IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO.

4, NOVEMBER 2002

307

Learning Feedback Controller Design of Switching Converters Via MATLAB/SIMULINK

Juing-Huei Su, Member, IEEE, Jiann-Jong Chen, Member, IEEE, and Dong-Shiuh Wu, Member, IEEE
AbstractThe application of MATLAB/SIMULINK is developed for simulation and learning feedback controller design of dcdc switching converters. With the help of its intuitive graphical user interface and some basic circuit theories, the corresponding SIMULINK models of the switching converter circuits can be easily constructed. Students can then use these models to learn and evaluate the closed-loop behavior of the entire system in the MATLAB/SIMULINK environment, after feedback controllers are devised by any classical or modern control theory. The accuracy of this approach is also verified by comparing the simulation results with the responses obtained from a buck-type dcdc switching converter circuit and existing experimental results [4]. Index Termsdcdc switching converter, feedback controller, MATLAB/SIMULINK.

Fig. 1. Boost-type dcdc switching converter.

I. INTRODUCTION HE SWITCHED-mode dcdc converters are power electronic systems that convert one level of electrical voltage into another level by switching action. They are very popular these days because of their high efficiency and smaller size [1], [7], [15]. Switched-mode dcdc converters are used extensively in personal computers, computer peripherals, and adapters of consumer electronic devices to provide dc voltages. The widespread use of switched mode dcdc converters in many electronic systems makes the fundamental understanding of them a necessity for many electronic system design engineers. Unfortunately, learning the design philosophy of the dcdc switching converters is interesting but difficult, because it includes many areas of knowledge, e.g., the converter circuits and electronics, linear and nonlinear control system theory [7], [15], magnetics, etc. Teaching the dcdc switching converters is a challenging undertaking because one cannot assume that all students enrolled in the class have solid prerequisite knowledge in so many areas. Therefore, to speed up the learning process, the application of user-friendly and powerful computer-aided simulation software tools to help students get acquainted with the dynamic behavior of the dcdc switching converter circuits is inevitable [1][3], [5], [10], [13]. Although the simulation program SPICE is quite popular among electronic engineers, the software package focuses primarily on the circuit level in constructing the entire circuit for simulation. Furthermore, the complexity of device models and the switching nature of the switching converter circuits make simulation difficult to conManuscript received August 25, 2001; revised December 27, 2001. The authors are with the Department of Electronic Engineering, Lunghwa University of Science and Technology, 330, Taiwan, R.O.C. (e-mail: suhu@mail.lhit.edu.tw). Digital Object Identifier 10.1109/TE.2002.803403

verge unless some of the default values of parameters are suitably changed [10]. Thus, SPICE is not suitable for students who need to learn quickly what the system dynamics are and to acquire the system viewpoints in linear and even nonlinear feedback controller designs. In this regard, the simulation environment MATLAB/SIMULINK is quite suitable for students to learn the feedback controller design techniques if the corresponding simulation models of dcdc switching converters can be constructed without too much effort and still give accurate results. It is, therefore, the purpose of this paper to propose an alternative way of teaching feedback controller designs of dcdc switching converters to students, so that it would be less difficult to learn. SIMULINK is a window-oriented dynamics modeling software package built on top of the MATLAB numerical workspace. An advantage is that models are entered as block diagrams with an intuitive graphical interface when the corresponding mathematical descriptions are available for the target systems. This application is not difficult to do for basic topologies of dcdc switching converters. Furthermore, a set of blocks with signal interconnections could be masked as a subsystem for convenience in the SIMULINK environment. The parameters of masked subsystems are then entered in dialog windows and can be changed interactively during a simulation. Simulation results can be viewed during the simulation via a virtual oscilloscope and then exported to the MATLAB workspace for subsequent off-line analysis. As stated in [6] and [13], the abundant library blocks that the SIMULINK modeling environment provides make construction of simple dynamical systems quite easy. This construction is also true for the design and verification of feedback controllers for dynamical systems. If the mathematical way of using Kirchhoffs laws to construct the corresponding dynamical systems is not favored, the MATLAB environment can also be used to develop mathematical models from inputoutput data via the system identification tools [2].

0018-9359/02$17.00 2002 IEEE

308

IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO. 4, NOVEMBER 2002

Fig. 3. Comparisons of simulation results obtained from SPICE and SIMULINK.

Fig. 2. The constructed boost-type converter block in SIMULINK.

II. SIMULINK MODELS CONSTRUCTION SWITCHING CONVERTERS

OF

dcdc

The boost-type dcdc switching converter with an ideal single-pole double-throw switch is an example of a SIMULINK model (Fig. 1). The dynamics of this converter operating in the continuous conduction mode (CCM) can be easily understood by applying Kirchhoffs voltage law on the loop containing the inductor and Kirchhoffs current law on the node with the capacitor branch connected to it [7], [15]. When the ideal switch is at position 1, the dynamics of the inductor current and the capacitor voltage are (1) When the ideal switch is at position 2, the dynamics of the circuit are

(2) and output voltage In both cases, capacitor voltage are related via the following equation: (3) The intuitive signal flow interface in SIMULINK makes this mathematical model and its corresponding masked subsystem very easy to create (see Fig. 2). The constraints that the inductor

current and the capacitor voltage can only be positive in the circuit are fulfilled by setting the integration blocks in the SIMULINK model to give only positive outputs. Therefore, the SIMULINK model for the boost-type dcdc switching converter is also valid when operating in the discontinuous conduction mode (DCM). To facilitate the subsequent simulation analysis and feedback controller verification, the pulse-width-modulation signal to control the ideal switch can also be built into the masked subsystem. Therefore, the inputs for the masked subsystem are duty ratio and input voltage, and the outputs are chosen to be inductor current, capacitor voltage, and output voltage. When double-clicking the pointer on the masked subsystem, students can enter parameter values of the switching converter circuit in a dialog window. The circuit parameters of the above boost-type switching con5 kHz, verter circuit are as follows: switching frequency 15 V, 0.25, 500 H, 200 F, 0.5 , 0.25 , and 10 . The simulation results can be seen to agree closely with those provided by SPICE in Fig. 3. These results validate the usefulness of the corresponding SIMULINK model of the boost-type switching converter. The nonidealities of the single-pole double-throw switch when realized via a transistor and diode could also be built into the corresponding SIMULINK subsystem. This process is accomplished by first modifying (1)(3) with simplified models of transistors and diodes and then constructing the subsystem in the SIMULINK environment accordingly. For example, MOSFETs would behave like a resistor and diode and can be modeled as a forward voltage drop in series with a resistor when they are in the ON-state. It is also possible to construct a SIMULINK model for zerocurrent-switching quasi-resonant converters (ZCS-QRCs), but it is not easy to construct simulation circuits in SPICE [10]. The buck-type half-wave ZCS-QRC model in SIMULINK is constructed without too much effort when the operating modes are 20 V, 1.6 H, identified. Fig. 4 (with parameters

SU et al.: LEARNING FEEDBACK CONTROLLER DESIGN OF SWITCHING CONVERTERS

309

(a)

(b)

(c)

(d)

Fig. 4. (a) The buck-type half-wave ZCS-QRC converter circuit. (b) The constructed buck half-wave ZCS-QRC converter block in SIMULINK. (c) Inductor currents in buck-type half-wave ZCS-QRC circuit. (d) Inductor currents and capacitor voltages in buck-type half-wave ZCS-QRC circuit.

310

IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO. 4, NOVEMBER 2002

0.064 F, 0.1 H, 0, 0, 0, 0, 0.1 F, 5 , and the switching frequency at 200 kHz) shows the buck-type half-wave ZCS-QRC circuit and its corresponding SIMULINK model. The simulation results also agree closely with those provided in [10]. Since the SIMULINK models for dcdc switching converters provide reasonably accurate simulation results, they could also be used to verify the modeling techniques for dcdc switching converters in both CCM and DCM operating conditions [7], [9], [11], [15]. Since the MATLAB toolboxes are abundant and the typical learning time for the manipulation of SIMULINKs intuitive graphical interface is short, the simulation tool is well suited to educational use. Also, the purpose of this paper is to provide an alternative way of learning the dynamic behavior and feedback controller designs of dcdc switching converters [14]. III. FEEDBACK CONTROLLER DESIGNS SWITCHING CONVERTERS
OF

Fig. 5.

Buck-type switching converter.

Now is the time to learn the design philosophies of linear averaged and nonlinear feedback controllers for the dcdc switching converters, after the corresponding SIMULINK models of the basic dcdc switching converters are created. A. Linear Averaged Feedback Controller Design For the sake of simplicity, the following discussion will be limited to the voltage-mode linear averaged feedback controller designs. Voltage-mode linear averaged feedback controllers for dcdc switching converters are mostly designed in the frequency domain, though some feedback controllers with more advanced control theory have been developed in the time domain recently [4], [12]. The design guidelines for the voltage-mode linear averaged feedback controllers of a given dcdc switching converter on a given operating condition could be summarized as the following steps. 1) Determine the steady-state operating condition to be controlled. The steady-state information of the switching converter circuit is then used to construct the corresponding linearized small signal mathematical model and its frequency response. The methods provided in [7][9], [11], and [15] may be used in this step. 2) Decide what the value of the 0-dB crossover frequency of the closed-loop gain should be. This value generally determines the bandwidth of the control to the output closed-loop system. Since the averaged mathematical models are accurate only up to one-third of the switching frequency [11], the value of the 0-dB crossover frequency should be lower than one-third of the switching frequency. 3) Choose an appropriate phase margin at the crossover frequency for the compensator to provide to meet both the stability and performance requirements. This task may take some iterations for students to check on different operating conditions. 4) Enlarge the gain below the crossover frequency of the loop gain so that the influences of disturbances (e.g., input

voltage variations and load variations) would be kept as small as possible. Output voltage steady-state error could be eliminated if an integrator is included in the feedback controller. 5) Verify whether the linear averaged feedback controller could deal with all the operating conditions in the specifications. Following the guidelines above, the voltage-mode linear averaged feedback controller can then be constructed by using any existing controller typefor example, phase lead, phase lag, or combined compensatorto meet the stability and performance requirements. Fortunately, these steps of feedback controller design could be accomplished quickly and easily in the MATLAB/SIMULINK environment. The sisotool (bode) command in the control system toolbox provides a graphical user interface so that the closed-loop frequency response can be interactively changed by online modifying of the pole-zero pattern of the feedback controller. When the desired frequency is obtained, users are also given the corresponding transfer function of the feedback controller in the same interface. Since the design of the linear averaged feedback controller is based only on a given operating condition, simulations or experimental prototypes should be conducted to see whether it could cope with the other operating conditions given in the specifications. Experimental prototype construction is usually a timeconsuming task; therefore, simulations are often taken to verify that the specifications are fulfilled to save time in the concept design stage. Since the SIMULINK models were introduced in Section II, students can easily see how the devised feedback controller works in the MATLAB/SIMULINK environment. The entire design procedure is illustrated in the following example, and the result is compared with an experimental circuit by using the voltage-mode PWM control IC TL494. Consider the buck-type switching converter in Fig. 5, with 12 V, 5 V, 0.25 , input voltage 0.25 , 150 H, 200 F, and 20 . If the switching frequency is set to 55 kHz, the circuit would operate in the CCM. and duty The steady-state operating point of the above switching converter operating in CCM cycle can be calculated via the following formula: (4) Equation (4) is derived by assuming that the derivatives of the and capacitor voltage are averaged inductor current zero without resorting to the method in [8]. The matrixes in

SU et al.: LEARNING FEEDBACK CONTROLLER DESIGN OF SWITCHING CONVERTERS

311

Fig. 6. Open-loop frequency response of buck converter in MATLAB/SIMULINK.

Fig. 7. Closed-loop frequency response of buck converter in MATLAB/SIMULINK.

(4) represent different operating modes (subscript 1 stands for transistor on, and subscript 2 stands for transistor off) of the switching converter circuit. Therefore

(5c) Using the element values of the above buck-type switching converter, the steady-state operating conditions can be simply 0.25 A, 5 V, calculated via (5) in MATLAB as 0.4219. The frequency response of the linearized and control to the output transfer function of the example switching circuit can then be plotted via the sisotool (bode) command and the switching converters linearized average model [7], as in Fig. 6. According to the above design guidelines, the 0-dB crossover frequency of the entire closed-loop frequency response is chosen to be about one-eighth of the switching

(5a) (5b)

312

IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO. 4, NOVEMBER 2002

(a)

(b) Fig. 8. (a) Verification setup in the SIMULINK environment for buck converter. (b) Parameters setting dialog box for the masked buck converter subsystem.

Fig. 9. Closed-loop control of buck-type converter via TL494.

frequency, i.e., 8 kHz. The linear averaged feedback controller is then designed to stabilize the linearized switching converter in the same interface provided by the sisotool command by

adjusting the pole-zero locations of the feedback controller. The corresponding transfer function of the feedback controller and gain/phase margins of the closed-loop system would also

SU et al.: LEARNING FEEDBACK CONTROLLER DESIGN OF SWITCHING CONVERTERS

313

Fig. 10.

Practical realization of feedback controller defined in (6).

be shown in the interface. Therefore, after some adjustments in the design environment, students can easily construct the following type 1 feedback controller (6)
(a)

which gives a phase margin of about 34 at a 0-dB crossover frequency of about 8.5 kHz. The closed-loop frequency response is shown in Fig. 7. For the above linear averaged feedback controller, students ought to check whether it still stabilizes the switching converter at different operating conditions, e.g., light load, heavy load, and different input voltages. To verify whether the linear averaged feedback controller meets the performance requirements, students can construct a SIMULINK model like the one in Fig. 8(a) with load change built into the SIMULINK switching converter model. Feedback controller design at this stage may require some iterations to tune the control parameters so that the closed-loop system gives satisfactory results during startup and abrupt load or input changes. The approach proposed in this paper could help save quite some time at this stage. If the simulation results meet the stability and performance requirements, the feedback controller is feasible. To assure students that this approach really works, the accuracy of the simulation results is then compared with the responses obtained from an experimental switching converter circuit constructed by using TL494 [16], which is shown in Fig. 9. The linear averaged feedback controller defined in (6) is realized in Fig. 10 via the error operation amplifier in TL494. The inputoutput relation of the circuit is (7) It can be seen from (7) that the circuit implements the desired linear averaged feedback controller in small signal. The simulation results and the experimental circuit output are compared in Fig. 11, showing close resemblance in the transient behavior when load abruptly changes from 2 K to 20 . B. Nonlinear Feedback Controller Design The framework described above is also quite suitable for learning feedback controller designs by applying more advanced control theory, as is illustrated in the following example. Example: In [4], several nonlinear feedback controllers for a boost-type dcdc switching converter are compared. One of the nonlinear feedback controllers, the sliding-mode controller illustrates how the above-mentioned approach could be applied to help students quickly get acquainted with the method and find out what the controllers strong and weak points are. A brief

(b) Fig. 11. (a) Output response simulation of the feedback controlled buck converter with load change at 8 ms. (b) Output response of the experimental feedback controlled buck converter with load change (using TL494).

introduction of the theoretical background and implementation issues of designing sliding-mode controllers would be given to students first, according to [17]. By using the previously developed boost-type dcdc switching converter subsystem, the authors constructed the digital sliding-mode controller [4] with a 60- s sampling period shown in Fig. 12(a). The sliding surface is chosen to be (8) and the corresponding control signal for the ideal switch in Fig. 1 is (9) To reduce the chattering phenomenon caused by the input signal (9) around the switching surface (8), hysteresis control method [17] is used. The entire feedback control system with input voltage variations in the SIMULINK environment is

314

IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO. 4, NOVEMBER 2002

This example helps students not only learn the sliding mode control theory but also get acquainted with some practical issues in implementation, e.g., the effect when the feedback controller is implemented digitally and the relay in Fig. 12(a) as a hysteresis controller to reduce the chattering phenomenon.

(a)

IV. CONCLUSION The need for undergraduate students to learn to apply computer-aided design tools to speed up the learning process in the dcdc switching converter design is increasing because the topic integrates many areas of knowledge. The computer-aided design software tool MATLAB/SIMULINK is shown in this paper to be quite useful in providing a simulation and verification environment for the feedback controller designs of the dcdc switching converters. Students can not only learn quickly to acquire the system viewpoint about the entire dcdc switching converter design process but also explore more advanced control techniques [4], [12], [17].

REFERENCES
[1] S. S. Ang, A practice-oriented course in switching converters, IEEE Trans. Educ., vol. 39, pp. 1418, Feb. 1996. [2] K.-T. Chau, A software tool for learning the dynamic behavior of power electronics circuits, IEEE Trans. Educ., vol. 39, pp. 5055, Feb. 1996. [3] T. H. Sloane, Laboratories for an undergraduate course in power electronics, IEEE Trans. Educ., vol. 38, pp. 365369, Nov. 1995. [4] G. Escobar, R. Ortega, H. Sira-Ramirez, J.-P. Vilain, and I. Zein, An experimental comparison of several nonlinear controllers for power converters, IEEE Control Syst. Mag., pp. 6682, Feb. 1999. [5] L. K. Wong, F. H. Leung, and P. K. S. Tam, A simple large-signal nonlinear modeling approach for fast simulation of zero-current-switch quasiresonant converters, IEEE Trans. Power Electron., vol. 12, pp. 437442, May 1997. [6] E. Allen, N. LaWhite, Y. Yoon, J. Chapman, and M. Ilic, Interactive object-oriented simulation of interconnected power systems using SIMULINK, IEEE Trans. Educ., vol. 44, pp. 8795, Feb. 2001. [7] R. W. Erickson, Fundamentals of Power Electronics. Norwell, MA: Kluwer-Academic, 1999. [8] B. K. H. Wong and H. S. Chung, Steady-state analysis of PWM DC/DC switching regulators using iterative cycle time-domain simulation, IEEE Trans. Ind. Electron., vol. 45, pp. 421432, June 1998. [9] , A systematic graphing technique for small-signal low-frequency characterization of PWM DC/DC converters, IEEE Trans. Ind. Electron., vol. 47, no. 1, pp. 4554, Feb. 2000. [10] J. Xu and M. Grotzbach, Time-domain analysis of half-wave zero-current switch quasiresonant converters by using SPICE, IEEE Trans. Ind. Electron., vol. 40, pp. 577579, Dec. 1993. [11] J. Sun, D. M. Mitchell, M. F. Greuel, P. T. Krein, and R. M. Bass, Average modeling of PWM converters operating in discontinuous conduction mode, IEEE Trans. Power Electron., vol. 16, pp. 482492, July 2001. [12] P. Midya, P. T. Krein, and M. F. Greuel, Sensorless current mode controlAn observer-based techniques for DCDC converters, IEEE Trans. Power Electron., vol. 16, pp. 522526, July 2001. [13] D. Logue and P. T. Krein, Simulation of electric machinery and power electronics interfacing using MATLAB/SIMULINK, in 7th Workshop Computers in Power Electronics, 2000, pp. 3439. [14] J.-H. Su, Power electronics course, Lunghwa Univ. of Science and Technology, Taiwan, R.O.C.. [15] M. H. Rashid, Power Electronics, Circuits, Devices and Applications, 2nd ed. Englewood Cliffs, NJ: Prentice-Hall, 1993. [16] Texas Instruments, Design switching voltage regulators with TL494,, Application Rep., 1998. [17] J. Y. Hung, W. Gao, and J. C. Hung, Variable structure control: A survey, IEEE Trans. Ind. Electron., vol. 40, pp. 222, Feb. 1993.

(b)

(c) Fig. 12. (a) Sliding-mode control of a boost-type converter with input voltage variations. (b) Step responses of a boost-type converter with a sliding-mode controller. (c) Simulation results of the sliding-mode control of a BOOSt-type converter with input voltage variations.

shown in Fig. 12(a). The parameters for the digital boost-type 10 V, switching converter are nominal input voltage 20 V, 170 mH, 1000 F, reference voltage 0.1 , 0.25 , and 100 . The input 0.3 s and returns to the nominal voltage changes to 13 V at 0.35 s. The simulation results in Fig. 12(b) and value when (c) agree with the experimental results in [4].

SU et al.: LEARNING FEEDBACK CONTROLLER DESIGN OF SWITCHING CONVERTERS

315

Juing-Huei Su (S87M93) was born in Tainan, Taiwan, R.O.C., on February 17, 1965. He received the B.S., M.S., and Ph.D. degrees in electrical engineering from the National Taiwan University, Taipei, Taiwan, in 1987, 1989, and 1993, respectively. From 1993 to 1995, he was a Military Officer in the army. In 1995, he was a Senior Engineer with Taian Electric Co., Ltd. Since 1996, he has been an Associate Professor in the Department of Electronic Engineering, Lunghwa University of Science and Technology, Taiwan. Currently, his research interests include robust control theory and power electronic systems.

Dong-Shiuh Wu (S94M96) was born in Yunlin, Taiwan, R.O.C., on September 1, 1964. He received the M.S. and Ph.D. degrees in electrical engineering from National Taiwan University, Taipei, Taiwan, in 1991 and 1995, respectively. From 1984 to 1986, he was a Second Lieutenant in the army of Taiwan. Since 1995, he has been an Associate Professor in the Department of Electronic Engineering, Lunghwa University of Science and Technology, Taiwan. His research interests are in analog and digital integrated circuits and systems.

Jiann-Jong Chen (S94M96) was born in Keelung, Taiwan, R.O.C., on July 23, 1966. He received the M.S. and Ph.D. degrees in electrical engineering from National Taiwan University, Taipei, Taiwan, in 1992 and 1995, respectively. During 19861988, he was an MP in the Chinese army. During 19941995, he was a Teacher in the Department of Electronic Engineering, Lunghwa University of Science and Technology, Taiwan, where he has been an Associate Professor since 1995. He has received one U.S. patent and one R.O.C. patent. His research interests are in analog and digital integrated circuits and systems.