Modulation techniques for AC/DC converters in ultrafast battery charger applications

11 
ABSTRACT 
Concerning today’s exigencies of electrical mains, the possibility to build a reliable 
technology of electric energy distributed storage in conjunction with the growing market of 
electric and hybrid vehicles (EV and HEV) address the actual producers’ interest mostly on 
batteries and charger station. Indeed, the UFC (ultrafast charger) technology has been 
becoming dominant in this ambit for last years, because these devices guarantee high density 
of power and a variety of combinations in terms of layout (bidirectional or unidirectional 
applications). 
According to the previous motivations, this thesis regards the report of a project of an 
ultrafast charger conceived by PEIC (Power Electronics Innovation Center) of the Turin 
Polytechnic, patented in symbiosis with “VISHAY Semiconductor Italia”, a leader company 
in the electronic and electric components market.  
The charger consists of two different converters: the first one is an AC/DC (active front end) 
converter while the second one a DC/DC converter (LLC resonant). This paper focuses on 
the AC/DC stage and highlights how to set up the most efficient control strategy for the 
structure. In particular, in this AC/DC application an AFE (active front end) converter has 
been realized. This topology represents an example of multi-level converter, with 3 output 
connections (p, m, n) instead of 2 (1, 0). It leverages bidirectional T-type switches for mid-
point connection.   
The principal points of the thesis deal with the advantages guaranteed by the converter 
(unitary cosφ, low impact on absorbed current by the grid so low THD and TDD, galvanic 
isolation of the battery, possibility to charge more vehicles contemporaneously, power 
sharing strategy, and so on) and also with the major issues affecting it (mid-point current 
control and DC-link balance).  
The main problem concerning the control of the structure regards the mid-point balancing. 
A suitable modulation technique and a proper voltage control are presented as methods to 
overcome this problematic. This modulation technique is known in technical literature as 
ZMPC (Zero-Mid-Point-Current) PWM and it is feasible for this application since it acts by 
nulling the periodic value of the mid-point current on the central connection of the DC-link.  
In the end, several considerations, supported by proper figures and comments, are presented 
in this paper and a final section reports further conclusions about the treated topics.
12 
INTRODUCTION 
The growing interest in electric mobility (HEV, PEV and EV) in conjunction with the 
possibility to achieve demand-side-management, have been the motivations leading the 
efforts of power electronic researches on ultrafast charger technologies for last years, in order 
to satisfy different market and users necessities. The main pointed out topics of interest 
regard: 
 the possibility to build an new electric infrastructure for users’ direct grid 
connection, in order to permit them to recharge their vehicles anywhere (like petrol 
stations) and in the lowest possible time (so with the highest possible current flow); 
 the necessity, for the grid owner, to create a real high capacity distributed storage 
that, in specific circumstances, could be used to guarantee a bidirectional power flow 
(the charger is as a load for the grid in case of vehicle recharge but it may act as a 
generator in case of electric energy peak demand). 
 the obligation to meet the EMI requirements for these kind of electronic devices; 
 the responsibility of today’s generation to evaluate the possibility for a new way of 
intending mobility as electric and sustainable (zero emission mobility challenge); 
The main issue tied with the proposal targets regards the huge required power density of this 
kind of chargers. They have to be not only compact but also powerful (up to 300 kW of 
power provided per single charger). As a first sight, it is not trivial to highlight that, in almost 
all the cases, these kind of devices are made up of more than a single unit. Usually, 5 or more 
units are connected in parallel and in this way the power absorbed by the single unit can be 
significantly lower than the whole power absorbed by the charger. Without this shrewdness, 
the realization of the charger could even be impossible.  
A simple solution might be searched in MV grid connection (up to 1 kV) but at the moment 
the state of art of semiconductors for such kind of applications can be really complex, due 
to the voltage rating of the components (IGBTs, MOSFETs switches and diodes). 
Consequently, the majority of producers prefer to deal with LV grid connected converters, 
providing then a higher current flow to fulfill the power request. Nevertheless, the power 
demand of these devices cannot be reached without the use of a specific topology for the 
converter and this is the reason why the interleaved topology is the preferred one. It is the 
most diffused and suitable since enables both to obtain the condition of LV grid connection
13 
and high current request in order to achieve the desired power demand. UFC (UltraFast 
Charger) is the name used in literature to describe these kind of devices.  
The UFC has to be connected in any time both to the grid and to loads (the vehicles to 
recharge). As a consequence, the UFC is indeed a back-to-back converter, with an AC/DC 
interface (for grid-converter connection) and a DC/DC stage (for converter-load connection). 
Between the stages of the converter is interposed a voltage DC-LINK in order to guarantee 
continuity of voltage for loads and also protection from voltage dips and swells originated 
by the grid.  
The purpose of the thesis is to provide an accurate and precise study, both simulative and 
experimental, of the AC/DC stage of an UFC, with a specific focus on the control of the 
converter, considering the main issues of the case, for example: 
 the optimal choose of the converter to use among different alternatives in 
literature (in terms of layouts, components and technologies); 
 the necessity to guarantee the immunity of the converter from the disturbances 
coming from the grid (in order to satisfy the EMI standards for the phase 
current absorbed by the grid in terms of TDD, Total Demand Distortion); 
 the critical point of the control of the converter; 
 the performances that may be obtained by the converter using a variety of 
different modulation techniques; 
In addition, a conclusive chapter is proposed to summarize the activities of the thesis and to 
provide further considerations about the treated topic and their eventual future scenarios.
14 
CHAPTER 1: STATE OF ART OF UFC 
1.1 Advantages of multi-level conversion  
Today’s state of art of UFC charger permits to recharge the battery pack of an electric vehicle 
in a reasonable time (400 km of autonomy in 10 minutes of recharge). In order to achieve 
this result, it is necessary to have a huge power density (at least 300 kW), to guarantee a 
sufficient current flow (up to 300 ARMS). As already suggested in the introduction, the use 
of MV grid (up to 1 kV) is not really feasible, due to the voltage rating of switches and 
semiconductors properties. For this reason, the use of classical 2-levels topologies for the 
converters within the charger could result in an inefficient structure or even in a not 
achievable one in certain cases. As a consequence, preferred topologies for these kind of 
devices are oriented to multi-level converters and interleaved topology. These are the two 
main features of the UFC that will be described in the case study of this thesis. In this 
paragraph and in the following one, it will be shown both the advantages of multi-level 
conversion and the mainly used topologies for this kind of applications. As a useful example, 
let introduce firstly, same practical measure of grid/battery connection in terms of voltage 
and current values [7]: 
 existent electric vehicle batteries may have a wide voltage range, usually: 
�� �������� =200÷1000 �� 
 considering a boost-type AC/DC converter topology (as it is normally the case), the 
actual value of voltage could float among 0.9 pu and 1.1 pu of the nominal value of 
voltage, thus: 
�� �������� = (400±10%) �� ������ ������������ ���� �� ���� ,������ =√3∙400∙0.9 �� ������ ≅625 �� ������  
The �� ����
 choice is of utmost importance, since it influences the voltage rating of the 
semiconductor devices and the DC/DC converter topology; 
 the maximum battery charging current impacts the DC/DC converter design and it 
is limited by the connector, according to the CCS Combo standard to: 
�� �������� = 350 �� ������ ���� 1000 �� 
 Thus, in this case: 
�� �������� = 565 �� ������ ���� 625 �� ������ 
 Where the last value of current (�� �������� ) represents the actual current absorbed by the 
grid at the fixed voltage of Vdc,min (for the assigned power of 350 kW).
15 
The power electronics supply of high-power electrical systems from the three-phase ac 
mains to the load is usually carried out in two stages, i.e., the mains ac voltage is first 
converted into a dc voltage and then adapted to the load voltage level with a DC/DC 
converter (with or without galvanic isolation). Therefore, the typical grid-load connection 
consists in [7]: 
 grid/transformer interface; 
 AC/DC stage (conversion stage); 
 Voltage DC - link; 
 DC/DC stage (for battery/grid connection); 
Fig. 1: Example of grid-battery connection. 
In the simplest case, the rectifier consists in a unidirectional three-phase diode rectifiers with 
capacitive smoothing of the output voltage and inductors on the AC or DC side (assuring 
voltage and current continuity for loads). Despite the low complexity and high robustness 
(no control, sensors, auxiliary supplies, or electromagnetic interference EMI filtering) of this 
concept must, this structure has to face against the disadvantages of relatively high effects 
on the mains and an unregulated output voltage directly dependent on the mains voltage 
level. Then, the AC/DC stage has to be chosen carefully in order to guarantee both load and 
grid specifications, for example: 
 unitary power factor (cosφ ≈ 1); 
 low THD and TDD (TDD < 0.05, in order to assume current waveform as 
sinusoidal); 
 high efficiency (so, reduced losses in nominal working conditions); 
 high power density (reduced devices and reactive components encumbrance). 
Consequently, a proper AC/DC structure has to be patented in order to match this 
requirements. The AFE converter (Active Front End) is a AC/DC converter derived from the 
one-phase PFC (Power Factor Corrector) that permits to mitigate the harmonic distortion of 
the current absorbed by the converter from the grid, in order to eliminate the disturbances on
16 
the load reducing further the dimensions and the cost of the reactive components (capacitors 
and inductors) necessary for the filtering operations. 
 
A primal split among AFE applications for UFC regards low frequency solutions and high 
frequency solutions. In both cases, it is necessary to provide an effective galvanic isolation 
between the grid and the load, definitely [7]. 
 
Fig. 2: LF solution with galvanic isolation. 
 
Fig. 3: HF solution with galvanic isolation. 
In LF applications the isolation is provided through the use of a dedicated isolation 
transformer that permit to isolate the load from the grid and further it keeps constant the 
voltage at the AC/DC stage input (in an independent grid manner). These kind of devices 
have the drawback of being bulky. Anyway, they enable the use of a non-isolated DC/DC 
converter stage (reducing costs and complexity of the system). On the contrary, in HF 
application, the isolation is provided by the DC/DC stage directly, since the system layout 
can be “transformer-less”. This is the direction followed by the majority of UFC constructor, 
since it guarantees the main advantages (low costs, high efficiency, compactness and so on). 
A high-frequency transformer is a special transformer used in HF application (fsw almost of 
kHz or even MHz) to guarantee galvanic isolation between grid and load or between two 
different levels of a multi-level converter.
17 
These kind of devices are characterized by: 
 reduced encumbrance in comparison with traditional LF transformer, due to the 
fact that for transformer the RMS value of the voltage depends on: 
�� ������ =
2�� √2
��������  (1.1) 
the flux density B [T], the frequency f [Hz], the surface S the flux density insists on 
[mm
2
] and the numbers of turns of the coil N [-]; therefore, at constant flux density 
B, if frequency f increases, the surface S decreases proportionally. 
 the iron losses Pfe are reduced thanks to the creation of internal air gaps within the 
conductors (sintering ferrite), that mitigate the circulation of parasitic current; 
 high power density (thanks to a reduced surfaces S). 
These are the major motivations that have been orienting the developing of AFE technology 
for last years. The AFE converters are examples of multi-level converters. Multi-level 
converters are converters in which the base layout is not represented by the canonical cell 
for hard commutation (combination of switch and diode for freewheeling operations) but 
they consist of the combination of more devices with a specific topology.  
The NPC circuits (isolated or non-isolated), the Vienna rectifier and the T-type rectifier are 
examples of 3-levels converters. The AC/DC stage is configured as follow 
 
Fig. 4: Model of a 2 levels and a 3 levels converter. 
With the use of a 2 levels converter it is possible to control switches only up and down (thus 
1 or -1 as modulation index or 1 and 0 as bits). The chance to have more possible states for 
switches (for example in a 3 levels converter switches state can vary between up, down and 
mid, thus p, n and m state) enables to increase the resolution of regulation in amplitude.