This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 1

Extension of Soft-Switching Region of

Dual-Active-Bridge Converter by Tunable Resonant
Tank
M. Yaqoob, Student Member, IEEE, K. H. Loo, Member, IEEE, and Y. M. Lai, Senior Member, IEEE

High Voltage Low Voltage

Abstract—Hard-switching induced switching loss can con- DC Bus DC Bus
tribute significantly to the power loss of an isolated bidirectional AC DC
dual-active-bridge (DAB) dc-dc converter operating at high DC DC
frequency. An LC-type series resonant DAB converter based DC
on switch-controlled-inductor (SCI) is proposed to mitigate the DAB Converter Storage Battery
Wind Power DC
loss arising from hard switching under wide-range variations DAB based Solid-State DC
in output voltage and current. Zero voltage switching (ZVS) DC Transformer DC
is achieved at the primary side (high voltage), while at the DC DAB Converter Super-Capacitor
secondary side (low voltage) zero current switching (ZCS) is PV Array
preferred to reduce excessive ringing due to circulating current DC
and switching loss. In order to achieve reduced conduction loss, AC
DC
a nominal operating point is chosen where the root-mean-square DC DAB based Solid-State AC Transmission
(rms) resonant tank current is the minimum. To validate the Transformer Lines
AC Bus
proposed topology and modulation scheme, an LC-type series Fuel Cell Stack for Utility Voltage
resonant DAB converter based on SCI operating at 100 kHz is
designed to interface a 400 V dc bus to a super-capacitor-based Fig. 1. Bidirectional power flow and distributed generation in a micro-grid
energy storage. Simulation and experimental results validate the system.
effectiveness of the proposed topology for charging/discharging a
super-capacitor with an output voltage variation of between 10
V to 48 V and a maximum rated power of 480 W. A maximum
efficiency of 94.6 % is achieved using the proposed topology and advancements in semiconductor and magnetic devices it has
modulation scheme. attracted more attention for its use in medium-voltage micro-
Index Terms—Dual-active-bridge converter, Zero voltage grid systems [4]–[7]. A DAB converter based on LC-type
switching, Zero current switching, Super-capacitor, Switch- series resonant tank uses an inductor and a capacitor as energy
controlled inductor. transfer elements [8], but under wide variations in output
voltage and current, this topology undergoes hard switching
I. I NTRODUCTION in certain operating regions and causes switching loss that
IDIRECTIONAL power flow is expected to be an essen- deteriorates converter’s performance in terms of efficiency.
B tial feature in future electric power systems where bidi-
rectional power conversion systems play a significant role. The
Hard switching in conventional DAB converter is mainly due
to the out-of-phase relationship between the series resonant
increasing penetration of renewable energy generation arises tank’s current and bridge voltages [9]. There is an increase in
due to its positive impact on the environment by providing low the trend of high-frequency operation of DAB converter due
carbon emissions and this has led to the development of micro- to the rise of new semiconductor devices with extremely low
grid systems [1], [2]. A generic micro-grid system is mainly gate charge such as SiC (Silicon-Carbide) and GaN (Gallium-
composed of distributed renewable energy resources and stor- Nitride) [11], [12]. When operating at high frequencies these
ages as shown in Fig. 1. These renewable energy resources devices can suffer from excessive switching loss.
are highly dynamic in nature and need to be equipped with Many modulation methods and converter topologies have
energy storages to achieve stable and un-interruptible power been proposed to extend soft-switching region but they cover
supply. Bidirectional power conversion systems can be used only limited operating regions. Single-phase-shift (SPS) mod-
to integrate these storage devices to the dc bus in a micro- ulation is used in [8] for the design of LC-type series
grid and in this context many bi-directional power converters resonant DAB converter. It uses one degree of freedom to
have been proposed. Dual-active-bridge (DAB) converter has modulate power and fails to provide full-range soft switching
shown many advantages over its counterpart topologies in especially under light load condition and non-unity voltage
terms of device stress, ease of control, galvanic isolation, and gain operation. In order to reduce circulating current, an
power density. It was first proposed in [3] and with recent extended-phase-shift (EPS) modulation is introduced in [13]–
[15] which uses an internal-phase-shift within the primary or
The authors are with the Department of Electronic and Informa- secondary bridges, and an external-phase-shift between the
tion Engineering, The Hong Kong Polytechnic University, Hong Kong
SAR. (e-mail: [email protected]; [email protected]; primary and secondary bridges of a DAB converter for power
[email protected];) modulation. Investigation of zero voltage switching (ZVS)

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 2

S 1 S 3 X Q 1 Q 3

L C n:1
V C C s V o
i p
+ +
i (t) +
v (t)
p
p
v' (t) s
v (t)
s

- -
-
S 2 S 4 Q 2
Q 4

Fig. 2. Conventional DAB converter based on LC-type series resonant tank.

and non-ZVS operating regions under dual-phase-shift (DPS) II. A NALYSIS OF C ONVENTIONAL AND P ROPOSED DAB
modulation has been presented in [16] and is implemented T OPOLOGY
in [17]–[19]. Power characterization of a DAB converter using A. Conventional Topology Based on LC Series Resonant Tank
triple-phase-shift (TPS) is presented in [20]–[22]. With EPS,
DPS, and TPS, however, soft switching can only be achieved A conventional LC-type series resonant DAB converter
for certain operating regions and also at the cost of complex is shown in Fig. 2. Switches S1 -S4 form the primary side
modulation schemes. Fundamental duty modulation (FDM) of the converter and are modulated to generate vp (t). The
and fundamental-optimal strategy (FOS) based on fundamental secondary side of the converter which is composed of Q1 -
component analysis (FCA) are explained in [23], [24] which Q4 generates a phase-shifted vs (t) as shown in Fig. 4. The
also achieve ZVS of all switches under certain operating magnitude and direction of power flow are determined by the
regions only. The main objective of FDM based on FCA is phase shift θ between vp (t) and vs′ (t). Power is assumed to
to reduce circulating current and computational complexities be transferred by the fundamental components of the resonant
involving on-line and off-line calculations as proposed in [15], tank current and bridge voltages, which leads to simplified
[20], [21]. analytical expressions of results [23]. Therefore, FCA is used
to determine the converter’s behavior, and an equivalent circuit
along with the corresponding phasor diagrams showing the
voltages and currents of interest are shown in Fig. 3. The
respective equations for the primary-reflected resonant tank
In this paper, an LC-type series resonant DAB converter current and bridge voltages are given by (1)-(3).
incorporating switch-controlled inductor (SCI) is proposed to
achieve soft switching for wide-range variations in output 4Vp
vp (t) = sin ωt (1)
voltage and current. ZVS at the primary side (high voltage) π
and zero current switching (ZCS) at the secondary side (low
′ 4M Vp
voltage) are achieved for charging/discharging a 48 V super- vs (t) = sin(ωt − θ) (2)
capacitor module from a 400 V dc bus. ZCS is preferred π
over ZVS at the secondary side, where high current flows, to 4Vp
avoid excessive ringing and current surge due to circulating ip (t) = (− cos ωt + M cos(ωt − θ)) (3)
πX
current arising from the out-of-phase relationship between
where M = nV Vp , X = ωL − ωC , θ is the phase shift between
s 1
bridge voltage and tank current at the secondary side leading
′
to significant energy stored in parasitic inductances [10], [25]. vp (t) and vs (t), and n is the transformer’s turn-ratio.
When designing the proposed converter, conduction loss has 1) Required Soft-Switching Conditions: From Fig. 2, in
been minimized by choosing the nominal operating point order to achieve ZVS operation during the switching instances
that gives the minimum rms resonant tank current at the of S1 and S4 , the corresponding body diodes should conduct
maximum value of output current over the entire range of before S1 and S4 are turned on, implying that ip (t) must be
voltage conversion ratio. In Section II, both the conventional negative when S1 and S4 are turned on. As for the secondary
LC-type series resonant DAB converter and the proposed side, ip (t) must be positive to achieve ZVS and zero for ZCS
topology based on SCI are analyzed to identify their ZVS when Q1 and Q4 are turned on. Similarly, for S2 and S3 ,
and ZCS conditions. To reduce conduction loss, the design ip (t) should be positive, and negative or zero for Q2 and
of nominal operating point with minimum rms tank current is Q3 . Operating waveforms depicting hard-switching and soft-
explained in Section III. The design of experimental prototype, switching behavior of the conventional DAB converter are
simulation and experimental results for charging/discharging a shown in Fig. 4. Thus, evaluating (3) at ωt = 0 (i.e., when
super-capacitor module are presented in Section IV. Finally, S1 and S4 are turned on) and θ (i.e., when Q1 and Q4 are
concluding remarks are given in Section V. turned on) lead to the required conditions for achieving ZVS

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 3

L ip C
vp
vp
vp v's ᶿ vx
vx
ᶿ v's ip
ᶿ
ip v's
(a) (b) (c)
Fig. 3. (a) Equivalent circuit of conventional DAB converter. (b) Phasor representation for hard switching. (c) Phasor representation for soft switching.

S1, S4 S1 , S 4 S2, S3 S2, S3 S1 , S 4 S1, S4 S2 , S 3 S2, S3

Q2, Q3 Q1 , Q4 Q1, Q4 Q2, Q3 Q2 , Q3 Q1, Q4 Q1 , Q4 Q2, Q3

p p
v

v
ωt ωt
ᶿ ᶿ
s s
v'

v'
ωt ωt

x x
v

ωt ωt

p p
i

ωt ωt

(a) (b)
Fig. 4. Operating waveforms of the DAB converter based on LC-type series resonant tank (a) Hard switching (b) Soft switching.

for the primary side switches and ZVS/ZCS for the secondary
side switches. n4Vp
is (t) = (− cos ωt + M cos(ωt − θ)) − isci (t) (6)
πX
4Vp 1 The general description and working principle of the SCI
ip (0) = (−1 + M cos θ) < 0 ⇒ cos θ < (4)
πX M have been discussed in [26] and operating waveforms illustrat-
ing the use of SCI in the proposed topology are shown Fig. 7.
θ 4Vp The phase shift β between the voltage across the SCI (i.e.,
ip ( ) = (− cos θ + M ) ≥ 0 ⇒ θ ≥ cos−1 M (5) vs (t)) and the driving signal for Ta determines the effective
ω πX
On the primary side, the required condition can be met for inductance and hence the current isci (t) flowing through Lsci .
all M ≤ 1 by ensuring
nVs(max)
≤ 1, but from (5) it can Setting β = 90o gives the maximum isci (t) and minimum
Vp
effective effective inductance of Lsci and at β = 180o SCI
be seen that the secondary-side switches will experience hard
becomes inactive with isci (t) = 0 and Lsci → ∞.
switching for wide-range variations in M and θ.
1) Required Soft Switching Conditions: The required con-
dition to achieve ZVS at the primary side would remain the
B. Proposed Topology Based on SCI same whereas the condition for achieving ZVS/ZCS at the
The proposed topology of DAB converter and the corre- secondary side is given by (7).
sponding FCA-based equivalent circuit are shown in Fig. 5
and Fig. 6 respectively. It can be seen that SCI is inserted θ n4Vp ′
is ( ) = (− cos θ + M ) − Isci ≥ 0 (7)
at the secondary side to split the secondary-reflected current ω πX
′
nip (t) into is (t) and isci (t), and their relationship is given by where Isci is the peak value of the current flowing through
(6). SCI which occurs at the zero crossing of vs (t) (i.e., t = ωθ )

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 4

and is given by (8). converter can be evaluated from either side of the resonant tank
and the results are given by (15) and (16).
′ Vs
Isci = (β − π) (8)
2πf Lsci
Pint,p.u (t) = vp,p.u (t)ip,p.u (t)
where T is the switching period. −16
The required β to achieve ZVS/ZCS is given by (9) and the = 2 (sin ωt cos ωt − M sin ωt cos(ωt − θ))
π Xp.u
corresponding phasor diagrams are shown in Fig. 6.
(15)
8n2 f Lsci 8n2 f Lsci cos θ
β≤ − +π (9) ∫ 2π
X MX 1
Pavg,p.u = vp,p.u (t)ip,p.u (t)
Furthermore, it can be seen from Fig. 7 that the switches 2π 0
(16)
Ta and Tb are both turned on and turned off at zero current 8M
= 2 sin θ
[26], hence ZCS is always achieved in the SCI switches. The π Xp.u
initial fixed value of Lsci is obtained from (9) by keeping the
The rated per-unit output current Io,p.u can also be obtained
SCI’s firing angle β = 90o at the required minimum M and θ
from (16) by dividing the per-unit average output power by
to produce the required output current. More explanation will
the per-unit output voltage. It can be seen from (17) that Io,p.u
be given in IV.
is independent of the output voltage.
The selection of the nominal operating θ for the conven-
tional DAB converter lies in the range of 0 to 90o , but under 8
wide-range variations in output voltage and current there is Io,p.u = sin θ (17)
π 2 Xp.u
a need to operate with low conduction loss (i.e., minimum
In the proposed topology, the impedance Xp.u is fixed, M
rms value of the primary-side resonant tank current under the
is dependent on the output voltage and θ is used to control
specified range of M ). The selection of nominal operating
the output current. The per-unit rms tank current Ip,rms,p.u
point θ for reduced conduction loss is discussed in the next
expressed in terms of Io,p.u can be evaluated by substituting
section.
Xp.u from (17) into (13) and the result is given by (18).
III. D ESIGN OF N OMINAL O PERATING P OINT FOR
R EDUCED C ONDUCTION L OSS AT M AXIMUM O UTPUT Ip,rms,p.u π
= √ (− cos ωp tp + M cos(ωp tp − θ)) (18)
C URRENT Io,p.u 2 2 sin θ
In the following analysis, normalized variables (i.e., per-unit From (18), the normalized behavior of Ip,rms,p.u /Io,p.u with
values) are used to generalize the results for all power levels. respect to variations in M can now be analyzed and their
To perform FCA in normalized form, the following base values quantitative relationship is plotted in Fig. 8a using (18). Again
are used: VB = Vp , ZB = n2 (Vo /Io ) and IB = VB /ZB . (10)- the normalization ensures that the result and any conclusion
(12) give the normalized forms of (1)-(3) for a generalized that it leads to will hold for all power levels. It is apparent from
analysis. Fig. 8a that a desired nominal operating point with reduced
conduction loss can be chosen by identifying the value of θ
4
vp,p.u (t) = sin ωt (10) that gives the minimum average value of Ip,rms,p.u /Io,p.u over
π a given range of M , as shown in (19).
′ 4M
vs,p.u (t) = sin(ωt − θ) (11) ∫ Mmax
π Ip,rms,p.u 1 Ip,rms,p.u
( )avg = dM
Io,p.u Mmax − Mmin Mmin Io,p.u
4 (19)
ip,p.u (t) = (− cos ωt + M cos(ωt − θ)) (12)
πXp.u
where Xp.u = X/ZB . IV. S IMULATION AND E XPERIMENTAL R ESULTS
The rms value of the primary-side resonant tank current In order to verify the feasibility of the proposed idea, a
ip,p.u (t) is derived from (12) by setting its√first derivative to DAB converter based on SCI is designed to charge/discharge
zero and dividing the peak tank current by 2, and the result a super-capacitor with a variation of M from 0.208 to 0.83.
is given by (13). The nominal operating point with reduced conduction loss
4 over the given range of M can be obtained by plotting (19)
Ip,rms,p.u = √ (− cos ωp tp + M cos(ωp tp − θ)) (13) against different values of θ as shown in Fig. 8b and such
πXp.u 2
an operating point is identified to be that with θ = 60o or
where ωp tp is given by (14). 1.047 rad. Hence, θ = 60o is used to calculate the value of
( ) Xp.u required to deliver the maximum rated per-unit output
−M sin θ
ωp tp = tan−1 +π (14) current. From (17), the desired value of Xp.u is 0.7019 Ω
1 − M cos θ and using this value of Xp.u , the variation of per-unit output
The instantaneous and average output power being trans- current with respect to θ is plotted using (17) and is shown
ferred from the primary side to the secondary side of the DAB in Fig. 9a. It can be seen that the maximum value of Io,p.u

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 5

S 1 S 3 X Q 1 Q 3

L C n : 1 ni (t)p i (t)
s

V C C s
V o
i p
+ +
+
v (t)
p
i (t)
p v' (t)
s L i (t) v (t) s
sci sci
-
- - v (t)
S 2 S 4
ab

Q
+T a
Q 2 4

T b

Fig. 5. Proposed topology based on SCI.

vp vp
L C nip is

isci
vp vs ᶿ vx vx
is
Lsci ᶿ is ᶿ vs
vs
nip nip
isci isci
(a) (b) (c)
Fig. 6. (a) Equivalent circuit of the proposed DAB converter. (b) Phasor representation for ZCS (b) Phasor representation for ZVS.

T/2
T/2
ᶿ
T/2 ᶿ
ᶿ vs
vs vs ωt
ωt ωt β = 180o
o o
β = 90o 90 < β < 180

vTa vTa ωt
vTa ωt ωt
vTb v Tb ωt
vTb ωt ωt

isci isci isci

ωt ωt ωt

vab vab vab

ωt ωt ωt

(a) (b) (c)

Fig. 7. Operating waveforms of the SCI for its use in DAB converter. (a) Firing-angle β = 90o . (b) Firing-angle 90o < β < 180o . (c) Firing-angle
β = 180o .

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 6

(a) (b)
Fig. 8. (a) Ip,rms,p.u /Io,p.u versus M for different choices of nominal operating points θ. (b) Average values of Ip,rms,p.u /Io,p.u versus θ within given
range of M .

(a) (b)
Fig. 9. (a) Variation of output current Io,p.u with respect to change from nominal operating point θ. (b) Operating region of the proposed DAB converter’s
topology.

occurs at the selected nominal operating point and θ should be the LC series resonant tank is X = Xp.u ZB = 232.1 Ω. The
decreased to reduce the output current, which is usually needed value of Lsci is chosen to be 4.71 µH with M = 0.2083 (Vo =
when it is required to charge/discharge super-capacitor slowly 10 V) at the maximum rated output current of Io = 10 A and
or as its voltage approaches the rated maximum value (i.e., 48 β = 90o .
V). Fig. 9a also compares the output current calculated using The hard-switching region covered by the chosen Lsci is
(17) to that generated by circuit simulation, where the actual shown in Fig. 9b plotted using (9) and the Region A above
current waveform is considered. The close agreement between the green curve undergoes soft switching by naturally meeting
the two results implies that the error incurred by applying FCA the required condition of (5). The relatively small adjustable
is small and power is mainly transferred by the fundamental Region C (i.e., depends on Lsci ) under the gridded Region B
component of the tank current. corresponds to low output current Io at low voltage that is not
The specifications of the proposed topology are given in Ta- covered by the SCI designed above as super-capacitor does
ble I based on which the required base values are VB = 400 V, not typically operate in this region and ZCS can be triggered
ZB = n2 (Vo2 /Po ) = 330.67 Ω. The equivalent impedance of with the aid of the energy stored in the MOSFETs’ parasitic

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 7

(a) (b) (c)

Fig. 10. Simulation results. (a) Charging with Vp = 400 V, Vs = 14 V, Io = 9.75 A, M = 0.29, θ = β = 134.2o , Ip,rms = 1.36 A, 60o ,
Is,rms = 11.2 A and Isci,rms = 1.25 A. (b) Charging with Vp = 400 V, Vs = 30 V, Io = 4.83 A, M = 0.625, θ = 25.7o , β = 151.8o ,
Ip,rms = 0.77 A, Is,rms = 6.13 A and Isci,rms = 1.35 A. (c) Charging with Vp = 400 V, Vs = 45 V, Io = 2.72 A, M = 0.937, θ = 15.1o , β = 180o ,
Ip,rms = 0.36 A, Is,rms = 3 A, and Isci,rms = 0 A.

X Transformer Sensors

Primary Side
SCI
Secondary Side
Super-Capacitor

PSoC based controller

Fig. 11. Hardware realization of the proposed DAB converter topology.

inductances during hard-switching operation at low output for the secondary-side switches for the depicted values of M
voltages [9]. During hard switching at low voltages there will and Io . It can be seen in Fig. 10c that the SCI essentially
still be ringing due to reverse recovery effects but the losses becomes inactive and ip (t) becomes identical to is (t) as the
are relatively low because of small reverse recovery charge and DAB converter can achieve natural soft switching at increased
parasitic inductances limit the slew rate of the currents through values of M and Io by operating in Region A of Fig. 9b. Under
the secondary side MOSFETs as shown in Fig. 12 [9], [10]. this condition the proposed SCI-based DAB converter reverts
If needed the value of Lsci can be redesigned to extend the to the conventional LC-type series resonant DAB converter
soft-switching region to Region C at the expense of increased with the SCI’s firing-angle β approaching 180o , the effective
Isci and hence conduction loss in the SCI switches. inductance Lsci approaching ∞ and isci (t) tending to zero as
shown in Fig. 10c.
A. Simulation Results
The proposed DAB converter is simulated using PSIM and B. Experimental Results
the main operating waveforms are shown in Fig. 10. Fig. 10a The hardware prototype of the proposed SCI-based DAB
corresponds to the case of charging the super-capacitor at a converter topology with the specifications given by Table
voltage of 14 V with the maximum rated output current of 10 I is shown in Fig. 11. The primary-side or high-voltage-
A, while Fig. 10b and Fig. 10c represent the cases of charging side switches are realized by IGBTs, and MOSFETs are
with reduced output current as the super-capacitor’s voltage used for the secondary-side or low-voltage-side switches.
approaches its rated value of 48 V. It is evident that ZVS is Programmable System-on-Chip (PSoC) is used to generate the
achieved for the primary-side switches and ZCS is achieved required PWM signals for the switches and to adjust the firing

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 8

(a) (b)
Fig. 12. Experimental results for charging at Vp = 400 V, Vs = 14 V, Io = 9.30 A, M = 0.29 and θ = 60o . (a) Without SCI, efficiency = 77.9 %,
Ip,rms = 10.2 A, Is,rms = 10.1 A and Isci,rms = 0 A. (b) With SCI, β = 133.8o , efficiency = 78.5 %, Ip,rms = 10.3 A, Is,rms = 10.2 A,
Isci,rms = 0.35 A, and total power loss of SCI Ptot,sci = 0.5 W [27], [28].

vp(t) (2 μs/div, 250 V/div)

vp(t) (2 μs/div, 250 V/div)
ip(t) (2 μs/div, 20 A/div)
ip(t) (2 μs/div, 20 A/div)

vs(t) (2 μs/div, 50 V/div)

vs(t) (2 μs/div, 50 V/div)
is(t) (2 μs/div, 20 A/div) is(t) (2 μs/div, 20 A/div)

isci(t) (2 μs/div, 5 A/div) isci(t) (2 μs/div, 5 A/div)

(a) (b)
Fig. 13. Experimental results for charging at Vp = 400 V, Vs = 30 V, Io = 4.9 A, M = 0.625 and θ = (a) Without SCI, efficiency = 88.73 %, 26o .
Ip,rms = 6.03 A, Is,rms = 6.05 A and Isci,rms = 0 A. (b) With SCI, β = 150o , efficiency = 90.38 %, Ip,rms = 6.21 A, Is,rms = 5.96 A,
Isci,rms = 1.55 A, and total power loss of SCI Ptot,sci = 1.6 W [27], [28].

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 9

vp(t) (2 μs/div, 250 V/div) vp(t) (2 μs/div, 250 V/div)

ip(t) (2 μs/div, 10 A/div) ip(t) (2 μs/div, 10 A/div)

vs(t) (2 μs/div, 50 V/div) vs(t) (2 μs/div, 50 V/div)

is(t) (2 μs/div, 10 A/div) is(t) (2 μs/div, 10 A/div)

isci(t) (2 μs/div, 5 A/div) isci(t) (2 μs/div, 5 A/div)

(a) (b)
Fig. 14. Experimental results for charging at Vp = 400 V, Vs = 45 V, Io = 3.1 A, M = 0.937 and θ = 16.6o . (a) Without SCI, efficiency = 91.80 %,
Ip,rms = 3.40 A, Is,rms = 3.41 A and Isci,rms = 0 A. (b) With SCI, β = 175o , efficiency = 94.3 %, Ip,rms = 3.50 A, Is,rms = 3.47 A,
Isci,rms = 0.33 A, and total power loss of SCI Ptot,sci = 0.49 W [27], [28].

vp(t) (2 μs/div, 250 V/div) vp(t) (2 μs/div, 250 V/div)

ip(t) (2 μs/div, 10 A/div) ip(t) (2 μs/div, 10 A/div)

vs(t) (2 μs/div, 50 V/div) vs(t) (2 μs/div, 50 V/div)

is(t) (2 μs/div, 20 A/div) is(t) (2 μs/div, 20 A/div)

isci(t) (2 μs/div, 10 A/div) isci(t) (2 μs/div, 10 A/div)

(a) (b)
Fig. 15. Experimental results for discharging at Vp = 400 V, Vs = 35 V, Io = 4.0 A, M = 0.73 and θ = −22.0o . (a) Without SCI, efficiency = 90.25 %,
Ip,rms = 5.01 A, Is,rms = 5.02 A and Isci,rms = 0 A. (b) With SCI, β = 155o , efficiency = 91.57 %, Ip,rms = 5.07 A, Is,rms = 4.70 A,
Isci,rms = 1.59 A, and total power loss of SCI Ptot,sci = 1.63 W [27], [28].

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 10

vp(t) (2 μs/div, 250 V/div) vp(t) (2 μs/div, 250 V/div)

ip(t) (2 μs/div, 10 A/div) ip(t) (2 μs/div, 10 A/div)

vs(t) (2 μs/div, 50 V/div) vs(t) (2 μs/div, 50 V/div)

is(t) (2 μs/div, 10 A/div) is(t) (2 μs/div, 10 A/div)

isci(t) (2 μs/div, 5 A/div) isci(t) (2 μs/div, 5 A/div)

(a) (b)
Fig. 16. Experimental results for charging at Vp = 400 V, Vs = 35 V, Io = 4.2 A, M = 0.73 and θ = 20.2o . (a) ZCS, β = 161o , efficiency = 92.40 %,
Ip,rms = 5.04 A, Is,rms = 4.91 A and Isci,rms = 0.97 A (b) ZVS, β = 154o , efficiency = 90.80 %, Ip,rms = 5.04 A, Is,rms = 4.89 A,
Isci,rms = 1.44 A, and total power loss of SCI Ptot,sci = 1.5 W [27], [28].

Fig. 17. Measured efficiencies of DAB converter with and without the use of SCI at M and θ falling within the conventional hard-switching region (i.e.,
region B) shown in Fig. 9b.

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 11

TABLE I
S PECIFICATIONS OF THE C ONVERTER P ROTOTYPE

Input Voltage Vi 400 V

Output Voltage Vo 10-48 V
Transformer ratio n 8.3
Proposed Resonant Tank (L, C and Lsci ) (753 µH, 6.6 nF and 4.71 µH)
Switching Frequency fs 100 kHz
Maximum Rated Output Current Io 10 A
Maximum Rated Output Power Pavg,o 480 W
Primary Side IGBTS STGW19NC60WD (600 V, 19 A)
Secondary Side and SCI MOSFETs FDPF085N10A (100 V, 40 A, 6.5 mΩ)
Gate Drivers Infineon 2ED020I12-F2
Inductor’s Core Shape and Material FC3 Ferrite (ETD-44 Core)
Transformer’s Core Shape and Material FC3 Ferrite (ETD-44 Core)
Super-capacitor Maxwell Technologies (48 V, 83 F)
Controller 32-bit ARM Cortex-M3 PSoC 5LP by Cypress Semiconductor
Voltage and Current Sensors LEM Sensors (LV25-P and LA55-P)

angle of SCI based on the sensed output voltage and current. respectively, over wide-range variations in output current
The measured waveforms of vp (t), vs (t), ip (t), is (t), and and output voltage. The proposed topology utilizes SCI in
isci (t) with SCI and without SCI are shown in Fig. 12, Fig. 13 such a way that the required ZVS and ZCS conditions are
and Fig. 14. Fig. 12 (M = 0.29 and Io = 9.30 A) depicts met in the conventional hard-switching region. The nominal
the case in which the improvement in efficiency by using operating point is chosen to minimize conduction loss over
SCI is less significant since at relatively low output voltages a wide range of M for improving system’s efficiency. To
the energy stored in parasitic inductances of the secondary- check the feasibility and validity of the proposed topology
side MOSFETs are generally large enough to initiate partial and its design method, simulation and experimental results
ZCS even under the condition of hard switching [9]. A lower at different values of super-capacitor’s output voltages and
efficiency of 91.80 % (M = 0.937 and Io = 3.1 A) is achieved output charging/discharging currents are presented, where a
without the use of SCI as compared to 94.3 % with the use of maximum efficiency of 94.6 % is obtained.
SCI and is shown in Fig. 14. As shown in Fig. 13 (M = 0.625
and Io = 4.9 A) a higher efficiency of 90.38 % is obtained ACKNOWLEDGMENT
due to the use of SCI and 88.73 % without the use of it. This research work was supported by the University Grants
Discharging of the super-capacitor with and without the use Committee of the Hong Kong Special Administrative Region,
of SCI is given in Fig. 15, which shows the effectiveness of Research Grants Council, under Early Career Scheme (ECS)
the proposed topology for reverse power flow as well. PolyU 5390/13E.
The measured waveforms show that ZVS and ZCS are
achieved for the primary-side and secondary-side switches, R EFERENCES
respectively. The reason for avoiding ZVS at the secondary-
[1] M. Liserre, T. Sauter and J. Hung, ”Future energy systems: Integrating
side switches is to reduce circulating current and parasitic- renewable energy sources into the smart power grid through industrial
induced ringing as shown in Fig. 16. The minor difference electronics,” IEEE Mag. Industrial Electronics, vol. 4, no. 1, pp. 18–37,
between the simulated and measured output current Io is due to 2010.
[2] J. Lai and D. J. Nelson, ”Energy management power converters in hybrid
the conduction loss caused by the parasitic resistances of semi- electric and fuel cells vehicles,” in Proceedings of the IEEE, vol. 95,
conductor devices, copper loss, and phase-drift phenomenon no. 4, pp. 766–777, 2007.
caused by the dead-time between two complementary switches [3] R. W. De Doncker, D. M. Divan and M. H. Kheraluwala, ”A three-
phase soft-switched high-power-density DC/DC converter for high-
in the bridge legs. Converter efficiencies with and without power applications,” IEEE Trans. Industry Applications, vol. 27, no. 1,
the use of SCI for the values of M and θ that fall into the pp. 63–73, 1991.
conventional hard-switching region are plotted in Fig. 17. The [4] S. Inoue and H. Akagi, ”A Bidirectional Isolated DCDC Converter as a
Core Circuit of the Next-Generation Medium-Voltage Power Conversion
increase in efficiency with increasing M can be explained System,” IEEE Trans. Power Electronics, vol. 22, no. 2, pp. 535–542,
by the fact that, at lower M , and hence at lower output 2007.
power (since the output current is constant and depends on [5] S. Inoue and H. Akagi, ”A Bidirectional DCDC Converter for an
Energy Storage System With Galvanic Isolation,” IEEE Trans. Power
θ only), the conduction loss due to the parasitic resistances Electronics, vol. 22, no. 6, pp. 2299–2306, 2007.
of semiconductor devices and PCB traces are constant for a [6] B. Zhao, Q. Song, W. Liu, and Y. Sun, ”Overview of dual-active bridge
given output current and poses more weight in efficiency as isolated bidirectional dcdc converter for high-frequency-link power-
conversion system,” IEEE Trans. Power Electronics, vol. 29, no. 8,
compared to the cases of higher M . pp. 4091–4106, 2014.
[7] B. Zhao, Q. Song, W. Liu, and Y. Sun, ”A Synthetic Discrete De-
V. C ONCLUSION sign Methodology of High-Frequency Isolated Bidirectional DC/DC
Converter for Grid-Connected Battery Energy Storage System Using
DAB converter operating at high frequency can experience Advanced Components,” IEEE Trans. Power Electronics, vol. 61, no. 10,
significant switching loss due to hard switching. A modified pp. 5402–5410, 2014.
[8] X. Li and A. K. S. Bhat, ”Analysis and Design of High-Frequency
topology of DAB converter is proposed to achieve ZVS Isolated Dual-Bridge Series Resonant DC/DC Converter,” IEEE Trans.
and ZCS for the primary-side and secondary-side switches, Power Electronics, vol. 25, no. 4, pp. 850–862, 2010.

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2654505, IEEE
Transactions on Power Electronics

IEEE TRANSACTIONS ON POWER ELECTRONICS 12

[9] F. Krismer and J. W. Kolar, ”Accurate Power Loss Model Derivation of M. Yaqoob (S’16) received his B.Eng. degree in
a High-Current Dual Active Bridge Converter for an Automotive Appli- Electronics Engineering from the National Univer-
cation,” IEEE Trans. Industrial Electronics, Vol. 57, No. 3, pp. 881–891, sity of Sciences and Technology, Islamabad, Pak-
2010. istan in 2012 and the M.Sc. degree in Electric Power
[10] F. Krismer, S. Round and J. W. Kolar, ”Performance Optimization of System and Its Automation from the North China
a High Current Dual Active Bridge with a Wide Operating Voltage Electric Power University, Beijing, China, in 2014.
Range,” Proc. of the 37th IEEE Annual Power Electronics Specialists He is currently working toward the Ph.D. degree in
Conference(PESC 2006), Jeju, South Korea, June 2006, pp. 1–7. the area of power electronics from The Hong Kong
[11] L. Xue, M. Mu, D. Boroyevich and P. Mattavelli, ”The optimal design of Polytechnic University, Hong Kong. His current re-
GaN-based Dual Active Bridge for bi-directional Plug-IN Hybrid Elec- search interests include high frequency medium volt-
tric Vehicle (PHEV) charger,” in Proc. IEEE Applied Power Electronics age power conversion systems, bidirectional isolated
Conf., 2015, pp. 602–608. DC/DC converters and advanced soft-switching techniques.
[12] L. Xue, D. Boroyevich and P. Mattavelli, ”Switching condition and loss Mr. Yaqoob serves as a reviewer of IEEE Transactions on Power Elec-
modeling of GaN-based dual active bridge converter for PHEV charger,” tronics, IEEE Transactions on Control Systems Technology and IET Power
in Proc. IEEE Applied Power Electronics Conf., 2016 pp. 1315–1322. Electronics.
[13] B. Zhao,Q. Yu, and W. Sun, ”Extended-phase-shift control of isolated
bidirectional dcdc converter for power distribution in micro-grid,” IEEE
Trans. Power Electronics, Vol. 27, No. 11, pp. 4667–4680, 2012.
[14] G. G. Oggier, G. O. Garcia and A. R. Oliva, ”Switching control strategy
to minimize dual active bridge converter losses,” IEEE Trans. Power
Electronics, vol. 24, no. 7, pp. 1826–1838, 2009.
[15] G. G. Oggier, G. O. Garcia and A. R. Oliva, ”Modulation strategy to
operate the dual active bridge DC-DC converter under soft switching
in the whole operating range,” IEEE Trans. Power Electronics, vol. 26,
no. 4, pp. 1228 - 1236, 2011.
[16] N. D. Dinh, N. D. Tuyen, G. Fujita and M. N. B. Muhtazaruddin,
”Investigation of ZVS condition for dual-active-bridge converter using
dual-phase-shift modulation,” in Proc. IEEE PES Asia-Pacific Power and
Energy Engineering Conf., 2014, pp. 1–6.
[17] H. Bai and C. Mi, ”Eliminate Reactive Power and Increase System Ef-
ficiency of Isolated Bidirectional Dual-Active-Bridge DCDC Converters
Using Novel Dual-Phase-Shift Control,” IEEE Trans. Power Electronics,
Vol. 23, No. 6, pp. 2905–2914, 2008.
[18] B. Zhao,Q. Yu, and W. Sun, ”Power Characterization of Isolated
Bidirectional Dual-Active-Bridge DCDC Converter With Dual-Phase- K. H. Loo (S’97–M’99) received the B.Eng. (Hons.)
Shift Control,” IEEE Trans. Power Electronics, Vol. 27, No. 9, pp. 4172– in electronic engineering and the Ph.D. degree from
4176, 2012. the University of Sheffield, U.K., in 1999 and 2002,
[19] B. Zhao,Q. Yu, and W. Sun, ”Efficiency Characterization and Optimiza- respectively.
tion of Isolated Bidirectional DCDC Converter Based on Dual-Phase- From 2002 to 2004, Dr. Loo was the Japan Society
Shift Control for DC Distribution Application,” IEEE Trans. Power for the Promotion of Science (JSPS) Postdoctoral
Electronics, Vol. 28, No. 4, pp. 1711–1727, 2013. Fellow at the Ehime University, Japan. In 2006,
[20] F. Krismer and J. W. Kolar, ”Efficiency-Optimized High-Current Dual he joined the Hong Kong Polytechnic University
Active Bridge Converter for Automotive Applications,” IEEE Trans. as an Instructor in the Faculty of Engineering. He
Industrial Electronics, vol. 59, no. 7, pp. 2745–2760, 2012. is now Assistant Professor in the Department of
[21] F. Krismer and J. W. Kolar, ”Closed Form Solution for Minimum Electronic and Information Engineering of the Hong
Conduction Loss Modulation of DAB Converters,” IEEE Trans. Power Kong Polytechnic University. He has been an Associate Editor for the IEEE
Electronics, vol. 27, no. 1, pp. 174–188, 2012. Transactions on Energy Conversion since 2013. His areas of research interest
[22] L. Corradini, D. Seltzer, D. Bloomquist, R. Zane, D. Maksimovi and include power electronics for LED lighting and renewable energy systems.
B. Jacobson, ”Minimum Current Operation of Bidirectional Dual-Bridge
Series Resonant DC/DC Converters,” IEEE Trans. Power Electronics,
vol. 27, no. 7, pp. 3266–3276, 2012
[23] W. Choi, K. M. Rho and B. H. Cho, ”Fundamental Duty Modulation of
Dual-Active-Bridge Converter for Wide-Range Operation,” IEEE Trans.
Power Electronics, vol. 31, no. 6, pp. 4048–4064, 2016.
[24] B. Zhao, Q. Song, W. Liu, and Y. Sun, ”Universal High-Frequency-
Link Characterization and Practical Fundamental-Optimal Strategy for
Dual-Active-Bridge DC-DC Converter Under PWM Plus Phase-Shift
Control,” IEEE Trans. Power Electronics, vol. 30, no. 12, pp. 6488–
6494, 2015.
[25] T. Hirose, M. Takasaki and Y. Ishizuka, ”A Power Efficiency Improve-
ment Technique for a Bidirectional Dual Active Bridge DCDC Converter
at Light Load,” IEEE Trans. Industry Applications, vol. 50, no. 6,
pp. 4047–4055, 2014.
[26] W. J. Gu and K. Harada, ”A new method to regulate resonant converters,”
IEEE Trans. Industry Applications, Vol. 3, No. 4, pp. 430–439, 1988.
[27] D. D. Graovac, M. Purschel and A. Kiep. (2012, Dec). MOSFET power
losses calculation using the data-sheet parameters. [Online]. Available: Y. M. Lai (M’92) received the B.Eng. degree in
http://application-notes.digchip.com/070/70-41484.pdf electrical engineering from the University of Western
[28] R. Erickson and D. Maksimovic, ”Transformer Design,” in Funda- Australia, Perth, Australia, in 1983, the M.Eng.Sc.
mentals of Power Electronics, 2nd ed. New York: Kluwer Academic degree in electrical engineering from University of
Publishers, 2004, pp. 582–583. Sydney, Sydney, Australia, in 1986, and the Ph.D.
degree from Brunel University, London, U.K., in
1997.
He is an Associate Professor with Hong Kong
Polytechnic University, Hong Kong, and his research
interests include computer-aided design of power
electronics and non-linear dynamics.

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.