IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 47, NO.

4, AUGUST 2000 787

Steady-State Analysis of an Interleaved Boost

Converter with Coupled Inductors
Po-Wa Lee, Member, IEEE, Yim-Shu Lee, Senior Member, IEEE, David K. W. Cheng,, Member, IEEE, and
Xiu-Cheng Liu

Abstract—Boost converters are widely used as power-factor-

corrected preregulators. In high-power applications, interleaved
operation of two or more boost converters has been proposed to
increase the output power and to reduce the output ripple. A major
design criterion then is to ensure equal current sharing among
the parallel converters. In this paper, a converter consisting of
two interleaved and intercoupled boost converter cells is proposed
and investigated. The boost converter cells have very good current
sharing characteristics even in the presence of relatively large
duty cycle mismatch. In addition, it can be designed to have small
input current ripple and zero boost-rectifier reverse-recovery loss.
The operating principle, steady-state analysis, and comparison
with the conventional boost converter are presented. Simulation
and experimental results are also given. Fig. 1. Converter consisting of two interleaved and intercoupled boost
converter cells.
Index Terms—Coupled inductors, current sharing, interleaved
boost converters.
loss and lower transistor switching-on loss [1], [2]. In order
to reduce the input current ripple, modified boost converter
I. INTRODUCTION
topologies have been proposed in [3]–[5].

W ITH the tightening requirements of power quality,

offline power supplies are required to operate at high
power factor and to draw low harmonic currents from the ac
In high-power applications (greater than 1.5 kW), boost con-
verters are often paralleled in an interleaved manner to increase
the output current and reduce the input current ripple. How-
mains. The conventional method of reducing input current ever, current sharing among the parallel paths is a major de-
harmonics using an LC input filter is no longer practically sign problem. It can be shown that, when two similar but inde-
sufficient to meet the requirements in many high-power ap- pendently controlled boost converters are connected in parallel
plications. Active power-factor correction using buck, boost, (with the same input and output voltages), the converter with a
buck–boost, Cuk and Sepic converters has been proposed. larger duty cycle may operate in CICM, while the other will then
Among these converters, the single-ended boost converter has automatically operate in DICM. Under this condition, any fur-
been widely adopted as a front-end power-factor-corrected ther additional loading current will be taken up by the converter
(PFC) regulator. The reasons for using boost converter are in CICM operation. Thus, current sharing is very sensitive to the
the simplicity in circuit and system design, reduced voltage mismatch in duty cycle. The design of current-sharing control
stress on devices, and high conversion efficiency compared to circuits has been discussed in [1], [6], [7]. A method of using
the other topologies. Further, the step-up conversion makes it hysteresis current control in a pair of boost converters with cou-
suitable for universal input voltage application (90 264 V). pled inductors has also been suggested in [8].
PFC boost converters operating in continuous inductor In this paper, a converter consisting of two interleaved and
current mode (CICM) have better utilization of power devices, intercoupled boost converter cells is proposed and studied.
lower conduction loss, and lower input current ripple. On the The converter can be designed to have a simple circuit, excel-
other hand, boost converters in discontinuous inductor current lent current sharing characteristics, low input current ripple,
mode (DICM) have lower boost-rectifier reverse-recovery and zero boost-rectifier reverse-recovery loss. The operating
principle, steady-state analysis, and comparison with the
Manuscript received September 5, 1998; revised September 17, 1999. Ab- conventional boost converter are described. The concept is
stract published on the Internet April 21, 2000. This work was supported by the demonstrated by simulation and experimental results.
Research Committee of The Hong Kong Polytechnic University, Hong Kong.
P.-W. Lee was with the Department of Electronic and Information Engi-
neering, The Hong Kong Polytechnic University, Kowloon, Hong Kong. He is
now with Popbridge Ltd., Hong Kong. II. OPERATING PRINCIPLE
Y.-S. Lee and D. K. W. Cheng are with the Department of Electronic and
Information Engineering, The Hong Kong Polytechnic University, Kowloon, Fig. 1 shows the schematic diagram of the proposed con-
Hong Kong. verter, consisting of two interleaved and intercoupled boost con-
X.-C. Liu is with the Department of Electrical Engineering and Applied Elec-
tronics, Tsinghua University, Beijing 100084, China. verter cells. The inductors and are closely coupled and
Publisher Item Identifier S 0278-0046(00)06832-5. with the same winding orientation. The coupled inductors can
0278–0046/00$10.00 © 2000 IEEE
788 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 47, NO. 4, AUGUST 2000

4) State d [Fig. 3(d)]:

The switch SW is closed at time . The current in
inductor starts to rise. continues to discharge. The
rate of change of is approximately given by

(7)

5) State e [Fig. 3(e)]:

Fig. 2. Equivalent circuit of the converter. At time , the inductor current rises at the rate of

(8)
be represented by an equivalent circuit with three uncoupled in-
ductors as illustrated in Fig. 2. The relationships of the inductors
are related by the following equations: where .
6) State f [Fig. 3(f)]:
(1) At time , SW is opened. discharges through the
output circuit. The rate of change of is
(2)

(3) (9)

where The switching cycle will be repeated when SW is turned

, inductances of the two inductors; on again at .
coupling coefficient; A careful study of the waveforms shown in Fig. 4 will reveal
, leakage inductances of the two inductors in the the following interesting facts.
equivalent circuit; 1) As far as the input current is concerned, the converter
mutual inductance. appears to operate in CICM (because is continuous).
Thus, the peak current stress of the inductors/switches and
Referring to the equivalent circuits for the different switching the input current ripple can be maintained relatively low.
states shown in Fig. 3 and the waveforms in Fig. 4, the operation 2) However, as far as the individual boost converter cells
of the converter can be explained as follows. are concerned, since and are discontinuous, these
1) State a [Fig. 3(a)]: converter cells are actually operating in DICM. This
At time , SW is closed. The current in the inductor helps solve the problem of uneven current sharing. Also,
starts to rise while continues to discharge. (The since the boost rectifier SD /SD is turned off before
current in was acquired in the last switching cycle.) SW /SW is turned on, the reverse-recovery loss of the
The rate of change of is approximately given by boost rectifiers is eliminated.
In the waveforms shown in Fig. 4, it has been purposely as-
sumed that the duty cycle of SW (denoted as ) is signifi-
(4)
cantly larger than that of SW (denoted as ). However, this
results in only a small change in the sharing of currents between
2) State b [Fig. 3(b)]: the two inductors ( and as shown in Fig. 1). This is be-
At time , falls to zero. continues to rise and cause of the fact that, for one half of the switching cycle flows
the rate of change of is through , and for the other half of the switching cycle flows
through . If is made relatively constant, and will be
approximately equal.
(5)
III. STEADY-STATE ANALYSIS
where . Assume that the current waveforms shown in Fig. 4 have
3) State c [Fig. 3(c)]: reached a steady state. From the waveform of shown in
At time , SW is opened. The energy stored in the Fig. 4, it can be found that
inductor is transferred to the load via the boost rectifier
SD . The rate of change of is

(6) (10)
LEE et al.: INTERLEAVED BOOST CONVERTER WITH COUPLED INDUCTORS 789

Fig. 3. Operation states and current paths of the converter.

From the waveform of shown in Fig. 4, it can be found that

(11)

Also from Fig. 4, the average values of and , denoted as

and , respectively, are found as

(13)
Assume that the converter is lossless, i.e., the input power
is equal to the output power

(14)
where is the load resistance.
In principle, (10)–(14) can be solved to find . However, in
(12) order to simplify the calculation, it is assumed that ,
790 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 47, NO. 4, AUGUST 2000

In the Appendix, it is also found [from (A27)] that the dif-

ference between and (which are the average values of
and , respectively) is given by

(18)

In (18), if , we have

(19)

It is interesting to note the following.

1) The difference between and is (approximately) pro-
portional to , and this difference can be reduced by
increasing the inductance .
2) When , we have (approximately)

[from (17)] (20)

[from (19) and (20)]. (21)

Therefore, as far as current sharing is concerned,

(for each switching transistor) is the best operating point.

IV. COMPARISON WITH CONVENTIONAL BOOST CONVERTER

Fig. 4. Current waveforms of the converter. AND DESIGN CONSIDERATIONS

In this section, the characteristics of the proposed converter

, , and . After will first be compared with those of the conventional boost con-
some algebraic manipulations and solving a quadratic equation verter (for both CICM and DICM operations). Some special de-
derived from (14), the expression for the output voltage is sign features of the proposed converter will then be briefly dis-
obtained as shown in (15) at the bottom of the page. [The de- cussed.
tailed derivation is given in the Appendix, where the expression
for is derived as (A21).] A. Comparison with Conventional Boost Converter
It should be noted that (15) is valid only if is real (having A comparison of the characteristics between the conventional
no imaginary part). Physically what it means is shown in (16) at converter and the proposed converter is given in Table I.
the bottom of the page.
When is smaller than the value given by (16), the output B. Magnetic Design
voltage will tend to approach the value of In the proposed converter, if the two inductors are more
or (17) tightly coupled, i.e., and are small compared with ,
the current waveforms become more rectangular. However,
which is the output voltage of a conventional boost converter in if and are too small, there is a tendency for SD and
CICM of operation with a duty cycle of ( ). SD to conduct simultaneously at states c and f [Fig. 3(c) and

(15)

(16)
LEE et al.: INTERLEAVED BOOST CONVERTER WITH COUPLED INDUCTORS 791

TABLE I
COMPARISON OF THE CHARACTERISTICS BETWEEN THE CONVENTIONAL CONVERTER AND THE PROPOSED CONVERTER

(f)]. The proposed converter will then behave like a single cedures are then much the same as those of conventional boost
boost converter with two active switches in parallel. In order to converters.
prevent this from happening, we need to keep
or larger than about 1 V, so that the conducting
inductor (either and ) will have a sufficiently large back V. SIMULATION RESULTS
EMF to reverse biase the diode rectifier connected to the other A PSpice simulation model of the proposed converter has
inductor. The practical coupling coefficient can be chosen been developed. Ideal switches and diodes are used in the model.
in the range of 0.9 0.98. In the practical construction of the The simulated waveforms are shown in Figs. 5 and 6.
intercoupled inductors, a lower value of can be achieved The simulation conditions are summarized as follows:
by physically separating the two windings and possibly also , F, H, ,
introducing an air gap between them, thus resulting in a larger V, and kHz. In Fig. 5, . In Fig. 6,
leakage magnetic flux. , .
The simulation results are as follows. In Fig. 5,
C. Feedback Design (peak) (peak) A, (average value of
) (average value of ) A, and V.
In the design of the feedback control circuit, the proposed In Fig. 6, (peak) A, (peak) A, (average
converter could be treated as a single boost converter with a value of ) A, (average value of ) A,
duty cycle of (assuming ). The design pro- and V.
792 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 47, NO. 4, AUGUST 2000

Fig. 5. Simulated current waveforms with matched duty cycles.

Fig. 7. Experimental waveforms with matched duty cycles. Vertical: trace 1,
trigger signal of SW (10 V/div); trace 2, inductor current i (2 A/div); trace
3. trigger signal of SW (10 V/div); trace 4, inductor current i (2 A/div).
Horizontal: time, 5 s/div.

Fig. 6. Simulated current waveforms with mismatched duty cycles.

It is interesting to note that, even in the presence of 40% mis-

match in duty cycles ( and ), the differ-
ence between the average inductor currents ( and ) is only Fig. 8. Experimental waveforms with mismatched duty cycles. Vertical: trace
1, trigger signal of SW (10 V/div); trace 2, inductor current i (2 A/div);
1.3% (1.52 and 1.54 A, respectively). trace 3. trigger signal of SW (10 V/div); trace 4, inductor current i (2 A/div).
Horizontal: time, 5 s/div.

VI. EXPERIMENTAL RESULTS

in a converter comprised of two interleaved and intercoupled
An experimental prototype circuit, as shown in Fig. 1, has
boost converter cells. The proposed circuit has the following ad-
been built. The component values used are the same as those for
vantages.
the simulation described in Section V. Fig. 7 shows the experi-
mentally measured inductor current waveforms for duty cycles • Even in the presence of large duty cycle mismatch, the
. The two inductor currents are practically the intercoupled cells have only small unbalance in current
same. Fig. 8 shows the measured inductor current waveforms sharing.
for and . The average values of these in- • It can be designed to have a small input current ripple and
ductor currents are 1.46 A and 1.50 A, respectively. They differ zero boost-rectifier reverse-recovery loss.
by only 3%. The calculated difference based on (14), (15), and • It does not require any current sensor.
(18) is also approximately 3%. • It is less costly because the two boost converter cells share
the same magnetic core.
Analytical expressions for steady-state operation have been
VII. CONCLUSION
derived to illustrate the excellent current sharing characteristics.
A method to equalize the current sharing between two par- The performance of the converter has also been demonstrated by
allel-connected boost converters has been proposed. This results simulation and experiments.
LEE et al.: INTERLEAVED BOOST CONVERTER WITH COUPLED INDUCTORS 793

APPENDIX
DC ANALYSIS OF INTERLEAVED BOOST CONVERTER
WITH COUPLED INDUCTORS

From the steady-state waveform of shown in Fig. 4, it can

be found that (A8)
Assume that the converter is lossless, i.e., the input power
is equal to the output power . We have
(A1) (A9)

From the waveform of in Fig. 4, it can also be found that (A10)

(A11)
(A12)
(A2) where
average input current;
From (A1) and (A2), and can be expressed as (A3) average current in (or );
and (A4), which are shown at the bottom of the page. Also from average current in (or );
Fig. 4, the average values of and , denoted as and load current;
respectively, are found as load resistance.
Substitution of (A7) and (A8) into (A12) gives

(A13)

(A5) The value of can be obtained by substituting (A3) and (A4)

into (A13), and solving the resulting quadratic equation.
To simplify the calculation, assume , ,
, and . Equation (A13) then
becomes

(A14)
(A6)
Rearranging (A5) and (A6), we have Since , , and are small, we further assume that
, , , and are negligible.
Equation (A14) then becomes

(A7) (A15)

(A3)

(A4)
794 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 47, NO. 4, AUGUST 2000

From (A3) and (A4), (A16)–(A19) can be found, as shown at Applying the assumptions that , , and are small
the bottom of the page. Substituting (A19) into (A15) and then and negligible,
rearranging, we have (A20) and (A21), shown at the bottom of
the page. Note that the other root of the quadratic equation (A20)
gives a less than , which is not a valid solution to a boost
converter.
It should also be noted that (A21) is valid only when is
sufficiently large so as to maintain the current waveforms shown
(A25)
in Fig. 4, i.e., (A22), shown at the bottom of the page.
When is smaller than the value given by (A22), the output
voltage will tend to approach the value of
Substitution of (A18) into (A25) gives
or (A23)
which is the output voltage of a conventional boost converter in
CICM operation with a duty cycle of ( ).
By substituting , ,
(A26)
and into (A7) and (A8), the difference
between and (which are the average values of and
respectively) is found as
In the proposed converter, the term ( ) is small. Equation
(A25) can, therefore, be further reduced to give the following
expression for the difference between the two average inductor
currents:

(A24) (A27)

(A16)

(A17)

(A18)

(A19)

(A20)

(A21)

(A22)
LEE et al.: INTERLEAVED BOOST CONVERTER WITH COUPLED INDUCTORS 795

REFERENCES Yim-Shu Lee (SM’98) received the M.Sc. degree

from the University of Southampton, Southampton,
[1] M. S. Elmore, “Input current ripple cancellation in synchronized, par- U.K., and the Ph.D. degree from the University
allel connected critically continuous boost converters,” in Proc. IEEE of Hong Kong, Hong Kong, in 1974 and 1988,
APEC’96, vol. 1, San Jose, CA, Mar. 1996, pp. 152–158. respectively.
[2] J. S. Lai and D. Chen, “Design consideration for power factor correc- He had been with Cable and Wireless, Rediffusion
tion boost converter operating at the boundary of continuous conduction Television, and the General Post Office, all in Hong
mode and discontinuous conduction mode,” in Proc. IEEE APEC’93, Kong, before joining The Hong Kong Polytechnic
San Diego, CA, Mar. 1993, pp. 267–273. University, Hong Kong, in December 1969 as a
[3] J. Wang, W. G. Dunford, and K. Mauch, “A comparison of modified member of the academic staff. Currently, he is the
boost converters with continuous inductor current mode and ripple free Chair Professor of Electronic Engineering in the
input current with conventional converters,” in Conf. Rec. IEEE-IAS Department of Electronic and Information Engineering. He is the author of
Annu. Meeting, vol. 2, San Diego, CA, Oct. 1996, pp. 878–885. Computer-Aided Analysis and Design of Switch-Mode Power Supplies (New
[4] J. Wang, W. G. Dunford, and K. Mauch, “A comparison between two York: Marcel Dekker, 1993). In addition, he has authored more than 100
proposed boost topologies and convention topologies for power factor published technical papers on the design of electronic circuits. His current
correction,” in Conf. Rec. IEEE-IAS Annu. Meeting, San Diego, CA, Oct. research interests include power electronics and computer-aided design of
1996, pp. 1210–1217. analog circuits.
[5] D. K. W. Cheng, X. C. Liu, and Y. S. Lee, “A new modified boost con- Prof. Lee is a Fellow of the Institution of Electrical Engineers, U.K., and the
verter with ripple free input current by using coupled inductors,” in Proc. Hong Kong Institution of Engineers.
IEE Int. Conf. Power Electronics and Variable Speed Drives, London,
U.K., Sept. 1998, pp. 592–599.
[6] B. A. Miwa, D. M. Otten, and M. F. Schlecht, “High efficiency
power factor correction using interleaving techniques,” in Proc. IEEE David K. W. Cheng (M’90) received the B.Sc. de-
APEC’92, vol. 1, Boston, MA, Feb. 1992, pp. 557–568. gree in electronic engineering from Brighton Univer-
[7] L. Balogh and R. Redl, “Power-factor correction with interleaved sity, Brighton, U.K., and the Ph.D. degree from the
boost converters in continuous-inductor-current mode,” in Proc. IEEE University of Hong Kong, Hong Kong, in 1975 and
APEC’93, San Diego, CA, Mar. 1993, pp. 168–174. 1992, respectively.
[8] J. W. Kolar, G. R. Kamath, N. Mohan, and F. C. Zach, “Self-ad- He held appointments with Pye Ether Ltd. (U.K.)
justing input current ripple cancellation of coupled parallel connected from 1975 to 1978 and Biccotest (U.K.) from 1978
hysteresis-controlled boost power factor correctors,” in Proc. IEEE to 1983. He is currently an Assistant Professor in
PESC’95, vol. 1, Atlanta, GA, June 1995, pp. 164–173. the Department of Electronic and Information En-
gineering, The Hong Kong Polytechnic University,
Hong Kong. His research interests include power
electronics and computer-aided design of electronic circuits.

Xiu-Cheng Liu received the B.Sc. and M.Sc. degrees

Po-Wa Lee (M’94) received the B.Eng (Hons.) and from Tsinghua University, Beijing, China, in 1983
Ph.D. degrees from the University of Warwick, War- and 1989, respectively.
wick, U.K., in 1991 and 1997, respectively, both in He was appointed as a Lecturer in the Department
electrical engineering. of Electrical Engineering, Tsinghua University, in
From 1997 to 1998, he was a Research Associate 1990. He was a Research Assistant at The Hong
in the Department of Electronic and Information En- Kong Polytechnic University, Hong Kong, from
gineering, The Hong Kong Polytechnic University, August 1996 to December 1997. He is currently an
Hong Kong. He is currently with Popbridge Ltd., Associate Professor in the Department of Electrical
Hong Kong, designing switch-mode power supplies. Engineering and Applied Electronics, Tsinghua
His research interests include magnetic system University. His recent research interests include
design, analysis, and control of power electronics. circuit theory and its applications and design of switch-mode power supplies.