International Journal of Applied Engineering Research ISSN 0973-4562 Volume 12, Number 7 (2017) pp.

1329-1333
© Research India Publications. http://www.ripublication.com

MCU-based Battery Management System for Fast Charging of IoT-based

Large-Scale Battery-Cells

Meng Di Yin, Jiae Youn, Jeonghun Cho, and Daejin Park*

School of Electronics Engineering, Kyungpook National University

Daehak-ro 80, Buk-gu, Daegu, Republic of Korea.

Abstract diagram, model implementation, finite state machine and

experiment environment are introduced in section II. Finally,
There is a growing demand for various applications requiring the implementation results and conclusion are discussed in
the large-scale battery cells, such as electrical vehicles and section III and IV, respectively.
industrial appliances. The Traditional constant current,
constant voltage (CC-CV) charging method cannot satisfy the
demand of fast charging. Pulse based battery-charging
approaches have been proved as fast charging method for the PROPOSED ARCHITECTURE
battery cells. Our work is focusing on build a testing
A. Schematic diagram for charging test
framework to evaluate the performance of pulse based
charging methods. In order to facilitate the progress of Using benefit from C code-generation features of the Simulink
evaluation, the charging control program is developed in and state flow model from MATLAB, the corresponding C
MATLAB environment and then the actual charging tests are code is obtained through automatic code generation from state
done on the customized hardware circuit board. As a case flow model. All of the low-level driver is developed in the
Arduino IDE (integrated development environment), such as
study, a newly-designed dynamic frequency and duty cycle
current sensor acquisition, temperature sensor acquisition, SPI
charging algorithm is evaluated by using the proposed testing communication control, PWM control and timer control code.
framework. The low level driver code is integrated with MATLAB
generated code and then they are download to the Arduino
Keywords: battery cells; electrical vehicle; frequency; duty UNO board [9].
cycle; testing framework; customized hardware
By using a LTC6802-2 battery stack monitoring chip [10], this
chip can monitor battery terminal voltage not more than 12
cells connected in serial. One other thing to note is that this
INTRODUCTION AND MOTIVATION chip must be powered at least 4 cells. In this project, 4 cells
Reducing gas emissions from cars is critical to management of are used and the registers of this chip need to be controlled by
air pollution. Electrical vehicle (EV) has the advantage of zero a microcontroller via SPI. In the experiment, the model
tailpipe emission but it has not been widely used because there INR18650-25R 2500mAh lithium battery as an object of
are many problems such as long charging time. There are a lot experimental research, which has the main parameters:
of studies [1][2] on how to increase the speed of battery maximum charge voltage and minimum discharge voltage are
charging such as variants of CC-CV charging, polarization 4.2V and 2.5V, respectively. Standard charge current and fast
curve charging and pulse charging methods [3][4]. The pulse charge current are 1250mA and 4000mA provided by the
charging method for battery cells has been recognized as a fast manufacturer. Considering the complexity and ease to
and efficient way to overcome the shortcoming of slow development, this project adopted the Arduino UNO board as
charging time [5]. The optimal frequency is determined in the control board shown in Figure 1. Microcontroller can
order to minimize battery impedance. The adaptation of the communicate with LTC6802-2 monitoring chip via SPI
controlled pulse duty cycle decreases the concentration of the interface. ACS712 is a hall effect-based linear current sensor,
polarization on battery cells. has bidirectional current measurement ability, convert the
sensing current into a 1.5V to 3.5V voltage as output. Output
Model-Based design (MBD) has the advantages of improving voltage responds directly according to the current magnitude.
the product quality and short development cycle [6]. It allows Charging current should be measured as frequently as possible
developers to create equivalent circuit of lithium battery and throughout the charging process to calculate the SOC of the
charging controller model in Simscape [7] and Stateflow [8], battery [11].
respectively. Using the Simulink toolbox, the simulation of
equivalent circuit and charging control models can be virtually To calculate the SOC, sampling interval of charge current is
run on computer to quickly assess the charging controller’s set as 10 milliseconds and the interval of cell voltage and
performance. In this way, designer-specific algorithm can be temperature acquisition set as 2 seconds. After request to
interactively modified and evaluated at starting point of design LTC6802-2 from the SPI master, measurement of 12 channels
flow rather than the final phase so that improve software takes 13 milliseconds. Modification of frequency and duty
reliability and reduce development time. The schematic cycle is shown at the bottom in Figure 1.

1329
 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 12, Number 7 (2017) pp. 1329-1333
© Research India Publications. http://www.ripublication.com

Figure 1. Schematic diagram for charging test

seconds, so as to obtain the optimal frequency and duty cycle.

B. Matlab/Simulink-based model implementation The charge controller is in charge of resetting the integrator
The evaluation framework is implemented using the block with the charge switch signal every 2 seconds based on
MATLAB/Simulink environment to design the proposed design requirements. Through observation and analysis of
battery charger system including the proposed searching battery state parameters, such as charging current, terminal
algorithm of the pulse frequency and the duty cycle, which is voltage, SOC, and temperature, achieved the most efficient
shown in Figure 2. The physical model of the battery cells is frequency and duty cycle for a fast pulse-based charging
developed using the SimscapeTM package, which requires a method. The charging time and temperature rise due to two
solver reconfiguration. important parameters in the charging algorithm. The rise of
The dotted box on the left in Figure 2 indicates variable temperature indicates the heating on internal resistance of the
declaration for the controlled frequency and duty cycle value. battery cells during the charge process.
The controlled voltage source block is selected from The third dashed line from the left shows the high-fidelity
SimscapeTM to model pulse voltage for the charge process of battery model provided by SimscapeTM. As a high-fidelity
the battery pack. Duty cycle and frequency, as the two input model design, also consider SOC and thermal effect on the
signals, can be adjusted by changing the gain coefficient of the modeled component. Each component of the equivalent circuit
gain blocks. The convective heat transfer block offers a way to model will change with SOC and temperature of the battery
exchange convective heat effect between the battery and the over the process of the battery charging. The amount of heat,
ambient environment. The second dotted box from the left which is generated on R0 and R1 as an output of the thermal
shows that ambient temperature is set at 20◦C, which can be model, is connected by the convective heat transfer. The
adjusted as an input variable. equivalent circuit model [12] consists of an ideal voltage
We designed a charge controller, described by a dashed line on source (Em), an RC block (R1, C1) and an internal resistance
the right. The charge controller decides whether to stop (R0), can describe the characteristics of the lithium battery as
charging based on two input signals: the SOC and the capacitance and resistance in Figure 3. The capacitance
temperature of the battery. The Cur_Inte signal is an characteristic of the battery cell causes different responses at
accumulated value provided by the integrator block at the different frequency pulses.
upper right corner to calculate the average current within

1330
 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 12, Number 7 (2017) pp. 1329-1333
© Research India Publications. http://www.ripublication.com

Figure 2. Matlab/Simulink-based model implementation for pulse-based charging

Figure 3. Equivalent circuit model for lithium battery

polarization curve by calling acc_cur check function. Adopt

C. Finite state machine of the proposed charger system the efficient pulse in the Charge state until the SOC is divisible
The implemented state flow of the proposed charger system is by 0.2% or 5%, then cycle back to the Search state once again.
shown in Figure 4. We proposed five states including Eventually, entering the Charge_Complete state means the end
Charge_Start, Search_Fre, Search_Duty, Charge and of the charge process. The user-defined functions
Charge_Complete state. The start state is a default state, update_param_fre and update_param_duty in
initializing all parameters and checking the battery status. The MATLAB/Simulink models are shown at the top left corner.
charger has to verify that the temperature and SOC of the In the function, get_param and set_param are used to get and
battery are under the respective thresholds of 45 ◦C and 80% change the gain coefficient of the gain blocks shown in Figure
by using the function of status_check at the lower left corner. 3 during the simulation.
Throughout the charge process, the battery status is checked at
regular intervals. Generated code by Simulink Coder support for Stateflow
which is integrated with low level code will be used in the
The optimal frequency is obtained in Search_Fre state by hardware-in-the-loop testing in the following section. Once all
calling the cur_check function. In Search_Duty state, the duty the above work is done, the new charging algorithm is ready
cycle is adjusted to ensure that the charging current abides by for the hardware test.
the polarization curve. Through the LUT, we can check the

1331
 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 12, Number 7 (2017) pp. 1329-1333
© Research India Publications. http://www.ripublication.com

Figure 4. Finite state machine of the proposed charger system

D. Experiment environment for charging test IMPLEMENTATION RESULTS

Figure 5. Experiment environment for charging test

Figure 6. Wave form captured by oscilloscope

The model KPS-600 regulated power supply as the charging

source, using MOSFET switch module to control the Take 500Hz frequency and 50% duty cycle as an example,
frequency and duty cycle. But the micro controller IO port obtained the output voltage signal from current sensor and
cannot control IRF540 MOSFET, indirectly control it via the corresponding PWM control signal via oscilloscope is
PS2801C optical coupler. PWM is generated by Arduino shown in Figure 6. One column means 1ms, pulse period is
UNO. The experiment environment setup here is shown in 2ms. There are two channels in this figure. One horizontal
Figure 5. Through the communication cable, Arduino UNO line of the first channel (blue line) means 500mV and the
board can report battery status to host computer, such as second channel (red line) means 2V. The first channel on
charging current, cell terminal voltage, cell temperature, behalf the magnitude of charging current, can be infer that
pulse frequency and duty cycle. charging current is approximately 1.35A (185mV/A)

1332
 International Journal of Applied Engineering Research ISSN 0973-4562 Volume 12, Number 7 (2017) pp. 1329-1333
© Research India Publications. http://www.ripublication.com

because current sensor output value is 2.5V when current is ACKNOWLEDGMENT

zero. The second channel on behalf the PWM control signal, This research was supported by the Basic Science Research
which has the same period and the amplitude is 5V. Program through the National Research Foundation of
Korea(NRF) funded by the Ministry of
Education(2014R1A6A3A04059410).

REFERENCES

[1] J. Li, “Design of a li-ion battery charger with cc-cv-ct

regualtion loop,” Electric Information and Control
Engineering, April 2011, pp. 4088–4091.
[2] M. D. Yin, J. Youn, D. Park, and J. Cho, “Efficient
frequency and dutycycle control method for fast pulse-
charging of distributed battery packs by sharing cell
status,” International Workshop on Ubiquitous Wireless
Sensor, August 2015, pp. 40–45.
[3] L.-R. Chen, “A design of an optimal battery pulse
charge system by frequency-varied technique,”
Industrial Electronics, IEEE Transactions, February
Figure 7. GUI interface for battery status observation 2007, pp. 398–405.
[4] L.-R. Chen, “Design of duty-varied voltage pulse
The graphical user interface (GUI) is developed in charger for improving li-ion battery-charging response,”
MATLAB [13]. Through the observation in the charging Industrial Electronics, IEEE Transactions, February
process, the status of the battery can be displayed in real 2009, pp. 480–487.
time is shown in Figure 7. With MATLAB support package [5] Purushothaman B K, Landau U, “Rapid charging of
for Arduino, Arduino board can communicate with lithium-ion batteries using pulsed currents,” Journal of
MATLAB by a USB cable. Through the defined The Electrochemical Society, 2006, pp. 533–542.
communication protocols, MATLAB can acquire real time [6] Broy, Manfred, Sascha Kirstan, Helmut Krcmar and
data such as charging current, pulse frequency, duty cycle, Bernhard Schätz. "What is the Benefit of a Model-
cell voltage and temperature from the Arduino board. In this Based Design of Embedded Software Systems in the
way, we can take further measures such as stop charging to Car Industry?." Emerging Technologies for the
prevent cell damage in case of the abnormal status. Evolution and Maintenance of Software Models, 2012,
343-369
CONLUSION [7] Miller S, Wendlandt J, “Real-Time simulation of
In order to speed up the evaluation of the implemented physical systems using Simscape,” MATLAB News
charger system, improve the evaluation quality and make the and Notes, 2010, pp. 1–13.
results of evaluation more comparable, a development [8] Arduino UNO and Genuino UNO,
methodology consisting of three steps in terms of https://www.arduino.cc/en/Main/ArduinoBoardUno
development procedure was proposed. We developed a [9] Kumar, Ashish, “Schaem: A method to extract
workflow to access the performance of pulse charging statechart representation of FSMs”, Advance
algorithm for multiple cells. Equivalent circuit model is Computing Conference, 2009, pp. 1556-1561.
adopted to simulate the lithium battery and charging strategy [10] Liao Y G, Fu D T, “Design and development of a
is built in Stateflow, which is a control logic tool developed teaching tool for lithium-ion battery management
by MathworksTM. A quick verification of the charging system,” Conference for Industry and Education
control logic can be performed and then integrated the Collaboration, 2015, pp. 1–12.
generated code from Statefow with the low level hand-code.
[11] Dai Haifeng Z X, “A smart current and voltage
Testing is done on the customized hardware for pulse
acquisition system with high accuracy for EV
charging. Sumsung 18650 type Li-ion battery as the test
case to evaluate the newly designed charging algorithm. A applications,” International Journal on Smart Sensing
design flow of developing fast pulse charging algorithm and and Intelligent Systems, December 2012, pp. 1–19.
a hardware platform linked with MATLAB are introduced [12] T. Huria, M. Ceraolo, J. Gazzarri, and R. Jackey, “High
in this paper. Developers are able to reuse our experience in fidelity electrical model with thermal dependence for
rapid design and validation as a reference throughout the characterization and simulation of high power lithium
development process. As a case study, we showed fast battery cells,” Electric Vehicle Conference, Mar 2012,
development capabilities of designer-specific battery- pp. 1–8.
charger algorithm and its evaluation experience using the Creating Graphical User Interfaces, https://
proposed hardware-software emulation framework. Through www.mathworks.com/help/pdf_doc/matlab/buildgui
the above work, we successfully developed the pulse .pdf
charging strategy and evaluated the charging performance in
three month ahead of schedule.

1333