o2—- BPW Loplso nal Feedback Control Systems Charles L. Phillips ” Prentice-Hall International, Inc.Contents PREFACE ii 1, INTRODUC? TION oO 1.1 The Control Problem, 4 ~ 71.2 The Laplace Transform, 5 1.3 ‘Theorems of the Laplace Transform, 12 1.4 Differential Equations and Transfer Functions, 15 1.5 Linearization, 17 1.6 Summary, 19 References, 20 Problems, 20 7 2. MODELS OF PHYSICAL SYSTEMS @) 2.1 System Modeling, 23 2.2 Electrical Circuits, 25 2-23 Block Diagrams and Signal Flow Graphs, 32 24 Mason's Gain Formula, 35 25 Mechanical Translational Systems, 40 2.6 Mechanical Rotational Systems, 45 27 Electromechanical Systems, 47 28 Some Special Manipulations, 52Contens, 29 A Temperature Control System, 54 2.10 Analogous Systems. $7 2.11 Transformers and Gears, 59 2.12 A Robotic Control System, 61 243 System Identification, 64 214 Summary, 64 References, 65 Problems, 65 ‘STATE VARIABLE MODELS a 3.1 State Variable Modeling, 73 - 3.2. Simulation Diagrams, 77 3.3 Solution of State Equations, 81 34 Transfer Functions, 89 3.5 Similarity Transformations, 91 3.6 Digital Simulation, 98 3.7 Analog Simulation, 102 3.8 Summary, 104 References, 105 Problems, 106 SYSTEM RESPONSES ia) 4.1 Time Response of First-order Systems, 114 4.2. Time Response of Second-order Systems, 118, 4.3. Time Response Specifications, 121 4.4 Frequency Response of Systems, 126 igher-order Systems. 133, 4.5 Response of 4.6 Reduced-order Models, 136 4.7 Summary. 138 References, 139 Problems. 139Contents, _& CONTROL SYSTEM CHARACTERISTICS fa) 5.1 A Closed-loop Control System, 145 5.2 Stability, 149 5.3 Sensitivity, 153 5.4 Disturbance Rejection, 157 5.5 Steady-state Accuracy, 162 5.6 Transient Response, 170 5.7 Closed-loop Frequency Response, 171 3.8 Summary, 172 References, 173 Problems, 173 - ‘STABILITY ANALYSIS (78) 6.1 The Routh-Hurwitz Stobility Criterion, 181 - 6.2 Roots of the Characteristic Equation, 190 6.3 Stability by Simulation, 191 64 Summary, 192 References, 192 Problems, 192 (7 meer-Locus ANALYSIS AND DESIGN 199 7.1 Root-locus Principles, 199 7.2 Some Root-locus Techniques, 204 73 Additional Rootlocus Techniques, 208 74 Additional Properties of the Root Locus, 217 7.5 Other Configurations. 221 7.6 — Rootlocus Design, 223 7.7 Phase-lead Design, | 227 7 Analytical Phase-lead Design, 230Contents vill 79 Phase-lag Designs 254 7.10 PID Design. 239 711 Analytical PID Design, 244 712 Complementary Root Locus, 247 7.13. Compensaior Realization, 250 7.14 Summary, 253 References, 253 Problems, 254 & FREQUENCY-RESPONSE ANALYSIS 8.1 Frequency Responses, 262 8.2 Bode Diagrams, 267 8.3 Additional Terms, 280 84 The Nyquist Criterion, 288 8.5 Application of the Nyquist Criterion, 295 8.6 Relative Stability and the Bode Diagram, 306 8.7 Closed-loop Frequency Response, 315 8.8 Summary, 322 References, 323 Problems, 323 @) (9. FREQUENCY-RESPONSE DESIGN 9.1 Control System Specifications, 333 9.2 Compensation, 337 9.3 Gain Compensation, 339 9.4 Phase-lag Compensation, 343, 9.5 Phase-lead Compensation, 348 9.6 Analytical Design, 352 9.7 Lag-lead Compensation, 360Contents x 9.8 PID Controller Design, 363 9.9 Analytical PID Controller Design, 368 9.10 PID Controller Implementation, 372 9.11 Summary. 375 References, 376 Problems, 376 10. DIGITAL CONTROL SYSTEMS (382 10.1 A Diserete-time System, 382 10.2. Transform Methods, 384 10.3. Theorems of the z-transform, 387 10.4 Solution of Difference Equations, 390 10.5 The Inverse z-transform, 394 10.6 Simulation Diagrams and Flow Graphs, 396 10.7 Stote Variables, 399 10.8 Solution of State Equations, 402 10.9 Summary, 406 References, 407 Problems, 407 11, SAMPLED-DATA SYSTEMS 413) 11 Sampled Data, 413 11.2 The Ideal Sampler, 415 113 Properties of the Starred Transform, 418 114 Data Reconstruction, 425 115 The Pulse Transfer Function, 426 116 — Open-loop Systems Containing Digital Filters. 431 11.7. Closed-loop Discrete-time Systems, 433 118 Transfer Functions for Closed-loop Systems. 43512, ANALYSIS AND DESIGN OF DIGITAL CONTROL 13. 19 11.10 Content, State Variables for Sampled-data Systems, 44 Summary, 445 References, 446 Problems, 446 SYSTEMS #1) 121 Two Examples, 451 ° 12.2 Diserete System Stability, 455 12.3 Jury's Test, 456 12.4 Mapping the s-plane into the z-plane, 458 12.5 Root Locus, 463 12.6 The Nyquist Criterion, 465 12.7 The Bilinear Transformation, 471 12.8 The Routh-Hurwitz Criterion, 475 129 The Bode Diagram, 476 12.10 Steady-state Accuracy, 478 12.11 Design of Digital Control Systems, 481 12.12 Phase-lag Design, 485 1213 Phase-lead Design, 490 12.14 Digital PID Controllers, 493 12.15 Rootlocus design, 498 12.16 Summary, S01 References, S01 Problems, 502 MODERN CONTROL DESIGN 508 13.1 Pole-Placement Design, 510 13.2 Ackermann’s Formula, 513 13.3. State Estimation, 518 13.4 Closed-loop System Characteristics, 527Chapter 1 Introduction This book is concerned with the analysis and design of closed-loop control systems, In the analysis of closed-loop systems, we are given the system, and we wish to determine its characteristics or behavior. In the design of closed-loop systems, we specify the desired system characteristics or behavior, and we must configure or synthesize the closed-loop system so that it exhibits these desired qualities. We define a closed-loop system as one in which certain of the system forcing signals (we call these inputs) ate determined, at least in part, by certain of the responses of the system (we call these outputs). Hence, the system inputs are @ function of the system outputs, and the system outputs are a function of the system inputs, A diagram representing the functional relationships in aclosed-loop system is given in Figure 1.1. ‘An example of a closed-loop system is the temperature-control system in the home. For this system we wish to maintain, automatically, the temperature of the living space in the home at a desired value, To control any physical variable, which we usually call a signal, we must know the value of this variable, that is, we must measure this variable. We call the system for the measurement of a variable 2 sensor, as indicated in Figure 1.2. In a home temperature-control system, the sensor is a thermostat, which indicates a low temperature by closing an electrical switch and an acceptable temperature by opening the same switch, We define the plant of a control system as that part of the system to be controlled. Itis assumed syxem, oyna inputs outputs igure 1.1 A closed-loop system,_———_of variable Sensor Variable in this example that the temperatures increased by activatinga gas furnace. Hen the plant input is the electrical signal that activates the furnace, and the pax butput signal is the actual temperature of the living area. The plant is representa $s shown in Figure 1.2. In the home-heating system, the output of each of the Systems is connected fo the input of the other, to form the closed loop. However, jin most closed-loop control systems, it is necessary to connect a third system inta the loop to obtain satisfactory characteristics for the total system. This additional system is called a compensator, a controller, or simply a filter, The usual form of a single loop closed-loop control system is given in Figure 1.3. The system input is a reference signal; usually we want the system output to be equal to this input. In the home temperature-control system, this input is the setting of the thermostat. If we want to change the temperature, we change the system input. The system output is measured by the sensor, and this measured value is compared with (subtracted from) the input. This difference signal is called the error signal, or simply the error. If the output is equal to the input, this difference is zero, and no signal reaches the plant. Hence the plant output remains at its current value. If the error is not zero, in a properly designed system the error signal causes a response in the plant such that the magnitude of the error is reduced. The compensator is a filter for the error signal, since usually satisfactory operation does not occur if the error signal is applied directly to the plant. Control systems are sometimes divided into two classes. If the object of the control system is to maintain a physical variable at some constant value in the presence of disturbances, we call this system a regulator. One example of a rg: Ulator control system is the speed-control system on the AC power generators of power utility companies. ‘The purpose of this control system is to maintain the speed of the generators at the constant value that results in the generated voltage having a frequency of 60 Hz in the presence of varying electrical power loads. Another example of a regulator control system is the biological system, which Maniputated Syeom ‘arable output Plone [compensator = Figure 1.3 Closed-loop conteol systeChap. 1 Introduction a maintains the temperature o aa °F in a erm Lhe Stand eitss of eonito systems is the servomecharia., Although this term was originally applied to a system that controlled @ mechanical position or motion, it is now often used to describe a control system in which a physical variable is required to follow, or track, some desired time function. An example of this type of system is an automatic aircraft landing system, in which the aircraft follows a amp 10 the desired touchdown point, A second example is the control systems of a robot, in which the robot hand is made to follow some desired path in space, The preceding is a very simplified discussion of a closed-loop control system. ‘The remainder of this book improves upon this description. In order to perform either mathematical analysis or design, itis necessary that we have a mathematical relationship between the input and the output for each of the blocks in the control system of Figure 1.3. The purpose of Chapter 2 is to develop these functional relationships for some common physical systems. Chapter 3 presents a different method of expressing these functional relationships. ‘We examine typical responses that occur in control systems in Chapter 4 and look at control system specifications in Chapter 5. Chapter 6 presents concepts and analysis techniques for system stability. The root locus, one of the principal methods of analysis and design, is developed in Chapter 7. Chapters 8 and 9 present a second principal analysis and design method, the frequency response. In Chapters 2 through 9 itis assumed that no signals appear in sampled form ‘and in particular that no digital computers are used in the control of the system ‘The systems considered in these chapiers are called analog systems, or continuous: data systems. Chapters 10 through 12 consider systems in which sampling does Occur, and these systems are called sampled-data systems. If a digitsl computer is used in the control of these systems, the systems ate then called digital control systems. The term discrete sysicms is elo used to refer to sampled-data systems and digital conirol systems. Chapter 13 presents an introduction to a different method of design of analog control systems, which is classified as a modem control procedure. In the systems of Chapters 2 through 13 all systems are assumed to be linear (linearity is defined in Chapter 2). However, physical systems are not linear. and {n general, nonlinear systems ace dificelt to analyze or design. Throughout this Book we siscuss the problems of the inaccurate representations in the functional relationships that we use to model physical systems. | However, for some systems, Toalimearities must be added to the system model to improve the accuracy of these Tanctional relationships. We consider some common nonlinearitiesand some anal ysis methods for nonlinear systems in Chapter 14 in the analysis of linear systems, we use the Laplace transform for analog sjstems and the z-transform for discrete systems. The latter part ofthis chapter presents the concepts and procedures of the Laplace transform, andthe z-1ransform Prepvered in Chapter 10. However, before the Laplace transform is discussed, the control problem is presented+ 4 Introduction Chap, 4 1.1. THE CONTROL PROBLEM We may state the control problem as follows. A physical system or process isto be accurately controlled through closed-loop. or feedback, operation. An output variable, called the response, is adjusted as required by the error signal. This error signal is the difference between the system response, as measured by a sensor, and the reference signal, which represents the desired system response Generally a controller, or compensator, is required to filter the error signal in order that certain control criteria, or specifications, be satisfied. These criteria may involve, but not be limited to 1, Disturbance rejection 2, Steady-state errors 3. Transient response characteristics, 4. Sensitivity to parameter changes in the plant Solving the control problem generally involves 1, Choosing sensors to measure the plant output, Choosing actuators to drive the plant 3. Developing the plant, actuator, and sensor equations (models) Designing the controller based on the developed models and the control criteria 5. Evaluating the design analytically, by simulation, and, finally, by testing the physical system 6. If the physical tests are unsatisfactory, iterating these steps x S Because of inaccuracies in the models, the tests on the physical control system are usually not satisfactory. The controls engineer must then iterate this design pro- cedure, using all tools available, to improve the system. Generally intuition, developed while experimenting with the physical system, plays an important part in the design process. ‘The relationship of mathematical analysis and design to physical-system design procedures is depicted in Figure 1.41]. In this text, all phases shown in the figure are discussed, but the emphasis is necessarily on the conceptual part of the pro- cedures—the application of mathematical concepts to mathematical models. In practical design situations, however, the major difficulties are in formulating the problem mathematically and in translating the mathematical solution back t0 the physical world. As stated earlier, many iterations of the procedures shown in Figure 1.4 are usually required in practical situations, Depending on the system and the experience of the designer. some of the steps listed earlier may be omitted. In particular, many control systems are plemented by choosing standard forms of controllers and experimentally deter ing the parameters of the controller by following a specified step-by-step PISoc. 1.2 The Laplace Transform s ‘ob foxmulation Soluen wansaton Figare 1-4 Mathematica solution of physica problems cedure with the physical system; no mathematical models are developed. This type of procedure works very well for certain control systems. For other systems, however, it does not, For example, a control system for a space vehicle obviously ‘cannot be designed in this manner. | In this book mathematical procedures are developed for she analysis and design of control systems. ‘The actual techniques may or may no! be of value the design of a particular control system. However, standard controllers are 1 Tzed in the developments in this book. Thus the analytical procedures develop the concepts of control system design and indicate the applizaion of each of the standard controllers. 4.2 THE LAPLACE TRANSFORM ‘A brief introduction to the Laplace transform fs presented in the remainder of is Chapter. We will see that the Laplace transform & useful tn the modeling of “Tingar time-invariant analog system as.a transfer Huneloh “The Laplace transform Thay alg ‘be used to solve for the response of this tyPe Of SsKem: however, We encrally use simulations (machine solutions of the Som ‘equations) for this pur- eral those readers wanting to delve more deeply into the Laplace ‘transform, {2}, 3}, and [4] are suggested for supplemental reading. ‘By definition, the Laplace transform of a function of time fois 21 ms) = 21/01 = [, foe ae wy were i indicates the Laplace transform. Note thatthe varahi te, has been Wisprated out of the equation and thatthe Lapiace transform is a function of the complex variable s.. The inverse Laplace transform is given by LPO riser ds fe = 2-1 = 35 |. (a) where £P-? indicates the inverse transform and j = V—l-6 Introduction Chap, y Equations (1-1) and (1-2) form the Laplace transform pair. Given a funetion fl. we integrate (1-1) to find its Laplace transform £6). ‘Then if this function FIs) is used to evaluate (1-2), the result will be the original value of f(t), The value of o in (1-2) is determined by the singularities of F(s) [2]. We seldom use (1-2) to evaluate an inverse Laplace transform; instead we use (1-1) to construc, ‘a table of transforms for common time functions, ‘Then, when possible, we use this table to find the inverse transform rather than integrating (1-2). ‘Asan example, we will find the Laplace transform of the exponential function From (1-1), Fs) = fe 1 “sta where Re(-) indicates the real part of the expression. Ofcourse, Laplace transform tables were derived long ago, and we will not derive any additional transforms, men ten “d= [ecw ar ste lo (13) Re(s + a) >0 TABLE 1.1. LAPLACE TRANSFORMS Nome Time Function (( Laplace Transform FG) Unit impulse 5) Unit step ua Unit ramp t inth-order ramp ” Exponential en inth-order exponential tre Sine sin be Cosine cos br Damped sine sin bt Damped cosine e-* cos br Diverging sine tin br Diverging cosine tee beSec. 1.2 The Laplace Transform 7 A short table of commonly required transforms is given in Table |.1. Appendix Chun rather extensive table of both Laplace transforms and 2-transforms. The rin share ihe table rea more complet table of Laplace transforms. 1¢ remainin; in this table will be useful when iit Te i {1 when we consider digital control From the definition of the Laplace transform, (1-1), LIKI] = KELSO] = KFS) (4) for k constant, and ZTAO + £0) LAO) + 2UA0) = Fils) + FG). (1s) The use of these two relationships greatly extends the application of Table 1.1 ‘We now present some examples of the Laplace transform and of the inverse Laplace transform. Fist, however, some definitions are required. Note from the definition of the Laplace transform, (I-1), that the value of the time function for time less than zero does not affect the transform. Thus we ean arbitrarily assign values to the function for negative time; to simplify certain theorems that wwe present later, we assign the value of zero to al ime functions for negative time We define the unit step function to simplify notation. By definition, the unit step function u(x) is wo“ (28 9 In Equation (1-3), the Laplace transform of ¢°** was derived. Note that the Laplace transform of e~*u() isthe same function. Thus for any function f(0), [MO] - LAO) = FO) (7) We have assumed that f() = 0 for ¢ <0; when the Laplace transform of f(é) is aerived we are actualy finding the Laplace transform of /()4(). Example 1.1 ‘The Laplace transform of the time function ft) = Sut) + 3e-™ ‘will now be found. From Table 1.1 and (1-4), 5 egfpulo) = SW] = 5 apes) sete 3g Then, from (15), 5,3 As) = £{5u0) + 3e1 = 2+ 5