A Self Adaptable Intelligent Battery Charger for Portable Electronics

Abeyratne S. G.1, Kuruppu S.S. 1, Divarathne C.M. 1, Handapangoda D.K. 1, Senevirathne S.M.A.J. 1, Dayaratne U.I. 1 1 Department of Electrical and Electronic Engineering, University of Peradeniya, Sri Lanka. sunil@ee.pdn.ac.lk, sandun@ee.pdn.ac.lk
Abstract- Proliferation of portable electronic equipment, that uses secondary battery types with different capacities & cell chemistries, has become a threat to the environment. The number of batteries disposed per day is on the rise; the associateddedicated-battery-charges naturally get discarded with the equipment, adding much to the pollution. The improper charging patterns that reduce the life cycle of batteries are also contributing to the problem. This paper introduces a concept and a methodology which can be used to reduce the environmental impact due to different batteries and their chargers. The proposed charger is adapting itself to, any battery type (Lead Acid, NiMH, Li-Ion) with any cell number. This finds the charger still useful, though the equipment is wasted and no longer useable. The proposed methodology identifies the battery type using its terminal electrical-characteristics. The method reveals the battery capacity, with no restrictions imposed by the charging state. Having identified the battery type, the proposed quantum table identifies the number of cells in the battery. Therefore, the battery can be properly maintained to get maximum cycles before being disposed. The operation and the control method of the charger are also explained.

However, these methods are less user friendly that they need various battery parameters to be provided by the user. This paper investigates and presents a user-friendly adaptable charging technique for different cell chemistries where the user intervention is not necessary. The preliminary investigation on this idea was presented by the authors in the forum [7]. II. PROPOSED SYSTEM A. Proposed Methodology Using the battery terminal characteristics, in real time, following parameters are identified: 1. Battery Type (Cell Chemistry) 2. Charge Capacity 3. Cell Number The identification of terminal characteristics is centered around the transient characteristics of the terminal-voltage recovery-upon sudden load removal. It has been identified in this experiment that the voltage-build up across the terminals, within a given time span, is basically unique for a given battery chemistry. The experimental setup employed is shown in Fig. 1.

I.

INTRODUCTION

Proper battery maintenance is vital for achieving the maximum life for a battery, not withstanding that premature battery failure could occur in numerous ways as explained in [1],[2]. Proper battery charging thus becomes important. The charger operation is, however, cell voltage and battery chemistry dependent. Therefore, each and every portable equipment purchase goes with a purchase of a specific battery charger too. At the time of discarding the equipment, the tailor-made charger would naturally render useless. This become an environmental hazard due to the toxic waste of the battery and also due to the electronics and associated materials. According to article [3] in United States alone, 426,000 cell phones are discarded per day, showing the severity of the impact on the environment. The concept of a universal charger is appropriate in the sense that it does not demand a charger per equipment purchase. However, the intelligence inside should be capable of identifying the battery chemistry, charge capacity, cell voltage and charging state for proper battery maintenance. Commercially available Integrated Circuits (IC) are capable of charging multiple battery chemistries in a single system. However, user intervention is necessary to input the battery type, cell voltage and charge capacity [4]. Several methods have been introduced to identify the battery capacity for a known battery type and cell number [5], [6].

Fig. 1 Experimental Setup for capturing the transient characteristic

The battery is connected (for 200s) and disconnected (for 200s) from the load using the relay shown in Fig.1 (load stepping). The relay is controlled using the computer via a data acquisition card (DAQ). The battery terminal voltage is sampled at a frequency of 1Hz and the data is stored in the computer through DAQ card. When the relay is turned on, the

978-1-4244-1666-0/08/$25.00 '2008 IEEE

1775

battery is made to discharge at a constant current IDC. The resulted battery voltage waveform is shown in Fig.2.

Fig. 4. Transient Voltage Difference vs Percentage Discharge for constant IDC for different battery capacities Fig. 2. Transient Voltage Recovery Behavior at load shedding over 200s intervals.

The period C to D represents the voltage decay when the battery is loaded with IDC. At point D, the relay is turned off, whereupon the battery open circuit voltage undergoes a transient. The open circuit voltage builds up to point E marking the end of the 200s period, where the relay is turned on again. Thereafter, the voltage starts dropping. The voltage VAB is measured and plotted against the percentage discharge corresponding to point A (total current discharged is measured and divided by the total nominal battery capacity). This is repeated for different discharge levels and marked as in Fig.3. Note that the above characteristic is valid for the given constant current, IDC.

Fig. 5 shows the Transient Voltage Difference (V1-V2 ) vs. Percentage Discharge for constant IDC for a given battery at two different temperatures.

Fig. 5. Transient Voltage Difference vs Percentage Discharge for constant IDC for a given battery but different temperatures

According to Fig. 5, it can be seen that this gradient is also independent of temperature. It should be noted that Fig. 2 through Fig. 5 are for Lead Acid battery type. Table I below shows the gradient for different battery capacities, terminal voltages and also for different discharging rates (IDC) of Lead Acid batteries.
TABLE I GRADIENT COMPARISON FOR LEAD ACID BATTERIES Fig. 3. Transient Voltage Difference vs. Percentage Discharge for constant I DC Terminal Voltage (V) 6 6 12 Battery Capacity (Ah) 4.5 4 7 Discharging Rate (A) 1 3 1 Gbat (V) 30.6 32.0 32.7

It could be noted that the curve section marked by M-N in Fig. 3 is approximately linear and it was found that the gradient, (named Gbat this point forward), is unique for a battery chemistry regardless of the capacity and terminal voltage as shown in Fig 4.

It can be seen that the gradient is almost constant for the given cell chemistry. The same experiment was done on other cell chemistries. Fig. 6 illustrates these for NiMH and Li-ion

1776

battery types. The results for the three battery chemistries are given in Table II.

(from M to N) and Qtot is the rated charge capacity of the battery. Vi are the corresponding transient voltage differences. Constant gradient Gbat is valid approximately over a span of 5% of the total charge. For example, in Fig. 7, MN represents charge span of 5% of total charge. For the period t,
D Q % = Q1 % - Q 2 % = 5

Q % can be generally expressed for a period of t as, DQ % =

I DC Dt 100 Qtot

Therefore for the period MN, DQ % =

I DC Dt 100 = 5 Qtot

Therefore, battery capacity Qtot can be found as,

I DC Dt 100 5
(V1 - V2 ) Qtot I DC Dt

(2)

By substituting Qtot in Eq. (1), Gbat can be found as, Gbat cal 1 = (3)

B.

Battery Type and Capacity Identification

Now, Gbatca11 is compared with Gbat in Table II. If the match is a bit far away refinement can be done selecting different V1-V2 and associated I DC Dt . C. Control Implementation for Faster Capacity Identification As explained earlier, the battery type is identified using the unique gradient. But as seen from Fig.7, battery type identification has to be delayed up until 87% discharge level is reached, if the discharge rate is at 1A. Therefore, the discharge rate has been increased to 3A to make the identification faster (70%). The comparison of characteristics at these two discharge rates is illustrated in Fig. 8.

Fig. 7. Transient Voltage Difference vs. Percentage Discharge showing V1 -V2 and Q1-Q2

Consider a general Transient Voltage Difference vs. Percentage Discharge characteristics shown in Fig. 7. The gradient, Gbat can be written as, (V1 - V2 ) (1) Gbat = (Q1 % - Q2 %) where Qi % = I DC Dt 100 for i = 1,2 with IDC = constant

Qtot

discharge current, t = time taken to discharge from Q1 to Q2

Fig. 8. Transient Voltage Difference vs. Percentage Discharge for two different IDC values for a given battery (6V 4.5Ah)

If the discharge rate is increased beyond 3A, identification can be done at an earlier stage (below 70% discharge point). However, care should be taken not to exceed the battery safe current limit by not exposing the battery to higher discharge rates for a longer period of time. Therefore, loading and unloading times of the battery should be reduced to a lower value than 200s, to load the battery with heavy currents for faster identification. It should be noted that the unique gradient, Gbat defined earlier remains unchanged regardless of the loading time (200s above). An algorithm that can be used in the discharge rate selection and accelerate the battery type identification is shown in Fig.9 below.
START LIMIT TO MAX YES SET INITIAL CURENT

microcontroller for cell number identification to be explained later. The proposed cell number identification is based upon the above quantum table after the battery type detection. The State of Charge (SOC) of a cell is strictly dependent upon the specific gravity of the electrolyte [8]. The unloaded state after loading provides a clear relationship with the state of charge of the battery for some battery chemistries. Once the cell number is identified, the SOC can be estimated measuring the open circuit voltage as given in [9],[10] for lead acid cells. A sample curve is given in Fig.10.

I-REF LIMIT EXCEEDED

NO

COMMENCE TIMED DISCHARGE Fig.10. Open Circuit Voltage vs. Percentage Discharge for a 6V lead acid battery.

I-REF INCREMENT

NO

REDUIRED GRADIENT ACHIEVED

III. IMPLEMENTATION OF THE CHARGER The overall charger arrangement is shown in Fig. 11.

YES STOP Fig. 9. Algorithm to accelerate battery type identification.

Vout

D. Cell Number and State of Charge Identification Noting the fact that, almost all battery chemistries reach their final terminal voltage when a small amount of current is injected to the battery, number of cells in series can be identified faster. Once the voltage settles at a specific value the actual number of cells can be identified by using a quantum table for battery voltages as shown in Table III.
TABLE III QUANTUM TABLE FOR CELL NUMBER DETECTION Cell No 1 2 3 4 5 6 7 8 Lead Acid 2.1 4.2 6.3 8.4 10.5 12.6 NiCd / NiMH 1.25 2.50 3.75 5.00 6.25 7.50 8.75 10.00 Li-Ion 4.1 8.2 12.3 -

Fig. 11. Overall charger arrangement

For the AC-DC converter, topology presented in [11] is adapted. Galvanic isolation is also there in this topology. Its specifications are given in Table IV.

The chemistry dependent, possible battery voltage levels for different cell numbers listed in Table III are stored in the

1778

TABLE IV CHARGER SPECIFICATIONS Feature Input Voltage (VRMS) Input Frequency Output Voltage Range Maximum Output Current Input Power Factor Minimum Requirement 95V 270V 48Hz 440Hz 2V 18V 1A > 0.8

Current mode control is used for pre charging, charging and end of charge detection of Ni chemistries and Lead acid batteries. Voltage mode control is used to detect the end of charge of Li-Ion batteries. The microcontroller is responsible for the intelligence required for battery type, charge capacity and cell number identification. A. Battery Name Plate Voltage Detection Implementation

procedure is adapted. After performing load stepping, V1-V2 (see Fig.7 for V1-V2) is checked to detect whether the battery SOC is to the left of point M in Fig.3. If it is to the left, load stepping is performed to identify Gbat If not pre charging is continued to get the SOC to the left of M, where upon load stepping can be done. Calculation of Gbat is done as shown in equation (1). For this V1-V2 and IDC are read in to the microcontroller. Qtot is calculated using equation (2). Next Gbatcal1 is calculated and compared with the values in Table II. If necessary, refinement can be done as explained earlier. Having known the Qtot, charging rate can be appropriately selected to suite the battery type for better battery life. C. End of Charge Detection Once the battery type and capacity has been identified, the charging can be performed while monitoring the well known and proven end of charge detection criteria for each type. IV. CONCLUSION A Self Adaptable Battery Charging technique for different cell chemistries has been proposed. It has the advantage of proper battery management using a single charger. A single charger can be used for different types of equipment with different battery types. The environmental pollution due to batteries and charging electronics can be reduced. The charger is easy to use since the user intervention is not necessary in parameter. REFERENCES
[1] [2] M.W. Migliaro, Maintaining maintenance free batteries, in Conf. Rec. Industrial and Commercial Power Systems Technical Conference, Chicago,IL,USA, May 7-11, 1989, pp 69-73. V. Alminauskas, Performance evaluation of lead acid batteries for use with solarpanels, in Photovoltaic Specialists Conference,1993.,Conference Record of the Twenty Third IEEE, May. 10-14, Louisville, KY. USA, pp 1258-1263, ISBN:0-7803-1220-1263. J Voelcker, Lithium batteries take to the road, and The Big Picture, by C. Jordan in IEEE Spectrum, September 2007 (INT) Issue. National Semiconductor, LM3647 Universal Battery Charger for Li-Ion, Ni-MH and Ni-Cd Batteries, Product data sheet, March 2001. T. Yanagihara, A. Kawamura, Residual Capacity Estimation of Sealed Lead- Acid Batteries for Electric Vehicles, in proceedings of the Power Conversion Conference Nagaoka 1997, Volume 2, Issue, 3-6 Aug 1997, pp 943-946 Microchip, Appl. Note 626. Universal Adapting Battery Charger, Final report for the International Future Energy Challenge 2007 by Sri Lankan team, IEEE APEC07, Disney Land Hotel, Anaheim, California. T.R. Compton, Battery Reference Book , 2nd Edition, ButtworthHeinmann LTD, 1995, ISBN 0-7506-2567-8 K. Kutluay, Y. Cadirci, Y.S. Ozkazanc, I. Cadirci, A New Online State-of-Charge Estimation and Monitoring System for Sealed Lead Acid Batteries in Telecommunication Power Supplies,IEEE Transaction on Industrial Electronics, Vol. 52, Issue. 5, October 2005. T. Yamazaki, K. Sakurai, K. Muramoto, Estimation of the Residual Capacity of Sealed Lead-Acid Batteries by Neural Network, in Telecommunications Energy Conference, INTELEC, 1998, pp210-214 D.B.Ekanayake, S.G. Abeyratne, A Compact Rectifier Stage with an Improved Power Factor, Transaction of the IEE-Sri Lanka, Volume 5, Number 1, September 2003,pp64-68.

(a) Timing Diagram

[3] [4] [5] (b) Voltage build-up during Pre Charging Fig. 12.

As shown in Fig.12 (a), initially SW1 is turned on and SW2 is turned off using the microcontroller. The battery is pre charged with a current of 100mA for 180 seconds. Next both SW1 and SW2 are turned off for 1 second. At the end of the 1 second period available battery voltage is recorded in to the microcontroller. The selection of 180 seconds is important to avoid local voltage plateaus observed for some battery chemistries. According to table 3, to uniquely find the cell number, knowing the battery type is essential. B. Battery Type and Capacity Detection Implementation As explained in section II, Gbat defines the battery type. Therefore, the values derived in Table II are stored in the microcontroller. To implement Gbat detection, following

[6] [7] [8] [9]

[10] [11]

1779