By Haoyu Wang, Amin Hasanzadeh,

and Alireza Khaligh

he U.S. transportation sector

T
consumes approximately 14 million
barrels of petroleum every day,
which is more than the total oil con-
sumption of any other nation in the
world. The most prominent sustainable solution to
profoundly reduce both oil consumption and
greenhouse gas emissions lies in grid-enabled elec-
tric vehicles (EVs) These vehicles are propelled
either partially or fully by electricity through energy
storage systems such as electrochemical batteries,
which need to be charged from the grid.
One of the most important realities that will facili-
tate the adoption of grid-enabled plug-in EVs (PEVs) is
the method by which these vehicles will be charged
and if the nation’s electrical infrastructure can sustain
the charging requirements of PEVs without the
need for a comprehensive evolution. Since nearly
all residences and businesses in the United States
are already equipped with a 120-Vac mains con-
nection, and most with a 240-Vac connection,
most charging will be done through onboard level 1
(120 Vac/16 A/1.92 kWMax) and level 2 (240 Vac/
80 A/19.2 kWMax) chargers when the vehicle is parked
either overnight at home or during the day at the
office. Even though onboard level 1 and level 2 charg-
ing would be the most prominent charging paradigm

Transportation
Electrification
Conductive charging of electrified vehicles.

Digital Object Identifier 10.1109/MELE.2013.2294238

Date of publication: 26 February 2014

46 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2013 2325-5987/13/$31.00©2013IEEE

 equipment (EVSE) is installed. Typically, the
EVSE includes conductors; EV connectors;
attachment plugs; communication proto-
cols; and all other fittings, devices, and
power outlets and/or apparatuses. The EVSE
serves as the energy transfer interface
between the premises’ wirings and the
PEVs’ onboard or off-board battery chargers.
Depending on the power level and the
required charging facilities, the charging of
EVs and plug-in hybrid EVs (PHEVs), cumu-
latively called PEVs, can be classified into
three different levels. The corresponding
power supplies, charging power levels,
EVSE costs, all-electric mile ranges for 1 h
of charge, and charging times are summa-
rized in Table 1.
In the United States, level 1 charging is
adapted for a single-phase, 120-V/16-A, 60-Hz
grid outlet. A level 1 charger is easy to inte-
grate on board. The installation and EVSE
cost for level 1 charging is estimated to be
US$500–800. The relatively low price makes
level 1 applicable to home charging. A major-
ity of the available EVs and PHEVs are
equipped with an onboard charger (OBC),
which is compatible with level 1 charging.
However, the low charging power increases
the charging time of the battery pack (up to
17 h) to charge a typical 25-kWh battery pack
from 20% state of charge (SoC) to full SoC.
Level 2 charging requires a single-phase
240-V power outlet, which is available in
most residential garages as well as public
facilities. Using the 6.6-kW OBC, it takes
3.5 h to charge a typical 25-kWh battery
for the next generation of PEVs, there is a crucial need to pack from 20% SoC to full SoC. In comparison to level 1,
d ep l oy h i g h - p owe r o f f - b o a rd l eve l 3 ( 2 0 8 – without compromising the convenience of accessibility,
600 Vac/400 A/240 kWMax) chargers. This would not only the level 2 charging time is much less. The installation and
eliminate the “range anxiety” phenomenon among drivers EVSE cost of level 2 charging is expected to be US$3,150–
but also enable long trips of considerable distance without 5,100. Although it is much higher than level 1 charging, it
substantial vehicle downtime, as level is still affordable to consumers and
3 chargers would serve a similar func- small businesses. Consequently, level
tion for PEVs as gas stations do for con- 2 charging is expected to be the
ventional internal combustion engine For high-power dominant charging method for pri-
(ICE) vehicles. This article focuses on vate and public facilities. Since Feb-
the conductive onboard and off-board onboard charging ruary 2011, five charging s­ tations
battery charging of PEVs.
applications, CCM is have been installed at the University
of Maryland, College Park. Each
Charging Levels and preferable because charging station is equipped with
Infrastructures one 120-V port for level 1 charging
A diagram of a typical PEV energy of its low peak and one 240-V port for level 2 charg-
transfer system is shown in Figure 1. A current and low ing. All of these stations are free and
low-frequency transformer is used to open to the public. Figure 2 shows a
deliver ac power from the grid to the current ratings. Tesla Model S being charged at one
neighborhood where the EV supply of these stations.

IEEE Elec trific ation Magazine / d ec emb er 201 3 47

 Electric
Utility
Power
System

Service Drop

Building EV Energy Transfer System

EV
Service EVSE Power Electronics and
dc Charge
Equipment Motor Controller
Meter Port
Electric Drive
ac Charge
Meter Branch Circuit Motor
Port
Grid Transformer
OBC
Battery Pack with Battery
Managing System
ac Electrical Energy dc Electrical Energy

Figure 1. A diagram of an EV energy transfer system (modified from Young et al.).

Level 3 fast charging extends the charging power to a market as of December 2013. All listed PEVs are equipped
much higher level (excess of 50 kW). Consequently, the with OBCs compatible with level 1 and level 2 charging. All
charging time is significantly reduced. The level 3 charging PEVs, except for the Tesla Model S, use a universal charge
stations are expected to charge a PEV battery pack to 80% connector, which is defined by the standard SAE J1772.
SoC in fewer than 30 min. One of the frontier EV manufac-
turers, Tesla Motors Inc., is aimed at fully charging its EVs in Battery Charging Strategies
5 min in the foreseeable future using its supercharging A Li-ion cell has a higher energy density than other bat-
­stations. To implement level 3 fast charging, an off-board tery chemistries such as lead acid cell, nickel cadmium
charger is necessary to convert three-phase 208 V–600-Vac cell, and Ni-metal hydride cell. In PEVs, the energy densi-
power to 200 V–450-Vdc power, which fits the voltage range ty and the weight of the battery are two of the most criti-
of the battery pack. However, level 3 charging comes with cal parameters that determine the electric range of the
extremely high costs, which include the installation, infra- vehicle. ­Consequently, the Li-ion cell has dominated the
structure, and maintenance costs. It should be noted that market of commercially available PEVs. This can be
the rapid charging of a battery pack can cause it to overheat observed in Table 2, as all the listed PEVs are equipped
and potentially decrease the battery life. Moreover, drawing with an Li-ion battery pack. Although extended life cycles,
ultrahigh power from the grid increases the demand from increased energy density, and a slight cost reduction have
the grid and might cause an overload of the local distribu- been achieved with the evolution of battery technology,
tion facilities. Consequently, level 3 charging is mainly the Li-ion battery pack is still the most expensive and
intended for commercial and public charging stations. heaviest component of a PEV.
Table 2 lists the charging characteristics and infrastruc- It is not only the battery chemistry that determines the
tures of some of the commercially available PEVs on the power level at which a cell can accept a charge but also

TABLE 1. Charging power levels.

Charging Power Supply Power Limit Installation and Typical Charging Range for 1 h of Charging Time
Level (kW) EVSE Cost Power (kW) Charge (mi)
EV PHEV
Level 1 120-Vac 1.92 US$500–800 1.4 3–4 ~17 h ~7 h
single phase
Level 2 240-Vac 19.2 US$3,150–5,100 3.3 (onboard) 8–10 ~7 h ~3 h
single phase 6.6 (onboard) 17–20 ~3.5 h ~1.4 h
Level 3 Three-phase 240 US$30,000– >50 50–60 30–45 ~10
208–600 16,000 (off-board) min min
Vac or dc

48 I E E E E l e c t ri f i c a t i on M a gaz ine / december 2013

the method used to charge the battery. Various methods
can be adopted to safely charge Li-ion batteries. The meth-
od shown in Figure 3(a) is called constant current–­
constant voltage (CC-CV), which is adopted by a majority
of the onboard battery chargers. The rate at which a bat-
tery is discharged relative to its maximum capacity is
known as the C-rate. The basic idea behind this technique
is that the battery is charged with a constant current, typi-
cally rated at 0.5–1 C. When the battery voltage reaches its
cutoff voltage, its voltage is kept fixed and it is charged at
this constant voltage until the current draw decreases to
the end charging current, e.g., 0.02 C, which represents a
full charge. For a deeply depleted battery pack, the b ­ attery Figure 2. A Tesla Model S being charged at the University of Mary-
land, College Park.
voltage might go down to 1 V/cell. In this case, a precharge
stage is required to charge the battery voltage from 1 V/
cell to 2.5 V/cell. In the precharge stage, the charging cur- techniques. A simplified example of the CC-CV method
rent is fixed at a constant rate (typically 0.1 C). with negative pulses is shown in Figure 4(a). Another
To increase the charge acceptance rate of the battery, approach uses a variable pulse charge strategy. In this
multistage CC–CV (MCC-CV) has been proposed. The prin- approach, the optimal pulse charge frequency is continu-
ciple is almost the same as CC-CV. However, instead of ously varied to distribute ions in the electrolyte evenly.
providing the battery with one constant current level, Between pulses, a variable rest period is applied to neu-
­several current steps are applied up to the cutoff voltage tralize and diffuse the ions. This rest period is defined by a
as shown in Figure 3(b). maximum power point tracker to determine the maxi-
In addition, a few other charging methods with less mum level of current acceptance for a given SoC in real
impact on battery polarization and higher charge accep- time. A typical waveform of the variable frequency associ-
tance rates are being investigated. One approach ated with pulse charging is shown in Figure 4(b). Using
­discharges the battery at specific time intervals during this method, the charge rate can be increased in compari-
charging to increase the charging acceptance. This son to conventional CC-CV and fixed-frequency pulse
approach can be applied to both CC-CV and MCC-CV charging methods.

TABLE 2. The charging characteristics and infrastructures of some of the commercially available
PEVs.
Vehicle EV Type Price Battery OBC (kW) E-Range (mi) Connector Type Charging Time (h)

Level 1 Level 2
Nissan EV US$35,200 24-kWh 3.3 100 SAE J1772 22 8
Leaf Li-ion JARI/TEPCO
BMW EV Lease only 32-kWh 7.2 100 SAE J1772 8–10 4–5
ActiveE Li-ion
Ford Focus EV US$39,200 23-kWh 6.6 76 SAE J1772 20 3–4
Li-ion
EV US$29,125 16-kWh 3.3 62 SAE J1772 22.5 7
Mitsubishi I
Li-ion JARI/TEPCO
Honda Fit EV Lease only 20-kWh 3.3 76 SAE J1772 6 3
Li-ion
Toyota PHEV US$32,000 4.4-kWh 3.3 15 SAE J1772 3 1.5
Prius Li-ion
Chevy Volt PHEV US$39,145 16-kWh 3.3 35 SAE J1772 10 4
Li-ion
Cadillac PHEV n/a 16.5-kWh 3.3 35 SAE J1772 n/a 4.5
ELR Li-ion
Tesla EV US$95,400 85-kWh 10 265 Mobile connec- 34 14
Model S Li-ion tor

Note: Specification data are based on public information and are subject to change.

IEEE Electrific ation Magazine / d ec emb er 201 3 49

 I I I f
CC CV

CC CV MCC CV

t t
t t
(a) (b)

Figure 3. The Li-ion battery charging techniques: (a) CC-CV and (b)
(a) (b)
MCC-CV.

Figure 4. The advanced fast-charging techniques: (a) CC-CV with a

Isolated Onboard PEV Chargers ­ egative pulse and (b) variable frequency pulse charge (Khaligh and
n
Dusmez).
The typical power architecture of an OBC is shown in
­Figure 5. Typically, an isolated OBC consists of two stages:
1) the first stage for ac/dc conversion and power factor boost PFC converter, where a full-bridge diode rectifier is
correction (PFC) and 2) the second stage for dc/dc conver- followed by a boost converter. PFC is achieved by controlling
sion and galvanic isolation. the duty cycle of the MOSFET to shape the inductor current
The first-stage ac/dc PFC converter typically consists of to be sinusoidal and in phase with the grid voltage. A sin-
an electromagnetic interference (EMI) input filter, rectifier, gle-phase PFC performs well for level 1 charging. However,
PFC converter, and dc link capacitor. for level 2 charging, the inductor
The PFC converter is controlled by a becomes bulky and the components’
high-frequency signal to regulate the current stress becomes high. An inter-
ac line current to follow the ac line Bidirectional power leaved boost PFC converter is prefera-
voltage and frequency. Ideally, the flow between the ble in level 2 chargers. Figure 6(b)
ac/dc PFC stage should be equivalent provides the schematic of a two-phase
to a resistive load to eliminate the grid and the vehicle interleaved boost PFC converter, whose
total harmonic distortion (THD) and interleaving legs are operated with 180°
maximize the power transfer.
has gained interest phase difference. The interleaved boost
Boost and its derivative topologies from academia and PFC converter has less current stress in
are commonly used in the PFC stage. each individual leg. With the ripple
This is because of their simple circuit industry. cancellation effect, both the input cur-
configurations, continuous input cur- rent ripple and output capacitor root
rent, and low THD. To be compatible mean square current can be reduced.
with universal grid voltages (85–265 V, The typical experimental waveforms of
47–70 Hz), typically the output voltage of the boost-type PFC an interleaved boost PFC stage design by the Power Elec-
stage is regulated at 390 V. There are three operation modes tronics Energy Harvesting and Renewable Energies Labora-
for boost-type PFC converters: 1) continuous conduction tory (PEHREL) at the University of Maryland are
mode (CCM), 2) discontinuous conduction mode, and 3) demonstrated in Figure 7. In comparison to the current rip-
boundary conduction mode (BCM). For high-power onboard ples in each individual inductor, the current ripple in the
charging applications, CCM is preferable because of its low input side is significantly reduced.
peak current and low current ratings. As an alternative, the diode bridge and boost converter
Six commonly used boost-type PFC stages are shown in can be integrated into one stage. Figure 6(c) shows a bridge-
Figure 6. Figure 6(a) shows the conventional s­ ingle-phase less boost PFC converter, where S1 and S2 are controlled
with the same gate signal. Figure 6(d)
is a bridgeless version of a two-phase
interleaved boost PFC converter. In
Onboard Figure 6(d), S1 and S2 are controlled
1Φ Grid ac/dc Isolated
PFC dc/dc Battery with one gate signal and S3 and S4 are
Input
Converter Converter Pack controlled with a second gate signal.
Filter
The two channels are shifted with a
180° phase difference. Using the
OBC bridgeless ­structure, the input diode
bridge can be eliminated. Fewer semi-
Figure 5. A block diagram of an isolated OBC. conductor devices mean less power

50 I E E E E l e c t ri f i c a t i on M a gaz ine / december 2013

 L1 D5 L1 D5

D1 D3 D1 D3 L2 D6
Cdc
Grid Load Grid Load
S1 Cdc
S1 S2
D2 D4 D2 D4

(a) (b)

L1 D1 D2 D3 D4
L1 D1 D2
Grid Cdc Grid
Cdc
Load L2 Load

S1 S2 S1 S2 S3 S4

(c) (d)
L1 D5

D1 D3 S1 D1 D2 S S3
Cdc1 L1 1
Cdc
Grid Load Grid L2 Load

S2 Cdc2 D2 D4 S2 S4
D2 D4

(e) (f)

Figure 6. The unidirectional ac/dc PFC stages: (a) a single-phase boost PFC converter, (b) a two-phase interleaved boost PFC converter,
(c) a bridgeless boost PFC converter, (d) a bridgeless interleaved boost PFC converter, (e) a multilevel boost PFC converter, and (f) a bridgeless
interleaved resonant boost PFC converter.

loss and higher conversion efficiency.

However, the EMI is higher for bridge-
less boost PFC converters and, thus, a vdc
bigger EMI filter is required.
Figure 6(e) shows a multilevel
boost PFC converter. Because of its iin
two-­level structure, the voltage stress 1
on each MOSFET is reduced to half.
Multilevel configuration has advan-
tages such as high power quality, iL1
reduced THD, and reduced EMI noise.
Moreover, its dc output voltage is 1
insensitive to load and supply distur-
bances. Figure 6(f) shows a bridgeless 2
interleaved resonant boost PFC con-
4 iL2
verter. This topology is proposed to
operate in BCM in level 2 onboard
charging applications. Diodes D1–D4
3
are operated with grid frequency.
1 10.0 A 2 500 mV 3 100 mV 4 500 mV
Thus, no fast diodes are required.
However, it has disadvantages in
Figure 7. The experimental waveforms of an interleaved boost PFC stage designed by PEHREL
terms of additional high side drives at the University of Maryland. From top to bottom: vdc (50 V/div), iin (10 A/div), iL1 (5 A/div), iL2
and a complicated control scheme. (5 A/div); time 10 ms/div.

IEEE Electrific ation Magazine / d ec emb er 201 3 51

 In high switching frequency appli-
Tek PreVu cations, MOSFETs are preferred
because of their fast switching speed
and no tail current. In hard switching
vgs3 vds3 topologies, higher switching frequen-
10 V/div 200 V/div cies would lead to high stress and
high EMI noise. Thus, soft switching
techniques, which include zero volt-
2 age switching (ZVS) and zero current
switching (ZCS), are desired. For MOS-
FETs, ZVS is more suitable because
vds4 vgs4 operation with ZVS eliminates both
200 V/div 10 V/div body diode reverse recovery and
semiconductor output capacitances
from inducing switching loss in MOS-
3 FETs. Figure 8 provides the ZVS wave-
t :2 µs/div forms of power MOSFETs, wherein
the drain source voltage (vds) decreas-
Figure 8. The experimental ZVS waveforms of a full-bridge inductor inductor capacitor (LLC) es to zero before the gate source volt-
stage designed by PEHREL at the University of Maryland. (Horizontal axis–time: 2 μs/div.) age (vgs) begins to rise.
Six commonly used isolated ZVS
A typical second-stage isolated dc/dc converter ­consists dc/dc topologies are summarized in Figure 9. Figure 9(a) is a
of a switching network, a high-frequency transformer, a full-bridge phase shift ZVS converter. S1 and S2 are driven
rectifier, and a low-pass filter. For frequency-modulated by two complementary 50% duty cycle gate signals. S3 and
­resonant converters, an additional resonant tank between S4 are also driven by two complementary 50% duty cycle
switching network and high frequency is required. gate signals. There is an adjustable phase difference

Lr n:1 Lo Lr n:1 Lo
S1 S3 S1 S3
D1 D3 Co Vbat + D1 D3 +
Vdc
+
Vdc Da C Vbat
+ o2
– Co1
–
S2 S4 D2 D4 – S2 S4 D2 D4 –

(a) (b)

Lr n:1 Lr n:1
S1 S3 S1 S3
D1 D3 + D1 D3 Vbat +
Vdc
+ Co Vbat Vdc + Lm Co
– –
S2 S4 D2 D4 – S2 S4 D2 D4 –
Cr
(c) (d)

n:1 Lo
Lr 1
S1 S3 D1 D3 Vbat + Lr 1 n1:1 D1 D3 Ch
Vdc + Co1 Dr
S1 S3
– S2 S4 Co2 Vbat +
D4C – Vdc Do
D2 o2 +
– Lr 2 n2:1 D5 D7
Lr 2 n:1 –
S5 S7 Co1
S2 S4 Cr
S6 S8
D2 D4 D6 D8

(e) (f)

Figure 9. The unidirectional isolated ZVS dc/dc converters in PEV battery charging applications: (a) a full-bridge phase-shift ZVS converter,
(b) a full-bridge trailing-edge ZVS converter, (c) a full-bridge ZVS converter with a capacitive output filter, (d) a full-bridge LLC resonant converter,
(e) a full-bridge interleaved ZVS converter with a voltage doubler, and (f) a full-bridge ZVS PWM resonant converter.

52 I E E E E l e c t ri f i c a t i on M a gaz ine / december 2013

between the gate signals of S1 and S2 and the gate signals
of S3 and S4. By controlling this phase difference, the output 4.0
420 V
voltage can be regulated. The advantages lie in its pulse-

Normalized Gain (nVo /Vdc)

3.5 fmin fp fmax
width modulation (PWM) operation and wide output volt-
3.0

Battery Voltage
age range. However, in light load conditions, two MOSFETs
in the lagging leg lose ZVS features. In addition, because of 2.5
high di/dt, there are reverse recovery losses in rectifier 2.0
ZVS ZVS
diodes. F
­ igure 9(b) is a full-bridge trailing-edge ZVS convert- Region 1
1.5 Region 2
er, which is a derivative of the phase-shift ZVS converter.
The difference between a full-bridge trailing-edge ZVS con- 1.0 100 V
ZCS Region
verter and a phase-shift ZVS converter mainly lies in the 0.5
0.25 0.5 0.75 1 1.25 1.50 1.75
switch pattern. S2 and S4 are driven by two complementary
Normalized Frequency (f /fp)
50% duty cycle gate signals, whereas S1 and S3 are driven by
two 180° phase difference and adjustable duty cycle gate
Figure 10. The dc voltage characteristics of an LLC converter.
signals. By controlling this duty cycle, the output voltage
can be regulated. The features of the full-bridge trailing-
edge ZVS converter are similar to those of the full-bridge
phase shifted ZVS converter. An additional clamp network turned on with ZVS and two MOSFETs are always turned
consisting of DC, RC, and CC is needed to clamp the ­voltage on with ZCS. The rectifier diodes are turned off with low
ringing due to diode junction capacitance with the leakage di/dt so that the reverse recovery losses can be minimized.
inductance of the transformer. The disadvantage mainly lies in the MOSFETs’ high peak
Figure 9(c) shows a full-bridge ZVS converter with a currents. Figure 9(e) shows the interleaved version of
capacitive output filter. The switching pattern is the same ­Figure 9(c), which offers benefits such as reduced current
as that of the trailing-edge ZVS converter. However, there ripple and reduced filter size. However, the number of gate
is no clamp network, and the output filter consists of a ­drivers, MOSFETs, and transformers are doubled, and the
capacitor. In this topology, two MOSFETs are always control scheme is more complicated.

S1 S1 S2
Grid Lo Grid ac/dc Lo1
ac/dc
PFC PFC
Co Vbat + Vbat +
Converter Converter Lo2 Co
D1
D1 D2
– –

(a) (b)

D5 L1 D6

D1 D3 S1 D1 D3 S1
– Co +
Grid L1 Grid S2 Vbat
D5
Co Vbat
+ –
D2 D4 D2 D4

(c) (d)

L1 C1 L1 C1

D1 D3 D1 D3
+ Co –
Grid S1 L2 Co Vbat Grid S1 Vbat
L2
– +
D2 D4 D2 D4

(e) (f)

Figure 11. The unidirectional nonisolated PEV battery chargers: (a) a two-stage buck, (b) a two-stage interleaved buck, (c) a buck–boost PFC,
(d) a noninverting buck–boost PFC, (e) an SEPIC PFC, and (f) a Cúk PFC.

IEEE Electrific ation Magazine / d ec emb er 201 3 53

 L1 C1 S1 S1 S3
Grid Grid L1
Cdc Cdc

Load Load
C2 S2 S2 S4

(a) (b)

Figure 12. The bidirectional ac/dc PFC stages: (a) a half-bridge bidirectional boost PFC and (b) a full-bridge bidirectional boost PFC.

Figure 9(d) shows a full-bridge LLC (inductor inductor convert the universal grid input to a fixed dc link voltage,
capacitor) resonant converter. S1 and S4 and S2 and S3 are which is higher than the maximum battery voltage. A buck
turned on and off complementarily with a dead band. Thus, converter is also used to step down the dc link voltage. In
the output of the full bridge is a square wave and is fed into this case, both the CCM and BCM modes of operation are
the resonant network. The output voltage is regulated by considered. CCM has lower current stress on each compo-
controlling the switching frequency. In the case of an induc- nent, while BCM has smaller switching losses. In Figure
tive resonant network, the primary 11(b), a two-phase interleaved noniso-
MOSFETs would be turned on with lated buck charger is demonstrated.
ZVS. Figure 10 provides the dc voltage–­ The resonant With this interleaving configuration,
frequency characteristics of the LLC output current ripples are mostly com-
converter. The benefits of the LLC converters are the pensated as they cancel each other
topology include: 1) short circuit pro-
preferred topologies out. In addition, the current stress on
tection, 2) good voltage regulation in each leg is reduced to half so that a
light load conditions, 3) the ability to for the second stage higher power level can be achieved.
operate with a ZVS over wide load Instead of two-stage configura-
ranges, and 4) no diode reverse recov- of onboard and tions, the PFC and dc/dc stages are
ery losses in ZVS region 1. ­However, off-board chargers integrated into one single stage. The
because of its high circulating current s ingle-stage PFC chargers have
­
at maximum gain, it is difficult to opti- because of their reduced power losses, but they have
mize the efficiency of the LLC convert- low-frequency (twice the grid fre-
er over an ultrawide voltage range (e.g., improved quency) ripples in the output. The
100–420 V). efficiencies. single-stage topologies must be
Figure 9(f) shows a full-bridge ZVS adaptable to the universal grid (85–
PWM resonant converter. The switch- 265 V, 47–70 Hz) from the input, and
ing pattern is the same as that of the wide battery voltage (100–420 V) from
full-bridge phase shift ZVS converter. A half-bridge LLC the output. Thus, the selected topology should be able to
resonant circuit shares the lagging leg with a full-bridge both step up and step down the input voltage. ­Figure 11(c)
phase-shift converter, which makes sure the lagging leg is a buck–boost PFC converter, which has buck and boost
MOSFETs are turned on with ZVS across the full load capabilities and a minimum number of components.
range. However, a secondary-side hybrid switching circuit However, its disadvantages lie in four aspects: 1) high side
is required to clamp the voltage overshoots that arise dur- drive is required; 2) the MOSFET has high v ­ oltage stress
ing the turn off of the rectifier diodes. (Vin + Vout), which means 1,200-V rating MOSFETs are
required; 3) the ground polarity is reversed on the output
Nonisolated PEV Chargers side; and 4) the input current is discontinuous, which
Although a two-stage structure with galvanic isolation has means a bulky EMI filter is required. Figure 11(d) is a non-
been a common topology, with an additional safety mar- inverting buck–boost PFC converter. Compared to a con-
gin, isolation is not a requirement for OBCs, according to ventional inverting buck–boost PFC converter, there are
standards such as SAE J1772. Hence, researchers have two improvements: 1) voltage stresses on MOSFETs are
studied the applicability of nonisolated chargers for PEVs. reduced and 2) the input ground polarity is the same as
Six different types of nonisolated battery chargers are that of the output side.
summarized in Figure 11. Figure 11(a) shows a two-stage Figure 11(e) and (f) demonstrates single-ended primary-
nonisolated EV charger. An ac/dc PFC converter is used to inductor converter (SEPIC) PFC and Cúk PFC converters,

54 I E E E E l e c t ri f i c a t i on M a gaz ine / december 2013

 S1 Lr
S1 S3 S5 S7
Vdc Lo Llk
n :1 + Vbat
+ Vdc Co
+
– Co Vbat + Lm
S2 –
–
– S2 S4 Cr S6 S8

(a) (c)

Lr 1
S1 S2 S1 S3 S5 S7
Lo1 n:1 Lr 2
Vdc Vdc + Vbat
+ Co
+
– Vbat +
S3 S4 Lo2 Co – –
S2 S4 Cr 1 S6 S8
Cr 2
–

(b) (d)

Figure 13. The bidirectional dc/dc converters in PEV battery charging applications: (a) a nonisolated bidirectional buck converter, (b) the two-phase
interleaved version of (a), (c) a bidirectional dual active-bridge LLC converter, and (d) a bidirectional dual active-bridge CLLC converter.

respectively. Compared to ­Figure 11(c) and (d), their input The bidirectional dc/dc stage could be either
currents are both continuous, meaning the size of the EMI ­ onisolated or isolated. Figure 13(a) shows a nonisolated
n
filter can be reduced. The SEPIC PFC converter is not invert- bidirectional buck converter. During energy transfer from
ing. However, the MOSFET has high the grid to the battery, S2 is off and
voltage stress (Vin + Vout), and 1,200-V S1 operates as the main switch. The
MOSFETs are required. The Cúk PFC The EVSE serves as circuit functions as a buck convert-
converter puts less voltage stress on er. When the energy is transferred
the switching MOSFET, but its ground
the energy transfer from the battery to the grid, S1 is off
polarity in the output is reversed. interface between and S2 operates as the main switch.
The circuit functions as a boost con-
Bidirectional PEV Chargers the premises wirings verter. Figure 13(b) shows the two-
Currently, all commercialized OBCs phase interleaved version of Figure
have a unidirectional power flow
and the PEVs’ onboard 13(a). With this interleaving configu-
from the grid to the vehicle. However, or off-board ration, output current ripple is
since most vehicles are parked an reduced and a higher power level
average of 95% of the time, it is fore- battery chargers. can be achieved.
seeable that batteries could be used Figure 13(c) shows a bidirectional
to let power flow from the vehicle to dual active-bridge LLC converter,
the grid. In this emerging vehicle-to-grid (V2G) technology, which is a derivative of the full-bridge LLC resonant con-
OBCs are required to have bidirectional power flow capa- verter [see Figure 9(d)]. When the energy is transferred
bility. When the vehicle is in idle mode, the battery can from the grid to the battery, the active bridge on the
feed power back to the grid.
A typical bidirectional PEV charger consists of two
­stages. The first stage is a bidirectional ac/dc PFC convert- TABLE 3. The U.S. DOE’s technical targets for
er, and the second stage is a bidirectional dc/dc converter. OBCs.
Figure 12(a) shows a single-phase half-bridge bidirec-
3.3-kW Charger
tional ac/dc PFC converter, while Figure 12(b) shows a sin-
gle-phase, full-bridge bidirectional ac/dc PFC converter. Year Cost Size (L) Weight Efficiency
Although half-bridge topology uses only two MOSFETs to (kg) (%)
achieve voltage doubling, it requires semiconductor devic-
2010 US$900– 6–9 9–12 90–92
es with higher voltage ratings. The full-bridge topology 1,000
can alleviate capacitor imbalances, but it comes with a
2015 US$600 4 4 93
higher number of semiconductor devices, increasing the
2022 US$330 3.5 3.5 94
cost and complexity of the control.

IEEE Electrific ation Magazine / d ec emb er 201 3 55

 4) the conversion efficiency must be
TABLE 4. The power levels for dc direct charging.
optimized across the full battery
Type of Charge Charger Power Level (kW) voltage ranges and different load
Heavy Duty SUV/Sedan Small Sedan conditions.
However, it is a challenging task to
Fast charge, 10 min, 100% SoC 500 250 125 satisfy all of the above-mentioned con-
Rapid charge, 15 min, 60% SoC 250 125 60 siderations simultaneously. Higher
Quick charge, 60 min, 70% SoC 75 35 20 switching frequency is associated with
PHEV, 30 min 40 20 10 smaller volt–second applied to the
magnetic component; consequently,
the flux variation is smaller and, hence,
secondary side of transformer functions as a full-bridge the corresponding core losses are reduced. However, the core
rectifier. When the energy is transferred from the battery loss and the switching loss increase with the increase in fre-
to the grid, the secondary-side active bridge functions as quency. With higher switching frequency, the conversion effi-
an inverter and the primary-side active bridge functions ciency could potentially degrade.
as a rectifier. ­Figure 13(d) shows a Another challenge comes from the
bidirectional dual active-bridge wide range of voltage variation of
capacitor inductor inductor capaci- The Li-ion cell the battery pack. Corresponding to the
tor (CLLC) converter. In this topology, depleted and full SoC, the voltage of the
there are two identical inductor has dominated battery pack varies from the cutoff volt-
capacitor (LC) networks on both the age to the charge voltage (e.g., 100–420 V).
primary and secondary side.
the market of This means that the dc/dc conversion
Bidirectional power flow between commercially stage must be able to adapt to this wide
the grid and the vehicle has gained voltage range. The PWM topologies have
interest from academia and industry. available PEVs. the advantage of easy regulation of the
However, it has not been implement- output voltage in a wide range. Howev-
ed in any commercial PEV on the mar- er, they also have the disadvantage of
ket. The challenges mainly lie in four aspects: 1) the an incomplete ZVS range. Frequency modulation resonant
additional cost of power electronics, 2) the possible chance of topologies have a full ZVS range. However, using currently
battery degradation due to frequent cycling (which might not known topologies, the efficiency of resonant topologies can
be the case in some battery chemistries as a few manufac- only be optimized over a limited range of output voltage. To
turers believe that slow discharge of the battery when it is overcome these challenges and develop an ultracompact,
fully charged would not cause degradation), 3) the require- highly efficient onboard charging system, the following
ment for metering from the utility company, and 4) the lack components and technologies need to be addressed:
of precise policies and standards as of December 2013. 1) Advanced magnetics material : The size of the
­magnetic component is constrained by the core loss
Goals and Challenges associated with high switching frequency. To solve
The U.S. Department of Energy’s (DOE’s) technical targets this problem, a more advanced magnetics material
for 3.3-kW level 2 OBCs are summarized in Table 3. To with smaller core losses in higher switching frequen-
design an ultracompact and highly efficient OBC, the fol- cies must be adopted.
lowing ­considerations must be taken into account: 1) a 2) Advanced packaging technique: The packaging is
high switching frequency is desired to reduce the volume directly relevant to the size of the onboard charging
and weight, 2) both step-down and step-up operations system. Advances in the packaging technology
should be realized to satisfy the wide output voltage range improve the space utilization and heat dissipation.
requirements, 3) the ZVS feature is desired to reduce the 3) Advanced cooling technique: Heat sinks take a large
switching losses and high-frequency EMI, and volume of the charger. The size of the heat sink is
directly determined by the cooling
method. Generally, active cooling
Onboard is better than passive cooling. Liq-
3Φ Grid ac/dc dc/dc Battery
Input uid cooling is preferred in the case
Filter Converter Converter Pack
of conventional silicon-based
power electronic interfaces.
4) Advanced switching power devic-
Off-Board Level-3 Charger es: Power losses from switching
power devices such as MOSFETs
Figure 14. A block diagram of an off-board level-3 charger. and diodes take a large portion of

56 I E E E E l e c t ri f i c a t i on M a gaz ine / december 2013

 Cdc1
S1 S3 S5
Grid L1 + Grid L1 +
dc/dc
L2 Vbat L2 dc/dc
Cdc Converter
L3 L3 Converter V
– – bat
S2 S4 S6
S1 S2 S3 Cdc2

(a) (b)

Figure 15. Level 3 chargers using (a) a six-switch boost rectifier and (b) a Vienna rectifier.

the total system losses. Advanced power devices, equipped with an active front-end rectifier (in the ac/dc
including wide-band-gap silicon carbide and gallium stage) can mitigate this problem, it increases the cost of the
nitride semiconductor devices with low on resistanc- charging station substantially. Alternatively, the degradation
es, high voltage ratings, faster switching speeds, and of distribution equipment can be reduced by employing a
high operating temperatures, help reduce the power smart charging strategy. The allowable harmonic and dc
losses and the thermal stress. injection into the grid are limited by IEEE 1547, SAE-J2894
5) Advanced converter topologies and control methods: IEC1000-3-2, and the U.S. National Electric Code (NEC) 690
The converter topology determines the circuit perfor- standards.
mances such as the ZVS feature, EMI, circulating cur- Different topologies are reported for three-phase off-
rent, conduction losses, and switching losses. An board chargers. Since the level 3 battery charger is typical-
optimized circuit topology and ly supplied by 480-V three-phase
control method would help opti- voltage, the grid voltage would be
mize the overall circuit perfor- Charging higher than the battery voltage. As
mance over the wide battery SoC mentioned earlier, an active PFC recti-
range. convenience and fier is required to alleviate grid power
charging time are quality issues. Hence, a three-phase
Off-Board Charging boost converter followed by a dc/dc
Level 3 charging, known as dc fast two of the most converter is required to charge the
charging, requires an off-board battery. Additionally, as mentioned in
charger, which is less constrained by important concerns the “Isolated Onboard PEV Chargers”
size. Fast charge, rapid charge, and from the consumer section, the galvanic isolation is
quick charge are a few of the important in the charger circuits for
commonly used terms for off-board, perspective. safety reasons. Even though PEV
as shown in Table 4. The high power charging standards do not mandate
level and high cost of off-board galvanic isolation as long as the
charging make it unfeasible for residential areas. Level 3 ground current is maintained within limited boundaries, if
charging is not compatible with commercial and residen- the battery is attached to the vehicle’s chassis, galvanic
tial outlets and, consequently, requires the installment of isolation is mandatory for safety reasons. Therefore, to
new charging infrastructure. Nonetheless, it is attractive avoid costly and complex shielding considerations for a
for commercial and public applications such as shopping nonisolated dc/dc converter, an isolated dc/dc converter is
centers, parking lots, hotels, highway rest areas, and preferred. Similar to OBCs, as shown in Figure 9, the reso-
ordinary filling stations. nant converters are the preferred topologies for the sec-
As shown in Figure 14, a level 3 charger is typically fed ond conversion stage of off-board charging because of
through a 208–600-Vac three-phase circuit. This system con- their improved efficiencies.
sists of an ac/dc stage and a dc/dc stage. Because of the high The three-phase six-switch boost and Vienna convert-
­current/power requirements, the ­off-board charging puts ers are the most suitable topologies for level 3 charger
power quality burdens on electric utility distribution sys- applications. Figure 15(a) shows the schematic of a com-
tems. These burdens can include voltage deviations, har- mon three-phase boost-type voltage–source converter
monic distortion, peak demand, and thermal loading on (VSC) followed by a resonant dc/dc converter. The VSC has
distribution power systems. In particular, the harmonic and a relatively simple structure, despite its high functionality,
dc injection can particularly increase distribution transform- that includes six semiconductor switches, three input
er losses resulting in thermal loading, which impacts trans- inductors, and one output filter capacitor to provide input
former aging accordingly. Although a charging system PFC and output voltage regulation simultaneously. In this

IEEE Electrific ation Magazine / d ec emb er 201 3 57

 boost converter, the voltage shape in each phase is inde- For Further Reading
pendent of the current as the current flows over the K. S. K. Young, C. Wang, and L. Y. Wang, “Electric vehicle
switch (insulated-gate bipolar transistor here) or its anti- battery technologies,” in Electric Vehicle Integration into
parallel diode. At the midpoint of each semiconductor leg Modern Power Networks. Berlin: Springer, 2013, pp. 15–56.
with respect to the midpoint across the output capacitor, a A. Khaligh and S. Dusmez, “Comprehensive topological
positive or a negative voltage may be produced. Therefore, analysis of conductive and inductive charging solutions
because of the two-level behavior of each semiconductor for plug-in electric vehicles,” IEEE Trans. Veh. Technol.,
leg, the phase of sinusoidal currents can be simply con- vol. 61, no. 8, pp. 3475–3489, Oct. 2012.
trolled. In fact, the current can be controlled to be in phase M. Yilmaz and P. Krein, “Review of battery charger
with grid voltage for grid-to-vehicle operation or anti- topologies, charging power levels and infrastructure for
phase with grid voltage for V2G operation. Hence, this con- plug-in electric and hybrid vehicles,” IEEE Trans. Power
figuration offers bidirectional operation. Electron., vol. 28, no. 5, pp. 2152–2169, May 2013.
The circuit in Figure 15(b) is a Vienna rectifier, which C.-Y. Oh, D.-H. Kim, D.-G. Woo, W.-Y. Sung, Y.-S. Kim, and
contains only three active switches. The switching of B.-K. Lee, “A high-efficient nonisolated single-stage on-
semiconductors is independent of the direction of phase board battery charger for electric vehicles,” IEEE Trans.
currents. The major advantage of the three-level Power Electron., vol. 28, no. 12, pp. 5746–5757, Dec. 2013.
­characteristic is that switches can be rated for only half of S. Dusmez, A. Cook, and A. Khaligh, “Comprehensive
the peak value of the line-to-line voltage. Furthermore, analysis of high quality power converters for level 3 off-
because of the three voltage levels, a smaller grid current board chargers,” in Proc. IEEE Vehicle Power and Propul-
fundamental ripple is generated, and the value of the sion Conf., 2011, pp. 1–10.
input boost inductances can be reduced. In addition, C. Botsford and A. Szczepanek, “Fast charging vs. slow
because of lower switching voltages, a lower conducted charging: Pros and cons for the new age of electric vehi-
EMI noise is generated. cles,” in Proc. Int. Battery, Hybrid and Electric Vehicle Symp. and
Expo. (EVS24), Stavanger, Norway, May 2009, pp. 1–9.
Conclusions SAE Standard for Electric Vehicle Conductive Charge
Although a great vision has been plotted for transporta- Coupler, SAE Standard 1772–2009. [Online]. Available:
tion electrification, there are several critical hurdles that http://www.arb.ca.gov/msprog/zevprog/stakeholders/
EVs must stride. Among these, charging convenience and infrastructure/finalsaej1772.doc
charging time are two of the most important concerns IEEE Standard for Interconnecting Distributed Resources
from the consumer perspective. Good power quality, com- with Electric Power Systems, IEEE Standard 1547–2003, 2003.
pact size, and high conversion efficiency are three of the SAE Standard for Power Quality Requirements for Plug-In
most important features desired from onboard level 1 and Electric Vehicle Chargers, SAE Standard J2894–2011, 2011.
level 2 PEV chargers. To ensure good power quality, the ac/ IEC Standard for Line Current Harmonic Measure-
dc stage must be featured with good PFC and THD reduc- ments, IEEE Standard 1000-3-2, 1995.
tion performances. Higher switching frequency, ZVS, and NEC Standard for Compliance for Photovoltaic Systems,
advanced packaging techniques are necessary to achieve U.S. National Electric Code (NEC) Standard 690.
smaller charger size. To achieve high converter efficiency,
advanced magnetics materials, power semiconductors, Biographies
and power electronic topologies must to be addressed. In Haoyu Wang is a Ph.D. candidate at the Power Electronic,
particular, the advanced control method must be imple- Energy Harvesting and Renewable Energies Laboratory at
mented to optimize the conversion efficiency across the the Electrical and Computer Engineering Department in the
full SoC range. Level 3 charging is able to reach a much University of Maryland. He is a Student Member of the IEEE.
higher power level and is regarded as an important alter- Amin Hasanzadeh is a postdoctoral research associate
native to level 1 and level 2 onboard charging. It is essen- at the Power Electronic, Energy Harvesting and Renewable
tial to alleviating range anxiety issues as it allows for Energies Laboratory at the Electrical and Computer Engi-
long-distance travel similar to conventional ICE-powered neering Department in the University of Maryland. He is a
vehicles. However, it requires an off-board charging station Member of the IEEE.
and is mainly intended for commercial and public Alireza Khaligh ([email protected]) is an assistant
­charging stations. professor and the director of the Power Electronic, Energy
Harvesting and Renewable Energies Laboratory at
Acknowledgment the Electrical and Computer Engineering Department
This work has been sponsored partly by the National in the University of Maryland. He is a Senior Member of
­Science Foundation Grant 1238985 and the Maryland the IEEE.
Industrial Partnerships Program, which are gratefully
acknowledged. 

58 I E E E E l e c t ri f i c a t i on M a gaz ine / december 2013