A new soft recovery PWM quasi-resonant ... - Semantic Scholar
456
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 49, NO. 2, APRIL 2002
A New Soft Recovery PWM Quasi-Resonant Converter With a Folding Snubber Network Jin-Kuk Chung, Student Member, IEEE, and Gyu-Hyeong Cho, Member, IEEE
Abstract—A new soft recovery (SR) quasi-resonant converter (QRC) having a multiple-order folding snubber network(MFSN) is introduced. It is obtained by combining a normal QRC with a folding snubber network of which the surrounding components are composed of passive devices only (diodes and capacitors). The reverse recovery loss of the main rectifier diode is eliminated by this method utilizing multiple resonance with a multiple-order folding snubber network. By realizing soft switching conditions, the proposed converter has pulsewidth modulation capability with high efficiency and is suitable for high-voltage and high-power dc to dc converter applications. Index Terms—Folding snubber network, quasi-resonant converter, soft switching.
I. INTRODUCTION
I
N DC-TO-DC switching converters, the active switch and power diode operate alternately. In other words, when the switch is ON, the diode is OFF and vice versa. The problem happens at the switching transient. Under hard switching conditions, a large recovery loss of the diode reduces the maximum safety operating region as well as efficiency. As the voltage and current ratings increase, the problem becomes more serious and system reliability cannot be guaranteed in most cases. To solve such a problem, an additional inductor is usually inserted in series with the loop formed by the switch and the diode, resulting in a soft switching converter [1]. Such soft switching converters can be classified into two groups. One is an active snubber type and the other is a passive snubber type. Switching converters of the passive snubber type are simpler and easier to realize than those of the active snubber type. In such a passive snubber type converter, the key ideas are, however, focused on how to suppress voltage stress and parasitic oscillation caused by the additional inductor and parasitic capacitance. All of the passive snubbers presented so far are not free from such an oscillation [2]. As a result, the active snubber appeared despite its additional control complexity [3]. In this paper, we suggest a new solution by proposing a new quasi-resonant converter (QRC) having an energy recovery network (ERN) and a multiple-order folding snubber network (MFSN). Manuscript received December 28, 2000; revised December 10, 2001. Abstract published on the Internet January 9, 2002. J.-K. Chung is with the Department of Electronics, Information and Communication, Daelim Collage, Anyang 430-514, Korea (e-mail: jkchung@ daelim.ac.kr). G.-H. Cho is with the Department of Electrical Engineering, Korean Advanced Institute of Science and, Technology, Taejon 305-701, Korea (e-mail: [email protected]). Publisher Item Identifier S 0278-0046(02)02876-9.
Fig. 1. Folding snubber and normal resonant circuit and its waveform. (a) Folding snubber and normal resonant circuit. (b) Resonant waveforms of current and voltage.
II. NEW SOFT RECOVERY BOOST CONVERTER A. Basic Folding Snubber Circuit Fig. 1(a) compares the normal series resonant circuit with a basic folding resonant circuit. If we assume the initial inductor is maximum ( ) and initial voltage is zero current , the standard waveforms of current with and induced voltage across the resonant inductor during one period are shown in Fig. 1(b), respectively. As is well known, and voltage in normal LC resthe waveforms of current onant circuits are sinusoidial waves, shown by dashed lines in in the resonant circuit inFig. 1(b). However, the voltage at the cluding the folding network jumps up to changes its polarity. There are time when the inductor current two characteristic impedances and angular resonant frequencies in one resonant period. However, the key operation in the folding resonant circuit is that the maximum negative voltage across the resonant inductor drops to half when the direction of the resonant current changes. As the negative resonant voltage at the inductor is folded to half, the network of three diodes and two capacitors is named the folding snubber network. We apply this characteristic to the soft switching operation in this proposed converter. B. New Multiple-Order Folding Snubber Network and Others Fig. 2 shows a proposed boost type soft recovery pulsewidth modulation quasi resonance converter (SR PWM QRC) that includes a multiple-order folding snubber network (MFSN) together with the energy recovery network (ERN) and zero-
0278-0046/02$17.00 © 2002 IEEE
CHUNG AND CHO: A NEW SOFT RECOVERY PWM QUASI-RESONANT CONVERTER
457
Fig. 2. Proposed SR PWM boost QR converter.
voltage-switching (ZVS) snubber. The ERN is composed of , , and and the ZVS snubber is composed of and . The MFSN is composed of , , , , , and a higher order FSN across . The function of and is to provide a zero-voltage turn-off condition to the switch . The ERN is inserted between the ZVS snubber for MOSFET. The resonant current in the auxiliary inductor charges through and during the ON interval of and restores energy to and returns to load through during the OFF-interval. As shown in Fig. 2, the MFSN is composed of the cascade strings of the basic folding snubber network. The MFSN provides several merits versus the proposed converter. First, similar to the conventional passive converter noted in [2], it also provides soft recovery condition to the main rectifier diode . in Second, FSN divides the voltage across the main diode half and suppresses the parasitic oscillation during the turn-off interval of . The stable operation without parasitic oscillation consequently enhances the reliability of the circuit and in) in creases the available output power. The center diode ( MFSN plays a very important role to divide the voltage induced through the inductor ( ) into the voltages of the two capaci). At the instant of division, the direction of the tors ( ) is changed from forward to revoltage across the diode ( verse as described above. This divided voltage suppresses parasitic oscillation, which can be observed in the conventional pas) of sive snubber type converter. Since this center diode ( the first-order FSN is, however, also not free from the reverse recovery problem during the procedure of dividing the voltage across the main diode ( ), it can create other problems as well. ) FSN can be reThis reverse recovery current of the diode ( moved by connecting an additional second-order FSN across the ) of the first order of the FSN. The third-order center diode ( ) of the FSN is similarly needed across the center diode ( second-order FSN and so on. As the number of orders of the folding snubber network increases, the magnitude of the resonance waveform, however, decays rapidly at a higher order FSN and the reverse recovery loss of the center diode of the FSN becomes negligible at the end. The MFSN induces the multiple numbers of half resonance across the diode , whose detail configuration is shown in Fig. 5(b). The numbers of half resonance correspond to the number of orders of FSN.
(a)
(b)
(c)
(d)
(e)
Fig. 3. Equivalent circuit in each mode of Fig. 2. (a) [t (c) [t t ]. (d) [t t ]. (e) [t t ].
t
]
. (b) [t
t
]
.
C. Analysis of Operation To simplify the analysis, we assumed several factors during one period of switching operation. The current of the input inductor is considered to be constant in one cycle operation and the diode is assumed to be ideal. The stray capacitance and switching time of the device are assumed to be zero. Then, one period of the switching operation can be divided into five modes described as follows. The mode diagrams corresponding to each time duration and the theoretical key waveforms (the current of resonant inductor , the voltage across the main diode of main diode ) are shown in Figs. 3 circuit , and current and 4, respectively. 1) Turn-on Modes: a) Mode 1: Linearly increased stage of the inductor current ( ) in the interval to . The equivalent circuit in this time duration is shown in Fig. 3(a). When the switch is closed at , the inductor current ramps up linearly to . The equations of diode and auxiliary inductor currents in this stage are respectively given by (1) (2)
458
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 49, NO. 2, APRIL 2002
(a)
Fig. 4. Theoretical waveforms of the proposed converter.
Initial conditions are assumed to be and . As the current in the resonant increases linearly, the switch is turned on at the inductor of the main diode zero current state at . As the current decreases linearly to zero at , the main diode turns off softly at . b) Mode 2: Resonant stage of the voltage across the capacitors in MFSN in the interval to . of inductor in the loop In this interval, the current , and MFSN induces resonances of a half cycle of the , that have different angular resonant frequencies and character). The first half cycle of istics in each duration ( resonance is induced by the first-order FSN. The second half cycle of resonance is induced by the second-order FSN. In other words, the numbers of half cycle resonance correspond to the numbers of order of the FSN. The time duration of resonance of the first—, second- and third-order FSN converter corresponds , to , and to of Fig. 4, to the duration of to respectively. Fig. 5 shows the detailed resonance waveforms of corresponding to the order of the FSN converter. In the inof the converter of the first-order FSN, terval to whose equivalent circuit is shown in Fig. 5(b), the inductor curforms resonance current through the paths of Fig. 6(a). rent across approaches to 2 and The overcharged voltage approaches . However, as soon as the resonance current becomes equal to at , the resothe resonant current that is alnance stops due to the strong driving input current , most constant during one period. As the resonance stops at becomes zero and the voltage across the voltage across from 2 at . If the the main diode jumps up to for concapacitance values are set to and the voltage across venience, the resonance current the main diode are given by
(b)
D
Fig. 5. The third-order FSN and voltage v across the main diode in proposed circuit including for each multiple order FSN. (a) The muliple-order FSN. (b) The resonance waveform of corresponding to the multiple-order FSN.
V
where
The FSN of the subsequent order is necessary to keep soft turn-off conditions for the center diode of the present order of FSN. Fig. 6 shows detailed resonance current paths and Fig. 5 for each order. shows the waveforms of resonant voltage The directions of the arrow in Fig. 6 are the direction of the resonant currents during each time interval at the multiple-order FSN converter. As the number of order of the FSN increases, the resonance amplitude and period of FSN decrease rapidly and as shown in Fig. 5(b). The characteristic impedance of (3) and (4) in each time angular resonance frequency duration are given by
where
durng
durng durng
durng (3) (4)
durng (5)
CHUNG AND CHO: A NEW SOFT RECOVERY PWM QUASI-RESONANT CONVERTER
459
(a)
(b)
(c)
(d)
(e) Fig. 6. Equivalent circuit of the third-order FSN of Fig. 5(a) in the interval of t to t (e) T interval.
1
where , , and are the capacitors in the first—, second– and third-order FSNs, respectively. c) Mode 3: Constant voltage and current stage in the interval to . The equivalent circuit in this time duration is shown in equals at , Fig. 3(c). When the resonance current the resonance stops and the induced voltage across the becomes zero. Therefore, the voltage of node A becomes zero across the main diode changes abruptly and the voltage to . The current directly flows through the from 2 supplies the switch to the ground and the energy stored in output load. 2) Turn-off Modes: d) Mode 4: Discharging of the folding snubber capacitor in the interval to . The equivalent circuit in this time duration is shown in is ( ). When the main switch Fig. 3(d), where is turned off at , of charges whose voltage from zero and decays increases to the output voltage to . becomes a resonantly to zero in the interval in this duration. On the other hand, the snubber of the switch discharges both of the capacitors and constant current in parallel. After the capacitors and are discharged turns on at . The equations of and to zero, the diode in this stage are
e) Mode 5: Feeding .
1T
interval. (b)
1T
interval. (c)
1T
interval. (d)
1T
interval.
The equivalent circuit in this time duration is shown in Fig. 3(e). When the voltage across the folding snubber capacitor becomes zero at , forward voltage is applied to the main flows to the load diode to turn it on. The inductor current through the main diode , completing one cycle of operation. III. EXPERIMENTAL RESULTS A. Experimental Results Applied to UPS System
(6)
The proposed circuit is applied to an auxiliary power source of an uninterrupted power supply (UPS) system having 2-kW output power. Fig. 7 shows the experimental waveforms of the circuit of Fig. 2 with 100-kHz switching frequency. The main parameters to be designed are only the values of the capacitors of the FSN and , which are determined by the minimum load current, maximum duty factor, and switching frequency. The experimental waveforms of the first-order FSN shown in Fig. 7(a) agree well with the theoretical waveforms of Fig. 4. While the across the main diode is observed to be approxvoltage imately constant in the time interval to , as shown by the encircled line in Fig. 7(a), an additional resonance waveform having five different values of characteristic impedance shown in (5) can be observed in the dotted area of the third-order FSN in Fig. 7(b). As discussed in the previous section, the magnitude of the resonance waveform decays rapidly for higher order FSN. The experimental waveforms shown in Fig. 7(b) agree well with the theoretical ones of Fig. 5(b).
(7)
B. Efficiency Measurement
where
to
. (a)
directly to the load in the interval
The efficiencies of power for both the normal snubberless boost converter, ZVT PWM converter [3], and the proposed converter of the multiple-order FSN have been measured. The , and comparison results are shown in Figs. 8 and 9, where , are the efficiency, output power, and input voltage, respectively. The difference of efficiency among the multiple-order FSN converter is negligible. But, as the reverse recovery current
460
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 49, NO. 2, APRIL 2002
Fig. 7. Experimental waveforms of the proposed circuit of Fig. 2. (a) Waveforms of the first-order FSN. (b) Waveforms of the third-order FSN.
Fig. 8. Comparison of efficiency versus output power P PWM [3], and proposed converters.
for normal, ZVT
of the center diode of the last-order folding snubber network limited the available maximum output power for each order of FNS of the proposed converter. Therefore, available maximum output power of the proposed converter for each multiple orders become different. Experimental results show that the safety operation region of the proposed converter is about 700 W for the first-order FSN, 1500 W for the second-order FSN and 3 kW for the third-order FSN, respectively. As the output power increases up to 250 W in Fig. 8, the reverse recovery current in the main caused the destruction of the MOSFET switch in diode the conventional converter. On the other hand, in the proposed FSN converters, the level of efficiency is below 0.4%, lower than that of the ZVT PWN converter [3]. However, it is not hard to obtain a good efficiency up to several kilowatts with negligible power losses of the center diodes of the FSN. Fig. 9 shows the comparison of efficiency for the above three converters under fixed output power and voltage, exhibiting an outcome similar to that of Fig. 8.
Fig. 9. Comparison of efficiency versus input voltage V for normal, ZVT PWM, and proposed converters in fixed output power and voltage.
C. Parasitic Oscillations An experimental, side-by-side comparison of a converter with a conventional snubber [3] and a converter with the proposed new snubber, the first-order FSN, was undertaken. Both converters are of similar configuration and function and the intent was to view and compare key waveforms as shown in Fig. 10. As can be seen, parasitic oscillation is observed in the time interval to , in the converter with the conventional snubber, to , of Fig. 10(b), Fig. 10(a), while in the same interval the converter with the new, first-order FSN, is free from parasitic oscillations. The oscillations in Fig. 10(a) are the result of a series resonance loop of biased junction capacitor of auxiliary diode and resonant inductor , during turn-on period - . As observed, in Fig. 10(b), the parasitic oscillation has been effectively damped by the first-order FSN, demonstrating the effectiveness of this snubber concept.
CHUNG AND CHO: A NEW SOFT RECOVERY PWM QUASI-RESONANT CONVERTER
461
Fig. 10. Experimental key waveforms of two converters. (a) Waveforms of the boost converter with passive snubber in [2]. (b) Waveforms of proposed the first-order FSN converter.
IV. CONVENIENT
ECONOMIC FOLDING SNUBBER NETWORK
AND
As the order of FSNs connected in series increases, the apparent complexity of this topology may dissuade the circuit designer from implementing this design concept. However, with the first– and higher order FSNs, voltage and stresses are also reduced, thereby drastically reducing size (and cost) of their components. The complexity of the active snubber is also avoided. For instance, with implementation of the second– and thirdorder FSNs, PWM MOSFET converters, with power output of the several kilowatts, are practical. As the output power is increased, the number of orders of the FSN is required to increase also. As the maximum output power achieved with available MOSFET devices in the semiconductor market is about 3 kW, the proposed converter of the third-order FSN actually becomes the maximum order to realize. With higher order FSNs, component and sizes are drastically reduced and modular configurations should be considered. V. CONCLUSION With the introduction of the new multiple-order FSN, combined with the energy recovery network (ERC), a boost converter with a new, multi-stage passive snubber is proposed. With this new snubber configuration, fully soft switching conditions are obtained. The MFSN and ERN reduce ringing voltages across capacitors and suppress the parasitic oscillations that occur in conventional converters using conventional snubbers. Experimental results are shown to prove the theoretical concepts. Enhanced reliability, coupled with switching operations that are free from parasitic oscillation, enables power output to increase up to several kilowatts. Additionally, this “ring-free” switching provides the opportunity to reduce the size (and cost) of EMI filters. This new snubber and concept is applicable to practical power supply systems, providing low cost and high efficiency.
REFERENCES [1] K. H. Liu and F. C. Y. Lee, “Quasiresonant converters—Topologies and characteristics,” IEEE. Trans. Power Electron., vol. PE-2, Jan. 1987. [2] M. 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 Semiconductors and Their Applications, Munich, Germany, 1992. [3] G. Hua, C. S. Leu, and F. C. Lee, “Zero-voltage-transition PWM converters,” in Proc. IEEE PESC’92, vol. 1, 1992, pp. 55–61. [4] C. J. Tseng and C. L. Chen, “Passive lossless snubbers for DC/DC converters,” in Proc. IEEE APEC’98, vol. 2, 1998, pp. 1049–1054. [5] M. H. Kheraluwala and S. A. El-Hamamsy, “Modified valley fill high power factor electronic ballast for compact florescent lamps,” IEEE Trans. Ind. Applicat., vol. 29, pp. 670–674, May/June 1993.
Jin Kuk Chung (M’77) received the B.S. degree from Hanyang University, Seoul, Korea, and the M.S. and Ph.D. degrees from Korea Advanced Institute of Science and Technology, Taejon, Korea, in 1972, 1977, and 2001, respectively. Since 1993, he has been with Daelim College, Anyang, Korea, where he is a Professor in the Department of Electronic, Information and Communication Engineering. He is interested in power electronics and analog integrated circuit design in industrial applications. He was with the R&D Center , Deawoo Electronic Company, Ltd., where he was engaged in ASIC design in home appliances and the defense industrial field from 1977 to 1990.
Gyu-Hyeong Cho (M’81) received the Ph.D. degree in electrical engineering from Korea Advanced Institute of Science and Technology (KAIST), Taejon, Korea, in 1981. He was with the Westinghouse R&D Center until 1983. Since 1984, he has been with KAIST, where he was appointed a Professor in 1991. During 1989, he was a Visiting Professor at the University of Wisconsin, Madison. His interests are in the areas of CMOS/BiCMOS analog-integrated circuits including A/D converters, smart-power ICs and RF ICs for wireless communications, at panel displays, etc. Dr. Cho is a member of the Institute of Electrical/Electronics Engineering of Korea.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 49, NO. 2, APRIL 2002
A New Soft Recovery PWM Quasi-Resonant Converter With a Folding Snubber Network Jin-Kuk Chung, Student Member, IEEE, and Gyu-Hyeong Cho, Member, IEEE
Abstract—A new soft recovery (SR) quasi-resonant converter (QRC) having a multiple-order folding snubber network(MFSN) is introduced. It is obtained by combining a normal QRC with a folding snubber network of which the surrounding components are composed of passive devices only (diodes and capacitors). The reverse recovery loss of the main rectifier diode is eliminated by this method utilizing multiple resonance with a multiple-order folding snubber network. By realizing soft switching conditions, the proposed converter has pulsewidth modulation capability with high efficiency and is suitable for high-voltage and high-power dc to dc converter applications. Index Terms—Folding snubber network, quasi-resonant converter, soft switching.
I. INTRODUCTION
I
N DC-TO-DC switching converters, the active switch and power diode operate alternately. In other words, when the switch is ON, the diode is OFF and vice versa. The problem happens at the switching transient. Under hard switching conditions, a large recovery loss of the diode reduces the maximum safety operating region as well as efficiency. As the voltage and current ratings increase, the problem becomes more serious and system reliability cannot be guaranteed in most cases. To solve such a problem, an additional inductor is usually inserted in series with the loop formed by the switch and the diode, resulting in a soft switching converter [1]. Such soft switching converters can be classified into two groups. One is an active snubber type and the other is a passive snubber type. Switching converters of the passive snubber type are simpler and easier to realize than those of the active snubber type. In such a passive snubber type converter, the key ideas are, however, focused on how to suppress voltage stress and parasitic oscillation caused by the additional inductor and parasitic capacitance. All of the passive snubbers presented so far are not free from such an oscillation [2]. As a result, the active snubber appeared despite its additional control complexity [3]. In this paper, we suggest a new solution by proposing a new quasi-resonant converter (QRC) having an energy recovery network (ERN) and a multiple-order folding snubber network (MFSN). Manuscript received December 28, 2000; revised December 10, 2001. Abstract published on the Internet January 9, 2002. J.-K. Chung is with the Department of Electronics, Information and Communication, Daelim Collage, Anyang 430-514, Korea (e-mail: jkchung@ daelim.ac.kr). G.-H. Cho is with the Department of Electrical Engineering, Korean Advanced Institute of Science and, Technology, Taejon 305-701, Korea (e-mail: [email protected]). Publisher Item Identifier S 0278-0046(02)02876-9.
Fig. 1. Folding snubber and normal resonant circuit and its waveform. (a) Folding snubber and normal resonant circuit. (b) Resonant waveforms of current and voltage.
II. NEW SOFT RECOVERY BOOST CONVERTER A. Basic Folding Snubber Circuit Fig. 1(a) compares the normal series resonant circuit with a basic folding resonant circuit. If we assume the initial inductor is maximum ( ) and initial voltage is zero current , the standard waveforms of current with and induced voltage across the resonant inductor during one period are shown in Fig. 1(b), respectively. As is well known, and voltage in normal LC resthe waveforms of current onant circuits are sinusoidial waves, shown by dashed lines in in the resonant circuit inFig. 1(b). However, the voltage at the cluding the folding network jumps up to changes its polarity. There are time when the inductor current two characteristic impedances and angular resonant frequencies in one resonant period. However, the key operation in the folding resonant circuit is that the maximum negative voltage across the resonant inductor drops to half when the direction of the resonant current changes. As the negative resonant voltage at the inductor is folded to half, the network of three diodes and two capacitors is named the folding snubber network. We apply this characteristic to the soft switching operation in this proposed converter. B. New Multiple-Order Folding Snubber Network and Others Fig. 2 shows a proposed boost type soft recovery pulsewidth modulation quasi resonance converter (SR PWM QRC) that includes a multiple-order folding snubber network (MFSN) together with the energy recovery network (ERN) and zero-
0278-0046/02$17.00 © 2002 IEEE
CHUNG AND CHO: A NEW SOFT RECOVERY PWM QUASI-RESONANT CONVERTER
457
Fig. 2. Proposed SR PWM boost QR converter.
voltage-switching (ZVS) snubber. The ERN is composed of , , and and the ZVS snubber is composed of and . The MFSN is composed of , , , , , and a higher order FSN across . The function of and is to provide a zero-voltage turn-off condition to the switch . The ERN is inserted between the ZVS snubber for MOSFET. The resonant current in the auxiliary inductor charges through and during the ON interval of and restores energy to and returns to load through during the OFF-interval. As shown in Fig. 2, the MFSN is composed of the cascade strings of the basic folding snubber network. The MFSN provides several merits versus the proposed converter. First, similar to the conventional passive converter noted in [2], it also provides soft recovery condition to the main rectifier diode . in Second, FSN divides the voltage across the main diode half and suppresses the parasitic oscillation during the turn-off interval of . The stable operation without parasitic oscillation consequently enhances the reliability of the circuit and in) in creases the available output power. The center diode ( MFSN plays a very important role to divide the voltage induced through the inductor ( ) into the voltages of the two capaci). At the instant of division, the direction of the tors ( ) is changed from forward to revoltage across the diode ( verse as described above. This divided voltage suppresses parasitic oscillation, which can be observed in the conventional pas) of sive snubber type converter. Since this center diode ( the first-order FSN is, however, also not free from the reverse recovery problem during the procedure of dividing the voltage across the main diode ( ), it can create other problems as well. ) FSN can be reThis reverse recovery current of the diode ( moved by connecting an additional second-order FSN across the ) of the first order of the FSN. The third-order center diode ( ) of the FSN is similarly needed across the center diode ( second-order FSN and so on. As the number of orders of the folding snubber network increases, the magnitude of the resonance waveform, however, decays rapidly at a higher order FSN and the reverse recovery loss of the center diode of the FSN becomes negligible at the end. The MFSN induces the multiple numbers of half resonance across the diode , whose detail configuration is shown in Fig. 5(b). The numbers of half resonance correspond to the number of orders of FSN.
(a)
(b)
(c)
(d)
(e)
Fig. 3. Equivalent circuit in each mode of Fig. 2. (a) [t (c) [t t ]. (d) [t t ]. (e) [t t ].
t
]
. (b) [t
t
]
.
C. Analysis of Operation To simplify the analysis, we assumed several factors during one period of switching operation. The current of the input inductor is considered to be constant in one cycle operation and the diode is assumed to be ideal. The stray capacitance and switching time of the device are assumed to be zero. Then, one period of the switching operation can be divided into five modes described as follows. The mode diagrams corresponding to each time duration and the theoretical key waveforms (the current of resonant inductor , the voltage across the main diode of main diode ) are shown in Figs. 3 circuit , and current and 4, respectively. 1) Turn-on Modes: a) Mode 1: Linearly increased stage of the inductor current ( ) in the interval to . The equivalent circuit in this time duration is shown in Fig. 3(a). When the switch is closed at , the inductor current ramps up linearly to . The equations of diode and auxiliary inductor currents in this stage are respectively given by (1) (2)
458
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 49, NO. 2, APRIL 2002
(a)
Fig. 4. Theoretical waveforms of the proposed converter.
Initial conditions are assumed to be and . As the current in the resonant increases linearly, the switch is turned on at the inductor of the main diode zero current state at . As the current decreases linearly to zero at , the main diode turns off softly at . b) Mode 2: Resonant stage of the voltage across the capacitors in MFSN in the interval to . of inductor in the loop In this interval, the current , and MFSN induces resonances of a half cycle of the , that have different angular resonant frequencies and character). The first half cycle of istics in each duration ( resonance is induced by the first-order FSN. The second half cycle of resonance is induced by the second-order FSN. In other words, the numbers of half cycle resonance correspond to the numbers of order of the FSN. The time duration of resonance of the first—, second- and third-order FSN converter corresponds , to , and to of Fig. 4, to the duration of to respectively. Fig. 5 shows the detailed resonance waveforms of corresponding to the order of the FSN converter. In the inof the converter of the first-order FSN, terval to whose equivalent circuit is shown in Fig. 5(b), the inductor curforms resonance current through the paths of Fig. 6(a). rent across approaches to 2 and The overcharged voltage approaches . However, as soon as the resonance current becomes equal to at , the resothe resonant current that is alnance stops due to the strong driving input current , most constant during one period. As the resonance stops at becomes zero and the voltage across the voltage across from 2 at . If the the main diode jumps up to for concapacitance values are set to and the voltage across venience, the resonance current the main diode are given by
(b)
D
Fig. 5. The third-order FSN and voltage v across the main diode in proposed circuit including for each multiple order FSN. (a) The muliple-order FSN. (b) The resonance waveform of corresponding to the multiple-order FSN.
V
where
The FSN of the subsequent order is necessary to keep soft turn-off conditions for the center diode of the present order of FSN. Fig. 6 shows detailed resonance current paths and Fig. 5 for each order. shows the waveforms of resonant voltage The directions of the arrow in Fig. 6 are the direction of the resonant currents during each time interval at the multiple-order FSN converter. As the number of order of the FSN increases, the resonance amplitude and period of FSN decrease rapidly and as shown in Fig. 5(b). The characteristic impedance of (3) and (4) in each time angular resonance frequency duration are given by
where
durng
durng durng
durng (3) (4)
durng (5)
CHUNG AND CHO: A NEW SOFT RECOVERY PWM QUASI-RESONANT CONVERTER
459
(a)
(b)
(c)
(d)
(e) Fig. 6. Equivalent circuit of the third-order FSN of Fig. 5(a) in the interval of t to t (e) T interval.
1
where , , and are the capacitors in the first—, second– and third-order FSNs, respectively. c) Mode 3: Constant voltage and current stage in the interval to . The equivalent circuit in this time duration is shown in equals at , Fig. 3(c). When the resonance current the resonance stops and the induced voltage across the becomes zero. Therefore, the voltage of node A becomes zero across the main diode changes abruptly and the voltage to . The current directly flows through the from 2 supplies the switch to the ground and the energy stored in output load. 2) Turn-off Modes: d) Mode 4: Discharging of the folding snubber capacitor in the interval to . The equivalent circuit in this time duration is shown in is ( ). When the main switch Fig. 3(d), where is turned off at , of charges whose voltage from zero and decays increases to the output voltage to . becomes a resonantly to zero in the interval in this duration. On the other hand, the snubber of the switch discharges both of the capacitors and constant current in parallel. After the capacitors and are discharged turns on at . The equations of and to zero, the diode in this stage are
e) Mode 5: Feeding .
1T
interval. (b)
1T
interval. (c)
1T
interval. (d)
1T
interval.
The equivalent circuit in this time duration is shown in Fig. 3(e). When the voltage across the folding snubber capacitor becomes zero at , forward voltage is applied to the main flows to the load diode to turn it on. The inductor current through the main diode , completing one cycle of operation. III. EXPERIMENTAL RESULTS A. Experimental Results Applied to UPS System
(6)
The proposed circuit is applied to an auxiliary power source of an uninterrupted power supply (UPS) system having 2-kW output power. Fig. 7 shows the experimental waveforms of the circuit of Fig. 2 with 100-kHz switching frequency. The main parameters to be designed are only the values of the capacitors of the FSN and , which are determined by the minimum load current, maximum duty factor, and switching frequency. The experimental waveforms of the first-order FSN shown in Fig. 7(a) agree well with the theoretical waveforms of Fig. 4. While the across the main diode is observed to be approxvoltage imately constant in the time interval to , as shown by the encircled line in Fig. 7(a), an additional resonance waveform having five different values of characteristic impedance shown in (5) can be observed in the dotted area of the third-order FSN in Fig. 7(b). As discussed in the previous section, the magnitude of the resonance waveform decays rapidly for higher order FSN. The experimental waveforms shown in Fig. 7(b) agree well with the theoretical ones of Fig. 5(b).
(7)
B. Efficiency Measurement
where
to
. (a)
directly to the load in the interval
The efficiencies of power for both the normal snubberless boost converter, ZVT PWM converter [3], and the proposed converter of the multiple-order FSN have been measured. The , and comparison results are shown in Figs. 8 and 9, where , are the efficiency, output power, and input voltage, respectively. The difference of efficiency among the multiple-order FSN converter is negligible. But, as the reverse recovery current
460
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 49, NO. 2, APRIL 2002
Fig. 7. Experimental waveforms of the proposed circuit of Fig. 2. (a) Waveforms of the first-order FSN. (b) Waveforms of the third-order FSN.
Fig. 8. Comparison of efficiency versus output power P PWM [3], and proposed converters.
for normal, ZVT
of the center diode of the last-order folding snubber network limited the available maximum output power for each order of FNS of the proposed converter. Therefore, available maximum output power of the proposed converter for each multiple orders become different. Experimental results show that the safety operation region of the proposed converter is about 700 W for the first-order FSN, 1500 W for the second-order FSN and 3 kW for the third-order FSN, respectively. As the output power increases up to 250 W in Fig. 8, the reverse recovery current in the main caused the destruction of the MOSFET switch in diode the conventional converter. On the other hand, in the proposed FSN converters, the level of efficiency is below 0.4%, lower than that of the ZVT PWN converter [3]. However, it is not hard to obtain a good efficiency up to several kilowatts with negligible power losses of the center diodes of the FSN. Fig. 9 shows the comparison of efficiency for the above three converters under fixed output power and voltage, exhibiting an outcome similar to that of Fig. 8.
Fig. 9. Comparison of efficiency versus input voltage V for normal, ZVT PWM, and proposed converters in fixed output power and voltage.
C. Parasitic Oscillations An experimental, side-by-side comparison of a converter with a conventional snubber [3] and a converter with the proposed new snubber, the first-order FSN, was undertaken. Both converters are of similar configuration and function and the intent was to view and compare key waveforms as shown in Fig. 10. As can be seen, parasitic oscillation is observed in the time interval to , in the converter with the conventional snubber, to , of Fig. 10(b), Fig. 10(a), while in the same interval the converter with the new, first-order FSN, is free from parasitic oscillations. The oscillations in Fig. 10(a) are the result of a series resonance loop of biased junction capacitor of auxiliary diode and resonant inductor , during turn-on period - . As observed, in Fig. 10(b), the parasitic oscillation has been effectively damped by the first-order FSN, demonstrating the effectiveness of this snubber concept.
CHUNG AND CHO: A NEW SOFT RECOVERY PWM QUASI-RESONANT CONVERTER
461
Fig. 10. Experimental key waveforms of two converters. (a) Waveforms of the boost converter with passive snubber in [2]. (b) Waveforms of proposed the first-order FSN converter.
IV. CONVENIENT
ECONOMIC FOLDING SNUBBER NETWORK
AND
As the order of FSNs connected in series increases, the apparent complexity of this topology may dissuade the circuit designer from implementing this design concept. However, with the first– and higher order FSNs, voltage and stresses are also reduced, thereby drastically reducing size (and cost) of their components. The complexity of the active snubber is also avoided. For instance, with implementation of the second– and thirdorder FSNs, PWM MOSFET converters, with power output of the several kilowatts, are practical. As the output power is increased, the number of orders of the FSN is required to increase also. As the maximum output power achieved with available MOSFET devices in the semiconductor market is about 3 kW, the proposed converter of the third-order FSN actually becomes the maximum order to realize. With higher order FSNs, component and sizes are drastically reduced and modular configurations should be considered. V. CONCLUSION With the introduction of the new multiple-order FSN, combined with the energy recovery network (ERC), a boost converter with a new, multi-stage passive snubber is proposed. With this new snubber configuration, fully soft switching conditions are obtained. The MFSN and ERN reduce ringing voltages across capacitors and suppress the parasitic oscillations that occur in conventional converters using conventional snubbers. Experimental results are shown to prove the theoretical concepts. Enhanced reliability, coupled with switching operations that are free from parasitic oscillation, enables power output to increase up to several kilowatts. Additionally, this “ring-free” switching provides the opportunity to reduce the size (and cost) of EMI filters. This new snubber and concept is applicable to practical power supply systems, providing low cost and high efficiency.
REFERENCES [1] K. H. Liu and F. C. Y. Lee, “Quasiresonant converters—Topologies and characteristics,” IEEE. Trans. Power Electron., vol. PE-2, Jan. 1987. [2] M. 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 Semiconductors and Their Applications, Munich, Germany, 1992. [3] G. Hua, C. S. Leu, and F. C. Lee, “Zero-voltage-transition PWM converters,” in Proc. IEEE PESC’92, vol. 1, 1992, pp. 55–61. [4] C. J. Tseng and C. L. Chen, “Passive lossless snubbers for DC/DC converters,” in Proc. IEEE APEC’98, vol. 2, 1998, pp. 1049–1054. [5] M. H. Kheraluwala and S. A. El-Hamamsy, “Modified valley fill high power factor electronic ballast for compact florescent lamps,” IEEE Trans. Ind. Applicat., vol. 29, pp. 670–674, May/June 1993.
Jin Kuk Chung (M’77) received the B.S. degree from Hanyang University, Seoul, Korea, and the M.S. and Ph.D. degrees from Korea Advanced Institute of Science and Technology, Taejon, Korea, in 1972, 1977, and 2001, respectively. Since 1993, he has been with Daelim College, Anyang, Korea, where he is a Professor in the Department of Electronic, Information and Communication Engineering. He is interested in power electronics and analog integrated circuit design in industrial applications. He was with the R&D Center , Deawoo Electronic Company, Ltd., where he was engaged in ASIC design in home appliances and the defense industrial field from 1977 to 1990.
Gyu-Hyeong Cho (M’81) received the Ph.D. degree in electrical engineering from Korea Advanced Institute of Science and Technology (KAIST), Taejon, Korea, in 1981. He was with the Westinghouse R&D Center until 1983. Since 1984, he has been with KAIST, where he was appointed a Professor in 1991. During 1989, he was a Visiting Professor at the University of Wisconsin, Madison. His interests are in the areas of CMOS/BiCMOS analog-integrated circuits including A/D converters, smart-power ICs and RF ICs for wireless communications, at panel displays, etc. Dr. Cho is a member of the Institute of Electrical/Electronics Engineering of Korea.
