Professional Documents
Culture Documents
1 - Skill Assesment Solutions of Control (Norman S Nice)
1 - Skill Assesment Solutions of Control (Norman S Nice)
CHAPTER 2 2.1 1 The Laplace transform of t is 2 using Table 2.1, Item 3. Using Table 2.2, Item 4, s 1 . Fs s 52 2.2 Expanding Fs by partial fractions yields: A B C D Fs s s 2 s 32 s 3 where, 5 10 B A 5 2 2 s 2s 3 S!0 9 ss 3 S!2 10 10 dFs 40 C ; and D s 32 ss 2 3 ds 9 10
S!3 s!3
Taking the inverse Laplace transform yields, 5 10 40 f t 5e2t te3t e3t 9 3 9 2.3 Taking the Laplace transform of the differential equation assuming zero initial conditions yields: s3 Cs 3s2 Cs 7sCs 5Cs s2 Rs 4sRs 3Rs
Collecting terms, s3 3s2 7s 5Cs s2 4s 3Rs Thus, Cs s2 4s 3 3 Rs s 3s2 7s 5 2.4 Gs Cross multiplying yields, d2 c dc dr 6 2c 2 r dt2 dt dt 2.5 1 s 1 A B C s2 s 4s 8 ss 4s 8 s s 4 s 8 1 1 B ; and ss 8 S!4 16 Cs 2s 1 2 Rs s 6s 2
Cs RsGs where
1 1 A S!0 32 s 4s 8 1 C ss 4 S!8 1 32 Thus, ct 2.6 Mesh Analysis Transforming the network yields,
s
1 1 1 e4t e8t 32 16 32
I9(s)
V1(s)
V(s)
+ _
Now, writing the mesh equations, s 1I1 s sI2 s I3 s Vs sI1 s 2s 1I2 s I3 s 0 I1 s I2 s s 2I3 s 0 Solving the mesh equations for I2(s), s 1 Vs 1 s 0 1 1 0 s 2 s2 2s 1Vs I2 s s 1 ss2 5s 2 s 1 s 2s 1 1 1 1 s 2 But, VL s sI2 s Hence, VL s or VL s s2 2s 1 2 Vs s 5s 2 Nodal Analysis Writing the nodal equations, 1 2 V1 s VL s Vs s 2 1 1 VL s Vs V1 s s s Solving for VL(s), 1 2 Vs s 1 1 Vs s2 2s 1Vs s VL s 1 s2 5s 2 2 1 s 2 1 1 s or VL s s2 2s 1 2 Vs s 5s 2 2.7 Inverting Gs Z2 s 100000 s Z1 s 105 =s s2 2s 1Vs s2 5s 2
2.8 Writing the equations of motion, s2 3s 1X1 s 3s 1X2 s Fs 3s 1X1 s s2 4s 1X2 s 0 Solving for X2(s), s2 3s 1 Fs 3s 1 0
3s 1Fs X2 s s2 3s 1 3s 1 ss3 7s2 5s 1 2 4s 1 3s 1 s Hence, X2 s 3s 1 3 Fs ss 7s2 5s 1 2.9 Writing the equations of motion, s2 s 1u1 s s 1u2 s Ts s 1u1 s 2s 2u2 s 0 where u1 s is the angular displacement of the inertia. Solving for u2 s, s2 s 1 Ts s 1 0 s 1Fs u2 s s2 s 1 s 1 2s3 3s2 2s 1 s 1 2s 2 From which, after simplication, u2 s 2.10 Transforming the network to one without gears by reecting the 4 N-m/rad spring to the left and multiplying by (25/50)2, we obtain,
T(t)
q1(t)
2s2
1 s1
1 N-m-s/rad 1 kg
qa(t)
1 N-m/rad
Writing the equations of motion, s2 su1 s sua s Ts su1 s s 1ua s 0 where u1 s is the angular displacement of the 1-kg inertia. Solving for ua s, s2 s Ts s sTs 0 ua s 2 3 s2 s s s s s s s 1 From which, 1 But, u2 s ua s: Thus, 2 ua s 1 Ts s2 s 1
u2 s 1=2 2 Ts s s 1 2.11 First nd the mechanical constants. 1 1 2 1 2 1 400 Jm Ja JL 5 4 400 1 1 2 1 Dm Da DL 7 5 800 5 4 400 Now nd the electrical constants. From the torque-speed equation, set vm 0 to nd stall torque and set Tm 0 to nd no-load speed. Hence, Tstall 200 vnoload 25 which, Kt Tstall 200 2 100 Ra Ea Ea 100 Kb 4 vnoload 25 Substituting all values into the motor transfer function, KT um s 1 Ra Jm 1 KT Kb 15 Ea s Dm s s s s Jm 2 Ra where um s is the angular displacement of the armature. 1 Now uL s um s. Thus, 20 uL s 1=20 15 Ea s s s 2
2.12 Letting u1 s v1 s=s u2 s v2 s=s in Eqs. 2.127, we obtain K K v1 s v2 s Ts J1 s D1 s s K K v2 s v1 s J2 s D2 s s From these equations we can draw both series and parallel analogs by considering these to be mesh or nodal equations, respectively.
J1
+
D1 1 K
J2
T(t)
D2 w2(t)
w1(t)
T(t)
J1
J2
1 D2
Parallel analog
2.13 Writing the nodal equation, C But, C1 v vo dv ir evr ev evo dv Substituting these relationships into the differential equation, dvo dv evo dv 2 it dt We now linearize ev. The general form is df dv f v f vo % dv vo (1) dv ir 2 it dt
Substituting the function, f v ev , with v vo dv yields, dev vo dv vo dv e % e dv vo Solving for evo dv, e Substituting into Eq. (1) ddv (2) evo evo dv 2 it dt Setting it 0 and letting the circuit reach steady state, the capacitor acts like an open circuit. Thus, vo vr with ir 2. But, ir evr or vr ln ir. Hence, vo ln 2 0:693. Substituting this value of vo into Eq. (2) yields ddv 2dv it dt Taking the Laplace transform, s 2dvs Is Solving for the transfer function, we obtain dvs 1 Is s2 or Vs 1 about equilibrium. Is s 2 CHAPTER 3 3.1 Identifying appropriate variables on the circuit yields
C1 iC1 v1(t) + iL L R iR C2
+
vo dv
vo(t)
iC2
(1)
Using Kirchhoffs current and voltage laws, 1 iC1 iL iR iL vL vC2 R vL vC1 vi 1 iC2 iR vL vC2 R Substituting these relationships into Eqs. (1) and simplifying yields the state equations as dvC1 1 1 1 1 vC1 iL vC2 vi RC1 C1 RC1 RC1 dt diL 1 1 vC1 vi L L dt dvC2 1 1 1 vC vC vi RC2 1 RC2 2 RC2 dt where the output equation is vo vC2 Putting the equations in vector-matrix form, 3 3 2 2 1 1 1 1 6 RC1 C1 RC1 7 6 RC1 7 7 7 6 6 7 7 6 6 1 1 7 7 6 6 _ x6 7vi t 0 0 7x 6 7 6 6 L 7 L 7 7 6 6 4 4 1 5 1 1 5 0 RC2 RC2 RC2 y 0 3.2 Writing the equations of motion s2 s 1X1 s
2
0 1 x
sX2 s
2
Fs X3 s 0
sX1 s s s 1X2 s
X2 s s s 1X3 s 0 Taking the inverse Laplace transform and simplifying, _ 1 _ 1 x1 x2 f x x 2 x1 x2 x2 x3 _ _ x 3 _ 3 x3 x2 x x Dening state variables, zi, _ _ _ z1 x1 ; z2 x1 ; z3 x2 ; z4 x2 ; z5 x3 ; z6 x3 Writing the state equations using the denition of the state variables and the inverse transform of the differential equation, z1 _ z2 _ z3 _ z4 _ z5 _ z6 _ z2 1 x _ x2 2 x _ x3 3 x _ _ 1 x1 x2 f z2 z1 z4 f x z4 _ _ x1 x2 x2 x3 z2 z4 z3 z5 z6 _ 3 x3 x2 z6 z5 z3 x
The output is z5. Hence, y z5 . In vector-matrix form, 2 3 3 2 0 0 1 0 0 0 0 617 6 1 1 0 1 0 0 7 6 7 7 6 6 7 6 0 0 0 1 0 0 7 7z 6 0 7 f t; y 0 0 6 _ z6 607 0 1 1 1 1 0 7 6 7 7 6 405 4 0 0 0 0 0 1 5 0 0 0 1 0 1 1 3.3
0 1
0 z
First derive the state equations for the transfer function without zeros. Xs 1 2 Rs s 7s 9 Cross multiplying yields s2 7s 9Xs Rs Taking the inverse Laplace transform assuming zero initial conditions, we get 7_ 9x r x x Dening the state variables as, x1 x _ x2 x Hence, _ x1 x2 _ x2 7_ 9x r 9x1 7x2 r x x Using the zeros of the transfer function, we nd the output equation to be, c 2_ x x1 2x2 x Putting all equation in vector-matrix form yields, ! ! 0 0 1 _ r x x 1 9 7 c 1 3.4 The state equation is converted to a transfer function using Gs CsI A1 B where A 4 4 ! ! 2 1:5 ; and C 1:5 ;B 0 0 1 2 x
0:625 :
10
1 s 2 s 4s 6 4 3s 5 4s 6
1:5 s4
Substituting all expressions into Eq. (1) yields Gs 3.5 Writing the differential equation we obtain d2 x 2x2 10 df t dt2 Letting x xo dx and substituting into Eq. (1) yields d 2 xo dx 2xo dx2 10 df t dt2 Now, linearize x2. xo dx from which xo dx2 x2 2xo dx o (3)
2
s2
(1)
(2)
x2 o
Substituting Eq. (3) into Eq. (1) and performing the indicated differentiation gives us the linearized intermediate differential equation, d 2 dx (4) 4xo dx 2x2 10 df t o dt2 The force of the spring at equilibrium is 10 N. Thus, since F 2x2 ; 10 2x2 from o which p xo 5 Substituting this value of xo into Eq. (4) gives us the nal linearized differential equation. p d 2 dx 4 5 dx df t dt2 Selecting the state variables, x1 dx _ x2 d x Writing the state and output equations _ x1 x2 y x1 p _ x2 x 4 5x1 df t d
11
CHAPTER 4 4.1 For a step input Cs 10s 4s 6 A B C D E ss 1s 7s 8s 10 s s 1 s 7 s 8 s 10 ct A Bet Ce7t De8t Ee10t 4.2 1 1 4 4 2:2 2:2 0:02 s; Ts 0:08 s; and Tr Since a 50; Tc a 50 a 50 a 50 0:044 s. 4.3 a. Since poles are at 6 j19:08; ct A Be6t cos19:08t f. b. Since poles are at 78:54 and 11:46; ct A Be78:54t Ce11:4t . c. Since poles are double on the real axis at 15 ct A Be15t Cte15t : d. Since poles are at j25; ct A B cos25t f. 4.4 p 400 20 and 2zvn 12; p b. vn 900 30 and 2zvn 90; p c. vn 225 15 and 2zvn 30; p d. vn 625 25 and 2zvn 0; a. vn 4.5 vn Now, T s p 361 19 and 2zvn 16; ; z 0:421: ; z 0:3 and system is underdamped . ; z 1:5 and system is overdamped . ; z 1 and system is critically damped . ; z 0 and system is undamped .
4 p 0:5 s and T p p 0:182 s. zvn vn 1 z2 From Figure 4.16, vn T r 1:4998. Therefore, Tr 0:079 s.
pzp 1z2
Finally, %os e
100 23:3%
12
4.6 a. The second-order approximation is valid, since the dominant poles have a real part of 2 and the higher-order pole is at 15, i.e. more than ve-times further. b. The second-order approximation is not valid, since the dominant poles have a real part of 1 and the higher-order pole is at 4, i.e. not more than ve-times further. 1 0:8942 1:5918 0:3023 a. ExpandingG(s) by partial fractions yields Gs . s s 20 s 10 s 6:5 But 0:3023 is not an order of magnitude less than residues of second-order terms (term 2 and 3). Therefore, a second-order approximation is not valid. 1 0:9782 1:9078 0:0704 . b. Expanding G(s) by partial fractions yields Gs s s 20 s 10 s 6:5 But 0.0704 is an order of magnitude less than residues of second-order terms (term 2 and 3). Therefore, a second-order approximation is valid. 4.8 See Figure 4.31 in the textbook for the Simulink block diagram and the output responses. 4.9 ! s 2 s5 a. Since sI A ; sI A1 2 1 3! s 5 s 5s 6 3 0 . BUs 1=s 1 ! 2 : s 4.7
Also,
1 The state vector is Xs sI A1 x0 BUs s 1s 2s 3 ! 2s2 7s 7 5s2 2s 4 . The output is Ys 1 3 Xs 2 s 1s 2s 3 s 4s 6 0:5 12 17:5 . Taking the inverse Laplace transform yields yt s1 s2 s3 0:5et 12e2t 17:5e3t . b. The eigenvalues are given by the roots of jsI Aj s2 5s 6, or 2 and 3. 4.10 a. Since sI A ! 1 s 2 s5 ; sI A1 2 2 s5 s 5s 4 2 ! 2 . Taking the s
Laplace transform of each term, the state transition matrix is given by 2 3 4 t 1 4t 2 t 2 4t e e 63e 3e 7 3 3 7: Ft 6 4 2 2 4t 1 t 4 4t 5 et e e e 3 3 3 3 2 3 4 tt 1 4tt 2 tt 2 4tt e e e e 63 7 3 3 3 7 and b. Since Ft t 6 4 2 2 4tt 1 tt 4 4tt 5 tt e e e e 3 3 3 3
13
3 2 t t 2 2t 4t ! 63e e 3e e 7 0 7: But 2t ; Ft tBut 6 4 1 e 4 2t 4t 5 t t e e e e 3 3 t Z Thus, xt Ftx0 Ft t 3 4 et e2t e4t 7 6 3 3 7: BUtdt 6 5 4 5 8 et e2t e4t 3 3
0
2 10
c. yt 2
1 x 5et e2t
CHAPTER 5 5.1 1 Combine the parallel blocks in the forward path. Then, push to the left past the pickoff s point.
s
R(s)
s2 + 1 s 1 s C(s)
Combine the parallel feedback paths and get 2s. Then, apply the feedback formula, s3 1 . simplify, and get, Ts 4 2s s2 2s 5.2 Gs 16 2 , where Find the closed-loop transfer function, Ts 1 GsHs s as 16 16 a and Hs 1. Thus, vn 4 and 2zvn a, from which z . and Gs ss a 8 % ln a 100 But, for 5% overshoot, z s 0:69. Since, z ; a 5:52. 8 % p2 ln2 100
14
R(s) +
s N2 (s)
N1 (s)
N3 (s)
N4 (s)
1 s
C(s)
N5 (s)
1 s
N6 (s)
N7 (s)
Draw nodes.
R(s) N1(s) N2(s) N5(s) N7(s) N3(s) N6(s) N4(s) C (s)
R(s) 1 N1 (s) s
N2 (s)
N3 (s)
1 N4 (s) s
C(s)
1 1 N5 (s) 1 s N6 (s)
N7 (s)
R(s)
1 s
C(s)
1 s s
15
5.4 Forward-path gains are G1G2G3 and G1G3. Loop gains are G1 G2 H1 ; G2 H2 ; and G3 H3 . Nontouching loops are G1 G2 H1 G3 H3 G1 G2 G3 H1 H3 and G2 H2 G3 H3 G2 G3 H2 H3 . Also, D 1 G1 G2 H1 G2 H2 G3 H3 G1 G2 G3 H1 H3 G2 G3 H2 H3 : Finally, D1 1 and D2 1. T k Dk Cs k yields Substituting these values into Ts Rs D G1 sG3 s1 G2 s Ts 1 G2 sH2 s G1 sG2 sH1 s1 G3 sH3 s 5.5 The state equations are, _ x1 2x1 x2 _ x2 3x2 x3 _ x3 3x1 4x2 5x3 r y x2 Drawing the signal-ow diagram from the state equations yields
1
1 s
x3
1 s
x2
1 s
x1
4 3
5.6 From Gs 100s 5 we draw the signal-ow graph in controller canonical form s2 5s 6 and add the feedback.
100
1 s
x1
1 s
x2
500
y
5 6
16
Writing the state equations from the signal-ow diagram, we obtain ! ! 105 506 1 x x r 0 1 0 y 100 500 x 5.7 From the transformation equations, P1 Taking the inverse, P Now, P1 AP ! ! ! 1 3 0:4 0:2 6:5 8:5 4 6 0:1 0:3 9:5 11:5 ! ! ! 3 3 2 1 P1 B 11 1 4 3 ! 0:4 0:2 CP 1 4 0:8 1:4 0:1 0:3 3 1 2 4 ! ! 6:5 8:5 3 _ z z u 9:5 11:5 11 y 0:8 5.8 First nd the eigenvalues. ! l 0 1 jlI Aj 0 l 4 ! 3 l 1 6 4 3 l2 5l 6 l 6 1:4 z ! 0:4 0:1 0:2 0:3 3 2 1 4 !
Therefore,
From which the eigenvalues are 2 and 3. Now use Axi lxi for each eigenvalue, l. Thus, ! ! ! x1 1 3 x1 l x2 4 6 x2 For l 2, 3x1 3x2 0 4x1 4x2 0 Thus x1 x2 For l 3 4x1 3x2 0 4x1 3x2 0
17
Thus x1 x2 and x1 0:75x2 ; from which we let ! 0:707 0:6 P 0:707 0:8 Taking the inverse yields P Hence, ! ! ! 2 0:707 0:6 1 3 4:2433 D P 0 4 6 0:707 0:8 5 ! ! ! 5:6577 4:2433 1 18:39 P1 B 5 5 3 20 ! 0:707 0:6 2:121 2:6 CP 1 4 0:707 0:8
1 1
5:6577 4:2433 5 5
5:6577 AP 5
0 3
Finally, 2 _ z 0 ! ! 0 18:39 z u 3 20
y 2:121 2:6 z
Since there are four sign changes and no complete row of zeros, there are four right halfplane poles and three left half-plane poles. 6.2 Make a Routh table. We encounter a row of zeros on the s3 row. The even polynomial is contained in the previous row as 6s4 0s2 6. Taking the derivative yields
18
24s3 0s. Replacing the row of zeros with the coefcients of the derivative yields the s3 row. We also encounter a zero in the rst column at the s2 row. We replace the zero with e and continue the table. The nal result is shown now as
s6 s5 s4 s3 s2 s1 s0 1 1 6 24 e 144=e 6 6 0 0 0 6 0 0 1 1 6 0 0 0 0 6 0 0 0 0 0 0
ROZ
There is one sign change below the even polynomial. Thus the even polynomial (4th order) has one right half-plane pole, one left half-plane pole, and 2 imaginary axis poles. From the top of the table down to the even polynomial yields one sign change. Thus, the rest of the polynomial has one right half-plane root, and one left half-plane root. The total for the system is two right half-plane poles, two left halfplane poles, and 2 imaginary poles. 6.3 Since Gs Ks 20 Gs Ks 20 ; Ts 3 2 6 Ks 20K ss 2s 3 1 Gs s 5s Form the Routh table.
s3 s2 s1 s0 1 5 30 15K 5 20K 6 K 20K
From the s1 row, K < 2. From the s0 row, K > 0. Thus, for stability, 0 < K < 2. 6.4 First nd 2 s jsI Aj 4 0 0 3 2 0 0 2 s 05 4 1 0 s 3 3 1 1 s 2 7 1 5 1 4 5 3 1 1 s 7 1 4 s 5
There are two sign changes. Thus, there are two rhp poles and one lhp pole.
19
CHAPTER 7 7.1 a. First check stability. Gs 10s2 500s 6000 10s 30s 20 3 2 1375s 6000 1 Gs s 70s s 26:03s 37:89s 6:085
Ts
Poles are in the lhp. Therefore, the system is stable. Stability also could be checked via Routh-Hurwitz using the denominator of T(s). Thus, 15ut : estep 1 15 15 0 1 lim Gs 1 1
s!0
15 15 2:1875 10 20 30 lim sGs s!0 25 35 15 30 30 1; since L15t2 3 15t2 ut : e parabola 1 2 0 s lim s Gs 15tut : eramp 1
s!0
b. First check stability. Ts Gs 10s2 500s 6000 5 1 Gs s 110s4 3875s3 4:37e04s2 500s 6000 10s 30s 20 s 50:01s 35s 25s2 7:189e 04s 0:1372
From the second-order term in the denominator, we see that the system is unstable. Instability could also be determined using the Routh-Hurwitz criteria on the denominator of T(s). Since the system is unstable, calculations about steady-state error cannot be made. 7.2 a. The system is stable, since Gs 1000s 8 1000s 8 2 1 Gs s 9s 7 1000s 8 s 1016s 8063
Ts
and Ka lim s2 Gs 0 b.
s!0
estep 1
20
eramp 1 e parabola 1
s!0
1 1 1 lim sGs 0
s!0
1 1 1 2 lim s Gs 0
7.3 System is stable for positive K. System is Type 0. Therefore, for a step input 1 12K 0:1. Solving for Kp yields K p 9 lim Gs ; from estep 1 s!0 1 Kp 14 18 which we obtain K 189. 7.4 System is stable. Since G1 s 1000, and G2 s eD 1 s 2 , s 4
7.5 1 s 1 : System is stable. Create a unity-feedback system, where He s s1 s1 The system is as follows:
R(s)
+ s s+1 Ea(s)
100 s+4
C(s)
Hence, the system is Type 0. Evaluating Kp yields Kp The steady-state error is given by estep 1 1 1 3:846e 02 1 K p 1 25
21
7.6 Since Gs Ks 7 1 ; e1 s2 2s 10 1 Kp 1 10 : 7K 10 7K 1 10
s 1 3
1 s6
!1
0 1
!#
"
0 3 !
1 6
!1
0 1
6 1 1 1 3 3
1 0
! 0 1
1 1
3 1 5 2 1 4 3 3 0
CHAPTER 8 8.1 a. 7 j9 27 j9 40:0339 5 j93 j9 7 j97 j9 37 j9 6 7 j94 j91 j9 66 j72 0:0339 j0:0899 0:096 < 110:7 944 j378
F7 j9
22
s-plane
M1
M2
M3
M4
M5
X 6
X 3
X 0
0:0339 j0:0899 0:096 <; 110:7 8.2 a. First draw the vectors.
jw X j3
j2 s-plane j1
s 3 2 1 0
j1
j2
j3
23
From the diagram, angles 180 tan1 3 3 tan1 180 108:43 108:43 180 : 1 1
b. Since the angle is 1808, the point is on the root locus. p p P pole lengths 12 32 12 32 10 c. K 1 P zero lengths 8.3 First, nd the asymptotes. sa ua poles zeros 2 4 6 0 4 #poles #zeros 30
8.4 a.
jw
j3
X s-plane
O 2
s 0 2
j3
24
b. Using the Routh-Hurwitz criteria, we rst nd the closed-loop transfer function. Gs Ks 2 2 Ts 1 Gs s K 4s 2K 13 Using the denominator of Ts, make a Routh table.
s2 s1 s0 1 K 40 2K 13 2K 13 0 0
We get a row of zeros for K 4. From the s2 row p K 4; s2 21 0. From with which we evaluate the imaginary axis crossing at 21. c. From part (b), K 4. d. Searching for the minimum gain to the left of 2 on the real axis yields 7 at a gain of 18. Thus the break-in point is at 7. e. First, draw vectors to a point e close to the complex pole.
jw s-plane j3 X
s 2 0 2
j3
At the point e close to the complex pole, the angles must add up to zero. Hence, angle th st from zero angle from pole in 4 quadrant angle from pole in 1 quadrant 180 , 3 90 u 180. Solving for the angle of departure, u 233:1. or tan1 4
25
8.5 a.
z = 0.5 X jw j4 s-plane
o
3 0 2
o
4
j4
b. Search along the imaginary axis and nd the 1808 point at s j4:06. c. For the result in part (b), K 1. d. Searching between 2 and 4 on the real axis for the minimum gain yields the break-in at s 2:89. e. Searching along z 0:5 for the 1808 point we nd s 2:42 j4:18. f. For the result in part (e), K 0:108. g. Using the result from part (c) and the root locus, K < 1. 8.6 a.
jw z = 0.591 s-plane
X 6
X 4
X 2
s
0
26
b. Searching along the z 0:591 (10% overshoot) line for the 1808 point yields 2:028 j2:768 with K 45:55. c. Ts 4 4 p p 1:97 s; T p 1:13 s; vn Tr 1:8346 from the jRej 2:028 jImj 2:768 rise-time chart and graph in Chapter 4. Since vn is the radial distance to the pole, p vn 2:0282 2:7682 3:431. Thus, Tr 0:53 s; since the system is Type 0, K 45:55 0:949. Thus, Kp 2 4 6 48 estep 1 1 0:51: 1 Kp
d. Searching the real axis to the left of 6 for the point whose gain is 45.55, we nd 7:94. Comparing this value to the real part of the dominant pole, 2:028, we nd that it is not ve times further. The second-order approximation is not valid. 8.7 Find the closed-loop transfer function and put it the form that yields pi as the root locus variable. Thus, 100 2 100 Gs 100 100 Ts s 1 Gs s2 pi s 100 s2 100 pi s 1 pi s s2 100 pi s . The following shows the root locus. Hence, KGsHs 2 s 100
jw s-plane
X j10
s
0
X j10
8.8 Following the rules for plotting the root locus of positive-feedback systems, we obtain the following root locus:
27
o 4
X 3
X 2
X 1
8.9 Ks 1 . Differentiating the The closed-loop transfer function is Ts 2 s K 2s K denominator with respect to K yields 2s Solving for @s @s @s K 2 s 1 2s K 2 s 1 0 @K @K @K @s @s s 1 K @s Ks 1 , we get . Thus, Ss:K : @K @K 2s K 2 s @K s2s K 2 10s 1 . ss 11
Now nd the closed-loop poles when K 20. From the denominator of Ts; s1;2 21:05; 0:95, when K 20. For the pole at 21:05, DK 1021:05 1 21:05 0:05 0:9975: Ds sSs:K K 21:0521:05 11 For the pole at 0:95, DK 100:95 1 0:95 0:05 0:0025: Ds sSs:K K 0:950:95 11
CHAPTER 9 9.1 a. Searching along the 15% overshoot line, we nd the point on the root locus at 3:5 j5:8 at a gain of K 45:84. Thus, for the uncompensated system, Kv lim sGs K=7 45:84=7 6:55.
s!0
Hence, eramp uncompensated 1 1=Kv 0:1527. b. Compensator zero should be 20x further to the left than the compensator pole. Arbitrarily select Gc s s 0:2 . s 0:01
c. Insert compensator and search along the 15% overshoot line and nd the root locus at 3:4 j5:63 with a gain, K 44:64. Thus, for the compensated system,
28
44:640:2 1 127:5 and eramp compensated 1 0:0078. 70:01 Kv eramp uncompensated 0:1527 19:58 d. 0:0078 eramp compensated Kv 9.2 a. Searching along the 15% overshoot line, we nd the point on the root locus at 3:5 j5:8 at a gain of K 45:84. Thus, for the uncompensated system, Ts 4 4 1:143 s: jRej 3:5
b. The real part of the design point must be three times larger than the uncompensated poles real part. Thus the design point is 33:5 j 35:8 10:5 j 17:4. The angular contribution of the plants poles and compensator zero at the design point is 130:8 . Thus, the compensator pole must contribute 180 130:8 49:2 . Using the following diagram,
jw
j17.4 s-plane
49.2 pc 10.5
we nd
17:4 tan 49:2 , from which, pc 25:52. Adding this pole, we nd pc 10:5 the gain at the design point to be K 476:3. A higher- order closed-loop pole is found to be at 11:54. This pole may not be close enough to the closed-loop zero at 10. Thus, we should simulate the system to be sure the design requirements have been met.
9.3 a. Searching along the 20% overshoot line, we nd the point on the root locus at 3:5 6:83 at a gain of K 58:9. Thus, for the uncompensated system, Ts 4 4 1:143 s: jRej 3:5
b. For the uncompensated system, Kv lim sGs K=7 58:9=7 8:41. Hence,
s!0
eramp uncompensated 1 1=Kv 0:1189. c. In order to decrease the settling time by a factor of 2, the design point is twice the uncompensated value, or 7 j 13:66. Adding the angles from the plants poles and the compensators zero at 3 to the design point, we obtain 100:8 . Thus, the compensator pole must contribute 180 100:8 79:2 . Using the following
29
diagram,
jw j13.66
s-plane
79.2 pc 7
we nd
13:66 tan79:2 , from which, pc 9:61. Adding this pole, we nd the gain pc 7 at the design point to be K 204:9. Evaluating Kv for the lead-compensated system: Kv lim sGsGlead K3=79:61 204:93=79:61 9:138:
s!0
Kv for the uncompensated system was 8.41. For a 10x improvement in steadystate error, Kv must be 8:4110 84:1. Since lead compensation gave us Kv 9:138, we need an improvement of 84:1=9:138 9:2. Thus, the lag compensator zero should be 9.2x further to the left than the compensator pole. s 0:092 . Arbitrarily select Gc s s 0:01 Using all plant and compensator poles, we nd the gain at the design point to be K 205:4. Summarizing the forward path with plant, compensator, and gain yields 205:4s 3s 0:092 Ge s : ss 79:61s 0:01 Higher-order poles are found at 0:928 and 2:6. It would be advisable to simulate the system to see if there is indeed pole-zero cancellation. 9.4 The conguration for the system is shown in the gure below.
R(s)
+
K
C(s)
Kf s
30
Minor-Loop Design: Kf . Using the following diagram, we nd s 7s 10 that the minor-loop root locus intersects the 0.7 damping ratio line at 8:5 j 8:67. The Im tan 45:57 , imaginary part was found as follows: u cos1 z 45:57 . Hence, 8:5 from which Im 8:67. For the minor loop, GsHs
jw z = 0.7
(-8.5 + j8.67) Im
s-plane X 7
X 10
8.5
The gain, Kf, is found from the vector lengths as pp K f 1:52 8:672 1:52 8:672 77:42 Major-Loop Design: Using the closed-loop poles of the minor loop, we have an equivalent forward-path transfer function of Ge s K K 2 : ss 8:5 j8:67s 8:5 j8:67 ss 17s 147:4
Using the three poles of Ge(s) as open-loop poles to plot a root locus, we search along z 0:5 and nd that the root locus intersects this damping ratio line at 4:34 j7:51 at a gain, K 626:3. 9.5 a. An active PID controller must be used. We use the circuit shown in the following gure:
Z2(s) I2(s) V1(s) Ia(s) + Vo(s)
Vi (s)
Z1(s)
I1(s)
31
Matching the given transfer function with the transfer function of the PID controller yields Gc s s 0:1s 5 s2 5:1s 0:5 0:5 s 5:1 s s 8 " 1 # R 2 C1 R C R2 C1 s 1 2 R 1 C2 s
In Eq. (2) we arbitrarily let C1 105 . Thus, R2 105 . Using these values along with Eqs. (1) and (3) we nd C2 100 mF and R1 20 kV. b. The lag-lead compensator can be implemented with the following passive network, since the ratio of the lead pole-to-zero is the inverse of the ratio of the lag pole-tozero:
R1 + vi (t) C1 R2 vo (t) C2 +
Matching the given transfer function with the transfer function of the passive lag-lead compensator yields Gc s s 0:1s 2 s 0:1s 2 2 s 0:01s 20 s 20:01s 0:2 1 1 s s R1 C1 R2 C2 1 1 1 1 s2 s R1 C1 R2 C2 R2 C1 R1 R2 C1 C2
32
(2)
Substituting Eqs. (1) and (2) in Eq. (3) yields 1 17:91 R 2 C1 Arbitrarily letting C1 100 mF in Eq. (1) yields R1 100 kV. Substituting C1 100 mF into Eq. (4) yields R2 558 kV. Substituting R2 558 kV into Eq. (2) yields C2 900 mF. CHAPTER 10 10.1 a. 1 1 ; G jv 2 j6v s 2s 4 8 v q Mv 8 v2 2 6v2 6v 1 : fv tan 8 v2 ! 6v 1 fv p tan : 8 v2 Gs
Bode Diagrams 0 20 Phase (deg); Magnitude (dB) 40 60 80 100 0 50 100 150 200 101 100 101 Frequency (rad/sec) 102
(4)
For v <
p 8, p 8,
For v < b.
33
c.
Nyquist Diagrams 0.08 0.06 0.04 0.02 0
Imaginary Axis
0.05
0.1
0.15
0.2
10.2
Asymptotic 20 dB/dec 40 40 dB/dec Actual 20 dB/dec 40 dB/dec
20 log M
60 80 100 120
0.1
10 Frequency (rad/s)
100
1000
Actual
45o/dec
Asymptotic
200 0.1 1 10 Frequency (rad/s) 100
1000
34
10.3 The frequency response is 1/8 at an angle of zero degrees at v 0. Each pole rotates 90 in going from v 0 to v 1. Thus, the resultant rotates 180 while its magnitude goes to zero. The result is shown below.
Im
w= 0
w=0
1 8
Re
10.4 a. The frequency response is 1/48 at an angle of zero degrees at v 0. Each pole rotates 90 in going from v 0 to v 1. Thus, the resultant rotates 270 while its magnitude goes to zero. The result is shown below.
Im
w = 6.63 w = 1 480 0
w=0 1 48
Re
b. Substituting jv into Gs
1 1 3 and 2 44s 48 s 2s 4s 6 s 12s 48 12v2 j44v v3 . The Nyquist diagram simplifying, we obtain G jv 6 v 56v4 784v2 2304 crosses the real axis when the imaginary part of G jv is zero. Thus, the Nyquist p diagram crosses the real axis at v2 44; or v 44 6:63 rad=s. At this fre1 . Thus, the system is stable for K < 480. quency G jv 480
35
10.5 If K 100, the Nyquist diagram will intersect the real axis at 100=480. Thus, 480 13:62 dB. From Skill-Assessment Exercise Solution 10.4, the 180 GM 20 log 100 frequency is 6.63 rad/s. 10.6 a.
60 80 100 20 log M 120 140 160 180 1 10 Frequency (rad/s) 100 1000
0 50 Phase (degrees) 100 150 200 250 300 1 10 Frequency (rad/s) 100 1000
b. The phase angle is 180 at a frequency of 36.74 rad/s. At this frequency the gain is 99:67 dB. Therefore, 20 logK 99:67, or K 96; 270. We conclude that the system is stable for K < 96; 270. c. For K 10; 000, the magnitude plot is moved up 20log10; 000 80 dB. Therefore, the gain margin is 99:67 80 19:67 dB. The 180 frequency is 36.7 rad/s. The gain curve crosses 0 dB at v 7:74 rad=s, where the phase is 87:1 . We calculate the phase margin to be 180 87:1 92:9 . 10.7 ln%=100 Using z q, we nd z 0:456, which corresponds to 20% overp2 ln2 %100 q p 4 1 2z2 4z4 4z2 2 5:79 rad=s. shoot. Using Ts 2; vBW Ts z
36
160 6750000 101250v2 j1350v2 1350v . 27 v6 2925v4 1072500v2 25000000 For a range of values for v, superimpose G jv on the a. M and N circles, and on the For both parts nd that G jv b. Nichols chart. a.
Im 3 G-plane
= 20
10.8
M = 1.3 M = 1.0
25 30
1.4 1.5
40
1.6 1.8 2.0 0.4
M = 0.7
50 70
0.5 0.6
Re
70 1 50 40 30 25 2 20
3 4 3 2
b.
Nichols Charts
0.25 dB 0.5 dB 1 dB 3 dB 6 dB 0 dB
50
100
150
200
200 dB 220 dB 350 300 250 200 150 Open-Loop Phase (deg) 100 50 0 240 dB
37
Plotting the closed-loop frequency response from a. or b. yields the following plot:
0 20
20 log M
100
1000
101
101
102
38
The open-loop frequency response is 7 at v 14:5 rad=s. Thus, the estimated bandwidth is vWB 14:5 rad=s. The open-loop frequency response plot goes through zero dB at a frequency of 9.4 rad/s, where the phase is 151:98 . Hence, the phase margin is 180 151:98 28:02 . This phase margin corresponds to z 0:25: Therefore; %OS e Ts 4 vBW z
zp=
p 2
1z
x100 44:4%;
p p Tp vBW 1 z2 10.10
The initial slope is 40 dB/dec. Therefore, the system is Type 2. The initial slope intersects 0 dB at v 9:5 rad=s. Thus, K a 9:52 90:25 and K p Kv 1. 10.11 10 10 , from which the zero dB frejv jv 1 vv j p 10 quency is found as follows: M p 1. Solving for v; v v2 1 10, or 21 v v after squaring both sides and rearranging, v4 v2 100 0. Solving for the roots, v2 10:51; 9:51. Taking the square root of the positive root, we nd the 0 dB frequency to be 3.08 rad/s. At this frequency, the phase angle, f v j 3:08 j 162 . Therefore the phase margin is 180 162 18 . b. With a delay of 0.1 s, a. Without delay, G jv f v j vT 3:08 j 3:080:1180=pi 162 17:65 179:65 : Therefore the phase margin is 180 179:65 0:35 . Thus, the system is stable. c. With a delay of 3 s, f v j vT 3:08 j 3:083180=pi 162 529:41 691:41 28:59 deg: Therefore the phase margin is 28:59 180 151:41 deg. Thus, the system is unstable. 10.12 Drawing judicially selected slopes on the magnitude and phase plot as shown below yields a rst estimate.
39
Phase(deg)
4 5 6 7 8 10
200 300
500
1000
We see an initial slope on the magnitude plot of 20 dB/dec. We also see a nal 20 dB/ dec slope with a break frequency around 21 rad/s. Thus, an initial estimate is 1 G1 s . Subtracting G1(s) from the original frequency response yields ss 21 the frequency response shown below.
Experimental Minus 1/s(s+21) 90 80 Gain(dB) Phase(deg) 70 60 50 40 100 80 60 40 20 0
4 5 6 7 8 10
20
30 40 50 70 100
200 300
500
1000
Frequency (rad/sec)
40
Drawing judicially selected slopes on the magnitude and phase plot as shown yields a nal estimate. We see rst-order zero behavior on the magnitude and phase plots with a break frequency of about 5.7 rad/s and a dc gain of about 44 dB 20 log5:7K, or K 27:8. Thus, we estimate G2 s 27:8s 7. Thus, Gs G1 sG2 s 27:8s 5:7 . It is interesting to note that the original problem was developed from ss 21 30s 5 . Gs ss 20 CHAPTER 11 11.1 The Bode plot for K 1 is shown below.
Bode Diagrams 60 80 100 120 Phase (deg); Magnitude (dB) 140 160 180
100
150
200
% log 100 A 20% overshoot requires z q 0:456. This damping ratio implies a % p2 log2 100 phase margin of 48.10, which is obtained when the phase angle 1800 48:10 131:9 . This phase angle occurs at v 27:6 rad=s. The magnitude at this frequency is 1 5:15 106 . Since the magnitude must be unity K 5:15106 194;200. 11.2 To meet the steady-state error requirement, K 1;942;000. The Bode plot for this gain is shown below.
41
100
150
200
% log 100 A 20% overshoot requires z s 0:456: This damping ratio % 2 log2 p 100 implies a phase margin of 48:1 . Adding 10 to compensate for the phase angle contribution of the lag, we use 58:1 . Thus, we look for a phase angle of 180 58:1 129:9 . The frequency at which this phase occurs is 20.4 rad/s. At this frequency the magnitude plot must go through zero dB. Presently, the magnitude plot is 23.2 dB. Therefore draw the high frequency asymptote of the lag compensator at 23:2 dB. Insert a break at 0:120:4 2:04 rad=s. At this frequency, draw 23:2 dB/dec slope until it intersects 0 dB. The frequency of intersection will be the low frequency s 2:04 break or 0.141 rad/s. Hence the compensator is Gc s Kc , where the gain s 0:141 is chosen to yield 0 dB at low frequencies, or Kc 0:141=2:04 0:0691. In summary, Gc s 0:0691 11.3 s 2:04 1;942;000 and Gs s 0:141 ss 50s 120
% log 100 A 20% overshoot requires z s 0:456. The required bandwidth is % 2 log2 p 100 q p 4 then calculated as vBW 1 2z2 4z4 4z2 2 57:9 rad/s. In order to Ts z K , we calculate meet the steady-state error requirement of Kv 50 50120 K 300;000. The uncompensated Bode plot for this gain is shown below.
42
100
150
200
The uncompensated systems phase margin measurement is taken where the magnitude plot crosses 0 dB. We nd that when the magnitude plot crosses 0 dB, the phase angle is 144:8 . Therefore, the uncompensated systems phase margin is 180 144:8 35:2 . The required phase margin based on the required damping ratio is 2z FM tan1 q 48:1 . Adding a 10 correction factor, the p 2z2 1 4z4 required phase margin is 58:1 . Hence, the compensator must contribute 1b 1 sin fmax ,b fmax 58:1 35:2 22:9 . Using fmax sin1 0:44. 1b 1 sin fmax 1 The compensators peak magnitude is calculated as Mmax p 1:51. Now nd the b frequency at which the uncompensated system has a magnitude 1=Mmax , or 3:58 dB. From the Bode plot, this magnitude occurs at vmax 50 rad=s. The compensators zero 1 1 is at zc . vmax p : Therefore, zc 33:2. T T b 1 zc The compensators pole is at pc 75:4. The compensator gain is chosen to b bT yield unity gain at dc. s 33:2 Hence, Kc 75:4=33:2 2:27. Summarizing, Gc s 2:27 , and s 75:4 300;000 Gs . ss 50s 120 % log 100 A 10% overshoot requires z s 0:591. The required bandwidth is % 2 log2 p 100 q p p then calculated as vBW p 1 2z2 4z4 4z2 2 7:53 rad=s. T p 1 z2 11.4
43
K , we calculate 830 K 2400. The uncompensated Bode plot for this gain is shown below. In order to meet the steady-state error requirement of Kv 10
Bode Diagrams 40 20 0 20 40 Phase (deg); Magnitude (dB) 60 80 100
100
150
200
Let us select a new phase-margin frequency at 0:8vBW 6:02 rad=s. The required phase margin based on the required damping ratio is FM tan1 2z q 58:6 . Adding a 5 correction factor, the required phase margin p 2z2 1 4z4 is 63:6 . At 6.02 rad/s, the new phase-margin frequency, the phase angle is which represents a phase margin of 180 138:3 41:7 . Thus, the lead compensator must contribute fmax 63:6 41:7 21:9 . 1b 1 sinfmax ,b 0:456. Using fmax sin1 1b 1 sinfmax We now design the lag compensator by rst choosing its higher break frequency one decade below the new phase-margin frequency, that is, zlag 0:602 rad=s. The lag compensators pole is plag bzlag 0:275. Finally, the lag compensators gain is Klag b 0:456. Now we design the lead compensator. The lead zero is the product of the new phase p p zlead margin frequency and b, or zlead 0:8 vBW b 4:07. Also, plead 8:93. b 1 Finally, Klead 2:19. Summarizing, b Glag s 0:456 s 0:602 ; s 0:275 Glead s 2:19 s 4:07 ; s 8:93 and K 2400:
44
CHAPTER 12 12.1 We rst nd the desired characteristic equation. A 5% overshoot requires % log p 100 z s 0:69. Also, vn p 14:47 rad=s. Thus, the T p 1 z2 % p2 log2 100 characteristic equation is s2 2zvn s v2 s2 19:97s 209:4. Adding a pole at n 10 to cancel the zero at 10 yields the desired characteristic equation,s2 19:97s 209:4s 10 s3 29:97s2 409:1s 2094. The compensated system matrix in phase-variable form is 2 3 0 1 0 5. The characteristic equation for this A BK 4 0 0 1 k1 36 k2 15 k3 system is jsI A BKj s3 15 k3 s2 36 k2 s k1 . Equating coefcients of this equation with the coefcients of the desired characteristic equation yields the gains as K k1 k2 k3 2094 373:1 14:97 : 3 2 1 1 The controllability matrix is CM B AB A2 B 4 1 4 9 5. Since 1 1 16 jCM j 80, CM is full rank, that is, rank 3. We conclude that the system is controllable. 12.2 First check controllability. The controllability matrix is CMz B AB A2 B 2 3 0 0 1 4 0 1 17 5. Since jCMz j 1, CMz is full rank, that is, rank 3. We conclude that 1 9 81 the system is controllable. We now nd the desired characteristic equation. A 20% % log 4 100 overshoot requires z s 0:456. Also, vn 4:386 rad=s. zTs % p2 log2 100 Thus, the characteristic equation is s2 2zvn s v2 s2 4s 19:24. Adding a pole n at 6 to cancel the zero at 6 yields the resulting desired characteristic equation, s2 4s 19:24s 6 s3 10s2 43:24s 115:45: s 6 s6 3 , we can write the Since Gs 2 191s 504 s 7s 8s 9 s 24s 2 3 2 3 0 1 0 0 0 1 5 ; Bp 4 0 5; Cp phase-variable representation as Ap 4 0 504 191 24 1 12.3 2
45
0 . The compensated system matrix3 phase-variable form is Ap Bp Kp in 0 1 0 4 5. The characteristic equation for this 0 0 1 504 k1 191 k2 24 k3 system is jsI Ap Bp Kp j s3 24 k3 s2 191 k2 s504 k1 . Equating coefcients of this equation with the coefcients of the desired characteristic equation yields the gains as Kp k1 k2 k3 388:55 147:76 14 . We now develop the transformation matrix to transform back to the z-system. 2 3 0 0 1 h i 1 17 5 and CMz Bz Az Bz A2 B z 4 0 z 1 9 81 2 3 0 0 1 h i CMp Bp Ap Bp A2 Bp 4 0 1 24 5: p 1 24 385 Therefore, 32 3 2 3 1 0 0 0 1 191 24 1 1 17 54 24 1 0 5 4 7 1 0 5 P CMz C1 Mx 9 81 1 0 0 56 15 1 2 3 1 0 0 Hence; Kz Kp P1 388:55 147:76 14 4 7 1 05 49 15 1 0 40 1 40:23 62:24 14 : 3 24 l1 1 0 For the given system ex A LCex 4 191 l2 0 1 5ex . The characteristic _ 504 l3 0 0 polynomial is given by jsI A LCj s3 24 l1 s2 191l2 s504 l3 . Now we nd the desired characteristic equation. The dominant poles from Skill-Assessment Exercise 12.3 come from s2 4s 19:24. Factoring yields 2 j3:9 and 2 j3:9. Increasing these poles by a factor of 10 and adding a third pole 10 times the real part of the dominant second-order poles yields the desired characteristic polynomial, s 20 j39s 20 j39s 200 s3 240s2 9921s 384200. Equating coefcients of the desired characteristic equation to the systems characteristic 2 3 216 equation yields L 4 9730 5. 383696 2 3 2 3 12.5 4 6 8 C The observability matrix is OM 4 CA 5 4 64 80 78 5, where 2 3 674 848 814 CA2 25 28 32 A2 4 7 4 11 5. The matrix is of full rank, that is, rank 3, since 77 95 94 jOM j 1576. Therefore the system is observable. 12.4 2 2
6 2 1
46
12.6 The system is represented in cascade form by the following state and output equations: 2 3 3 2 0 7 1 0 6 7 7 6 _ z 4 0 8 1 5z 4 0 5u 0 y 1 0 0 0 z 2 9 1
3 2 3 Cz 1 0 0 The observability matrix is OMz 4 Cz Az 5 4 7 1 0 5, where A2 z 2 Cz Az 49 15 1 2 3 49 15 1 1 1 4 0 64 , 17 5. Since Gs s 7s 8s 9 s3 24s2 191s 504 0 0 81 we can write the observable canonical form as 2 3 2 3 24 1 0 0 _ x 4 191 0 1 5x 4 0 5u 504 0 0 1 y 1 0 0 x 2 3 2 3 Cx 1 0 0 The observability matrix for this form is OMx 4 Cx Ax 5 4 24 1 0 5, 2 Cx Ax 385 24 1 2 3 385 24 1 where A2 4 4080 191 0 5: x 12096 504 0 We next nd the desired characteristic equation. A 10% overshoot requires % log 4 100 z s 0:591. Also, vn 67:66 rad=s. Thus, the character zTs 2 % p2 log 100 istic equation is s2 2zvn s v2 s2 80s 4578:42. Adding a pole at 400, or 10 n times the real part of the dominant second-order poles, yields the resulting desired characteristic equation, s2 80s 4578:42s 400 s3 480s2 36580s 1:831x106 . For the system represented in observable canonical form ex _ 2 3 24 l1 1 0 Ax Lx Cx ex 4 191 l2 0 1 5ex . The characteristic polynomial is given 504 l3 0 0 by jsI Ax Lx Cx j s3 24 l1 s2 191 l2 s 504 l3 . Equating coefcients of the desired characteristic equation to the systems characteristic equation 2 3 456 yields Lx 4 36; 389 5: 1; 830; 496
47
Now, develop the transformation matrix between the observer canonical and cascade forms. 3 2 31 2 1 0 0 1 0 0 7 6 7 6 P O1 OMx 4 7 1 05 1 0 5 4 24 Mz 385 24 1 15 1 32 3 2 1 0 0 1 0 0 1 0 6 76 7 6 4 7 1 0 54 24 1 0 5 4 17 1 56 15 1 385 24 1 81 9 2 49 Finally, 2 3 0 7 0 5: 1
1 0 Lz PLx 4 17 1 81 9 12.7
3 3 2 32 3 2 456 456 0 456 0 54 36; 389 5 4 28; 637 5 % 4 28; 640 5. 1; 540; 000 1; 539; 931 1 1; 830; 496
We rst nd the desired characteristic equation. A 10% overshoot requires % log 100 z s 0:591 % p2 log2 100 p Also, vn p 1:948 rad=s. Thus, the characteristic equation is s2 T p 1 z2 2 s2 2:3s 3:79. Adding a pole at 4, which corresponds to the 2zvn s vn original systems zero location, yields the resulting desired characteristic equation, s2 2:3s 3:79s 4 s3 6:3s2 13s 15:16. ! ! ! ! ! _ A BK BKe x 0 x x r; and y C 0 , Now, _ xN C 0 xN 1 xN where ! ! 0 1 0 0 A BK k1 k2 7 9 1 7 ! 0 1 7 k1 9 k2 C 4 1 ! ! 0 0 k Bke 1 e ke Thus, 2 3 2 _ x1 0 1 6 7 6 _ 4 x2 5 4 7 k1 9 k2 _ xN 4 1 32 3 2 1 3 ! 1 0 9 k1 0 k2 !
0 76 7 ke 54 x2 5 r; y 4 1 0 xN
x1
6 7 0 4 x2 5. xN
x1
48
1 9 k2 1
3 0 ke 5 0
3 0 ke 5 s3 9 k2 s2 7 k1 ke s 4ke s
Equating this polynomial to the desired characteristic equation, s3 6:3s2 13s 15:16 s3 9 k2 s2 7 k1 ke s 4ke Solving for the ks, K 2:21 2:7 and ke 3:79:
CHAPTER 13 13.1 f t sinvkT; f t sinvkTdt kT; k0 1 1 e jvkT e jvkT ekTs kTs F s sinvkTe 2j k0 k0 1 k Ts jv k 1 eTs jv e 2 j k0 1 1 But, xk 1 x1 k0 Thus,
F s ! ! 1 1 1 1 eTs e jvT eTs e jvT 2 j 1 eTs jv 1 eTs jv 2 j 1 eTs e jvT eTs e jvT e2Ts ! sinvT z1 sinvT eTs 1 eTs 2cosvT e2Ts 1 2z1 cosvT z2
1
Fz z 1z 2 z z 0:5z 0:7z 0:9 46:875 Fz 46:875 z z z 114:75 68:875 z 0:5 z 0:7 z 0:9 z z z 114:75 68:875 ; z 0:5 z 0:7 z 0:9
49
13.3
8 , Since Gs 1 eTs ss 4 & ' & ' & ' 8 z1 A B z1 2 2 1 z z : Gz 1 z z ss 4 z s s4 z s s4 2 2 Let G2 s . Therefore, g2 t 2 2e4t , or g2 kT 2 2e4kT . s s4 Hence, G2 z 2z 2z 2z1 e4T . z 1 z e4T z 1z e4T z1 21 e4T G2 z . z z e4T
Therefore, Gz
1 1:264 . For T s, Gz 4 z 0:3679 13.4 Add phantom samplers to the input, feedback after H(s), and to the output. Push G1(s)G2(s), along with its input sampler, to the right past the pickoff point and obtain the block diagram shown below.
R(s)
G1(s)G2(s)
C(s)
H(s)G1(s)G2(s)
G1 G2 z . 1 HG1 G2 z
20 Gs 20 4 4 . Let G2 s . Taking the inverse s5 s ss 5 s s 5 Laplace transform and letting t kT, g2 kT 4 4e5kT . Taking the z-transform 4z 4z 4z1 e5T yields G2 z . z 1 z e5T z 1z e5T z1 41 e5T Gz 41 e5T G2 z . Finally, Tz . 5T z 1 Gz z 5e5T 4 z e
Now, Gz
The pole of the closed-loop system is at 5e5T 4. Substituting values of T, we nd that the pole is greater than 1 if T > 0:1022 s. Hence, the system is stable for 0 < T < 0:1022 s. 13.6 Substituting z s1 into Dz z3 z2 0:5z 0:3, we obtain Ds s1 s3 8s2 27s 6. The Routh table for this polynomial is shown below.
50
Since there is one sign change, we conclude that the system has one pole outside the unit circle and two poles inside the unit circle. The table did not produce a row of zeros and thus, there are no jv poles. The system is unstable because of the pole outside the unit circle. 13.7 Dening G(s) as G1(s) in cascade with a zero-order-hold, ! ! s 3 3=20 1=4 2=5 201 eTs : Gs 201 eTs ss 4s 5 s s 4 s 5 Taking the z-transform yields ! 1=4z 2=5z 5z 1 8z 1 1 3=20z Gz 201 z : 3 4T 4T 5T z1 ze ze z e5T 1 ze Hence for T 0:1 second, K p lim Gz 3, and K v lim z 1Gz 0, and z!1 T z!1 1 2 K a 2 lim z 1 Gz 0. Checking for stability, we nd that the system is stable T z!1 Gz 1:5z 1:109 has poles inside the for T 0:1 second, since Tz 1 Gz z2 0:222z 0:703 unit circle at 0:957 and 0:735. Again, checking for stability, we nd that the system Gz 3:02z 0:6383 has is unstable for T 0:5 second, since Tz 1 Gz z2 2:802z 0:6272 poles inside and outside the unit circle at 0:208 and 3:01, respectively. 13.8 Draw the root locus superimposed over the z 0:5 curve shown below. Searching along a 54:3 line, which intersects the root locus and the z 0:5 curve, we nd the point 0:58754:3 0:348 j0:468 and K 0:31.
z-Plane Root Locus
1.5
0.5
1.5
2.5
1.5
0.5
1.5
51
100
150
200
We nd that the zero dB frequency, vFM , for Ge jv is 39 rad/s. Using Astroms guideline the value of T should be in the range, 0:15=vFM 0:0038 second to 0:5=vFM 0:0128 second. Let us use T 0:001 second. Now nd the Tustin 2z 1 into Gc s transformation for the compensator. Substituting s Tz 1 2:38s 25:3 with T 0:001 second yields s 60:2 z 0:975 : Gc z 2:34 z 0:9416 13.10
Xz 1899z2 3761z 1861 2 . Cross-multiply and obtain z2 1:908z Ez z 1:908z 0:9075 0:9075Xz 1899z2 3761z 1861Ez. Solve for the highest power of z operating on the output, X(z), and obtain z2 Xz 1899z2 3761z 1861 Gc z
52
Ez 1:908z 0:9075Xz. Solving for X(z) on the left-hand side yields Xz 1899 3761z1 1861z2 Ez 1:908z1 0:9075z2 Xz. Finally, we implement this last equation with the following ow chart:
e*(t) x*(t) 1899 Delay 0.1 second + e*(t-0.1) 3761 + + Delay 0.1 second e*(t-0.2) 1861 0.9075 + Delay 0.1 second x*(t-0.2) 1.9.08 x*(t-0.1) Delay 0.1 second