MAX17055 MODELGAUGE M5 EZ USER GUIDE

Transcription

1 MAX17055 MODELGAUGE M5 EZ USER GUIDE Single-Cell ModelGauge m5 Fuel Gauge December 2016 Rev.0, December 2016 Maxim Integrated Page 1

2 2016 Maxim Integrated Products, Inc. All rights reserved. No part of this documentation may be reproduced nor distributed in any form or by any means, graphic, electronic, or mechanical, including but not limited to photocopying, scanning, recording, taping, ing, or storing in information storage and retrieval systems without the written permission of Maxim Integrated Products, Inc. (hereafter, Maxim ). Products that are referenced in this document such as Microsoft Windows may be trademarks and/or registered trademarks of their respective owners. Maxim makes no claim to these trademarks. While every precaution has been taken in the preparation of this document, individually, as a series, in whole, or in part, Maxim, the publisher, and the author assume no responsibility for errors or omissions, including any damages resulting from the express or implied application of information contained in this document or from the use of products, services, or programs that may accompany it. In no event shall Maxim, publishers, authors, or editors of this guide be liable for any loss of profit or any other commercial damage caused or alleged to have been caused directly or indirectly by this document. Rev. 0, Month 2016 Rev.0, December 2016 Maxim Integrated Page 2

3 CONTENTS 1 MAX17055 ModelGauge m5 EZ User Guide User Guide Overview MAX17055 Overview ModelGauge m5 Algorithm OCV Estimation and Coulomb Count Mixing Empty Compensation Converge-to-Empty % and Empty Hold End-of-Charge Detection Fuel Gauge Learning Predicting Run Time for a Hypothetical Load Lithium Iron Phosphate Support (LiFePO4) ModelGauge m5 Standard Register Formats Analog Measurements Current Measurements Temperature Measurements Voltage Measurements Dynamic Battery Power Dynamic Power Performance Dynamic Power Output Registers Dynamic Power Configuration Registers Typical Applications Circuit Serial Number Feature Determining Fuel Gauge Accuracy Initial Accuracy ModelGauge m5 Registers ModelGauge m5 Algorithm Battery Parameters ModelGauge m5 Output Registers ModelGauge m5 Algorithm Configuration Registers ModelGauge m5 Algorithm Additional Registers Status and Configuration Registers Soft-Wakeup (Command Register) (60h) Modes of Operation I 2 C Bus System Hardware Configuration I/O Signaling I 2 C Protocols Trademark Information Rev.0, December 2016 Maxim Integrated Page 3

4 1 MAX17055 ModelGauge m5 EZ User Guide 1.1 User Guide Overview This User Guide provides extended description of the MAX It includes detailed descriptions of the register set, extended features, as well as application guidance. The MAX17055 data sheet describes the basic feature set as well as the minimal register set need to support the plug-and-play ModelGauge m5 EZ performance MAX17055 Overview The MAX17055 is a low power 7μA operating current fuel gauge IC that implements Maxim ModelGauge m5 EZ algorithm. ModelGauge m5 EZ makes fuel gauge implementation easy by eliminating battery characterization requirements and simplifying host software interaction. The MAX17055 measures voltage, current, and temperature to produce fuel gauge results. The MAX17055 uses either an external thermistor or internal die temperature to measure temperature of the battery pack. The ModelGauge m5 EZ robust algorithm provides tolerance against battery diversity. This robustness enables simpler implementation for most applications and batteries by avoiding time-consuming battery characterization. The ModelGauge m5 algorithm combines the short-term accuracy and linearity of a coulomb-counter with the long-term stability of a voltage-based fuel gauge, along with temperature compensation to provide industry-leading fuel gauge accuracy. The MAX17055 automatically compensates for aging, temperature, and discharge rate and provides accurate state-of-charge (SOC) in percentage (%) or milliampere-hours (mah) over a wide range of operating conditions. The MAX17055 ensures that fuel gauge error always converges to 0% as the cell approaches empty. The MAX17055 provides an accurate estimate of time-to-empty (TTE) and time-to-full (TTF) and provides three methods for reporting the age of the battery: reduction in capacity, increase in battery resistance, and cycle odometer. The MAX17055 provides the following additional features detailed in this User Guide: PCB metal sense resistor with automatic temperature compensation Support for special chemistries such as LiFePO4 Dynamic Power technology guides in throttling the processor (or other load) optimally to maintain battery above a minimum voltage while maximizing performance Time-to-Empty estimation calculated either with constant power or constant current Rev.0, December 2016 Maxim Integrated Page 4

5 1.2 ModelGauge m5 Algorithm VOLTAGE FUEL GAUGE VOLTAGE CURRENT OCV TEMPERATURE COMPENSATION LEARN OCV CALCULATION (OCV OUTPUT) OCV TABLE LOOKUP (% REMAINING OUTPUT) COULOMB COUNTER (mah OUTPUT) TIME RELAXED CELL DETECTION CAPACITY LEARN (mah PER PERCENT) REDUNDANT FUEL GAUGE SOURCES EMPTY DETECTION MIXING ALGORITHM (mah OUTPUT) MixCap REGISTER MixSOC REGISTER CURRENT + TEMP APPLICATION EMPTY COMPENSATION (BASED ON TEMPERATURE, DISCHARGE RATE, AND VOLTAGE) - + VCELL END OF CHARGE DETECTION APPLICATION OUTPUTS: RepSOC REGISTER RepCap REGISTER AvSOC REGISTER AvCap REGISTER TTE/AtTTE/TTF REGISTERS FullCap REGISTERS CELL CHEMISTRY OUTPUTS: VFOCV REGISTER Cycles REGISTER RCELL REGISTER FullCAPNom REGISTER Age REGISTER Figure 1. ModelGauge m5 flow diagram OCV Estimation and Coulomb Count Mixing The core of the ModelGauge m5 algorithm is a mixing algorithm that continuously combines the voltage fuel gauge estimation with the coulomb counter. Unlike traditional coulomb counters, which lose state information after reset, ModelGauge m5 initially uses the voltage fuel gauge. As the cell progresses through cycles in the application, coulomb counter accuracy improves and the mixing algorithm alters the weighting so that the coulomb counter result is dominant. See Figure 1. The resulting output from the mixing algorithm does not suffer accumulation drift from current measurement offset error and is more stable than a stand-alone OCV estimation algorithm. See Figure 2. Initial accuracy depends on the relaxation state of the cell. The highest initial accuracy is achieved with a fully relaxed cell. Rev.0, December 2016 Maxim Integrated Page 5

6 Figure 2. Error is filtered and bounded by combination of ModelGauge and coulomb counter. 100% OCV AND COULOMB COUNT MIXING RATIO COULOMB COUNT INFLUENCE SERVO MIXING OCV INFLUENCE 0% CELL CYCLES Figure 3. Voltage fuel gauge vs. coulomb counter influence vs. first cycles. Rev.0, December 2016 Maxim Integrated Page 6

7 1.2.2 Empty Compensation As the temperature and discharge rate of an application changes, the amount of charge available to the application also changes. The ModelGauge m5 algorithm distinguishes between remaining capacity of the cell and remaining capacity of the application and reports both results to the user. The MixCap output register tracks the charge state of the cell. This is the theoretical mah of charge that can be removed from the cell under ideal conditions extremely low discharge current and no concern for cell voltage. This result is not affected by application conditions such as cell impedance or minimum operating voltage of the application. ModelGauge m5 continually tracks the expected empty point of the application in mah. This is the amount of charge that cannot be removed from the cell by the application because of minimum voltage requirements and internal voltage drops. The IC subtracts the amount of charge not available to the application from the MixCap register and reports the result in the AvCap register. Since available remaining capacity is highly dependent on discharge rate, the AvCap register can be subject to large instantaneous changes as the application load current changes. The resulting remaining capacity or percentage can increase, even while discharging, if the load current suddenly drops or temperature increases. This result, although correct, can be very counterintuitive to the host software or end user. The RepCap output register contains a filtered version of AvCap that removes any abrupt changes in remaining capacity. RepCap converges with AvCap over time to correctly predict the application empty point while discharging or the application full point while charging. Figure 3 shows the relationship of these registers. LOAD INCREASES MixCap REGISTER ABSOLUTE mah STATE OF BATTERY NOT CONSIDERING TEMPERATURE AND DISCHARGE RATE (I.E., CAPACITY AVAILABLE IF VERY LIGHT LOAD) CAPACITY (mah) AvCap REGISTER AVAILABLE CAPACITY OF THE CELL UNDER PRESENT CONDITIONS INCREASE IN AVAILABLE CAPACITY WHEN UNDER LOAD IS COUNTERINTUITIVE TO USERS AND OPERATING SYSTEMS RepCap REGISTER REPORTED CAPACITY WITH NO SUDDEN JUMPS AND CORRECT FORECAST OF EMPTY EMPTY POINT, DEFINED BY EMPTY VOLTAGE TIME Figure 4. Handling changes in empty Converge-to-Empty The MAX17055 includes a feature that guarantees the fuel gauge output smoothly converges to 0% as the cell voltage approaches the empty voltage. As the cell voltage approaches the target empty voltage (AvgVCell approaches VEmpty), the IC smoothly adjusts the rate of change of RepSOC so that the fuel gauge reports 0% at the same time that the cell voltage reaches empty. This prevents early or late empty reporting by the fuel gauge, maximizing application run-time. See Figure 5. Rev.0, December 2016 Maxim Integrated Page 7

8 AvgVCell VEmpty RepSOC ESTIMATION TOO HIGH RepSOC RepSOC ESTIMATION TOO LOW IDEAL RepSOC RepSOC RATE OF CHANGE ADJUSTED SO THAT IT REACHES 0% AS THE CELL VOLTAGE REACHES VEmpty 0% Figure 5. Converge-to-empty % and Empty Hold The MAX17055 supports two modes that limit the RepSOC % reported until a specific condition is reached. 99% Hold: This feature limits RepSOC to not exceed 99% until a charge termination event is detected. See End-Of-Charge Detection section for more details. Empty Hold: This feature limits RepSOC to not fall below a configurable percentage (1% default) until empty voltage is crossed. This can be useful with operating systems that force system shutdown at a particular battery percentage. A Windows computer, for example, can force system shutdown or hibernate when the fuel gauge crosses 5%. So setting Empty Hold to 6% can guarantee deeper discharge to a specified voltage level (EmptyVoltHold), and thereby often extend runtime. See SOCHold register description for more information. 20% 15% VOLTAGE REFERENCE LATE EmptyVoltHold 10% 5% EARLY EmptySOCHold HOLD TIME VEmpty Figure 6. SOC hold conceptual drawing. Rev.0, December 2016 Maxim Integrated Page 8

9 SOCHold Register (D3h) The SOCHold register configures operation of the hold before empty feature and also the enable bit for 99% hold during charge. The default value for SOCHold is 0x1002. Table 1 shows the SOCHold register format. Table 1. SOCHold (D3h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D %HoldEn EmptyVoltHold EmptySOCHold EmptyVoltHold: The positive voltage offset that is added to VEmpty. At VCell = VEmpty + EmptyVoltHold point, the empty detection/learning is occurred. EmptyVoltHold has an LSb of 10mV giving a range of 0 to 1270mV. EmptySOCHold: It is the RepSOC at which RepSOC is held constant until the EmptyVoltHold condition is crossed. After empty detection occurs, RepSOC update continues as expected. EmptySOCHold has an LSb of 0.5% with a full range of 0 to 15.5%. 99%HoldEn: Enable bit for 99% hold feature during charging. When enabled, RepSOC holds a maximum value of 99% until full qualified is reached End-of-Charge Detection The IC detects the end of charge when the application current falls into the band set by the IChgTerm and VFSOC is above the FullSOCThr. By monitoring both the Current and AvgCurrent registers, the device can reject false end-of-charge events such as application load spikes or early charge-source removal. See Figure 7. When a proper end-of-charge event is detected, the device learns the FullCapRep register based on the RepCap register. If the old FullCapRep value is too high, it is adjusted on a controlled downward slope near the end of charge as defined by the MiscCfg.FUS setting until it reaches RepCap. If the old FullCapRep is too low, it is adjusted upward to match RepCap. This prevents the calculated state of charge from reporting greater than 100%. See Figure 8. Charge termination is detected by the IC when the following conditions are met: VFSOC register > FullSOCThr register AND IChgTerm x < Current register < IChgTerm x 1.25 AND IChgTerm x < AvgCurrent register < IChgTerm x 1.25 Rev.0, December 2016 Maxim Integrated Page 9

10 CHARGING AvgCurrent CURRENT 1.25 x IChgTerm 0mA x IChgTerm DISCHARGING HIGH CURRENT LOAD SPIKES DO NOT GENERATE END OF CHARGE DETECTION BECAUSE CURRENT AND AVERAGE CURRENT READINGS DO NOT FALL INTO THE DETECTION AREA AT THE SAME TIME. CHARGING AvgCurrent CURRENT 1.25 x IChgTerm 0mA x IChgTerm DISCHARGING EARLY CHARGER REMOVAL DOES NOT GENERATE END OF CHARGE DETECTION BECAUSE CURRENT AND AVERAGE CURRENT READINGS DO NOT FALL INTO THE DETECTION AREA AT THE SAME TIME. Figure 7. ModelGauge m5 rejects false end-of-charge events. Rev.0, December 2016 Maxim Integrated Page 10

11 CHARGING AvgCurrent CURRENT 1.25 x IChgTerm 0mA x IChgTerm DISCHARGING CORRECT END OF CHARGE DETECTION AREA CASE 1: OLD FullCAPRep TOO HIGH NEW FullCapRep CASE 2: OLD FullCAPRep TOO LOW RepCap Figure 8. FullCapRep learning at end of charge. ` Fuel Gauge Learning The IC periodically makes internal adjustments to cell characterization and application information to remove initial error and maintain accuracy as the cell ages. These adjustments always occur as small undercorrections to prevent instability of the learning process and prevent any noticeable jumps in the fuel-gauge outputs. Learning occurs automatically without any input from the host. In addition to estimating the state of charge, the IC observes the battery s relaxation response and adjusts the dynamics of the voltage fuel gauge. Registers used by the algorithm include: Application Capacity (FullCapRep Register): This is the total capacity available to the application at full, set through the IChgTerm and FullSOCThr registers as described in the End-of-Charge Detection section. See the FullCapRep register description. Cell Capacity (FullCapNom Register): This is the total cell capacity at full, including some capacity that sometimes is not available to the application due to high loads and/or low temperature. The IC periodically compares percent change based on an open-circuit voltage measurement vs. coulomb-count change as the cell charges and discharges, maintaining an accurate estimation of the pack capacity in mah as the pack ages. See Figure 9. Voltage Fuel-Gauge Adaptation: The IC observes the battery s relaxation response and adjusts the dynamics of the voltage fuel gauge. This adaptation adjusts the RComp0 register during qualified cell relaxation events. Empty Compensation Adaptation: The IC updates internal data whenever cell empty is detected (VCell < VEmpty) to account for cell age or other cell deviations from the characterization information. Rev.0, December 2016 Maxim Integrated Page 11

12 100% RELAXATION EVENTS 90% 80% D%4 70% D%1 VFSOC (%) 60% 50% 40% 30% 20% D%2 D%5 OBSERVED SIZE OF BATTERY: FullCapNom = dqacc dpacc x 100% 10% 0% D%3 WHERE: 1200mAh DQ4 dqacc = DQ1 + DQ2 + DQ3... dpacc = D%1 + D%2 + D% mAh DQ1 1000mAh 900mAh DQ5 COULOMB COUNT (mah) 800mAh 700mAh 600mAh 500mAh 400mAh 300mAh 200mAh DQ2 100mAh DQ3 0mAh Figure 9. ModelGauge m5 learns full capacity during arbitrary cycling Predicting Run Time for a Hypothetical Load The MAX17055 provides an AtRate register which can be used to predict the remaining percentage or time associated with different load conditions. This can be used for system power-management decisions, to prevent or limit the load according to the battery capability or runtime requirements. The AtRate function allows host software to see theoretical remaining time or capacity for any hypothetical load current. AtRate can be used for power management by limiting system loads depending on present conditions of the battery. To use AtRate, set AtRateEn = 1 in Config2. To query the remaining capacity or time associated with a hypothetical load, write the load to AtRate (negative value) and read the result from AtTTE, AtAvSOC, or AtAvCap. Host software should wait 351ms in active mode or 11.25s in hibernate after writing the AtRate register before reading any of the result registers. Exit hibernate with soft-wakeup command to provide faster answers. See the Command register (60h) Lithium Iron Phosphate Support (LiFePO4) The MAX17055 supports lithium iron phosphate (LiFePO4) batteries with a special model configuration. To produce good state-of-charge (%) accuracy performance, it is necessary to characterize and model the specific LiFePO4 cells being used. The MAX17055 provides additional algorithm support specifically for the challenges associated with LiFePO4 and other flat OCV chemistries. The OCV curve of LiFePO4 is much flatter than conventional lithium cobalt chemistries, which produces a greater sensitivity to the algorithm's interpretation of cell voltage and open-circuit voltage. A LiFePO4 battery has two voltage plateaus. Using the ScOcvLim register as a keep-out window, the fuel gauge limits OCV interpretation within a specified OCV voltage range. The default keep-out window is 3.275V to 3.350V. Rev.0, December 2016 Maxim Integrated Page 12

13 To configure the MAX17055 for LiFePO4 support, do the following: 1. Characterize the battery. The characterization data should be translated by Maxim into a battery model. 2. Configure the ScOcvLim register according to the desired OCV keep-out window. 3. Configure ModelCfg.ModelID = 6 to enable LiFePO4 mode. 4. Load the rest of the battery model. Refer to the MAX17055 Software Implementation Guide. ScOcvLim Register (D1h) Initial Value: 0x479E This register only has usage when ModelCfg.ModelID is selected as 6 (LiFePO4). Table 2 shows the register format. Table 2. ScOcvLim (D1h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 OCV_Low_Lim OCV_Delta OCV_Low_Lim: Defines the lower limit for keep-out OCV region. A 5mV resolution gives a 2.56 to 5.12V range. Lower limit voltage of OCV keep-out region is calculated as 2.56V + OCV_Low_Lim x 5mV. Default value is 0x8F. OCV_Delta: Defines the delta between lower and upper limits for keep-out OCV region. A 2.5mV resolution gives a 0 to 320mV range. Upper limit voltage of OCV keep-out region is calculated as 2.56V + OCV_Low_Lim x 5mV + OCV_Delta x 2.5mV. Default value is 0x1E. Default OCV_low is 3275mV and OCV_high is 3350mV. 1.3 ModelGauge m5 Standard Register Formats Unless otherwise stated during a given register's description, all IC registers follow the same format depending on the type of register. See Table 3 for the resolution and range of any register described hereafter. Note that current and capacity values are displayed as a voltage and must be divided by the sense resistor to determine Amps or Amp-hours. It is strongly recommended to use the nrsense (1CFh) register to store the sense resistor value for use by host software. REGISTER TYPE LSb SIZE MINIMUM VALUE Table 3. Standard Register Formats MAXIMUM VALUE NOTES Capacity 5.0μVH/R SENSE 0.0μVH mVH/R SENSE Equivalent to 0.5mA with a 0.010Ω sense resistor. Percentage 1/256% 0.0% % 1% LSb when reading only the upper byte. Voltage mV 0.0V V Current μV/R SENSE -51.2mV/R SENSE mV/R SENSE Signed 2's complement format. Equivalent to μA with a 0.010Ω sense resistor. Temperature 1/256 C C C Signed 2's complement format. 1 C LSb when reading only the upper byte. Resistance 1/4096 Ω 0.0 Ω Ω Time 5.625s 0.0s h Special Format details are included with the register description. 1.4 Analog Measurements To properly fuel gauge a battery the MAX17055 continually monitors the battery voltage, battery temperature, and current flow into and out of the battery. The following sections detail how these measurements occur Current Measurements Current flow through the battery is determined by making voltage measurements between the CSP and CSN pins. An external sense resistor placed between these pins creates a positive voltage when the battery is charged and a negative voltage when the battery is discharged. The resulting value is reported in μv or μvh and must be divided by the sense resistor value in ohms to convert to current. Rev.0, December 2016 Maxim Integrated Page 13

14 The value of the sense resistor determines the range and the resolution of current values that can be reported. The MAX17055 has a maximum measurement range of ±51.2mV and a reporting resolution of μV. Table 4 shows the measurement ranges and resolutions of several common-sense resistor values. Table 4. Measurement Range and Resolution vs. Sense Resistor Value SENSE RESISTOR (mω) MEASUREMENT RANGE (A) MEASUREMENT RESOLUTION (μa) 20.0 ± ± ± (board trace) ± ± CGain Register (2Eh) and COff (2Fh) Registers Initial Value: CGain = 0x0400 and COff = 0x0000 The CGain and COff registers adjust the gain and offset of the current measurement result. The current measurement A/D is factory trimmed to data-sheet accuracy without the need for the user to make further adjustments. The default power-up settings for CGain and COff apply no adjustments to the Current register reading. For specific application requirements, the CGain and COff registers can be used to adjust readings as follows: Current register = Current A/D reading (CGain register/0400h) + COff register Copper Trace Current Sensing The MAX17055 has the ability to measure current using a copper board trace instead of a traditional sense resistor. The main difference being the ability to adjust to the change in sense resistance over temperature. The MAX17055 EV kit includes a PCB trace that demonstrates this functionality. Board-to-board variation makes it challenging to use PCB current-sensing to achieve the normally stringent requirements for fuel gauging. However, the MAX17055 can meet this challenge due to the robust ModelGauge m5 EZ algorithm, plus PCB compensations provided in the MAX To configure the MAX17055 for accurate PCB current-sensing, configure the following: 1. Set CGTempCo to 0x20C8, corresponding to copper at 0.4% per C. Set CGTempCo to 0 to disable copper current sensing when using a normal sense resistor. 2. Set CGAIN according to the room-temperature resistance associated with the PCB trace. Configure CCurve to compensate for the self-heating of the trace. The PCB trace temperature mismatches the MAX17055 die temperature (used for compensation), especially at high currents. This self-heating is also compensated by the MAX17055 using the CCurve configuration register. For 1oz copper, a length to width ratio of 6:1 creates a Ω sense resistor that is suitable for most applications. PCB manufacturing process can produce a trace-resistance variation of ±20%. ModelGauge m5 adapts to this variation and reports SOC accurately. The adaptation is similar to supporting battery-to-battery, full-capacity variation. Curve Register (B9h) The upper half of the Curve register applies curvature correction current measurements made by the IC when using a copper trace as the sense resistor. See the Temperature Measurements section for a description of the lower half of the register. Table 5. Curve (B9h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 Metal Trace Curve Thermistor Curve Rev.0, December 2016 Maxim Integrated Page 14

15 Initial Value: 0x0 CGTempCo Register (B8h) If CGTempCo is nonzero then CGTempCo is used to adjust current measurements for temperature. CGTempCo has a range of 0% to % per C with a step size of /0x10000 percent per C. If a copper trace is used to measure battery current, CGTempCo should be written to 0x20C8 or 0.4% per C, which is the approximate temperature coefficient of a copper trace Temperature Measurements The MAX17055 can measure and report its own internal temperature or report an external temperature by using an NTC thermistor divider network connected to the THRM and AIN pins. Only one measurement path can be active at any time. When measuring temperature externally, the MAX17055 takes a raw percentage reading and converts that value to temperature using gain (TGain register), offset (TOff register), and second order curve adjustment (Curve register). This allows an accurate temperature to be reported over a variety of different NTC thermistor options. Internal temperature measurement does not use TGain, TOff, or Curve. AIN Register (27h) External temperature measurement on the AIN pin is compared to the THRM pin voltage. The MAX17055 stores the result as a ratio-metric value from 0% to 100% in the AIN register with an LSB of %. The TGain, TOff, and Curve register values are then applied to this ratiometric reading to convert the result to temperature. TGain (2Ch) Register/TOff (2Dh) Register Initial Value: TGain = 0xEE56 and TOff = 0x1DA4 The TGain, TOff, and Curve registers are used to calculate temperature from the measurement of the AIN pin with an accuracy of ±3 C over a range of -40 C to +85 C. Table 6 lists the recommended TGain, TOff, and Curve register values for common NTC thermistors. Table 6. Thermistor Configuration THERMISTOR R 25C (kω) BETA RECOMMENDED RECOMMENDED RECOMMENDED TGain Toff Curve.TCurve Semitec 103AT-2, Murata NCP15XH103F03RC xEE56 0x1DA4 0x0025 Fenwal LAG-A xF49A 0x16A1 0x0064 TDK Type F xF284 0x18E8 0x0035 Default Value: 0x0025 Curve Register (B9h) The lower half of the Curve register applies thermistor measurement curvature correction to allow thermistor measurements to be accurate over a wider temperature range. A ±3 C accuracy can be achieved over a -40 C to 85 C operating range. See the Copper Trace Current Sensing section for a description of MCurve. Rev.0, December 2016 Maxim Integrated Page 15

16 Table 7. Curve (B9h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 MCurve TCurve Voltage Measurements Measurement of the battery voltage is performed by determining the voltage difference between the V BATT and CSP pins. 1.5 Dynamic Battery Power To achieve better run-time and CPU performance, the MAX17055 supports a variant of Intel Turbo Boost Technology that provides the ondemand battery capability used for managing pulse-loads such as CPU loads. A CPU requires the battery to deliver short pulses of high power. To support these high pulses without system undervoltage, the MAX17055 indicates the peak power levels that can be taken from the battery. The host can use this information to set its maximum current in accordance with battery power capability. In many 1-Series applications, the system requires at least 3.3V to operate correctly. By configuring the MAX17055 for Dynamic Battery Power, the system's loads can be controlled or limited to stay within the battery's capability and ensure that a minimum system voltage (MinSysVolt) is not crossed until the battery is a very low state. The implementation of Intel Dynamic Battery Power Technology v2.0 relies on new functions and corresponding registers. This document defines those new functions. The implementation in the MAX17055 includes all of the same registers as the Intel spec. However, it uses different LSB's and addresses to match the current/voltage/resistance LSBs used elsewhere in the MAX The LSB is µV for voltage, mA for current, and 0.8mW for power and 0.244mΩ for resistance Dynamic Power Performance Figure 10 shows the performance of dynamic battery power. In this test, the load was limited according to the feedback from the MaxPeakCurrent register. The MAX17055 continues to update the MaxPeakCurrent output as the state-of-charge, load, or temperature changes. The performance of the MaxPeakCurrent register ensures that the voltage is limited to, but does not not fall below the MinSysVolt setting DYNAMIC POWER PERFORMANCE MinSysVolt = 3.4V MININUM VOLT AvgVolt VOLTAGE (V) 0 AvgCurrent CURRENT (ma) MEASURED PEAK CURRENT MPPCurrent FROM GAUGE TIME (hours) 2.2 Figure 10. Dynamic battery power limits voltage spikes when the load limits by MaxPeakCurrent. Rev.0, December 2016 Maxim Integrated Page 16

17 1.5.2 Dynamic Power Output Registers MPPCurrent() (D9h) The MAX17055 estimates the maximum instantaneous peak current of the battery pack in ma, which the battery can support for up to 10ms, given the external resistance and required minimum voltage of the voltage regulator. The MPPCurrent() value is negative and updates every 175ms. SPPCurrent() (DAh) The MAX17055 estimates the sustained peak current of the battery pack in ma, which the battery can support for up to 10s, given the external resistance and required minimum voltage of the voltage regulator. The SPPCurrent() value is negative and updates every 175ms. MaxPeakPower() Register (D4h) The MAX17055 estimates the maximum instantaneous peak output power of the battery pack in mw, which the battery can support for up to 10ms, given the external resistance and required minimum voltage of the voltage regulator. The MaxPeakPower() value is negative (discharge) and updates every 175ms. LSB is 0.8mW. Calculation: MaxPeakPower = MPPCurrent() x AvgVCell SusPeakPower() Register (D5h) The fuel gauge estimates the sustainable peak output power of the battery pack in mw, which the battery supports for up to 10s, given the external resistance and required minimum voltage of the voltage regulator. The SusPeakPower() value is negative and updated each 175ms. LSB is 0.8mW. Calculation: SusPeakPower = SPPCurrent() x AvgVCell Dynamic Power Configuration Registers The following registers provide the battery capability estimates from the Dynamic Power calculations. PackResistance() (D6h) When the MAX17055 is installed host-side, simply set PackResistance to zero, since the MAX17055 can observe the total resistance between it and the battery. However, when the MAX17055 is installed pack-side, configure PackResistance according to the total non-cell pack resistance. This should account for all resistances due to cell interconnect, sense resistor, FET, fuse, connector, and other resistance between the cells and output of the battery pack. The cell internal resistance should not be included and is estimated by the MAX x1000 is 1000mΩ, which results in an LSB of mΩ per LSB. SysResistance() (D7h) Set SysResistance according to the total system resistance. This should include any connector and PCB trace between the MAX17055 and the system at risk for dropout when the voltage falls below MinSysVolt. Rev.0, December 2016 Maxim Integrated Page 17

18 SysResistance() is initialized to a default value upon removal or insertion of a battery pack. Writes with this function overwrite the default value. 0x1000 is 1000mΩ, which results in an LSB of mΩ per LSB. MinSysVoltage() (D8h) Set MinSysVoltage according to the minimum operating voltage of the system. This is generally associated with a regulator dropout or other system failure/shutdown. The system should still operate normally until this voltage. MinSysVoltage() is initialized to the default value (3.0V). Writes with this function overwrite the default value. RGain Register (43h) Initial Value: 0x8080 The RGain register sets the value of RGain1 and RGain2 during DBPT register calculation. Table 8 shows the register format. Table 8. RGain (43h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 RGain1 RGain2 SusToMaxRatio RGAIN provides additional limitations to ensure that the dynamic power outputs (MPPCurrent, MaxPeakPower, SPPCurrent, SusPeakPower) are conservative, thereby preventing any risk of system shutdown. RGain1: Gain resistance used for peak current and power calculation. RGain1 = 80% % x RG1. The range of RGain1 is between 80~120%. RGain2: Gain resistance used for peak current and power calculation. RGain2 = 60% + 5% x RG2. The range of RGain1 is between 60~140%. SusToMaxRatio: Used to calculate the maximum ratio between SPPCurrent to MPPCurrent. The maximum value of SPPCurrent = MPPCurrent x (0.75-SusToPeakRatio x 0.04). 1.6 Typical Applications Circuit The application example in Figure 11 shows the MAX17055 monitoring a captive battery. An external thermistor measurement circuit is not required since measurements are made internal to the IC. Cell voltage is measured from V BATT to CSP and current measurements are made differentially from CSN to CSP. Results are reported on the SDA, SCL, and ALRT lines. The other pin connections to the IC are not used in this application and can be left unconnected. Rev.0, December 2016 Maxim Integrated Page 18

19 BATTERY P SYSTEM PK+ 0.1µF BATT MAX17055 ALRT SDA HOST µp PROTECTION 0.1µF REG CSP SCL CSN PK- PCB TRACE OR 1mΩ 1000mΩ Figure cell application circuit. 1.7 Serial Number Feature Each MAX17055 provides a unique serial number ID. To read this serial-number, clear AtRateEn and DPEn in Config2. After the read operation is completed, the MAX17055 sets Status2.SNReady flag to indicate that serial number read operation is completed. 128-bit serial information read from MAX17055 overwrites Dynamic Power and AtRate output registers. To continue Dynamic Power and AtRate operations after reading serial number, the host should set Config2.AtRateEn and Config2.DPEn to 1. Table 9. Accessing The Serial Number ADDRESS Config2.AtRateEn = 1 Config2.AtRateEn = 0 && Config2.DPEn = 1 Config2.DPEn = 0 0xD4 MaxPeakPower Serial Number Word0 0xD5 SusPeakPower Serial Number Word1 0xD9 MPPCurrent Serial Number Word2 0xDA SPPCurrent Serial Number Word3 0xDC AtQResidual Serial Number Word4 0xDD AtTTE Serial Number Word5 0xDE AtAvSoc Serial Number Word6 0xDF AtAvCap Serial Number Word7 1.8 Determining Fuel Gauge Accuracy To determine the true accuracy of a fuel gauge, as experienced by end users, the battery should be exercised in a dynamic manner. The enduser accuracy cannot be completely understood with only simple cycles. To challenge a correction-based fuel gauge, such as a coulomb counter, test the battery with partial loading sessions. For example, a typical user can operate the device for 10 minutes and then stop use for an hour or more. A robust test method includes these kinds of sessions many times at various loads, temperatures, and duration. Refer to the Application Note 4799: Cell Characterization Procedure for a ModelGauge m3/modelgauge m5 Fuel Gauge. Rev.0, December 2016 Maxim Integrated Page 19

20 1.8.1 Initial Accuracy The IC uses the first voltage reading after power-up or after cell insertion to determine the starting output of the fuel gauge. It is assumed that the cell is fully relaxed prior to this reading; however, this is not always the case. If there is a load or charge current at this time, the initial reading is compensated using a default battery resistance of 40mΩ to estimate the relaxed cell voltage. If the cell was recently charged or discharged, the voltage measured by the IC may not represent the true state of charge of the cell, resulting in initial error in the fuel gauge outputs. In most cases, this error is minor and is quickly removed by the fuel gauge algorithm during the first hour of normal operation. 1.9 ModelGauge m5 Registers Table 10 shows the user memory space. Register locations shown in gray are reserved locations and should not be written to. PAGE/ WOR D Table 10. MAX17055 Register Map 00h 10h 20h 30h 40h B0h D0h 0h Status FullCapRep TTF Reserved UserMem1 Status2 RSense/UserMem3 1h VAlrtTh TTE DevName Reserved Reserved Power ScOcvLim 2h TAlrtTh QRTable00 QRTable10 QRTable20 QRTable30 ID/UserMem2 Reserved 3h SAlrtTh FullSocThr FullCapNom Reserved RGain AvgPower SOCHold 4h AtRate RCell Reserved DieTemp Reserved IAlrtTh MaxPeakPower 5h RepCap Reserved Reserved FullCap dqacc Reserved SusPeakPower 6h RepSOC AvgTA Reserved Reserved dpacc CVMixCap PackResistance 7h Age Cycles AIN Reserved Reserved CVHalfTime SysResistance 8h Temp DesignCap LearnCfg RComp0 Reserved CGTempCo MinSysVoltage 9h VCell AvgVCell FilterCfg TempCo ConvgCfg Curve MPPCurrent Ah Current MaxMinTemp RelaxCfg VEmpty VFRemCap HibCfg SPPCurrent Bh AvgCurrent MaxMinVolt MiscCfg Reserved Reserved Config2 ModelCfg Ch QResidual MaxMinCurr TGain Reserved Reserved VRipple AtQResidual Dh MixSOC Config TOff FStat QH RippleCfg AtTTE Eh AvSOC IChgTerm CGain Timer Reserved TimerH AtAvSOC Fh MixCap AvCap COff ShdnTimer Reserved Reserved AtAvCap Rev.0, December 2016 Maxim Integrated Page 20

21 VCell AvCap/AvSOC CURRENT RepCap/RepSOC ANALOG INPUTS TEMPERATURE AvgVCell AvgCurrent MixCap/MixSOC FullCap FullCapRep APPLICATION SPECIFIC CELL CHARACTERIZATION INFORMATION AvgTemperature VEmpty DesignCap IChgTerm Characterization Table Characterization Table CHARACTERIZATION TABLES QRTable00,10,20,30 FullSOCThr RComp0 FullCapNom TempCo ModelGauge m5 EZ ALGORITHM RippleCfg ConvgCfg CVCfg LearnCfg FilterCfg RelaxCfg MiscCfg ALGORITHM CONFIGURATION FullCapNom RCell TTE/TTF/AtTTE VFOCV/VFSOC VRipple AGE CYCLES FullCapNom CYCLES RComp0 TempCo LEARNED INFORMATION ModelGauge ALGORITHM OUTPUTS IAvgEmpty AtRate Figure 12. ModelGauge m5 registers ModelGauge m5 Algorithm Battery Parameters The following registers are inputs to the ModelGauge m5 algorithm and store characterization information for the application cells as well as important application specific specifications. They are described only briefly here. Contact Maxim for information regarding cell characterization. VEmpty Register (3Ah) Initial Value: 0xA561 (3.3V/3.88V) The VEmpty register sets thresholds related to empty detection during operation. Table 11 shows the register format. Table 11. VEmpty (3Ah) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 VE VR VE (Empty Voltage Target, During Load): The fuel gauge provides capacity and percentage relative to the empty voltage target, eventually declaring 0% at VE. A 10mV resolution gives a 0 to 5.11V range. This value is written to 3.3V after reset. VR (Recovery Voltage): Sets the voltage level for clearing empty detection. Once the cell voltage rises above this point, empty voltage detection is reenabled. A 40mV resolution gives a 0 to 5.08V range. This value is written to 3.88V, which is recommended for most applications. Rev.0, December 2016 Maxim Integrated Page 21

22 Register Type: Capacity DesignCap Register(18h) The DesignCap register holds the expected capacity of the cell. This value is used to determine age and health of the cell by comparing against the measured present cell capacity. ModelCfg Register (DBh) The ModelCFG register controls basic options of the EZ algorithm. Table 12 shows the register format. Table 12. ModelCFG (DBh) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 Refresh VChg 0 0 ModelID Refresh: Set 1 to command the model refreshing. After firmware executes the command, it will be cleared by firmware. ModelID: Choose from one of 3 common classifications of lithium cobalt batteries supported by EZ, without characterization. For the majority of batteries, use ModelID = 0. ModelID = 0: Use for most lithium cobalt-oxide variants (a large majority of lithium in the market-place). ModelID = 6: Use for lithium iron-phosphate (LiFePO4). However, since LiFePO4 is a challenging chemistry, custom characterization is recommended. VChg: Set VChg = 1 for 4.35V or 4.4V models. Set VChg = 0 for 4.2V models. Register Type: Current IChgTerm Register (1Eh) Initial Value: 0x0640 (250mA on 10mΩ) The IChgTerm register allows the device to detect when a charge cycle of the cell has completed. IChgTerm should be programmed to the exact charge termination current used in the application. The device detects end of charge if all the following conditions are met: VFSOC register > FullSOCThr register AND IChgTerm x < Current register < IChgTerm x 1.25 AND IChgTerm x < AvgCurrent register < IChgTerm x 1.25 See the End-of-Charge Detection section for more details. Register Type: Percentage Initial Value: 95% FullSOCThr Register (13h) The FullSOCThr register gates detection of end-of-charge. VFSOC must be larger than the FullSOCThr value before IChgTerm is compared to the AvgCurrent register value. The recommended FullSOCThr register setting for most custom characterized applications is 95% (default, 0x5F05). For EZ Performance applications the recommendation is 80% (0x5005). See the IChgTerm register description and the End-Of- Charge Detection section for details. Table 13 shows the register format. Table 13. FullSOCThr (13h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 FullSOCThr Rev.0, December 2016 Maxim Integrated Page 22

23 XTable0 (90h) to XTable15 (9Fh) Registers Cell characterization information used by the ModelGauge algorithm to determine capacity versus operating conditions. This table comes from battery characterization data. OCVTable0 (80h) to OCVTable15 (8Fh) Registers Cell characterization information used by the ModelGauge algorithm to determine capacity versus operating conditions. This table comes from battery characterization data. QRTable00 (12h) to QRTable30 (42h) Registers The QRTable00 to QRTable30 register locations contain characterization information regarding cell capacity under different application conditions. RComp0 Register (38h) The RComp0 register holds characterization information critical to computing the open-circuit voltage of a cell under loaded conditions. TempCo Register (39h) The TempCo register holds temperature compensation information for the RComp0 register value ModelGauge m5 Output Registers Register Type: Capacity RepCap Register (05h) RepCap or reported remaining capacity in mah. This register is protected from making sudden jumps during load changes. Register Type: Capacity AtAvCap Register (DFh) The AtAvCap register holds the estimated remaining capacity of the cell based on the theoretical load current value of the AtRate register. The value is stored in terms of µvh and must be divided by the application sense-resistor value to determine the remaining capacity in mah. Register Type: Percentage RepSOC Register (06h) RepSOC is the reported state-of-charge percentage output for use by the application GUI. Rev.0, December 2016 Maxim Integrated Page 23

24 Register Type: Percentage AtAvSOC Register (DEh) The AtAvSOC register holds the theoretical state of charge of the cell based on the theoretical load of the AtRate register. The register value is stored as a percentage with a resolution of 1/256 % per LSB. The high byte indicates 1% resolution. Register Type: Capacity FullCapRep Register (10h) This register reports the full capacity that goes with RepCap, generally used for reporting to the GUI. Most applications should only monitor FullCapRep, instead of FullCap or FullCapNom. A new full-capacity value is calculated at the end of every charge cycle in the application. Register Type: Capacity FullCap Register (35h) FullCap is the full discharge capacity compensated according to the present conditions. A new full-capacity value is calculated continuously as application conditions change (temperature and load). Register Type: Capacity FullCapNom Register (23h) This register holds the calculated full capacity of the cell, not including temperature and empty compensation. A new full-capacity nominal value is calculated each time a cell relaxation event is detected. This register is used to calculate other outputs of the ModelGauge m5 algorithm. Register Type: Time TTE Register (11h) The TTE register holds the estimated time to empty for the application under present temperature and load conditions. The TTE value is determined by relating AvCap with AvgCurrent. The corresponding AvgCurrent filtering gives a delay in TTE, but provides more stable results. AtTTE Register (DDh) Register Type: Time The AtTTE register can be used to estimate time to empty for any theoretical load entered into the AtRate register. Register Type: Time TTF Register (20h) The TTF register holds the estimated time to full for the application under present conditions. The TTF value is determined by learning the constant current and constant voltage portions of the charge cycle based on experience of prior charge cycles. Time to full is then estimated by comparing present charge current to the charge termination current. Operation of the TTF register assumes all charge profiles are consistent in the application. Cycles Register (17h) Rev.0, December 2016 Maxim Integrated Page 24

25 The Cycles register maintains a total count of the number of charge/discharge cycles of the cell that have occurred. The result is stored as a percentage of a full cycle. For example, a full charge/discharge cycle results in the Cycles register incrementing by 100%. The Cycles register accumulates fractional or whole cycles. For example, if a battery is cycles 10% x 10 times, then it tracks 100% of a cycle. The Cycles register has a full range of 0 to cycles with a 1% LSb. Initial Value: 0x0002 Status Register (00h) The Status register maintains all flags related to alert thresholds and battery insertion or removal. Table 14 shows the Status register format. Table 14. Status (00h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 Br Smx Tmx Vmx Bi Smn Tmn Vmn dsoci Imx X X Bst Imn POR X POR (Power-On Reset): This bit is set to a 1 when the device detects that a software or hardware POR event has occurred. This bit must be cleared by system software to detect the next POR event. POR is set to 1 at power-up. Imn and Imx (Minimum/Maximum Current Alert Threshold Exceeded): These bits set to a 1 whenever a Current register reading is below (Imn) or above (Imx) the IAlrtTh thresholds. These bits may or may not need to be cleared by system software to detect the next event. See Config.IS bit description. Imn and Imx are cleared to 0 at power-up. Vmn and Vmx (Minimum/Maximum Voltage Alert Threshold Exceeded): These bits set to a 1 whenever a VCell register reading is below (Vmn) or above (Vmx) the VAlrtTh thresholds. These bits may or may not need to be cleared by system software to detect the next event. See Config.VS bit description. Vmn and Vmx are cleared to 0 at power-up. Tmn and Tmx (Minimum/Maximum Temperature Alert Threshold Exceeded): These bits set to a 1 whenever a Temperature register reading is below (Tmn) or above (Tmx) the TAlrtTh thresholds. These bits may or may not need to be cleared by system software to detect the next event. See Config.TS bit description. Tmn and Tmx are cleared to 0 at power-up. Smn and Smn (Minimum/Maximum SOC Alert Threshold Exceeded): These bits set to a 1 whenever SOC is below (Smn) or above (Smx) the SAlrtTh thresholds. These bits may or may not need to be cleared by system software to detect the next event. See Config.SS and MiscCFG.SACFG bit descriptions. Smn and Smx are cleared to 0 at power-up. dsoci (State of Charge 1% Change Alert): This is set to 1 whenever the RepSOC register crosses an integer percentage boundary such as 50.0%, 51.0%, etc. Must be cleared by host software. Bst (Battery Status): Useful when the IC is used in a host side application. This bit is set to 0 when a battery is present in the system and set to 1 when the battery is absent. Bst is set to 0 at power-up. Bi (Battery Insertion): Useful when the IC is used in a host-side application. This bit is set to a 1 when the device detects that a battery has been inserted into the system by monitoring the AIN pin. This bit must be cleared by system software to detect the next insertion event. Bi is cleared to 0 at power-up. Br (Battery Removal): Useful when the IC is used in a host side application. This bit is set to a 1 when the system detects that a battery has been removed from the system. This bit must be cleared by system software to detect the next removal event. Br is cleared to 0 at power-up. X (Don t Care): This bit is undefined and can be logic 0 or 1. Register Type: Percentage Age Register (07h) The Age register contains a calculated percentage value of the application s present cell capacity compared to its original design capacity. The result can be used by the host to gauge the battery pack health as compared to a new pack of the same type. The equation for the register output is: Age Register(%) = 100% x (FullCapRep register/designcap register) Rev.0, December 2016 Maxim Integrated Page 25

26 For example, if DesignCap = 2000mAh and FullCapRep = 1800mAh, then Age = 90% (or 0x5A00) Initial Value: 0x0000 TimerH (BEh) and Timer (3Eh) Registers TimerH and Timer provide a long-duration time count since last POR. 3.2 hour LSb gives a full scale range for the register of up to years. The Timer register LSb is 175.8ms giving a full-scale range of 0 to 3.2 hours. TimerH and Timer can be interpreted together as a 32-bit timer. Register Type: Resistance Initial Value: 0x0290 (160mΩ) RCell Register (14h) The RCell register provides the calculated internal resistance of the cell. RCell is determined by comparing open-circuit voltage (VFOCV) against measured voltage (VCell) over a long time period while under load or charge current. Initial Value: 0x0000 VRipple Register (BCh) The VRipple register holds the slow average RMS ripple value of VCell register reading variation compared to the AvgVCell register. The default filter time is 22.5 seconds. See RippleCfg register description. VRipple has an LSb weight of 1.25mV/ ModelGauge m5 Algorithm Configuration Registers The following registers allow operation of the ModelGauge m5 algorithm to be adjusted for the application. It is recommended that the default values for these registers be used. Register Type: Current AtRate Register (04h) Host software should write the AtRate register with a negative two s-complement 16-bit value of a theoretical load current prior to reading any of the at-rate output registers (AtTTE, AtAvSOC, AtAvCap). Initial Value: 0xCEA4 FilterCfg Register (29h) The FilterCfg register sets the averaging time period for all A/D readings, for mixing OCV results and coulomb count results. It is recommended that these values are not changed, unless absolutely required by the application. Table 15 shows the FilterCfg register format. Table 15. FilterCfg (29h) Format D15 D14 D13 D12 D11 D10 D9 D8 D7 D6 D5 D4 D3 D2 D1 D0 1 1 TEMP MIX VOLT CURR Rev.0, December 2016 Maxim Integrated Page 26