1

Power Stages

The EV Charging
Infrastructure Designbook: Volume 1

Power Stages
 Content
The variety of charger types......................................................................... 4

Connector variations........................................................................................ 4

Connector and cable technology -

the future............................................................................................................... 6

AC wall box EV chargers............................................................................... 6

AC wall box power architecture ..................................................................7

DC EV chargers.................................................................................................. 8

Variants.................................................................................................................... 8

Uni- and bi-directional energy flow.......................................................... 9

DC charger hardware...................................................................................... 9

EMI Filter stage ..................................................................................................10

Line rectification and power factor

correction...............................................................................................................11

The DC link............................................................................................................11

Main power conversion and isolation stage......................................... 12

Auxiliary power...................................................................................................14

Sensing current..................................................................................................14

2
Version 01, September 2022

The EV Charging
Infrastructure Designbook:

Power Stages
This is the first in a series of guides for designers of fixed
EV charging stations, whether at the component level or
utilizing a modular approach. In this part, the range of
power stage options are examined with an overview of their
function and what technologies and conversion topologies
might be employed. In further design guides, the control
functions, connectivity and software are considered.

T
he accelerating move to adoption of electric vehicles installations showing the highest growth. As a result,
in the last few years has generated a paradigm shift in 2021, the worldwide average was 10 EVs per public
in battery and power conversion technologies. charger and 2.4kW per EV. Home charger installation
Power density of energy storage and conversion rates are higher still, with one estimate putting total
efficiency have always been a goal in any application, but installed at 35 million globally by 2026. With prices at
in EVs, these parameters make the difference between $600 to over $1000 each, the market for ‘wall boxes’
e-mobility being fundamentally viable or not. Advances in alone is in the order of tens of $ billions.
battery chemistry, motors and traction drive electronics have
now resulted in range per charge that is generally acceptable.

Now, the roll-out rate of charge stations, their associated

infrastructure and time to charge have come to the fore, with
the International Energy Agency (IEA) reporting nearly half
a million public chargers installed in 2021 [1]. The increase is
exponential, year on year, at around 50%, with fast charger

3
 The variety of charger types Connector variations

In this design guide, we will consider off-board wired chargers, In 2011, as an initiative to generate a standard for AC-only,
including home single-phase AC connections at 240/120V DC-only and combination connectors, a ‘Combined Charging
supplying a maximum of about 8kW to an on-board charger. System’ or CCS (now at CCS 2.0) was proposed for a single
Also considered are three-phase AC and DC output home connector pattern that could be used worldwide, now
wall boxes delivering up to around 20kW and roadside DC formalized under IEC 62196: ‘Plugs, socket-outlets, vehicle
fast chargers currently rated up to 350kW, with higher power connectors and vehicle inlets – Conductive charging of
on the horizon. Charging is also characterized by ‘Levels’ and electric vehicles’.
their types of connectors: Level 1 at 120V AC from a domestic
source provides 12-20A, a charging rate of around 6.5 km/ The ‘Charging Interface Initiative’ or CharIN [2] was also set
hour and duration 30+ hours. up by an association of car makers, suppliers and interested
parties to promote CCS. First CCS-compliant charging stations
Level 2 is at 220/240VAC at up to 80A, providing around 20 were rated at 50kW, then 150kW in 2015 and now 350kW is
to 50 km/hour in 8 to 12 hours. available with higher power possible for the future. The CCS
standard defines a ‘Control Pilot Contact’ (CP) which allows
Level 3 is DC fast charging from a dedicated roadside station, bi-directional communication between charger and EV with
powered from three-phase AC and providing 5 to 25 miles Pulse Width Modulation (PWM) basic signaling (BS) according
range per minute with a typical charge time of 30 minutes to to IEC 61851-1 ‘Electric vehicle conductive charging system’.
80% full. ‘Superchargers’, informally called ‘Level 3+’ or Level
4, are still relatively rare but improve charge rates still further High Level Communication (HLC) based on the ISO/IEC 15118
to just a few minutes, used typically with 800V commercial series and DIN 70121 allows for more complexity such as
vehicle batteries. Figure 1 summarizes. scheduling of the charge rate for user convenience and by
the utility supplier for grid load-balancing. Other connection
standards for DC charging are CHAdeMO in Japan and
the GB/T standard used in the China. Table 1 shows CCS 2
functionality for AC and DC charging.

Energy Supply
Power level Types Charger Location Typical Use Expected Power Level Charging Time
Interface

On-board 1-phase (AC Charging at home or 1.4kW (12A) 11-36 hours

Level 1 (Slow) Convenience outlet
charging) office 1.9kW (20A) 4-11 hours

Level 2 4Kw (17A) 2- 6 hours

On-board 1- or 3-phase Charging at private or Dedicated power
(Intermediate or 8kW (32A) 2-3 hours
(AC charging) public outlets connector
slow) 19.2kW (80A) 1-4 hours

50kW
Level 3 Off-board
Commercial, same as Dedicated power 100kW 0.4-1 hours
(Fast/less than 1 3-phases
gas station connector 250kW 0.2-0.5 hours
hour) (DC charging)
350kW

Figure 1: The levels of EV chargers

4
 Feature CCS 2.0 DC CCS 2.0 AC

Vehicle Connector Combo 1 or 2 (IEC 62196-3) Type 1 or 2 (IEC 62196-2)

Type 1 or 2 (IEC 62196-2)

Vehicle Inlet Combo 1 or 2 (IEC 62196-3)
Combo 1 or 2 (IEC 62196-3)

Basic Signaling:
High Level Communication: · IEC 61851-1:2010 ED2
· DIN SPEC 70121:2014 (<80 kW)
Charging Communication
· ISO/IEC 15118-2:2014 ED1 (<350 kW) Basic Signaling:
· ISO/IEC 15118-3:2015 ED1 (<350 kW) · ISO/IEC 15118-2:2014 ED1
· ISO/IEC 15118-3:2015 ED1

With Basic Signaling only:

· Reactive
Loading Balance Reactive and Scheduled
With High Level Communication:
· Reactive and/or
· Scheduled

With Basic Signaling only:

· External Payment
Charge
External Payment and/or Plud and Charge
Authorization
(mandatory from 2020 on) With High Level Communication:
Mode
· External Payment and/or
· Plug and Charge (mandatory from 2020 on)

Charging Station IEC 61851-1 IEC 61851-1

Table 1: CCS 2.0 specifications for AC and DC charging

The CCS specification references connectors Type 1, Type 2, Invalid combinations do not mechanically mate, such as a
Combo 1 and Combo 2. Type 1, a North American standard, Combo 2 coupler and a Type 2 inlet. Figures 2 and 3 show
was the first to be used for single-phase AC connection and the combinations. (Source - Wikipedia).
Type 2 is able to charge with three-phase AC, as defined in
European standard IEC 62196-2. New connector designs were Couplers and inlets have to withstand everyday outdoor use,
adopted for combined high voltage DC and AC charging: so must be robust. Manufacturers such as Amphenol have
Combo 1 and 2, associated with Type 1 and 2. Power inlets designed their parts for more than 10,000 cycles (unloaded)
on EVs have all pins populated corresponding to valid and their Excel Mate CC -200A IEC Combo parts are rated at
charge connections, while the connectors on the charging 1000V/200A DC, with IP44 environmental protection (mated)
cable just have the pins fitted for that source, AC or DC. and incorporate a temperature sensor.

TYPE 1 COMBO 1 TYPE 2 COMBO 2

Inlet/Connector Inlet/Connector

TYPE 1 TYPE 2

AC Charging
AC Charging
Does not mate Singe Phase or Does not mate
Singe Phase
Three Phase

COMBO 1 COMBO 2

AC Charging
AC Charging
DC Charging Singe Phase or DC Charging
Singe Phase
Three Phase

Figure 2: Mating options for Type 1 and Combo 1 couplers Figure 3: Mating options for Type 2 and Combo 2 couplers

5
 Connector and cable technology - AC wall box EV chargers
the future
The simplest and slowest charging method is from an AC
Cables are routinely rated today at 200A, but this figure is ‘wall box’ typically in a home, workplace or street-side
set to increase with ultra-fast chargers providing 500kW, installation. This routes mains AC directly to the on-board
or 500A at 1000V. Even higher power, 900kW, is planned by AC-DC charger in the vehicle, allowing up to around 8kW
China, with its ‘Chao ji’ connector and charging standard. charge rate from single-phase supplies, or more unusually,
This would be more relevant however to high-capacity around 20kW from a three-phase supply. Vehicles that can
batteries in trucks and buses. In this system, the maximum accept three-phase AC are not common and even if they
voltage proposed is 1500V and current 600A. do, there may be a limit set by the car, 11kW for example
for the Tesla model 3. In the home or work-place, bill-
This potentially gives recharge rates comparable to a gas ing functionality is not typically needed, perhaps only an
pump, but represents a challenge for the cable and con- RFID or key function for authentication, so it might seem
nectors. CCS Combo connectors are seen as potentially that the cable could just connect to a domestic or office
able to handle higher than 200A but only with active cool- socket. However, direct connection to a wall socket, de-
ing, where liquid is circulated through the cable and con- fined as ‘Mode 1 charging’, typically only intended for elec-
nector contacts. This allows the cable to be lighter, more tric bikes and scooters, is seen as potentially dangerous.
flexible and easier to handle.
This is due to the possibility of overload and fire and is
Resistive losses are also minimized by keeping tempera- prohibited by law in the US, UK and other countries in
ture rises lower. The higher voltages used for higher charge public places. These countries require the charging to
rates present a greater shock hazard, so connectors also meet ‘Mode 2’ connection at least, according to standard
need the pilot line to signal that mating is complete and IEC 61851. The standard covers the ‘Mechanical, electrical,
weather-sealed before the charging voltage is applied. communications, EMC and performance requirements for
This also prevents arcing and degradation of the connec- EV supply equipment used to charge electric vehicles, in-
tor contacts. cluding light electric vehicles’. In the standard, Mode 2 de-
fines connection to a standard AC supply socket, but with
a control pilot function, that is, communication between
the EV and a control unit in-line with the cable. A system
for personal protection against electric shock between the
standard plug and EV is also required. Mode 3 connection
is for a wall box with permanent, dedicated connection to
an AC supply, again with a control pilot function. A ‘street-
side’ AC charger will additionally have control and billing
functionality using either in-car connectivity or a mobile
device. Mode 4 is for DC output from a wall box or road-
side charger, which we will discuss later.

T1 T2 RELAY

L in L out

Transient
suppression
Figure 4:
The power flow N in N out
through an AC wall Current
GFCI Test
monitor Disconnect
box EV charger AC monitor and GFCI Trip control
aux supply To EV

Pilot
Measurement, control, communications
AC-DC aux
supply

E E
6
 1st Stage 2nd Stage 3rd Stage
IEC 61643-11
Class I Class II Class III

IEC 60664-1 Category IV Category III Category II Category I

Maximum Impuls Withstand
6kV 4kV 2.5kV 1.5kV
300Vrms System Voltage

Symbol IV III II I

TNC End
230V/400V/50Hz Equipment

DIN VDE 0675 Class B Class C Class D

IEC 62305-1 Zone Divisions LPZ 0A (LPZ 0B) LPZ 1 LPZ 2 LPZ 3

Figure 5: Over-voltage categories. For EV charger installations IEC 60664-1 applies

AC wall box power architecture are typically used, but may be supplemented by gas dis-
charge tubes (GDTs) at OVC IV. Devices that combine a
It is clear from the typical size of an in-line control unit series GDT and varistor in one package are available such
or AC wall box that they do not feature galvanic isolation as the ISOMOV TM range from Bourns. Varistors can exhibit
between the AC supply and EV connection in normal op- a wear-out mechanism, because of continuous leakage
eration – this would require a large multi-kW inverter or an current in normal conditions and from energy absorption
extremely large 50/60Hz transformer. However, a means of during transient events. This is avoided in the ISOMOV TM
mechanical disconnection is required, typically with relay range, as the GDT provides a break in any leakage path,
contacts in both line and neutral. These go open when stopping degradation and the risk of eventual ‘flame-out’
an overload, short circuit or ground fault is detected. To and fire, which would otherwise mandate a separate fuse
sense these conditions, current transformers are placed or thermal cut-out.
in differential and common mode positions on the power
connections (Figure 4). The disconnect function can also The 50/60Hz current transformers, T1 and T2 in Figure 4, op-
be used for scheduled charging, controlled either by an erate under different conditions. T1 measures line current,
interface on the control unit/wall-box or remotely by which can be tens of amps normally and momentarily hun-
wireless communication to a cell phone app, or possibly dreds under fault conditions, tested at 230A peak for 100µs
through powerline communications along the AC supply. according to IEC 61851-1. A part has to be specified to not
The control, measurement and communications options saturate under normal conditions at high temperatures
will be discussed in a following design guide. and could be a strip-wound toroidal type using specialty
steel, for small size and high saturation flux density. PCB
In Figure 4, the varistor shown indicates that line transient mount types are available from KEMET for example, with
suppression is needed. The level of suppression is defined their MR series featuring continuous ratings to 125A. Wired
in IEC 61851-1 and varies with the charger mode. For ex- versions are also offered, but in either case, the AC power
ample, chargers must withstand 1.2µs/50µs impulses at connection would typically be a wire passed through the
their AC inputs, tested according to IEC 60664-1 at volt- toroid aperture. Current measurement is used to measure
ages for ‘Over-Voltage Category (OVC) IV’, the highest, for energy usage and to detect over current and short circuits.
EV charger equipment connected to a public AC supply. Resistive sensing could be used, but causes relatively high
A domestic wall box permanently connected (Mode 3) is losses and the sensed signal is not isolated from the AC
subject to OVC III levels and a pluggable domestic charger mains. Hall effect sensors provide isolation but are relatively
(Mode 2) is OVC II. Figure 5 summarizes the voltages and expensive and non-linear compared with resistors or trans-
categories. Over-Voltage Categories for other standards, formers, suffer from poor temperature stability and have
IEC 61643-11, DIN VDE 0675 and IEC 62305 are included offset errors that have to be trimmed out. For AC, current
for information. At IEC 60664-1 OVC II and III, varistors transformers are therefore a good solution.

7
 T2 operates as a Ground Fault Current Indicator (GFCI), with though. Any auxiliary supply must operate at the higher
normal line and neutral current causing flux cancellation. A voltage input and the wide input modular supplies from
secondary voltage is only generated when there is a differ- Aimtec are candidates, with their rated input from 82 to
ence between line and neutral currents indicating an extra 528VAC, which can operate from line to line or line to neu-
fault current to ground. Because of the flux cancellation, tral in three-phase systems. A disconnect relay is required
the part can be physically smaller than T1. Typically, an for all three phases, but there are single relay parts avail-
extra winding is added, driven from the controller, to force able such as the new HE-R product from Panasonic that
a ground fault indication for test purposes. A smaller part include four normally open contacts able to switch up to
in the KEMET MR series could also be used in this position. 40A. An optional low current, normally-closed contact is
also incorporated for feedback. The relay is certified to IEC
The relay shown in Figure 4 performs the disconnect 62955 and has a coil holding power of less than 0.5W.
function under fault conditions or for scheduled connec-
tion. The contacts are shown driven from one coil but two
separate relays are also seen. The contacts must withstand
repeated test surge currents to IEC 61851-1 as mentioned DC EV chargers
and must have very low contact resistance for minimum
loss, while maintaining safety separation. Additionally, the Variants
relay should comply with IEC 61810-1, characterized for
50,000 cycles minimum with a contact category CC 2 (a DC charging, or ‘Mode 4’ connects directly to the EV bat-
high load where contact arcing can occur). An important tery and bypasses the limited power capability of an on-
consideration is the relay lifetime, which can be limited by board AC-DC charger, for faster recharge time. This does
contact degradation and resistance increase due to arc- mean that the charger is either specific to a vehicle type,
ing. Latest designs such as the G9KB series from OMRON such as Tesla or must be able to vary output voltage ac-
incorporate special techniques for arc minimization, with cording to the particular EV and its installed battery, which
their qualification tests showing large design margins, could currently be between 200V and 920V. In practice, in
enabling the company to offer a long warranty period. For all DC chargers, there must be detailed communication
safety, the relay has to be ‘normally open’ so that it defaults using the pilot connection between the EV and charger,
to the disconnected state. Power is therefore dissipated in to signal correct and complete mating, the initial state
the coil in operation, but this can be minimized by ener- of charge, appropriate charge rate and a cut off when
gizing at the rated coil voltage then reducing to a ‘holding’ completed or if a fault occurs such as battery temperature
value of around 50%. With AC wall boxes and in-line con- becoming too high.
trol units being ever-smaller and often fitted in hostile en-
vironments, the reduction in dissipation significantly helps DC chargers can be fitted as home installations but are
to keep internal temperatures lower and reliability/lifetime expensive and require a three-phase supply which is not
higher. PCB-mount versions of relays are also attractive for usually available. They would still be limited to about 22kW
compact form-factors. by domestic wiring, but could have a place, where the on-
board AC-DC charger of a particular vehicle limits power
The isolated auxiliary supply shown in the block diagram to less than 2kW. Again, in this scenario, billing functionality
provides a few watts for the control and monitoring func- is not a requirement but connectivity would be provided
tions. It is typically a board-mounted module and may to enable user authentication, charge scheduling and en-
need to be Over-Voltage Category III rated, depending ergy use monitoring.
on the installation position. Alternatively, external transient
suppression components can be added. Suitable modular In public areas such as car parks and fuel stations, ‘Level 3’
power supply products can be found in the Murata BAC DC chargers might be provided with ratings up to about
range or from Aimtec in their AME range. 150kW, powered from the local low-voltage three-phase
network, which limits total power available. Chargers will
Three-phase AC wall boxes for domestic installation will typically be stand-alone and assembled from paralleled
normally operate from nominal 380/400VAC and be de- sub-units, rated at 15-25kW each. This allows installations
signed for Over-Voltage Category III and IEC 61581 similarly to be economically scaled to the power available and al-
applies as it is relevant for supply voltages up to 1000VAC. lows for some fault tolerance with reduced output.
Appropriate creepage and clearance are necessary for
the higher line to line voltages seen, and multiple sense Level 4+ or ultra-fast DC chargers would only be seen in
transformers are required for running current measure- charger parks, perhaps in the larger highway service sta-
ment. Ground fault detection can still use one transformer tions or ‘Park-and-Ride’ facilities. These can be rated up

8
to 350kW currently (400kW for CHAdeMO) so more than a some central battery bulk storage, topped up by solar or
few stations would represent too high a load on the local other alternative energy. Excess EV battery energy could
low voltage three-phase network. For this reason, power be used to add to this battery capacity. It would not re-
is typically taken from the medium voltage network at place it, as there needs to be storage when the EV is not
11kV to 33kV, with a local, dedicated transformer drop- present. A separate bi-directional inverter would in this
ping the voltage down to three-phase network levels, case allow back-feeding from the whole system to the
400VAC in Europe. utility supply for cash credit. Level 3 DC chargers at home
or at a designated parking spot could be bi-directional
but again the extra protection equipment is a barrier
Uni- and bi-directional energy flow and like Level 4 DC charging, the fast charge rates are
for users that need quick ‘refueling’ perhaps on a longer
Many of the power conversion technologies used in EV trip, where there is no value in losing, rather than gaining
charging, both AC and DC, are capable of bi-directional battery range.
energy flow, that is, energy could be returned from the
EV back to the source, sometimes called ‘V2G’. In prac-
tice, there are only a few instances when this is attrac-
tive. For AC wall boxes, this would be prohibited without DC charger hardware
certified protection measures to prevent ‘islanding’, the
inadvertent back-feeding of AC into the utility network, The general outline of an EV DC charger is shown in Figure
endangering operatives who might think they are work- 6. Input is normally three-phase AC passing through an
ing on ‘dead’ connections. The principle does operate EMI filtering stage (not shown) before rectification, power
with domestic solar energy installations for example, factor correction, a DC link, then conversion with isolation
but an EV back-feeding less than 2kW is not very worth- (although sometimes without) to the required battery
while considering the extra protection equipment that voltage. We will consider the power stages in this design
would be needed. A DC charger however, is more easily guide – control and communication functions will be
coupled into a ‘mini’ or ‘micro’ home grid where there is covered a following document.

1200V SiC
Diodes

Secondary Battery
3-Phase DC-DC
Rectifier/PFC Rectification

Current
1200V FRD Sense Discrete IGBTs & HV FETs
BLDC Fan
with low RDS(on)/fast
Diodes Gate Drivers Gate Drivers Moter
switching ot PIMs providing
improved system reliability
High Current BLDC Motor
full protection Control
Gate Drivers
Voltage
µCU Isolation ADC
High Precision Sense
Op-Amps
Communications Touch Screen Wide Body
Controller Optocouplers
Auxiliary Display
LCD
RF PLC CAN Power
Backlighting

Figure 6: Three-phase EV DC charger outline (Source Onsemi)

9
 The implementation of the various power stages depends propriate for OVC III and IV installations. Above 0.1µF, for
on the power level, with different conversion topologies ‘pluggable’ inputs, X capacitors must also discharge within
having trade-offs in efficiency, potential EMI, complexity, a maximum time to a safe voltage. Suitable capacitors are
cost, functionality and ability to be bi-directional. In this available from TDK Electronics in their new B32922…B32924
section, as well as EMI filtering, we will consider the ‘Totem- xxx series, Kyocera/AVX in their FVC series, KEMET with
Pole PFC’ stage, ‘Vienna rectifier’ and ‘Active Front End’ for their R46K13 X2 series or the Murata DE1 series rated X1 to
the combined rectification and power factor correction 760VAC, suitable for three-phase applications.
function. For the main DC-DC conversion stage the ‘LLC’
and ‘CLLC’ converters are discussed, in stacked and inter- Capacitors from lines to ground must be ‘Y’ types, pro-
leaved forms, along with the ‘hard-switched full bridge’, viding ‘Basic’ insulation at the system operating voltage.
‘phase-shift full bridge’ and ‘dual active bridge’ topologies. These attenuate common-mode noise but are limited in
It is not possible to say that any one topology is ‘best’ for a value by leakage current considerations. IEC 61851-1 sets
particular power level as high-power chargers are typically a limit of 3.5mA total from all leakage current sources for
assembled from lower power ‘sub-units’ and each topology Class I, earthed chargers, tested at 1.1x rated maximum
can be interleaved and arranged as ‘multi-level’ for higher input voltage. The TDK 81123 series includes examples of
power. The best efficiency low power topology could there- suitable ‘Y’ capacitors as do the KEMET PME295 series.
fore be chosen for example, but cost and control complex-
ity can increase to an unaffordable level when arranged in The common-mode coupled inductor T1 in Figure 7 is
multiples for higher power. Selection of a topology must typically wound on a high-permeability ferrite core with
therefore be done with overall system constraints in mind. flux from operating current in each line cancelling, so
high winding inductance, in the order of mH can be de-
signed, for maximum common-mode current attenua-
EMI Filter stage tion. Winding resistance and dissipation is traded against
wire and overall component size but very high induct-
At any load power level, an EMI filter stage (Figure 7), works ances can have a low self-resonance with self and stray
to attenuate transients and bursts on the input, typically at capacitances, reducing the attenuating effect. It can
Category III or IV levels, with up to 6kV impulses present, therefore be better to have multiple stages of filtering,
as well as to attenuate conducted emissions from the EV each with high self-resonance for an overall better effect.
charger. The IEC 61000 ‘Electromagnetic Compatibility’ Along with the ‘X’ capacitors, series inductors L1, L2 and
series of standards apply. As in the AC wall boxes, varistors L3 attenuate differential mode noise and there is no flux
would be commonly employed for transient suppres- cancellation, so only small inductance values are practi-
sion with integrated or separate gas discharge tubes to cal, typically tens of µH, using iron powder toroid cores.
reduce leakage and component degradation with time EMC standards for conducted emissions use test methods
and transient events. The ISOMOV TM range from Bourns that combine differential and common mode noise into
is again a good choice. Capacitor and inductor networks one measurement so both modes must be attenuated.
are then employed to further reduce transients and also Sometimes, the common-mode coupled inductor can be
more importantly, to attenuate differential and common wound with deliberately high leakage inductance, which
mode emissions from the charger back to the supply. helps with series-mode noise filtering and can be a more
Capacitors across the lines must be ‘X’ safety-rated so economical solution than having separate series-mode
that they withstand high impulse voltages and restricted inductors. This component will often be a custom design
in value to avoid excessive current draw under no load or due to the specific characteristics required for a particular
standby operation. X1-rated capacitors have the highest application and suppliers such as TDK and KEMET offer a
transient voltage withstand (4kV tested) and would be ap- design service as well as standard components.

Fuse 1 L1
L1 in L1 out

C1 C4 C9

‘X’
‘X’ ‘Y’
Fuse 2 L2
L2 in L2 out

C2 C3 C5 C6

‘X’ ‘X’ ‘X’ ‘X’

Fuse 3 L3
L3 in L3 out
Varistors/GDTs
Figure 7: C7 C8

A typical three phase Earth ‘Y’ ‘Y’

EMI filter

10
Line rectification and power factor using a three-level circuit such as the Vienna rectifier (Figure
correction 8, right). The switches only see half of the DC output voltage
so the six switches can be lower loss types. In its simplest form
An overriding concern in power conversion stages is ef- shown, the Vienna rectifier uses six diodes which do see the
ficiency – high values lead to smaller, cheaper products full voltage and control is relatively complex. However, effi-
and energy savings. In EV chargers, the user is billed for ciency is high and lower loss/cost 650V-rated switches can be
the energy provided, not the total energy taken from the used for up to 480VAC three-phase input. The basic circuit is
supply, which includes losses from less than perfect con- not bi-directional however, but it can be made so with added
version efficiency, so every watt dissipated in the charger complexity and cost. Suitable SiC switches from Infineon are
is a cost to the energy provider and the environment, and available at 1200V and 650V rating in TO-247 packages up to
needs to be minimized. In all chargers, AC line voltage has 47A, while Onsemi also offers parts, with one device for exam-
to be initially converted to DC before a normally isolated ple, the NTBG020N120SC1 rated at 98A/1200V in a D2PAK-7L
conversion stage and traditionally this would be achieved package. Both manufacturers also offer suitable SiC diodes.
with a bridge rectifier. However, this loses significant power
from two diodes conducting at any one time, so ‘bridge-
less’ rectification and power factor correction circuits are The DC link
the solution. There are different options depending on
power level; for single phase operation to around 20kW, a The output of the rectification/power factor correction stage
‘Totem-Pole PFC’ (TPPFC) stage might be used, for three- is high voltage DC. This is regulated to be a constant value,
phase to around 50kW, an ‘Active Front End’ (AFE) can be higher than the peak of the AC input at typically 800VDC.
used and the ‘Vienna Rectifier’ can also be considered. Some arrangements allow for the DC voltage to be varied,
Each topology can be interleaved or stacked in a multi- which can ease the design of the subsequent power stage
level arrangement to increase power rating, or to reduce when it is required to vary its output for different battery volt-
component stress at a given power level. Power factor ages. A capacitor is placed on the DC link which has several
correction is achieved by pulse width modulation of the functions: it provides attenuation of high frequency ripple
switches, to force sinusoidal current draw. from the preceding power factor correction stage; it provides
a source for the high frequency ripple current into the follow-
The TPPFC stage is popular, (Figure 8, left) with just one diode ing DC-DC stage and it also acts as energy storage for ride-
drop in operation, although a fully synchronous version is through on power dips and interruptions. In practice, multiple
possible, replacing the diodes with MOSFETs. The circuit can paralleled capacitors are used, perhaps electrolytic types for
be controlled to be bi-directional and is ‘hard-switched’ at energy storage and film/ceramic types for high frequency
high power, operating in ‘Continuous Current Mode’ (CCM), ripple provision, which can be tens of amps in scale. Correct
so there are potentially reverse recovery losses in switch positioning of the film/ceramic capacitors is vital, to avoid
body diodes. Silicon MOSFETs can be used, but in these, ripple current/spikes producing voltage transients across
body diode reverse recovery losses are relatively high, so connection inductance. The capacitors should have very low
SiC devices are preferred, with their much lower reverse re- internal inductance (ESL) for the same reason, along with low
covery charge. The three-phase AFE circuit (Figure 8, mid- internal resistance (ESR) to avoid excessive heating. KEMET has
dle) is similar and it is clearer how it can be bi-directional, parts available which suit the application, including press-fit
with the circuit’s resemblance to a reversed, three-phase types, making change-out easier if required. The company
inverter. Both the TPPFC and AFE are ‘two-level’ circuits, ap- also offers a ceramic part in C0G dielectric for surface mount
plying full output DC voltage to the switches, so they need assembly, with 150°C rating and extremely low ESR and ESL,
to be high-voltage types. Switch on-resistance and con- the KC-LINK series.
duction losses scale with rated voltage, so there is benefit in

Totem Pole PFC DC Link Active Front End DC Link Vienna Rectifier DC Link

Figure 8: Bridgeless rectifier/PFC front end options

11
 Electrolytic capacitors for energy storage inevitably have a Main power conversion and
finite lifetime and will often be screw-terminal types to allow isolation stage
easy routine replacement. High temperature-rated parts will
typically be used to maximize service life and may be banked In the main conversion stage, efficiency is again a prime
in series/parallel arrangements to allow more common lower concern and resonant converter topologies are popular.
voltage types to be used. In three-level PFC stages such as The variable frequency basic ‘LLC’ converter (Figure 9 left)
the Vienna rectifier, the output capacitor is anyway required suits the lower powers and is uni-directional. Alternatively,
to be split into two series parts. Many parts are available such the CLLC converter (Figure 9 right) allows bi-directional en-
as those from Panasonic, TDK, Vishay and more. ergy flow. LLC converters can be ‘stacked’ (Figure 10) and
implemented as half or full bridges to limit voltage stress,
so that for example, 650V devices can be used on 800V DC
links. This is a workable solution with silicon super junction

DC Link DC Link DC Out

LLC Converter CLLC Converter

DC Out
Figure 9:
Variable frequency Lr
Lr Lr
converter options - Cr Cr
LLC, left and CLLC, Lm
Lm

right

Cr

Full Bridge Interleaved LLC 3 - Level LLC

DC out DC in
DC in
DC out
Figure 10:
LC converters
can be stacked
and interleaved
to reduce
component
stresses

DC Out
DC Link
Full-Bridge Converter
Figure 11:
With SiC
devices, a
traditional,
Cblock
simpler full-
bridge converter
may be viable

12
(Si-SJ) MOSFETs but component count is high and control disproportionately. Core material can also be selected for
very complex. If SiC devices are used however, a simpler, minimum loss at the chosen frequency and for particular
uni-directional, full bridge, hard-switched topology can temperature ranges.
be considered (Figure 11) with 1200V-rated SiC MOSFETs,
without a major efficiency penalty, due to the ultra-fast Capacitors seen in series with transformers in the topolo-
switching and low recovery charge of body diodes. Other gies discussed are often relatively low values but typically
possible and popular topologies are the uni-directional pass very high ripple current and, unlike the DC link ca-
resonant ‘Phase Shift Full Bridge’ or bi-directional ‘Dual pacitor, are subject to high dV/dt. They should therefore be
Active Bridge’. chosen carefully. Polypropylene types from the Panasonic
ECW range or the Vishay MKP1848C range for example, are
All of the topologies mentioned will operate at high fre- good candidates.
quency to enable smaller transformers and filter induc-
tors. This makes selection of the switches critical to keep Output rectifier stages can simply use diodes in uni-
dynamic losses low. Along with headline voltage, current directional converters, as the voltage is high and current
and on-resistance ratings, devices are offered with Figures relatively low for the power levels involved, but efficiency
of Merit (FoMs) to enable comparisons. Notable metrics would suffer. Also, high voltage silicon diodes have rela-
are RDS(on).Qoss for resonant converters, allowing short tively high forward voltage drop and reverse recovery
deadtimes and high frequency operation; RDS(on).Qrr for charge adding to losses. SiC diodes have near zero re-
hard-switched converters such as the TPPFC stage, mini- covery losses but have a high forward drop of well over a
mizing body diode loss, and RDS(on).Qg, affecting light volt and are expensive at high current ratings. For better
load efficiency. Infineon CoolMOS and CoolSiC devices in efficiency, synchronous rectifiers can be used with SiC
discrete and module packages are good choices, as are MOSFETs for best results. For bi-directional operation,
parts from Onsemi. this is anyway necessary. For uni-directional operation,
synchronous rectification gate drive can be as simple as
The isolation transformer in the main conversion stage an auxiliary winding off the transformer, but dead time
steps the DC link voltage up or down to the battery level. should be minimized and tightly controlled, as body diode
For each design, the transformer is most likely to be a cus- ‘commutation’ conduction can cause excessive losses
tom, to extract the maximum efficiency from the topol- during these periods. This is a particular problem with
ogy used. Ferrite cores will be used for high frequencies SiC MOSFETs whose body diodes have a forward drop
and conventional bobbin-wound types are still a practical of several volts. For this reason, active, timed control of
solution, as they facilitate the required creepage, clear- the synchronous rectifiers is preferable and again for bi-
ance and insulation required. For some topologies, such directional operation this is anyway necessary. Similar SiC
as the conventional hard-switched full bridge converter, MOSFETs can be used as synchronous rectifiers to those
it is important to minimize leakage inductance, which can for the primary side switches, as the voltages and current
be achieved by interleaving the primary and secondary ratings will be similar in EV battery charging applications.
windings, at the expense of more complex insulation.
Other topologies such as the LLC, require the equivalent Output capacitors, like DC link capacitors, are likely to
of leakage inductance to function and this can be inte- be a parallel combination of values to achieve sufficient
grated into the transformer by careful physical layout and capacitance at the high voltage rating, with an overall
controlled, loose coupling. In many cases however, this low equivalent series resistance. Ripple current is set by
is a compromise and a separate series discrete inductor the converter topology and ripple voltage is produced
is fitted adding to cost, but improving performance. For by this ripple current across the series combination of
example, in the LLC converter, an integrated inductor of a capacitance and ESR. As the current is likely to be tens of
practical value can force a wider range of frequency vari- amps at maximum load, the capacitors need to be high-
ation for regulation, making EMI control more difficult and performance low ESR types. With high frequency switch-
losses in the transformer higher. Manufacturers of mag- ing, ceramic capacitors can be used with inherently longer
netic components such as Bourns and can assist in the de- lifetime than electrolytic types. However, they have less
sign process, but should be involved early, as transformer capacitance per unit volume and high voltage types are
design involves many trade-offs and opportunities for less common and more expensive.
performance improvement when considered with the sys-
tem as a whole. For example, switching frequency directly
affects core and AC losses in windings so although higher
values might seem to give smaller magnetics size, if losses
are higher, then cooling sizing and costs could increase

13
 Auxiliary power than 5 milliohm contact resistance and certification to UL
60947-1 and EN 61810-10, required for energy storage sys-
As in the AC wall box, an auxiliary power supply is needed tems. For a contactor with ratings at 300A, for example, the
for control, monitoring and communications functions. TE Connectivity 60-311-12 part is specified at 800VDC and
Again, this could be a small module from, for example 200,000 switching cycles at rated load. Solid state circuit
Aimtec and could be a PCB-mount type for DC EV charger breakers could be considered for the application, but volt-
sub-units to around 25kW. The converter input must be age drop with existing technologies could produce unac-
compatible with the three-phase supply, for example ceptable running losses. Back-to-back SiC JFETs however
rated to 305VAC for phase to neutral connection in up to are becoming a feasible alternative.
480VAC three-phase systems. A pre-certified bought-in
part saves design time and development cost over a dis-
crete solution and will often be smaller as well. In DC EV Sensing current
chargers, the AC supply is most likely to be Over-Voltage
Category III or even IV so the auxiliary supply should either As well as sensing gross over-currents and short circuits,
have protection included or have external transient sup- a means of accurately measuring current and voltage is
pression provisioned. necessary. The resultant power and energy use calculated
over a given charge time is what the user is billed for, so
the accuracy of measurement is important. The charger
Output disconnect manufacturer is obliged to err on the side of underbilling,
given component tolerances and sensing errors, so these
As well as galvanic isolation from the utility supply pro- must be minimized in the measurement circuitry. Resistive
vided by the charger transformer, as in an AC wall box, monitoring of current can be accurate but is lossy, so sim-
there must be a means of disconnecting the charger out- ple Hall effect sensors could be employed, but these are
put in case of faults. Relays or contactors are used, with the subject to drift and require a regular calibration routine.
latter more expensive and with higher contact resistance, Differential Hall effect sensors have very good stability and
and because they are interrupting DC, arc suppression is accuracy, typically +/-2.5%, such as the Infineon XENSIV TM
important and the contacts must be rated accordingly to range. Alternatively, a fluxgate sensor with closed-loop
avoid undue degradation and material deposition. Up to compensation and differential measurement can be used
600V/50A, the OMRON G9KB series of PCB-mount relays for ‘parts-per-million’ accuracy at extra cost.
are an option, with the latest arc cut-off technology, less

Conclusion

This guide has just touched on the elements of the design of the power stages in AC and DC EV chargers. For
minimum cost and maximum performance, in volume production, a ground-up design would be the best solu-
tion, leveraging latest technologies. As in any power conversion system, skills would be needed in analog and digi-
tal techniques, thermal design, EMC, safety compliance, control theory, communications and more. This has tech-
nical risk however and development, test and qualification time for a fast DC charger would be measured in tens
of man-years. Modular solutions, perhaps based on ‘reference design platforms’ from companies such as Infineon
with pre-tested sub-units give quicker time to market and lower initial cost to enter the market. As mentioned pre-
viously, it is typical to assemble multiple conversion stages in parallel to achieve high power and as well as provid-
ing redundancy and scalability, this enables the use of more commonly available components for the power, cur-
rent and voltage levels encountered. As a single example of many, DC current sensors rated to more than 55A for
a 22kW/400V charger sub-unit are commodity parts available as PCB surface-mount components sized 8 x 8mm,
whereas a single sensor for 150kW/400V rated at 375A would be hard to find, expensive and chassis-mount, adding
to assembly costs.

In a following design guide, we will consider the control circuitry necessary for operation of the power stages and
for the charger functionality for the end user and energy supplier. Further guides will address connectivity and
software.

14
 Questions?

If you have any questions about the

topics discussed in this guide, or you’re
ready to take the next step with your
design, reach out to our EV charging
experts - they’re some of the best in the
industry.

References

[1] 
https://www.iea.org/reports/global-ev-outlook-2022/trends-in-charging-infrastructure

[2] https://www.charin.global/
http://www.ev-chargi.ng