BATTERIES are used as voltage sources in many applications

Transcription

1 2900 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012 Single-Magnetic Cell-to-Cell Charge Equalization Converter With Reduced Number of Transformer Windings Sang-Hyun Park, Student Member, IEEE, Ki-Bum Park, Member, IEEE, Hyoung-Suk Kim, Student Member, IEEE, Gun-Woo Moon, Member, IEEE, and Myung-Joong Youn, Senior Member, IEEE Abstract In this paper, a new cell-to-cell charge equalization converter using a multiwinding transformer is proposed. The proposed scheme achieves the direct cell-to-cell charge transportation by buck boost and flyback operation. In this operation, the adjacent two cells share either a current path or a tap of multiwinding transformer. Therefore, the number of windings ia cut in half in comparison to the number of batteries, resulting in a small circuit size. To verify the operation of the proposed charge equalization converter, an experiment with a lithium-ion battery stack is performed. Index Terms Battery equalizer, cell-to-cell charge equalization, multiwinding transformer. I. INTRODUCTION BATTERIES are used as voltage sources in many applications such as artificial satellite, hybrid electric vehicles, electric vehicles, uninterruptible power supplies, and photovoltaic systems. These applications generally require a high voltage source. However, the terminal voltage of one battery cell is relatively low. One-cell voltages of the lead acid battery, Ni Cd battery, Ni MH battery, and lithium-ion battery are 2.0, 1.2, 1.2, and 3.7 V, respectively. To achieve the required voltage level, series-connected battery stacks are being utilized in these applications. During operation, the batteries are charged or discharged repetitively. Since the batteries have different chemical and electrical characteristics during manufacturing, there can be a cell mismatch problem. Also, different ambient temperature and asymmetrical degradation with aging can also cause a cell mismatch problem. This problem leads to large nonuniformities in a cell charge level after several cycles of charge and discharge operations. While charging a series-connected battery Manuscript received July 7, 2011; revised September 23, 2011; accepted November 20, Date of current version March 16, This paper was presented at the International Conference on Power Electronics, Jeju, Korea, in May Recommended for publication by Associate Editor S. Williamson. S.-H. Park, H.-S. Kim, G.-W. Moon, and M.-J. Youn are with the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology, Yuseong-gu, Daejeon , Korea ( gbs@powerlab.kaist.ac.kr; hskim27@powerlab.kaist.ac.kr; gwmoon@ee.kaist. ac.kr; mmyoun@ee.kaist.ac.kr). K.-B. Park was with the Korea Advanced Institute of Science and Technology, Yuseong-gu, Daejeon , Korea. He is now with ABB Corporate Research Center, Baden-Dättwil, 8050 Zurich, Switzerland ( parky@powerlab.kaist.ac.kr). Color versions of one or more of the figures in this paper are available online at Digital Object Identifier /TPEL stack, some cells can be fully charged before others. Without any control, such as charge equalization, the fully charged batteries will be overcharged in a short time. Then, the energy storage capacity of overcharged batteries severely decreases, and in a worst case, there may be an explosion or a fire. Therefore, battery charge equalization circuits are required for a series-connected battery stack to enhance the lifetime of the application. Numerous charge equalization schemes have been proposed and well summarized [1] [5]. They are classified into two categories: dissipative method and nondissipative method. In a dissipative method, the unbalanced energy of each battery is converted to heat by a resistive shunt circuitry. A resistive shunt is connected across each cell in a battery stack. The current drawn by the resistive shunt is proportional to the cell terminal voltage. The more charged battery has a higher terminal voltage than other batteries, and the resistive shunt circuit consumes more energy. As a result, charge balancing can be achieved. However, this method cannot regulate the shunt current precisely and the shunt elements have the additional losses by a resistive component. To regulate the shunt current precisely, a new scheme using an individual circuit equalizer (ICE) has been proposed [6], [7]. In this scheme, each cell has its own ICE, which consists of a switch and a resistor. By controlling this switch in ICE, the equalization circuit can be connected selectively to discharge a particular battery. However, it also has additional losses in the resistor similar to the first method. In order to prevent the additional losses, a nondissipative charge equalization converter using dc/dc topologies has been proposed [8] [17]. Nondissipative methods are further divided into three categories: charge type, discharge type, and charge discharge type [4]. In a charge type, the equalizing current is extracted from the battery stack and the current flows into each battery until the voltage of each battery reaches a threshold voltage. This scheme employs a multiwinding transformer, which has a single magnetic core with secondary taps for each cell. Therefore the multiwinding transformer requires as many windings as the number of batteries [8], [9]. In a discharge type, the equalizing current is drained from individual batteries and the current charges whole battery stack. This scheme is equipped with flyback or buck boost topology because a high voltage conversion ratio is required. Each cell has a switch and a diode with high voltage stress [10]. In charge type and discharge type, the equalization current is diverted along many paths, which results in a longer equalization /$ IEEE

2 PARK et al.: SINGLE-MAGNETIC CELL-TO-CELL CHARGE EQUALIZATION CONVERTER 2901 Fig. 1. Method using bidirectional nondissipative current diverter. Fig. 2. Switched capacitor method. II. CELL-TO-CELL CHARGE EQUALIZATION CONVERTER In a charge discharge type, the equalizing current is extracted from the most charged battery and the current flows into another battery. This scheme achieves a cell-to-cell charge transportation and small circulating current. A. Adjacent Cell-to-Cell Method In a typical cell-to-cell method, the charge moves between adjacent batteries. Fig. 1 shows the method using a bidirectional nondissipative current diverter [11]. In this method, the charge moves from the most charged battery to an adjacent battery by a buck boost operation. Fig. 2 shows the switched capacitor method [12] [14]. The equalizing path is controlled by single-pole double-throw switches and the charge moves from one battery to an adjacent battery through the capacitor C E. Because the transferred charge is proportional to the difference between the terminal voltages of adjacent batteries, this method takes a long equalization time compared with other charge equalization schemes. To limit the surge current, resistors R ON are added. In these typical cell-to-cell methods, when the target battery and the source battery are nonadjacent, the charge transportation is achieved in several steps. Therefore, this scheme has the disadvantages of long equalization time and low efficiency. B. Direct Cell-to-Cell Method To overcome the disadvantages of adjacent cell-to-cell methods, a cell-to-cell method using a common energy storage component is introduced. Through the common energy storage component such as a capacitor, this method achieves the direct cell-to-cell charge transportation between any two batteries in the battery stack, which results in a relatively short equalization time. Fig. 3 shows the flying capacitor charge shuttling method [16]. The charge moves from the most charged battery to the least charged battery through common capacitor C fly. Similar to the switched capacitor method, the transferred Fig. 3. Flying capacitor charge shuttling method. charge is proportional to the difference between the terminal voltages of the most and least charged batteries, which results in a long equalization time. Fig. 4 shows a bidirectional flyback converter with dc-link capacitor [17]. In this scheme, the charge moves from the most charged battery to the least charged battery through a dc-link capacitor by flyback operations. In comparison with the flying capacitor charge shuttling method, a bidirectional flyback converter with dc-link capacitor has the advantage of a short equalization time. But this method requires as many transformers as the number of batteries and additional link capacitors. The number or size of the magnetic components makes the system bulky. C. Proposed Cell-to-Cell Method A new cell-to-cell charge equalization converter using a multiwinding transformer is proposed in this paper. In this scheme, the multiwinding transformer has a reduced number of windings to improve the size problem, and direct cell-to-cell

3 2902 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012 Fig. 4. Bidirectional flyback converter with dc-link capacitor. charge transportation can be achieved. Fig. 5 shows the proposed charge equalization converter. Each cell is connected to a bidirectional switch set. The upper switch blocks the discharging current path and the lower switch blocks the charging current path. An odd-numbered cell and its neighboring cell share one multiwinding tap. By using a shared multiwinding transformer, direct cell-to-cell charge transportation can be achieved. When a source cell (the most charged battery) is an odd-numbered cell and a target cell (the least charged battery) is an even-numbered cell, and vice versa, the charge transportation can be achieved in one step. If the source cell and the target cell are adjacent, the equalizing current is transferred by a buck boost operation using the magnetizing inductor of the transformer. If the source cell and the targeted cell are separated by other cells, the equalizing current is transferred by a flyback operation using the part of multiwinding transformer. When the source cell and the target cell are all odd- or even-numbered cells, the charge transportation can be achieved in two steps consisting of buck boost and flyback operations in a sequential pattern. In the proposed scheme, the total number of windings in the transformer is reduced to half the number of cells in the battery pack. The cell-to-cell charge transportation is achieved by selecting the switching patterns of buck boost, flyback, or both. The analysis of the operational principles of this method follows. III. OPERATIONAL PRINCIPLES In the proposed charge equalization converter, the equalizing paths have three kinds of operational cases. In case 1, battery charge moves from an odd- to an adjacent even-numbered cell, or vice versa. In this case, the charge is transferred by a buck boost operation in one step. In case 2, battery charge moves from an odd- to a nonadjacent even-numbered cell, or vice versa. In this case, the charge is transferred by a flyback operation in one step. In case 3, battery charge moves from one odd-numbered cell to another, or from one even-numbered cell to another. In Fig. 5. Proposed charge equalization converter. this case, the charge cannot be moved in one step. Therefore, the charge is moved by buck boost and flyback operations in sequence. In this section, the detailed analysis of each case is presented. For the convenience of analysis, several assumptions are given as follows. 1) All the switches are ideal except the body diode. 2) The terminal voltages of battery cells are constant during a switching cycle. 3) The magnetizing inductance of multiwinding transformer can be treated as the reflected inductance L m. A. Case 1: Charge Transfer From an Odd- to Adjacent Even- Numbered Cell The equalizing current is transferred by a buck boost operation. A magnetizing inductor and leakage inductor of multiwinding transformer is operated as an output inductor of the buck boost converter. The mode analysis of case 1 is presented assuming that the source cell is B 1 and the target cell is B 2. Switches Q 1d and Q 2d are always turned on to provide the current path between B 1 and B 2. Switch Q 2u is always turned off because the body diode of Q 2u is used as a rectifier diode. Fig. 6 shows the current path of case 1 and Fig. 7 shows the gate signals and the key waveforms of case 1. Mode 1(t 0 t 1 ): When switch Q 1u is turned on, mode 1 starts and the charge is extracted from B 1. The voltage of B 1 is applied to L m and L lkg1, and the equalizing current is built up. The current path of mode 1 is shown in Fig. 6(a). According to the assumptions, the equalizing current increases with a constant slope as follows: i B 1 (t) V B 1 L m + L lkg1 (t t 0 ). (1)

4 PARK et al.: SINGLE-MAGNETIC CELL-TO-CELL CHARGE EQUALIZATION CONVERTER 2903 Fig. 6. Current paths of case 1. (a) Mode 1: discharging B 1. (b) Mode 2: charging B 2. Fig. 8. Current paths of case 2. (a) Mode 1: discharging B 1. (b) Mode 2: charging B 4. follows: V B 1 E B 2,charge E B 1,discharge 1 (DT s ) 2. (4) 2 L m + L lkg1 Mode 3(t 2 t 3 ): When the inductor current reaches zero, mode 3 starts. In this mode, a resonance occurs until mode 1 starts again. Fig. 7. Key waveforms of case 1. The maximum current of mode 1 can be expressed as follows: V B i peak 1 DT S. (2) L m + L lkg1 Mode 2(t 1 t 2 ): When switch Q 1u is turned off, mode 2 starts. The equalizing current flows through switch Q 2d and the body diode of Q 2u. The energy stored in L m and L lkg1 flows into B 2. As shown in Fig. 6(b), the terminal voltage of B 2 is applied to L m and L lkg1 in opposite direction. The equalizing current decreases with a constant slope as follows: V B i L (t) 1 V B DT S 2 (t t 1 ). (3) L m + L lkg1 L m + L lkg1 When the inductor current becomes zero, the charge transportation will be stopped by the body diode of Q 2d. During mode 2, all the energy stored in L m and L lkg1 moves to B 2. The whole transferred energy in one switching cycle can be expressed as B. Case 2: Charge Transfer Between Nonadjacent Odd- and Even-Numbered Cells The equalizing current is transferred by a flyback operation. For the analysis, it is assumed that the source cell is B 1 and the targeted cell is B 4. In this case, T 1 :T 2 of the multiwinding transformer are operated as a flyback transformer. Switches Q 1d and Q 4d are always turned on to provide the current path between B 1 and B 4. Switch Q 4u is always turned off because the body diode of Q 4u is used as a rectifier diode. Switch Q 1u determines the current build up period. Fig. 8 shows the current path and Fig. 9 shows the key waveforms of case 2. Mode 1(t 0 t 1 ): When switch Q 4d is turned on, mode 1 starts. The terminal voltage of B 1 is applied to T 1 and the inductor current of L m is built up. The energy of B 1 is stored in L m.the current path of mode 1 is shown in Fig. 8(a). The equalizing current and the maximum value of equalizing current are the same as (1) and (2) in case 1, respectively. Mode 2(t 1 t 2 ): When switch Q 1u is turned off, mode 2 begins. The energy stored in L m is transferred to B 4 through transformer (T 1 :T 2 ) by a flyback operation. Fig. 8(b) shows the current path of mode 2. In this mode, switch Q 1u has a voltage spike due to the current in L lkg1. Therefore, switch Q 1u requires additional snubber circuit to limit the large voltage stress. Mode 3(t 2 t 3 ): When the inductor current reaches zero, mode 3 starts. In this mode, a resonance occurs until mode 1 starts again. C. Case 3: Charge Transfer Between Odd- or Between Even- Numbered Cells In this case, because the charge cannot be transferred directly by buck boost or flyback operation, the charge transportation can be achieved in two steps that consist of buck boost and flyback operations. The charge moves from a target cell to an adjacent cell (intermediate cell) in the first step by a buck boost

5 2904 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012 Fig. 9. Key waveforms of case 2. operation. In the next step, the charge moves from an intermediate cell to a target cell by a flyback operation. Consequently, the charge is transferred from a source cell to a target cell in two steps. For the analysis, it is assumed that the source cell is B 1 and the target cell is B 3. Because the charge can move from a source cell to B 2 by a buck boost operation, B 2 is selected as an intermediate cell. Fig. 10 shows the current path and Fig. 11 shows the switching patterns and simplified current waveforms of case 3. Mode 1(t 0 t 1 ): In first step, the charge from B 1 is transferred into B 2 by a buck boost operation through L m and L lkg1. When switch Q 1u is turned on, mode 1 starts. The terminal voltage of B 1 is applied to L m and L lkg1, and the equalizing current increases with a constant slope. The current path of mode 1 is shown in Fig. 10(a). Mode 2(t 1 t 2 ): When switch Q 1u is turned off, mode 2 starts. The terminal voltage of B 2 is applied to L m and L lkg1 in opposite direction, and the energy stored in L m and L lkg1 flows into B 2. The body diode of Q 2u is used as a rectifier diode. The current path of mode 2 is shown in Fig. 10(b). During modes 1 and 2, Q 1d and Q 2d are always turned on to provide the current path, and the other switches are always turned off. All the operations are the same as case 1. Mode 3(t 2 t 3 ): After the first step ends, the charge flows from B 2 into B 3 by a flyback operation through multiwinding transformer (T 1 :T 2 ). When switch Q 2u is turned on, mode 3 starts. The terminal voltage of B 2 is applied to T 1,asshownin Fig. 10(c), and the equalizing current is built up. Mode 4(t 3 t 4 ): When Q 2u is turned off, mode 4 begins. The terminal voltage of B 3 is applied to T 2 in the opposite direction and the energy stored in L m flows into B 3, as shown in Fig. 10(d). The body diode of Q 3u is used as a rectifier diode for the flyback Fig. 10. Current paths of case 3. (a) Mode 1: discharging B 1. (b) Mode 2: charging B 2. (c) Mode 3: discharging B 2. (d) Mode 4: charging B 3. operation. During modes 3 and 4, switches Q 2d and Q 3d are always turned on to provide the current path, and the other switches are always turned off. Similar to case 2, switch Q 2u has a large voltage spike because there is a leakage current in L lkg1. IV. COMPARATIVE STUDY ON THE PROPOSED CHARGE EQUALIZATION CONVERTER A. Equalization Speed The equalization speed is one of the major design parameters for a cell balancing circuit. Except for the switched-capacitor method and the flying capacitor charge shuttling method, a cellto-cell charge equalization converter, especially direct cell-tocell method, has a high equalization speed compared with the other types of cell balancing circuits. When the power rating of equalizing converter is fixed, the transferred power among the batteries in one switching cycle decides the equalization speed. It is assumed that each battery stack consists of N seriesconnected cells, and the transferable power rating of equalizing converter is P 0. With these assumptions, the average transferable power of the unit switching cycle is derived for each equalization converter type. In the charge type, the net charging power of target cell and the net discharging power of other cells except the target cell can be expressed as follows: P target,charge type P 0 P 0 (5) N P source,charge type P 0 N. (6) The equalization speed is determined by (5) when one cell is less charged than the other batteries, while the other batteries

6 PARK et al.: SINGLE-MAGNETIC CELL-TO-CELL CHARGE EQUALIZATION CONVERTER 2905 TABLE I REQUIRED SWITCHING CYCLE FOR CHARGE DISCHARGE TYPE TABLE II COMPARISON OF EQUALIZATION PERFORMANCE Fig. 11. Key waveforms of case 3. are balanced in a same charge level. In this case, the charge type has the best equalization performance. When one cell is more charged than the others while the others are balanced, (6) determines the equalization speed. This is the worst case in the equalization performance of the charge type. In the discharge type, in a similar way, the net discharging power of source cell and the net charging power of other cells except source cell can be expressed as follows: P source,discharge type P 0 P 0 (7) N P target,discharge type P 0 N. (8) When one cell is more charged or less charged than the other balanced cells in the battery stack, the equalization speed of the discharge type is determined by (7) or (8), respectively. In the charge discharge type, the transferred power during one switching cycle is always P 0. In the adjacent cell-to-cell method, if the source cell is separated from the target cell by several other cells, it takes several switching cycles to transfer the charge from the source cell to the target cell. Table I shows the required switching cycles according to each position of the source cell and the target cell. From Table I, the possible number of cases and the sum of all required cycles for N series-connected battery stack can be presented as follows: # of Case N 2 N (9) ( N 1 ) N 2 Cycle 2 k + k k1 N 1 2 k1 k1 k(k +1) 2 N(N 1)(N +1). (10) 3 By (9) and (10), the average switching cycle to complete the charge transportation is given as follows: Cycle ave Cycle # of Case N +1. (11) 3 Consequently, the average transferable power in one switching cycle is presented as follows: P source,adjacent type P target,adjacent type Cycle ave 3P 0 N +1. (12) The equalization speed of the adjacent charge discharge type is decided by (12). In the direct cell-to-cell method, the charge always moves from the source cell to the target cell in one switching cycle. Therefore, the average transferable power is P 0. In the proposed scheme, to reduce the number of multiwindings, the charge is transferred in one switching cycle or two switching cycles. Therefore, the charge transportation is completed on average in 1.5 cycles and the average transferable power in one switching cycle is presented as follows: P source,proposed P target,proposed P (13) As summarized in Table II, the direct cell-to-cell method has the highest equalization speed for the same power rating of the equalization converter. The proposed scheme also has a relatively high equalization speed compared to other types. B. Devices/Components In this section, a comparison among the conventional equalization schemes and the proposed equalization scheme based on the number of active switches, component ratings, and number of magnetic components will be presented for a battery stack P 0

7 2906 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012 TABLE III COMPARISON BETWEEN THE CONVENTIONAL AND PROPOSED SCHEMES of N series-connected batteries. The average voltage of each cell is assumed to be V B. Table III lists the comparison for these schemes. The equalization methods using a centralized flyback converter (scheme #1) [8] and using a centralized forward converter with a multiwinding transformer (scheme #2) [9] are charge types and move the charge by flyback of the forward operation using a multiwinding transformer. These schemes require a small number of active switches but require the same number of multiwindings as the number of cells. The equalization method using isolated flyback converters (scheme #3) [8] and the current diverters using forward converters with a centralized multiwinding transformer (scheme #4) [10] are discharge types. Scheme #3 moves the charge using the same number of transformers as the number of cells. Scheme #4 requires the same number of multiwindings as the number of cells. The nondissipative current diverter (scheme #5) and the bidirectional nondissipative current diverter (scheme #6) [11] employ the adjacent cell-to-cell method, which is one of the charge discharge types and moves the charge between adjacent cells by a buck boost operation. These schemes require as many inductors as the same number of cells. The bidirectional flyback converter with dc-link capacitor (scheme #7) [17] is the direct cell-to-cell method, which is one of the charge discharge types and the charge moves between batteries through a dc-link capacitor by flyback operations. This scheme requires as many transformers as the same number of cells and additional dc-link capacitors. However, the proposed circuit needs a multiwinding transformer, which has a small size because it requires N/2 windings. Consequently, the reduced size of magnetic components leads to the advantages of small size and low cost as well as fast equalization time, as mentioned in Section IV A. Fig. 12. Implemented prototype of proposed scheme with six-cell lithium-ion battery stack. TABLE IV EXPERIMENTAL PARAMETERS V. EXPERIMENTAL RESULTS To verify the operational principles of the proposed charge equalization converter, a prototype has been implemented for a six-cell lithium-ion battery stack. Fig. 12 shows the implemented prototype and Table IV summarizes the parameters of the prototype. The voltage sensor for each cell is configured with a filter capacitor and differential Op-amp circuit. The microcontroller receives the cell voltage information by analog-to-digital converter and calculates the pulsewidth modulation (PWM) duty cycles. The PWM gate signal is applied to MOSFETs through gate drivers. The initial difference of cell voltages in the battery stack is V, which is about a 20% state of charge (SOC) gap. The target difference is set as 0.04 V, which is 1% of the nominal lithium-ion battery voltage (about 5.9% SOC gap).

8 PARK et al.: SINGLE-MAGNETIC CELL-TO-CELL CHARGE EQUALIZATION CONVERTER 2907 A. Design Consideration The transferred output power of one cell is designed to be approximately 2 W and the average equalizing current I ave is 0.5 A. In this calculation, the nominal battery voltage is assumed to be about 4.0 V. When the terminal voltages of the target and source batteries are V B,discharge and V B,charge, respectively, and T s is the switching period, the condition to guarantee the discontinuous conduction mode (DCM) of buck boost and flyback operation is expressed as follows: V B,discharge V B,charge DT s (1 D)T s 0. (14) L m + L lkg L m + L lkg If the lithium-ion battery is in a normal operation region, the open-circuit battery voltage lies between 3.4 and 4.2 V. With this voltage, the duty cycle for DCM operation is under approximately 44.7%. Including the margin, the constant duty cycle D is set as 40%. The peak equalizing current I peak in the buck boost and flyback operations is expressed as follows: 2 I peak I ave 2.5A. (15) D The magnetizing inductance of a multiwinding transformer L m is designed to satisfy the power rating and is expressed as follows: DT s L m V B,discharge μh. (16) I peak When the charge is transferred by a flyback operation, the large voltage spike is applied to the upper switch Q iu of each current path. Therefore, an additional snubber circuit is required to reduce the voltage spike. To limit the voltage stress of the switch to twice the target battery voltage plus the source battery voltage, an RCD snubber is designed as follows: R sn V 2 sn 1 2 L lkgi 2 peak (V sn/v sn V B,source )(1/T s ) C sn 2kΩ (17) V snt s ΔV sn R sn 100 nf. (18) B. SOC Estimation The SOC is directly related to the open-circuit voltage (OCV) V OC of the battery. Therefore, the charge equalization can be considered as the OCV equalization. When the battery is charged or discharged, the OCV and measured terminal voltage V T have the difference due to the diffusion characteristics and internal resistances. Fig. 13 shows the commonly used simplified cell model. R Diff and C Diff are the parameters, which are related to slow dynamics of charge diffusion. By controlling the timing of voltage sensing, the effect of C Diff can be cancelled. The internal resistor R C and R D are the fast dynamic components in charging and discharging operations, respectively. When the average charging and discharging currents are equal to I ave and ( I ave ), respectively, the estimated OCVs in charging and discharging processes can be expressed as follows: V OC,charge V T V Diff V D V T I ave (R Diff + R C ) (19) Fig. 13. Simplified cell model. V OC,discharge V T V Diff V D V T + I ave (R Diff + R D ). (20) The OCV in charging process V OC,charge is less than V T, and the OCV in discharging process V OC,discharge is larger than V T. The terms of (R Diff + R C ) and (R Diff + R D ) can be measured experimentally. The calculation of OCV estimation is implemented by a microcontroller. C. Switching Pattern Implementation The source battery and the target battery are selectable by controlling the switching pattern. Each battery has two switches in its current path. When a battery has to be discharged, its upper switch determines a switching duty and its lower switch must always be turned on during switching cycles. When a battery has to be charged, its upper switch must be turned off and its lower switch must be turned on during switching cycles. As shown in Fig. 14, these switching patterns are valid for either buck boost or flyback operation. The rules of the switching pattern are presented in Table V. In the prototype, the microcontroller makes two kinds of switching patterns and selects the output ports, which are connected to gate drivers. D. Experimental Results Fig. 15 shows the gate signals and experimental waveforms of case 1, where B 1 is the source cell and B 2 is the target cell. In this case, the equalization converter operates as a buck boost converter, as shown in Fig. 14(a). The switching patterns of switches Q 1u, Q 1d, Q 2u, and Q 2d are shown in Fig. 15(a) and these switches belong to the current path of B 1 and B 2. Switch Q 1u determines the duty cycle and build-up duty. Switches Q 1d and Q 2d are always turned on to provide the current path. The other switches are always turned off. Fig. 15(b) shows the experimental waveforms of battery current and voltage stress on the upper switches. When switch Q 1u is turned on, the discharging current of B 1 is built up with a constant slope. After switch Q 1u is turned off, the equalizing current from B 1 flows into B 2 and the current decreases with a constant slope. The slope of current increase or decrease is proportional to the applied battery terminal voltage. As shown in Fig. 15(b), there is a difference between ascending slope and descending slope. In a practical case, the terminal voltage of discharging battery is increased compared with its own OCV. In a charging battery, the terminal voltage is decreased. Therefore, the slope of current B 2 is steeper. Fig. 16 shows the gate signals and experimental waveforms of case 2. When B 1 is the source cell and B 4 is the target cell, the equalization converter operates as a flyback converter. Fig. 16(a)

9 2908 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012 Fig. 14. Switching patterns of each operation case. (a) Buck boost operation. (b) Flyback operation. TABLE V SWITCHING PATTERNS Fig. 16. (a) Switching patterns. (b) Key waveforms of case 2. Fig. 15. (a) Switching patterns. (b) Key waveforms of case 1. shows the switching patterns of switches Q 1u, Q 1d, Q 4u, and Q 4d. Similar to case 1, switch Q 1u determines the duty cycle and buildup duty. Switches Q 1d and Q 4d are always turned on and the other switches are always turned off. Fig. 16(b) shows the experimental waveforms of battery current and voltage stress on the upper switches. When switch Q 1u is turned on, the charge flows from B 1 and the magnetizing current is built up. After switch Q 1u is turned off, the magnetizing current flows into B 4 by transformer T 1 :T 2. The slopes of current B 1 and B 4 are also different. The maximum voltage stress and voltage spike on the upper switches is constrained to a maximum of 35 V by a snubber circuit, which is added to the upper switches of each battery. Fig. 17 shows the gate signals and experimental current waveforms of case 3. When the source cell is B 1 and the target cell is B 3, the charge moves from B 1 to B 2 in the first step, and moves from B 2 to B 3 in the next step. Fig. 17(a) shows the switching patterns of switches Q 1u, Q 1d, Q 2u, Q 2d, Q 3u, and Q 3d.Inthe first step, switch Q 1u makes the duty cycle, and switches Q 1d and Q 2d are turned on. A buck boost operation is performed in first step. After that, switch Q 2u determines the duty cycle, and switches Q 2d and Q 3d are turned on for a flyback operation. Fig. 17(b) shows the experimental waveforms of battery current. In the first step, the current flows from B 1 into B 2.Inthe second step, the current flows from B 2 into B 3. Consequently, the charge moves from B 1 to B 3 in two steps. Fig. 18 shows the equalization performance of the proposed charge equalization converter. The cell voltages are measured every second. When the equalization circuit is driven for

10 PARK et al.: SINGLE-MAGNETIC CELL-TO-CELL CHARGE EQUALIZATION CONVERTER 2909 TABLE VI SOC DISTRIBUTION OF THE LITHIUM-ION BATTERY CELLS Fig. 17. (a) Switching patterns. (b) Key waveforms of case 3. batteries. In this paper, a new cell-to-cell charge equalization converter with a reduced number of transformer windings is proposed. In the proposed circuit, the charge is transferred by buck boost operation, flyback operation, or both of them in sequence. The operation principle is analyzed and the experimental results are presented to verify the analysis. The proposed circuit achieves the high-speed equalization by the direct cellto-cell charge transportation. The number of multiwindings is cut in half, which results in small circuit size and low production cost. Therefore, the proposed circuit can be used widely for lithium-ion battery applications, which need fast equalization. APPENDIX The measured transferable power of the proposed charge equalization converter P 0 is 1.20 W. For the same power rating of equalizing converter, the estimated equalization times of the other equalization methods are presented as follows. Fig. 18. Equalization performance of proposed charge equalization converter. 150 min, the charge balance is achieved. The distributions of battery voltages and their SOCs are summarized in Table VI. From the experimental results, the initial SOC gap of 21.3% among six cells decreases to about 5.8% at the end of equalization. This SOC gap of 5.8% is equivalent to approximately 0.04 V and this value satisfies the target. Under the same condition of the initial SOC gap, the equalization times of the other charge shuttling mechanisms such as the charge type, the discharge type, the adjacent cell-to-cell method, and the direct cell-to-cell method are estimated as 303, 241, 247, and 96 min, respectively. These results come from the analysis in the Appendix. VI. CONCLUSION Series-connected lithium-ion batteries are used in many applications. These applications require the charge equalization circuit to solve the SOC imbalance problems of lithium-ion A. Charge Type For a six lithium-ion battery stack, the transferable power of the target cell and the source cell of the charge type is expressed as follows: P target,charge type 5P 0 6. (21) P source,charge type P 0 6. (22) Because the SOCs of all the batteries are different, all the batteries except B 1 are operated as both the target battery and the source battery during the equalization process. But the most charged battery B 1 is operated only as the source battery until the end of equalization. Therefore, the equalization time is determined by the SOC change of B 1, ΔSOC B 1 and the transferable power of the source cell (22). From the experimental result of the proposed method, the SOC of B 1 changes from 68.4% to 57.9%. For this condition, the equalization time of the charge type is estimated as follows: t charge type ΔQ B 1 I eq C ΔSOC B 1 P source,charge type /V B,nominal 2.6Ah 10.5% h (23) (1.20/6 3.7)A

11 2910 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012 where I eq is an equalization current, C is the nominal capacity, and V B,nominal is the nominal voltage of lithium-ion battery. Under the same experimental condition of the proposed method, the equalization time of the charge type is about 303 min. B. Discharge Type For a six lithium-ion battery stack, the transferable power of the target cell and the source cell of the discharge type is expressed as follows: P target,discharge type P 0 (24) 6 P source,discharge type 5P 0 6. (25) Similar to the charge type, the least charged battery B 6 is operated only as the target battery until the end of equalization. Therefore, the equalization time is determined by the SOC change of B 6, ΔSOC B 6 and the transferable power of the target cell (24). From the experimental results of the proposed method, the SOC of B 6 changes from 47.1% to 52.1%. For this condition, the equalization time of the discharge type is estimated as follows: t discharge type ΔQ B 6 I eq C ΔSOC B 6 ηp target,discharge type /V B,nominal 2.6Ah 5.0% h (26) ( /6 3.7)A where η is the efficiency of the equalizer, which is assumed to be 0.6. Under the same experimental condition of the proposed method, the equalization time of the discharge type is about 241 min. C. Adjacent Cell-to-Cell Method In the adjacent cell-to-cell method, the transferable power of the target cell and the source cell for the six cells is expressed as follows: P source,adjacent type P target,adjacent type 3P 0 7. (27) From the result of the proposed method, B 1, B 2, and B 3 are operated as the source battery. Therefore, the equalization time is determined by the total SOC change of these batteries and the transferable power of the source cell (27). The SOCs of B 1, B 2, and B 3 change from 68.4% to 57.9%, from 63.7% to 57.0%, and from 61.2% to 56.4%, respectively. For this condition, the equalization time of the adjacent cell-to-cell method is estimated as follows: t adjacent type ΔQ total I eq C (ΔSOC B 1 + ΔSOC B 2 + ΔSOC B 3 ) P source,adjacent type /V B,nominal 2.6Ah (10.5% + 6.7% + 4.8%) (3 1.20/7 3.7)A h (28) Under the same experimental condition of the proposed method, the equalization time of the adjacent cell-to-cell method is about 247 min. D. Direct Cell-to-Cell Method In the direct cell-to-cell method, the transferable power of the target cell and the source cell are P 0. Similar to the adjacent cell-to-cell method, the equalization time is determined by the total SOCs of the source batteries and the transferable power of the source cell P 0. The equalization time of the direct cell-to-cell method is estimated as follows: t directtype ΔQ total I eq C (ΔSOC B 1 + ΔSOC B 2 + ΔSOC B 3 ) P 0 /V B,nominal 2.6Ah (10.5% + 6.7% + 4.8%) (1.20/3.7)A h. (29) Under the same experimental condition of the proposed method, the equalization time of the direct cell-to-cell method is about 96 min. REFERENCES [1] Y.-S. Lee and M.-W. Cheng, Quasi-resonant zero-current-switching bidirectional converter for battery equalization applications, IEEE Trans. Power Electron., vol. 21, no. 5, pp , Sep [2] N. H. Kutkut and D. M. Divan, Dynamic equalization techniques for series battery stacks, in Proc. 18th Annu. Int. Telecommun. Energy Conf., Boston, MA, Oct. 1996, pp [3] J. Cao, N. Schoeld, and A. Emadi, Battery balancing methods: A comprehensive review, in Proc. IEEE Veh. Power Propulsion Conf., Sep. 2008, pp [4] H.-S. Park, C.-E. Kim, C.-H. Kim, G.-W. Moon, and J.-H. Lee, A modularized charge equalizer for an HEV lithium-ion battery string, IEEE Trans. Ind. Electron., vol. 56, no. 5, pp , May [5] Y.-S. Lee and M.-W. Cheng, Intelligent control battery equalization for series connected lithium-ion battery strings, IEEE trans. Ind. Electron., vol. 52, no. 5, pp , Oct [6] D. Bjork, Maintenance of batteries-new trends in batteries and automatic battery charging, in Proc. Int. Telecommun. Energy Conf., Toronto, ON, Canada, Oct. 1986, pp [7] B. Lindemark, Individual cell voltage equalizers (ICE) for reliable battery performance, in Proc. 13th Int. Telecommun. Energy Conf.,Kyoto,Japan, Nov. 1991, pp [8] H. Schmidt and C. Siedle, The charge equalizer-a new system to extend battery lifetime in photovoltaic system, U.P.S. and electric vehicles, in Proc. 15th Int. Telecommun. Energy Conf., Paris, France, Sep. 1993, pp [9] N. H. Kutkut, D. M. Divan, and D. W. Novotny, Charge equalization for series-connected battery strings, IEEE Trans. Ind. Appl., vol. 31, no. 3, pp , May/Jun [10] N. H. Kutkut, Non-dissipative current diverter using a centralized multiwinding transformer, in Proc. 28th Annu. IEEE Power Electron. Spec. Conf., 1997, vol. 1, pp [11] N. H. Kutkut, A modular nondissipative current diverter for EV battery charge equalization, in Proc. IEEE Appl. Power Electron. Conf., 1998, vol. 2, pp [12] A. Baughman and M. Ferdowsi, Double-tiered switched-capacitor battery charge equalization technique, IEEE Trans. Ind. Electron., vol. 55, no. 6, pp , Jun [13] J. W. Kimball and P. T. Krein, Analysis and design of switched capacitor converters, in Proc. Appl. Power Electron. Conf. Expo., Mar. 2005,vol. 3, pp

12 PARK et al.: SINGLE-MAGNETIC CELL-TO-CELL CHARGE EQUALIZATION CONVERTER 2911 [14] C. Pascual and P. T. Krein, Switched capacitor system for automatic series battery equalization, in Proc. IEEE Appl. Power Electron. Conf. Expo., Feb. 1997, pp [15] S. West and P. T. Krein, Equalization of valve-regulated lead-acid batteries: Issues and life test results, in Proc. 22nd Annu. Int. Telecommun. Energy Conf., Phoenix, AZ, Sep. 2000, pp [16] X. Wei and B. Zhu, The research of vehicle power Li-ion battery pack balancing method, in Proc. IEEE 9th Int. Electron. Meas. Instruments Conf., Beijing, China, Aug. 2009, pp [17] C. Karnjanapiboon, K. Jirasereeamornkul, and V. Monyakul, High efficiency battery management system for serially connected battery string, in Proc. IEEE Int. Symp. Ind. Electron., Seoul, Korea, Jul. 2009, pp [18] D. V. Cadar, D. M. Petreus, and M. Patarau, An energy converter method for battery cell balancing, in Proc. IEEE Int. Spring Semin. Electron. Technol., Warsaw, Poland, May 2010, pp [19] B. T. Kuhn, G. E. Pitel, and P. T. Krein, Electrical properties and equalization of lithium-ion cells in automotive applications, in Proc. IEEE Vehicle Power Propuls. Conf., Chicago, IL, Sep. 2005, pp [20] M. Tang and T. Stuart, Selective buck-boost equalizer for series battery packs, IEEE Trans. Aero. Electron. Syst., vol. 36, no. I, pp , Jan [21] H.-S. Park, C.-H. Kin, K.-B. Park, G.-W. Moon, and J.-H. Lee, Design of a charge equalizer based on battery modularization, IEEE Trans. Veh. Technol., vol. 58, no. 7, pp , Sep [22] S. Sheldon, S. Williamson, C. Rimmalapudi, and A. Emadi, Electrical modeling of renewable energy sources and energy storage devices, J. Power Electron., vol. 4, no. 2, pp , Apr [23] S. Pang, J. Farrell, J. Du, and M. Barth, Battery state-of-charge estimation, in Proc. Amer. Control Conf., Arlington, VA, 2001, pp [24] J. Chiasson and B. Vairamohan, Estimating the state of charge of a battery, IEEE Trans. Control Sys. Tech., vol. 13, no. 3, pp , May [25] G.-B. Koo, Design guidelines for RCD snubber of flyback converters, Fairchild Semiconductor Corp., Seoul, Korea, Application Note AN-4147, Rev , [26] I.-S. Kim, A technique for estimating the state of health of lithium batteries through a dual-sliding-mode observer, IEEE Trans. Power Electron., vol. 25, no. 4, pp , Apr [27] L. Maharjan, T. Yamagishi, H. Akagi, and J. Asakura, Fault-tolerant operation of a battery-energy-storage system based on a multilevel cascade PWM converter with star configuration, IEEE Trans. Power Electron., vol. 25, no. 9, pp , Sep [28] I. Aharon and A. Kuperman, Topological overview of powertrains for battery-powered vehicles with range extenders, IEEE Trans. Power Electron., vol. 26, no. 3, pp , Mar [29] H. Qian, J. Zhang, J.-S. Lai, and W. Yu, A high-efficiency grid-tie battery energy storage system, IEEE Trans. Power Electron., vol. 26, no. 3, pp , Mar [30] H. Zhou, T. Bhattacharya, D. Tran, T. S. T. Siew, and A. M. Khambadkone, Composite energy storage system involving battery and ultracapacitor with dynamic energy management in microgrid applications, IEEE Trans. Power Electron., vol. 26, no. 3, pp , Mar Ki-Bum Park (S 07 M 10) was born in Korea, in He received the B.S., M.S., and Ph.D. degrees in electrical engineering from the Korea Advanced Institute of Science and Technology, Daejeon, Korea, in 2003, 2005, and 2010, respectively. He is currently a Scientist at ABB Corporate Research Center, Baden-Dättwil, Switzerland. His research interests include power converters, server power system, high power density adapter, battery management system, and display driver circuit. Dr. Park received the 2nd Prize paper award from the International Telecommunications Energy Conference (INTELEC), in Hyoung-Suk Kim (S 09) was born in Korea, in He received the B.S. degree in electronics engineering from Pusan National University, Pusan, Korea, in He is currently working toward the Ph.D. degree in electrical engineering at Korea Advanced Institute of Science and Technology, Daejeon, Korea. His current research interests include dc dc power converters, digital controller design, LED color control, and battery equalizers. Gun-Woo Moon (S 92 M 00) was born in Korea, in He received the M.S. and Ph.D. degrees in electrical engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea, in 1992 and 1996, respectively. He is currently a Professor in the Department of Electrical Engineering, KAIST. His research interests include modeling, design and control of power converters, soft-switching power converters, resonant inverters, distributed power systems, power-factor correction, electric drive systems, driver circuits of plasma display panels, and flexible ac transmission systems. Dr. Moon is a member of the Korean Institute of Power Electronics, Korean Institute of Electrical Engineers, Korea Institute of Telematics and Electronics, Korea Institute of Illumination Electronics and Industrial Equipment, and Society for Information Display. Sang-Hyun Park (S 09) was born in Korea, in He received the B.S. and M.S. degrees in electrical engineering from the Korea Advanced Institute of Science and Technology, Daejeon, Korea, in 2005 and 2007, respectively, where he is currently working toward the Ph.D. degree. His research interests include power converters, digital power control, server power system, and charge equalization converter. Mr. Park is a member of the Korea Institute of Power Electronics. Myung-Joong Youn (S 74 M 78 SM 98) was born in Seoul, Korea, in He received the B.S. degree from Seoul National University, Seoul, Korea, in 1970, and the M.S. and Ph.D. degrees in electrical engineering from the University of Missouri, Columbia, in 1974 and 1978, respectively. In 1978, he joined the Air-Craft Equipment Division, General Electric Company, Erie, PA, where he was an Individual Contributor on Aerospace Electrical System Engineering. Since 1983, he has been a Professor at the Korea Advanced Institute of Science and Technology, Daejeon, Korea. His research interests include power electronics and control, which include the drive systems, rotating electrical machine design, and high-performance switching regulators. Dr. Youn is a member of the Institution of Electrical Engineers, U.K., the Korean Institute of Power Electronics, the Korean Institute of Electrical Engineers, and the Korea Institute of Telematics and Electronics.