Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267965006

Characterizing the Effects of Inductor Coupling on the


Performance of an Interleaved Boost Converter

Article · January 2009

CITATIONS READS

32 4,347

6 authors, including:

Austin Page Biswajit Ray


Massachusetts Institute of Technology Bloomsburg University
3 PUBLICATIONS 87 CITATIONS 28 PUBLICATIONS 361 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Biswajit Ray on 10 June 2015.

The user has requested enhancement of the downloaded file.


March 30 - April 2, 2009 CARTS USA 2009 Jacksonville FL

Characterizing the Effects of Inductor Coupling on the Performance of


an Interleaved Boost Converter

Hiroyuki Kosai
UES Inc.
Dayton, OH

Seana McNeal, Austin Page, Brett Jordan, Jim Scofield


Air Force Research Laboratory
WPAFB, OH

Biswajit Ray
Bloomsburg University of Pennsylvania
Bloomsburg, PA

Abstract
Interleaved power converter topologies have received increasing attention in recent years for
high power and high performance applications. The advantages of interleaved boost
converters include increased efficiency, reduced size, reduced electromagnetic emission,
faster transient response, and improved reliability. The front end inductors in an interleaved
boost converter are magnetically coupled to improve electrical performance and reduce size
and weight. In this paper we focus on developing useful design equations for interleaved
boost converters for operation under continuous inductor current mode (CICM). The effects
of inverse and direct inductor coupling on the key converter performance parameters such as
inductor ripple current, input ripple current, minimum load current requirement for achieving
CICM operation, and dc and ac flux levels are quantified and experimentally verified.
Compared to the direct coupling configuration, inverse coupling provides the advantages of
lower inductor ripple current and negligible dc flux levels in the core. On the other hand,
direct inductor coupling provides the advantages of lower input ripple current and reduced ac
flux levels. Therefore, application specific design trades will necessarily involve a
consideration of higher ac flux density dependent core loss for inversely coupled inductors
versus the higher winding loss associated with a direct coupled winding. Additionally,
higher levels of conducted electromagnetic emission, associated with the higher input ripple
current in an inverse coupled configuration, may also be a consideration at higher power
levels. However, due primarily to core saturation concerns and magnetic volume and weight
savings advantages, the inverse coupled configuration is recommended for most high power
applications.
Introduction
Interleaved buck and boost converters have been studied in recent years to improve power
converter performance in terms of efficiency, size, conducted electromagnetic emission, and
transient response. The benefits of interleaving include high power capability, modularity,
and improved reliability. However, an interleaved topology improves converter performance
at the cost of additional inductors, power switching devices and output rectifiers. Since the
inductor is the largest and heaviest component in a power boost converter, the use of a
coupled inductor instead of multiple discrete inductors is preferable. Coupled inductors also
offer additional advantages such as reduced core and winding loss as well as improved input
and inductor current ripple [1].

Performance improvements of voltage regulator modules, using a coupled inductor in an


interleaved buck converter, have been shown to include improved efficiency and transient
response [2, 3], when operated in continuous inductor current mode (CICM) and applying
the concept of equivalent inductance. In another report, an interleaved power boost
converter, under discontinuous inductor current mode (DICM) operation, was shown to
enable a reduction in the size of the coupled inductor, but at the cost of increased ripple
current and parasitic ringing [4]. Generalized steady state studies of multiphase interleaved
boost converters, under CICM operation have also been accomplished. Specifically [5]
compared input and inductor ripple currents to output ripple voltage, and [6] showed
improved converter efficiency and dynamic response using a signal flow modeling approach.
Additionally, a simplified approach to the modeling of coupled inductors was recently
reported in [7], and [8] demonstrated that current mode control in an interleaved boost
converter could be used to ensure balanced current sharing between legs.

The focus of the present paper is on developing useful design equations for CICM operation
of an interleaved boost converter. The effects of inductor coupling on the key converter
performance parameters such as; inductor ripple current, input ripple current, minimum load
current requirement for achieving CICM operation, and dc and ac flux levels in the coupled
inductor are analyzed, quantified, and then experimentally verified for both inverse and
direct coupled configurations. The following section provides detailed analysis supported by
graphical comparison of inductor coupling type on converter performance. Subsequent
sections provide experimental verification of key performance parameters as a function of
inductor coupling type.

Performance analysis based on inductor coupling type


In this section, the inductor currents and the equivalent inductances for the converter were
derived under CICM operation for a duty ratio greater than 0.5. These equivalent
inductances were then used to study the effect of inductor coupling on the converter’s
inductor and input ripple currents, minimum load current requirement for CICM operation,
and dc and ac flux levels in the inductor.
Inductor current and equivalent inductance analysis
The schematic of an inverse-coupled interleaved boost converter is shown in Fig. 1. For the
coupled inductor, the mutual inductance is M and the self inductances (L1 and L2) are
assumed to be equal with a value of L, and the coupling factor (k) defined as M/L.

The basic relationship among voltage and current for the coupled inductor is given by (1) and
(2).

di1 di
v1 = L −M 2
dt dt
(1)

di2 di
v2 = L −M 1
dt dt
(2)

A set of theoretical waveforms for CICM operation of the interleaved boost converter is
shown in Fig. 2 for a duty ratio (D) greater than 0.5. Applying volt-sec balance to either
winding of the coupled inductor provides the well-known dc voltage gain equation [Vout/Vin =
1/(1-D)].

Using equations (1) and (2) and the dc voltage gain relation, equations (3) and (4) were
derived in which a l and cl ,where the subscript l denotes the time interval, were solved
simultaneously for each interval. Then bl and d l values were determined by matching the
current at each boundary between the intervals. The dc component of the current through
each winding ( I dc ) is given by Vin /[ 2 R L (1 − D) 2 ] .

i1 = al t + b/ + I dc
(3)
i2 = cl t + d / + I dc
(4)
The results are shown in Table I for the inverse-coupled (DC flux balanced) configuration.
The corresponding expressions for direct-coupled inductor configurations are obtained by
substituting k with –k in Table I.
Table I: Coefficients for instantaneous current expressions for inverse-coupled boost
converter

Time interval

I II III IV
{0 ≤ t ≤ (D- {(D-0.5)Ts ≤ t ≤ {0.5Ts ≤ t ≤ DTs} {DTs ≤ t ≤ Ts}
0.5)Ts} 0.5Ts}

al Vin Vin ( D − 1 + kD) Vin Vin ( D − k + kD)


− −
L(1 − k ) L(1 − D)(1 − k )2 L(1 − k ) L(1 − D)(1 − k 2 )

bl VinTs ( D − k + kD VinTs D(kD + D − 1) VinTs ( D + kD + k ) VinTs (1 + D)(kD + D −


− 2 2

2 L(1 − k ) 2 L(1 − k )(1 − D) 2 L(1 − k 2 ) 2 L(1 − k 2 )(1 − D)

cl Vin Vin ( D − k + kD) Vin Vin ( D − 1 + kD)


− −
L(1 − k ) L(1 − D)(1 − k 2 ) L(1 − k ) L(1 − D)(1 − k 2 )

dl VinTs (kD + D − 1) VinTs D(kD + D − k ) VinTs (1 + D + kD) VinTs (1 + D)(kD + D −


− 2 2

2 L(1 − k ) 2 L(1 − k )(1 − D) 2 L(1 − k 2 ) 2 L(1 − k 2 )(1 − D)
The results of Table I can also be expressed in terms of equivalent inductances [2,3] as
shown in Table II. Again, the corresponding equivalent inductance expressions for the
direct-coupled configuration are obtained by simply substituting the coupling factor k with –k
in Table II. These equivalent inductances are used next in quantifying input and inductor
ripple currents as well as in determining the minimum load current requirement for CICM
operation.

Table II: Equivalent inductance expressions for inverse-coupled boost converter

Time interval

I II III IV
{0 ≤ t ≤ (D- {(D-0.5)Ts ≤ t ≤ {0.5Ts ≤ t ≤ {DTs ≤ t ≤ Ts}
0.5)Ts} 0.5Ts} DTs}

Leq1 (1 − k) L 1− k 2 (1 − k) L 1− k 2
(leg 1) L L
D 1− D
1− k 1− k
1− D D

Leq2 (1 − k) L 1− k 2 (1 − k) L 1− k 2
(leg 2) L L
1− D D
1− k 1− k
D 1− D

Ripple current and minimum load current analysis for CICM operation
The peak-to-peak inductor ripple current is governed by the same equivalent inductance
value corresponding to time interval IV for leg-1 and time interval II for leg-2. For the
inverse-coupled configuration, the peak-to-peak inductor ripple current is given by
Vout D(1 − D)
I L , ripple , p − p = .
1− k2
fsL
1− D
1− k
D
(5)
The corresponding inductor ripple current ratio between inverse and direct-coupling
configurations is given by (6) with the associated graph shown in Fig. 3.

1− D
I L ,ripple, p − p ,inverse _ coupling 1− k
= D
I L ,ripple, p − p , direct _ coupling 1− D
1+ k
D
(6)

Therefore, the inductor ripple current for the inverse-coupling configuration is always less
than the direct-coupling configuration; this ripple becomes significantly higher for the direct-
coupling case when operating with a duty ratio closer to 0.5 and as the coupling ratio
approaches unity.

The input ripple current, on the other hand, is controlled by the equivalent inductance
corresponding to time interval I. An expression for input ripple current for the inverse-
coupling configuration is given by (7),

2Vout ( D − 0.5)(1 − D)
I in, ripple, p − p = ,
f s L(1 − k )
(7)

while the ripple current ratio between two coupling configurations is given by (8).

I in,ripple, p − p ,inverse _ coupling 1+ k


=
I in,ripple, p − p ,direct _ coupling 1− k
(8)

The input ripple current for the inverse-coupling configuration is always greater than the
direct-coupling configuration. As shown in Fig. 4, the difference in input ripple current
between the two configurations is independent of duty ratio, and increases dramatically with
coupling factors above 0.6.
Another key advantage of the inverse-coupled configuration is the ability to maintain CICM
operation over a wider load range compared to the direct-coupled configuration. For a given
value of L, the normalized minimum load current (Iout,min_N) for maintaining CICM operation
for the inverse-coupled configuration is given by (9).

⎛ 1− D ⎞
D(1 − D) 2 ⎜1 − k⎟
I out ,min ⎝ D ⎠
I out ,min_ N ≡ =
⎛ Vout ⎞ (1 − k 2 )
⎜⎜ ⎟⎟
⎝ Lf s ⎠
(9)
The corresponding plots for both inverse and direct-coupled configurations are shown in Fig.
5. It can be noted that for a fixed coupling and duty ratios, the CICM operation can be
maintained down to a lower value of load current for the inverse-coupling case compared to
the direct-coupling case. Since the minimum load current for CICM operation is directly
proportional to the inductor ripple current, equation (6) and Fig. 3 also represent the ratio of
the minimum load current for CICM operation between inverse and direct-coupled
configurations.

Fig. 1: Interleaved boost converter circuit.

Fig. 2: Theoretical waveforms of the


interleaved boost converter.
Fig. 3: Inductor ripple current ratio between
inverse and direct coupling configurations.

Fig. 4: Input ripple current ratio between


inverse and direct coupling configurations.

Fig. 5: Normalized minimum load current for CICM operation.


Magnetic flux density analysis for CICM operation
Calculations for magnetic flux density and the associated currents that generate the flux are
considered next by using a transformer-like equivalent circuit of the coupled inductor shown
in Fig. 6. The basic voltage equations (1) and (2) are rearranged next for deriving explicit
expressions for voltage across the mutual and leakage inductances. Also, an expression for
the flux, φ1 and φ2 , crossing each winding and the number of turns for each winding, N
can be expressed at the same time.

dφ1 di d (i1 − i2 ) dφ 2 di d (i1 − i2 )


v1 = N = (L − M ) 1 + M v2 = N = (L − M ) 2 − M
dt dt dt dt dt dt
(11)
(10)
⎛ di di ⎞
vM = M⎜ 1 − 2 ⎟
⎝ dt dt ⎠
(12)
Using the expressions above, voltages across the mutual and leakage inductances for both
inverse and direct-coupled configurations can be derived. The results are shown in Table III
for the specified time intervals.
Table III: Voltage across mutual and leakage inductances
Time interval

I II III IV
{0 ≤ t ≤ (D- {(D-0.5)Ts ≤ t ≤ {0.5Ts ≤ t ≤ {DTs ≤ t ≤ Ts}
0.5)Ts} 0.5Ts} DTs}
vM 0 0 kVin
(Inverse −
coupled) kVin (1 + k )(1 − D )
(1 + k )(1 − D)
vM 2kVin k (2 D − 1)Vin 2kVin k (2 D − 1)Vin
(Direct − −
coupled) (1 + k ) (1 + k )(1 − D) (1 + k ) (1 + k )(1 − D)

vLK Vin (kD + D − 1)Vin Vin (kD + D − k )Vin


(Inverse − −
coupled) (1 + k )(1 − D) (1 + k )(1 − D)

vLK Vin (1 − k ) (kD + 1 − D)Vin Vin (1 − k ) (−kD + k + D)V


(Direct −
coupled) (1 + k ) (1 + k )(1 − D) (1 + k ) (1 + k )(1 − D)
For CICM operation, the voltages across the windings are identical for both direct and
inverse polarization configurations throughout the period; therefore the total ac magnetic flux
crossing each winding is identical. The ac voltages across the windings are the sum of the
voltage across the mutual inductance and the voltage across the leakage inductance. Since
the induced voltage is related to flux by Faraday’s law, the mutual flux and the leakage flux
across the winding can therefore be determined.

The main source of flux leakage is introduced by core gaps. For the configuration where two
windings are positioned next to and separated by a gap on the center leg, the majority of
leaking flux exits from the gap. Outside these windings, flux associated with the leakage
inductance takes a shorter path (i.e. through the air) and is uncoupled. The flux associated
with the mutual inductance travels through both set of windings, and a large portion of the
flux remains in the core. The winding orientation for both coupling configurations is shown
in Figure 7. Figure 8 shows the relationship between the voltage across the mutual
inductance and the voltage across the leakage inductance. Also shown is the current
associated with the mutual inductance. It should be noted that the ac flux in the core is not
related to the converter’s dc output current.

In the dc environment where two windings share the dc current equally, the inverse-coupled
configuration cancels most flux in the core since the flux generated by the two windings has
opposing polarity. This opposing flux configuration makes the flux in the core away from
the windings substantially small. At the core directly underneath the windings, the resultant
flux is (1 − k ) LI dc / N , where I dc is dc current through the winding. For the direct-coupled
configuration, the flux generated by two windings will sum rather than canceling within the
core, which is (1 + k ) LI dc / N . As a result, direct-coupling configurations will saturate the
core more readily when operated at a high output current. The dc flux consideration
represents a significant benefit to using inverse-coupled configuration in high power dc
applications.

Fig. 7: Winding orientation for inverse


(left) and direct (right) coupled
Fig. 6: Transformer-like equivalent configurations.
circuit of the coupled inductor.
(a) Inverse coupled configuration

b) Direct coupled configuration

Fig. 8: Voltages across the mutual and


leakage inductances, and current
generating mutual flux.

Useful expressions for peak-to-peak magnetic flux density, which are associated with both
mutual flux and leakage flux, are shown in Table IV. For these expressions, the effective
cross-sectional area of the core is denoted as Ae, and the number of turns for each winding is
denoted as N.

It can be observed that for a given input voltage, the mutual flux density is independent of
duty ratio for inverse-coupled configuration, and it is duty ratio dependent for direct-coupled
configuration. Additionally, the mutual flux density ratio between inverse and direct-
coupled configurations is 1 /(2 D − 1) . As shown in Fig. 9, compared to the direct-coupled
case, the mutual flux level can be significantly higher in the inverse-coupled configuration
for duty ratio values closer to 0.5. Accordingly, the mutual flux induced core losses would
be significantly higher in the inverse-coupled configuration.
Table IV: Currents generating mutual flux and flux density expressions for the coupled
inductor

Coupling ( i1 − i 2 ) for inverse Mutual flux Leakage flux


configuratio density density
coupled
n (peak-to-peak) (peak-to-peak)
( i1 + i 2 ) for direct
coupled
(peak-to-peak)
Inverse kVin kVin Vin (kD + D − k )
coupled M (1 + k ) f s (1 + k ) f s NAe (1 + k ) f s NAe
Direct kVin (2 D − 1) kVin (2 D − 1) Vin (−kD + k + D)
coupled M (1 + k ) f s (1 + k ) f s NAe (1 + k ) f s NAe
Fig. 9: Peak-to-peak ac flux ratio between inverse and direct coupling configurations.

Experimental Results
An interleaved boost power converter was prototyped and tested to verify the presented
analysis and design equations; specifically, inductor ripple current, input ripple current,
minimum load current for CICM operation, and magnetic flux density in the coupled
inductor. Each leg of the 100V/270V, 2 kW power stage, shown in Fig. 10, consists of four
MOSFETs (IRFP27N60) and four Schottky diodes (CSD20030D). The output filter
capacitance consists of two 30 µF X7R ceramic capacitors. The coupled inductor was
designed with a single pair of N27 ferrite E-E cores (E70/33/32) from EPCOS. An air gap of
3 mm was used in all three legs of the core, and each winding of 12 turns was placed in the
center leg. Based on the measured open and short circuit inductances, the calculated self
inductance (L) is 42 µH and mutual inductance (M) is 27.5 µH representing a coupling factor
(k) of 0.65. Each leg of the converter was switched at a frequency of approximately 70 kHz
with a phase shift of 180o between the two legs.
Inductor ripple current
Recorded inductor ripple current waveforms are shown in Figs. 11 and 12 for inverse and
direct-coupling configurations, respectively. The measured inductor ripple currents are 6.2 A
and 14 A for inverse and direct-coupling configurations, respectively. Based on (5), the
corresponding calculated values are 7 A and 15.7 A. The measured values are about 11%
less than the calculated ones primarily due to voltage drop in the inductor winding resistance,
MOSFETs’ on-resistance, and rectifiers’ forward voltage drop not accounted for in the
analysis. The ratio between the measured ripple currents is 0.44 and the corresponding
theoretical value from (6) is 0.45.

Input ripple current


Recorded input ripple current waveforms are shown in Figs. 13 and 14 for inverse and direct-
coupling configurations, respectively. The measured input ripple currents are 6.6 A and 1.5
A for inverse and direct-coupling configurations, respectively. Based on (7), the
corresponding calculated values are 7.6 A and 1.6 A. Therefore, the measured values are
about 13% and 6% less than the calculated ones for inverse and direct-coupled
configurations, respectively. This deviation again is contributed primarily by various voltage
drops in the converter not accounted for in the analysis. The ratio between the measured
ripple currents is 4.4 and the corresponding theoretical value from (8) is 4.7.

Minimum load current for CICM operation


The inductor current waveforms shown in Figs. 11 and 12 are for the minimum load current
necessary to maintain CICM operation. The corresponding experimental load currents are
2.1 A and 5.0 A for inverse and direct-coupling configurations, respectively. Using (9),
these numbers can be compared to the calculated values of 2.6 A and 5.6 A, respectively.
Again, discrepancy between the measured and calculated values is due to voltage drop in
non-ideal passive components and semiconductor devices. The ratio between the measured
minimum load currents for CICM operation is 0.42 and the corresponding theoretical value
from (6) is 0.45.

Magnetic flux density in the core


AC magnetic field generated in the core was experimentally studied by recording the voltage
induced to a one turn pickup coil closely wrapped at the middle of the center leg gap. The
coupled inductor used in measuring the flux density had the following parameters: N = 12,
Ae = 6.81x10-4 m2, L = 44 µH, M = 27 µH, k = 0.61, and the relevant circuit parameters were:
Vin = 10 V, D = 0.63, and fs = 75.4 kHz. The majority of the flux associated with the leakage
inductance escapes at the gaps. Therefore, the flux crossing the pickup coil was associated
with mutual inductance and some leakage inductance. The experimentally obtained flux
density waveforms for inverse and direct-coupled configurations are shown in Figs. 15 and
16, respectively. The measured flux density waveforms were very similar to Fig. 8.

From Figs. 15 and 16, the experimental peak-to-peak flux densities were 6.4 mT and 1.8 mT
for inverse and direct-coupled configurations, respectively. From Table IV, the
corresponding calculated peak-to-peak flux densities are 6.2 mT and 1.6 mT. The results
proved that almost all flux associated with leakage inductance exited before reaching the
pickup coil, and the measured flux was almost entirely associated with mutual inductance. In
addition, the measured ratio of flux densities between inverse and direct-coupled
configurations was 3.6 whereas the predicted ratio, 1 /(2 D − 1) , was 3.8. The experimental
investigation, therefore, confirmed the validity of the flux analysis presented in the previous
section.
Direct coupling
Time: 2 8 s/div i1 (5 A/div)

0
i2 (5 A/div)
Fig. 10: Prototype of the interleaved boost
converter.
0
Inverse coupling
Time: 2 s/div
iin (2 A/div)

Fig. 14: Input current waveform for direct-coupling


(Vin = 30 V/Iin = 14 A/Vout = 78 V/Iout = 5.0 A
4

Flux density (mT)


0
1

0
Fig. 11: Inductor current waveforms for inverse- -1
coupling -2

(Vin = 30 V/Iin = 6.0 A/Vout = 80 V/Iout = 2.2 A). -3

-4
0 5 10 15 20 25
Inverse coupling Time (µs)
Time: 2 ℵs/div i1 (2 A/div)

Fig. 15: Measured ac magnetic flux density at


0 the center gap for inverse-coupled configuration.
i2(2 A/div)
1.5

1
0
Flux density (mT)

0.5

Fig. 12: Inductor current waveforms for direct- 0

coupling -0.5
(Vin = 30 V/Iin = 14 A/Vout = 78 V/Iout = 5.0 A).
-1
Direct coupling
Time: 2 ° s/div -1.5
0 5 10 15 20 25
iin (1 A/div)
Time (µs)

Fig. 16: Measured ac magnetic flux density at


the center gap for direct-coupled configuration

DC offset: 10 A
0

Fig. 13: Input current waveform for inverse-


coupling
(Vin = 30 V/Iin = 6.0 A/Vout = 80 V/Iout = 2.2 A
Conclusions
Useful design equations for a magnetically coupled interleaved boost converter under
continuous inductor current mode (CICM) operation were presented. Compared to the
direct-coupling configuration, the inverse-coupling configuration provides the advantages of
lower inductor ripple current and very low dc flux in the core. The direct-coupling
configuration, on the other hand, provides the advantages of lower input ripple current and ac
flux level. Therefore, the ac flux-density dependent core losses would be higher in an
inverse-coupled converter whereas the winding loss would be higher in a direct-coupled
converter. Magnetic saturation due to dc flux is a greater design challenge with direct-
coupled converters. Accordingly, an inverse-coupled configuration is the logical choice for
high power interleaved boost converters. Input current ripple is higher in an inverse-coupled
converter resulting in a relatively higher level of conducted electromagnetic emission and/or
increased size of input capacitive filter.

References
1. W. Wen and Y. Lee, “A two-channel interleaved boost converter with reduced core
loss and copper loss,” Proc. IEEE Power Electronics Spec. Conf., June 2004, pp.
1003-1009.
2. P. Wong et al., “Performance improvements of interleaving VRMs with coupling
inductors,” IEEE Trans. Power Electronics, Vol. 16, No. 4, pp. 499-507, July 2001.
3. P. Wong et al., “Investigating coupling inductors in the interleaving QSW VRM,”
Proc. IEEE Applied Power Electronics Conf., Feb. 2000, pp. 973-978.
4. X. Huang et al., “Parasitic ringing and design issues of digitally controlled high
power interleaved boost converters,” IEEE Trans. Power Electronics, Vol. 19, No.
5, pp. 1341-1352, Sept. 2004.
5. H. Shin et al., “Generalized steady-state analysis of multiphase interleaved boost
converter with coupled inductors,” IEE Proc. Elect. Power Appl., Vol. 152, No. 3,
pp. 584-594, May 2005.
6. M. Veerachary, “Analysis of interleaved dual boost converter with integrated
magnetics: signal flow graph approach,” IEE Proc. Elect. Power Appl., Vol. 150,
No. 4, pp. 407-416, July 2003.
7. J. Gallagher, “Designing coupled inductors,” Power Electronics Technology, April
2006, pp. 14-21.
8. R. Giral et al., “Interleaved converters based on CMC,” IEEE Trans. Power
Electronics, Vol. 14, No. 4, pp. 643-652, July 1999.

View publication stats

You might also like