IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 42, NO.

6, DECEMBER 1995

595

Soft-Switching Techniques in PWM Converters

Guichao Hua and Fred C. Lee, Fellow, ZEEE
Abstruct-Xecently, a number of soft-switching pulse-widthmodulated (PWM) converter techniques have been proposed, aimed at combining the desirable features of both the conventional PWM and resonant converters while avoiding their respective limitations. In this paper, three classes of zerovoltage soft-switchingPWM converters (namely the zero-voltageswitched (ZVS) quasi-square-wave converters, ZVS-PWM converters, and zero-voltage-transition PWM converters) and two classes of zero-current soft-switching PWM converters (namely, the zero-current-switched PWM converters and zero-current-transition PWM converters) are reviewed, and their merits and limitations are assessed. Experimental results of several prototype of converters are presented to illustrate each class of converters.

I. INTRODUCTION 0 accommodate the ever increasing requirements for smaller size, lighter weight, and higher efficiency power supplies, switched-mode power conversion technologies have evolved from the basic PWM converters to resonant converters, quasi-resonant converters, multi-resonant converters, and most recently to soft-switching PWM converters. The PWM converters process power by interrupting the power flow by means of abrupt switching. This hard-switching operation results in significant switching losses, switching noise, and switching stresses, especially at high frequencies. To improve switching conditions for semiconductor devices in PWM converters, several resonant techniques were proposed. The resonant converters, which include the traditional series and parallel resonant converters, class-E converters [11, quasiresonant converters (QRC's) [2], [3], 161, and multi-resonant converters 141-[7], process power in a sinusoidal or quasisinusoidal form. The power switches are commutated with either zero-voltage switching (ZVS) or zero-current switching (ZCS), thus switching losses and stresses of the resonant converters are significantly reduced in comparison with the PWM converters. However, due to the resonant nature of the current and voltage waveforms, the operation of resonant converters usually involves high circulating energy which results in a substantial increase in conduction losses [8]. In addition, due to wide linefload range, most resonant converters operate with a wide switching frequency range, thus making the circuit design difficult to optimize. As a compromise between the PWM and resonant techniques, various soft-switching PWM converter techniques were proposed recently aimed at achieving soft-switching
Manuscript received September 1, 1994. G. Hua is with Virginia Power Technologies Inc., Blacksburg, VA 24060 USA. F. C. Lee is with the Virginia Power Electronics Center, Department of Electrical Engineering, Virginia Polytechnlc Institute and State University, Blacksburg, V 24061 USA. A IEEE Log Number 9415125.

TO T1

T2 T3

T O

Fig. 1. Circuit diagram and waveforms of the buck ZVSQSC.

without a significantly increase in circulating energy. Generally, a soft-switching converter utilizes some form of resonant technique to soften the switching transition. When switching transition is completed, the converter reverts back to the familiar PWM mode of operation so that the circulatory energy can be minimized. Thus switching losses are reduced at the cost of a minimal increase of conduction losses. In this paper, the operation principles of three classes of zero-voltage soft-switching PWM converters (including the ZVS quasi-square-wave converters (QSC's) 191-1111, ZVSPWM converters 1121-[ 141, and zero-voltage-transition (ZVT) PWM converters [15]-[22]) and two classes of zero-current soft-switching PWM converters (including the ZCS-PWM converters [21], [23], [24], and zero-current-transition (ZCT) PWM converters [21], 1261) are presented, and their ad-

0278-0046/95$04.00 0 1995 IEEE

596

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 42, NO. 6, DECEMBER 1995

vantages and disadvantages are compared. The experimental results of several prototype converters are also presented.

placed in series with the MOSFETs and a second, fastrecovery, diode be placed antiparallel to the MOSFETs [ 111. Fig. 2 shows a four-module, multiphase ZVS bidirectional 11. ZERO-VOLTAGE SOFT-SWITCHCNG PwM CONVERTERS battery charger/&scharger designed for the NASA EOS satelVarious studies have concluded that for high-frequency lite [Ill. The input is 120-V regulated bus, and the output power conversion where power MOSFETs are used as power is battery. The battery voltage varies between 64-84 V, switch(es), ZVS is the preferred operating mode. In this depending on the depth of discharge. The maximum load is section, three classes of zero-voltage soft-switching PWM 1800 W. Due to the interleaving structure, the input and output converters, namely the ZVS-QSCs, ZVS-PWM converters, current ripples are greatly reduced. During both charge mode and discharge mode, the converter efficiency is above 96% at and ZVT-PWM converters, are reviewed. full load. A. ZVS Quasi-Square- Wave Converters In a ZVS-QRC, a resonant inductor is employed to achieve soft-switching for the active switch. The resonant inductor is required to be of sufficient magnitude to store enough energy to discharge the devices junction capacitance. ZVS is difficult to achieve at light load. In a ZVS-QSC, however, the filter inductor is utilized to serve as the resonant inductor to achieve ZVS for the power switches. The circuit diagram and key waveforms of the buck ZVS-QSC is given in Pig. 1, where C, is the resonant capacitor that incorporates the output capacitances of the transistors, and S1 is used for constant frequency and bidirectional power flow. Detailed operation of this converter is referred to in [lo], [Ill. The features of the ZVS-QSCs are summarized as follows: Advantages: ZVS for both transistors, minimum switch voltage stresses: Both switches are subjected to minimum voltage stresses, the same as those in their PWM counterpart. This is a very desirable feature for high-frequency conversion where MOSFETs are used, since power MOSETs favor the ZVS operating mode, and their on-resistance is strongly dependent on their voltage ratings. Bidirectional power flow: Since both switches in a ZVS-QSC are bidirectional switches, a ZVS-QSC is naturally suited for bidirectional power conversion. For instance, the circuit shown in Fig. 1 can be used for a battery chargeddischarger [113. Disadvantages: High transistor peak current: To achieve ZVS, the peak turn-off current of the main switch has to be more than two times as high as that in its PWM counterpart. As a result, the conduction losses of the switches are increased by approximately 40% when compared to its PWh4 counterpart. In addition, the high turn-off current of the main switch tends to increase the turn-off loss. This technology is not deemed desirable when minority-carrier power devices, such as IGBTs and BJTs, are used as the power switches. Transformer leakage inductance is not utilized, and * High input and output current ripples. One good application of the ZVS-QSC shown in Fig. 1 is for battery charge/discharge. Due to ZVS, the MOSFET body diode naturally commutates and does not suffer from reverse recovery problem. Proper implementation of a conventional bidirectional converter, however, requires that a diode be B. ZVS-PWM Converters The ZVS-QRC technique eliminates the capacitive turn-on loss which plagues PWM converters and ZCS-QRCs [3]. However, the ZVS-QRC technique has several limitations. First, the power switch in a single-ended ZVS-QRC suffers from an excessive voltage stress which is proportional to the load range. Second, a wide switching frequency range is required for a ZVS-QRC to operate with a wide input voltage and load range. The wide frequency range makes design optimization of the power transformer, inpudoutput filters, and control circuits difficult. Another limitation of the ZVS-QRC technique is severe parasitic ringing between the resonant inductor and the diode junction capacitance, which results in an increase of the switching noise and a possible instability in the closed-loop system [4]. The ZVS-PWM converters can be considered as hybrid circuits of ZVS-QRCs and PWM converters. A ZVS-PWM converter can be derived by adding an auxiliary switch across the resonant inductor in a ZVS-QRC. As an example, Fig. 3 shows the circuit diagram and the key waveforms of the ZVSPWM buck converter, where L, is the resonant inductor and S1 is the auxiliary switch. Detailed operation of this converter is addressed in [12]. The use of the auxiliary switch essentially creates a freewheeling stage (T1-T2) within the quasi-resonant operation. The advantages of this additional freewheeling stage are twofold. First, it enables constant-frequency operation by controlling the time interval of this freewheeling stage. Second, the freewheeling stage could occupy a substantial portion of a cycle so that the proposed circuit resembles that of a conventional PWM converter. Resonant operation takes place only during a small portion of a cycle and is used only to create a ZVS condition for the power switch. In this way, the circulating energy required for ZVS-QRC operation can be significantly reduced. Simply by adding an auxiliary switch across the resonant inductor in ZVS-QRCs, a family of ZVS-PWM converters can be derived. The advantages and disadvantages of the ZVS-PWM converters are summarized below: Advantages: ZVS for the power switch, low current stress of the power switch. Disadvantages: * High voltage stress of the power switch: One limitation of the single-ended ZVS-PWM converters is that the power switch suffers from a high voltage

HUA AND LEE: SOFT-SWITCHING TECHNIQLES IN PWM CONVERTERS

597

p^

Fig. 2. Four-module, multiphase battery charger/discharger using ZVS-QSC buc!dboost topology.

stress that is proportional to the load range under which ZVS is maintained [12]. The rectifier diode is not operated with favorable switching condition. Among the family of ZVS-PWM converters, a particular member is the well-known full-bridge (FB) ZVS-PWM converter [13], [14] as shown in Fig. 4. This converter shares the same topology as the FB-ZVS-QRC. However, its operation differs from the latter by the use of phase-shift control, which essentially creates an extra freewheeling operation stage during which the resonant inductor current travels through the upper or lower two switches. No auxiliary switch is needed in this particular topology. As a bridge-type converter, the power switches in this converter are subjected to minimum voltage stress the same as those in its PWM counterpart. The FB-ZVSPWM converter is deemed favorable for many high-power conversion applications. For example, Fig. 4 shows the power stage circuit diagram of a 100-kHz, 8-kW FB-ZVS-PWM converter with active clamp [14]. The specifications of the breadboarded converter are the following: input voltage, Vi = 325-400 V, output voltage, Vo = 360 V fully isolated, and output power Po = 8 kW.

The experimental waveforms of the breadboarded converter at full load are also shown in Fig. 4. It can be seen that all the primary waveforms are quite clean because of phase-shift control and ZVS operation. In addition, due to the use of the active clamp network, the secondary waveforms are also free of parasitic oscillations. The measured efficiency of the converter at full load is 95.8%.
C. Zero-Voltage-Transition PWM Converters

One common characteristic of resonant-type topologies (as well as the ZVS-PWM and ZCS-PWM converter topologies) is that they all employ a resonant inductor in series with the power switch or the rectifier diode to shape switch voltagekurrent waveforms. Soft-switching is achieved by utilizing the resonance between this resonant inductor and certain resonant capacitors, which are usually in parallel with the semiconductor devices. Due to the fact that these resonant elements are placed in the main power path, the resultant resonant converters are always subjected to inherent problems. First, since the resonant inductor is subjected to bidirectional voltage, it inevitably generates additional voltage stress on the semiconductor devices. Second, since all the power flows through the resonant inductor, substantial circulating energy is always created, which significantly increases conduction

598

EEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 42, NO. 6, DECEMBER 1995

SI

"

DS

'S

Vp, 100 V/div Ip, 20 Ndiv

Vrect, 200 Wdiv

I
To TI T2 T3

I T4

II m

IC, 10 Ndiv

Fig. 3. Circuit diagram and waveforms of the ZVS-PWM buck converter.

Fig. 4. Power stage circmt diagram of the 100-kHz, 8-kW FB-ZVS-PWM converter with active clamp and its experimental waveforms.

losses. In addition, the energy stored in the resonant inductor strongly depends on the line voltage and load current. Therefore, soft-switching condition is sensitive to line voltage and load current changes. This is why most resonant converters are unable to maintain soft-switching for a wide line and load range. To alleviate the above-mentioned limitations, it is necessary to remove the resonant element(s) from the main power path. Instead of using a series resonant element, an altemative way is to use a shunt resonant network across the power switch. Dunng the switching transition, the shunt resonant network is activated to create a partial resonance to achieve ZVS or ZCS. When switching transition is over, the circuit simply reverts back to the familiar PWM operating mode. In this way, the converter can achieve soft-switching while preserving the advantages of the PWM converter. In fact, the above-suggested concept of using a shunt resonant network has been adopted in the ZVT-PWM and the ZCT-PWM converters [151, [161, [24]. Among various zero-voltage soft-switching PWM techniques, the recently developed ZVT-PWM technique is deemed most desirable. By using an auxiliary shunt network, a ZVT-PWM converter achieves soft-switching for both the transistor and the rectifier diode while minimizing their

voltage and current stresses [15]-[20]. This is a technology proposed for the first time, to realize soft-switching in PWM converters without imposing additional switch voltage and current stresses or conduction losses. As an example, Fig. 5 shows the circuit diagram and the key waveforms of the boost ZVT-PWM converter. It differs from a conventional boost PWM converter by possessing an additional shunt resonant network consisting of a resonant inductor (&), an auxiliary switch (Sl), and a diode (Dl). C, is the resonant capacitor, which incorporates the output capacitance of the power switch and the junction capacitance of the rectifier. Detailed operation of this converter is described in [16]. The features of the ZVT-PWM boost converter are summarized as follows: Advantages: 0 Soft-switching for both the transistor and rectifier diode: In addition to the power switch, the rectifier diode in a ZVT-PWM converter is also commutated with ZVS. This feature makes the ZVT-PWM technique particularly attractive for high-voltage conversion applications, where the rectifier diodes suffer from severe reverse recovery

HUA AND LEE SOm-SWITCHING TECHNIQUES IN PWM CONVERTERS

599

load current and input voltage. At light load or high line, ZVS is usually difficult to maintain [12]. The situation is opposite in a ZVT-PWM converter. As long as a ZVT converter is designed to operate with soft-switching at full load and low line, soft-switching operation will be ensured for the whole load and line range [15], [16]. Disadvantages: Transformer leakage inductance not utilized: Similarly to the ZVS-QSC technique, the limitation of TO S the isolated ZVT-PWM converters is that they do not utilize the leakage of the power transformer. Therefore, the transformer should be designed with a minimum leakage. Hard switching for the auxiliary switch: One limitation of the ZVT-PWM technique is that the auxiliary switch does not operate with soft-switching. However, the switching losses involved in the operation of the auxiliary resonant branch are typically much lower than those of a PWM converter. First, the major switching loss that occurs in the ZVT-PWM converter is the capacitive turn-on Ioss of the auxiliary switch (S1). It is much lower than the capacitive turn-on loss of the main switch in a PWM converter due to the fact that S1 only handles a much lower rms current, and thus a smaller MOSFET with lower output capacitance can be used as S1. The turn-off loss of S1 is negligible if S1 is implemented by a MOSFET, and its gate-drive impedance is sufficiently low. Second, the auxiliary diode To T I T2 T3 T4 T5 T6 TO of a ZVT-PWM converter always operates with ZCS. Thus it does not suffer from a reverse-recovery problem. Fig. 5. Circuit diagram and key Waveforms of ZVT-PWM boost converter. For a PWM converter, however, the reverse-recovery loss of the rectifier diode normally dominates the total switching loss in high voltage applications, where p-n problems. For instance, in a PFC boost circuit, both the junction diodes are used. For this reason, the ZVT-PWM power switch and the rectifier diode are subjected to converters are particularly deemed attractive for highhigh voltage. With the conventional PWM, or the ZVSvoltage applications where the reverse-recovery problem QRC, or the ZVS-PWM technique employed, due to the of the rectifier diode is of important concern. reverse recovery of minority-carrier rectifier diode, the switching loss, EM1 noise, and device failure problems Fig. 6 shows the circuit diagram and typical experimental become more pronounced. Therefore, implementing soft- waveforms of a 100-kHz, 600-W PFC circuit using the ZVTswitching for both the transistor and the rectifier diode in PWM boost topology [16], [17]. To reduce the cost of the such a circuit is particularly rewarding [16]-[20]. circuit, an IR TO-220 package IGBT, IRGBC30U, is used as the power switch. To reduce the IGBT turn-off loss and further Minimum switch voltage and current stresses: The voltage and current waveforms of the switches in a reduce EM1 noise, a 4.4-nF external resonant capacitor (C,) is ZVT-PWM converter are essentially square-wave except used to soften the switching actions. The breadboarded circuit during the tum-on and turn-off switching intervals, when is regulated at 380 V output with a 90-260 VAC universal ZVS transition takes place. Both the power switch and input range. To compare the performance of the ZVT PFC the rectifier diode are subjected to minimum voltage and circuit with that of the PWM circuit, the auxiliary resonant current stresses. In addition, the ZVT time intervals can branch is removed. Fig. 7 shows the efficiency measurement be very short with respect to the new converter resembles of two PFC circuits. It can be seen that the ZVT technique that of the boost PWM converter during most portions of significantly improves the circuit efficiency. Moreover, owing the cycle. Circulating energy employed to realize ZVS is to soft-switching operation, the ZVT circuit also significantly therefore minimal. In addition, a low-current MOSFET reduces circuit switching noise. Because of low voltage and can be used for the auxiliary switch as it only handles current stresses of the semiconductor devices, the ZVT-PWM technique is well-suited for high-power applications. For insmall amounts of resonant-transition energy. stance, the ZVT-PWM boost converter has been successfully Soft-switching for wide line and load range: One drawback of ZVS-QRC and ZVS-PWM tech- employed to implement a 6-kW PFC circuit with a 180-560 niques is that the ZVS condition is strongly dependent on VAC input range.
Lf
D

600

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 42, NO. 6, DECEMBER 1995

Lf

MUR850

98

38OV

96

94

15uH

92

,
#

, ,

I
100
150

200
Line Voltage (VAC)

250

300

Efficiency comparison of 100-kHz IGBT PFC circuits using ZVT-PWh4 and conventional PWM boost converters. Fig. 7.

1231. The ZCS-PWM converters can be considered the hybrid circuits of the ZCS-QRCs and the PWM converters. Contrary to a ZVS-PWM converter, the transistor in a ZCS-PWM The concept of ZVT as illustrated in the boost ZVT-PwE/I converter operates with ZCS and the rectifier diode with ZVS. converter can be extended to any switched-mode power con- Fig. 8 shows the circuit diagram and typical waveforms of versiodinversion topology. As an example, the ZVT technique the ZCS-PWM boost converter. It differs from the ZCS-QRC has been successfully employed to implement a 10-kW three- buck topology by possessing an auxiliary switch (Sl) that is phase rectifierhverter [181. in series with the resonant capacitor. The advantages and disadvantages of the ZCS-PWM con1 1 ZERO-CURRENT 1. SOFTSWITCHING PWM CONVERTERS verters are summarized as follows: Advantages: Due to continuous improvement of switching characteristics, ZCS for the transistor, lower conduction losses, and lower cost, IGBTs are gaining ZVS for the rectifier diode. wide acceptance in switched-mode power convertershnverters. Since IGBT is a minority-carrier devices, it exhibits a cmDisadvantages: rent tail at turn-off which causes considerably high tumHigh voltage stress of the rectifier diode: off switching losses. To operate IGBTs at relatively high The rectifier diode experiences a voltage stress twice switching frequencies, either the ZVS or the ZCS technique as high as that in its PWM counterpart. can be employed to reduce switching losses. Basically, ZVS Severe parasitic ringing across the power switch: eliminates the capacitive turn-on loss, and reduces the turn-off Due to the series resonant inductor and the output caswitching loss by slowing down the voltage rise and reducing pacitance of the switch, the power switch in a ZCS-PWM the overlap between the switch voltage and switch current. converter sees severe parasitic voltage ringing at turnThis technique can be effective when applied to a fast IGBT off. To reduce switch voltage stress and switching noise, with a relatively small current tail. However, employing ZCS dissipative snubbers are usually required to suppress this technique eliminates the voltage and current overlap by forcing type of parasitic ringing. the switch current to zero before the switch voltage rises. * ZCS sensitive to line voltage and load change [26]. ZCS is deemed more effective than ZVS in reducing IGBT Among the family of ZCS-PWM topologies, one particuswitching losses, particularly for slow devices [24], [25]. In larly interesting topology is the FB-ZCS-PWM converter [24], this section, two classes of zero-current soft-switching PWM which is a dual circuit of the FB-ZVS-PWM converter. All converters namely the ZCS-PWM converters [21], [23], [24], the semiconductor devices operate with soft-switching and are and ZCT-PWM converters [21], [26], are briefly reviewed. subjected to low voltage and current stresses associated with those in their PWM counterparts. Thus switching losses are A. Zero-Current-Switched PWM Converters greatly reduced without a significant increase of conduction Simply by applying circuit duality to the ZVS-PWM con- losses. In practice, this converter can be suited for high-power verters,, a family of ZCS-PWM converters are derived [21], applications.
Fig. 6. Circuit diagram of the lOO-ldIz, 600-W PFC circuit using ZVT-PWM converter and IGBT and its typical operating waveforms.

HUA AND LEE SOIT-SWITCHING TECHNIQUES IN PWM CONVERTERS

60 1

'Lr

To Ti T2

T3

T4

TO

Fig. 9. Circuit diagram and key waveforms of the ZCT-PWM boost converter

To TI

Tz

T3

T4 Ts

To

Fig. 8. Buck ZCS-PWM converter and its waveforms.

B. Zero-Current-Transition PWM Converters The concept of using a shunt resonant network has been applied to implement ZCT [26]. For example, Fig. 9 shows the circuit diagram and key waveforms of the ZCT-PWM boost converter. The converter differs from a conventional PWM boost converter by the introduction of a resonant branch, which consists of a resonant inductor, L,, a resonant capacitor, C,, an auxiliary switch, S1, and an auxiliary diode, D1. This resonant branch is active only during a short switching-transitiontime to create the ZCS condition for the main switch. It is interesting to note that the basic ZCS mechanism of a ZCT converter is
somewhat similar to that of the McMurray converter [25].

The features of the ZCT-PWM converters are summarized in the following: Advantages: ZCS for the power switch; Low voltage/current stresses of the power switch and rectifier diode:

The ZCT-PWM technique implements ZCS turn-off for the power transistor without penalizing the voltage stresses of both the power transistor and the rectifier diode. Although the main switch current waveform exhibits a resonant peaking, it does not increase the conduction loss, since the average current through the power switch (IGBT) is essentially the same compared with its PWM counterpart; Minimal circulating energy: It is revealed that regardless of the line and load changes, the energy stored in the resonant tank will always be adaptively adjusted so that it is only slightly higher than what is needed for creating the ZCS condition 1261. wide line and load ranges for ZCS. Disadvantages: The rectifier diode operates with hard switching: For high output voltage applications, using a fastrecovery diode as the rectifier diode is required. A 100-kHz, 1-kW ZCT-PWM boost converter was implemented to illustrate the operation of ZCT-PWM converters [26]. The circuit is regulated at 400 V output with a 200-300 V input range. In the breadboarded converter, the main power

602

EEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 42, NO. 6, DECEMBER 1995

resonant converters. Thus switching losses are reduced at a minimd increase of conduction losses. 98 Three classes of zero-voltage soft-switching PWM converters are reviewed in this paper. The ZVS-QSC technique offers ZVS for the power switches without increasing their 96 voltage stress. The limitation of the ZVS-QSC technique associates with the high peak current stresses of the switches, which are more than twice those in their PWM counterparts. 94 The ZVS-PWM technique is an extension of the ZVS-QRC \ technique. For single-ended topologies, the disadvantage of \ the Z V S - P W technique is that the main switch suffers from 92 a high voltage stress which is proportional to the load range under which ZVS is maintained. A particular topology in this family, the bridge-type ZVS-PWM topology, is deemed most 90 desirable, since the switches are subjected to minimum voltage 200 400 600 800 Io00 stresses. This circuit is widely used today in medium to high output Power (w) power applications. Fig. 10. Efficiency comparison of 100 kHz ZCT-PWM and PWM boost Among the introduced three classes zero-voltage softconverters using IGBT. switching PWM convertegs, the ZVT-PWM converters are deemed most desirable since they combine the advantages switch is implemented by an IR fast-series IGBT, IRGPF40 of both the PWM and resonant techniques while overcoming (VCE = 600 V, IC = 40 A, t, = 37 nS, and T f = 420 nS, their respective drawbacks. Since the shunt resonant network rated for up to 8 kHz switching frequency operation). Since is activated only during switching transition time, both the the auxiliary switch only handles a little resonant transition transistor and the rectifier in a ZVT-PWM converter operate energy, a small MOSFET, IRF830, is employed. L, and C, with ZVS and are subjected to minimum voltage and current are selected at 10 uH and 8.2 nF, respectively. It is shown stresses. In addition, soft-switching operation of the ZVTthat ZCS is always maintained when the line voltage or load PWM converters can be easily maintained for a wide line current changes in a wide range. It is shown that the IGBT and load range. The ZVT-PWM technique can be used for turn-off current tail in the ZCT-PWM converter is essentially many practical power conversiodinversion applications. This alleviated [26]. Fig. 10 shows the efficiency measurements of technology is widely used today for PFC applications. For high-power applications where minority-carrier devices ZCT and PWM boost converters. It can be seen that the ZCT such as BJTs, IGBTs, and GTOs are used as the power technique significantly improves the efficiency. switches, the ZCS technique is deemed more desirable than ZVS in reducing switching losses. Two families of zero-current IV. SUMMARY soft-switching converters are introduced in this paper. As a Switching losses, stresses, and noise due to circuit parasitics dual circuit technology to the ZVS-PWM technique, the ZCSare inherent in the conventional PWM technique and are PWM converters offer ZCS for the power switch and ZVS for major factors that restrict PWM converters from operating at the rectifier diode. The limitations of the ZCS-PWM technique, a higher frequency for sizelweight reduction and for perfor- which include severe parasitic ringing on power switch, high mance improvement. To alleviate these problems, numerous voltage stress of the rectifier diode, and the line and load resonant techniques were developed. Typically, a resonant dependence of the ZCS condition, are associated with the use converter incorporates a certain type of resonant network of the resonant inductor in series with the power switch. By into a PWM topology to shape the switch voltage/cment applying a shunt resonant network, the ZCT-PWM convertwaveforms so that the power switches are commutated with ers implement ZCS turn-off for the power transistor while either ZVS or ZCS. The improved switching conditions enable retaining the advantages of conventional PWM converters. In the resonant converters to operate at much higher frequencies addition, ZCT operation is independent of line and load conwith significantly reduced switching losses, stresses, and noise. ditions, and the circulating energy of the converter is always Unfortunately, due to high circulating energy, the switches maintained minimum. These features make the ZCT-PWM in most resonant converters experience either high voltage technique attractive for many applications where the minoritycarrier devices such as IGBTs and MCTs are employed. stresses, high current stresses, or both. To facilitate soft-switching operation while preserving the merits of the PWM technique, a number of soft-switching REFERENCES PWM techniques were proposed. As hybrid circuits between [11 R. Redl, B. Mohar, and N. 0. Sokal, Class E resonant regulated dc/dc the PWM and resonant converters, the soft-switching PWM power converters: Analysis of operation and experimental results at 1.5 converters utilize some form of partial resonance to soften the MHz, IEEE Trans. Power Electron., vol. PE-1, no. 2, pp. 111-120, Apr. 1986. switching process. When switching transition is completed, the [2] K. H. Liu, R. h g a n t i , and F. C. Lee, Quasiresonant convertconverter reverts back to the familiar PWM mode of operation ers-Topologies and charactenstics, IEEE Trans. Power Electron., so that the circulatory energy can be minimized compared to vol. PE-2, pp. 62-74, 1987.
Efficiency (%)

HUA AND LEE: SOIT-SWITCHING TECHNIQUES IN PWM CONVERTERS

603

[3] K. H. Liu and F. C. Lee, Zero-voltage switching technique in dc-dc converters, in IEEE Power Electron. Spec. Con$ Rec., 1986, pp. 58-70. [4] W. A. Tabisz and F. C. Lee, Zero-voltage-switching multi-resonant technique-A novel approach to improve performance of higbfrequency quasiresonant converters, in IEEE Power Electron. Spec. Conf Rec.,-1988, pp. 9-17. W. A. Tabisz, M. M. Jovanovic, and F. C. Lee, High-freauencv multi- -~ resonant converter technology and its applications, in Int. Con$ Power Electron. Variable Speed Drives, 1990, pp. 1-8. F. C. Lee, W. A. Tabisz, and M. M. Jovanovic, Recent developments in high-frequency quasiresonant and multi-resonant converter technologies, in 3rd European Con$ Power Electron. Applicat., 1989, pp. 401410. M. M. Jovanovic, R. Farrington, and F. C. Lee, Constant-frequency multi-resonant converters, in Virginia Power Electron. Center (VPEC) Sem. Proc., 1989, pp. 49-55. G. Hua and F. C. Lee, An overview of soft-switching techniques for PWM converters, EPE J., vol. 3, no. 1, Mar. 1993. V. Vorperian, Quasi-square wave converters: Topologies and analysis, IEEE Trans. Power Electron., vol. 3, no. 2, pp. 183-191, Apr. 1988. C. P. Henze, H. C. Martin, and D. W. Parsley, Zero-voltage-switching in high frequency power converters using pulse width modulation, in IEEE Applied Power Electronics Con$ Proc., 1988, pp. 3340. D. Sable, F. C. Lee, and B. H. Cho, A zero-voltage-switching bidirectional battery charger/dischargerfor the NASA EOS satellite, in IEEE Appl. Power Electron. Con$ Proc., 1992. G. Hua and F. C. Lee, A new class of ZVS-PWM converters, in High Frequency Power Conversion Con$, 1991, pp. 244-251. 0. D. Patterson and D. M. Divan, Pseudo-resonant full-bridge dcdc converter, in IEEE Power Electron. Spec. Con$ Rec., 1987, pp. 424430. J. A. Sabate, V. Vlatkovic, R. Ridely, and F. C. Lee, High-voltage, high-power, ZVS, full-bridge PWM converter employing an active switch, in IEEE Applied Power Electron. Conf Proc., 1991, pp. 158-1 63. G. Hua, C. Leu, and F. C. Lee, Novel zero-voltage-transition PWM converters, in Virginia Power Electronics Center (VPEC) Sem. Proc., 1991, pp. 81-88. G. Hua, C. Leu, Y. Jiang, and F. C. Lee, Novel zero-voltage-transition PWM converters, IEEE Trans. Power Electron., Apr. 1994; also in IEEE Power Electron. Spec. Con$ Rec., 1992, pp. 55-61. M. Jovanovic, C. Zhou, and P. Liao, Evaluation of active and passive snubber techniques for applications in power-factor-correction boost converters, in 6th Int. Conf Power Semiconductor, Applicat. (ELECTRONICA 92), Munich, Germany, 1992. V. Vlatkovic, D. Borojevic, and F. C. Lee, A new zero-voltage transition, three-phase PWM inverterhectifier circuit, in IEEE Power Electron. Spec. Con5 Rec., 1993, pp. 868-873. R. Streit and D. Tollik, High efficiency telecom rectifier using a novel soft-switched boost-based input current shaper, in 13th Int. Telecommunicat. Energy Con$ Proc., 1991, pp. 720-726.

[20] D. C. Martins, F. Seixas, J. A. Brilhante, and I. Barbi, A family of dc-to-dc converter using a new ZVS commutation cell, in IEEE Power Electron. Soec. Conf Rec.. 1993. DO. 524-530. [21] G . Hua, Sbft-switching technique; for PWM converters, Ph.D. dissertation, Virginia Polytechnic Institute and State University, Blacksburg, VA, 1994. [22] J. Bazinet and J. OConnor, Analysis and design of a zero-voltagetransition power factor correction circuit, in IEEE Appl. Power Electron. Con$ Proc., 1994. [23] I. Barbi, J. C. Bolacell, D. C. Martins, and F. B. Libano, Buck quasiresonant converter operating at constant frequency: Analysis, design, and experimentation, in IEEE Power Electron. Spec. Con$ Rec., 1989, pp. 873-880. [24] G. Hua and F. C. Lee, Novel full-bridge zero-current-switched PWM converter, in 4th European Con$ Power Electron. Applicat., 1991, vol. 2, pp. 29-34. [25] K. Wang, G. Hua, and F. C. Lee, Switching losses of IGBTs in softswitching PWM converters, in IEEE Power Electron. Spec. Conf Rec., 1994. [26] G. Hua, X. Yang, Y. Jiang, and F. C. Lee, Novel zero-current-transition PWM converters, in IEEE Power Electron. Spec. Con$ Rec., 1993, pp. 538-544. [27] W. McMurray and D. P. Shattuck, Silicon-controlled rectifier inverter with improved commutation,AIEE Trans., vol. 80, pt. 1, pp. 531-542, 1961.

Guichao Hua received the B.S. and M.S. degrees in electrical engineering from Zhejiang University in China in 1985 and 1988, respectively, and the Ph.D. degree from Virginia Tech in 1994. From 1988-1989, he was employed as a power supply design engineer at Watt Power Supply Corporation in China. He joined the Virginia Power Electronics Center (VPEC) in 1989 and became a research associate in 1991 and a research scientist in 1994. He is currently the Vice President of Virginia Power Technology Corp. in Blacksburg, VA. He has published more than 45 papers and holds four U.S. patents. His research interests include high-frequency power conversion, new converterhnverter topologies, power-factor correction circuits, distributed power systems, piezoelectric transformers, and UPS systems.

Fred C. Lee @72-M74M77-SM83-F90), for a photograph and hiography, see p. 71 of the February 1995 issue of this TRANSACTIONS.