Multi-loop Sliding-Mode Control for a Battery

Charger Using a Quadratic Buck Converter

Oswaldo López-Santos∗ , F. Flores-Bahamonde† , C. A. Torres-Pinzón‡ , Reham Haroun∗ , Hugo Valderrama-Blavi∗ ,
Juan A. Garriga§ and Luis Martı́nez-Salamero∗
∗ Group of Automatic Control and Industrial Electronics, Rovira i Virgili University, Tarragona, Spain
† Energy Transformation Center, Faculty of Enigineering, Universidad Andres Bello, Chile
‡ Faculty of Electronic Engineering, Universidad Santo Tomás, Colombia
§ Department of Computer Science and Industrial Engineering, University of Lleida, Lleida, Spain.
email: [email protected]

Abstract—In this paper, the quadratic buck converter (QBC) technology refers to the so called fast-chargers and ultrafast-
is proposed as competitive alternative to implement a battery chargers.
2023 IEEE 8th Southern Power Electronics Conference (SPEC) | 979-8-3503-2112-8/23/$31.00 ©2023 IEEE | DOI: 10.1109/SPEC56436.2023.10408511

charger. Since, QBC is a high order system, the required control Battery chargers can be classified as isolated or non-isolated
is designed to follow the conventional constant-current constant-
voltage regime by means of three loops. Namely, i) an inner-loop and can be powered by alternating current (AC chargers) or
operating in sliding mode to control the current of the closest direct current (DC chargers). The AC chargers normally are
inductor to the input port providing the properly stability of composed of two power conversion stages, one performing
the system, ii) a first outer loop designed to regulate the battery rectification ensuring a unitary power factor and other perform-
voltage providing the reference of the inner loop, and finally iii) ing regulation of the DC output voltage and current. In turn,
a second outer loop to regulate the battery current modifying
the reference of the voltage loop. Proportional Integral (PI) the DC chargers can be implemented using a single stage of
controllers are used in both outer loops. Simulation results are DC-DC conversion which is fed by a primary system providing
presented validating the theoretical study. a regulated DC output. The selection of the adequate DC-DC
power converter in both cases (output stage) depends on the
Index Terms—Sliding Mode Control, Quadratic Buck Con- constraints of the application mainly related with the input and
verter, Battery charger
output voltage ranges, the rated power and the requirement of
galvanic isolation. For instance, the quadratic buck converter
I. I NTRODUCTION (QBC) may be a suitable candidate in applications where: a)
the DC input voltage is various times higher than the battery
Battery energy storage is widely used in mobile systems
array voltage; b) galvanic isolation is not mandatory because it
operating with a certain temporary energy autonomy such
is provided by a previous conversion stage; and c) the required
as cell phones, drones and electric vehicles (EVs) [1]- [2].
output current is high. The main reason is the ability of this
For these applications, the batteries are replenished using
converter to provide a high step-up current gain without the
battery chargers which are composed of a unidirectional power
need of transformers [5]- [6].
converter and its control system. The main function of these
The energy processed by a battery charger is transferred
chargers is to provide the necessary energy to the batteries
from a primary power source to the battery following a
favoring a good utilization to increase its lifetime. The voltage
charging method. Most of the methods perform the charge in a
and power levels of the chargers depend on the application
sequence of stages in which the charger applies a constant DC
and are defined by the size of the battery array and the
current or a constant DC voltage. The classical method consist
expected charging time. For example, battery chargers for the
in applying a constant current to the battery while it reaches a
electric vehicle application can provide between 100 and 800
predefined voltage level and then apply that voltage until the
V with an output power of up to 350 kW. Depending on
battery is completely charged. This method is well known as
the relation between the battery capacity and the transferred
the constant current – constant voltage (CC-CV) method [7].
power, the charging time varies from fractions of an hour to
Other methods can use multiple stages of constant current with
several hours [3]- [4]. The advances increasing the maximum
increasing references like in the case of [8] and [9].
output power of the chargers in the market has allowed to
Sliding mode control has been used in power converters
replenish the battery of an EV in less than fifteen minutes. This
proving to be a superior alternative in terms of simplicity of
implementation, speed of response and robustness [10]–[12].
This work has been partially sponsored by the Spanish Ministerio de
Ciencia e Innovación under grants MSCA IF EF-ST 2020 / PCI2021- This type of control can be applied using nested loops (multi-
122066-2B and PID2019-111443RB-I00, and partially by the Chilean Agencia loop control) in which an inner loop ensures operation in a
Nacional de Investigación y Desarrollo (ANID) through projects FONDECYT sliding mode while an outer loop impose regulation or tracking
Iniciación 11220863 and SERC Chile (CONICYT/FONDAP/15110019) as
well as by Universidad Santo Tomás through grant project no. BOG-2023- in a variable of interest [13] - [14]. A key aspect in the multi-
AI015. loop implementation is the choice of the variables involved in

Authorized licensed use limited to: UNIVERSIDAD SANTO TOMAS. Downloaded on April 16,2025 at 17:55:37 UTC from IEEE Xplore. Restrictions apply.
 S1
L1 L2

iL1 (t) iL2 (t)

D2 C1 Rbat
u(t)
vg (t) + D3 C2
−
+
D1 vbat
−

Fig. 1: Schematic circuit diagram of a quadratic buck converter feeding a battery type load.

the sliding surface of the inner loop since the stability depends of the outer loops to impose the desired CC-CV battery
on it. As it can be verified in [15], [16] and [17], despite the charging regime. Finally, simulation results are presented in
complexity of the power converter, the use of a single inductor Section V and conclusions are presented in Section VI.
current in the sliding surface is a simple and effective choice.
Also, in [6] although the use of the current of an inductor
S1
in the sliding surface does not allow a direct stabilization of L1 L2

the converter because the constant power nature of the load, it iL1 (t) iL2 (t)
facilitates the stabilization by the outer loop. Conventionally, D2 C1 Rbat
u(t)
the outer loop of the control system is configured using a +
vg (t) D3 C2
proportional-integral (PI) controller which offers the solution −

with the best cost-benefit trade-off. D1

+
vbat
−
This paper proposes a multi-loop control method based
on sliding mode to control the QBC as a battery charger.
(a) ON-state.
The proposal involves an inner loop of sliding mode control
S1
and two nested outer regulation loops to perform the above L1 L2

mentioned CC-CV charging method. The sliding surface iL1 (t) iL2 (t)
of the inner loop involves only the inductor current closer D2 C1 Rbat
u(t)
to the input port of the converter. The first outer loop +
vg (t) D3 C2
implemented using a PI controller regulates the battery −

voltage at a reference value which is fixed for the constant D1

+
vbat
−
voltage interval. During the constant current interval, the more
external outer loop also implemented using a PI controller
modifies the voltage reference in order to impose the charging (b) OFF-state.
current limitation. The synthesis of the controllers is assisted
by the Control Systems Designer of MATLAB considering Fig. 2: Quadratic buck circuit configurations.
the basic fundamentals for design of cascade controllers.
The main contributions of this paper are: 1) the theoretical II. S WITCHED - MODEL OF THE QBC
analysis of the proposed control applied to the quadratic The electrical circuit of the QBC proposed for battery
buck converter feeding a battery, which is represented by a charging is shown in Fig. 1. As it can be noted, the circuit
voltage source in series with a resistor, thus extending the is composed by the inductors L1 and L2 , the capacitors C1
work presented in [6], and 2) the application of the proposed and C2 , the diodes D1 , D2 and D3 , and the controlled switch
multi-loop controller for the quadratic buck converter in the S1 which is governed by the control signal u(t). As stated, this
battery charging application extending the work reported circuit provides a wider conversion ratio than the conventional
in [17]. The rest of the paper is organized as follows: Section buck converter, since the dc gain between the output voltage
II presents the circuit analysis used to derive the model of the and the input voltage is a quadratic function, described by
quadratic buck converter. After that, section III presents the
theoretical analysis of the inner current control loop developed VC2
using sliding mode control. Section IV develops the synthesis = D2 (1)
Vg

Authorized licensed use limited to: UNIVERSIDAD SANTO TOMAS. Downloaded on April 16,2025 at 17:55:37 UTC from IEEE Xplore. Restrictions apply.
where VC2 and Vg are the average values of the output and The coordinates of the equilibrium point are given by:
input voltage, and D is the duty cycle of the converter which
is defined as the average value of the control signal u(t). As ∗ T
x∗ = i∗L1 , vC
∗
, i∗L2 , vC
 
1 2
(6)
can be observed, the converter is also connected to a battery q
modeled as the controlled voltage source vbat connected in where i∗L1 = k ∗ , vC∗
= Vbat Vg , i∗L2 = k ∗
p Vg
and
√ 3 1
√ Vbat
series with the resistance Rbat . Although the resistance Rbat Vbat +k∗ Rbat Vg
∗
slightly varies during the charging process, in the subsequent vC 2
= Vo = √
V
.
bat
analysis it is considered constant.
The QBC has only one active switch to perform the voltage Taking into account the equations (4), (5), and (6), the ideal
conversion and its operation is equivalent to have two cascaded sliding dynamics can be linearized as:
buck converters sharing the same control signal. Thus, through
the control signal u(t) = {0, 1} two configurations are de
vC 1
k + c11 ddtk
= a11 veC1 + a12eiL2 + a13 veC2 + b11 e
e
obtained for operation in continuous conduction mode (CCM). dt
Figure 2(a) shows the ON−state configuration associated to de
iL2
= k + c21 ddtk
a22 veC1 + a21eiL2 + a23 veC2 + b21 e (7)
e
u(t) = 1, and Figure 2(b) the OFF−state configuration, dt
corresponding to u(t) = 0.
de
vC 2
The state vector of the converter is given in (2). Operating k + c31 ddtk
= a33 veC1 + a32eiL2 + a33 veC2 + b31 e
e
dt
in CCM, the dynamics of the QBC can be modelled by means
of the bilinear system equation (3).
 T The parameters aij , bij , cij , (i, j ∈ {1, 2, 3}) are defined as:
x(t) = iL1 (t) vC1 (t) iL2 (t) vC2 (t) (2) q
∗
a11 = − √k a12 = − C11 VVbat
C1 Vg Vbat g
diL1 v vg
dt
= − LC11 + L1
u
a13 = 0
dvC1 i L1 i L2
dt
= C1
− C1
u b11 = 1 L1 √
c11 = − C k∗
C1 1 Vg Vbat
(3)
diL2 vC1 vC2
= u −
q
2 Vbat
dt L2 L2 a21 = L2 Vg a22 = 0

dvC2 i L2 (vC2 −vbat ) a23 = − L12

dt
= C2
− Rbat C2
q
L1 Vbat
b21 = 0 c21 = L2 Vg

III. S LIDING MODE CONTROL 1

a31 = 0 a32 = C2
The switching surface considered in this application is
defined as S(x) = iL1 − k(t), where k(t) is provided by
a33 = − Rbat1 C2
an outer control loop. Introducing the invariance conditions
S(x) = 0 and dS(x)
dt = 0 in (4) results in the equivalent control b31 = 0 c31 = 0
ueq (x) bounded as follows:
" #
L1 dk
dt + vC1
0 < ueq (x) = <1 (4) By applying the Laplace transform to (7), the transfer
Vg
function given by (8) is obtained:
Now, substituting u by ueq (x) in (3) results in the following
ideal sliding dynamics:
ṽC2 (s) β2 s 2 + β1 s + β0
Gvk (s) = = 3 (8)
dvC1 i i v k̃(s) s + α2 s2 + α1 s + α0
dt
= k
C1
− CL11 vLg2 dk
dt
− CL12 vCg1 = g1 (x)
2 where
diL2 L1 vC1 dk vC v
dt
= L2 vg dt
+ 1
L2 vg
− LC22 = g2 (x) (5)
β2 = a32 c21 β1 = a32 (c11 a21 − a11 c21 )
dvC2 i L2 (vC −vbat ) β0 = a32 b11 a21 α2 = −a11 − a33
dt
= C2
− R2bat C2 = g3 (x)
α1 = (a11 a33 − a23 a32 − a12 a21 )
α0 = (a11 a23 a32 + a12 a21 a33 )

Authorized licensed use limited to: UNIVERSIDAD SANTO TOMAS. Downloaded on April 16,2025 at 17:55:37 UTC from IEEE Xplore. Restrictions apply.
 Fig. 3: Proposed battery charger control architecture.

IV. BATTERY CHARGING CONTROL TABLE I: System parameters

A. Control architecture and charging method Symbol Parameter Value
P Rated power 540 W
The control scheme depicted in Fig. 3 is proposed to Vo Output voltage 40 -54 V
apply the CC-CV charging method based on sliding mode Vbat Battery voltage 40 -54 V
Ibat Output current 10 A
control. This is achieved by using an inner loop based on a Vg Input voltage 380 V
sliding mode control and two nested outer loops regulating Rbat Battery resistance 0.1 Ω
the voltage and current of the battery. Specifically, the L1 Input inductor 1.2 mH
L2 Output inductor 0.3 mH
sliding mode controller acting on the inductor current iL1 C1 Input capacitor 0.3 mF
has been implemented by means of an hysteresis comparator C2 output capacitor 0.1 mF
which enforces a finite switching frequency. The reference
of this control loop, k(t), is given by a first outer loop,
which is composed of a classical PI compensator that acts
minimizing the output voltage error. The reference of this Kpn s + Kin
loop is obtained as the sum of the desired voltage for the
Gcn (s) = (9)
s
CV interval denoted as Vref and the contribution of the most
outer loop regulating the current of the battery and denoted as where n = {i, v} indicates the voltage or current loop,
Vcomp . Since this current is a function of the output voltage respectively.
and the instantaneous static impedance of the battery, the The specifications for the voltage regulation loop are a
current limitation during the CC interval can be provided settling time of 0.1 s without overshoot. The transfer function
by another PI compensator acting on the battery current used for the design is Gvk (s) (Expression (8)). Thus, the
error and tracking the reference Iref . As can be observed in closed loop transfer function of the voltage regulation loop
the scheme of Fig. 3, the current error is saturated at zero, is obtained as follows:
thus, the contribution of this loop through Vcomp takes only
negative values. Gvk (s)Gcv (s)
Gvbat (s) = (10)
1 + Gvk (s)Gcv (s)
B. Synthesis of the controllers The gains of the voltage regulation loop were obtained as
In order to design the PI controllers of the respective Kpv = 0.34 and Kiv = 265.78. Considering that the more
nested loops, the parameters of the converter are described in outer loop needs to be at least ten times slower than the
Table I. As a result, the parameters of the third order transfer subsequent nested loop to satisfy conventional constraints of
function of expression (8) become: cascade controllers design, the specifications for the current
regulation loop are a settling time of 1 s without overshoot.
β2 = 13.298 × 103 β1 = −1.052 × 106 The output voltage of the converter is applied to the battery
β0 = 73.879 × 109 α2 = 100 × 103 model in which the current is defined by:
α1 = 43.705 × 106
α0 = 248.25 × 109 vC2 − vbat
In this way, using the output to control transfer function and ibat = (11)
Rbat
following a conventional design method, the PI controllers of
the first and second outer loops can be designed. The syn- Then, the transfer function used for the design is as follows:
thesis of these controllers is assisted by the Control Systems
Designer of MATLAB© by using the root locus design method. Gvbat (s)
The PI controllers follow the form: Giv (s) = (12)
Rbat

Authorized licensed use limited to: UNIVERSIDAD SANTO TOMAS. Downloaded on April 16,2025 at 17:55:37 UTC from IEEE Xplore. Restrictions apply.
 The gains of the current regulation loop were obtained as
Kpi = 0.0034 and Kii = 0.6. The closed loop transfer
function of the current regulation loop is obtained as follows:

Giv (s)Gci (s)

Gibat (s) = (13)
1 + Giv (s)Gci (s)
It is worth to mention that although the order of the
resulting dynamics of this control loop is five, it is enforced
to approximately have the dynamics of a first order system.
V. S IMULATION RESULTS
The power converter circuit and battery models have been (a) iL1
simulated using PSIM© software according to the circuit
scheme described in Fig. 1. The first simulation has been
performed in order to verify the stable operation of the
converter. The parameters of the converter defined in Table
I are used. The sliding mode controller is implemented by
means of a hysteresis comparator composed of two simple
comparators and one S-R flip-flop and defined for a band of
± 0.5 A. Using this implementation, the switching frequency
varies depending of the operation point but is constrained to
dozens of kHz. Measurements are acquired when the output
voltage is 54 V and a value of 53 V in reached in the variable
vbat of the battery model. Fig. 4 shows some high frequency
cycles of the converter variables detailing the ripples and (b) iL2
average values.
The second simulated scenario simulated has been designed
to verify the correct operation of the proposed control, i.e.,
following the CC-CV charging method. The battery model
considers a capacity of 2.5 Ah, and the internal voltage vbat
evolves as a function of the state of charge from 0 % to 100 %.
As it can be seen in Fig. 5, in the first interval, the system
operates in the CC interval following a set-point of 10 A
provided by the action of the outermost control loop. During
this interval of approximately 850 s, the voltage applied to
the battery increases from 42 V until it reaches the voltage
reference value of 54 V. Thereafter, this voltage reference is
maintained constant until the battery is fully charged while the (c) vC1
current decreases exponentially until zero. The last interval has
a duration of about 300 s.
VI. C ONCLUSIONS
In this paper, the quadratic buck converter has been studied
in the application of battery charging. Due to the higher
order of the system, a multi-loop controller using three nested
loops has been proposed involving one sliding mode current
controller and two conventional PI controllers. The selected
sliding surface uses the inearest nductor current to the input
port because the individual selection of the other variables
leads to an unstable behavior and the use of a linear combi-
nation of more than one variable implies added complexity
(d) vC2
and does not improve the achieved performance. Simulated
results show that the studied converter operates properly
Fig. 4: Steady-state current and voltage waveforms.
following the conventional CC-CV charging method by using
the proposed control scheme. A laboratory prototype is under
implementation in order to provide experimental validation.

Authorized licensed use limited to: UNIVERSIDAD SANTO TOMAS. Downloaded on April 16,2025 at 17:55:37 UTC from IEEE Xplore. Restrictions apply.
 Fig. 5: Battery voltage and current during a complete CC-CV changing cycle.

Further research contemplates the study of the interleaving [12] D. Zambrano-Prada, A. El Aroudi, L. Vázquez-Seisdedos, and
operation [18] of this converter for battery charging. L. Martinez-Salamero, “Polynomial sliding surfaces to control a boost
converter with constant power load,” IEEE Trans. Circuits Syst. I: Reg.
Papers, vol. 70, no. 1, pp. 530–543, Jan. 2023.
R EFERENCES [13] E. Vidal-Idiarte, C. Carrejo, J. Calvente, and L. Martı́nez-Salamero,
“Two-loop digital sliding mode control of dc–dc power converters based
[1] J. Cao and A. Emadi, “Batteries need electronics,” IEEE Ind. Electron. on predictive interpolation,” IEEE Trans. Ind. Electron., vol. 58, no. 6,
Mag., vol. 5, no. 1, pp. 27–35, Mar. 2011. pp. 2491–2501, June 2011.
[2] K. Pham, J. Leuchter, R. Bystricky, M. Andrle, N. Pham, and V. Pham, [14] O. Lopez-Santos, G. Garcia, L. Martinez-Salamero, J. Avila-Martinez,
“The study of electrical energy power supply system for uavs based on and L. Seguier, “Non-linear control of the output stage of a solar
the energy storage technology,” Aerospace, vol. 9, no. 9, pp. 500–529, microinverter,” Intl. J. Control, vol. 90, no. 1, pp. 90–109, 2017.
Sep. 2022. [15] O. Lopez-Santos, L. Martinez-Salamero, G. Garcia, H. Valderrama-
[3] M. Safayatullah, M. Elrais, S. Ghosh, R. Rezaii, and I. Batarseh, Blavi, and T. Sierra-Polanco, “Robust sliding-mode control design for
“A comprehensive review of power converter topologies and control a voltage regulated quadratic boost converter,” IEEE Trans. Power
methods for electric vehicle fast charging applications,” IEEE Access, Electron., vol. 30, no. 4, pp. 2313–2327, Apr. 2015.
vol. 10, pp. 40 753–40 793, 2022. [16] Z. Chen, “Pi and sliding mode control of a cuk converter,” IEEE Trans.
[4] S. Rivera, S. Kouro, S. Vazquez, S. M. Goetz, R. Lizana, and E. Romero- Power Electron., vol. 27, no. 8, pp. 3695–3703, Aug. 2012.
Cadaval, “Electric vehicle charging infrastructure: From grid to battery,” [17] O. Lopez-Santos, D. Zambrano Prada, Y. Aldana-Rodriguez,
IEEE Ind. Electron. Mag., vol. 15, no. 2, pp. 37–51, Jun. 2021. H. Esquivel-Cabeza, G. Garcia, and L. Martinez-Salamero, “Control of
[5] J. Morales-Saldana, J. Leyva-Ramos, E. Carbajal-Gutierrez, and a bidirectional cuk converter providing charge/discharge of a battery
M. Ortiz-Lopez, “Average current-mode control scheme for a quadratic array integrated in dc buses of microgrids,” Applied Computer Sciences
buck converter with a single switch,” IEEE Trans. Power Electron., in Engineering. WEA 2017. Commun. Comput. Inf. Sci., vol. 742, pp.
vol. 23, no. 1, pp. 485–490, Jan. 2008. 495–507, Aug. 2017.
[6] C. Torres-Pinzón, F. Flores-Bahamonde, J. Garriga-Castillo, [18] S. Ozeri, D. Shmilovitz, S. Singer, and L. Martinez-Salamero, “The
H. Valderrama-Blavi, R. Haroun, and L. Martı́nez-Salamero, “Sliding- mathematical foundation of distributed interleaved systems,” IEEE
mode control of a quadratic buck converter with constant power load,” Trans. Circuits Syst. I: Reg. Papers, vol. 54, no. 3, pp. 610–619, Mar.
IEEE Access, vol. 10, pp. 71 837–71 852, 2022. 2007.
[7] B.-Y. Chen and Y.-S. Lai, “New digital-controlled technique for battery
charger with constant current and voltage control without current feed-
back,” IEEE Trans. Ind. Electron., vol. 59, no. 3, pp. 1545–1553, Mar.
2012.
[8] J. Deng, S. Li, S. Hu, C. Mi, and R. Ma, “Design methodology of llc
resonant converters for electric vehicle battery chargers,” IEEE Trans.
Vehicular Technol., vol. 63, no. 4, pp. 1581–1592, May 2014.
[9] D. Lyu, T. Soeiro, and P. Bauer, “Impacts of different charging strategies
on the electric vehicle battery charger circuit using phase-shift full-
bridge converter,” in Proc. IEEE 19th Intl. Power Electron. Motion
Control Conf. (PEMC), 2021.
[10] S.-C. Tan, Y. Lai, and C. Tse, “General design issues of sliding-mode
controllers in dc–dc converters,” IEEE Trans. Ind. Electron., vol. 55,
no. 3, pp. 1160–1174, Mar. 2008.
[11] L. Martinez-Salamero, A. Cid-Pastor, A. ElAroudi, R. Giral, J. Calvente,
and G. Ruiz-Magaz, “Sliding-mode control of dc-dc switching convert-
ers,” Proceedings of IFAC World Conference, Milano, Italy, vol. 44,
no. 1, pp. 1910–1916, 2011.

Authorized licensed use limited to: UNIVERSIDAD SANTO TOMAS. Downloaded on April 16,2025 at 17:55:37 UTC from IEEE Xplore. Restrictions apply.