In critical applications like very low temperatures with dynamic load, fuel gauges based on the Impedance Track™ algorithm may report a sudden state of charge (SOC) jump from several or tens of percent to zero towards the end of discharge, which causes system abruptly shut-down. This application report discusses the SOC jump issue, explains the possible root causes, and accordingly gives feasible solutions to eliminate it.
Impedance Track is a registered trademark of Texas Instruments.
All other trademarks are the property of their respective owners.
Battery fuel gauge is used to calculate the remaining capacity and full-charged capacity under any given load. It can tell the host side when the battery is going to be empty or full, what state of the battery right now, how much life time has left. This information will help the host side to do smarter power management for the whole system and also ensure safe charge/discharge. An accurate gauge is critical for the longer battery runtime and life span.
Impedance Track technology is a fuel-gauging algorithm developed by Texas Instruments. What makes it unique and more accurate than other solutions is the self-learning mechanism. IT algorithm continually measures and updates battery impedance and full chemical capacity. IT-based gauge can maintain its accuracy even as the battery ages. However, if the gauge was improperly configured, the gauge may perform badly in some applications. A possible issue encountering is the SOC jump issue.
Figure 1(a) describes a low temperature discharge test of the portable device using TI gauge BQ27Z561-R1. There is a significant SOC drop from 40% to 0% that causing the device abruptly shut down. Essentially, SOC jump issue is an overestimation issue. Due to overestimation, the gauge ‘thinks’ there is still 40% of capacity left before the voltage reaches Termination Voltage (EDV). But gauge is forced to report SOC=0% once the voltage hits the EDV, causing a 40% drop of reported SOC at the EDV.
The undesired drop can be eliminated using following methods of this document, the optimization result is shown in Figure 1(b), which will surely improve the experience and prevent data loss due to abnormal shutdown.
(a) 40% SOC Jump (before optimization) |
(b) No SOC Jump (after optimization) |
Figure 2 shows that the Impedance Track gauge uses a battery model to estimate how much capacity is still available until the cell voltage reaches EDV. It is a model-based algorithm. Battery model (including impedance model, temperature model, transient model…) needs to be determined before using the gauge. To solve reported SOC jump issue, firstly you should guarantee the model parameters are well extracted.
Texas Instruments has a database of thousands of battery models, as shown in Figure 3. Each model has a unique number referred to as the chemistry identifier, or ‘ChemID’. It is important that the ChemID programmed into the gauge was either generated by TI for that battery or a close match to an existing ChemID in TI database using our online tool - GPCCHEM. The ChemID identification requires running a relax-discharge-relax test while logging data using the gauge’s GUI (bqStudio) and then using GPCCHEM tool with the logged data to identify a close match. If there is no match (model error > 3%), then the cells have to be sent to TI for characterization and ChemID generation. Contact a local field applications engineer if cells have to be sent to TI.
SOC jump issue is more severe in very low temperature test, such as -15℃. Note that TI standardizes the modeling temperatures to 0℃, 25℃ and 50℃. ChemID was not operated at lower temperatures. it is recommended that the cells be characterized at 0℃ and that low temperature optimization tests can be performed using the online tool - GPCRB.
GPCRB is a dedicated tool to optimize low temperature performance of IT gauges. For IT gauges, resistance in data flash is normalized to 25°C as shown in Equation 1.
Where, DOD is the depth of discharge, R[DOD] is the measured resistance at a given DOD, Rb[DOD] is the temperature coefficient of impedance change at a given DOD stored as a reserved data flash table, and T is temperature in °C. GPCRB tool can modify the Rb values to improve Ra calculation accuracy in low temperature test, as shown in Figure 4, so as to improve SOC accuracy and relieve SOC jump problem at low temperature. This tool also obtains thermal model parameters that do not update in gauges, which helps with high rate tests accuracy. This tool provides Ra0_charge value as well, that helps to reach 100% SOC during charge more accurately.
GPCRB tool requires two log files of a relax-discharge-relax test performed under load and temperature conditions similar to an actual device, and also requires the gg file exported from your gauge EVM or device PCB using bqStudio after chosen ChemID data has been programmed. This tool will generate a new chemistry file that provides both improved resistance temperature compensation and learned values.