[Power Switch Tips] Accurately predict the remaining battery power and runtime of your device

"30 Tips on Power Switch Design" collects 30 tips on power supply design written by experts from TI. To facilitate reading and learning for engineer friends, we will share these tips with you on the WeChat platform. Stay tuned.

Tip 10

Battery Capacity: How to Accurately Predict Remaining Battery Charge and Runtime in Portable Devices

challenge

在过去的几年里，诸如笔记本电脑、手机以及媒体播放器等便携式设备的数量显著增长。这些具有更多特性与功能的设备要求更高的电量，所以电池必须能够提供更多的能量以及更长的运行时间。对于电池供电的系统而言，最大的挑战在于电池的运行时间。通常，电子系统设计人员通常将注意力集中在提高 dc-dc 电源转换效率上以此来延长电池的运行时间，而往往会忽略与电源转换效率和电池容量同等重要的电池电量监测计的精确度问题。如果电池电量监测计的误差范围是±10%，那么就会有相当于 10% 的电池容量或运行时间损失掉。然而，电池的可用电量与其放电速度、工作温度、老化程度以及自放电特性具有函数关系。此外，传统的电池电量监测计还要求对电池进行完全充电和完全放电以更新电池容量，但是这在现实应用中很少发生，因而造成了更大的测量误差。因此，在电池运行周期内很难精确预测电池剩余容量及工作时间。

solution

The most common technique used today for battery charge monitoring is coulomb counting or integrating the current flowing into and out of the battery. This works very well for a new battery that has just been fully charged. However, as the battery ages and self-discharges, this method becomes less effective. There is no way to measure the self-discharge rate. Therefore, a predefined self-discharge rate formula is usually used to correct for it. This method is not very accurate because the self-discharge rate varies from battery to battery and one model does not fit all batteries. Another drawback of the coulomb counting algorithm is that the total capacity of the battery can only be updated if it is fully discharged immediately after a full charge. Portable device users rarely fully discharge the battery, so the actual capacity may be greatly reduced before the update is completed.

The second method uses the relationship between battery voltage and state of charge (SOC) to monitor battery capacity. This method seems intuitive, but the battery voltage is only highly correlated with the SOC or battery capacity when there is no load current applied to the battery. This is because if a load current is applied, there will be a voltage drop across the internal impedance of the battery. The battery impedance increases by 1.5 times for every 100°C drop in temperature. In addition, as the battery ages, significant impedance-related issues arise. A typical lithium-ion battery will double its DC impedance after completing 100 charge and discharge cycles. Finally, the battery has a very large time constant transient response to step-load changes. After the load is applied, the battery voltage will gradually decrease at different rates over time and gradually increase after the load is removed. After only 15.0% of the standard charge and discharge cycles (500) have been completed, a very effective voltage-based algorithm may cause up to 50% error for a new battery.

Based on Impedance TrackingTM

Battery fuel monitoring based on technology From the above results, it can be seen that it is impossible to achieve 1% battery capacity estimation regardless of whether it is the coulomb counting algorithm or the battery fuel monitoring based on the battery voltage correlation algorithm. Therefore, TI has developed a new battery fuel monitoring algorithm-Impedance TrackingTM technology, which combines the advantages of the coulomb counting algorithm and the voltage correlation algorithm. When the notebook computer system is in sleep or shutdown mode, its battery and its battery pack are in an idle state without load. At this time, there is a very accurate correlation between the battery open circuit voltage (OCV) and SOC. This correlation gives the exact starting position of SOC. Since all self-discharge activities are reflected in the process of battery OCV reduction, there is no need for self-discharge correction. Before the portable device is turned on, the accurate SOC usually depends on the measurement of the battery OCV. When the device is in active mode and the load is connected, the coulomb counting algorithm based on current integration is executed. The coulomb counter measures the amount of charge passed and integrates it to continuously calculate the SOC value.

Figure 1: Estimated maximum total capacity Qmax of a battery

Figure 1 shows the update of the total battery capacity measurement. The total battery capacity is calculated from two OCV readings at P1 and P2 when the battery voltage change before and after charge and discharge is small enough and the battery is in full idle state. Before the battery is completely discharged at P1, the SOC value can be obtained as follows:

When the battery is fully discharged and the pass charge is DQ, the SOC value can be obtained by the following formula:

Subtracting the two equations, we get:

in

In the formula, by measuring the OCV of the battery at P1 and P2 respectively, SOC1 and SOC2 can be obtained from the correlation between the battery OCV and SOC. From this equation, it can be seen that the total capacity of the battery can be determined without going through a full charge and discharge cycle. After connecting an external load, the impedance of each battery can be measured by measuring the battery voltage difference under load conditions. The voltage difference divided by the connected load current can give the low-frequency battery impedance.

Figure 2: Comparison of the remaining capacity predicted by the fuel gauge bq20z80 algorithm based on real-time updated battery impedance and the actual remaining capacity

In addition, when the measurement is performed using a model that describes the temperature effect, the magnitude of the impedance is related to the temperature. With this impedance information, we can predict the termination voltage, so that we can accurately calculate the remaining capacity at all loads or temperatures. With this battery impedance signal, we can determine the remaining capacity by using a voltage simulation method in the firmware. This simulation method first calculates the current SOCstart value, and then calculates the future battery voltage value when the load current is the same and the SOC value continues to decrease. When the simulated battery voltage is lower than the battery termination voltage (typical value is 3.0V/each cell), obtain the SOC value corresponding to this voltage and record it as SOCfinal. The remaining capacity RM can be obtained by the following formula:

.

Figure 2 illustrates how the bq20z80 accurately predicts the remaining charge of the battery. The error in the remaining charge prediction is less than 1.0%. This error rate persists throughout the life of the battery pack. Conclusion The Impedance Tracking™ based battery fuel gauge combines the advantages of coulomb counting based algorithms and voltage correlation based algorithms to achieve the best battery fuel monitoring accuracy. By measuring the OCV in the idle state, an accurate SOC value can be obtained. Since all self-discharge activity is reflected in the process of the battery's OCV reduction, no self-discharge correction is required. When the device is in active mode and a load is connected, the current integration based coulomb counting algorithm is executed. The battery impedance is updated through real-time measurement, and the impedance tracking technology also allows us to save the time-consuming battery automatic learn cycle. Therefore, 1% battery fuel monitoring accuracy is achieved throughout the battery life.