IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 13, NO. 3, MAY 1998 501 A Family of Continuous-Conduction-Mode Power-Factor-Correction Controllers Based on the General Pulse-Width Modulator Zheren Lai, Student Member, IEEE, and Keyue Ma Smedley, Senior Member, IEEE Abstract— This paper presents a family of constant-switching-frequency pulse-width-modulated controllers for singlephase power-factor-correction (PFC) circuits that operate at continuous-conduction mode (CCM). Both trailing- and leadingedge pulse-width modulation (PWM) are used. These controllers do not require the multiplier and rectified-line-voltage sensor, which are needed by traditional control methods, and they can be implemented with a unified control circuit. Controller examples are analyzed and verified experimentally. Index Terms— PFC, power factor correction, PWM, rectifier, switching converter. I. INTRODUCTION A SINGLE-phase diode bridge followed by a dc–dc converter with proper control forms a rectifier with active power-factor correction (PFC). If the controller forces the input current to have the same shape as the input voltage so that the input impedance appears to be resistive, that rectifier is called a resistor emulator. The resistor emulator not only requires a near-unity power factor, but also low harmonic contents in the line current. There are two traditional approaches to control a resistor emulator, namely, the voltagefollower approach and the multiplier approach [1]. The voltage-follower approach realizes a resistor emulator with the constant-duty-ratio or the constant-ON-time control, such as a flyback [2] or a Cuk [3] converter operating at discontinuous-conduction mode (DCM), or a boost converter at the boundary of DCM and continuous-conduction mode (CCM) [4]. The control circuit is simply a voltage-mode pulsewidth-modulation (PWM) chip which does not even require a current sensor. However, the DCM or the boundary operation causes a large current stress on semiconductors and demands more effort to attenuate the current ripple so as to have a satisfactory low electromagnetic interference (EMI) to the line. The input-current ripple of the DCM Cuk converter introduced in [3] can be eliminated by inductor coupling, however, the current stress on semiconductors remains the same. At high-power applications, the current stress and current ripple become too large for a single DCM converter to operate efficiently. Therefore, the voltage-follower approach is not suitable for high-power application. Manuscript received December 16, 1996; revised August 18, 1997. Recommended by Associate Editor, T. Sloane. The authors are with the Department of Electrical and Computer Engineering, University of California, Irvine, CA 92697 USA. Publisher Item Identifier S 0885-8993(98)03426-7. The multiplier approach requires relatively complicated control circuitry. As shown in Fig. 1(a), this approach needs a multiplier, current sensor, and sensor of the input voltage The control method is based on the current mode control. The current reference is the rectified line voltage with its amplitude , the output of the modulated by the modulation voltage feedback compensator. In contrast to the voltage-follower approach, resistor emulators with the multiplier approach operate in CCM so they are suitable for high-power applications. Examples of the multiplier approach include the averagecurrent mode control [5] and the peak-current mode control [6]. The peak-current mode control usually has larger current distortion due to the current ripple, however, its current sensor can be implemented with a current transformer (CT). The CT has fewer losses, thus, it is more desirable in high-power applications. Additional current distortion may result due to the nonlinearity of the multiplier. A number of papers have been dedicated to simplifying the control of the CCM converters [7]–[11]. These papers eliminated the multiplier and the input-voltage sensor in the multiplier approach. Some methods fulfilled that purpose under the penalty of higher current distortion [7]. The nonlinearcarrier (NLC) control, proposed in [8] and [9] for the boost converter and other topologies, initiated a new approach for the resistor emulator. This new approach utilizes the property of the quasi-steady-state operation of the CCM converters to simplify the control circuitry, hence, it may be called the quasisteady-state approach. The linear peak-current mode (LPCM) control proposed in [10] is another example of this new approach. It was found that the LPCM control was a subset of the family of controllers proposed independently in [11], shortly after [10] was published, as application examples of a general PWM modulator [11]. The block diagram of this new approach, shown in Fig. 1(b), demonstrates that these control methods have the similar complexity as the traditional current mode control for dc–dc converters. Advantages of the current mode control, such as the cycle-by-cycle current limitation, are preserved. This paper discusses the common property of the quasi-steady-state approach and further analyzes and verifies the family of controllers proposed in [11]. The common property of this quasi-steady-state approach is discussed first in Section II. Then, the general PWM modulator is reviewed in Section III. The generality of this modulator can be demonstrated with the unified quasi-steady-state approach for resistor emulators presented in Section IV. Section V takes 0885–8993/98$10.00 1998 IEEE 502 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 13, NO. 3, MAY 1998 (a) (b) Fig. 1. Block diagram of PFC circuits at CCM operation (a) with the multiplier-approach controller and (b) with the proposed controllers. the simplest controller among this family as an example and analyzes the stability of the control method and the linecurrent distortion when a small-ripple assumption is not valid. The same treatment is applicable to other converters and controllers. Section VI verifies this family of controllers experimentally with four control methods for resistor emulators with various dc–dc converters. Finally, conclusions are drawn in Section VII. Notation conventions are as follows unless indicated explicitly. Capital letters are used for quantities associated with the steady state of the dc–dc converter, lowercase letters represent time-variant variables, and a quantity in a pair of angular brackets is the local average of the quantity, i.e., the average in each switching cycle. conversion ratio of the dc–dc converter, and is the modulation voltage, as shown in Fig. 1(b), for controlling the amplitude of the line current. When the CCM dc–dc converter is stable, the steady-state duty ratio satisfies (3) is the output voltage and ’s for five topologies where are shown in Table I. Capital characters are used here to indicate the switching-frequency steady state. Substituting (3) into (2) yields the quasi-steady-state approximation (4) is a constant over a line cycle if the output capacitance is large enough, therefore, if is also a constant in that line cycle, is proportional to and the emulated resistance (5) II. QUASI-STEADY-STATE APPROACH A dc–dc converter will reach a steady state if its inputs and load remain unchanged for a long time and the control law governing the converter is stable. The line frequency in Fig. 1 is usually well below the switching frequency, hence, the input voltage of the dc–dc converters can be approximated as a constant in a few consecutive switching periods. The converter tries to converge to its steady state if the control law is stable. Even though the converter can never reach the true switchingfrequency steady state, it operates in the neighborhood, that is, in the quasi-steady state. The important property of the quasisteady-state operation is that all quantities can be approximated with their steady-state values. The control objective of the resistor emulator is to force the input current of the dc–dc converter to be proportional to the input voltage so that the input impedance is resistive. In other words, the local average of the input current (1) is the emulated resistance. can be controlled by where modulating the duty ratio This objective can be accomplished, in general, with the control law (2) is the due to the quasi-steady-state operation, where equivalent current-sensing resistance, is the voltage- can regulate so as to control the input current. Voltage Implementation of (2) relies on the PWM modulator. Various implementations have been shown in [8]–[10]. These implementations as well as some new ones can be unified with the general PWM modulator proposed in [11]. III. THE GENERAL PWM MODULATOR The general PWM modulator is shown in Fig. 2(a). It contains a constant-frequency clock generator, a flip flop (FF), a comparator (CMP), and two stages of an integrator with reset. The number in the circle represents a constant gain of two. and are control inputs to the modulator, and the output can be taken either from or of FF. The integrator performs normal integration unless the control input is at logical high state, which resets the integrator output to zero as result. The time constant of the integrator is selected to equal the switching period The operation waveforms of the general modulator are shown in Fig. 2(b). Each switching period is initiated by the constant-frequency clock. The output of FF rises to logical high state as a clock-pulse arrives and each integrator starts integrating its input signal. When the voltage at the noninverting input of CMP reaches the voltage at the inverting input, CMP outputs a logical high signal, resetting FF and both integrators. LAI AND SMEDLEY: FAMILY OF CONTROLLERS BASED ON PULSE-WIDTH MODULATOR (a) 503 (b) Fig. 2. The two-stage general PWM modulator and its operation waveforms. Assuming that leads to and are constant in a switching period (6) for time from the beginning of a cycle to the moment when Thus, the duty ratio of satisfies (7) The modulation of ’s pulse width is called trailing-edge modulation because the leading edge of pulse, as shown in Fig. 2(b), is synchronized by the clock, and the trailing edge is modulated according to values of and On the contrary, ’s pulse width is leading-edge modulated, and its pulse duty ratio satisfies (8) In a resistor emulator, and can be approximated as constants in a switching cycle, thus, (6)–(8) are valid. In and the next section, we will show how to derive from (2) for various power topologies. The general modulator provides the flexibility of choosing leading- or trailing-edge modulation. For some applications, using leading-edge modulation may have more advantages over using trailing-edge modulation. IV. UNIFIED IMPLEMENTATION OF QUASI-STEADY-STATE APPROACH A family of resistor-emulator controllers described in the format of (7) or (8) can be derived from (2) for various topologies. In this section, the boost, buck–boost, and Cuk topologies are used as examples to demonstrate the derivation and the application of the general modulator. A. The Boost Converter Topology The power stage of the boost topology is given in Fig. 3, where represents the transfer function of the output feedback compensator and is the output of the current sensor. In this topology (9) Substituting the above equation and yields in Table I to (2) (10) Fig. 3. The boost resistor emulator block diagram. The trailing-edge modulation requires the control law to be in the format of (7). Reorganizing (10) gives (11) and Comparing to (7) yields (11) is a control law of average-current mode control because the sensed current is an averaged quantity, which is the same as that in [5]. When the current ripple in the inductor is negligible, the is approximately equal to the average inductor current instant inductor current Thus, one can replace the averagecurrent sensor with a simpler instant-current sensor. In this paper, we assume that the ripple is negligible so that the average and instant-inductor-current controllers are equivalent, called the inductor-current controller. The inductor-current controller for the trailing-edge-modulated boost topology is shown in Fig. 4(a). When this assumption is not valid, the instant-inductor-current control will result in more line-current distortion. The distortion due to the current ripple is analyzed for one example shortly. Inductor-current sensing is usually accomplished with a resistor in series with the inductor, as shown in Fig. 4(a). It is more efficient if a switch-current CT is used. For the boost converter, it is a common practice to replace the inductor current with the switch current in the control circuit because they have the equal value when the switch is on. That results in the instant-switch-current controller, shown in Fig. 4(b). The instant-switch-current control is the same method as the LPCM control [10]. The switch-current control has one disadvantage, that is, the sensed current signal is subject to switching noise. 504 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 13, NO. 3, MAY 1998 Equation (15) is the fundamental control law for the NLC control. Fig. 4(c) shows the average-switch-current controlled boost resistor emulator. Up to this point, the three control circuits derived (four if assume the average and instant inductor-current controls are two different ones) are trailing-edge modulated. One can also find the leading-edge modulation version of these three controllers by converting each control law to the format of (8). Notice (10) is already in this format. Comparing (10) to (8) yields and equal to and zero, respectively. Assuming negligible current ripple, (10) is the inductor-current control law. Fig. 4(d) illustrates the leadingedge-modulated inductor-current controller. The modulation output is from For the leading-edge modulation, the modulation decision is made during the off period of the active switch, as shown in Fig. 8. During that period, the diode current is equal to the inductor current, hence, one can use a CT to sense the diode current instead of the inductor current. The resulting instant-switch-current-controlled boost converter is shown in Fig. 4(e). For noise-immunity reason, one can also with use the average-switch-current sensor. Replacing TABLE I INPUTS TO THE GENERAL PWM MODULATOR FOR PROPOSED RESISTOR-EMULATOR CONTROLLERS (16) results in (17) Usually, one has to use a low-pass filter after the sensor to avoid control error caused by the noise. Employing the average-switch-current sensor, as used in the NLC control [8], enhances the noise immunity. The average-switch-current sensor, as shown in Fig. 4(c), is equivalent to the switch-current sensor cascading with an integrator with reset. Its output (12) where is the sensing capacitance and is the turns ratio of the CT (primary: secondary ). The equivalent sensing resistance (13) The average-switch-current control law can be found by integrating each term in (11) one more time. In fact, the average switch current relates the inductor current by (14) for the boost converter. Substituting (11) leads to from (14) into in (15) The average-switch-current controller is shown in Fig. 4(f). Another three controllers are derived for the boost converter that are leading-edge modulated. These controllers are listed in Table I for application convenience. Inductor-current controllers and switch-current controllers are listed separated because they are not always equivalent, as shown later in the Cuk converter. Notice that some controllers with , therefore, only one stage of integrator-with-reset is actually necessary. Thus, the modulator can be further simplified for these controllers. The leading-edge modulation for the boost converter has some advantages compared to the trailing-edge modulation. 1) The current-sensing circuitry for some boost-derived topologies with multiple switches is simplified. Fig. 5 shows some of these topologies employed in literature [14]–[16]. For trailing-edge modulation, at least two switch-current CT’s are required due to multiple active switches in the topology. For leading-edge modulation, there is a common path for the two or more diode currents, thus, they can be sensed with one CT, as shown in each figure. 2) The switching ripple current in the output filtering capacitor can be reduced. The boost converter is usually used as a preregulator, and the postregulator is normally trailing-edge modulated. This results in less switching ripple current in the capacitor as indicated in [13]. 3) For high-power applications where two boost converters have to operate in parallel, one may parallel a leading-edge-modulated converter with a trailing-edgemodulated one. This may lead to inductor-current ripple and line-current distortion cancellation. LAI AND SMEDLEY: FAMILY OF CONTROLLERS BASED ON PULSE-WIDTH MODULATOR 505 (a) (b) (c) (d) (e) (f) Fig. 4. The boost resistor emulators with various control methods based on the general PWM modulator. B. The Buck–Boost Topology For the buck–boost topology, the derivation starts from the general control law in (2) as well. As shown in Fig. 6 (18) Replacing in (2) with and substituting from Table I directly leads to the average-switch-current control law for trailing-edge modulation (19) For the buck–boost converter, (14) is still valid, hence, one can find the inductor-current control law (20) When the switch is on, , hence, replacing yields the switch-current controller. with Leading-edge-modulated controllers can be derived with the same procedures as for the boost converter. The derivation results are listed in Table I. C. The Cuk Topology The Cuk as well as the Sepic and Zeta topologies are slightly different from the previous two because they have two inductors. We use the Cuk converter as an example to demonstrate the derivation of the controllers. As shown in Fig. 7, the input current is the same as the input inductor current, thus, by substituting , one can find that (21) which is the inductor-current control law. Attention needs to be paid to the value of the switch current. Unlike the previous 506 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 13, NO. 3, MAY 1998 Fig. 7. The Cuk resistor emulator block diagram. Fig. 8. Operation waveforms for the instant-switch-current-controlled boost resistor emulator with leading-edge modulation. Leading-edge-modulated controllers were also found and listed in Table I. The same procedures can be applied to other topologies to find out their resistor-emulator controllers. For the Sepic and Zeta converters, the derived results are the same as the Cuk converter as given in Table I. Notice that at least one controller for each topology in Table I can be implemented with single-integrator modulators because Fig. 5. Boost-derived PFC topologies employed in [14]–[16]. V. STABILITY AND DISTORTION ANALYSIS In order for the resistor emulator to operate successfully, the control law needs to be stable. Furthermore, the line current will have distortion when instant-current control is employed due to the switching-frequency inductor-current ripple. In this section, we will take the switch-current-controlled boost topology with leading-edge modulation as an example to analyze the stability and distortion. Analysis needs only to carry out for dc steady state. Actual quantities can be approximated with the steady-state results because of the quasi-steady-state operation. Fig. 6. The buck–boost resistor emulator block diagram. two topologies, the steady-state switch current (22) and the switch is on. Thus, one can find the switch-current control law to be (23) The average-switch-current control law can be found by reThe control law happens to be the placing same as (19). These derivation results are listed in Table I. A. Stability Analysis Mapping theory [12] is used for this analysis. Assume the the peak inductor current for input voltage is a dc quantity and the inductor current changes linearly the th cycle is with time, as shown in Fig. 8. The moment when FF is reset and satisfies (24) LAI AND SMEDLEY: FAMILY OF CONTROLLERS BASED ON PULSE-WIDTH MODULATOR Replacing with , where 507 , leads to (25) where hand is the switching frequency. On the other (26) hence (27) where Fig. 9. Calculated line current at various power levels. (28) and determines the stability of the current loop [12]. When (29) the circuit will converge to its steady state. Any perturbation will disappear gradually. If the inductor current, as shown in Fig. 8, is higher than the steady-state value, it takes longer to reach for a given value of , thus, the for inductor has longer time to discharge so as to reduce its reaches value. If the inductor current is too small, sooner, resulting a larger duty ratio, hence, the inductor current increases. Subswitching-frequency harmonic oscillation occurs when (29) is not satisfied. The steady state can be found out by letting (26) equal zero, therefore Fig. 10. Theoretical THD versus power levels. (30) Combining (5), (28)–(30) gives another form of the stability condition, that is, (31) In PFC applications, the stability is guaranteed when the line 0.5 The possible situation voltage is at the region that for the instability to occur is when the converter operates at the region near the peak line voltage and simultaneously at light load. The instability is localized and does not spread from one line cycle to another. With certain minimal load condition, the unstable situation can be excluded. B. Line-Current Distortion The line-current distortion is found by solving the steadystate expressions. Combining (25) with (30) yields the steadystate peak inductor current (32) The valley inductor current is For constant and the valley current is proportional to the input voltage for the leading-edge-modulated boost resistor emulator. In the trailing-edge-modulation version of this method, the peak inductor current is proportional to , as shown in [10], thus, paralleling two boost converters with these two controllers, respectively, will result in inductor-current ripple and line-current distortion cancellation. The average inductor current is then found (34) Equation (34) can be used to find the actual average line current. An example is given with the following circuit parameters. For V line voltage, H kHz, and V The normalized line current is calculated and shown in Fig. 9 for and , respectively. Taking as the full load, the line current total harmonic distortion (THD) versus power level is plotted in Fig. 10. The distortion improvement over the peak-current control with multiplier approach [6] is obvious. VI. EXPERIMENTAL VERIFICATIONS (33) Three converters with four different controllers were built and tested. 508 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 13, NO. 3, MAY 1998 (a) (b) (c) Fig. 11. Experimental waveforms at different load power levels. Top: duty ratio (0.5/div). Middle: line voltage (110 Vrms ). Bottom: line current, 5 A/div, 2 A/div, and 0.5 A/div, respectively, for (a)–(c). (a) At full load, (b) 40% of full load, and (c) 10% of full load. A. The Boost Resistor Emulator with Leading-Edge Instant-Switch-Current Control An experimental circuit has been built according to Fig. 4(e) and tested. The line voltage is 110 V , and the output voltage is 220 V The switching frequency is 100 kHz. The inductance of the boost converter is 520 H An EMI filter was inserted between the diode bridge and the boost converter, similar to the one shown in Fig. 13, with an inductance of 44 H and a capacitance of 0.68 F The feedback compensator has a proportional-integral (PI) transfer function. Experimental waveforms for a full load of 350 W, 40% of full load, and 10% of the full load are shown in Fig. 11(a)–(c), respectively. The waveforms in each figure, from top to bottom, are the duty ratio measured with a Tektronix time-to-voltage converter TVC501, line voltage, and line current with its scale marked in the figure, respectively. Notice that in Fig. 11(a) and (b) the current distortion is not significant while in (c) the distortion is noticeable. The current shape in Fig. 11(c) is similar to the calculated waveform shown in Fig. 9. It is difficult to compare the measured THD with the theoretical THD since the line voltage itself has some distortion, however, one can still see the trend of distortion as load decreases. B. Flyback Resistor Emulator with Trailing-Edge Average-Switch-Current Control Fig. 12 shows the flyback PFC converter of trailing-edge average-switch-current control (the feedback compensator is not shown). As indicated in [10], this control method is unconditionally stable. An experimental circuit was built and tested. The major 63 kHz, the primary circuit parameters are as follows: Fig. 12. Flyback PFC converter with trailing-edge average-switch-current control. Fig. 13. Experimental waveforms of the flyback converter at 100-W power output. Top: duty ratio (0.63/div). Middle: line voltage (110 Vrms ). Bottom: line current (1 A/div). inductance is 340 H, the flyback transformer turns ratio is 1:1, H, F, line voltage is 110 V , and the output voltage is regulated at 50 V The experimental LAI AND SMEDLEY: FAMILY OF CONTROLLERS BASED ON PULSE-WIDTH MODULATOR Fig. 14. Cuk converter with instant-inductor-current control. 509 Fig. 16. Cuk converter with average-switch-current control. Fig. 15. Experimental waveforms of the Cuk converter with instant-inductor-current control at 150-W power output. Top: duty ratio (0.5/div). Middle: line voltage (110 Vrms ). Bottom: line current (2 A/div). Fig. 17. Experimental waveforms of the Cuk converter with average-switch-current control at 150-W power output. Top: duty ratio (0.5/div). Middle: line voltage (110 Vrms ). Bottom: line current (2 A/div). result is demonstrated in Fig. 13. The line-current distortion is insignificant. VII. CONCLUSIONS C. Cuk Resistor Emulators To verify the control methods for the Cuk, Sepic, and Zeta converters, a trailing-edge instant-inductor-current and an average-switch-current controlled Cuk PFC converter were built and examined. The circuits are illustrated in Figs. 14 and 16, respectively, with their feedback compensators not shown. The same power stage were used for both control methods. The switching frequency is 100 kHz, and other major parameters are shown in the figures. The line voltage was 110 V , and both outputs were regulated at 50 V Figs. 15 and 17 are the experimental waveforms at 150-W power output. From top to bottom, the waveforms are the duty ratio, line voltage, and line current, respectively. Notice that both line currents are closely following the line voltage. It was observed that for the average-switch-current control, the line current had some ringing at around 2 kHz when the rectified line voltage changed slope rate abruptly, i.e., when the line voltage crosses zero or is at its peak for the distorted line voltage. The ringing is still noticeable in Fig. 17 after resistive damping is used to clean up the waveforms. With instantinductor-current control, the ringing is less severe for the same power stage. In other words, different control methods may provide different dynamic performance for the same power stage. Further investigation of the dynamics is beyond the scope of this paper. A family of resistor emulator controllers for the quasisteady-state approach are presented in this paper based on the general PWM modulator. The property and general control law for the quasi-state-state approach are discussed. The derivation procedures are given in detail with derivation results for five commonly used converters listed in Table I. Both trailing- and leading-edge modulation can be realized at constant switching frequency. Leading-edge modulation can sometimes lead to simpler control circuitry as demonstrated in the boost converter example. Both leading- and trailing-edge modulation have three basic control circuits for one converter according to the way that the current is measured. Physically, the average-switch-current control is equivalent to the instant-switch-current control in the sense that the former is the integration of the latter. The PFC circuits are ideal resistor emulators when the switching-frequency inductor-current ripple is zero. In practice, the line-current distortion due to the current ripple can be analyzed for each specific control method. For this family of controllers, the rectified-line-voltage sensor, error amplifier in the current loop, and multiplier in the voltage feedback loop that exist in a traditional CCM PFC circuit are eliminated, hence, the control circuitry is simplified. The performance of these PFC circuits is comparable to or improved over those traditional CCM converters with the multiplier-approach control. The most important advantage of these controllers is that their implementation is unified. 510 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 13, NO. 3, MAY 1998 One can see from the three converters and four control methods experimentally verified in this paper that they share an identical PWM modulator. Therefore, these controllers are well suited for integrated-circuit implementation. [14] P. N. Enjeti and R. Martinez, “A high performance single phase ac to dc rectifier with input power factor correction,” in APEC’93, pp. 190–195. [15] A. Pietiewicz and D. Tollik, “New high power single-phase power factor corrector with soft-switching,” in Intelec’96, pp. 114–119. [16] E. X. Yang, Y. Jiang, G. Hua, and F. C. Lee, “Isolated boost circuit for power factor correction,” in APEC’93, pp. 196–203. ACKNOWLEDGMENT The authors would like to thank Dr. D. Maksimovic of the University of Colorado at Boulder for correcting errors in Table I in an earlier version of this paper. REFERENCES [1] J. Sebastian, M. Jaureguizar, and J. Uceda, “An overview of power factor correction in single-phase off-line power supply systems,” in IECON’94, pp. 1688–1693. [2] R. Erickson, M. Madigan, and S. Singer, “Design of a simple highpower-factor rectifier based on the flyback converter,” in APEC’90, pp. 792–801. [3] M. Brokovic and S. Cuk, “Input current shaper using Cuk converter,” in INTELEC’92, Washington, DC, Oct. 1992, pp. 532–539. [4] J. Lai and D. Chen, “Design consideration for power factor correction boost converter operating at the boundary of continuous conduction mode and discontinuous conduction mode,” in APEC’93, pp. 267–273. [5] L. Dixon, “Average current mode control of switching power supplies,” presented at the Unitrode Power Supply Design Seminar, 1990. [6] R. Redl and B. P. Erisman, “Reducing distortion in peak-currentcontrolled boost power-factor correctors,” in APEC’94, pp. 576–583. [7] D. Maksimonvic, “Design of the clamped-current high-power-factor boost rectifier,” in APEC’94, pp. 584–590. [8] D. Maksimonvic, Y. Jang, and R. Erickson, “Nonlinear-carrier control for high power factor boost rectifiers,” in APEC’95, pp. 635–641. [9] R. Zane and D. Maksimovic, “Nonlinear-carrier control for highpower-factor rectifiers based on flyback, Cuk or SEPIC converters,” in APEC’96, pp. 814–820. [10] J. P. Gegner and C. Q. Lee, “Linear peak current mode control: A simple active power factor correction control technique for continuous conduction mode,” in PESC’96, June 1996, pp. 196–202. [11] Z. Lai and K. M. Smedley, “A general constant frequency pulse-width modulator and its applications,” in HFPC’96, Las Vegas, NV, Sept. 1996. [12] F. C. Moon, Chaotic Vibrations—An Introduction for Applied Scientists and Engineers. New York: Wiley, 1987. [13] R. A. Mammano, “New developments in high power factor circuit topologies,” in HFPC’96, Las Vegas, NV, Sept. 1996. Zheren Lai (S’95) received the B.S. and M.S. degrees in electrical engineering from Zhejiang University, China, in 1989 and 1993, respectively. He is currently working toward the Ph.D. degree at the University of California (UCI), Irvine. From 1989 to 1992, he worked on active power filters. He was a Research Engineer at the Electrical and Electronic Engineering Company, Hangzhou, China, for a short period in 1993. Since joining UCI in the fall of 1993, his research has been concentrated in the control of switching converters, including switching audio power amplifiers and power-factor-corrected rectifiers. Keyue Ma Smedley (SM’97) received the B.S. and M.S. degrees in electrical engineering from Zhejiang University, China, in 1982 and 1985, respectively, and the Master and Ph.D. degrees in electrical engineering from the California Institute of Technology, Pasadena, in 1987 and 1991, respectively. She was an Engineer at the Superconducting Super Collider from 1990 to 1992, where she designed specific ac–dc conversion systems for all accelerator rings. She joined the Faculty of Electrical and Computer Engineering at the University of California, Irvine, in 1992, where she established a state-of-the-art power electronics laboratory. Her research interests include modeling, control, topologies, and integration of switching converters, inverters, class-D power amplifiers, softswitching techniques, power-factor-correction methods, laser current sources, power conversion for alternative energy sources, etc. She currently holds the U.S. patent for one-cycle control. Dr. Smedley is the Chair of the Constitution and Bylaws Committee of the IEEE Power Electronics Society, an Associate Editor of the IEEE TRANSACTIONS ON POWER ELECTRONICS, an Associate Editor of the the Advisor of the IEEE UCI Student Chapter, a Faculty Member of Eta Kappa Nu, and a Member of the Power Sources Manufacturer’s Association.