A Fast Charging Multi-C Technique for Mobile Devices

Ramy Awwad1, Ripan Das2, Tawfic Arabi2, Hazem Hajj1

1Department of Electrical and Computer Engineering 1 American University of Beirut

Beirut, Lebanon 1{raa105,hh63}@aub.edu.lb

2Intel Cooperation Hillsboro, OR-USA

2 {Ripan.das, Tawfik.r.arabi}}@intel.com

Abstract— Users of mobile devices typically charge their devices over night. In these situations, it is acceptable to the users for the phone to take several hours to charge. But at times, such as when traveling, users want fast charging for their devices, and would like to have the battery maintain high charging capacity for an extended period without need to recharge. However, the battery’s charging capacity diminishes after many charging and discharging cycles. At that point, the mobile device starts requiring more frequent charging, which becomes very disruptive for travelers. As a result, fast charging and maintaining battery quality and capacity retention over extended number of charging-discharging cycles is a valuable user experience for mobile device users. Traditional constant current constant voltage(CC-CV) charging methods require higher density batteries to be charged with lower charging current leading to long charging time up to 4.5 hrs – 5hrs. In this paper, we propose an alternative faster approach to battery charging based on multi-rate decisions without compromising battery longevity. Our experiments show that the battery charge time can be increased by ~ 3x, leading to 1.2hrs time needed to charge. The results also show that the battery can be charged beyond 100% rated capacity to almost 102.5% and still maintaining lower skin temperature.

Keywords—Energy, Battery, Mobile, Device,(Multi-C, CC-CV.

I. INTRODUCTION Selection of batteries is a key design decision for mobile devices such as phones and tablets since they typically occupy about 71% of the total system design area and account for almost 9% of total system bill of material (BOM) cost. This is illustrated in Fig. 1 for the iPad3 tablet.

Figure 1: Ipad3 system design showing PCB and the battery

Users of mobile devices typically charge their devices overnight. In these situations, it is acceptable to the users for the phone to take several hours to charge. But at times, such as when traveling, users want fast charging for their devices, and would like to have the battery maintain high charging capacity for an extended period without need to recharge. However, the battery’s charging capacity diminishes after many charging and discharging cycles. At that point, the mobile device starts requiring more frequent charging, which becomes very disruptive for travelers. As a result, fast charging and maintaining battery quality and capacity retention over extended number of charging-discharging cycles is a valuable user experience for mobile device users. Traditional constant current constant voltage(CC-CV) charging methods require higher density batteries to be charged with lower charging current leading to long charging time up to 4.5 hrs – 5hrs. There is a need to devise new mechanism for faster battery charging without compromising battery longevity. Battery charge time is dependent on the type of battery used and its characteristics. For example, the higher the capacity (w-hr), the longer the charge rates. Battery density is a factor in battery dimension, and the possibility for heating with excessive charging. Table I shows the specifications for two known tablets, the iPad2 and the iPad3. It can be seen that the tablets use almost 2x higher density battery compared to notebook such as MacBook.

TABLE I. Battery Feature Differences in Tablets and Notebooks Product name Dimension Capacity

w-hr Density (w-

hr/L) IPAD2 4.22x2.48x0.09 25 453.6 IPAD3 4.92x2.55x0.16 44 435.9

Mac Book 13’’ 8.25x3.1x0.59 60.5 245

To support the small-factor devices, there is a clear trend towards using higher density batteries going forward for phones and tablets. These high density-batteries are traditionally changed with a special approach that requires different phases of constant current and constant voltage, called

978-1-4673-5328-1/12/$31.00©2012 IEEE

CC-CV methodologies. Fig. 2 illustrates the impact of different CC-CV charging approaches on charge time. Longevity can be assessed from battery specifications’ retention graph, which shows the total number of charging and discharging cycles of batteries without losing retention capacity.

Figure 2: Retention capacity over cycle counts and charging time expected using different charge-rates with traditional CC-CV methodology

With CC-CV battery health and retention are compromised with faster charging, As a result, to maintain retention capacity, the best option for longevity is to have extremely slow charging of the battery. . The general direction for new higher density batteries is to have the batteries charged with lower rates, resulting in very long charge time. As an example, iPad3 takes 8 hours of charge time. This direction will have negative reception with the users, especially business travelers who want to charge their tablet or mobile device faster, preferably within the 1 hour typically available between connecting flights. As a result there is a conflicting user requirement to maintain fast charge rate and long longevity This paper proposes an alternative and more efficient technique in comparison to CC-CV. The key idea is to use different charge rates at different energy capacity levels while charging the battery to maintain high charge speed but also maintain longevity by reducing stress on battery cells. Charge rate is also varied based on battery specifications. The rest of the paper is organized as follows. Section II discusses related work in battery charging. The Proposed Multi-C rate charging implementation Technique is described in section III Section V presents different experimental results measuring performance with the proposed methodlogy and also with traditional CC-CV method. Summary and conclusion are captured in section V.

II. RELATED WORK Battery charging techniques have received increasing attention from research communities. In [2], Chen provides a new methodology for Li-Ion batteries. His method is called Phase-Locked Battery Charger (PLBC), and is based on a Phase Locked Loop (PLL) feedback mechanism. This algorithm transforms the voltage difference across the battery and input adaptor to a frequency via Voltage Control Oscillator (VCO) called feedback phase. In addition, the charging current is controlled by monitoring the difference between the feedback phase and the input phase. The charging time remains long at 150 minutes even for a small battery (600mAh). In [3], the

concept of dynamic CC is applied. An optimization technique based on linear programming is used to assign the optimum charging rate for every State of Charge (SOC). Experimental result on a 1000mAh Li-Ion battery showed an improvement in the charging time by 18.25% comparing to 1C charging rate. In [4], a fuzzy–based five- steps battery charger is presented for Li-Ion batteries. The algorithm is based on adjusting the charging current based on the temperature T and temperature change ∆T. Experimental results showed that the method outperformed the (CC/CV) and the conventional five stages for thermal performance. It showed a decrease in the temperature around (0.8CO, 0.5CO) and (30%,-1.25%) for 1C and conventional five steps respectively. However, the technique did not use CC charging, which is required for lithium ion batteries. In [5], the paper provides an advanced algorithm called NeuFuz. Fast charging is associated with multiple trip points based approach to control voltage, temperature, rate of change for voltage and temperature and associated charge coulomb meter. The algorithm had the capability to deliver up to 5C burst charging for a battery of 700mAh capacity. Some other researchers focused their interest in applying a universal charger for different kinds of batteries (Li-Ion, and Ni-Cd) [6, 7]. The advantages of the methods explained in [6, 7] were to prevent transforming the energy into heat. The method also helped the battery cells to be relaxed and to accept more energy. However, the drawback of applying these techniques is that they require more time to complete the charging process since there is a rest period during the charging process. Hence, more time is needed regardless of the charger input source whether from a a wall adapter or USB. The methodology proposed in this paper will address the challenges of having a fast charge time, and maintaining longevity. The method can be also extended into a universal charger with embedded firmware or driver for control.

III. PROPOSED MULTI-C RATE CHARGING IMPLEMENTATION TECHNIQUE

This section describes the approach to providing for batteries: (a) reduced charging time, (b) longer lifetime, and (c) lower operating temperature. Some of these requirements impose conflicting constraints. For faster charging time, we want to have the highest charging rate possible since the higher the charging rate, the faster the charging time as illustrated in Fig. 3. For longer longevity, we want to reduce the charging rate since a lower charging rate produces higher retention and therefore longer discharge time as shown in Fi. 2. For lower temperature, we also want reduced charge rate, since higher charge rate can produce more heat. So the goal of the method is to find the right balance for the charge rate to provide fastest battery charge and best longevity.

Figure 3: Different charging rate for ATL battery

A. Proposed Method

In a traditional CC-CV mode, a battery charger can be programmed to charge with the profile shown in Fig.3, where the two phases are specified for the constant charge rate and the maximum CV voltage. There are five charging phases:

1. Pre-charging Phase: The first stage of charging is called pre-charging where battery needs to be charged within specified max current till voltage goes above minimum power levels, also called dead-battery voltage level.

2. Ramp to Low Voltage Phase: After the pre-charging stage, the traditional way to charge Li-Ion batteries is to apply a constant current phase (CC) followed by a constant voltage (CV). The next level reached is called the battery Low-Voltage level.

3. High CC Charge Phase: From the Low-Voltage level, the charger performs CC mode until the battery voltage first hits its maximum potential voltage.

4. High CC Voltage Stability phase: At that point, the voltage goes into a cycle of dropping and then picking up again as shown in green in Fig.5. This phase continues briefly until the voltage stabilizes at its highest level. In this phase, the charge rate is kept at its previous phase CC level.

5. Exponential CC Drop Phase: Then, the charger switches to CV mode, where charge rate drops exponentially until the current decreases to a predetermined lowest value; also called the termination current. This value is typically equal to 2% of the fast charging current of the battery as suggested in [1] as an indication for end of charging process.

Figure 4: Flow diagram for Multi-C rate charging methodology.

Figure 5: Charging time for Different charging profile for ATL battery

In this paper, we change the last two phases of the charging cycle, and introduce a variable CC charge mechanism during the voltage stability phase, and that extends longer than the traditional method, and reduces the exponential CC drop phase. The goal is to try to improve the longevity without compromising charge speed by having several sub-phases of dropping charge rates. For these two phases, we propose to balance the choice of charge rate to achieve best tradeoff of charging speed with longevity. As a result, we propose a new seven-phase approach, called the Multi-C battery charging method as shown in Fig.(4,5). The first two phases of the traditional CC-CV approach remain the same, however the last three phases are replaced by the following five phases:

3. High CC Charging Phase: Starting with the Low-Voltage level, we propose to use the highest charge rate possible at constant CC (typically a 1C level), until maximum voltage. We stop this phase when the power level reaches a high enough level that can allow us to keep fast charge time, but reduce charge rate for longevity. In our work, we tried different levels of power (6 w-hr, 6.5 w-hr, 7 w-hr, and 7.5 w-hr for a maximum 9 w-hr battery), and determined 7.5 w-hr to be a good choice. This was the 83% energy level of the battery.

4. First Drop from high CC Charge Phase: Then starting with this new level, the charge rate remains at CC, but the rate level is dropped to the next lower CC level,

typically at 0.7 CC level. This is continued until the charge reaches its next charge level, which was 8 w-hr in this paper.

5. Second Drop from high CC Charge Phase: At that point, step 5 is repeated again by dropping the CC charge rate further to the next level, and this phase continues until the next charge level is reached, which was 8.5 w-hr for this paper.

6. Third Drop from high CC Charge Phase Again from the new level, step 5 is repeated by dropping the CC charge rate further, until the maximum desired charging capacity is reached (at 9 w-hr).

7. Exponential CC Drop Phase From that point, the charger switches to CV mode, where charge rate drops exponentially until the current decreases to a predetermined lowest value; also called the termination current. This value is typically equal to 2% of the fast charging current of the battery. This last step is similar to the CC-CV charging model, however, it lasts a lot less as shown in Fig.5.

Figure 6: CC-CV method vs. Multi-C rate charging. Shows why Multi-c rate can accept more energy than 1C charging rate

The Multi-C rate approach provides the same fast charging as the high CC in CC-CV approach, but while preserving battery longevity as will be shown in the experiments section. The experiments will show that the charging methodology is both fast and safe. When the current profile of the Multi-C goes from charging at 1C to 0.7C, the voltage of the battery drops to a lower value, which enables the battery to accept more charge and it is illustrated in the green shaded area. The same thing applies for the other drop phases as shown in Fig.6.

B. Proposed Setup

To support the proposed Multi-C method, two key measurements need to be tracked, and these are the battery charge level, and the battery voltage. To collect and track these

measurements, we propose the setup shown in Fig. 7. This shows a block diagram implementation for a tablet system with external charger and fuel gauge. The battery charge level, measured through its energy level, is monitored from the fuel gauge. It is like an indicator of how much energy remaining in the battery if it is empty or full. Also, it communicates with the System On Chip (SOC) via I2C protocol. And the later will enable or disable the charger according to the data sent by the fuel gauge.

Figure 7: Proposed system setup for collecting required measurements in support of

the Multi-C method.

It is worth noting that the fuel gauge and temperature monitors shown in the setup of Fig.7 cancan be easily implemented with a driver or ASL/BIOS when the system is on.

In regards to the frequency of data collection, we propose to set a charger timer to decide when to collect data. The following methodology is suggested

• Step1: Every time charger timer expires, wake up the system. Timer can be set to expire every 60 seconds or more. So waking the system every 60 seconds when the battery is charging does not violate Operating System (e.g. windows) requirements and also it does not hurt battery life since system is charging. The platform monitoring unit (PMU) should not turn on the display if the user has not pressed the power up or home button key.

• Step2: After the system wakes up, an interrupt source is used to charge the sub-system. The so battery driver is also loaded, to check the fuel gauge and temperature.

• Step 3: The measurement are then used to implement the multi-rate charging profile.

Granularity of charging steps should be based on battery specifications and cost of driver /FW polling to check fuel gauge. The right tradeoff should be selected.

IV. EXPERIMENTS AND RESULTS To evaluate the proposed algorithm, we conducted several experiments to show the benefits of our approach. In the first experiment (sub-section B), we captured the charging time and compared it to the (CC-CV) method. The goal in the first experiment is to show that our methodology achieves charging time comparable to the CC-CV method with a high 1C. In the second experiment (sub-section C), we show that the proposed method can achieve higher capacity compared to 1C in the CC phase. We show the effect of the discharging cycles which represent the life time of the battery. The last experiment (sub-section D) captures the effect of the charging algorithm performed on the battery temperature. Before the experiments are presented, the setup is first described in sub-section A..

A. Experiment Setup In support of the setup suggested for tracking battery fuel gauge and energy several specific components were used. The charger was selected as and Arbin battery tester. The battery used was a Li-Ion. Experiments were performed on Samsung and ATL batteries with specifications shown in Table II. Fig.8 shows the overall experimental flow. Charging algorithm is written as a script in a batch file. The script includes charging rate, commands to check on different parameters such as voltage, and capacity levels. The safety parameters of the battery are also specified according to specifications. The charger executes each line in the batch file and all the charging parameters of the battery such as: voltages, current, charging capacity, charging energy, discharging energy, discharging capacity are saved and exported to an excel sheet for further analysis.

TABLE II. Battery Specification Used for Measurements Battery type Samsung ATL

Battery dimensions 1100x600x20 mm 92.0 X 46.5 X 3.8 mm

Nominal voltage 3.7V 3.7V Cut off voltage 3.00V 3.00V

Capacity 2.31Ah 2.88Ah Charging capacity

rate From 0.5C to 2C From 0.5C to 3C

Figure 8: Experimental methodology for testing charging algorithms

B. Evaluating Charge Speed In this experiment, charging times for different C rate were measured on the selected ATL battery. Fig.9 (a) shows the charging currents and energy profiles with the proposed Multi-C, with the 1C charge level, and with the 0.2C charge rate. As

the battery reaches the desired Vmax, the current is reduced exponentially so that charging voltage remains the same or lower than Vmax. Charging is terminated when charging current falls below 125mA. As obvious from Fig.9 (b) to achieve a charging level of 96% of the energy capacity, 0.2C charging takes about ~3.2hours, while the proposed Multi-C method takes ~only 2 minutes more time compared to the fixed 1C charging with CC-CV. Fig.10 shows charging time required to charge the battery to 80%, 90% and also 100% of the total capacity. It can be seen that Multi-C rate charging provides almost similar charge time as 1C, and ~3x shorter charging time compared to 0.2C.

Figure 9: (a) Charging current for different charging profile

Figure 9: (b) Charging energy for different charging profile

Table III compares 1C and 0.2C with CC-CV and

Multi-C rate charging. It can be seen that 1C rate charging pumps almost 26% of energy during CV stage i.e. after the battery reaches max voltage. In comparison either Multi-C-rate and 0.2C rate CC-CV charging pumps only 5% energy during CV at max voltage stage which explains better battery safety, temperature performance and also battery longevity with 0.2C CC-CV or multi C rate charging.

TABLE.III. Energy Gained During Constant Current or Constant Voltage Stage as Shown in Figure 9(b)

Figure 10: Charge time taken to charge to different energy levels (80%, 90% and 100% of total capacity) with different charging profiles. 0.2C, 0.7C and 1C were used with CC-CV and also was used multi-rate charging From charging time experiments above it is obvious that Multi-c rate charging overcomes the slow charging time associated with 0.2C charge rate CC-CV method and provides almost equal charge time advantage as the 1C charge rate CC-CV method.

C. Evaluating Longevity This experiment was designed to understand impact to battery discharge capacity as a result of charging and discharging over multiple numbers of cycles. Both 1C charging with traditional CC-CV methodology and Multi-C charging (1C ->0.7C->0.5C->0.2C) were applied on the same ATL battery to charge and discharge for 20 cycles.

Figure 11: (a) Discharge capacity with 1C charging rate shown for 20 cycles of

charging and discharging

Fig 11: (b) Discharge capacity with Multi-C rate charging rate shown for 20 cycles of

charging and discharging.

This experiment showed the effect of discharging the battery for 20 cycles. As shown in Fig.11 (b), the discharging curves for Multi-C rate almost overlap each other. This indicates very little degradation (~0.02%) in discharge capacity over time. On the other hand, for the 1C charging rate the discharge capacity of the 1st cycle and 20th cycle are different. This change indicates degradation in the energy level by 4% of the total discharging energy as a result of 20 cycles of charging and discharging. This indicates one of the key reasons why using 1C rate charging is avoided with CC-CV. On the other hand Multi-C rate charging provides similar benefit of fast charge time and also does not degrade the battery as much.

TABLE V. Multi-Rate Charging Vs CC-CV

Table V shows key performance difference between traditional charging and the methodology described in this paper. Different charge rates in the Multi-C rate should be picked based on battery characteristics As a summary, the multi-rate (1C, 0.7C, 0.5C, 0.2C) charging methodology based on energy level of the battery is able to reduce charging time from 4.5 hours to 1.5 hours compared to CC-CV 0.2C charging while avoiding battery longevity issues associated with charging the battery with 1C charging which also delivers close to 1.5 hours of charging time. This method is also capable of charging the battery to 102.5% energy and exceed battery life performance target of 1000 cycles required by Microsoft win8.

D. Evaluating Temperature In this experiment, we monitored the rise in the temperature for both charging rate (1C, Multi-C). The proposed method Multi-

Charged Energy for ATL Battery Charging

rate Energy (CC)

Energy (CV)

Total Energy

Capacity gain in the CC phase

1C 6.67 2.36 9 74% MC 8.61 0.38 9 96% 0.2 8.43 0.37 8.8 96%

Charging Method

Charge time (25 W.hr battery)

capacity lost after 20 cycles

Capacity level battery can be

charged to 1C (CC-CV) 1.45 hours 4% 98%

0.2C (CC-CV) 4.5 hours 0.02% 102% Proposed Multi-C

Rate charging 1.5 hours 0.02% 102.50%

C rate shows an average decrease in temperature by (1%) compared to 1C. The results are shown in Fig.12.

Figure 12: Temperature vs. charging time

V. SUMMARY Experimental results show while Multi-C rate delivers almost similar fast charge time compared to higher single C-rate CC-CV charging. It does so without harming battery longevity unlike single high C-rate CC-CV Another benefit for Multi-C rate charging is maintaining about a degree cooler temperature in comparison to single C rate charging while delivering similar charge time. While Multi-C rate charging was tested and verified to optimize tablet system battery cells, it can be extended to all products using Li-Ion battery cells. Depending on the battery type and fuel gauge capability, granularity can be increased to maximize benefits. . More research is needed to characterize exact nature of chemical and electrical reactions that take place during charge and discharge processes.

ACKNOWLEDGEMENTS Authors Appreciate help from extended team members such as Swapna Marupaka for helping with high density cell samples, Andy Keates for valuable test suggestions. Authors sincerely appreciate help from PCCG battery testing team members Madeleine Ong, Gang Ji for proving valuable training with the tools.

VI. REFERENCES

[1] Hussein, A.A.-H, and Batarseh, I, "A Review of Charging Algorithms for Nickel and Lithium Battery Chargers," IEEE Transactions on Vehicular Technology, vol.60, no.3, pp.830-838, March 2011.

[2] Liang-Rui Chen; , "PLL-Based Battery Charge Circuit Topology," IEEE Transactions on Industrial Electronics, vol.51, no.6, pp. 1344- 1346, Dec. 2004. [3] Lan-Ron Dung, and Jieh-Hwang Yen, "ILP-Based Algorithm for Lithium-ion Battery Charging Profile” ,IEEE International Symposium on ," Industrial Electronics (ISIE), July 2010. [4] Jia-Wei Huang, Yi-Hua Liu, Shun-Chung Wang, and Zong-Zhen Yang , "Fuzzy-Control-Based Five-Step Li-ion Battery Charger,". International Conference on Power Electronics and Drive Systems ( PEDS), 2009. [5] Dilip, S, and Stolitzka, D, "Advanced Algorithms and Hardware for Intelligent Batteries," Conference on Battery Applications and Advances, Proceedings of the Tenth Annual, Jan 1995. [6] Hussein, A.A.-H, Pepper, M.Harb, A, and Batarseh, I, "An Efficient Solar Charging Algorithm for Different Battery Chemistries," Conference on Vehicle Power and Propulsion, Sept. 2009. [7] Tsenter, B. and Schwartzmiller. F, "Universal Charge Algorithm for Telecommunication Batteries," Conference on Battery Applications and Advances, Twelfth Annual Jan 1997.