Design and implementation of a battery test system with

Design and implementation ofa battery test system withenergy recycling technique

Chang-Hua Lin1, Hwa-Dong Liu1, Yu-Liang Lin2a),and Tung-Chin Pan11 Department of Electrical Engineering, National Taiwan University of Science and

Technology, No. 43, Sec. 4, Keelung Rd., Da’an Dist. Taipei, Taiwan2 Institute of Nuclear Energy Research,

No. 1000, Wenhua Rd., Longtan Dist. Taoyuan, Taiwan

a) [email protected]

Abstract: In this study, a microcontroller-based battery test system for

power battery is realized. The system is composed by a microcontroller,

a sampling circuit, a human interface and a resonant load. The proposed

resonant load having wide-range slew-rate and continuous loading features

is used to verify the dynamic characteristics of the power battery thus can

recycle energy in diagnostic process. Furthermore, the system can provide

loading current according to battery specifications. The proposed system has

both low cost and portable feature. Finally, this work provides analysis of

operation principle, and test results to verify the theoretical feasibility.

Keywords: power battery, resonant load, energy recycling

Classification: Electron devices, circuits and modules

References

[1] X. Wang, et al.: “A multi-cell battery pack monitoring chip based on 0.35-µmBCD technology for electric vehicles,” IEICE Electron. Express 12 (2015)20150367 (DOI: 10.1587/elex.12.20150367).

[2] D. Gharavian, et al.: “ZEBRA battery SOC estimation using PSO-optimizedhybrid neural model considering aging effect,” IEICE Electron. Express 9(2012) 1115 (DOI: 10.1587/elex.9.1115).

[3] D. Park, et al.: “Fast battery charger MCU with adaptive PWM controller usingruntime tracking of polarization curve,” IEICE Electron. Express 13 (2016)20160131 (DOI: 10.1587/elex.13.20160131).

[4] Q. Zhang, et al.: “An area-efficient and high speed multiplexer for batterymonitor system,” IEICE Electron. Express 13 (2016) 20160120 (DOI: 10.1587/elex.13.20160120).

[5] C. H. Yang, et al.: “Switching-mode battery test system,” IEEE IS3C’14 (2014)(DOI: 10.1109/IS3C.2014.164).

[6] K. I. Hwu and Y. T. Yau: “Active load for burn-in test of buck-type DC-DCconverter with ultra-low output voltage,” IEEE APEC’08 (2008) (DOI: 10.1109/APEC.2008.4522788).

[7] H. Ma, et al.: “Energy recycling load system with a high gain DC-DC converterfor ultra low voltage power supplies,” IEEE ISIE’13 (2013) (DOI: 10.1109/ISIE.2013.6563619).

© IEICE 2017DOI: 10.1587/elex.14.20170115Received February 9, 2017Accepted March 8, 2017Publicized March 28, 2017Copyedited April 25, 2017

1

LETTER IEICE Electronics Express, Vol.14, No.8, 1–6

http://dx.doi.org/10.1587/elex.12.20150367

















http://dx.doi.org/10.1109/IS3C.2014.164




http://dx.doi.org/10.1109/APEC.2008.4522788




http://dx.doi.org/10.1109/ISIE.2013.6563619





[8] M. K. Kazimierczuk: “Analysis of class-E zero-voltage-switching rectifier,”IEEE Trans. Circuits Syst. 37 (1990) 747 (DOI: 10.1109/31.55033).

[9] Q. Cheng, et al.: “High-efficiency parallel-circuit class-E power amplifier withdistributed T-shaped compensation circuit,” IEICE Electron. Express 13 (2016)20160570 (DOI: 10.1587/elex.13.20160570).

[10] L. Roslaniec, et al.: “Design of single-switch inverters for variable resistance/load modulation operation,” IEEE Trans. Power Electron. 30 (2015) 3200(DOI: 10.1109/TPEL.2014.2331494).

1 Introduction

In recent years, due to the energy shortage and the increase of people’s environ-

mental awareness, the electric vehicle [1] includes hybrid vehicles, electric ve-

hicles, electric scooter, etc. becomes one of the main items for development in

science and technology. The lithium-ion batteries as a mainstream power source for

various electric vehicles have turned a popular research topic, especially the

research focus on state of charge (SOC) and state of health (SOH) [2, 3]. A typical

battery test system captures the battery’s characteristics from the sampling circuit

[4], and the electronic load discharge energy of battery by constant current (CC) or

step current. However, the load behavior of above-described discharge test method

is very different compared with practical load conditions. Furthermore, there are a

lot of power waste during the test process and this power consumption also brings

about thermal issue consequently the heat dissipation system is required, moreover

it also results problems in bulky system, high cost, complication, etc. If the energy

consumption from conventional battery test method during testing process can be

effectively recycling, it can save a considerable amount of energy and comply with

the energy conservation policy. Generally, there are two battery test schemes with

energy recycling in the industry. The first is additional battery scheme [5], the

discharge energy from test battery is charged to another chargeable battery by

designated charger. The second is by way of grid-tied DC/AC inverter [6, 7], it

returns the discharge energy from test battery to the power grid. However, there

still exists a power loss in those two solutions. This study aims to improve the

shortcoming of the aforementioned scheme by resonant load. The energy is

recycling in the negative half cycle of sinusoidal current. A low loss energy

recycling technique can be achieved without too much additional circuit to imple-

ment energy recycling demand.

2 Scheme description

Fig. 1 shows the proposed battery test system with energy recycling technique. The

whole system is composed of one resonant load, an MCU, voltage and current

sampling circuit, DC offset circuit, USB module, and auxiliary power. And the

control rule of MCU (dsPIC33FJ64GS606) is shown in Fig. 2.

There are two reasons to adopt a resonant load, one is the estimation of SOC

and SOH can be calculated by the instantaneous voltage and current of battery.

Therefore, the resonant sinusoidal current can cover various load conditions for

battery. Another reason is the resonant load can provide dynamic loading in the


2

IEICE Electronics Express, Vol.14, No.8, 1–6

http://dx.doi.org/10.1109/31.55033

http://dx.doi.org/10.1109/31.55033

http://dx.doi.org/10.1109/31.55033





http://dx.doi.org/10.1109/TPEL.2014.2331494




positive half-cycle and recycle energy in the negative half-cycle to achieve a low

loss battery test technique as shown in Figs. 3(a)–(b). Fig. 3(c) shows the average

current in a half-cycle.

3 Resonant load

The resonant load is achieved by class E topology [8, 9, 10], as shown in Fig. 4(a),

which consists of two inductors (L1 and L2), two capacitors (C1 and C2), and one

power switch (S). The class E topology is simple structure and can implement high

efficiency by zero voltage switching (ZVS).

The input inductor (L1) of conventional class E topology is in series with VS

and becomes a current source hence the L1 does not participate in resonant

operation. The proposed technique using battery as voltage source (VS) employ

the bidirectional load current for testing and recycling energy of battery. Therefore,

Fig. 1. Proposed battery test system with energy recycling technique.

Fig. 2. The control rule of MCU.

(a) (b) (c)

Fig. 3. (a) Extract energy from battery. (b) Recycle energy. (c) Averagecurrent

(a) (b)

Fig. 4. (a) Class E topology. (b) Simplified circuit of resonant load.


3


the L1 participates with resonant operation as shown in Fig. 4(b). Moreover, the

load (Ro) is short circuit in order to reduce power loss. For simplifying the circuit,

L2 and C2 are equivalent to a capacitive device Co, which can be represented as:

Co ¼ C2

1 � !2L2C2

ð1ÞAfterward we combine the Co and C1 to equivalent Ceq ¼ C1 þ Co and the resonant

frequency can be expressed as:

fr ¼ 1

2�ffiffiffiffiffiffiffiffiffiffiffiffiL1Ceq

p ð2Þ

The switching frequency fs of resonant load are set identical with resonant

frequency fr (in proposed case is 38 kHz). Based on the above design, the iL1 is

symmetrical in each half-cycle.

There are three distinct intervals in circuit operation.

Interval I [t0–t1]:

The power switch (S) is turned off, supposing the initial current iL1ðt0Þ ¼ 0 and

the voltage of Ceq reaches the maximum as VCeqðt0Þ ¼ VCeq max. The Ceq starts

resonant with L1, and consequently Ceq transfers energy to L1, the equivalent circuit

and Laplace transform circuit are shown in Fig. 5(a), (b), respectively.

From kirchhoff’s voltage laws (KVL), Fig. 5(b) can determine the equation as:

Vs

sþ L1 � IL1ðt0Þ � VCeqðt0Þ

s¼ IL1ðsÞ sL1 þ 1

sCeq

� �ð3Þ

VceqðsÞ ¼ VCeqðt0Þs

þ IL1ðsÞ � 1

sCeqð4Þ

By the inverse Laplace transformation, we can get the inductor current iL1(t) and

vCeq(t) as:iL1ðtÞ ¼ Ceq!0ðVs � VcmaxÞ sin!0ðt � t0Þ ð5Þ

vCeqðtÞ ¼ Vs þ ðVcmax � VsÞ cos!oðt � t0Þ ð6ÞWhen t ¼ t1, the vCeqðt ¼ t1Þ is resonate to zero and just in time to make the power

switch (S) bring about zero voltage switching.

Interval II [t1–t2]:

The power switch (S) is turned on with ZVS then VS magnetizes L1, the iL1(t) rising

linearly and energy stored in L1. The equivalent circuit and Laplace transform

circuit are also described in Fig. 6(a), (b), respectively.

From KVL, and VCeqðt1Þ ¼ 0, Fig. 6(b) can be obtained as:

Vs

sþ L1 � IL1ðt1Þ ¼ IL1ðsÞ � sL1 ð7Þ

By the inverse Laplace transformation, the inductor current iL1(t) can be shown as:

iL1ðtÞ ¼ Vs

L1� t þ IL1ðt1Þ ð8Þ

(a) (b)

Fig. 5. Interval I. (a) Equivalent circuit. (b) Laplace transform circuit.


4


Where iL1(t1) can be indicated as:

IL1ðt1Þ ¼ Ceq!oðVs � VcmaxÞ sin!oðt1 � t0Þ ð9ÞWhen t ¼ t2, the power switch (S) is turned off and this interval is terminated.

Interval III [t2–t3]:

This interval’s equivalent circuit shown in Fig. 7(a), the circuit is the same as

Fig. 5(a), nevertheless, the initial current iL1(t2) and the initial voltage of VCeq(t2)

are different from the prior two intervals. The energy stored in L1 would resonant to

Ceq therefore Laplace transform circuit can be exhibited in Fig. 7(b) and the KVL

be derived as:Vs

sþ L1 � IL1ðt2Þ � VCeqðt2Þ

s¼ IL1ðsÞ sL1 þ 1

sCeq

� �ð10Þ

VceqðsÞ ¼ VCeqðt2Þs

þ IL1ðsÞ � 1

sCeqð11Þ

By the inverse Laplace transformation, the inductor current iL1(t) and voltage of

equivalent VCeq(t) are described as:

iL1ðtÞ ¼ Ceq!oVs sin!oðt � t2Þ þ IL1ðt2Þ � cos!oðt � t2Þ ð12ÞVCeqðtÞ ¼ Vs þ ð�VsÞ cos!0ðt � t2Þ þ 1

Ceq!0

� IL1ðt2Þ� �

� sin!0ðt � t0Þ ð13Þ

IL1ðt2Þ ¼ Vs

L1� t2 þ Ceq!oðVs � VcmaxÞ sin!oðt1 � t0Þ ð14Þ

When t ¼ t3, the power switch (S) is turned on and this interval is completed.

4 Experimental verification

A laboratory prototype is designed and tested as shown in Fig. 8(a) to verify the

feasibility of proposed battery test system with energy recycling technique. Owing

to the energy is recycled under battery testing process and the power switch

achieves ZVS, hence heatsink is not required, consequently the volume of proto-

type can be effectively downsized and to achieve the portable purpose.

For comparing with conventional CC test in battery, the adopted battery is

dicharged by a battery-analyzer (GWInstek GBT-2636 and this system accuracy is

within 3�5%) with constant current 0.2C until cut-off voltage and energize battery

through 0.2C to fully charged in one charge cycle. The GBT-2636 will record the

voltage, current, stored energy of battery in charge-dicharge process. Based on the

(a) (b)

Fig. 6. Interval II. (a) Equivalent circuit. (b) Laplace transform circuit.

(a) (b)

Fig. 7. Interval III. (a) Equivalent circuit. (b) Laplace transform circuit.


5


recorded experimental data to calculate the capacity of battery within a charging

interval is defined as Q1, and the remaining capacity of battery after a discharging

interval is defined as Q2, accordingly the energy loss is defined as Qloss. Then the

energy recycling rate is defined as:

�recycle ¼ Q2

Q1

% ð15ÞProposed resonant load is used to load sinusoidal current, therefore the energy is

sank and backed to battery in one cycle, only exits part of energy loss. From

Fig. 3(c) the average value of sinusoidal current in positive half-cycle is:

Iavg ¼

Z �

0

Im � sinð!tÞdð!tÞ�

¼ 2Im�

ð16ÞSupposing the peak current (Im) is 1000mA, thus the average current (Iavg) is equal

636.62mA. For example in a 60 minute test time, the positive and negative half-

cycle each has half time which means the 30 minute test time can charging Q1 ¼318:31mAh. From the recorded experimental data, the remain capacity after 30

minute test time is Q2 ¼ 280:11mAh which means the �recycle in proposed

technique is 88% (this efficiency includes power consumption of entire system

with peripheral circuit). The test results using conventional CC and proposed

resonant load for battery with 1–60 minute test time are systematically listed in

Fig. 8(b). The CC test in battery consumes all of energy, and in contrast with the

proposed resonant load implementing half-cycle testing and another half-cycle for

energy recycling, the results show the energy loss (Qloss ¼ Q1 � Q2 ¼ 38:2mAh in

a 60 minute test time) during the entire test process can be effectively reduced.

5 Conclusion

In this paper, a practical battery test system for power battery is fabricated. The

resonant load achieves load test and energy recycling for battery. The proposed

system not only achieves the wide-range slew-rate for battery testing but also brings

about the function of energy recycling without grid-tied equipment. Also the

charger, additional equipment, and backup batteries are not required as a result

the total cost of battery test system is significantly decreased. Eventually, the

experimental verification shows the energy recycling rate of proposed technique

can reach 88%.

Acknowledgment

The financial support from MOST 105-2221-E-011-111.

(a) (b)

Fig. 8. Prototype and energy consumption of CC load vs. resonant load.


6


Documents

Design and implementation of a battery test system with