This circuit calculates battery life expectancy by outputting a voltage proportional to estimated remaining battery capacity. It can be adapted to different battery voltages and types by selecting resistors R1 and R2. The voltage varies from 0-2.5V as capacity decreases from 100% to 0%. The circuit uses a linear model based on least squares approximations of popular battery discharge curves to estimate remaining charge from battery voltage. It then translates the estimated charge into estimated life expectancy by making the current-to-voltage gain inversely proportional to load current.
This circuit calculates battery life expectancy by outputting a voltage proportional to estimated remaining battery capacity. It can be adapted to different battery voltages and types by selecting resistors R1 and R2. The voltage varies from 0-2.5V as capacity decreases from 100% to 0%. The circuit uses a linear model based on least squares approximations of popular battery discharge curves to estimate remaining charge from battery voltage. It then translates the estimated charge into estimated life expectancy by making the current-to-voltage gain inversely proportional to load current.
This circuit calculates battery life expectancy by outputting a voltage proportional to estimated remaining battery capacity. It can be adapted to different battery voltages and types by selecting resistors R1 and R2. The voltage varies from 0-2.5V as capacity decreases from 100% to 0%. The circuit uses a linear model based on least squares approximations of popular battery discharge curves to estimate remaining charge from battery voltage. It then translates the estimated charge into estimated life expectancy by making the current-to-voltage gain inversely proportional to load current.
Calculates Battery Life Expectancy W. Stephen Woodward CIRCLE 523 Marine Sciences, Venable Hall, CB3300, University of North Carolina, Chapel Hill, NC 27599-3300; e-mail: woodward@unc.edu
he convenience and reliability of TB = QB/IB. Measurement of IB is easy The circuit in Figure 1 is based on this
T portable electronic devices de-
pend on efficient management of batteries and their limitations. One uni- enough, but we also need QB to predict battery life expectancy. The simplest way to obtain QB is to es- idea. It outputs a 0- to 2.5-V signal that is proportional to the estimated re- maining battery capacity. The circuit versal, not to mention irritating, limita- timate it from the battery output voltage. can be adapted to different battery volt- tion of all (chemical) batteries is their Unfortunately, the exact relationship of ages and types by proper selection of R1 finite capacity and lifetime. Moreover, battery voltage to charge is a dauntingly and R2. This allows IQ to vary from 50 while the inevitability of eventual dis- messy nonlinear function of many para- to 0 µA as battery charge varies from charge is bad enough, the effects of sud- meters. Included among these are initial 100% to 0%. Then, the fixed-gain cur- den and unexpected battery failure are capacity, temperature, load current, dis- rent-to-voltage converter A2 produces usually much worse. This makes real- charge history, and battery chemistry (al- the calibrated output voltage. time estimation of remaining battery kaline, NiCd, LiMnO2, etc.). Even so, a Listed below are the least-square energy and life expectancy a useful fea- usable approximation of QB can be ob- R1/R2 approximations for four of the ture in portable systems. tained by using a simple linear model most popular battery chemistries A battery’s expected lifetime at any derived from a least-squares method. (normalized to NCELLS = the number given moment is equal to remaining These are fit to the actual discharge of cells in the battery; VR = the 1.25-V battery charge divided by load current: curves of various popular battery types. reference voltage). 174 ELECTRONIC DESIGN • September 5, 2000 IDEAS FOR DESIGN
VB Alkaline: R1 = 12.2K * NCELLS
VBATT R2 = VR * R1 / (0.72 * NCELLS−1) IB LiMnO2:R1 = 16.8K * NCELLS +5 V R1 R2 = VR * R1 / (1.85 * NCELLS−1) 73.2k NiCd: R1 = 6.83K * NCELLS 300k R2 = VR * R1 / (0.82 * NCELLS−1) +5 V 6 8 Pb-Acid: R1 = 6.76K * NCELLS – 7 Q1 R2 = VR * R1 / (1.43 * NCELLS−1) VR 5 1.25 V + A1 2N5087 + 1/2LT1495 Translation of estimated battery 9-V alkaline QB R2 LT1634 (6-cell) battery 22.1k charge to estimated life expectancy is IQ R4 more complicated due to the required (0-50 mA) 40.2k VQ arithmetic division of charge by load 2 current. Figure 2 shows one way to im- – 1 A2 plement this calculation by making A2’s 3+ 1/2LT1495 transimpedance gain variable (inversely 4 R3 R5 proportional to IB) instead of fixed as in 10k 10k 2.5 V = full charge Figure 1. Key to this method is the addi- 0 V = dead tion of digitally controlled potentiome- ter DCP1 (a Xicor X9511 “PushPot”). In operation, comparators A3 and A4 1. This circuit outputs a 0- to 2.5-V voltage that is proportional to the estimated remaining match VS, the voltage developed across battery capacity. It can be adapted to different battery voltages and chemistries. load-current sense resistor R4, to the voltage at DCP1’s VL pin. As the load current increases, VB A3 asserts DCP1’s PD To load (decrement) control pin. R1 R1 This causes the VW wiper to +5 V 73.2k 73.2k step toward the VL pin and drive RL and VI toward zero. +5 V Conversely, as load cur- 150k rent decreases, A4 asserts 4 DCP1’s PU (increment) pin, 6 A1 VR – Q1 Q2 driving the wiper toward VH 1.25 V 5 2N5087 + 7 2N5087 and RL toward 10k. This es- tablishes a feedback loop + R2 R2 that continuously main- LT1634 22.1k 9V 22.1k tains: RL = 10k(1−ILR4/0.05 0.5 A-hr QB Alkaline IQ V). In other words, as IL goes 0-50 mA from 0 to full-scale defined R6 240k +5 V by 0.05 V/R4, RL goes from IB 10k to zero. The internal 2 1 7 8 structure of DCP1 inher- R6 14 8 VCC DCP1 ently establishes RH = 10k− A4 A3 x9511W PD PU ASE
240k RL, so RH = 10k*ILR4/0.05.
+ – + – Since A2’s transimpedance R3 R8 12 13 10 9 VL 10k VH 14k V OUT = 1 V/hour- gain is given by R32/ RH, the 10.2k 6 VSS 3 remaining gain is R32/(10k*ILR4/0.05) VS VW R7 = ~10k/I L . A2’s gain is, as 390 4 5 2 A2 promised, inversely propor- – 1 R4 3+ tional to IL. 0.05 With the component val- 11 ues shown, Figure 2 will work R3 with a standard 9-V, 500-mA- 14k hr alkaline battery and 100 A1,..., A4 = 1/4 LT1496 mA < IB < 1 A. Together, they generate an output scaled to RL RH 1 V/hr of remaining battery RL+RH = 10k life. Appropriate selection of resistors R1 through R4 can accommodate virtually any 2. Translation of estimated battery charge to estimated life expectancy is accomplished by making A2’s battery and any desired scale transimpedance gain variable (inversely proportional to IB) instead of fixed as in Figure 1. factor. 176 ELECTRONIC DESIGN • September 5, 2000