥ ⎢⎥ ⎢ −(k1 + 189) −(k2 + 111) −(k3 + 19) ⎥ ⎢1 ⎥ ⎣ ⎦ ⎣⎦ −1 which yields the same result as iii(d). iv. a. G ( s) = 30( s + 2)( s + 3) 1 =3 * (30s 2 + 150 s + 180) 2 ( s + 4)( s + 5)( s + 6) s + 15s + 74s + 120 532 Chapter 12: Design via State Space b. c. 0 1 0 ⎡ ⎤ ⎡0 ⎤ ⎢ ⎥ x + ⎢0 ⎥ r ; y = 180 150 30 x x=⎢ 0 0 1 [ ⎥ ⎢⎥ ⎢ −(k1 + 120) −(k2 + 74) −(k3 + 15) ⎥ ⎢1 ⎥ ⎣ ⎦ ⎣⎦ • d. 30 s 2 + 150s + 180 T (s) = 3 s + (k3 + 15) s 2 + (k2 + 74) s + (k1 + 120) e. 0 1 0 ⎡ ⎤ ⎡0⎤ ⎢ ⎥ ; B = ⎢ 0⎥ ; C = 180 150 30 T ( s ) = C(sI − A) B; A = ⎢ 0 0 1 [ ⎥ ⎢⎥ ⎢ −(k1 + 120) −(k2 + 74) −(k3 + 15) ⎥ ⎢1 ⎥ ⎣ ⎦ ⎣⎦ −1 which yields the same result as iv(d). v. G ( s) = s 2 + 8s + 15 1 =4 * (s 2 + 8s + 15) 2 2 3 (s + 4 s + 10)(s + 3s + 12) s + 7 s + 34s 2 + 78s + 120 Solutions to Problems 533 a. b. c. 0 1 0 0 ⎡ ⎤ ⎡0 ⎤ ⎢ ⎥ ⎢⎥ • 0 0 1 0 ⎢ ⎥ x + ⎢0 ⎥ r ; y = [15 8 1 0] x x= ⎢ ⎥ ⎢0 ⎥ 0 0 0 1 ⎢ ⎥ ⎢⎥ ⎢ −(k1 + 120) −(k2 + 78) −(k3 + 34) −(k4 + 7) ⎥ ⎣1 ⎦ ⎣ ⎦ d. s 2 + 8s + 15 T ( s) = 4 s + (k4 + 7) s 3 + (k3 + 34) s 2 + (k2 + 78) s + (k1 + 120) 534 Chapter 12: Design via State Space e. 0 1 0 0 ⎡ ⎤ ⎡0 ⎤ ⎢ ⎥ ⎢⎥ 0 0 1 0 ⎥ ; B = ⎢0 ⎥ ; C = [15 8 1 0] T ( s) = C(sI − A) −1 B; A = ⎢ ⎢ ⎥ ⎢0 ⎥ 0 0 0 1 ⎢ ⎥ ⎢⎥ ⎢ −(k1 + 120) −(k2 + 78) −(k3 + 34) −(k4 + 7) ⎥ ⎣1 ⎦ ⎣ ⎦ which yields the same result as v(d). 2. i a. The output is Since, y = (30s 2 + 270 s + 420) x1 = 30 x1 + 270 x1 + 420 x1 = 30 x2 + 270 x2 + 420 x1 = 30(−5 x2 + x3 ) + 270 x2 + 420 x1 = 420 x1 + 120 x2 + 30 x3 b. 1 ⎡0 ⎢0 −5 T ( s ) = C(sI − A) B; A = ⎢ ⎢ − k1 − k2 ⎣ −1 T ( s) = ⎤ ⎡0⎤ ⎥ ; B = ⎢0 ⎥ ; C = 420 120 30 1 [ ⎥ ⎢⎥ ⎢1 ⎥ −(k3 + 3) ⎥ ⎦ ⎣⎦ 0 30( s + 2)( s + 7) s (k3 + 8) s 2 + (5k3 + k2 + 15) s + k1 3 Solutions to Problems 535 ii a. 1 u 1 5 r 1 s x3 1 s 1 -2 1 s 3 x 1 7 -10 -k 2 -k 1 ⎡0 T(s) = C(sI − A) B; A = ⎢ 0 ⎢−k ⎣ −1 1 = 2 1 -2 -k 3 b. x 1 0 −k 2 0 ⎤ ⎡0 ⎤ ⎥; B = ⎢ 0 ⎥; C = [-3 1 1] 1 ⎢5 ⎥ − (5k 3 + 2)⎥ ⎣⎦ ⎦ 5(s2 + 3s + 7) s3 + (5k 3 + 4)s2 + (10k 3 + k 2 + 14)s + (50k 3 + k 1 + 20) 3. i a. 50( s 2 + 7 s + 25) 6.25 27.5 71.25 = − + G ( s) = s( s + 10)( s + 20) s s + 10 s + 20 y 536 Chapter 12: Design via State Space b. Writing the state equations: • x1 = −20 x1 + 71.25u • x 2 = −10 x2 − 27.5u • x3 = 6.25u But, u = -k1x1 - k2x2 - k3x3 + r . Substituting into the state equations, • x1 = (−20 − 71.25k1 ) x1 − 71.25k2 x2 − 71.25k3 x3 + 71.25r • x 2 = 27.5k1 x1 + (−10 x2 + 27.5k2 ) x2 + 27.5k3 x3 − 27.5r • x3 = −6.25k1 x1 − 6.25k2 x2 − 6.25k3 x3 + 6.25r Therefore, T(s) = C(sI - A)-1B, where −71.25k2 −71.25k3 ⎤ ⎡(−20 − 71.25k1 ) ⎡ 71.25 ⎤ ⎢ ⎥ ; B = ⎢ −27.5⎥ ; C = 1 1 1 27.5k1 (−10 x2 + 27.5k2 ) 27.5k3 ⎥ A=⎢ [ ⎢ ⎥ ⎢ ⎢ 6.25 ⎥ −6.25k1 −6.25k2 −6.25k3 ⎥ ⎣ ⎦ ⎣ ⎦ Hence, T ( s) = 200( s 2 + 7 s + 25) 4 s 3 + (120 + 285k1 − 110k2 + 25k3 ) s 2 + (800 + 2850k1 − 2200k2 + 750k3 ) s + 5000k3 Solutions to Problems 537 ii a. G(s) = 50(s + 3)( s + 4) 50 300 300 = − + ( s + 5)(s + 6)(s + 7) s + 5 s + 6 s + 7 x1 400 950 -30 -5 1 -k 1 x2 1 100 -300 9 1 -15 -6 -k 2 400 300 9 -k 1 x 3 3 -7 b. Writing the state equations: • x1 = −5 x1 + 50 u • x2 = −6 x 2 − 300u • x3 = − 7 x3 + 300u But, u = − k1 x1 − k2 x 2 − k 3 x3 + r Substituting into the state equations, collecting terms, and converting to vector-matrix form yields −50 k2 − 50k 3 ⎤ ⎡ −(5 + k1 ) ⎡ 50 ⎤ x = ⎢ 300 k1 (300 k2 − 6) 300k3 ⎥ x + ⎢ −300 ⎥r ⎢− (300 k1 + 7) ⎢ 300 ⎥ −300 k2 300k3 ⎥ ⎣ ⎦ ⎣ ⎦ • y = [1 1 1]x Therefore, T(s) = C(sI - A)-1B, or T (s ) = 50 s2 + 1750 s + (6900 − 88200 k1 ) s3 + ( 300k3 − 300k 2 + k1 + 11)s2 + 2(1475k3 − 750k 2 + 3k1 − 7350 k3 k1 + 7350 k2 k1 + 15) s + 300k3 (23 − 294k1 ) 4. The plant is given by 538 Chapter 12: Design via State Space G ( s) = 20 20 =3 2 ( s + 1)( s + 3)( s + 7) s + 11s + 31s + 21 The characteristic polynomial for the plant with phase-variable state feedback is s 3 + ( k3 + 11) s 2 + ( k2 + 31) s + ( k1 + 21) = 0 The desired characteristic equation is ( s + 53.33)( s 2 + 10.67 s + 106.45) = s 3 + 64s 2 + 675.48s + 5676.98 based upon 10% overshoot, Ts = 0.5 second, and a third pole ten times further from the imaginary axis than the dominant poles. Comparing the two characteristic equations, k1 = 5655.98, k2 = 644.48, and k3 = 53 . 5. a. The system in controller canonical form is: ⎡ − a n − 1 − a n− 2 ⎢1 0 A=⎢ ⎢ ⎢0 0 ⎣ − a1 − a 0 ⎤ ⎡1 ⎤ ⎢0 ⎥ ⎥ 0 0 ⎥;B = ⎢ ⎥;C = c c 1 2 c3 c4 ⎢0 ⎥ ⎥ ⎢0 ⎥ 1 0⎥ ⎣⎦ ⎦ [ The characteristic equation of the plant is: sn + an-1sn-1 + . . . + a1s + a0 = 0 Forming the closed-loop system by feeding back each state variable and the input to u forming u = -Kx + r where K = [k1 k2 . . . kn] and substituting u into the state equation, we obtain . x = Ax + B u = ( A − BK )x + B r Forming A - BK: ⎡ − (a n − 1+ k1 ) ⎢ 1 A − BK = ⎢ ⎢ 0 ⎣ −(a n− 2 0 0 + k2 ) − (a + k n − 1 ) − ( a + k n ) ⎤ 1 0 ⎥ 0 0 ⎥ ⎥ 1 0 ⎦ The characteristic equation is: sn + (an-1 + k1)sn-1 + (an-2 + k2). . . + (a1 + kn-1)s + (a0 + kn) = 0 Assuming a desired characteristic equation, sn + dn-1sn-1 + dn-2sn-2+ . . . + d2s2 + d1s + d0 = 0 Equating coefficients, Solutions to Problems 539 di = ai + kn-i ; i = 0, 1, 2, ... n-1 from which kn-i = di – ai (1) b. The desired characteristic equation is s3 + 15.9 s2 + 136.08s + 413.1 = 0 the characteristic equation of the plant is s3 + 5s 2 + 4 s + 0 = 0 Using Eq. (1) above, k3-i = di – ai. Therefore, k3 = d0 – a0 = 413.1 – 0 = 413.1; k2 = d1 – a1 = 136.08 – 4 = 132.08; k1 = d2 – a2 = 15.9 – 5 = 10.9. Hence, K = [10.9 132.08 413.1] 6. Using Eqs. (4.39) and (4.34) to find ζ = 0.5169 and ωn = 7.3399, respectively. Factoring the denominator of Eq. (4.22), the required poles are -3.7942 ± j6.2832. We place the third pole at -2 to cancel the open loop zero. Multiplying the three closed-loop pole terms yields the desired characteristic equation: 100s2 + 2200s + 4000 , the controller s3 + 9.5885s2 + 69.0516s + 107.7493 = 0. Since G(s) = 3 s + 8 s2 + 19 s + 12 ⎡− 8 − 19 − 12⎤ ⎡1 ⎤ ⎢ ⎢⎥ ⎥ 0 0 ⎥ ; B = ⎢0 ⎥ ; C = [100 2200 4000]. The first row of A canonical form is A = ⎢ 1 ⎢0 ⎢0 ⎦ ⎥ 1 0⎦ ⎣ ⎣⎥ contains the coefficients of the characteristic equation. Thus comparing the first row of A to the desired characteristic equation and using the results of Problem 5, k1 = -(9.5885 - 8) = 1.5885; k2 = -(69.0516 - 19) = 50.0516; and k3 = -(107.7493 - 12) = 95.7493. 7. The plant is given by G(s) = 20( s + 2) 20 s + 40 =3 s( s + 4)( s + 6) s + 10s 2 + 24 s + 0 The characteristic polynomial for the plant with phase-variable state feedback is s 3 + ( k3 + 10) s 2 + ( k 2 + 24) s + ( k3 + 0) The desired characteristic equation is ( s + 20)( s 2 + 4s + 11.45) = s 3 + 24 s 2 + 91.45s + 229 based upon 10% overshoot, Ts =2 seconds, and a third pole ten times further from the imaginary axis than the dominant poles. Comparing the two characteristic equations, k1 = 229, k2 = 67.45, and k 3 = 14 . 540 Chapter 12: Design via State Space 8. Drawing the signal-flow diagram, 1 r u 1 1 s x3 1 -2 1 s x2 1 1 s x 1 10 y -4 -k 3 -k 2 -k 1 Writing the state equations yields the following A matrix: from which, The desired characteristic equation is (s + 80)(s2 + 16s + 183.137) = s3 + 96s2 + 1463.1s + 14651 based upon 10% overshoot, Ts = 0.5 second, and a third pole ten times further from the imaginary axis than the dominant poles. Comparing the two characteristic equations, , k1 = 14651, k2 = 1095.1, and k3 = 90. 9. Expand G(s) by partial fractions and obtain G(s) = 20 1.67 5 6.67 = + − s( s + 4)( s + 6) s s+4 s+6 Drawing the signal-flow diagram with state feedback Solutions to Problems 541 Writing the state equations yields the following system matrix: −1.67k3 ⎤ ⎡ −1.67k1 −1.67k2 A = ⎢ −5k1 −(5k2 + 4) 5k3 ⎥ ⎢ ⎥ ⎢ 6.67k1 6.67k2 (6.67k3 − 6) ⎥ ⎣ ⎦ Evaluating the characteristic polynomial yields, sI − A = (−6.67k3 + 5k2 + 1.67 k1 + 10) s 2 + (−26.68k3 + 30k2 + 16.7 k1 + 24) s + 40.08k1 From Problem 7, the desired characteristic polynomial is s 3 + 24 s 2 + 91.45s + 229 . Equating coefficients and solving simultaneously yields k1 = 5.71, k2 = −4.58, and k 3 = −4.10 . 10. Writing the state equation and the controllability matrix for the system yields . x= -5 -1 b 1 -3 x+ b b 1 u;C 2 M = B AB = b 1 2 -5b +b 1 -b -3b 1 2 2 The controllability matrix has a zero determinant if b2 = b1. 11. The controllability matrix is given by Eq. (12.26) for each of the following solutions: a. -2 0 1 0 -2 0 0 A= 0 -3 ;B= 0 1 ;C 1 M 0 1 1 -2 -5 4 1 = -3 9 ; det C M = 0 ; s y st em i s u n co n t ro ll ab le b. -2 0 0 0 0 A= 1 -2 0 -3 ;B= 0 1 ;C 1 M 1 -2 -4 4 1 = 0 1 -3 9 ; det C M = - 1 ; s y st em i s co n t ro ll ab le c. -4 A= 1 0 0 0 1 0 0 -3 ;B= 0 2 1 ;C M = 0 2 2 1 -7 -3 1 -3 9 ; det C M = 7 ; s yst em i s cont ro llabl e 542 Chapter 12: Design via State Space d. -4 0 1 0 0 1 -5 A= 0 -3 1 0 ;B= ;C 1 M 1 0 -4 1 17 -8 1 = -8 44 1 -2 -2 4 ; de t C -4 0 -6 16 0 36 ; det C M = -5 ; s y s t em i s co n t ro l l ab l e e. 0 -6 A= 1 -5 1 -2 ;B= ;C M = M = 0 ; s y s t em i s u n co n t ro l l ab l e f. A= -4 0 0 0 -5 0 0 0 -6 ;B= 1 0 1 ;C M = 1 0 1 ; det C This system can also be determined uncontrollable by inspection. 12. Program: '(d)' A=[-4 1 0;0 0 1;-5 0 -3] B=[1;0;1] Cm=ctrb(A,B) Rank=rank(Cm) pause '(f)' A=[-4 0 0;0 -5 0;0 0 -6] B=[1;0;1] Cm=ctrb(A,B) Rank=rank(Cm) Computer response: ans = (d) A= -4 0 -5 1 0 0 0 1 -3 -4 1 -8 17 -8 44 B= 1 0 1 Cm = 1 0 1 Rank = 3 M =0 ; s yst em i s u nco n trollable Solutions to Problems 543 ans = (f) A= -4 0 0 0 -5 0 0 0 -6 -4 0 -6 16 0 36 B= 1 0 1 Cm = 1 0 1 Rank = 2 13. From Eq. (12.46) we write the controller canonical form: A cc ⎡− 8 −17 −10 ⎤ ⎡1 ⎤ ⎥ ⎢ ⎢⎥ 0 0 ⎥ ; B cc = ⎢0 ⎥ . =⎢1 ⎢0 ⎢0 ⎦ 1 0⎥ ⎦ ⎣ ⎣⎥ The controllability matrices are found using Eq. (12.35). For the original system of Eq. (12.44), C Mz ⎡0 ⎢ = ⎢0 ⎢1 ⎣ ⎡1 −8 47⎤ 0 1⎤ ⎢ ⎥ ⎥ 1 − 3⎥ . For the controller canonical form, C Mcc = ⎢0 1 −8⎥ . The transformation ⎢0 0 1 ⎥ −1 1 ⎥ ⎣ ⎦ ⎦ ⎡0 0 1 ⎤ ⎥ 1 5 ⎥ . Comparing the first row of Acc with the desired ⎢ 1 7 10 ⎦ ⎥ ⎣ ⎢ matrix is, P = CMzCMcc-1 = ⎢0 characteristic equation, Eq. (12.50), Kcc = [ -2 Kz = KccP-1 = [-20 10 -4 10]. Transforming back to the original system, -2]. 14. Drawing the signal-flow diagram for the plant in cascade form yields u 1 s 1 z3 1 -10 1 s z2 1 1 s 1 z1 6 -3 -8 Writing the A and B matrices for the z system, Az -3 1 0 Bz 0 y 544 Chapter 12: Design via State Space 0 0 -8 0 1 -10 0 1 Writing the A and B matrices for the x (phase-variable) system, Ax 0 0 -240 1 0 -134 Bx 0 0 1 0 1 -21 Phase-Variable Form From the phase variable from, the characteristic polynomial is s3 + 21s2 + 134s + 240. Finding the controllability matrices and their determinants for the z and x systems shows that there is controllability, CMz Controllability Matrix of z Controllability Matrix of x CMx 0 0 1 0 0 1 0 1 -18 0 1 -21 1 -10 100 1 -21 307 Det(CMz) -1 Det(CMx) -1 Using Eq. (12.39), the transformation matrix P and its inverse are found to be Transformation Matrix z=Px P PINV 1 0 0 1.00 0.00 0.00 3 1 0 -3.00 1.00 0.00 24 11 1 9.00 -11.00 1.00 Using the given transient requirements, and placing the third closed-loop pole over the zero at -6 yields the following desired closed-loop characteristic polynomial: (s2 + 8s + 45.78)(s + 6) = s3 + 14s2 + 93.78s + 274.68 Using the phase-variable system with state feedback the characteristic polynomial is s3 + (k3 + 21)s2 + (k2 + 134)s + (k1 + 240) Equating the two characteristic polynomials yields the state feedback vector for the x system as Kx Controller for x 34.68 -40.22 -7 Using Eq. (12.42), Kz Controller for z 92.34 36.78 -7 15. Program: A=[-3 1 0;0 -8 1;0 0 -10]; B=[0;0;1]; %Generate system matrix A %Generate input coupling matrix B Solutions to Problems 545 C=[3 1 0]; %Generate output coupling matrix C D=0; %Generate matrix D Po=10; %Input desired percent overshoot Ts=1; %Input desired settling time z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); %Calculate required damping ratio wn=4/(z*Ts); %Calculate required natural %frequency [num,den]=ord2(wn,z); %Produce a second-order system that %meets transient requirements r=roots(den); %Use denominator to specify %dominant poles poles=[r(1) r(2) -6]; %Specify pole placement for all %poles. %A few tries at the the third-pole %value shows T(s) with a closed%loop zero at -7. %Thus, choose the third pole to %cancel this zero. K=acker(A,B,poles) %Calculate controller gains in z%system Anew=A-B*K; %Form compensated A matrix Bnew=B; %Form compensated B matrix Cnew=C; %Form compensated C matrix Dnew=D; %Form compensated D matrix [numt,dent]=ss2tf(Anew,Bnew,Cnew,Dnew); %Form T(s) 'T(s)' %Display label T=tf(numt,dent) %Display T(s) poles=pole(T) %Display poles of T(s) Computer response: K= 92.3531 36.7844 -7.0000 ans = T(s) Transfer function: -3.553e-015 s^2 + s + 6 -----------------------------s^3 + 14 s^2 + 93.78 s + 274.7 poles = -4.0000 + 5.4575i -4.0000 - 5.4575i -6.0000 16. Expanding by partial fractions, G (s ) = ( s + 6) 0.085714 0.2 0.28571 = − − ( s + 3)(s + 8)(s + 10) (s + 3) (s + 8) (s + 10 ) Writing the A and B matrices for the z system with ki's set to zero, Az -3 0 0 Bz 0.085714 546 Chapter 12: Design via State Space 0 0 -8 0 0 -10 0.2 -0.28571 Writing the A and B matrices for the x (phase-variable) system, Ax 0 0 -240 1 0 -134 Bx 0 0 1 0 1 -21 Phase-Variable Form From the phase variable from, the characteristic polynomial is s3 + 21s2 + 134s + 240. Finding the controllability matrices and their determinants for the z and x systems shows that there is controllability, CMz Controllability Matrix of z CMx Controllability Matrix of x 0.085714 -0.257142 0.771426 0 0 1 0.2 -1.6 12.8 0 1 -21 -0.28571 2.8571 -28.571 1 -21 307 Det(CMz) 0.342850857 -1 Det(CMx) Using Eq. (12.39), the transformation matrix P and its inverse are found to be P 6.85712 6 -6.85704 Transformation Matrix z=Px 1.542852 0.085714 2.6 0.2 -3.14281 -0.28571 PINV 0.33 -1.00 3.00 -0.50 4.00 -32.00 -0.25 2.50 -25.00 Using the given transient requirements, and placing the third closed-loop pole over the zero at -6 yields the following desired closed-loop characteristic polynomial: (s2 + 8s + 45.78)(s + 6) = s3 + 14s2 + 93.78s + 274.68 Using the phase-variable system with state feedback the characteristic polynomial is s3 + (k3 + 21)s2 + (k2 + 134)s + (k1 + 240) Equating the two characteristic polynomials yields the state feedback vector for the x system as Kx 34.7062 Controller for x -40.2156 -7 Using Eq. (12.42), Kz 30.78443595 Controller for z 45.7845 65.78543678 Solutions to Problems 547 17. Draw signal-flow diagram showing state variables, z, at the output of each subsystem and the state variables, w, at the output of the integrators. u 1 1 s z3 -3 1 1 s 1 z1 6 1 s z2 1 w2 w3 -8 -10 z1 1 y w1 Recognizing that z2 = 6 w2 − 8w2 + w3 = − 2w2 + w3 , we can write the state equations for w as ⎡ − 10 − 2 1 ⎤ ⎡ 0⎤ − 8 1 ⎥ w + ⎢ 0⎥ u w=⎢ 0 ⎢0 ⎢1 ⎥ 0 − 3⎥ ⎦ ⎣ ⎣⎦ • y = [1 0 0]w Writing the relationship between z and w yileds 0⎤ 1⎥ w = P −1w 1⎥ ⎦ ⎡1 z = ⎢0 ⎢0 ⎣ 0 −2 ⎡1 P = ⎢0 ⎢0 ⎣ 0 0⎤ −0.5 0.5⎥ 0 1⎥ ⎦ 0 Thus Converting the state equations in w to state equations in z, we use Eqs. (5.87) and obtain the A matrix and B vector as Az -10 0 0 1 -8 0 Bz 0 1 1 0 3 -3 Writing the A and B matrices for the x (phase-variable) system, Ax 0 0 -240 1 0 -134 0 1 -21 Bx 0 0 1 Phase-Variable Form From the phase variable from, the characteristic polynomial is s3 + 21s2 + 134s + 240 Finding the controllability matrices and their determinants for the z and x systems shows that there is controllability, 548 Chapter 12: Design via State Space CMz 0 1 1 Controllability Matrix of z 1 -5 -3 -15 31 9 -8 Det(CMz) CMx 0 0 1 Controllability Matrix of x 0 1 -21 1 -21 307 -1 Det(CMx) Using Eq. (12.39), the transformation matrix P and its inverse are found to be Transformation Matrix z=Px 1 0 16 1 18 1 P 6 60 80 PINV -0.25 2.50 -25.00 -0.13 0.75 -3.50 0.13 -0.75 4.50 Using the given transient requirements, and placing the third closed-loop pole over the zero at -6 yields the following desired closed-loop characteristic polynomial: (s2 + 8s + 45.78)(s + 6) = s3 + 14s2 + 93.78s + 274.68 Using the phase-variable system with state feedback the characteristic polynomial is s3 + (k3 + 21)s2 + (k2 + 134)s + (k1 + 240) Equating the two characteristic polynomials yields the state feedback vector for the x system as Kx 34.68 Controller for x -40.22 -7 Using Eq. (12.42), Kz 65.78 18. Controller for z -10 3 Using Eqs. (4.39) and (4.34) to find ζ = 0.5169 and ωn = 18.3498 respectively. Factoring the denominator of Eq. (4.22), the required poles are -9.4856 ± j15.708. We place the third pole 10 times further at -94.856. Multiplying the three closed-loop pole terms yields the desired characteristic equation: s3 + 114s2 + 2136s + 31940 = 0. Representing the plant in parallel form: Apar 0⎤ ⎡ 3.125⎤ ⎡0 0 = ⎢ 0 − 4 0 ⎥ ; Bpar = ⎢ −6.25⎥ ; Cpar = [1 1 1]. Using Eq. (12.26), ⎢ 3.125⎦ ⎢ 0 0 − 8⎥ ⎥ ⎦ ⎣ ⎣ C Mpar 0 0⎤ ⎡ 3.125 = ⎢ − 6.25 25 − 100⎥ , which is controllable since the determinant is 7812.5. Since ⎢ 3.125 −25 200 ⎦ ⎥ ⎣ ⎡ −12 100 , the controller canonical form is A cc = ⎢ 1 G (s ) = 3 s + 12s 2 + 32s ⎢0 ⎣ −32 0 1 0⎤ 0⎥ ; Bcc 0⎥ ⎦ ⎡1 ⎤ ⎢⎥ = ⎢0 ⎥ ; ⎢⎥ ⎣0 ⎦ Solutions to Problems 549 −12 1 ⎡1 C = [0 0 100 ]. Using Eq. (12.26), CMcc = ⎢0 ⎢0 ⎣ 0 112 ⎤ −12 ⎥ , which is controllable since the ⎥ 1⎦ determinant is 1. The first row of Acc contains the coefficients of the characteristic equation. Comparing the first row of Acc to the desired characteristic equation and using the results of Problem 5, (12 + k1) = 114; (32 + k2)= 2136; and (0 + k3)= 31940. Hence Kcc =[31940 2104 102]. The ⎡100 37.5 −50 transformation matrix is, P = CMparCMcc-1 = ⎢ 0 ⎢ 0 12.5 ⎣ 3.125⎤ −6.25⎥ . Transforming back to the ⎥ 3.125⎦ original system, Kpar = KccP-1 = [319.396 251.5184 216.2255]. 19. G(s) = 1 1 =3 2 s(s + 3)( s + 7) s + 10 s + 21s + 0 Writing the A and C matrices for the observer canonical system, Az -10 -21 0 1 0 0 Cz 1 0 1 0 0 0 The characteristic polynomial is s3 + 10s2 + 21s + 0. Now check observability by calculating the observability matrix and its determinant. OMz 1 -10 79 Observability Matrix of z 0 1 -10 Det(OMz) 0 0 1 1 Using the given transient requirements, and placing the third closed-loop pole 10 times further from the imaginary axis than the dominant poles yields the following desired characteristic polynomial: (s + 300)(s 2 + 60s + 5625) = s 3 + 360s 2 + 23625 s + 1687500 Equating this polynomial to Eq. (12.67), yields the observer gains as: Lz 350 23604 1687500 Observer for z 550 Chapter 12: Design via State Space 20. Using Eqs. (4.39) and (4.34) to find ζ = 0.5912 and ωn = 19.4753 respectively. Factoring the denominator of Eq. (4.22), the required poles are -11.513 ± j15.708. We place the third pole 20 times further at –230.26. Multiplying the three closed-loop pole terms yields the desired characteristic equation: s3 + 253.28s2 + 5681.19s + 87334.19 = 0. ⎡ − 20 1 0 ⎤ ⎡0⎤ Representing the plant in observer canonical form: A oc = ⎢ − 108 0 1 ⎥ ; Boc = ⎢ 0 ⎥ ; ⎢ − 144 0 0 ⎥ ⎢10 ⎥ ⎦ ⎣ ⎣⎦ Coc = [1 0 0]. The first column of Aoc contains the coefficients of the characteristic equation. Comparing the first column of Aoc to the desired characteristic equation and using Eq. (12.67), l1 = 253.28 - 20 = 233.28; l2 = 5681.19 - 108 = 5573.19; and l3 = 87334.19 - 144 = 87190.19. Hence, Loc = [233.28 5573.19 87190.19]T. 21. The A, L, and C matrices for the phase-variable system are: 0⎤ ⎡0 1 A = ⎢0 0 1⎥ ⎢ 0 − 21 − 10⎦ ⎥ ⎣ C = [1 0 0 ] ⎡ l1 ⎤ L = ⎢ l2 ⎥ ⎢l ⎦ ⎣⎥ 3 Hence, ⎡ λ + l1 λ − ( A − LC) = ⎢ l2 ⎢l ⎣ 3 or ⎤ λ −1 ⎥ 21 λ + 10⎥ ⎦ −1 0 λ − ( A − LC) = λ3 + (10 + l1 )λ2 + (21 + 10l1 + l2 )λ + (21l1 + 10l2 + l3 ) From Problem 19, the desired characteristic polynomial is λ3 + 360λ2 + 23625λ + 1687500. Equating coefficients yields: 10 + l1 = 360; ( 21 + 10 l1 + l2 ) = 23625; (21l1 + 10l2 + l3 ) = 1687500 Solving successively, l1 = 350; l2 = 20104; l3 = 1479110 22. The A, L, and C matrices for the phase-variable system are: Solutions to Problems 551 1⎤ ⎡0 ⎡ l1 ⎤ ; C = [2 1]; L = ⎢ ⎥ A=⎢ ⎦ ⎣ −45 − 14⎥ ⎣l2 ⎦ Hence, l1 − 1 ⎤ ⎡ λ + 2l1 λ − ( A − LC ) = ⎢ ⎦ ⎣ 2l2 + 45 l2 + λ + 14⎥ or λ 2 + (2l1 + l2 + 14 )λ + (2l2 − 17l1 + 45) From the problem statement, the desired characteristic polynomial is λ2 + 144λ + 14400. Equating coefficients yields, (2l1 + l2 + 14) = 144; (2l2 − 17l1 + 45) = 14400 Solving simultaneously, l1 = −671.2; l2 = 1472.4 23. The A matrix for each part is given in the solution to Problem 11. Each observability matrix is calculated from Eq. (12.79). a. b. c. d. e. f. 24. Program: '(a)' A=[-2 0 1;0 -2 0;0 0 -3] C=[5 5 5] Om=obsv(A,C) Rank=rank(Om) '(f)' A=[-4 0 0;0 -5 0;0 0 -6] %Form compensated A matrix %Form compensated C matrix %Form observability matrix %Find rank of observability %matrix %Form compensated A matrix 552 Chapter 12: Design via State Space C=[1 1 1] Om=obsv(A,C) Rank=rank(Om) %Form compensated C matrix %Form observability matrix %Find rank of observability Computer response: ans = (a) A= -2 0 0 0 -2 0 1 0 -3 5 5 5 5 -10 20 5 -10 20 5 -10 20 -4 0 0 0 -5 0 0 0 -6 1 1 1 1 -4 16 1 -5 25 1 -6 36 C= Om = Rank = 1 ans = (f) A= C= Om = Rank = 3 25. Representing the system in state space yields . 0 -1 = x 1 -2 x+ 0 1 u; y= c 1 c 2 x Using Eq. (12.79), c O M = 1 -c 2 c 2 (c -2c ) 1 2 Thus, the system is unobservable if c1 = c2. and det O M 2 2 = c - 2c c + c 1 12 2 Solutions to Problems 553 26. The A and C matrices for the system represented in cascade form is Az -20 0 0 1 -13 0 Cz 1 0 1 -5 0 0 The characteristic polynomial found from the transfer function of the plant is s3 + 38s2 + 425s + 1300 From this characteristic polynomial, we can write observer canonical form of the state equations. The A and C matrices of the observer canonical form are given below as Ax -38 -425 -1300 1 0 0 0 1 0 Cx 1 0 0 To test observability, we write the observability matrices for both systems and show that both observability matrices have non zero determinants. Using Eq. (12.79), OMz 1 -20 400 Det(OMz) Observability Matrix of z 0 1 -33 0 0 1 1 OMx 1 -38 1019 Observability Matrix of x 0 1 -38 0 0 1 1 Det(OMx) Using Eq. (12.89), we obtain the transformation matrix, P, and its inverse as P 1 -18 25 Transformation Matrix z=Px 0 0 1 0 -5 1 PINV 1.00 18.00 65.00 0.00 1.00 5.00 Using the characteristic polynomial given in the problem statement, the plant’s characteristic equation, and Eq. (12.67), the observer for the observer canonical system is Lx 562 39575 1498700 Observer for x 0.00 0.00 1.00 554 Chapter 12: Design via State Space Using Eq. (12.92), the observer for the cascade system is found to be Lz 562 29459 1314875 Observer for z 27. Program: A=[-20 1 0;0 -13 1;0 0 -5] B=[0;0;1] C=[1 0 0] D=0 poles=roots([1 600 40000 1500000]) L=acker(A',C',poles); 'L' L' Computer response: A= -20 0 0 1 -13 0 0 1 -5 0 0 B= 0 0 1 C= 1 D= 0 poles = 1.0e+002 * -5.2985 -0.3508 + 0.4001i -0.3508 - 0.4001i ans = L ans = 1.0e+006 * 0.0006 0.0295 1.3149 Solutions to Problems 555 28. Expanding the plant by partial fractions, we obtain G(s) = 1 0.008333 0.017857 0.0095238 = − + ( s + 5)(s + 13)(s + 20 ) (s + 5) (s + 13) ( s + 20) The A and C matrices for the system represented in parallel form is Az -5 0 0 0 -13 0 Cz 1 0 0 -20 1 1 The characteristic polynomial found from the transfer function of the plant is s3 + 38s2 + 425s + 1300 From this characteristic polynomial, we can write the observer canonical form of the state equations. The A and C matrices of the observer canonical form are given below as Ax -38 -425 -1300 1 0 0 Cx 1 0 1 0 0 0 To test observability, we write the observability matrices for both systems and show that both observability matrices have non zero determinants. Using Eq. (12.79), OMz 1 -5 25 Det(OMz) Observability Matrix of z 1 -13 169 1 -20 400 -840 OMx 1 -38 1019 Observability Matrix of x 0 1 -38 Det(OMx) 0 0 1 1 Using Eq. (12.89), we obtain the transformation matrix, P, and its inverse as P 0.2083333 -3.017857 3.8095238 Transformation Matrix z=Px PINV -0.04166667 0.008333333 1.00 1.00 1.00 0.232142857 -0.01785714 33.00 25.00 18.00 -0.19047619 0.00952381 260.00 100.00 65.00 Using the characteristic polynomial given in the problem statement, the plant’s characteristic equation, and Eq. (12.67), the observer for the observer canonical system is Lx 562 Observer for x 556 Chapter 12: Design via State Space 39575 1498700 Using Eq. (12.92), the observer for the parallel system is found to be Lz 10957.29167 -19271.4821 8876.190476 29. Observer for z Use Eqs. (4.39) and (4.42) to find ζ = 0.5912 and ωn = 135.328 respectively. Factoring the denominator of Eq. (4.22), the required poles are -80 ± j109.15. We place the third pole 10 times further at -800. Multiplying the three closed-loop pole terms yields the desired characteristic equation: s3 + 960s2 + 146313.746s + 14650996.915 = 0. ⎡ − 18 1 0 ⎤ 50 ⎥ ⎢ , the plant in observer canonical form is: A oc = ⎢ − 99 0 1 ⎥ ; Since G(s) = 3 s + 18s2 + 99s + 162 ⎥ ⎢ ⎣ −162 0 0 ⎦ B oc ⎡1 ⎡0⎤ 0 0⎤ ⎥ ⎢ ⎢⎥ = ⎢ 0 ⎥ ; Coc = [1 0 0]. Using Eq. (12.79), OMoc = ⎢− 18 1 0⎥ , which is observable since ⎥ ⎢ ⎢⎥ ⎣ 225 − 18 1⎦ ⎣50 ⎦ 50 the determinant is 1. Since G(s) = 3 , the phase-variable form is s + 18s2 + 99s + 162 A pv ⎡0 ⎤ ⎡0 1 0⎤ ⎢⎥ ⎢ ⎥ =⎢ 0 0 1 ⎥ ; Bpv = ⎢0 ⎥ ; C = [50 0 0 ]. Using Eq. (12.79), ⎢1 ⎦ ⎢− 162 − 99 − 18⎥ ⎣⎥ ⎣ ⎦ ⎡50 0 0 ⎤ ⎥ ⎢ OMpv = ⎢ 0 50 0 ⎥ , which is observable since the determinant is 125000. The first column of ⎥ ⎢ ⎣ 0 0 50 ⎦ Aoc contains the negatives values of the coefficients of the characteristic equation. Comparing the first column of Aoc to the desired characteristic equation and using Eq. (12.67), l1 = 960-18 = 942; l2 = 146313.746-99 = 146214.746; and l3 = 14650996.915-162= 14650834.915. Hence, Loc = [942 146214.746 14650834.915]. The transformation matrix is, ⎡ 0. 02 ⎢ P = OMpv-1OMoc = ⎢−0. 36 ⎢ ⎣ 4. 5 0⎤ ⎥ 0⎥ ⎥ − 0. 36 0. 02⎦ 0 0. 02 Transforming back to the original system, Lpv = PLoc = [18.84 2585.175 244618.39]T. Solutions to Problems 557 30. s+2 . The open-loop transfer function of the plant is T(s) = C(sI-A)-1B = 2 s -s-2 Using Eqs. (12.115), the closed-loop state equations with integral control is The characteristic polynomial is s3 + (k2-1)s2 + (k2 + k1 + ke - 2)s + 2ke The desired characteristic polynomial is calculated from the desired transient response stated in the problem. Also, the third pole will be placed to cancel the zero at -2. Hence, the desired characteristic polynomial is (s + 2)(s2 + 16s + 183.137) = s3 + 18s2 + 215.14s + 366.27 Equating coefficients of the characteristic polynomials yields, ke = 183.135, k2 = 19, k1 = 15.005 31. s+3 . The open-loop transfer function of the plant is T(s) = C(sI-A)-1B = 2 s +7s+10 Using Eqs. (12.115), the closed-loop state equations with integral control is The characteristic polynomial is s3 + (k2 + 7)s2 + (2k2 + k1 + ke + 10)s + 3ke The desired characteristic polynomial is calculated from the desired transient response stated in the problem. Also, the third pole will be placed to cancel the zero at -3. Hence, the desired characteristic polynomial is Equating coefficients of the characteristic polynomials yields, ke = 183.137, k2 = 12, k1 = 14.003 558 Chapter 12: Design via State Space SOLUTIONS TO DESIGN PROBLEMS 32. Writing the A and B matrices for (G(s) represented in phase-variables form, A 0 0 1.30E+06 1 0 4551 B 0 0 10 0 1 -286 From the phase-variable from, the characteristic polynomial is s3 + 286s2 - 4551s - 1301586. Finding the controllability matrix and it’s determinant shows that there is controllability, CM 0 0 10 0 10 -2860 Det(CM) -1000 10 -2860 863470 Using the given transient requirements, and arbitrarily placing the third closed-loop pole more than 5 times further than the dominant pair at -50 yields the following desired closed-loop characteristic polynomial: (s2 + 16s + 134.384)(s + 50) = s3 + 66s2 + 934.4s + 6719.2 Using the phase-variable system with state feedback the characteristic polynomial is s3 + (k3 + 286)s2 + (k2 - 4551)s + (k1 - 1301586) Equating the two characteristic polynomials yields the state feedback vector for the phase-variable system as K 1308305.2 5485.4 -220 33. Controller design: The transfer function for the plant is 5 5 G(s) = (s+0.4)(s+0.8)(s+5) = 3 s +6.2s2+6.32s+1.6 The characteristic polynomial for the plant with phase-variable state feedback is s3 + (6.2 + k3)s2 + (6.32 + k2)s + (1.6 + k1) Using the given transient response of 5% overshoot and Ts = 10 minutes, and placing the third pole ten times further from the imaginary axis than the dominant pair, the desired characteristic equation is Solutions to Design Problems 559 (s + 4)(s2 + 0.8s + 0.336) = s3 + 4.8s2 + 3.536s + 1.344. Comparing the two characteristic equations, k1 = - 0.256, k2 = -2.784, and k3 = -1.4. Observer design: The A and C matrices for the system represented in phase-variable form is Az 0 0 -1.6 1 0 -6.32 0 1 -6.2 Cz 5 0 0 The characteristic polynomial found from the transfer function of the plant is s3 + 6.2s2 + 6.32s + 1.6 From this characteristic polynomial, we can write the dual phase-variable form of the state equations. The A and C matrices of the dual phase-variable form are given below as Ax -6.2 1 0 -6.32 0 1 -1.6 0 0 Cx 1 0 0 To test observability, we write the observability matrices for both systems and show that both observability matrices have nonzero determinants. Using Eq. (12.79), OMz 5 0 0 Observability Matrix of z 0 0 5 0 0 5 125 Det(OMz) OMx 1 -6.2 32.12 Observability Matrix of x 0 0 1 0 -6.2 1 1 Det(OMx) Using Eq. (12.89), we obtain the transformation matrix, P, and its inverse as P 0.2 -1.24 6.424 Transformation Matrix z=Px 0 0 0.2 0 -1.24 0.2 PINV 5.00 31.00 31.60 0.00 5.00 31.00 0.00 0.00 5.00 Using the characteristic polynomial given in the problem statement, the observer for the dual phasevariable system is 560 Chapter 12: Design via State Space Lx 41.8 347.28 1342.4 Using Eq. (12.92), the observer for the cascade system is found to be Lz 8.36 17.624 106.376 34. a. Using the following signal-flow graph, 1 u s 800 1 x3 20000 s x2 1 1 s x1 1 y -100 -800 the plant is represented in state space with ⎡0⎤ ⎡0 1 0⎤ ⎢ ⎢ ⎥ ⎥ A = ⎢0 −100 20000⎥ ; B = ⎢ 0 ⎥ ; and C = [1 0 0]. ⎢ ⎢ ⎥ ⎥ 0 − 800 ⎦ ⎣800 ⎦ ⎣0 Using Eq. (12.26), CM ⎡0 0 1. 6 E 07 ⎤ ⎥ ⎢ = ⎢ 0 1. 6 E 07 −1. 44 E10 ⎥ ⎥ ⎢ ⎣800 − 6. 4 E 05 5. 12 E08 ⎦ The system is controllable since the determinant of CM = -2.04e17. Use Eqs. (4.39) and (4.42) to find ζ = 0.5912 and ωn = 135.3283 respectively. Factoring the denominator of Eq. (4.22), the required poles are -80 ± j109.15. Place the third pole 10 times farther at = 800. Multiplying the three closedloop pole terms yields the desired characteristic equation s3 + 960s2 + 1.463E05s + 1.4651E07 = 0. Since the plant's characteristic equation is s3 + 900s2 + 80000s, we write the plant in controller canonical form as ⎡− 900 − 80000 0 ⎤ ⎡1 ⎤ ⎥ ⎢ ⎢⎥ 0 0 ⎥ ; B cc = ⎢0 ⎥ ; and Ccc = [0 0 1.6E07] A cc = ⎢ 1 ⎢ ⎢⎥ 1 0⎥ ⎣0 ⎣0 ⎦ ⎦ The controllability matrix for controllable canonical form is C Mcc ⎡1 −900 730000⎤ ⎢ ⎥ 1 −900 ⎥ = ⎢0 ⎢ ⎥ 0 1⎦ ⎣0 Solutions to Design Problems 561 Comparing the first row of Acc to the desired characteristic equation and using the results of Problem 5, k1 = -(900 - 960) = 60; k2 = -(80000 - 1.463E05) = 66300; and k3 = -(0 - 1.465E07) = 1.465E07. Hence. Kcc = [60 66300 1.465E07] The transformation matrix is, ⎡0 150 1. 6E 07⎤ ⎥ ⎢ 0⎥ P = CMCMcc-1 = ⎢ 0 1. 6E 07 ⎥ ⎢ 0⎦ ⎣800 8E 04 Transforming back to the original system, K = KccP-1 = [9.1569E-01 3.7696E-03 7.5E-02] The controller compensated system is ⎡0 1 ⎢ − 100 A - BK = ⎢ 0 ⎢−732. 55 −3. 0157 ⎣ 0⎤ ⎡0⎤ ⎥ ⎥ ⎢ 20000⎥ ; B = ⎢ 0 ⎥ ; C = [1 0 0] ⎥ ⎢800 ⎦ −860 ⎥ ⎦ ⎣ b. To evaluate the steady-state error, use Eq. (7.89) where ⎡0 1 ⎢ A - BK = ⎢ 0 − 100 ⎢−732. 55 −3. 0157 ⎣ 0⎤ ⎥ 20000⎥ ⎥ −860 ⎦ is the system matrix. Thus, s I − A −B K −1 = s 2 + 960 s + 1.4631 ×10 5 − 14651040 s + 860 2 + 860 s s − 732.55 s − 73255 − 3.0157 s − 732.55 1 s 3 + 960 s 2 + 1.4631 ×10 5 s + 14651040 20000 20000 s s 2 + 100 s The steady-state error is given by sR(s)[1 − C s I − A − B K − 1 B ] as s->0 For a step input, R(s) = 1/s. Since 1− C s I − A − B K −1 B = 1− 1 s 3 + 960 s 2 + 1.4631×10 5 s + 14651040 16000000 for a step input e(∞) = -0.092073. Using Eqs. 12.115, the system with integral control is: 0 AI = 1 0 0 0 − 100 20000 0 − 800 K 1 −1 − 800 K 2 0 − 800 K 3 − 800 0 800 K e 0 0 ;BI = 0 ; 0 1 C I = 1, 0, 0, 0 Assume the following desired characteristic equation: (s3 + 960s2 + 1.463E05s + 1.4651E07)(s + 1000) = s 4 + 1960 s 3 + 1.1063×106 s 2 + 1.6096×108 s + 1.4651×1010 = 0, 562 Chapter 12: Design via State Space which is the desired characteristic equation from part (a) plus an additional pole at -1000. But the integral controlled system characteristic equation is |sI - AI| = s 4 + 100 8 K 3 + 9 s 3 + 80000 K 3 + 200 K 2 + 1 s 2 + 16000000 K 1 s + 16000000 K e Equating coefficients to the desired characteristic equation 100 8 K 3 + 9 = 1960; 80000 K 3 + 200 K 2 + 1 = 1.1063×106 ; 16000000 K 1 = 1.6096×108 ; 16000000 K e = 1.4651 ×10 10 Solving for the controller gains: K e = 915.69 ; K 1 = 10.06; K 2 = 0.05752; and K 3 = 1.325. Substituting into AI yields the integral controlled system. 0 0 0 0 − 100 20000 0 − 8048.2 − 46.016 − 1860 −1 AI = 1 0 0 0 ; B I = 0 ; C I = 1, 0, 0, 0 0 7.3255 ×10 5 0 1 Finding the characteristic equation as a check yields s 4 + 1960 s 3 + 1.1063×106 s 2 + 1.6096×108 s + 1.4651×1010 which checks with the desired characteristic equation. Now check the steady-state error using Eq. (7.89) using the integral controlled system. We find the error is zero. c. Program: 'Controller Compensated' A=[0 1 0;0 -100 20000;-732.55 -3.0157 -860]; B=[0;0;800]; C=[1 0 0]; D=0; S=ss(A,B,C,D) step(S) title('Controller Compensated') pause 'Integral Controller' A=[0 1 0 0;0 -100 20000 0;-8048.2 -46.016 -1860 7.3255e05;-1 0 0 0]; B=[0;0;0;1]; C=[1 0 0 0]; D=0; S=ss(A,B,C,D) step(S) title('Integral Controller') Computer response: ans = Controller Compensated a= x1 x2 x3 x1 0 1 0 x2 0 -100 2e+004 x3 -732.5 -3.016 -860 Solutions to Design Problems 563 b= u1 x1 0 x2 0 x3 800 c= x1 x2 x3 y1 1 0 0 d= u1 y1 0 Continuous-time model. ans = Integral Controller a= x1 0 0 -8048 -1 x1 x2 x3 x4 x2 x3 x4 1 0 0 -100 2e+004 0 -46.02 -1860 7.326e+005 0 0 0 b= x1 x2 x3 x4 u1 0 0 0 1 c= x1 x2 x3 x4 y1 1 0 0 0 d= u1 y1 0 Continuous-time model. 564 Chapter 12: Design via State Space Solutions to Design Problems 565 35. Program: %Enter G(s) numg=0.072*conv([1 23],[1 0.05 0.04]); deng=conv([1 0.08 0.04],poly([0.7 -1.7])); 'G(s)' G=tf(numg,deng) 'Plant Zeros' plantzeros=roots(numg) %Input transient response specifications Po=input('Type %OS '); Ts=input('Type settling time '); %Determine pole location z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); wn=4/(z*Ts); %wn=pi/(Tp*sqrt(1-z^2)); [num,den]=ord2(wn,z); r=roots(den); poles=[r(1) r(2) plantzeros(2) plantzeros(3)] characteristiceqdesired=poly(poles) %Find controller canonical form of state-space representation of G(s) 'Controller Canonical Form' [Ac Bc Cc Dc]=tf2ss(numg,deng) %Design controller gains Kc=acker(Ac,Bc,poles) Acnew=Ac-Bc*Kc Bcnew=Bc Ccnew=Cc Dcnew=Dc characteristiceqcontroller=poly(eig(Acnew)) %Transform to phase-variable form P=[0 0 0 1;0 0 1 0;0 1 0 0;1 0 0 0]; 'Phase-variable form' Ap=inv(P)*Ac*P Bp=inv(P)*Bc Cp=Cc*P Dp=Dc Kp=acker(Ap,Bp,poles) Apnew=Ap-Bp*Kp Bpnew=Bp Cpnew=Cp Dpnew=Dp characteristiceqphase=poly(eig(Apnew)) [numt,dent]=ss2tf(Apnew,Bpnew,Cpnew,Dpnew); T=tf(numt,dent); 'T(s)' T=minreal(T) step(T) 'T(s)' Tzpk=zpk(T) 'Poles of T(s)' p=pole(T Computer response: ans = 566 Chapter 12: Design via State Space G(s) Transfer function: 0.072 s^3 + 1.66 s^2 + 0.08568 s + 0.06624 --------------------------------------------s^4 + 1.08 s^3 - 1.07 s^2 - 0.0552 s - 0.0476 ans = Plant Zeros plantzeros = -23.0000 -0.0250 + 0.1984i -0.0250 - 0.1984i Type %OS 10 Type settling time 0.5 poles = -8.0000 +10.9150i 0.1984i -8.0000 -10.9150i -0.0250 + 0.1984i characteristiceqdesired = 1.0000 16.0500 183.9775 9.7969 ans = Controller Canonical Form Ac = -1.0800 1.0000 0 0 1.0700 0 1.0000 0 0.0552 0 0 1.0000 0.0476 0 0 0 1.6596 0.0857 0.0662 Bc = 1 0 0 0 Cc = 0.0720 Dc = 0 Kc = 7.3255 -0.0250 - Solutions to Design Problems 567 14.9700 185.0475 9.8521 7.3731 -16.0500 -183.9775 1.0000 0 0 1.0000 0 0 -9.7969 0 0 1.0000 -7.3255 0 0 0 0.0857 0.0662 Acnew = Bcnew = 1 0 0 0 Ccnew = 0.0720 1.6596 Dcnew = 0 characteristiceqcontroller = 1.0000 16.0500 183.9775 9.7969 1.0000 0 0 0.0552 0 1.0000 0 1.0700 0 0 1.0000 -1.0800 0.0857 1.6596 0.0720 ans = Phase-variable form Ap = 0 0 0 0.0476 Bp = 0 0 0 1 Cp = 0.0662 Dp = 0 Kp = 7.3255 568 Chapter 12: Design via State Space 7.3731 9.8521 185.0475 14.9700 1.0000 0 0 1.0000 0 0 -9.7969 -183.9775 0 0 1.0000 -16.0500 Apnew = 0 0 0 -7.3255 Bpnew = 0 0 0 1 Cpnew = 0.0662 0.0857 1.6596 0.0720 183.9775 9.7969 Dpnew = 0 characteristiceqphase = 1.0000 16.0500 ans = T(s) Transfer function: 0.072 s + 1.656 -----------------s^2 + 16 s + 183.1 ans = T(s) Zero/pole/gain: 0.072 (s+23) -------------------(s^2 + 16s + 183.1) ans = Poles of T(s) p= -8.0000 +10.9150i -8.0000 -10.9150i 7.3255 Solutions to Design Problems 569 36. Program: %Enter G(s) numg=0.072*conv([1 23],[1 0.05 0.04]); deng=conv([1 0.08 0.04],poly([0.7 -1.7])); 'Uncompensated Plant Transfer Function' 'G(s)' G=tf(numg,deng) 'Uncompensated Plant Zeros' plantzeros=roots(numg) %Input transient response specifications Po=input('Type %OS '); Ts=input('Type settling time '); %Determine pole location z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); wn=4/(z*Ts); %wn=pi/(Tp*sqrt(1-z^2)); [num,den]=ord2(wn,z); r=roots(den); 'Desired Observer Poles' poles=[r(1) r(2) plantzeros(2) plantzeros(3)]' 'Desired Characteristic Equation of Observer' poly(poles) %Find phase variable form of state-space representation of Estimated Plant %Find controller canonical form [Ac Bc Cc Dc]=tf2ss(numg,deng); %Transform to phase-variable form of Uncompensated Plant P=[0 0 0 1;0 0 1 0;0 1 0 0;1 0 0 0]; 'Phase-variable form of Estimated Plant' Ap=inv(P)*Ac*P 570 Chapter 12: Design via State Space Bp=inv(P)*Bc Cp=Cc*P Dp=Dc %Design observer gains for phase variables 'Observer gains' Lp=acker(Ap',Cp',poles)' 'Error System Matrix' Aep=Ap-Lp*Cp 'Error System Eigenvalues' eig(Aep) 'Error Characteristic Polynomial' poly(eig(Aep)) Computer response: ans = Uncompensated Plant Transfer Function ans = G(s) Transfer function: 0.072 s^3 + 1.66 s^2 + 0.08568 s + 0.06624 --------------------------------------------s^4 + 1.08 s^3 - 1.07 s^2 - 0.0552 s - 0.0476 ans = Uncompensated Plant Zeros plantzeros = -23.0000 -0.0250 + 0.1984i -0.0250 - 0.1984i Type %OS 10 Type settling time 0.5/15 ans = Desired Observer Poles poles = 1.0e+002 * -1.2000 -1.2000 -0.0003 -0.0003 + + 1.6373i 1.6373i 0.0020i 0.0020i ans = Desired Characteristic Equation of Observer Solutions to Design Problems 571 ans = 1.0e+004 * 0.0001 0.0240 4.1218 0.2070 ans = Phase-variable form of Estimated Plant Ap = 0 0 0 0.0476 1.0000 0 0 0.0552 0 1.0000 0 1.0700 0 0 1.0000 -1.0800 0.0857 1.6596 0.0720 0.0003 -0.0071 0.1636 -4.3139 0.0000 -0.0003 0.0072 -0.1873 Bp = 0 0 0 1 Cp = 0.0662 Dp = 0 ans = Observer gains Lp = 1.0e+004 * -0.0002 0.0043 -0.0986 2.5994 ans = Error System Matrix Aep = 1.0e+004 * 0.0000 -0.0003 0.0065 -0.1722 0.0001 -0.0004 0.0084 -0.2227 0.1648 572 Chapter 12: Design via State Space ans = Error System Eigenvalues ans = 1.0e+002 * -1.2000 -1.2000 -0.0003 -0.0003 + + - 1.6373i 1.6373i 0.0020i 0.0020i ans = Error Characteristic Polynomial ans = 1.0e+004 * 0.0001 0.0240 4.1218 0.2070 0.1648 37. a. Using Eqs. (12.115), the system with integral control is: 0 0 0 0 0 0 1 0 0 0 0 0 1 0 − K 1 + 0.0476 − 0.06624 AI = 1 − K 2 + 0.0552 − 0.08568 − K 3 + 1.07 − 1.6596 − K 4 − 1.08 − 0.072 Ke 0 0 0 ;BI = 0 ; 0 1 C I = 1, 0, 0, 0, 0 Assume the following desired characteristic equation, s + 8 + 10.915 i s + 8 − 10.915 i s + 0.025 + 0.1984 i s + 0.025 − 0.1984 i s + 23 = s 5 + 39.05 s 4 + 553.13 s 3 + 4241.3 s 2 + 232.65 s + 168.43 which is the desired characteristic equation from Problem 35 plus an additional pole at -23, the closed-loop zero. But the integral controlled system characteristic equation is |sI - AI| = s 5 + K 4 + 1.08 s 4 + K 3 + 0.072 K e − 1.07 s 3 + K 2 + 1.6596 K e − 0.0552 s 2 + K 1 + 0.08568 K e − 0.0476 s + 0.06624 K e Equating coefficients to the desired characteristic equation K 4 + 1.08 = 39.05; K 3 + 0.072 K e − 1.07 = 553.13 ; K 2 + 1.6596 K e − 0.0552 = 4241.3 ; K 1 + 0.08568 K e − 0.0476 = 232.65 ; and 0.06624 K e = 168.43 Solving for the controller gains K 1 = 14.829; K 2 = 21.328; K 3 = 371.12; K 4 = 37.97 and K e = 2542.8 Substituting into AI yields the integral controlled system, Solutions to Design Problems 573 0 0 0 − 14.781 − 0.06624 AI = 1 0 0 − 21.272 − 0.08568 0 1 0 − 370.05 − 1.6596 0 0 1 − 39.05 − 0.072 0 0 0 2542.8 0 0 0 ;BI = 0 ; 0 1 C I = 0.06624 , 0.08568 , 1.6596 , 0.072 , 0 Finding the characteristic equation as a check yields s 5 + 39.05 s 4 + 553.13 s 3 + 4241.3 s 2 + 232.65 s + 168.43 which checks with the desired. Now check the steady-state error using Eq. (7.89) using the integral controlled system. We find the error is zero. b. Program: %Design with Integral Control 'State-Space Representation of System with Integral Control' AI=[0 1 0 0 0;0 0 1 0 0;0 0 0 1 0;... -14.781 -21.272 -370.05 -39.05 2542.8;... -0.06624 -0.08568 -1.6596 -0.072 0] BI=[0;0;0;0;1] CI=[0.06624 0.08568 1.6596 0.072 0] DI=0 %Convert to transfer function [numt,dent]=ss2tf(AI,BI,CI,DI); 'Integral Control Transfer Function' 'T(s)' T=tf(numt,dent) 'Integral Control Transfer Function Zeros' roots(numt) 'Integral Control Transfer Function Poles' roots(dent) step(T) title('Step Response with Integral Controller') Computer response: ans = State-Space Representation of System with Integral Control AI = 1.0e+003 * 0 0 0 -0.0148 -0.0001 BI = 0 0 0 0 0.0010 0 0 -0.0213 -0.0001 0 0.0010 0 -0.3700 -0.0017 0 0 0.0010 -0.0390 -0.0001 0 0 0 2.5428 0 574 Chapter 12: Design via State Space 1 CI = 0.0662 0.0857 1.6596 0.0720 0 DI = 0 ans = Integral Control Transfer Function ans = T(s) Transfer function: -1.421e-014 s^4 + 183.1 s^3 + 4220 s^2 + 217.9 s + 168.4 -------------------------------------------------------s^5 + 39.05 s^4 + 553.1 s^3 + 4241 s^2 + 232.6 s + 168.4 ans = Integral Control Transfer Function Zeros ans = 1.0e+016 * 1.2883 -0.0000 -0.0000 + 0.0000i -0.0000 - 0.0000i ans = Integral Control Transfer Function Poles ans = -22.9998 -8.0001 -8.0001 -0.0250 -0.0250 +10.9151i -10.9151i + 0.1984i - 0.1984i Solutions to Design Problems 575 38. a. The open-loop block diagram is Y h-Y cat Spring displacement Desired force Input 1 voltage 1 100 1000 Input transducer Fup 0 . 7883 ( s + 53 . 85 ) 82300 Actuator 2 2 ( s + 15 . 47 s + 9283 )( s + 8 . 119 s + 376 .3 ) Pantograph dynamics From Chapter 3, the state-space representation for [Yh(s) – Ycat(s)]/Fup(s) is 1 0 0⎤ ⎡0⎤ ⎡0 ⎢0⎥ ⎢ -9353 -14.286 769.23 14.286 ⎥ Ý f x=⎢ x+⎢ 0 0 0 1⎥ 0 ⎥ up ⎢ 0.0581⎥ ⎢406.98 7.5581 -406.98 -9.3023⎥ ⎣ ⎦ ⎣ ⎦ y = [0.94911 0 0 0]x Spring Fout 576 Chapter 12: Design via State Space ⎡y h ⎤ . ⎢y h ⎥ where y = yh - ycat and x = ⎢ ⎥ yf ⎢.⎥ ⎣y f ⎦ Let vi represent the input voltage shown on the diagram. Thus, fup = vi/1000. Also, fout = 82300(yh - ycat). Thus, fout = 82300y Substituting fup and fout into the state-equations above yields 1 0 0⎤ 0 ⎡ ⎤ ⎡0 ⎢ ⎥ ⎢ -9353 -14.286 769.23 14.286 ⎥ 0 Ý x=⎢ x+⎢ ⎥ vi 0 0 0 1⎥ 0 ⎢ 0.0581x10-3 ⎥ ⎢406.98 7.5581 -406.98 -9.3023⎥ ⎣ ⎦ ⎣ ⎦ fout = [78,112 0 0 0]x Thus, 1 0 0⎤ ⎡0 ⎢ -9353 -14.286 769.23 14.286 ⎥ A=⎢ 0 0 0 1⎥ ⎢ 406.98 7.5581 -406.98 -9.3023⎥ ⎣ ⎦ 0 ⎡ ⎤ ⎥ ⎢ 0 B=⎢ ⎥ 0 ⎢ 0.0581x10-3 ⎥ ⎣ ⎦ K = [k1 k2 k3 k4 ] Hence, A-BK = [0 , 1 , 0 , 0 [-9350 , -14.3 , 769 , 14.3 [0 , 0 , 0 , 1 [407 - 0.0000581 k1 , 7.56 - .0000581 k2 , -407 - 0.0000581 k3 , -9.30 - 0.0000581 k4] Solutions to Design Problems 577 and | A-BK| = s4 + (0.0000581 k4 + 23.60) s3 + (0.00083083 k4 + 0.00083083 k2 + 9781.882 + 0.0000581 k3) s2 + (0.00083083 k1 + 81141.36 + 0.543235 k4 + 0.00083083 k3 + 0.0446789 k2) s +(0 .0446789 k1 +0.3492467 107 + 0.543235 k3) Input transient response specifications, Po = 20 Ts = 1 yields poles at -4.0000 + 7.8079i, -4.0000 - 7.8079i, -53.8500, -50.0000 Thus, the desired characteristic equation is s4 + 112s3 + 3600s2 + 29500s + 207000 = 0 We now equate the coefficients of |A-BK| to the coefficients of the desired characteristic equation. For compactness we solve for the coefficients, K, using the form FK = G, where F= 0 0 0.00083083 0 .0446789 0 0.00083083 0.0446789 0 0 0.0000581 0.0000581 0.00083083 0.00083083 0.543235 0.543235 0 and G= 88.4 –6181.882 -51641.36 -3285467 Solving for K using K = F-1G K= -4.8225e8 -0.1131e8 0.3361e8 0.0152e8 b. Integral Control Design A= 1.0e+03 * 0 -9.3530 0 0.4070 B= 1.0e-04 * 0 0 0 0.5810 0.0010 -0.0143 0 0.0076 0 0.7692 0 -0.4070 0 0.0143 0.0010 -0.0093 578 Chapter 12: Design via State Space C= 78112 0 0 0 Aaug = A-BK = [0 , 1. , 0 , 0 , 0 [-9350. , -14.3 , 769. , 14.3 , 0 [0 , 0 , 0 , 1. , 0 [407. - 0.0000581 k1 , 7.56 - 0.0000581 k2 , -407. - 0.0000581 k3 , -9.30 - 0.0000581 k4 , 0.0000581 Ke [-78100. , 0 , 0 , 0 , 0] Desired poles Po = 20 Ts = 1 Determine pole location poles = -4.0000 + 7.8079i, -4.0000 - 7.8079i, -53.85, -50 , -50 Desired characteristic equation s5 + 162s4 + 0.919e4s3 + 0.210e6s2 + 0.168e7s + 0.104e8 System characteristic equation |sI-Aaug| = s5 + (23.60 + 0.0000581 k4) s4 + (0.00083083 k4 + 0.00083083 k2 + 0.0000581 k3 + 9781.882) s3 + (0.00083083 k1 + 81141.36 + 0.0446789 k2 + 0.543235 k4 + 0.00083083 k3) s2 + (0.0446789 k1 + 64.887823 Ke + 0.543235 k3 + 0.3492467 107 ) s + 3489.42209 Ke Solving for Coefficients, K, using FK = G as in (a), where F 0 0 0 8.3083e-04 8.3083e-04 4.4679e-02 4.4679e-02 0 0 0 G= 1.3840e+02 -5.9188e+02 1.2886e+05 -1.8125e+06 1.0800e+07 Thus, K= -1.0157e+09 -8.6768e+06 7.9827e+07 2.3821e+06 0 5.8100e-05 8.3083e-04 5.4324e-01 0 5.8100e-05 0 8.3083e-04 0 5.4324e-01 0 0 6.4888e+01 0 3.4894e+03 Solutions to Design Problems 579 3.0951e+03 Step Response 1.2 1 Amplitude 0.8 0.6 0.4 0.2 0 0 0.4 0.8 Time (sec.) 1.2 THIRTEEN Digital Control Systems SOLUTIONS TO CASE STUDIES CHALLENGES Antenna Control: Transient Design via Gain a. From the answer to the antenna control challenge in Chapter 5, the equivalent forward transfer function found by neglecting the dynamics of the power amplifier, replacing the pots with unity gain, and including the integration in the sample-and-hold is Ge(s) = 0.16K s (s + 1.32) 2 But, z-1 Thus, Ge(z) = 0.16K z Gz , or, (z+0.95696) Ge(z) = 7.659x10-4K (z-1) (z-0.87634) b. Draw the root locus and overlay it over the ζ = 0.5 (i.e. 16.3% overshoot) curve. . Solutions to Case Studies Challenges 581 We find that the root locus crosses at approximately 0.93 ± j0.11 with 7.659x10-4K = 8.63x10-3. Hence, K = 11.268. c. 1 (7.659x10−4 K)(1.95696) lim(z − 1)Ge (z) = = 0.1366; T z →1 0.12366 1 e( ∞ ) = = 7.321 Kv Kv = d. Program: T=0.1; numf=0.16; denf=[1 1.32 0 0]; 'F(s)' F=tf(numf,denf) numc=conv([1 0],numf); denc=denf; %Input sampling time %Numerator of F(s) %Denominator of F(s) %Display label %Display F(s) %Differentiate F(s) to compensate %for c2dm which assumes series zoh %Denominator of continuous system %same as denominator of F(s) %Form continuous system, C(s) %Cancel common poles and zeros %Convert to z assuming zoh C=tf(numc,denc); C=minreal(C,1e-10); D=c2d(C,T,'zoh'); 'F(z)' D=minreal(D,1e-10) %Cancel common poles and zeros and display rlocus(D) pos=(16.3); z=-log(pos/100)/sqrt(pi^2+[log(pos/100)]^2); zgrid(z,0) title(['Root Locus with ' , num2str(pos), ' Percent Overshoot Line']) [K,p]=rlocfind(D) %Allows input by selecting point on %graphic Computer response: ans = F(s) Transfer function: 0.16 -------------s^3 + 1.32 s^2 ans = F(z) Transfer function: 0.0007659 z + 0.000733 ---------------------z^2 - 1.876 z + 0.8763 Sampling time: 0.1 Select a point in the graphics window selected_point = 9.2969e-001 +1.0219e-001i K= 9.8808e+000 582 Chapter 13: Digital Control Systems p= 9.3439e-001 +1.0250e-001i 9.3439e-001 -1.0250e-001i Antenna Control: Digital Cascade Compensator Design a. Let the compensator be KGc(s) and the plant be Gp (s ) = 0.16 . For 10% overshoot and a s( s + 1.32) peak time of 1 second, ζ = 0.591 and ωn = 3.895, which places the dominant poles at –2.303 ± j3.142. If we place the compensator zero at –1.32 to cancel the plant’s pole, then the following geometry results. j3.142 s-plane X -pc Hence, pc = 4.606. Thus, Gc (s ) -2.303 = X K (s + 1.32 ) 0.16 K and Gc (s )G p ( s ) = . Using the ( s + 4.606) s(s + 4.606) product of pole lengths to find the gain, 0.16K = (3.896)2, or K = 94.87. Hence, Solutions to Case Studies Challenges 583 Gc (s ) = 94.87(s + 1.32) . Using a sampling interval of 0.01 s, the Tustin transformation of Gc(s) (s + 4.606) is Gc (z ) = 93.35( z − 0.9869) 93.35z − 92.12 = . ( z − 0.955) z − 0.955 b. Cross multiplying, (z - 0.955)X(z) = (93.35z – 92.12)E(z) Solving for the highest power of z operating on X(z), zX(z) = (93.35z – 92.12)E(z) + 0.955X(z) Solving for X(z), X(z) = (93.35 – 92.12z-1)E(z) + 0.955z-1X(z) Implementing this equation as a flowchart yields the following diagram c. Program: 's-plane lead design for Challenge - Lead Comp' clf %Clear graph on screen. 'Uncompensated System' %Display label. numg=0.16; %Generate numerator of G(s). deng=poly([0 -1.32]); %Generate denominator of G(s). 'G(s)' %Display label. G=tf(numg,deng); %Create G(s). Gzpk=zpk(G) %Display G(s). pos=input('Type desired percent overshoot '); %Input desired percent overshoot. z=-log(pos/100)/sqrt(pi^2+[log(pos/100)]^2); %Calculate damping ratio. Tp=input('Type Desired Peak Time '); %Input desired peak time. wn=pi/(Tp*sqrt(1-z^2)); %Evaluate desired natural frequency. b=input('Type Lead Compensator Zero, (s+b). b= '); %Input lead compensator zero. done=1; %Set loop flag. while done==1 %Start loop for trying lead %compensator pole. a=input('Enter a Test Lead Compensator Pole, (s+a). a = '); %Enter test lead compensator pole. numge=conv(numg,[1 b]); %Generate numerator of Gc(s)G(s). denge=conv([1 a],deng); %Generate denominator of Gc(s)G(s). Ge=tf(numge,denge); %Create Ge(s)=Gc(s)G(s). clf %Clear graph on screen. 584 Chapter 13: Digital Control Systems rlocus(Ge) %Plot compensated root locus with %test lead compensator pole. axis([-5 2 -8 8]); %Change axes ranges. sgrid(z,wn) %Overlay grid on lead-compensated %root locus. title(['Lead-Compensated Root Locus with ' , num2str(pos),... '% Overshoot Line, Lead Pole at ', num2str(-a),... ' and Required Wn']) %Add title to lead-compensated root %locus. done=input('Are you done? (y=0,n=1) '); %Set loop flag. end %End loop for trying compensator %pole. [K,p]=rlocfind(Ge); %Generate gain, K, and closed-loop %poles, p, for point selected %interactively on the root locus. 'Gc(s)' %Display label. Gc=K*tf([1 b],[1 a]) %Display lead compensator. 'Gc(s)G(s)' %Display label. Ge %Display Gc(s)G(s). 'Closed-loop poles = ' %Display label. p %Display lead-compensated system's %closed-loop poles. f=input('Give pole number that is operating point '); %Choose lead-compensated system %dominant pole. 'Summary of estimated specifications for selected point on lead' 'compensated root locus' %Display label. operatingpoint=p(f) %Display lead-compensated dominant %pole. gain=K %Display lead-compensated gain. estimated_settling_time=4/abs(real(p(f))) %Display lead-compensated settling %time. estimated_peak_time=pi/abs(imag(p(f))) %Display lead-compensated peak time. estimated_percent_overshoot=pos %Display lead-compensated percent %overshoot. estimated_damping_ratio=z %Display lead-compensated damping %ratio. estimated_natural_frequency=sqrt(real(p(f))^2+imag(p(f))^2) %Display lead-compensated natural %frequency. s=tf([1 0],1); %Create transfer function, "s". sGe=s*Ge; %Create sGe(s) to evaluate Kv. sGe=minreal(sGe); %Cancel common poles and zeros. Kv=dcgain(K*sGe) %Display lead-compensated Kv. ess=1/Kv %Display lead-compensated steady%state error for unit ramp input. 'T(s)' %Display label. T=feedback(K*Ge,1) %Create and display lead-compensated %T(s). 'Press any key to continue and obtain the lead-compensated step' 'response' %Display label pause step(T) %Plot step response for lead %compensated system. title(['Lead-Compensated System with ' ,num2str(pos),'% Overshoot']) %Add title to step response of PD %compensated system. pause 'z-plane conversion for Challenge - Lead Comp' clf %Clear graph. 'Gc(s) in polynomial form' %Print label. Gcs=Gc %Create Gc(s) in polynomial form. 'Gc(s) in polynomial form' %Print label. Solutions to Case Studies Challenges 585 Gcszpk=zpk(Gcs) %Create Gc(s) in factored form. 'Gc(z) in polynomial form via Tustin Transformation' %Print label. Gcz=c2d(Gcs,1/100,'tustin') %Form Gc(z) via Tustin %transformation. 'Gc(z) in factored form via Tustin Transformation' %Print label. Gczzpk=zpk(Gcz) %Show Gc(z) in factored form. 'Gp(s) in polynomial form' %Print label. Gps=G %Create Gp(s) in polynomial form. 'Gp(s) in factored form' %Print label. Gpszpk=zpk(Gps) %Create Gp(s) in factored form. 'Gp(z) in polynomial form' %Print label. Gpz=c2d(Gps,1/100,'zoh') %Form Gp(z) via zoh transformation. 'Gp(z) in factored form' %Print label. Gpzzpk=zpk(Gpz) %Form Gp(z) in factored form. pole(Gpz) %Find poles. Gez=Gcz*Gpz; %Form Ge(z) = Gc(z)Gp(z). 'Ge(z) = Gc(z)Gp(z) in factored form' %Print label. Gezzpk=zpk(Gez) %Form Ge(z) in factored form. 'z-1' %Print label. zm1=tf([1 -1],1,1/100) %Form z-1. zm1Gez=minreal(zm1*Gez,.00001); '(z-1)Ge(z)' %Print label. zm1Gezzpk=zpk(zm1Gez) pole(zm1Gez) Kv=300*dcgain(zm1Gez) Tz=feedback(Gez,1) step(Tz) title('Closed-Loop Digital Step Response') %Add title to step response. Computer response: ans = s-plane lead design for Challenge - Lead Comp ans = Uncompensated System ans = G(s) Zero/pole/gain: 0.16 ---------s (s+1.32) Type desired percent overshoot 10 Type Desired Peak Time 1 Type Lead Compensator Zero, (s+b). b= 1.32 Enter a Test Lead Compensator Pole, (s+a). a = Are you done? (y=0,n=1) 0 Select a point in the graphics window selected_point = -2.3045 + 3.1056i ans = 4.606 586 Chapter 13: Digital Control Systems Gc(s) Transfer function: 93.43 s + 123.3 --------------s + 4.606 ans = Gc(s)G(s) Transfer function: 0.16 s + 0.2112 -----------------------s^3 + 5.926 s^2 + 6.08 s ans = Closed-loop poles = p= -2.3030 + 3.1056i -2.3030 - 3.1056i -1.3200 Give pole number that is operating point 1 ans = Summary of estimated specifications for selected point on lead ans = compensated root locus operatingpoint = -2.3030 + 3.1056i gain = 93.4281 estimated_settling_time = 1.7369 estimated_peak_time = 1.0116 estimated_percent_overshoot = 10 Solutions to Case Studies Challenges 587 estimated_damping_ratio = 0.5912 estimated_natural_frequency = 3.8663 Kv = 3.2454 ess = 0.3081 ans = T(s) Transfer function: 14.95 s + 19.73 --------------------------------s^3 + 5.926 s^2 + 21.03 s + 19.73 ans = Press any key to continue and obtain the lead-compensated step ans = response ans = z-plane conversion for Challenge - Lead Comp ans = Gc(s) in polynomial form Transfer function: 93.43 s + 123.3 --------------s + 4.606 ans = Gc(s) in polynomial form Zero/pole/gain: 93.4281 (s+1.32) ---------------- 588 Chapter 13: Digital Control Systems (s+4.606) ans = Gc(z) in polynomial form via Tustin Transformation Transfer function: 91.93 z - 90.72 --------------z - 0.955 Sampling time: 0.01 ans = Gc(z) in factored form via Tustin Transformation Zero/pole/gain: 91.9277 (z-0.9869) -----------------(z-0.955) Sampling time: 0.01 ans = Gp(s) in polynomial form Transfer function: 0.16 -----------s^2 + 1.32 s ans = Gp(s) in factored form Zero/pole/gain: 0.16 ---------s (s+1.32) ans = Gp(z) in polynomial form Transfer function: 7.965e-006 z + 7.93e-006 -----------------------z^2 - 1.987 z + 0.9869 Sampling time: 0.01 ans = Gp(z) in factored form Zero/pole/gain: Solutions to Case Studies Challenges 589 7.9649e-006 (z+0.9956) ---------------------(z-1) (z-0.9869) Sampling time: 0.01 ans = 1.0000 0.9869 ans = Ge(z) = Gc(z)Gp(z) in factored form Zero/pole/gain: 0.0007322 (z+0.9956) (z-0.9869) ------------------------------(z-1) (z-0.9869) (z-0.955) Sampling time: 0.01 ans = z-1 Transfer function: z-1 Sampling time: 0.01 ans = (z-1)Ge(z) Zero/pole/gain: 0.0007322 (z+0.9956) -------------------(z-0.955) Sampling time: 0.01 ans = 0.9550 Kv = 9.7362 Transfer function: 0.0007322 z^2 + 6.387e-006 z - 0.0007194 ---------------------------------------z^3 - 2.941 z^2 + 2.884 z - 0.9432 Sampling time: 0.01 590 Chapter 13: Digital Control Systems Solutions to Case Studies Challenges 591 592 Chapter 13: Digital Control Systems ANSWERS TO REVIEW QUESTIONS 1. (1) Supervisory functions external to the loop; (2) controller functions in the loop 2. (1) Control of multiple loops by the same hardware; (2) modifications made with software, not hardware; (3) more noise immunity (4) large gains usually not required 3. Quantization error; conversion time 4. An ideal sampler followed by a sample-and-hold 5. z = esT 6. The value of the time waveform only at the sampling instants 7. Partial fraction expansion; division to yield power series 8. Partial fraction 9. Division to yield power series 10. The input must be sampled; the output must be either sampled or thought of as sampled. 11. c(t) is c*(t) = c(kT), i.e. the output only at the sampling instants. 12. No; the waveform is only valid at the sampling instants. Instability may be apparent if one could only see between the sampling instants. The roots of the denominator of G(z) must be checked to see that they are within the unit circle. 13. A sample-and-hold must be present between the cascaded systems. 14. Inside the unit circle 15. Raible table; Jury's stability test 16. z=+1 17. There is no difference. 18. Map the point back to the s-plane. Since z = esT, s = (1/T) ln z. Thus, σ = (1/T) ln (Re z), and ω = (1/T) ln (Im z). 19. Determine the point on the s-plane and use z = esT. Thus, Re z = eσT cos ω, and Im z = eσT sin ω. 20. Use the techniques described in Chapters 9 and 11 and then convert the design to a digital compensator using the Tustin transformation. 21. Both compensators yield the same output at the sampling instants. SOLUTIONS TO PROBLEMS 1. a. f(t) = e-at; f*(t) = ∞ ∑e-akTδ(t-kT) ; F*(s) = k=0 ∞ ∑e-akT e-kTs = 1 + e-aT e-Ts + e-a2T e-2Ts + . . . Thus, k=0 F(z) = 1 + e-aT z-1 + e-a2T z-2 + . . . = 1 + x-1 + x-2 + . . . where x = e-aT z-1. Solutions to Problems 593 1 1 z = = . 1 - x-1 1 - e-aT z-1 z - e-aT ∞ ∞ b. f(t) = u(t); f*(t) = ∑δ(t-kT) ; F*(s) = ∑ e-kT s = 1 + e-Ts + e-2Ts + . . . k=0 k=0 But, F(z) = Thus, F(z) = 1 + z-1 + z-2 + . . Since 1 = 1 + z-1 + z-2 + z-3, F(z) = 1 - z-1 ∞ ∑z-k = k=0 1 = 1 - z-1 z z-1 . ∞ ∞ k=0 k=0 ∑(kT)2e-akTδ(t-kT) ; F*(s) = T2 ∑ k2 e-akT e-kTs c. f(t) = t2 e-at; f*(t) = ∞ ∑ k2 (e-(s+a)T)k = T2 ∞ ∑ k2 xk = T2 k=0 = T2(x + 4x2 + 9x3 + 16x4 + . . . ) , where x = e- (s+a)T. k=0 Let s1 = x + 4x2 + 9x3 + 16x4 + . . . Thus, xs1 = x2 + 4x3 + 9x4 + 16x5 + . . . Let s2 = s1 - xs1 = x + 3x2 + 5x3 + 7x4 + . . .Thus, xs2 = x2 + 4x3 + 9x4 + 16x3 + . . . Let s3 = s2 - xs2 = x + 2x2 + 2x3 + 2x4 + . . . Thus xs3 = x2 + 2x3 + 2x4 + 2x3 + . . . Let s4 = s3 - xs3 = x + x2. Solving for s3, x + x2 s3 = 1 - x and s3 x + x2 s2 = 1 - x = (1 - x)2 and s2 x + x2 s1 = 1 - x = (1 - x)3 Thus F*(s) = T2 s1 = T2 -(s+a)T+ e-2(s+a)T) x + x2 2 (e = 3 =T (1 - e-(s+a)T)3 (1 - x) T 2 [z −1 e −aT + z −2 e −2aT ] T 2 ze −aT [z + e − aT ] = − aT 3 −3 −aT 3 (z − e ) z (z − e ) ∞ d. f(t) = cos(ωkT); f*(t) = ∑ cos(ωkT )δ (t k =0 ∞ ∑ cos(ωkT )e k =0 (e jωkT + e − jωkT )e −kTs 1 ∞ T ( s − jω ) − k T ( s + jω ) − k ) +( e ) = ∑ (e ∑ 2 2 k=0 k =0 ∞ = − kT ) ; F*(s) = But, − kTs 594 Chapter 13: Digital Control Systems ∞ ∑ x-k k=0 = 1 . 1 - x-1 Thus, F*(s) = = 1⎡ 1 1 ⎤ 2 − e −Ts ( e jωT + e− jωT ) ⎤= 1⎡ + − T ( s − jω ) − T ( s + jω ) − T ( s − jω ) − T ( s + jω ) −T ( s − jω ) − T (s + jω ) ⎥ ⎦ 2 ⎢1 − e −e )+e e 2 ⎣1 − e 1−e ⎣ ⎦ 1⎡ 2 − e − Ts ( 2 cos(ωT ) 1 − z −1 cos(ωT ) ⎤ = 2 ⎢ 1 − e −Ts (e jωT + e − jωT ) + e −2 Ts ⎥ 1 − 2 z −1 cos(ωT ) + z −2 ⎣ ⎦ Therefore, F( z ) = z (z − cos(ω T)) z − 2 z cos(ωT ) + 1 2 2. Program: syms T a w n '(a)' 'f(kT)' f=exp(-a*n*T); pretty(f) 'F(z)' F=ztrans(f); pretty(F) %Construct symbolic objects for %'T', 'a','w', and 'n'. %Display label. %Display label. %Define f(kT). %Pretty print f(kT) %Display label. %Find z-transform, F(z). %Pretty print F(z). '(b)' 'f(kT)' f=exp(-0*n*T); pretty(f) 'F(z)' F=ztrans(f); pretty(F) %Display label. %Display label. %Define f(kT) %Pretty print f(kT) %Display label. %Find z-transform, F(z). %Pretty print F(z). '(c)' 'f(kT)' f=(n*T)^2*exp(-a*n*T); pretty(f) 'F(z)' F=ztrans(f); pretty(F) %Display label. %Display label. %Define f(kT) %Pretty print f(kT) %Display label. %Find z-transform, F(z). %Pretty print F(z). '(d)' 'f(kT)' f=cos(w*n*T); pretty(f) 'F(z)' F=ztrans(f); pretty(F) %Display label. %Display label. %Define f(kT) %Pretty print f(kT) %Display label. %Find z-transform, F(z). %Pretty print F(z). Computer response: ans = (a) ans = f(kT) exp(-a n T) Solutions to Problems 595 ans = F(z) z ------------------------/ z \ exp(-a T) |--------- - 1| \exp(-a T) / ans = (b) ans = f(kT) 1 ans = F(z) z ----z-1 ans = (c) ans = f(kT) 22 n T exp(-a n T) ans = F(z) 2 T z exp(-a T) (z + exp(-a T)) -----------------------------3 (z - exp(-a T)) ans = (d) ans = f(kT) cos(w n T) ans = F(z) (z - cos(w T)) z --------------------2 z - 2 z cos(w T) + 1 596 Chapter 13: Digital Control Systems 3. a. F( z ) = z( z + 3)( z + 5) ( z − 0.4 )(z − 0.6)(z − 0.8) 275.5 504 F( z) 229.5 + − = z − 0.4 z − 0.6 z − 0.8 z F( z ) = 275.5z 229.5 z 504 z + − z − 0.4 z − 0.6 z − 0.8 f (kT ) = 229.5(0.4 ) − 504(0.6 ) + 275.5(0.8) , k = 0, 1, 2, 3,. . . k b. F( z) = k k (z + 0.2)( z + 0.4) ( z − 0.1)(z − 0.5)(z − 0.9) F( z) 1.778 4.6875 7.875 4.9653 + − + =− z − 0.1 z − 0.5 z − 0.9 z z F( z ) = −1.778 + 4.6875z 7.875 z 4.9653 z − + z − 0.1 z − 0.5 z − 0.9 f (kT ) = 4.6875( 0.1) − 7.875(0.5) + 4.9653 (0.9) , k = 1 , 2, 3,. . . k c. k k ( z + 1)( z + 0.3)( z + 0.4) z ( z − 0.2)( z − 0.5)( z − 0.7) F ( z) ( z + 1)( z + 0.3)( z + 0.4) =2 z z ( z − 0.2)( z − 0.5)( z − 0.7) 38.1633 72 60 26.1633 1.7143 = − + − − z − 0.7 z − 0.5 z − 0.2 z z2 38.1633z 72 z 60 z 1.7143 F ( z) = − + − 26.1633 − z z − 0.7 z − 0.5 z − 0.2 k k k F = 38.1633(0.7) − 72(0.5) + 60(0.2) for k = 2, 3, 4,… =1 for k = 1 =0 for k = 0 F ( z) = 4. Program: '(a)' syms z k F=vpa(z*(z+3)*(z+5)/((z-0.4)*(z-0.6)*(z-0.8)),4); pretty(F) f=vpa(iztrans(F),4); pretty(f) '(b)' syms z k F=vpa((z+0.2)*(z+0.4)/((z-0.1)*(z-0.5)*(z-0.9)),4); pretty(F) Solutions to Problems 597 f=vpa(iztrans(F),4); pretty(f) '(c)' syms z k F=vpa((z+1)*(z+0.3)*(z+0.4)/(z*(z-0.2)*(z-0.5)*(z-0.7)),4); pretty(F) f=vpa(iztrans(F),4); pretty(f) Computer response: ans = (a) z (z + 3.) (z + 5.) ----------------------------------(z - .4000) (z - .6000) (z - .8000) n n n 275.5 .8000 - 504.0 .6000 + 229.5 .4000 ans = (b) (z + .2000) (z + .4000) ----------------------------------(z - .1000) (z - .5000) (z - .9000) n n n -1.778 charfcn[0](n) + 4.965 .9000 - 7.875 .5000 + 4.688 .1000 ans = (c) (z + 1.) (z + .3000) (z + .4000) ------------------------------------z (z - .2000) (z - .5000) (z - .7000) n n -1.714 charfcn[1](n) - 26.16 charfcn[0](n) + 38.16 .7000 - 72.00 .5000 n + 60.00 .2000 598 Chapter 13: Digital Control Systems 5. a. By division Instant 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 By Formula Value k 1 9.8 31.6 46.88 53.4016 53.43488 49.64608 44.043776 37.90637056 31.95798733 26.5581568 21.84639857 17.83896791 14.48905384 11.72227881 9.456567702 7.612550239 6.118437551 4.911796342 3.939668009 3.15787423 2.529983782 2.026197867 1.622284879 1.298623886 1.039376712 0.831787937 0.665602292 0.532584999 0.4261299 0.34094106 Value 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 1 9.8 31.6 46.88 53.4016 53.43488 49.64608 44.043776 37.90637056 31.95798733 26.5581568 21.84639857 17.83896791 14.48905384 11.72227881 9.456567702 7.612550239 6.118437551 4.911796342 3.939668009 3.15787423 2.529983782 2.026197867 1.622284879 1.298623886 1.039376712 0.831787937 0.665602292 0.532584999 0.4261299 0.34094106 Solutions to Problems 599 b. By division Instant 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 By Formula Value k 1 2.1 2.64 2.766 2.6859 2.51571 2.313354 2.1066276 1.90826949 1.723594881 1.554311564 1.400418494 1.261145687 1.13541564 1.022066337 0.919955834 0.828008315 0.745231516 0.670720381 0.603654351 0.54329192 0.48896423 0.440068558 0.396062078 0.356456058 0.320810546 0.288729538 0.259856608 0.233870959 0.210483869 0.189435485 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Value 1.00002 2.100018 2.6400162 2.76601458 2.685913122 2.51572181 2.313364629 2.106637166 1.908278099 1.723602629 1.554318538 1.40042477 1.261151336 1.135420724 1.022070912 0.919959951 0.828012021 0.745234852 0.670723383 0.603657053 0.543294352 0.488966419 0.440070528 0.39606385 0.356457653 0.320811982 0.28873083 0.259857771 0.233872006 0.210484811 0.189436333 600 Chapter 13: Digital Control Systems c. 6. a. G (s ) = (s + 4) 0.6667 0.3333 = + ( s + 2)( s + 5) s+2 s +5 G( z) = For T = 0.5 s, G( z) = 0.6667z 0.3333z z(z − 0.1774) + = z − 0.3679 z − 0.082085 ( z − 0.3679)( z − 0.082085) b. G (s ) = (s + 1)(s + 2) 0.1667 0.6667 1.5 = − + s (s + 3)(s + 4) s s +3 s +4 G( z) = For T = 0.5 s, G( z) = 0.6667z 0.3333z + −5 T z − e −2 T z − e 0.1667z 0.6667z 1.5z − −3T + z−1 z −e z − e−4 T 0.1667z 0.6667 z 1.5 z z(z − 0.29675)(z − 0.8408) − + = z−1 z − 0.22313 z − 0.13534 ( z − 1)(z − 0.22313)( z − 0.13534 ) Solutions to Problems 601 c. 20 1.25 1.25s + 3.57 1.25 1.25( s + 3) = − = − ( s + 3)( s + 6 s + 25) s + 3 s 2 + 6s + 25 s + 3 ( s + 3) 2 + 42 z z 2 − zae − aT cos ωT G ( z ) = −1.25 − 1.25 2 z − e − aT z − 2 ze − aT cos ωT + e −2 aT G(s) = 2 ω = 4; T = 0.5 , z z 2 + 0.0929 z G ( z ) = −1.25 − 1.25 2 z − 0.2231 z + 0.1857 z + 0.0498 z ( z + 0.2232) = 0.395 ( z − 0.2231)( z 2 + 0.1857 z + 0.0498) For a = 3; d. G ( s) = 15 0.1852 0.2083 s + 0.9978 = − + 0.02314 2 s ( s + 1)( s + 10s + 81) s s +1 s + 10s + 81 2 0.1852 0.2083 ( s + 5) − 0.5348 56 − + 0.02314 s s +1 ( s + 5) 2 + 56 z z z 2 − zae − aT cos ωT ze − aT sin ωT G ( z ) = 0.1852 − 0.2083 + 0.02314 2 − 0.0124 2 z −1 z − eβT z − 2 ze − aT cos ωT + e −2 aT z − 2 ze− aT cos ωT + e −2 aT = For a = 5; G ( z ) = 0.1852 = β =1; ω = 56 ; T = 0.5 , z z z 2 + 0.0678 z z − 0.2083 + 0.02314 2 + 0.0005748 2 z −1 z − 0.6065 z + 0.1355 z + 0.006738 z + 0.1355 z + 0.006738 0.00004 z 4 + 0.05781z 3 + 0.02344 z 2 + 0.001946 z 0.05781z 3 + 0.02344 z 2 + 0.001946 z ≈ ( z − 1)( z − 0.6065) z 2 + 0.1355 z + 0.006738 ( z − 1)( z − 0.6065) z 2 + 0.1355 z + 0.006738 ( = 0.05781 ) ( ) z 3 + 0.4055 z 2 + 0.0337 z z ( z + 0.2888)( z + 0.1167) = 0.05781 2 ( z − 1)( z − 0.6065) z + 0.1355 z + 0.006738 ( z − 1)( z − 0.6065) z 2 + 0.1355 z + 0.006738 ( ) ( 7. Program: '(a)' syms s z n T Gs=(s+4)/((s+2)*(s+5)); 'G(s)' pretty(Gs) %'g(t)' gt=ilaplace(Gs); %pretty(gt) gnT=compose(gt,n*T); %'g(kT)' %pretty(gnT) Gz=ztrans(gnT); Gz=simplify(Gz); %Construct symbolic objects for %'s', 'z', 'n',and 'T'. %Form G(s). %Display label. %Pretty print G(s). %Display label. %Find g(t). %Pretty print g(t). %Find g(nT). %Display label. %Pretty print g(nT). %Find G(z). %Simplify G(z). ) 602 Chapter 13: Digital Control Systems %'G(z)’ %pretty(Gz) Gz=subs(Gz,T,0.5); Gz=vpa(simplify(Gz),6); Gz=vpa(factor(Gz),6); 'G(z) evaluated for T=0.5' pretty(Gz) %Display label. %Pretty print G(z). %Let T = 0.5 in G(z). %Simplify G(z) and evaluate numerical %values to 6 places. %Factor G(z). %Display label. %Pretty print G(z) with numerical %values. '(b)' Gs=(s+1)*(s+2)/(s*(s+3)*(s+4)); %Form G(s) = G(s). 'G(s)' %Display label. pretty(Gs) %Pretty print G(s). %'g(t)' %Display label. gt=ilaplace(Gs); %Find g(t). %pretty(gt) %Pretty print g(t). gnT=compose(gt,n*T); %Find g(nT). %'g(kT)' %Display label. %pretty(gnT) %Pretty print g(nT). Gz=ztrans(gnT); %Find G(z). Gz=simplify(Gz); %Simplify G(z). %'G(z)’ %Display label. %pretty(Gz) %Pretty print G(z). Gz=subs(Gz,T,0.5); %Let T = 0.5 in G(z). Gz=vpa(simplify(Gz),6); %Simplify G(z) and evaluate numerical %values to 6 places. Gz=vpa(factor(Gz),6); %Factor G(z). 'G(z) evaluated for T=0.5' %Display label. pretty(Gz) %Pretty print G(z) with numerical %values. '(c)' Gs=20/((s+3)*(s^2+6*s+25)); %Form G(s) = G(s). 'G(s)' %Display label. pretty(Gs) %Pretty print G(s). %'g(t)' %Display label. gt=ilaplace(Gs); %Find g(t). %pretty(gt) %Pretty print g(t). gnT=compose(gt,n*T); %Find g(nT). %'g(kT)' %Display label. %pretty(gnT) %Pretty print g(nT). Gz=ztrans(gnT); %Find G(z). Gz=simplify(Gz); %Simplify G(z). %'G(z)’ %Display label. %pretty(Gz) %Pretty print G(z). Gz=subs(Gz,T,0.5); %Let T = 0.5 in G(z). Gz=vpa(simplify(Gz),6); %Simplify G(z) and evaluate numerical %values to 6 places. Gz=vpa(factor(Gz),6); %Factor G(z). 'G(z) evaluated for T=0.5' %Display label. pretty(Gz) %Pretty print G(z) with numerical %values. '(d)' Gs=15/(s*(s+1)*(s^2+10*s+81)); %Form G(s) = G(s). 'G(s)' %Display label. pretty(Gs) %Pretty print G(s). %'g(t)' %Display label. gt=ilaplace(Gs); %Find g(t). %pretty(gt) %Pretty print g(t). gnT=compose(gt,n*T); %Find g(nT). %'g(kT)' %Display label. %pretty(gnT) %Pretty print g(nT). Gz=ztrans(gnT); %Find G(z). Gz=simplify(Gz); %Simplify G(z). %'G(z)' %Display label. Solutions to Problems 603 %pretty(Gz) Gz=subs(Gz,T,0.5); Gz=vpa(simplify(Gz),6); Gz=vpa(factor(Gz),6); 'G(z) evaluated for T=0.5' pretty(Gz) %Pretty print G(z). %Let T = 0.5 in G(z). %Simplify G(z) and evaluate numerical %values to 6 places. %Factor G(z). %Display label. %Pretty print G(z) with numerical %values. Computer response: ans = (a) ans = G(s) s+4 --------------(s + 2) (s + 5) ans = G(z) evaluated for T=0.5 z (z - .177350) 1.00000 ---------------------------(z - .0820850) (z - .367880) ans = (b) ans = G(s) (s + 1) (s + 2) ----------------s (s + 3) (s + 4) ans = G(z) evaluated for T=0.5 z (z - .296742) (z - .840812) 1.00000 -----------------------------------(z - .135335) (z - .223130) (z - 1.) ans = ans = (c) ans = G(s) 20 ----------------------2 (s + 3) (s + 6 s + 25) ans = G(z) evaluated for T=0.5 604 Chapter 13: Digital Control Systems (z + .223130) z .394980 ----------------------------------------2 (z - .223135) (z + .185705 z + .0497861) ans = (d) ans = G(s) 15 -------------------------2 s (s + 1) (s + 10 s + 81) ans = G(z) evaluated for T=0.5 (z + .289175) (z + .116364) z .0578297 -------------------------------------------------------2 (z - .606535) (z - .999995) (z + .135489 z + .00673794) 8. a. Thus, Hence, Letting T = 0.3, b. Solutions to Problems 605 Thus, Hence, Letting T = 0.3, c. Ge ( z ) = Ga ( z )G ( z ) where Ga ( z ) is the answer to part (a) and G ( z ) , the pulse transfer function for with a zero-order-hold will now be found: Thus, Hence, 1 in cascade s+3 606 Chapter 13: Digital Control Systems Letting T = 0.3, Thus, Ge ( z ) = Ga ( z )G ( z ) = 0.19464 6.482 z + 3.964 ( z − 1)(4.482 z − 1)(2.46 z − 1) 9. a. Add phantom samplers at the input, output, and feedback path after H(s). Push G2 (s) and its input sampler to the right past the pickoff point. Add a phantom sampler after G1(s). Hence, R(s) C(s) + G (s) - G (s) 1 2 H (s) G (s) 2 G1(z)G2(z) From this block diagram, T(z) = 1 + G (z)HG (z) . 1 2 b. Add phantom samplers to the input, output, and the output of H(s). Push G1(s)G2(s) and its input sampler to the right past the pickoff point. Add a phantom sampler at the output. R(s) C(s) + G (s) G (s) 1 - H(s) G (s) G (s) 1 2 2 Solutions to Problems 607 G1G2(z) From this block diagram, T(z) = 1 + G G (z)H(z) . 12 10. Add phantom samplers after G1(s), G3(s), G4(s), H1(s), and H2(s). Push G1(s) and its sampler to the left past the summing junction. Also, push G4(s) and its input sampler to the right past the pickoff point. The resulting block diagram is, Converting to z transforms, 608 Chapter 13: Digital Control Systems G2G3 ( z ) 1 ⎡ ⎤ * ⎢ ⎥ (1 + G2G3 ( z ) H1 ( z )) (1 + H 2G4 ( z )) ⎥ C ( s) = RG1 ( z )G4 ( z ) ⎢ 1 G2G3 ( z ) ⎢1 + * H G G ( z) ⎥ ⎢ (1 + G2G3 ( z ) H1 ( z )) (1 + H 2G4 ( z )) 3 1 4 ⎥ ⎣ ⎦ = RG1 ( z )G4 ( z )G2G3 ( z ) (1 + G2G3 ( z ) H1 ( z ))(1 + H 2G4 ( z )) + G2G3 ( z ) H 3G1G4 ( z ) 11. Push gain of 2 to the left past the summing junction and add phantom samplers as shown. Push the z.o.h. and 1 to the right past the pickoff point. Also, add a phantom sampler at the s(s+1) output. Add phantom samplers after the gain of 2 at the input and in the feedback. Also, represent the z.o.h. as Laplace transforms. Solutions to Problems 609 Push the last block to the right past the pickoff point and get, Find the z transform for each transfer function. G1 ( s ) = 2 transforms into G1 ( z ) = 2 . H1 ( s ) = (1 − e − sT ) 2 2⎤ ⎡2 2 = (1 − e − sT ) ⎢ 2 − + s ( s + 1) s s + 1⎥ ⎣s ⎦ 2 transforms into z −1 ⎡ Tz z z⎤ Tz − Te −T + ze −T − z − e−T + 1 H1 ( z ) = 2 −2 +2 =2 z ⎢ ( z − 1) 2 z −1 z − e −T ⎥ ( z − 1)( z − e −T ) ⎣ ⎦ H 2 ( s ) = (1 − e − sT ) 2 2 2 2⎤ ⎡2 = (1 − e − sT ) ⎢ − − 2 + 3⎥ s ( s + 1) s⎦ ⎣ s s +1 s 3 transforms into H 2 ( z) = z −1 ⎡ 2z 2z 2Tz T 2 z ( z + 1) ⎤ − − + ⎢ ⎥ z ⎣ z − 1 z − e −T ( z − 1) 2 ( z − 1)3 ⎦ 610 Chapter 13: Digital Control Systems = (T 2 − 2e−T + 2 − 2T ) z 2 + (4e−T − 4 + 2Te−T + 2T + T 2 − T 2 e−T ) z + (2 − 2e−T − 2Te −T − T 2 e−T ) ( z − 1)2 ( z − e−T ) G2 ( s) = (1 − e − sT ) 1 s ( s + 1) 3 transforms into 1 H 2 ( z) 2 Thus, the closed-loop transfer function is ⎡ ⎤ 1 T ( z ) = G1 ( z )G2 ( z ) ⎢ ⎥ ⎣ 1 + H1 ( z ) + H 2 ( z ) ⎦ 12. z-1 G(z) = z z ⎧s2(s1+ 1)⎫ . ⎨ ⎬ ⎩ ⎭ Using Eq. (13.49) T (1 - e-T) (T-1+e-T)z + (1-e-T-Te-T) G(z) = z - 1 -T = z-e (z-1)(z-e-T) But, G(z) (T-1+e-T) z + (1-e-T-Te-T) T(z) = 1 + G(z) = 2 z + (T-2) z + (1-Te-T) The roots of the denominator are inside the unit circle for 0<T<3.923. 13. Program: numg1=10*[1 7]; deng1=poly([-1 -3 -4 -5]); G1=tf(numg1,deng1); for T=5:-.01:0; Gz=c2d(G1,T,'zoh'); Tz=feedback(Gz,1); r=pole(Tz); rm=max(abs(r)); if rm<=1; break; end; end; T r rm Computer response: T= 3.3600 r= -0.9990 -0.0461 -0.0001 Solutions to Problems 611 -0.0000 rm = 0.9990 >> T= 3.3600 r= -0.9990 -0.0461 -0.0001 -0.0000 rm = 0.9990 14. 612 Chapter 13: Digital Control Systems 15. a. First, check to see that the system is stable. Solutions to Problems 613 Since the closed-loop poles are inside the unit circle, the system is stable. Next, evaluate the static error constants and the steady-state error. K p = lim G( z ) = 0.5 z →1 e * ( ∞) = 1 2 = 1+ Kp 3 Kv = 1 lim(z − 1)G( z) = 0 T z →1 e (∞ ) = Ka = 1 2 2 lim( z − 1) G (z ) = 0 T z →1 e ( ∞) = 1 =∞ Ka e* (∞ ) = 1 =0 1 + Kp * * 1 =∞ Kv b. K p = lim G( z ) = ∞ z →1 Kv = 1 lim(z − 1)G( z) = 10 T z →1 e (∞ ) = Ka = 1 2 2 lim( z − 1) G (z ) = 0 z →1 T e ( ∞) = * * 1 = 0.1 Kv 1 =∞ Ka 614 Chapter 13: Digital Control Systems c. K p = lim G( z ) = 2.03 z →1 1 = 0.33 1+ Kp e * (∞ ) = Kv = 1 lim(z − 1)G( z) = 0 T z →1 e (∞ ) = Ka = 1 2 2 lim( z − 1) G (z ) = 0 T z →1 e ( ∞) = * * 1 =∞ Kv 1 =∞ Ka d. K p = lim G( z ) = ∞ z →1 e* (∞ ) = 1 =0 1 + Kp 1 1 * lim(z − 1)G( z) = 10 e (∞ ) = = 0.1 T z →1 Kv 1 1 2 * Ka = 2 lim( z − 1) G(z ) = 0 e ( ∞ ) = =∞ z →1 T Ka Kv = 16. Program: T=0.1; numgz=[0.04406 -0.03624 -0.03284 0.02857]; dengz=[1 -3.394 +4.29 -2.393 +0.4966]; 'G(z)' Gz=tf(numgz,dengz,0.1) 'Zeros of G(z)' zeros=roots(numgz) 'Poles of G(z)' Solutions to Problems 615 poles=roots(dengz) %Check stability Tz=feedback(Gz,1); 'Closed-Loop Poles' r=pole(Tz) M=abs(r) pause 'Find Kp' Gz=minreal(Gz,.00001); Kp=dcgain(Gz) 'Find Kv' factorkv=tf([1 -1],[1 0],0.1); %Makes transfer function %proper and yields same Kv Gzkv=factorkv*Gz; Gzkv=minreal(Gzkv,.00001); %Cancel common poles and %zeros Kv=(1/T)*dcgain(Gzkv) 'Find Ka' factorka=tf([1 -2 1],[1 0 0],0.1);%Makes transfer function %proper and yields same Ka Gzka=factorka*Gz; Gzka=minreal(Gzka,.00001); %Cancel common poles and %zeros Ka=(1/T)^2*dcgain(Gzka) Computer response: ans = G(z) Transfer function: 0.04406 z^3 - 0.03624 z^2 - 0.03284 z + 0.02857 ----------------------------------------------z^4 - 3.394 z^3 + 4.29 z^2 - 2.393 z + 0.4966 Sampling time: 0.1 ans = Zeros of G(z) zeros = -0.8753 0.8489 + 0.1419i 0.8489 - 0.1419i ans = Poles of G(z) poles = 1.0392 0.8496 + 0.0839i 0.8496 - 0.0839i 0.6557 ans = 616 Chapter 13: Digital Control Systems Closed-Loop Poles r= 0.9176 0.9176 0.7573 0.7573 + + - M= 0.9332 0.9332 0.7765 0.7765 ans = Find Kp Kp = -8.8750 ans = Find Kv Kv = 0 ans = Find Ka Ka = 0 17. First find G(z) 0.1699i 0.1699i 0.1716i 0.1716i Solutions to Problems 617 Next, plot the root locus. 18. 9.68 Step Sum Zero-Order Hold Gain 1 2 +4s+3 s Transfer Fcn Scope 618 Chapter 13: Digital Control Systems 19. Program: numgz=0.13*[1 1]; dengz=poly([1 0.74]); Gz=tf(numgz,dengz,0.1) Gzpkz=zpk(Gz) Tz=feedback(Gz,1) Ltiview Computer response: Transfer function: 0.13 z + 0.13 ------------------z^2 - 1.74 z + 0.74 Sampling time: 0.1 Zero/pole/gain: 0.13 (z+1) -------------(z-1) (z-0.74) Sampling time: 0.1 Transfer function: 0.13 z + 0.13 ------------------z^2 - 1.61 z + 0.87 Sampling time: 0.1 Solutions to Problems 619 20. Program: %Digitize G1(s) preceded by a sample and hold numg1=1; deng1=poly([-1 -3]); 'G1(s)' G1s=tf(numg1,deng1) 'G(z)' Gz=c2d(G1s,0.1,'zoh') %Input transient response specifications Po=input('Type %OS '); %Determine damping ratio z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)); %Plot root locus rlocus(Gz) zgrid(z,0) title(['Root Locus']) [K,p]=rlocfind(Gz) %Allows input by selecting point on graphic pause 'T(z)' Tz=feedback(K*Gz,1) step(Tz) Computer response: ans = G1(s) Transfer function: 620 Chapter 13: Digital Control Systems 1 ------------s^2 + 4 s + 3 ans = G(z) Transfer function: 0.004384 z + 0.003837 ---------------------z^2 - 1.646 z + 0.6703 Sampling time: 0.1 Type %OS 16.3 Select a point in the graphics window selected_point = 0.8016 + 0.2553i K= 9.7200 p= 0.8015 + 0.2553i 0.8015 - 0.2553i ans = T(z) Transfer function: 0.04262 z + 0.0373 ---------------------z^2 - 1.603 z + 0.7076 Sampling time: 0.1 Solutions to Problems 621 622 Chapter 13: Digital Control Systems 21. 13.12 25.46 deg 9 deg Solutions to Problems 623 22. z-1 1 z-1 First find G(z). G(z) = K z z{ 2 }=K z s (s+1)(s+3) z{ − 1 18 1 +1 1 −4 1 +1 1 } s +3 2 s +1 9 s 3 s 2 1z 4z 1z −1z z −1 30 9 2 18 + + − For T=0.1, G(z) = K z − 0.74082 z − 0.90484 z − 1 z − 1 2 z z + 0.24204 z + 3.3828 = 0.00015103 K . Plotting the root locus and overlaying z − 1 z − 0.74082 z − 0.90484 the 20% overshoot curve, we select the point of intersection and calculate the gain: 0.00015103K = 1.789. Thus, K = 11845.33. Finding the intersection with the unit circle yields 0.00015103K = 9.85. Thus, 0 < K < 65218.83 for stability. 23. z-1 (s+2) z-1 First find G(z). G(z) = K z z{ 2 } = K z z{ 1 1 + 1 1 − 5 1 + 2 1 } = 18 s (s+1)(s+3) s +3 2 s +1 9 s 3 s 2 1z 1z 5z 2z 18 2 −9 + +3 For T=1, G(z) = K z − 1 z − 0.049787 z − 0.36788 z − 1 z − 1 z = 0.29782K 2 z − 0.13774 z + 0.55935 . Plotting the root locus and overlaying the Ts = 15 z − 1 z − 0.049787 z − 0.36788 second circle, we select the point of intersection (0.4 + j0.63) and calculate the gain: 0.29782K = 1.6881. Thus, K = 5.668. Finding the intersection with the unit circle yields 0.29782K = 4.4. Thus, 0 < K < 14.77 for stability. 624 Chapter 13: Digital Control Systems Root Locus 1 0.5 0 -0.5 -1 -1 0 Real Axis -0.5 0.5 1 24. Substituting Eq. (13.88) into Gc(s) and letting T = 0.01 yields (z − 0.9959)(z − 0.5625) 1180z 2 − 1839z + 661.1 = 1180 Gc (z) = 2 ( z + 1)(z − 1) z −1 25. Since %OS = 10%, ζ = %OS - ln ( 100 ) 4 = 0.591. Since Ts = = 2 seconds, ζωn %OS π2 + ln2 ( 100 ) ωn = 3.383 rad/s. Hence, the location of the closed-loop poles must be –2 ± j2.729. The summation of angles from open-loop poles to –2 ± j2.729 is –192.99o. Therefore, the design point is not on the root locus. A compensator whose angular contribution is 192.99o - 180o = 12.99o is required. Assume a compensator zero at -5 canceling the pole at -5. Adding the compensator zero at –5 to the plant’s poles yields –150.690 at to –2 ± j2.729. Thus, the compensator’s pole must contribute 1800 – 150.69 = 29.310. The following geometry results. j2.729 29.31 -2 Thus, 2.729 = tan( 29.310 ) pc − 2 Solutions to Problems 625 Hence, pc = 6.86.Adding the compensator pole and zero to the system poles, the gain at the design point is found to be 124.3. Summarizing the results: Gc (s) = 124.3(s + 5) . Substituting ( s + 6.86) Eq. (13.88) into Gc(s) and letting T = 0.01 yields Gc (z) = 123.2z − 117.2 123.2(z − 0.9512) = z − 0.9337 (z − 0.9337) 26. Program: 'Design of digital lead compensation' clf %Clear graph on screen. 'Uncompensated System' %Display label. numg=1; %Generate numerator of G(s). deng=poly([0 -5 -8]); %Generate denominator of G(s). 'G(s)' %Display label. G=tf(numg,deng) %Create and display G(s). pos=input('Type desired percent overshoot '); %Input desired percent overshoot. z=-log(pos/100)/sqrt(pi^2+[log(pos/100)]^2); %Calculate damping ratio. rlocus(G) %Plot uncompensated root locus. sgrid(z,0) %Overlay desired percent overshoot %line. title(['Uncompensated Root Locus with ' , num2str(pos),... '% Overshoot Line']) %Title uncompensated root locus. [K,p]=rlocfind(G); %Generate gain, K, and closed-loop %poles, p, for point selected %interactively on the root locus. 'Closed-loop poles = ' %Display label. p %Display closed-loop poles. f=input('Give pole number that is operating point '); %Choose uncompensated system %dominant pole. 'Summary of estimated specifications for selected point on' 'uncompensated root locus' %Display label. operatingpoint=p(f) %Display uncompensated dominant %pole. gain=K %Display uncompensated gain. estimated_settling_time=4/abs(real(p(f))) %Display uncompensated settling %time. estimated_peak_time=pi/abs(imag(p(f))) %Display uncompensated peak time. estimated_percent_overshoot=pos %Display uncompensated percent %overshoot. estimated_damping_ratio=z %Display uncompensated damping %ratio. estimated_natural_frequency=sqrt(real(p(f))^2+imag(p(f))^2) %Display uncompensated natural %frequency. numkv=conv([1 0],numg); %Set up numerator to evaluate Kv. denkv=deng; %Set up denominator to evaluate Kv. sG=tf(numkv,denkv); %Create sG(s). sG=minreal(sG); %Cancel common poles and zeros. Kv=dcgain(K*sG) %Display uncompensated Kv. ess=1/Kv %Display uncompensated steady-state %error for unit ramp input. 'T(s)' %Display label. T=feedback(K*G,1) %Create and display T(s). step(T) %Plot step response of uncompensated %system. title(['Uncompensated System with ' ,num2str(pos),'% Overshoot']) %Add title to uncompensated step %response. 626 Chapter 13: Digital Control Systems ‘Press any key to go to lead compensation' %Display label. pause Ts=input('Type Desired Settling Time '); %Input desired settling time. b=input('Type Lead Compensator Zero, (s+b). b= '); %Input lead compensator zero. done=1; %Set loop flag. while done==1 %Start loop for trying lead %compensator pole. a=input('Enter a Test Lead Compensator Pole, (s+a). a = '); %Enter test lead compensator pole. numge=conv(numg,[1 b]); %Generate numerator of Gc(s)G(s). denge=conv([1 a],deng); %Generate denominator of Gc(s)G(s). Ge=tf(numge,denge); %Create Ge(s)=Gc(s)G(s). wn=4/(Ts*z); %Evaluate desired natural frequency. clf %Clear graph on screen. rlocus(Ge) %Plot compensated root locus with %test lead compensator pole. axis([-10,10,-10,10]) %Change lead-compensated root locus %axes. sgrid(z,wn) %Overlay grid on lead-compensated %root locus. title(['Lead-Compensated Root Locus with ' , num2str(pos),... '% Overshoot Line, Lead Pole at ', num2str(-a),... ' and Required Wn']) %Add title to lead-compensated root %locus. done=input('Are you done? (y=0,n=1) '); %Set loop flag. end %End loop for trying compensator %pole. [K,p]=rlocfind(Ge); %Generate gain, K, and closed-loop %poles, p, for point selected %interactively on the root locus. 'Gc(s)' %Display label. Gc=tf([1 b],[1 a]) %Display lead compensator. 'Gc(s)G(s)' %Display label. Ge %Display Gc(s)G(s). 'Closed-loop poles = ' %Display label. p %Display lead-compensated system's %closed-loop poles. f=input('Give pole number that is operating point '); %Choose lead-compensated system %dominant pole. 'Summary of estimated specifications for selected point on lead' 'compensated root locus' %Display label. operatingpoint=p(f) %Display lead-compensated dominant %pole. gain=K %Display lead-compensated gain. estimated_settling_time=4/abs(real(p(f))) %Display lead-compensated settling %time. estimated_peak_time=pi/abs(imag(p(f))) %Display lead-compensated peak time. estimated_percent_overshoot=pos %Display lead-compensated percent %overshoot. estimated_damping_ratio=z %Display lead-compensated damping %ratio. estimated_natural_frequency=sqrt(real(p(f))^2+imag(p(f))^2) %Display lead-compensated natural %frequency. s=tf([1 0],1); %Create transfer function, "s". sGe=s*Ge; %Create sGe(s) to evaluate Kv. sGe=minreal(sGe); %Cancel common poles and zeros. Kv=dcgain(K*sGe) %Display lead-compensated Kv. ess=1/Kv %Display lead-compensated steady%state error for unit ramp input. Solutions to Problems 627 'T(s)' T=feedback(K*Ge,1) %Display label. %Create and display lead-compensated %T(s). 'Press any key to continue and obtain the lead-compensated step' 'response' %Display label pause step(T) %Plot step response for lead %compensated system. title(['Lead-Compensated System with ' ,num2str(pos),'% Overshoot']) %Add title to step response of PD %compensated system. pause 'Digital design' %Print label. T=0.01 %Define sampling interval. clf %Clear graph. 'Gc(s) in polynomial form' %Print label. Gcs=K*Gc %Create Gc(s) in polynomial form. 'Gc(s) in polynomial form' %Print label. Gcszpk=zpk(Gcs) %Create Gc(s) in factored form. 'Gc(z) in polynomial form via Tustin Transformation' %Print label. Gcz=c2d(Gcs,T,'tustin') %Form Gc(z) via Tustin transformation. 'Gc(z) in factored form via Tustin Transformation' %Print label. Gczzpk=zpk(Gcz) %Show Gc(z) in factored form. 'Gp(s) in polynomial form' %Print label. Gps=G %Create Gp(s) in polynomial form. 'Gp(s) in factored form' %Print label. Gpszpk=zpk(Gps) %Create Gp(s) in factored form. 'Gp(z) in polynomial form' %Print label. Gpz=c2d(Gps,T,'zoh') %Form Gp(z) via zoh transformation. 'Gp(z) in factored form' %Print label. Gpzzpk=zpk(Gpz) %Form Gp(z) in faactored form. pole(Gpz) %Find poles of Gp(z). Gez=Gcz*Gpz; %Form Ge(z) = Gc(z)Gp(z). 'Ge(z) = Gc(z)Gp(z) in factored form' %Print label. Gezzpk=zpk(Gez) %Form Ge(z) in factored form. 'z-1' %Print label. zm1=tf([1 -1],1,T) %Form z-1. zm1Gez=minreal(zm1*Gez,.00001); %Cancel common factors. '(z-1)Ge(z)' %Print label. zm1Gezzpk=zpk(zm1Gez) %Form & display (z-1)Ge(z) in %factored form. pole(zm1Gez) %Find poles of (z-1)Ge(z). Kv=10*dcgain(zm1Gez) %Find Kv. Tz=feedback(Gez,1) %Find closed-loop %transfer function, T(z) step(Tz) %Find step reponse. title('Closed-Loop Digital Step Response') %Add title to step response. Computer response: ans = Design of digital lead compensation ans = Uncompensated System ans = G(s) 628 Chapter 13: Digital Control Systems Transfer function: 1 ------------------s^3 + 13 s^2 + 40 s Type desired percent overshoot 10 Select a point in the graphics window selected_point = -1.6435 + 2.2437i ans = Closed-loop poles = p= -9.6740 -1.6630 + 2.2492i -1.6630 - 2.2492i Give pole number that is operating point 2 ans = Summary of estimated specifications for selected point on ans = uncompensated root locus operatingpoint = -1.6630 + 2.2492i gain = 75.6925 estimated_settling_time = 2.4053 estimated_peak_time = 1.3968 estimated_percent_overshoot = 10 estimated_damping_ratio = 0.5912 estimated_natural_frequency = Solutions to Problems 629 2.7972 Kv = 1.8923 ess = 0.5285 ans = T(s) Transfer function: 75.69 --------------------------s^3 + 13 s^2 + 40 s + 75.69 ans = Press any key to go to lead compensation Type Desired Settling Time 2 Type Lead Compensator Zero, (s+b). b= 5 Enter a Test Lead Compensator Pole, (s+a). a = Are you done? (y=0,n=1) 0 Select a point in the graphics window selected_point = -1.9709 + 2.6692i ans = Gc(s) Transfer function: s+5 ------s + 6.8 ans = Gc(s)G(s) Transfer function: s+5 ---------------------------------s^4 + 19.8 s^3 + 128.4 s^2 + 272 s ans = Closed-loop poles = 6.8 630 Chapter 13: Digital Control Systems p= -10.7971 -5.0000 -2.0015 + 2.6785i -2.0015 - 2.6785i Give pole number that is operating point 3 ans = Summary of estimated specifications for selected point on lead ans = compensated root locus operatingpoint = -2.0015 + 2.6785i gain = 120.7142 estimated_settling_time = 1.9985 estimated_peak_time = 1.1729 estimated_percent_overshoot = 10 estimated_damping_ratio = 0.5912 estimated_natural_frequency = 3.3437 Kv = 2.2190 ess = 0.4507 ans = T(s) Solutions to Problems 631 Transfer function: 120.7 s + 603.6 -------------------------------------------s^4 + 19.8 s^3 + 128.4 s^2 + 392.7 s + 603.6 ans = Press any key to continue and obtain the lead-compensated step ans = response ans = Digital design T= 0.0100 ans = Gc(s) in polynomial form Transfer function: 120.7 s + 603.6 --------------s + 6.8 ans = Gc(s) in polynomial form Zero/pole/gain: 120.7142 (s+5) -------------(s+6.8) ans = Gc(z) in polynomial form via Tustin Transformation Transfer function: 119.7 z - 113.8 --------------z - 0.9342 Sampling time: 0.01 ans = Gc(z) in factored form via Tustin Transformation 632 Chapter 13: Digital Control Systems Zero/pole/gain: 119.6635 (z-0.9512) ------------------(z-0.9342) Sampling time: 0.01 ans = Gp(s) in polynomial form Transfer function: 1 ------------------s^3 + 13 s^2 + 40 s ans = Gp(s) in factored form Zero/pole/gain: 1 ------------s (s+8) (s+5) ans = Gp(z) in polynomial form Transfer function: 1.614e-007 z^2 + 6.249e-007 z + 1.512e-007 -----------------------------------------z^3 - 2.874 z^2 + 2.752 z - 0.8781 Sampling time: 0.01 ans = Gp(z) in factored form Zero/pole/gain: 1.6136e-007 (z+3.613) (z+0.2593) -------------------------------(z-1) (z-0.9512) (z-0.9231) Sampling time: 0.01 ans = 1.0000 0.9512 0.9231 ans = Ge(z) = Gc(z)Gp(z) in factored form Zero/pole/gain Solutions to Problems 633 1.9308e-005 (z+3.613) (z-0.9512) (z+0.2593) ------------------------------------------(z-1) (z-0.9512) (z-0.9342) (z-0.9231) Sampling time: 0.01 ans = z-1 Transfer function: z-1 Sampling time: 0.01 ans = (z-1)Ge(z) Zero/pole/gain: 1.9308e-005 (z+3.613) (z+0.2593) -------------------------------(z-0.9342) (z-0.9231) Sampling time: 0.01 ans = 0.9342 0.9231 Kv = 0.2219 Transfer function: 1.931e-005 z^3 + 5.641e-005 z^2 - 5.303e-005 z - 1.721e-005 ----------------------------------------------------------z^4 - 3.809 z^3 + 5.438 z^2 - 3.45 z + 0.8203 Sampling time: 0.01 634 Chapter 13: Digital Control Systems Solutions to Problems 635 + 636 Chapter 13: Digital Control Systems SOLUTIONS TO DESIGN PROBLEMS 27. a. Push negative sign from vehicle dynamics to the left past the summing junction. The computer will be the area inside the large box with the inputs and outputs shown sampled. G(s) is the combined rudder actuator and vehicle dynamics. Also, the yaw rate sensor is shown equivalently before the integrator with unity feedback. 0.25(s+0.437) where G(s) = (s+2)(s+1.29)(s+0.193) . Solutions to Design Problems 637 b. Add a phantom sampler at the output and push G(s) along with its sample and hold to the right past the pickoff point. G(s) Move the outer-loop sampler to the output of s and write the z transforms of the transfer functions. where 0.25(s+0.437) G1(s) = (1 - e-Ts) s(s+2)(s+1.29)(s+0.193) and 0.25(s+0.437) . G2(s) = (1 - e-Ts) 2 s (s+2)(s+1.29)(s+0.193) Now find the z transforms of G1(s) and G2(s). For G1(z). For G2(z): 638 Chapter 13: Digital Control Systems Now find the closed-loop transfer function. First find the equivalent forward transfer function. Thus, Ge(z) T(z) = 1 + G (z) e Substituting values, c. Using Ge(z), plot the root locus and see where it crosses the unit circle. Solutions to Design Problems 639 The root locus crosses the unit circle when 3.8642x10-5K = 5.797x10-4, or K = 15. 28. a. First find G(z). z-1 G(z) = K z z{ s 2 1 } s + 7 s + 1220 2 7 s + 3.5 − 34.4 1207.8 − 7 1 + 1220 1 ) } z-1 = K z z{6.7186x10-7( 2 2 s s s + 3.5 + 1207.8 For T = 0.1, 2 z z z z + 0.66582 z z-1 + 7.8472 + 122 −7 = K z {6.7186x107(7 2 2 z −1 z −1 z + 1.3316 z + 0.49659 z + 1.3316 z + 0.49659 G(z) = K 7.9405 ×10 −5 z + 0.63582 + 0.49355 i z + 0.63582 − 0.49355 i z − 1 z + 0.66582 + 0.2308 i z + 0.66582 − 0.2308 i 2 )} 640 Chapter 13: Digital Control Systems b. c. The root locus intersects the unit circle at -1 with a gain, 7.9405x10-5K = 10866, or 0 < K < 136.84x106. d. Program: %Digitize G1(s) preceded by a sample and hold numg1=1; deng1=[1 7 1220 0]; 'G1(s)' G1s=tf(numg1,deng1) 'G(z)' Gz=c2d(G1s,0.1,'zoh') [numgz,dengz]=tfdata(Gz,'v'); 'Zeros of G(z)' roots(numgz) 'Poles of G(z)' roots(dengz) %Plot root locus rlocus(Gz) title(['Root Locus']) [K,p]=rlocfind(Gz) Computer response: ans = G1(s) Solutions to Design Problems 641 Transfer function: 1 -------------------s^3 + 7 s^2 + 1220 s ans = G(z) Transfer function: 7.947e-005 z^2 + 0.0001008 z + 5.15e-005 ---------------------------------------z^3 + 0.3316 z^2 - 0.8351 z - 0.4966 Sampling time: 0.1 ans = Zeros of G(z) ans = -0.6345 + 0.4955i -0.6345 - 0.4955i ans = Poles of G(z) ans = 1.0000 -0.6658 + 0.2308i -0.6658 - 0.2308i Select a point in the graphics window selected_point = -0.9977 K= 1.0885e+004 p= -0.9977 -0.0995 + 0.2330i -0.0995 - 0.2330i See part (b) for root locus plot. 29. z-1 20000 z-1 a. First find G(z). G(z) = K z z{ 2 } = K z z{2 1 − 2 1 + 200 12 } s (s+100) s s + 100 s z z −1 −2 z +2 z +2 For T = 0.01, G(z) = K 2 z −1 z − 0.36788 z z −1 642 Chapter 13: Digital Control Systems = 0.73576 K z + 0.71828 . z − 1 z − 0.36788 b. Plotting the root locus. Finding the intersection with the unit circle yields 0.73576K = 1.178. Thus, 0 < K < 1.601 for stability. c. Using the root locus, we find the intersection with the 15% overshoot curve (ζ = 0.517) at 0.5955 + j0.3747 with 0.73576K = 0.24. Thus K = 0.326. d. Program: %Digitize G1(s) preceded by a sample and hold numg1=20000; deng1=[1 100 0]; 'G1(s)' G1s=tf(numg1,deng1) 'G(z)' Gz=c2d(G1s,0.01,'zoh') [numgz,dengz]=tfdata(Gz,'v'); 'Zeros of G(z)' roots(numgz) 'Poles of G(z)' roots(dengz) %Input transient response specifications Po=input('Type %OS '); %Determine damping ratio z=(-log(Po/100))/(sqrt(pi^2+log(Po/100)^2)) %Plot root locus rlocus(Gz) zgrid(z,0) title(['Root Locus']) [K,p]=rlocfind(Gz) %Allows input by selecting point on graphic. Computer response: ans = G1(s) Solutions to Design Problems 643 Transfer function: 20000 ----------s^2 + 100 s ans = G(z) Transfer function: 0.7358 z + 0.5285 ---------------------z^2 - 1.368 z + 0.3679 Sampling time: 0.01 ans = Zeros of G(z) ans = -0.7183 ans = Poles of G(z) ans = 1.0000 0.3679 Type %OS 15 z= 0.5169 Select a point in the graphics window selected_point = 0.5949 + 0.3888i K= 0.2509 p= 0.5917 + 0.3878i 0.5917 - 0.3878 644 Chapter 13: Digital Control Systems 30. G PID ( s) = 0.5857(s + 0.19)(s + 0.01) . s Substituting Eq. (13.88) with T = 1/3 second, Gc (z) = 3.632z2 − 7.028z + 3.397 3.632(z − 0.9967)(z − 0.9386) = 2 (z + 1)( z − 1) z −1 Drawing the flow diagram yields Solutions to Design Problems 645 3.632 -7.028 0 3.397 -1 T = 1/3 second 31. a. From Chapter 9, the plant without the pots and unity gain power amplifier is 64.88 (s+53.85) Gp(s) = (s2 + 8.119s + 376.3) (s^2 + 15.47s + 9283) The PID controller and notch filter with gain adjusted for replacement of pots (i.e. divided by 100) was Gc(s) = 26.82 (s+24.1) (s+0.1) (s^2 + 16.s + 9200) --------------------------------------------------------------s (s+60)^2 Thus, Ge(s) = Gp(s)Gc(s) is 1740.0816 (s+53.85)(s2 + 16s + 9200)(s+24.09)(s+0.1) Get(s) = s (s2 + 8.119s + 376.3) (s^2 + 15.47s + 9283)(s+60)2 A Bode magnitude plot of Ge(s) shows ωc = 36.375 rad/s. Thus, the maximum T should be in the range 0.15/ωc to 0.5/ωc or 4.1237e-03 to 1.3746e-02. Let us select T = 0.001. Performing a Tustin transformation on Gc(s) yields 5.166e04 z4 - 2.041e05 z3 + 3.029e05 z2 - 2.001e05 z + 4.963e04 Gc(z) = -----------------------------------------------------------------------------------------Z4 - 1.883 z3 - 0.1131 z2 + 1.883 z - 0.8869 b. Drawing the flowchart 646 Chapter 13: Digital Control Systems 5.166e4 -1.883 -2.041e5 + ++ - -- 3.029e5 -0.1131 1.883 -2.001e5 -0.8869 4.963e4 T = 0.001 c. Program: syms s 'Compensator from Chapter 9' T=.001 Gc=26.82*(s^2+16*s+9200)*(s+24.09)*(s+.1)/(s*((s+60)^2)); Gc=vpa(Gc,4); [numgc,dengc]=numden(Gc); numgc=sym2poly(numgc); dengc=sym2poly(dengc); Gc=tf(numgc,dengc); 'Gc(s)' Gczpk=zpk(Gc) 'Gc(z)' Gcz=c2d(Gc,T,'tustin') 'Gc(z)' Gczzpk=zpk(Gcz) 'Plant from Chapter 9' Solutions to Design Problems 647 Gp=64.88*(s+53.85)/[(s^2+15.47*s+9283)*(s^2+8.119*s+376.3)]; Gp=vpa(Gp,4); [numgp,dengp]=numden(Gp); numgp=sym2poly(numgp); dengp=sym2poly(dengp); 'Gp(s)' Gp=tf(numgp,dengp) 'Gp(s)' Gpzpk=zpk(Gp) 'Gp(z)' Gpz=c2d(Gp,T,'zoh') 'Gez=Gcz*Gpz' Gez=Gcz*Gpz Tz=feedback(Gez,1); t=0:T:1; step(Tz,t) pause t=0:T:50; step(Tz,t) Computer response: ans = Compensator from Chapter 9 T= 0.0010 ans = Gc(s) Zero/pole/gain: 26.82 (s+24.09) (s+0.1) (s^2 + 16s + 9198) ------------------------------------------s (s+60)^2 ans = Gc(z) Transfer function: 5.17e004 z^4 - 2.043e005 z^3 + 3.031e005 z^2 - 2.002e005 z + 4.966e004 ---------------------------------------------------------------------z^4 - 1.883 z^3 - 0.1131 z^2 + 1.883 z - 0.8869 Sampling time: 0.001 ans = Gc(z) Zero/pole/gain: 51699.4442 (z-1) (z-0.9762) (z^2 - 1.975z + 0.9842) ---------------------------------------------------(z+1) (z-1) (z-0.9417)^2 Sampling time: 0.001 ans = 648 Chapter 13: Digital Control Systems Plant from Chapter 9 ans = Gp(s) Transfer function: 64.88 s + 3494 ---------------------------------------------------s^4 + 23.59 s^3 + 9785 s^2 + 8.119e004 s + 3.493e006 ans = Gp(s) Zero/pole/gain: 64.88 (s+53.85) ---------------------------------------------(s^2 + 8.119s + 376.3) (s^2 + 15.47s + 9283) ans = Gp(z) Transfer function: 1.089e-008 z^3 + 3.355e-008 z^2 - 3.051e-008 z - 1.048e-008 ----------------------------------------------------------z^4 - 3.967 z^3 + 5.911 z^2 - 3.92 z + 0.9767 Sampling time: 0.001 ans = Gez=Gcz*Gpz Transfer function: 0.000563 z^7 - 0.0004901 z^6 - 0.005129 z^5 + 0.01368 z^4 - 0.01328 z^3 + 0.004599 z^2 + 0.0005822 z - 0.0005203 -------------------------------------------------------------------------z^8 - 5.85 z^7 + 13.27 z^6 - 12.72 z^5 - 0.6664 z^4 + 13.25 z^3 - 12.74 z^2 + 5.317 z - 0.8662 Sampling time: 0.001 Solutions to Design Problems 649 ...
View Full Document

This note was uploaded on 06/03/2008 for the course MEM 255/355 taught by Professor Yousuff during the Spring '08 term at Drexel.

Ask a homework question - tutors are online