The Performance of Feedback Control Systems 5.1 Introduction 231 5.2 Testinput Signals 232 5.3. Performance of aSecond-Order System 234 5.4 Effects of a Third Pole and a Zero on the Second-Order System Response 240 5.5 Estimation of the Damping Ratio 245 5.6 The s-Plane Root Location and the Transient Response 246 5.7 The Steady-State Error of Feedback Control Systems 247 5.8 The Steady-State Error of Nonunity Feedback Systems 252 5.9 Performancelndices 254 5.10 The Simplification of Linear Systems 263 5.11 Design Example: Hubble Telescope Pointing Control 266 5.12 System Performance Using Matias 269 5.13 Sequential Design Example: Disk Drive Read System 273 5.14 Summary 277 PREVIEW The ability to adjust the transient and steady-state response of a feedback control system is a beneficial outcome of the design of control systems. One of the first steps in the design process is to specify the measures of performance. In this chapter we introduce the common. time-domain specifications such as percent overshoot, settling time, time to peak, time to rise, and steady-state tracking error. We will use selected input signals such as the step and ramp to test the response of the control system. The correlation between the system perfor- mance and the location of the system transfer function poles and zeros in the s-plane is discussed. We will develop valuable relationships between the performance specifications and the natural frequency and damping ratio for second-order systems. Relying on the no- tion of dominant poles, we can extrapolate the ideas associated with second-order systems to those of higher order. The concept of a performance index that represents a system’s performance by a single number (or index) will be considered. We will present a set of popular quantitative perfor- mance measures that adequately represent the performance of the control system. This approach will enable us to adjust the system for optimal performance. The chapter con- cludes with a performance analysis of the Sequential Design Example: Disk Drive Read System. 2305.1 Introduction 231 3.1 INTRODUCTION The ability to adjust the transient and steady-state performance is a distinct advantage of feedback control systems. To analyze and design a control system, we must define and measure its performance. Then based on the desired performance of the control system, the system parameters may be adjusted to provide the desired response. Because control systems are inherently dynamic, their performance is usually specified in terms of both the transient response and the steady-state response. The transient response is the response that disappears with time. The steady-state response is that which exists a long time fol- lowing any input signal initiation. The design specifications for control systems normally include several time-response indices for a specified input command as well as a desired steady-state accuracy. Often in the course of any design the specifications are revised to effect a compromise. Therefore specifications are seldom a rigid set of requirements, but rather a first attempt at listing a desired performance. The effective compromise and adjustment of specifications are graph- ically illustrated in Fig. 5.1. The parameter p may minimize the performance measure M, if we select p as a very small value. However this results in large measure M ,, an undesir- able situation. If the performance measures are equally important, the crossover point at Pmin Provides the best compromise. This type of compromise is normally encountered in control system design. It is clear that if the original specifications called for both M, and M, to be zero, the specifications could not be simultaneously met and would have to be altered to allow for the compromise resulting with pj, [1, 12, 17, 23]. The specifications stated in terms of the measures of performance indicate to the de- signer the quality of the system. In other words, the performance measures help to answer the question, How well does the system perform the task for which it was designed? Performance Performance 100 measure, M, measure, M 90 80 | 70 60 50 40 30 5.00 4.80 4.60 4.40 4.20 4.00 «,T, 3.80 3.60 3.40 3 + 3.20 5 0.10.2 03 04 0.5 0.6 0.7 08 09 10 FIGURE 5.1 oo ; . Two performance Damping ratio, ¢ Measures versus Parameter p. FIGURE 5.8 Percent overshoot and normalized peak time versus damping ratio ¢ for a second-order system (Eq. 5.8).232 FIGURE 5.2 Test input signals: (a) step, (b) ramp, (c) parabolic. Chapter5 The Performance of Feedback Control Systems ro re) r(t) A (a) (b) ©) 5.2 TEST INPUT SIGNALS The time-domain performance specifications are important indices because control systems are inherently time-domain systems. That is, the system transient or time performance is the response of prime interest for control systems. It is necessary to determine initially whether the system is stable, by utilizing the techniques of ensuing chapters. If the system is stable, the response to a specific input signal will provide several measures of the perfor- mance. However, because the actual input signal of the system is usually unknown, a stan- dard test input signal is normally chosen. This approach is quite useful because there is a reasonable correlation between the response of a system to a standard test input and the system’s ability to perform under normal operating conditions. Furthermore, using a stan- dard input allows the designer to compare several competing designs. Many control sys- tems experience input signals that are very similar to the standard test signals. The standard test input signals commonly used are the step input, the ramp input, and the parabolic input. These inputs are shown in Fig. 5.2. The equations representing these test signals are given in Table 5.1, where the Laplace transform can be obtained by using ‘Table 2.5. The ramp signal is the integral of the step input, and the parabola is simply the integral of the ramp input. A unit impulse function is also useful for test signal purposes. The unit impulse is based on a rectangular function f, (1) such that Ii feraf 6, -Sstss: felt) = 2 : 0, otherwise, TABLE 5.1 Test Signal Inputs Test Signal r(t) R(s) Step r@®) R(s) = Ae Ramp rt) R(s) = AL? Parabolic r(t) R(s) = 2A/5°5.2 Test Input Signals where € > 0. As € approaches zero, the function f, (t) approaches the unit impulse function 5(t), which has the following properties: fr 6()dt = 1, as S(t — a)g(t) dt = g(a). (5.1) The impulse input is useful when one considers the convolution integral for an output y(t) in terms of an input r(t), which is written as Deed f a(t — 7)r(7)dr = £-'{G(s)R(s)}. (5.2) This relationship is shown in block diagram form in Fig. 5.3. If the input is a unit impulse function, we have y(t) = f g(t — 7)d(r)dr. (5.3) The integral has a value only at 7 = 0, and therefore yQ) = go), the impulse response of the system G(s). The impulse response test signal can often be used for a dynamic system by subjecting the system to a large-amplitude, narrow-width pulse of area A. The standard test signals are of the general form re) = #, (5.4) and the Laplace transform is (5.5) Hence one may relate the response to one test signal to the response of another test signal of the form of Eq. (5.4). The step input signal is the easiest to generate and evaluate and is usually chosen for performance tests. 233234 Chapter 5 The Performance of Feedback Control Systems URE 5.3 ~ fopen-loop) stem RG) -———>} Gls) re Ms) nou: eedback). Consider the response of the system shown in Fig. 5.3 for a unit step input when Gs) = ——. 9 = G1 Then the output is Ys) = —2-— s(s + 10)" the response during the transient period is y() = 0.90 = e= 1), and the steady-state response is y@) = 0.9. If the error is E(s) = R(s) — Y(s), then the steady-state error is ey = lim sE(s) = 0.1. 0Table 2.3 Important Laplace Transform Pairs f(t) Fis) Step function, u() i : 1 é sta sin wt ae, Teoh : s ce Fea a nt ! at 5 #00) = an S*F(s) — Ff) - HF) inp function 5(+) * sin wot cos wt 4a = a)? +o) sin(wt + 6), @ = tant en sin am V1 — ft, <1 TEE inlet - 9), @ = tan’ — ie yar $n sin(w,V1 — ft + 6), b= cosy g <1 =a) +o? iP =... = EMO) F(s) 1 f% Fat na 1 ——_ (s + a)? + a? sta (sta +o? sta (stay +o 2 3? + Buns + of it s[(s + a)? + a] os 3(s* + 2fw,5 + wa) sta s[(s + a)? + w]6th page of notes (s + B/MIy) AS + Ba M¥o) Ys) = (2 + (b/M)s + k/M) 2 + 2o,5 + wy FIGURE 2.9 Ansplaneplot 2 X------- JVI =F of the poles and zeros of. Y(s). where ¢ is the dimensionless damping ratio, and w, is the natural frequency of the sys- tem. The roots of the characteristic equation are al > i rine LS 55 = —Lo, + 0, VET, (231) where, in this case, w, = VE/M and ¢ = b/(2VKM). When £ > 1, the roots are real; when g <1, the roots are complex and conjugates. When ¢ = 1, the roots are repeated and real, and the condition is called critical damping. When ¢ <1, the response is underdamped, and @ Sig = —f@, + jo,VT— & (2.32) c The s-plane plot of the poles and zeros of ¥(s) is shown in Fig. 2.9, where @ = e . cos~' ¢. As { varies with w,, constant, the complex conjugate roots follow a circular locus, t as shown in Fig. 2.10. The transient response is increasingly oscillatory as the roots ap- proach the imaginary axis when £ approaches zero. M The inverse Laplace transform can be evaluated using the graphical residue evaluation. The partial fraction expansion of Eq. (2.30) is | ¥(s) ky + (2.33) s) = +, 5 ee (s- 5) G 2 my FIGURE 2.10 ¢=0_ je FIGURE 2.11 The locus of roots lies, Evaluation of the as ¢ varies with <1 ie residue Kk, ©, constant. — / eeSince s, is the complex conjugate of s,, the residue k, is the complex conjugate of k, so that we obtain -—h_,_ 4 GaGa, where the asterisk indicates the conjugate relation. The residue k, is evaluated from Fig. 2.11 as , = Gols: + 2%.) _ (o)Mre” 1 (s) — s*) 988 Myeit2 * 1 1 ey (2.34) where M, is the magnitude of (s, + 2¢«,), and M, is the magnitude of (s; — s*).* In this case, we obtain = lenge) o 1 Qa, VT — @ ei?) 2 where @ = cos~! ¢. Therefore F giial2—6)? ae Go) wif Finally, we find that (using 8 = V1 — £2) y(t) = kes + kent ella2-0), Yo 7m, Np fente~ Fs (ella Lantg lB 4. ehtal2- Me —Cenntg—ionBt) 21-2 = veo enSet sin (w, VT = 2 + 8). (2.35) (2.36) (2.37) | || | | — w, = 1.4142, = 0.3535 | | | | 1 —_ T 0.05 —0.05 -0.10 —0.15 -0.20 0 Time (sec)5.3 PERFORMANCE OF A SECOND-ORDER SYSTEM page 235A Let us consider a single-loop second-order system and determine its response to a unit step input. A closed-loop feedback control system is shown in Fig. 5.4. The closed-loop output is Gi (agg) = —_* wy), (5.6) %9) = TT G@ S+pstK Utilizing the generalized notation of Section 2.4, we may rewrite Eq. (5.6) as ¥(s) (5.7) With a unit step input, we obtain Ys) = a (5.8) s(s? + 2fw,s + 3)" for which the transient output, as obtained from the Laplace transform table in Appen- dix A, is yQ=1- ao sino, Bt + 6), (5.9) where B = VI — £2,6 = cos~'¢, and 0 <£ < 1. The transient response of this second- order system for various values of the damping ratio ¢ is shown in Fig. 5.5. As £ decreases, the closed-loop roots approach the imaginary axis, and the response becomes increasingly oscillatory, The response as a function of £ and time is also shown in Fig. 5.5(b) for a step input. FIGURE 5.4 Closed-loop control system (with feedback).FIGURE 5.5 (a) Transient response of a second-order system (Eq. 5.9) for a step input. (b) The transient response of a second-order system (Eq. 5.9) for a step input as a function of ¢ and wt. (Courtesy of Professor R. Jacquot, University Wyoming.) 5.3 Performance of a Second-Order System y(t) Output 235Ys) = 236 Chapter 5 The Performance of Feedback Control Systems The Laplace transform of the unit impulse is R(s) = 1, and therefore the output for an impulse is CH iC) = ee 2Ca,8 + w2” (5.10) which is T(s) = Y(s)/R(s), the transfer function of the closed-loop system. The transient response for an impulse function input is then xO = gee sin w, Bt, (5.11) which is simply the derivative of the response to a step input. The impulse response of the second-order system is shown in Fig. 5.6 for several values of the damping ratio, ¢. The designer is able to select several alternative performance measures from the transient re- sponse of the system for either a step or impulse input. 2 w2 On 5 is) = = ———"*—____ (5.10) — Ys) = s¥(s) P ( RCSER Ore ae) see ed eee ore) FIGURE 5.6 10 Response of a second-order os system for an oe impulse function input. a o2l xo. ®, 0.0 -0.2 -04 -0.6 -085.3 Performance of a Second-Order System 235 wo236 Chapter5 The Performance of Feedback Control Systems Standard performance measures are usually defined in terms of the step response of a system as shown in Fig. 5.7. The swiftness of the response is measured by the rise time, T,, and the peak time, T,,. For underdamped systems with an overshoot, the 0—100% rise time is a useful index. If the system is overdamped, then the peak time is not defined, and the 10-90% rise time, T,,,is normally used. The similarity with which the actual response matches the step input is measured by the percent overshoot and settling time T,. The percent overshoot, P.O., is defined as Mp, — fv fe for a unit step input, where M,, is the peak value of the time response, and fv is the final value of the response. Normally fv is the magnitude of the input, but many systems have a P.O. = x 100% (5.12) = Ben Ts e < 0.02 z Sent | A ai (on 454 ys 1-e L4 (xt +9) | pe Bee Bout ar = 0.62 \roo2 = \7 € a5.3 Performance of a Second-Order System 237 Overshoot | wo E a Tp ie Time Tr Peak Settling FIGURE 5.7 hs time time Step response of a +—T,: K . . control system Rise time Vall, puadtoe verdanpod (Eq. 5.9). 7 7 final value significantly different from the desired input magnitude. For the system with a unit step represented by Eq. (5.8), we have fv = 1. The settling time, 7,, is defined as the time required for the system to settle within a certain percentage 6 of the input amplitude. This band of +6 is shown in Fig. 5.7. For the second-order system with closed-loop damping constant ¢w,, with a response described by Eq. (5.9), we seek to determine the time, T,, for which the response remains within 2% of the final value. This occurs approximately when e—funTs < 0.02 or Therefore we have (5.13) Hence we will define the settling time as four time constants (that is, r = I/w,) of the dominant roots of the characteristic equation. The steady-state error of the system may be measured on the step response of the system as shown in Fig. 5.7. The transient response of the system may be described in terms of two factors: 1. The swiftness of response, as represented by the rise time and the peak time. 2. The closeness of the response to the desired response, as represented by the overshoot and settling time. As nature would have it, these are contradictory requirements, and a compromise must be obtained. To obtain an explicit relation for M,, and T, as a function of £, one can differen-238 Chapter5 The Performance of Feedback Control Systems 100 : 5.00 ooh ft tt go = 80}-\~ Percent —-— 4 4.60 3 70 | \overshoor_ 440 2 90} 420 FIGURE 5.8 2 50 4.00 0,7, Percent overshoot and normalized 3.80 peak time versus damping ratio ¢ for ce aay a second-order = 20 3.40 system (Eq. 5.8). + 3.20 ———__——=—_} 3.00 0.7 08 09 1.0 Damping ratio, £ 0.0 0.102 03 04 05 tiate Eq. (5.9) and set it equal to zero. Alternatively one can utilize the differentiation prop- erty of the Laplace transform, which may be written as ay) ef a s¥(s) when the initial value of y(t) is zero. Therefore we may acquire the derivative of y(t) by multiplying Eq. (5.8) by s and thus obtaining the right side of Eq. (5.10). Taking the inverse transform of the right side of Eq. (5.10), we obtain Eq. (5.11), which is equal to zero when «, Bt = 7. Therefore we find that the peak time relationship for this second-order system is (5.14) and the peak response is M,= 1+ (5.15) Therefore the percent overshoot is ~tmVT=et i = My, ~ fv BO. = 100e eit | O40) if Vel POs ae x 100% (5.12) v ‘The percent overshoot versus the damping ratio ¢ is shown in Fig. 5.8. Also, the normalized peak time, w,T,,, is shown versus the damping ratio £ in Fig. 5.8. The percent overshoot versus the damping ratio is listed in Table 5.2 for selected values of the damping ratio. TABLE 5.2 Percent Peak Overshoot Versus Damping Ratio for a Second-Order System $$$ Damping ratio 09 0.8 07 0.6 05 0.4 03 Percent overshoot 02 15 46 9.5 16.3 25.4 37.2Te = peck tide Fabs ye) due to shefo mfuct ancl Ai flare teak. it te y Ge) ard Lek it to Zero or ¥(s) = 3 Nos tod (5.8) Ys) = s¥(s) = (ear eee (5.10) 4 @) is vetthens Wor Han the pala HoPowes— Sell yo = no sin cnt (3.1) ‘pile B yen| eo then Ain Wut =O Vpn yes tart Wupt = ——> equation (1) Thus TVe= w = SW eg oft anena38 Mer C peek a) -Swyt ya) \- ee Oo" bon Cionpt +O P= Vins? seepage 2354 peck occurs ee wo, ft = Ap __eprevious page quation (1) pa (Tre) = AR Tene + enT Ame = — sind - Font 4 (6) = pa bere ene P sO =F AA A 3 goatee [Al = ¥oxt = \re time eal. response > da xara) —- st 1-$2 4 led= \+e ee : = - SE to. = Mrs ea eo a Ce \238 Chapter5 The Performance of Feedback Control Systems (5.14) and the peak response is M,, = 1 + efi (5.15) Therefore the percent overshoot is P.O. = 100¢ ~s#/VI- (5.16) ‘The percent overshoot versus the damping ratio ¢ is shown in Fig. 5.8. Also, the normalized peak time, w,T,, is shown versus the damping ratio ¢ in Fig. 5.8. The percent overshoot versus the damping ratio is listed in Table 5.2 for selected values of the damping ratio. TABLE 5.2 Percent Peak Overshoot Versus Damping Ratio for a Second-Order System —_—_ Damping ratio 09 0.8 0.7 0.6 0.5 0.4 0.3 Percent overshoot 0.2 15 4.6 95 16.3 25.4 37.25.3 Performance of a Second-Order System 239 35 T 7 T 1 3.0 + | 25 —+— 1 oT, Linear approximation 20}— 7, = 216s 0.60 ct t IGURE 5.9 15 ALi | ob iormalized rise me T,, versus ora second-order 10 i ystem 0 01 02 0S 04 05 O06 OF 08 09 1 o Again we are confronted with a necessary compromise between the swiftness of response and the allowable overshoot. The swiftness of step response can be measured as the time it takes to rise from 10% to 90% of the magnitude of the step input. This is the definition of the rise time T,, shown in Fig, 5.7. The normalized rise time w, 7,, versus ¢ (0.05 = £ = 0.95) is shown in Fig. 5. Although it is difficult to obtain exact analytic expressions for T,,, we can utilize the linear approximation: 2.16¢ + 0.60 On (5.17) which is accurate for 0.3 = £ = 0.8. This linear approximation is shown in Fig. 5.9. ‘The swiftness of a response to a step input as described by Eq. (5.17) is dependent on £ and w,. For-a given £, the response is faster for larger @,, as shown in Fig. 5.10. Note that the overshoot is independent of a; —~ —— FIGURE 5.10 The step response ‘org = 0.2 forw,= tandw, = 10. Amplitude Time (seconds)240 Chapter5 The Performance of Feedback Control Systems rad/s, FIGURE 5.11 The step response % "os. 10~*15-~~=30~~«23~~=*S for es ae a . = Time (seconds) For a given w,,, the response is faster for lower ¢, as shown in Fig. 5.11. The swiftness of the response, however, will be limited by the overshoot that can be accepted.E5.2_ The engine, body, and tires of a racing vehicle affect the acceleration and speed attainable [11]. The speed control of the car is represented by the model shown in Fig. E5.2. (a) Calculate the steady- Engine and tires ( “. 100, Ys) ae G+ 2s +5) Speed comand (s + 2p(s + 5) Spee FIGURE E5.2 Racing car speed control. state error of the car to a step command in speed. (b) Calculate overshoot of the speed to a step command. Answer: (a) é,, = A/I 1; (b) P.O. = 20.8%oS = — = i i S42) (s#5) lL +*«#6& loo . a [+ (s+2) (#8) (s+) Csts) +1? Res : 3 +2)CS+S) a @ (2) = lin SEO c lin S G +a)Cs*s) Ag sve Se GS +2\(s+D +4190 8S = 10 A jok A —_——_ = —_—_—_ So 10 +100 }to If ys) . q 108 ec) @ 746 = (S+2)(5 #9 _ (oe _ 122 7 (5 #2968 +55 (S429 (845 4 60° loo ag 7 +119 a gts tlio = gt 2 Suse We w= dno . vo 2¥un= 7 sp $ = ie 20.3887 10 2 ales A) = aS Le RD= eS att §, a Ro,= loe Are 2 30.97 -~80/ee = ae 4 le = \+e i <= Mr = 1,329 if fv=1 p62 MrcYO _ 4324 - te —_—_—_—— «00 + 46.2% ye veE5.9 A second-order control system has. the closed- loop transfer function T(s) = Y(s)/R(s). The sys- tem specifications for a step input follow: (1) Percent overshoot S 5%. (2) Settling time < 4 seconds. (3) Peak time T,, < 1 second. Show the permissible area for the poles of T(s) in order to achieve the desired response. Use a 2% settling criterion to determine settling time.TGs) = — S74 28s 40> -T re: P.0.= looe EAT 26 & 5 5 \i-* = bn loo 2.746 —+ = 0.6 5 4 Look at the 6th page ae 4 a 4 of this set of notes Is = Bon for equations. —~ Ew. rl 7-2 =, <1 a Vi F* Wr Vi-5* > 7 S*42SwnSt+wr => $3. 47> — 5, +. jo. =t cose Ss B= CoS S = cod (0.64 = YL"5.4 EFFECTS OF A THIRD POLE AND A ZERO ON THE SECOND-ORDER SYSTEM RESPONSE The curves presented in Fig. 5.8 are exact only for the second-order system of Eq. (5.8). eo, 7 ( With a unit step input ) Ys) = 5G? + 2,5 + oF) (5.8) However, they provide a remarkably good source of data because many systems possess a dominant pair of roots, and the step response can be estimated by utilizing Fig. 5.8. This approach, although an approximation, avoids the evaluation of the inverse Laplace trans- formation in order to determine the percent overshoot and other performance measures. For example, for a third-order system with a closed-loop transfer function 1 (s? + 2¢s + Ilys + 1)’ the s-plane diagram is shown in Fig. 5.12. This third-order system is normalized with Ts) = (5.18) FIGURE 5.12 An s-plane diagram of a third- order system. jo Tm» = roots of the closed-loop / system fina how 3, 0.0 0.10.2 03 04 05 06 0.7 08 09 10°” Damping ratio, ¢ FIGURE 5.8 Percent overshoot and normalized peak time versus damping ratio ¢ for a second-order system (Eq. 5.8).5.4 Effects of a Third Pole and a Zero on the Second-Order System Response 241 TABLE 5.3 Effect of a Third Pole (Eq. 5.18) for ¢ = 0.45 ——— al Percent Settling Y y Overshoot Time* 225 0.444 0 9.63 15 0.666 3.9 63 09 LT 123 8.81 04 2.50 18.6 8.67 0.05 20.0 20.5 837 0 % 20.5 8.24 * Note: Settling time is normalized time, w,7, and utilizes a 2% criterion. @,, = 1. It was ascertained experimentally that the performance as indicated by the percent overshoot, P.O., and the settling time, T,, was represented by the second-order system curves when [4] aly toe A = l/y| = 10\Z@,|. In other words, the response of a third-order system can be approximated by the dominant roots of the second-order system as long as the real part of the dominant roots is less than Yio of the real part of the third root [17, 23]. Using a computer simulation, when ¢ = 0.45, one can determine the response of a system to a unit step input. When y = 2.25, we find that the response is overdamped because the real part of the complex poles is — 0.45, whereas the real pole is equal to —0.444. The settling time to within 2% of the final value is found via the simulation to be 9.6 seconds, If y = 0.90 or I/y = 1.11 is compared to fw, = 0.45 of the complex poles, we find that the overshoot is 12% and the settling time is 8.8 seconds. If the complex roots were dominant, we would expect the overshoot to be 20% and the settling time to be 4/fw,, = 8.9 seconds. The results are summarized in Table 5.3. We must also note that the performance measures of Fig. 5.8 are correct only for a transfer function without finite zeros. If the transfer function of a system possesses finite zeros and they are located relatively near the dominant complex poles, then the zeros will Taterially affect the transient response of the system [5].Oars = RG) S42 YOd= Sear for RGD = ent shep inpat Yee) = lim taal: = yoo A + & Sz S As+8S +282) Si A+rBz0 Bae gs: 2821 Reeve _ =e \ Y¥O* She + yay -be™ ab > ZC-€ ea) a YOO 2 7d = a = ECETNCT ES) for a>2 : ee a2 y (e) = lin 8 Laeetscexas| bat % acd=*s 2 x \ A B / (s) = ot = r CstanG+2) § gra San 8 Az | a el (34298 Ise-a ~ Casncas ~ Z-a” B= al 7 -GQ C3+adS \se-2 7 ZC a-2) s +5 \ 6 ray (s+ |s=0 2a ° 2eee eee st Yo = g-a (S+a) 2Ca-2) S*2 “2's yO zat a -2t YOe) = en ~ Zena © a 2 iho aspa veo az2or2 -2t = yas yeh 4 be £02") te. the term e* transiedts die al fact as compared fo the tremaente t em.We must also note that the performance measures of Fig. 5.8 are correct only for a transfer function without finite zeros. If the transfer function of a system possesses finite zeros and they are located relatively near the dominant complex poles, then the zeros will materially affect the transient response of the system [5]. The transient response of a system with one zero and two poles may be affected by the location of the zero [5]. The percent overshoot for a step input as a function of a/fw,, when ¢ = 1. is given in Fig. 5.13(a) for the system transfer function (w3/ay(s + a) 1) 24 eee | feeds The actual transient response for a step input is shown in Fig. 5.13(b) for selected values of a/w,,. The actual response for these selected values is summarized in Table 5.4 when $= 045. TABLE 5.4 The Response of a Second-Order System with a Zero and { = 0.45 Percent Settling Peak allw, Overshoot Time Time 3 23.1 8.0 3.0 2 39.7 7.6 2.2 1 89.9 10.1 18 05 210.0 10.3 15 ee Wiese Tine uenahl esas i a ee Note: Time is normalized as w,f and settling time is based on a 2% criterion. 5.00 4.80 5 4.60 (5.8) 2 440 @ 5 4.20 Y(s) * 3 a (s) (s? + 22,5 + w2) 4.00 wT, 3.80 3.60 FIGURE 5.8 3.40 Percent overshoot and normalized 3.20 peak time versus 0 3.00 damping ratio ¢ for 0.0 0.10.2 03 04 05 0.6 0.7 08 0.9 1.0" a second-order D: v4 system (Eq. 5.8). aE Sams242 Chapter5 The Performance of Feedback Control Systems w tO: = 20) 50 Percent overshoot (@) 100 200 Time (w,f) FIGURE 5.13 (a) Percent overshoot as a function of f and «when a second- order transfer function contains a zero. (From R. N. Clark, Introduction to Automatic Control Systems, New York, Wiley, 1962, redrawn with permission.) (b) The response for the second- order transfer function with a zero for four values of the ratio (alfan): A= 5, B=2,C=1,and D=0.5when £= 0.45.Wr ey 2 Ler Bre 2EWnS + We 2 . Ca Ti) = eo + (22)8 S242 Ens toe S24 EWS tOw if input Rak) then? 4st berm on LHS re presets the unit step tS pense- ara ferm on LHS represents the dlerivat ihe of the step esponse which is the anit wapalee ve sponse hiwes “K- Tf the zero, a>>Ewn, the zere ‘se very far to the Lekt of the S-plane , then khe transiets of the impulse pesponse will wot influence. the step response and +hus can be neglected ies the wragnibude of the \wepulse response ie multiplied by Ya Reet au ct a lange number, Obviously , a approaches Eun Cthe zero is nolonger Sar +o the \eft of the S-plane>, then the Impulse pas ponse will TWPluence Cadd tod the step response.EXAMPLE 5.1 Parameter selection A single-loop feedback control system is shown in Fig. 5.14. We desire to select the gain K end the parameter p so that the time-domain specifications will be satisfied. The transient response to a step should be as fast as is attainable while retaining an overshoot of less than 5%, Furthermore, the settling time to within 2% of the final value should be less than 4 seconds. The damping ratio £ for an overshoot of 4.3% is 0.707. This damping ratio is shown graphically as a line in Fig. 5.15. Because the settling time is 4 s—= 4s, fo, i, we require that the real part of the complex poles of T(s) be fo, = 1. This region is also shown in Fig. 5.15. The region that will satisfy both time-domain re- quirements is shown cross-hatched on the s-plane of Fig. 5.15. IGURE 5.14 ingle-loop 2edback control ystem. Ms) FIGURE 2.9 An s-plane plot ¥@) = (s + 2fe,)(¥o) of the poles and 2 + WLo,s + w2 zeros of Y(s). ey AGURE 5.15 specifications and dot locations on ne s-plane.244 Chapter 5 The Performance of Feedback Control Systems When the closed-loop roots are chosen as r, = —1 + jl and#, = —1 — jl, then we have T, = 4s and an overshoot of 4.3%. Therefore £ = 1/\/2 and w, = 1/¢ = V2. The -loop transfer function is Go) K 2 T= TS G@ 7 taps KO + Yo,s + oF Hence we require that K = @2 = 2 and p = 2¢w, = 2. A full comprehension of the correlation between the closed-loop root location and the system transient response is im- portant to the system analyst and designer. Therefore we shall consider the matter more fully in the following sections. mEXAMPLE 5.2 Dominant poles of T(s) 245 Consider a system with a closed-loop transfer function oF Y) “Hs + a) RO = T(s) ~ G+ 2fu,s + w2\1 + 75)" Both the zero and the real pole may affect the transient response. If a >> a, and 1 << {w,, then the pole and zero will have little effect on the step response. Assume that we have 62.5(s + 2.5) MS) = aE 6s + NG + 6. Note that the de gain is equal to 1 (T(0) = 1), and we expect zero steady-state error for a step input. We have {w,, = 3, 7 = 0.16, and a = 2.5. The poles and the zero are shown on the s-plane in Fig. 5.16. As a first approximation, we neglect the real pole and obtain 1) ~ 10+ 25) (s) = SO. (s? + 65 + 25) FIGURE 5.16 jo The poles and Zeros on the x i4 s-plane for a third- order system. -3 x O ° 6.25 =iLS Bs “4 FIGURE 2.9w2=25 2fw, =2XxEx5 =6 akfw,) = en 0.833 t=06 0.6x5 We now have ¢ = 0.6 and w, = 5 for dominant poles with one accompanying zero for which a4Zw,) = 0.833. Using Fig. 5.13(a), we find that the percent overshoot is 55%. We expect the settling time to within 2% of the final value to be pa “= to. OGG) 1.33.8. Using a computer simulation for the actual third-order system, we find that the percent overshoot is equal to 38% and the settling time is 1.6 seconds. Thus the effect of the damp- ing of the third pole of T(s) is to dampen the overshoot and increase the settling time (hence the real pole cannot be neglected). = 10 Se, FIGURE 5.13 (a) Percent overshoot as a function of ¢ and «when a second- order transfer | function contains a 1 2 5 10 20 50 100 200 500-1000 zero. Percent overshoot @) 1%) = (w3/a)(s + a) 2o,s + w2246 Chapter5 The Performance of Feedback Control Systems 5.6 THE s-PLANE ROOT LOCATION AND THE TRANSIENT RESPONSE ‘The transient response of a closed-loop feedback control system can be described in terms of the location of the poles of the transfer function. The closed-loop transfer function is written in general as 1) < 12 = ZPAS)AG) Rs) ACs) where A(s) = 0 is the characteristic equation of the system. For the single-loop system of Fig. 5.14, the characteristic equation reduces to 1 + G(s) = 0. It is the poles and zeros of 7(s) that determine the transient response. However, for a closed-loop system, the poles of T(s) are the roots of the characteristic equation A(s) = 0 and the poles of 2 P,(s)A ,(s). The output of a system (with gain = 1) without repeated roots and a unit step input can be formulated as a partial fraction expansion as M N y= t+ D+ Asta; Es? + 2a,5 + 621) where the A,, B, and C, are constants. The roots of the system must be either s = —o, or complex conjugate pairs such as s = —a, + jw,. Then the inverse transform results in the transient response as a sum of terms: M ” yt) = 1 + D Aen! + SD Dyew a sin (ot + 4,)5 (5.22) it = where D, is a constant and depends on B,, C,, ay and «,. The transient response is com- posed of the steady-state output, exponential terms, and damped sinusoidal terms. For the response to be stable—that is, bounded for a step input—one must require that the real part of the roots, —a; and —a,, be in the left-hand portion of the s-plane. The impulse response for various root locations is shown in Fig. 5.17. The information imparted by the location of the roots is graphic, indeed, and usually well worth the effort of determining the location of the roots in the s-plane. It is important for the control system analyst to understand the relationship between the complex-frequency representation of a linear system, through the poles and zeros of its transfer function, and its time-domain response to step and other inputs. Many of the analy- sis and design calculations in such areas as signal processing and control are done in the complex-frequency plane, where a system model is represented in terms of the poles andrf 5.7 The Steady-State Error of Feedback Control Systems jo zeros of its transfer function 7(s). On the other hand, system performance is often analyzed by examining time-domain responses, particularly when dealing with control systems. The capable system designer will be able to envision the effects on the step and im- pulse responses of adding, deleting, or moving poles and zeros of T(s) in the s-plane. Like- wise, the designer should be able to visualize what changes should be made in the poles and zeros of T(s) in order to effect desired changes in the model’s step and impulse responses. An experienced designer is aware of the effects of zero locations on system response. The poles of 7(s) determine the particular response modes that will be present, and the zeros of T(s) establish the relative weightings of the individual mode functions. For ex- ample, moving a zero closer to a specific pole will reduce the relative contribution of the mode function corresponding to the pole. A computer program can be developed that allows a person to specify arbitrary sets of poles and zeros for the transfer function of a linear system. Then the computer will evaluate and plot the system’s impulse and step responses individually. It will also display them in reduced form along with the pole—zero plot. Once the program has been run for a set of poles and zeros, the user can modify the locations of one or more of them. Plots are then presented showing the old and new poles and zeros in the complex plane and the old and new impulse and step responses. 247 FIGURE 5.17 Impulse response for various root locations in the s-plane. (The conjugate root is not shown.)5.7 THE STEADY-STATE ERROR OF FEEDBACK CONTROL SYSTEMS RS) + es) Re Table 5.5 Ce. +50) — or und eee G edbeck system. => Gls) = KT Gs42) ot Th, Cs+PR> Ke 2 lim GG) posthen error constant g20 Ky = lyn 3 GG) velo eihy error Constant ed 4% \vw 97 GG) acecltian error context Seo Kp G@@) oe oo . te a | GO | snl ke 8 | 0 | 4 Tyree | Tore! Typestep vp steady _stete error: Qe. 2 lies Sete Se \ eS seo THGG)S 1 + Go) Ke \s ded ied by Ke = Mme Gis) = GOOD Smo 2 Ess = \+ Ke For a type 0 system, : KTG+t2) _ Ke = Lee ars end \ For a type Aor Vagher system 5 Kee lim —KTGHZD 2 co (N24) Pr go sf TCS+PO Summary , Gor a unt step mpl, Cas Ther type O Systems Egg = O Soe type h or higher systemsStebie Velociky Error Coelhicvect Ky unt Taymp pak, steady-state error: : ' \ ess = iw S$ —li—.t = lw 36) geo V+G(s) 8% so Ko Ss detined 2 Tyree | Tres Tare ky = bie SGD eG | oe oe | ky 0 | GO _ ; KN G+z) _ Fora type © system, Ky = im 8 ee Sto eles KEGtZD =k, Fer a type L system, S>o g Tt CS+ PH) For a type 2 KUNCS*Z) _ gg (nz2) : a\lws or higher system, Ky Ne Sar Gstr Summary , See a uwk ramp vipat, Ces = ae = co Coes type } system Cc = et Sor type t system 2 Ky Ki Lis o Gor type 2 oF higher system = EssStekic Acceleration Error Cocfficiet Ke Unit — parabolic input Caccelerection wept), steacy-s ate error: é = lms _! a a \ _—! 5 Smo \+GQcs) 3? — Mie s* G(s) Ka \s defied by: Kas \w ae s7o ' i 1 €ss = Ke ‘or e tom = lms” KMGt2ZD 2 FE. a typ O systom, Ka |e 7 eaves ° lua S7_KEG*Z9) 2 Fer a bape b system, est ie so sm (s+PY hy wet KIT CSHZ K For atype 2 ayston, Kas We S! (sepa Vor otyre Se tw S KTM S+2 (NZS) or higher system, seo NT (S+Pk? Summary , Sor a umit parabolic inp, sg = Of fee type O ond type 4 systems ot Css = x — type 2 systems éss= 0 foe dupe 3 or highee systems250 ep Kv Ka Chapter 5 The Performance of Feedback Control Systems TABLE 5.5 Summary of Steady-State Errors Number of In- Input tegrations in G(s), Type Step, r(t) = A, Ramp, At, Parabola, Number R(s) = A/s Als? At?/2, A/s® A “ 7 0 ly = Infinite Infinite 1+K, 1 es = 0 z Infinite A = 0 & 2 es =0 x \Typeo Tare! Ture (GO | oe | Teo Gl 9 ° GO oo Tope! 2 | GED OO 0 GO an | @ | 6 { ° ‘The steady-state error is infinite for one integration; for two integrations, N = 2, we obtain A A 6. = = =, (5.31) KTTz./TIp, Ke where K, is designated the acceleration error constant. The acceleration error constant is K, = lim s?G(s). 50 When the number of integrations equals or exceeds three, then the steady-state error of the system is zero. Control systems are often described in terms of their type number and the error con- stants, K,, K,, and K,. Definitions for the error constants and the steady-state error for the three inputs are summarized in Table 5.5. The usefulness of the error constants can be illustrated by considering a simple example.A low-inertia plotter is shown in Fig. E5.5(a). This system may be represented by the block dia- gram shown in E5.5(b) [18]. (a) Calculate the Rs) Ys) Position Velocity Position feedback (b) FIGURE E5.5 (a) The Hewlett-Packard x-y-plotter, (Courtesy of Hewlett-Packard Co.) (b) Block diagram of plotter. steady-state error for a ramp input. (b) Select a value of K that will result in zero overshoot to a step input but as rapid response as is attainable. Plot the poles and zeros of this system and dis- cuss the dominance of the complex poles. What overshoot for a step input do you expect?pot lye 2 5 a: oe ESS RO + Cea] Y¢s) posit. part (a) applying masons loop tale | ¢ 100 Ye =T@-2 8 seg loo = _ loo | _ {oo = 2 iG 237 KS Bs $*+ 100KS +100 Qa = bins [ R= NAN hws Lt-Te] RS) = lw sti-TqT 4 4. soe Li-T] St Bie RED= Se ep pt = \iw \ - —2? | 4 ae Gaara cae Na le | AGES Stoo KS \4 See St¢iookS ti00 1S eal —Stivek lA = $70 St loo KS +100 =RID + a yes) ace = st loo \o al i \ a KS $CStlook) Tw 4 system! A sre J 4 Take | Met Mion 6a) step wepat — > @ss70 eas Aate ramp spt Ess = whee Ky= lms Ged smo \an Ss lsan nie S CS + loo a K Ku . ss = AKE5.5 A low-inertia plotter is shown in Fig. ES.5(a). This system may be represented by the block dia- gram shown in Fig. E5.5(b) [18]. (a) Calculate the x Ys Rs) Position Velocity Position feedback (b) FIGURE ES5.5 (a) The Hewlett-Packard x--plotter. (Courtesy of Hewlett-Packard Co.) (b) Block diagram of plotter. steady-state error for a ramp input. (b) Select a value of K that will result in zero overshoot to a step input but as rapid response as is attainable. loo YO. Ti) = se ee jt Retin BD 190 S*+ 100KS +1loob) Sy 10eKS +100 = S+2E OS +00 ZEW, = lOOK ee look _ JooKk ZwWn Zxlo SK Fer zero ovesheot, te Aystem shold be oni haath, derpeel eee Keg = oe The closer loop system hes no zeros and tHe poles are: $1,S2= ~ 50K zy Clook}* = 4xloo NON. 2 ws za = -SokK + Jo V2sK*-1 ake percest ovtshat to a step ype is LS P.o. = loo exp =e" = 100 exp | SE ean on o< 5 =| Vy 25k O< KS 0-2 Aud Po.=o co K2Z0.2E5.13_ A feedback system is shown in Fig. E5.13. (a) Determine the steady-state error for a unit step when K = 0.4 and G,(s) = 1. (b) Select an appropriate value for G ‘p(S) So that the steady-state error is equal to zero for the unit step input. K Ris) 4} G(s) WtD Ys) (s +3) (s+0.1) FIGURE E5.13 Feedback system. Ris) 4 Y(s) 6.7 A negative feedback system has a loop transfer function Kis 2) 'H(s) = 5 GH(s) us) a) Find the value of the gain when the ¢ of the ‘osed-loop roots is equal to 0.707. (b) Find the value of the gain when the closed-loop system has wo roots on the imaginary axis.ESAS . 82 feel — > “ y¢s) - $CS42) a) Gp=t Keo.d4 delemmne Css Sor RODE EE K Tis) = YO Gp * Scs4ar RS) ~ :) S+3 \+ (sem ao TS = Cr $(se2Csrony 4 KOS 43) esse Ins Le-y] = Vows [v-T]R 2 bie L1-7] for REVS Gpk Cod C.F SS $5 BK Sue Geel Kt0-4 Ess= |= 0.033 = 0.967 b) Oe ee SO O=\- Ge o> Gp = 30 cei2/,yes) : isi a RG) GK 4s) G9 OED Ys) ) — th $($42) (s +3) (s+0.1) R(s) Y(s) = S Rate Ya) LH | ] = Lpuivalink to SD +t aay ob = rime . a = __s_ = S$ Cs4e> Sp(s+0.1) ay “$43 Se S40 \+6CR-S) \+ Sli Gp (S400) Gp S401) _ Qt kK Cs +000 Ge Ges (S40. (S42) + Gp kL sCar- + 3-01 Ge | t= GeK (sto. S (gro. 1)CS42 + KES + 3-165 | nonce * type b system, rot type “Type” of systen. dat oy apples to "Unity Feedbsok spstems. eci2/\ —— fr new ed ess = \+ Kp Ke= lin Gos) S2o Zw >= 0-1 GeK KL 43-016, | Gi = St — = 0.0348 2-4 l 0.467 Ess | + 0034S b) Coe Ess =O : ; | _ 3-0.1 Gp ss eee : 2-0.1GP Oo > -ol Gp Gp= 30252 Chapter5 The Performance of Feedback Control Systems 5.8 THE STEADY-STATE ERROR OF NONUNITY FEEDBACK SYSTEMS A general feedback system with nonunity feedback is shown in Fig. 5.21 with H(s) not equal to unity. For a system in which the feedback is not unity, the units of the output, ¥(s), are usually different from the output of the sensor, For example, a speed control system is shown in Fig. 5.22, where H(s) = K. The constants K, and K, account for the conversion of one set of units to another set of units (here we convert rad/s to volts). We can select K, and thus we set K, = K, and move the block for K, and K, past the summing node. Then we obtain the equivalent block diagram shown in Fig. 5.23. Thus we obtain a unity feed- back system as desired. Let us return to the system of Fig. 5.21 with H(s). The case where ne OLS (eaceay has a de gain for H(s) of lim H(s) = Ky. 0 The factor K’, is a conversion-of-units factor. If we set K, = K,, then the system is trans- formed to that of Fig. 5.23 (for the de gain or steady-state calculation). Then the error of the system shown in Fig. 5.23 is E(s), where E(s) = Rts) — ¥(s) = [1 — T(s)|RG), (5.35) FIGURE 5.21 Anonunity Rs) —>| GAs) Ys) feedback system. His) FIGURE 5.22 ROs) WA A speed control Desired Gt) sped system. speed ‘ (rad/s) (rads Ky FIGURE 5.23 a ; The speed control i) 3) system of Fig. 5.22‘) (rad/s) with K, = Ka.5.8 The Steady-State Error of Nonunity Feedback Systems 253 since Y(s) = T(s)R(s). Note that Pe Sea) OS T+ KGS and therefore 1 = Rs). 29 = TK Gok Then the steady-state error for a unit step input is 1 ey = lim sE(s) = (5.36) 0 1 + KG) In general we can always determine the actual system error by using Eq. (5.23). EXAMPLE 5.4 Steady-state error Let us determine the appropriate value of K, and calculate the steady-state error for a unit step input for the system shown in Fig. 5.21 when oo =” AS) = G+ 5) and HW) «20 = a = C510 * Gist We can rewrite H(s) as HG) = —2— oOi+)D Selecting K, = Ky = 2, we can use Eq, (5.36) to determine _ 1 DD fs “14 K,GQ 1+ 28) 17° or 5.9% of the magnitude of the step input. =254 Chapter5 The Performance of Feedback Control Systems EXAMPLE 5.5 Feedback system Let us consider the system of Fig. 5.24, where we assume we cannot insert a gain K, following R(s) as we did for the system of Fig. 5.21. Then the actual error is given by FIGURE 5.24 * r Asystem witha RU) wD Ys) sriback H(s). ~ 2 Ez Eq. (5.35), E(s) = [1 — T(s)]R(). Let us determine an appropriate gain K so that the steady-state error to a step input is minimized. The steady-state error is eee yt 390 s where Gs) _ K(s + 4) TS) = TT G6@H@ 6+ D644) 2K Then we have The steady-state error for a unit step input is e, = [1 — TO). Thus, to achieve zero steady-state error, we require that 4K OS g2K = or 8 + 2K = 4K. Thus K = 4 will yield a zero steady-state error. m254 Chapter5 The Performance of Feedback Control Systems. 5.9 PERFORMANCE INDICES Increasing emphasis on the mathematical formulation and measurement of control system performance can be found in the recent literature on automatic control. Modern control theory assumes that the systems engineer can specify quantitatively the required system performance. Then a performance index can be calculated or measured and used to evaluate the system’s performance. A quantitative measure of the performance of a system is nec- essary for the operation of modern adaptive control systems, for automatic parameter op- timization of a control system, and for the design of optimum systems. Whether the aim is to improve the design of a system or to design a control system, a performance index must be chosen and measured. A performance index is a quantitative measure of the performance of a system and is chosen so that emphasis is given to the important system specifications. A system is considered an optimum control system when the system parameters are adjusted so that the index reaches an extremum value, commonly a minimum value. A performance index, to be useful, must be a number that is always positive or zero. Then the best system is defined as the system that minimizes this index. A suitable performance index is the integral of the square of the error, ISE, which is defined as T ISE = [ er(t)dt. (3.37) ‘The upper limit 7 is a finite time chosen somewhat arbitrarily so that the integral ap- proaches a steady-state value. It is usually convenient to choose T as the settling time, Te The step response for a specific feedback control system is shown in Fig. 5.25(b), and the error in Fig. 5.25(c). The error squared is shown in Fig. 5.25(d), and the integral of the error squared in Fig. 5.25(e). This criterion will discriminate between excessively over- damped and excessively underdamped systems. The minimum value of the integral occurs for a compromise value of the damping. The performance index of Eq. (5.37) is easily adapted for practical measurements because a squaring circuit is readily obtained. Further- more, the squared error is mathematically convenient for analytical and computational purposes.255 5.9 Performance Indices 'GURE 5.25 re calculation of 2 integral juared error. (a) rt)256 Chapter5 The Performance of Feedback Control Systems Another readily instrumented performance criterion is the integral of the absolute mag- nitude of the error, IAE, which is written as s IAE = I lear. (5.38) This index is particularly useful for computer simulation studies. To reduce the contribution of the large al error to the value of the performance integral, as well as to emphasize errors occurring later in the response, the following index has been proposed [6]: co YT f . ITAE = lh tle(t)|dt. (5.39) This performance index is designated the integral of time multiplied by absolute error, ITAE. Another similar index is the integral of time multiplied by the squared error, ITSE: 7 ITSE = [ te?(t)dt. (5.40) The performance index ITAE provides the best selectivity of the performance indices: that is, the minimum value of the integral is readily discernible as the system parameters are varied. The general form of the performance integral is T r= f fle), r(), yO). dt, (5.41) where fis a function of the error, input, output, and time. One can obtain numerous indices based on various combinations of the system variables and time. It is worth noting that the minimization of LAE or ISE is often of practical significance. For example, the minimiza- tion of a performance index can be directly related to the minimization of fuel consumption for aircraft and space vehicles. Performance indices are useful for the analysis and design of control systems. Twe examples will illustrate the utility of this approach.256 Chapter5 The Performance of Feedback Control Systems EXAMPLE 5.6 Performance criteria A single-loop feedback control system is shown in Fig. 5.26, where the natural frequency is the normalized value, w, = 1. The closed-loop transfer function is then 1 s2+ 2s 41° Three performance indices—ISE, ITSE, and ITAE—calculated for various values of the damping ratio ¢ and for a step input are shown in Fig. 5.27. These curves show the selec- T(s) = (5.42) FIGURE 5.26 Single-loop feedback control “ system. 0.0 0.2 04 06 08 10 12 14 16 18 20 o FIGURE 5.27 Three performance criteria fora second-order system tivity of the ITAE index in comparison with the ISE index. The value of the damping ratio £ selected on the basis of ITAE is 0.7, which, for a second-order system, results in a swift response to a step with a 4.6% overshoot. m5.9 Performance Indices 257 EXAMPLE 5.7 Space telescope control system The signal-flow graph of a space telescope pointing control system is shown in Fig. 5.28 [11]. We desire to select the magnitude of the gain K; to minimize the effect of the dis- FIGURE 5.28 Disturbance Dis) A space telescope pointing control system. (a) Block diagram. (b) Flow Ris) graph. Ys) Auitude Position feedbac Ky (a) Disturbance Dis) RG) O Position feedback (b) FIGURE 5.29 The performance indices of the telescope control system versus Ks.258 Chapter5 The Performance of Feedback Control Systems turbance D(s). The disturbance in this case is equivalent to an initial attitude error. The closed-loop transfer function for the disturbance is obtained by using Mason's signal-flow: gain formula as follows: YO) _ PyA() D(s) Aw) i (5.43) s(s + K,K3) s° + K)K3s + K\K,K, Typical values for the constants are K, = 0.5 and K,K>K, = 2.5. Then the natural fre- quency of the vehicle is f, = V25/2a = 0.25 cycles/s. For a unit step disturbance, the minimum ISE can be analytically calculated. The attitude y(t) is ‘TO MO) = a sn(8 + *));: (5.44) where B = K;V(K3/8) — 5. Squaring y(t) and integrating the result, we have 210 osxy { camera position = position (b) FIGURE 5.31 The foreground camera, which may be either a film or video camera, is trained on the blue cyclorama stage. The electronic servocontrol installation permits the slaving, by means of electronic servodevices, of the two cameras. The background camera reaches into the miniature set with a periscope lens and instantaneously reproduces all movements of the foreground camera in the scale of the miniature. The video control installation allows the composite image to be monitored and recorded live. (Part (a) reprinted with permission from Electronic Design 24, 11, May 24, 1976. Copyright © Hayden Publishing Co., Inc., 1976.) 1.0 | 08 response 06 é 4 FIGURE 5.30 (c) Continued Normalized time262 Chapter 5 The Performance of Feedback Control Systems The locations of the closed-loop roots dictated by the ITAE system are shown in Fig. 5.32. The damping ratio of the complex roots is ¢ = 0-44. However, the complex roots 597 do not dominate. The actual response to a step input using a computer simulation showed the overshoot to be only 2% and the settling time to within 2% of the final value equal to 0.75 second. 1000 iS) =< = Ge 7086 + 521 + j10.68)(s + 5.21 — 710.68) o -8 4 ee fe ee FIGURE 5.32 The closed-loop Z| roots of a minimum i ‘TAE system. For a ramp input, the coefficients have been determined that minimize the ITAE crite- rion for the general closed-loop transfer function [6]: bis + by Ts) = : = ae bs"! + --- + bys + by (5.50) This transfer function has a steady-state error equal to zero for a ramp input. The opti- mum coefficients for this transfer function are given in Table 5.7. The transfer function, Eq. (5.50), implies that the plant G(s) has two or more pure integrations, as required to provide zero steady-state error. = TABLE 5.7 The Optimum Coefficients of T(s) Based on the ITAE Criterion for a Ramp Input Ss? + 3.20,5 53 + 1.75@,s? + 3.25wis + w3 St + 241w,s? + 4.9335? + 5.1dwis + wi S* + 2.19w,s* + 6.50m35? + 6.30wis? + 5.24wis + wi wr5.10 The Simplification of Linear Systems 263 5.11 DESIGN EXAMPLE: HUBBLE TELESCOPE POINTING CONTROL The orbiting Hubble space telescope is the most complex and expensive scientific instru- ment that has ever been built. Launched to 380 miles above the earth on April 24, 1990, the telescope has pushed technology to new limits. The telescope’s 2.4 meter (94.5-inch) mirror has the smoothest surface of any mirror made, and its pointing system can center it on a dime 400 miles away [9, 10, 21]. The mirror had a spherical aberration that was largely corrected during a space mission in December 1993 [24]. Consider the model of the telescope-pointing system shown in Fig. 5.34. The goal of the design is to choose K, and K so that (1) the percent overshoot of the output to a step command, r(r), is less than or equal to 10%, (2) the steady-state error to a ramp command is minimized, and (3) the effect of a step disturbance is reduced. Since5.11 Design Example: Hubble Telescope Pointing Control 267 Disturbance D(s) Telescope Amplifier 1 dynamics Ris) t Ys) Command Pointing (a) Dis) Gs) Ris) Ys) (b) Dis) Ys) tc) FIGURE 5.34 Response tor) (a) The Hubble a telescope Pointing system, (b) reduced block diagram, (c) system design, and (d) system Fesponse toa “nit step input mand and & wait step disturbance input. w@) 1 (seconds) (da)268 Chapter 5 The Performance of Feedback Control Systems the system has an inner loop, block diagram reduction can be used to obtain the simplified system of Fig, 5.34(b). Mason's signal-flow gain formula can be used to obtain the output due to the two inputs of the system of Fig. 5.34(b): ¥(s) = T(s)R(s) + [T(/KID(), (5.67) where 15) =< KOO). KG) 1+ KG) 1+ Lis) The error E(s) is 4 5 Gls) . EO) = TS TBRO - Tew: (5.68) First let us select K and K, to meet the percent overshoot requirement for a step input, R(s) = AA. Setting D(s) = 0, we have KG(s) Ys) = —_p, = Tyee (5.69) ——K___fA\__ KA MEEK) + Ks} 52 $ Kis + K\ sy To set the overshoot less than 10%, we select £ = 0.6 by examining Fig. 5.8 or using Eg. (5.16) to determine that the overshoot will be 9.5% for ¢ = 0.6. We next examine the steady-state error for a ramp, r(r) = Br, r= 0, using (Eq. = tim {8} 8 - oe ey LSKGG)| (K/R,)” 6.70) The steady-state error due to a unit step disturbance is equal to — 1/K. (Can you show this?) The transient response of the error due to the step disturbance input can be reduced by increasing K (see Bq. 5.68). In summary, we seek a large K and a large value of (K/K) to obtain a low steady-state error for the ramp input (see Eq. 5.70). However, we also re- quire £ = 0.6 to limit the overshoot. For our design, we need to select K. The characteristic equation of the system is (with £= 0.6) (s? + 2fa,5 + w2) = (s? + 20.6)w,5 + K). (6.71) Therefore w, = VK, and the second term of the denominator of Eq. (5.69) requires K, = 2(0.6)w,. Then K, = 1.2VK, or the ratio K/K, becomes K K VK K, 12VK. 12° Selecting K = 25, we have K, = 6 and K/K, = 4.17. If we select K = 100, we have K, = 12 and K/K, = 8.33. Realistically we must limit K so that the system’s operation remains linear. Using K = 100, we obtain the system shown in 34(c). The responses of the system to a unit step input command and a unit step disturbance input are shown in Fig. 5.34(d). Note how the effect of the disturbance is relatively insignificant.5.12 System Performance Using MatLas 100 90 80 Percent —— 7 5.00 4.80 “| 4.60 70 overshoot 60 440 4.20 4.00 3.80 3.60 3.40 3.20 Damping ratio, £ P.O. = 100¢ ~s#/V1~2?_ 0 3.01 0.0 0.10.2 0.3 04 0.5 0.6 0.7 0.8 0.9 1.0 10 (5.16) On T, 269 FIGURE 5.8 Percent overshoot and normalized peak time versus damping ratio ¢ for a second-order system (Eq. 5.8). Finally we note that the steady-state error for a ramp input (see Eq. 5.70) is B Oss ~ 333 7 0.12B. This design, using K = 100, is an excellent system.