This paper presents my research on the charging and protection circuit for a lithium iron phosphate (LiFePO4) power battery pack used in electric vehicles and light electric vehicles. The study is motivated by the rapid development of electric mobility and the need for reliable, efficient, and safe battery management systems (BMS) for small and medium-sized power tools and two-wheelers. In this work, I first review the fundamentals of lithium power batteries and their management systems, and then elaborate on the hardware design of a two-stage switching power supply with active power factor correction (PFC) and a DC/DC converter. I also discuss the protection and control hardware based on the C8051F020 microcontroller, including voltage, current, temperature, and cell voltage sensing circuits. In the software part, I propose a practical state-of-charge (SOC) estimation method based on ampere-hour integration with compensation for temperature, discharge rate, aging, and self-discharge, combined with open-circuit voltage correction. The communication between the battery management system and external hosts is implemented using the SMBus protocol. The design is validated by PSPICE simulations, which demonstrate that the charging circuit meets the requirements for a 48V/20Ah LiFePO4 battery pack. This work provides a practical reference for the replacement of lead-acid batteries with lithium power batteries in electric vehicle battery applications.
1. Introduction
The global increase in the number of automobiles has aggravated the greenhouse effect and air pollution caused by exhaust emissions. According to a report, motor vehicles emit more than 200 million tons of harmful gases into the atmosphere each year, accounting for more than 60% of the total atmospheric pollution. In addition to air pollution, noise pollution from traffic is also harmful to human health. Moreover, petroleum is a non-renewable resource, and the energy shortage problem is becoming increasingly serious. In China, oil production was 160 million tons in 2000, while imports of crude oil and refined oil reached 73 million and 30 million tons, respectively, costing 25 billion US dollars. By 2010, demand was projected to reach 360 million tons, with imports covering 45% of the total. Therefore, many countries have invested heavily in developing battery-powered vehicles to reduce pollution and energy dependence. Electric motors are more efficient than internal combustion engines and produce less noise. The Chinese government has supported electric vehicle research as a major national project.
Currently, electric bicycles and electric motorcycles are popular in many cities, but the development of electric vehicles is still at a research stage. One of the major bottlenecks is the energy storage component — the battery. Conventional valve-regulated lead-acid batteries are inexpensive and simple to manufacture, but they are heavy, bulky, have low specific energy, and a short cycle life (about 80–120 cycles in practice). Furthermore, lead pollution contradicts the goal of environmental protection. Therefore, lithium-based power batteries have emerged as the leading candidates to replace lead-acid batteries in electric vehicle battery applications. Among them, lithium iron phosphate (LiFePO4) offers excellent safety, long cycle life, high discharge current capability, and low cost. However, the charging and protection circuits must be specifically designed according to the battery chemistry and characteristics. An inadequate charging strategy can lead to overcharge, overheating, and even fire. Therefore, dedicated charging and protection circuits are essential for the safe and efficient operation of electric vehicle battery packs.
In this paper, I present the design of a charging and protection circuit for a 48V/20Ah LiFePO4 battery pack consisting of 15 cells in series. The maximum allowable charging and discharging currents are 20A. The operating temperature range is 0–50°C. The designed system must implement charging, safety protection (overvoltage, overcurrent, overtemperature, etc.), battery status monitoring (voltage, current, temperature, capacity), and communication functions. I first describe the battery characteristics, then propose a complete hardware architecture, and finally discuss the software algorithm for SOC estimation and SMBus communication.
2. Lithium Power Battery
2.1 Overview of Lithium Power Batteries
A power battery is an energy storage device that can provide substantial power for a relatively long time. The most common power battery is the valve-regulated lead-acid type, but its drawbacks are significant. Lithium-ion batteries have been widely used in portable electronics, but they cannot be directly scaled up for power applications because conventional lithium-ion batteries are prone to thermal runaway when high discharge currents are drawn. Power batteries, however, are not simply larger versions of portable lithium-ion batteries. They require specialized manufacturing processes, materials, and safety features to deliver high power reliably.
The performance of a lithium battery depends largely on the materials used for the electrodes, electrolyte, separator, and positive electrode. The positive electrode material is particularly important because it accounts for about 40% of the total battery cost. Several candidates for cathode materials are listed in Table 1.
| Cathode material | Theoretical capacity (mAh/g) | Actual capacity (mAh/g) | Operating voltage (V) | Safety | Cost |
|---|---|---|---|---|---|
| LiCoO2 | 274 | 140–150 | 3.7 | Fair | High |
| LiNiO2 | 274 | 190–210 | 2.5–4.2 | Poor | Medium |
| LiMn2O4 | 148 | 90–120 | 3.0–4.0 | Good | Low |
| LiMnO2 | 286 | 200 | 3–4.5 | Good | Low |
| LiFePO4 | 170 | 110–165 | 3.2 | Very good | Low |
Among these, LiFePO4 exhibits the best safety performance and has reached about 95% of its theoretical capacity in recent experiments. It offers high thermal stability, long cycle life (over 500 cycles with more than 95% capacity retention), and a flat discharge voltage plateau. It is considered the most promising cathode material for electric vehicle battery applications. The main drawback of LiFePO4 is its poor low-temperature performance: the discharge capacity at 0°C is about 78% of that at 23°C, and at -20°C it drops to 65%.
2.2 Characteristics of LiFePO4 Power Battery
The internal structure of a LiFePO4 battery consists of a LiFePO4 positive electrode (with aluminum foil as the current collector), a polymer separator, a carbon (graphite) negative electrode (with copper foil), and an organic electrolyte, all hermetically sealed in a metal casing. The nominal voltage of a single cell is 3.2V, the charge termination voltage is 3.6V, and the discharge termination voltage is 2.1V. The specific energy is 60–75 W/kg, and the energy density is 220–240 W/L.
Experimental data show that LiFePO4 batteries have the following advantages: high-efficiency output (standard discharge 2–5C, continuous high-current discharge up to 10C, and pulse discharge up to 20C for 10s); good performance at high temperatures (structural integrity maintained even at 160°C); excellent safety (no combustion or explosion under internal or external abuse); extremely long cycle life (more than 500 cycles with over 95% capacity retention); and fast charging capability (2C current). Compared with lead-acid batteries, LiFePO4 batteries are smaller and lighter (two-thirds the volume and one-third the weight for the same capacity) and environmentally friendly.
The primary disadvantages include poor low-temperature performance and relatively lower energy density compared with LiCoO2-based cells. Nonetheless, LiFePO4 is currently the most viable option for electric vehicle battery systems, including electric bicycles, electric mopeds, and hybrid vehicles.
3. Battery Management System (BMS)
3.1 Introduction to BMS
A battery management system (BMS) is essential for monitoring the state of a battery pack, protecting it from improper use, estimating the state of charge, and managing the charging/discharging process. The BMS performs dynamic monitoring of voltage, current, temperature, and other parameters; calculates the remaining capacity; protects the battery against overcharge, overdischarge, overcurrent, short circuit, and overtemperature; and ensures that the battery operates within its safe operating area. In this work, the BMS is divided into two parts: a power circuit (charging part) and a control/protection circuit, as shown conceptually in the system block diagram.
3.2 System Structure
The power circuit consists of a switching power supply and a PWM integrated controller. The control/protection part is based on a microcontroller (MCU) and includes voltage sensing, current sensing, temperature sensing, capacity estimation, alarm/display, and communication interfaces. Two main approaches exist for intelligent battery control: dedicated integrated circuits (ICs) and microcontroller-based systems. Dedicated ICs are cost-effective for small single-cell or dual-cell applications but are not scalable to high-power battery packs with many series cells. Therefore, I chose a microcontroller-based design because of its flexibility and expandability.
3.3 Functions of the BMS
The complete BMS should implement the following functions:
- State monitoring and data output: measure battery voltage, temperature, current, and internal resistance; compute the SOC in real time.
- Cell equalization: compensate for differences between cells to improve pack performance and prolong life.
- Temperature control: use cooling or heating to keep the battery in a suitable temperature range.
- Remaining capacity estimation: accurate SOC calculation is essential for user convenience and battery health.
- Safety protection: overvoltage, undervoltage, overcurrent, short circuit, and overtemperature protection.
- Communication: exchange data with external devices.
4. Design of the Charging Part
4.1 Switching Power Supply Basics
The charging part must convert the AC mains input into a controlled DC voltage/current suitable for the battery. A switching power supply (SMPS) is preferred over a linear regulator because of its higher efficiency, smaller size, and lower weight. The SMPS consists of a power stage and a PWM controller. The main components include a rectifier, a power factor correction (PFC) circuit, a high-frequency transformer, an output rectifier and filter, and a switch-mode controller. A typical block diagram shows the feedback path from the output to the PWM comparator.
There are many switching power supply topologies: buck, boost, buck-boost, single-ended forward, single-ended flyback, push-pull, half-bridge, and full-bridge. The choice depends on the output power level. Table 2 summarizes the advantages, disadvantages, and power ranges of common topologies.
| Type | Isolation | Advantages | Disadvantages | Power range (W) |
|---|---|---|---|---|
| Buck | No | Simple, high efficiency | Not suitable for high current | ≤50 |
| Boost | No | Simple, high efficiency | Not suitable for high current | ≤30 |
| Buck-Boost | No | Simple, high efficiency | Not suitable for high current | ≤30 |
| Single-ended forward | Yes | Reliable, simple drive | Transformer utilization low | 50–200 |
| Single-ended flyback | Yes | Simple transformer, few parts | High switch voltage stress | 20–100 |
| Push-pull | Yes | Bidirectional transformer excitation | Current imbalance can saturate core | 100–500 |
| Half-bridge | Yes | Simple transformer, low switch voltage | Floating gate drive required | 100–700 |
| Full-bridge | Yes | Simple transformer, high power | Complex drive, four high-voltage switches | 500–2000 |
PWM control can be voltage-mode or current-mode. Voltage-mode control is simple but has slow dynamic response to input voltage changes. Current-mode control, including peak current control and average current control, offers faster response but may require slope compensation. For this design, I selected average current-mode control because it has good noise immunity and can be applied to any topology.
4.2 Selection of the Overall Charger Structure
I designed a two-stage charger: a boost PFC preregulator followed by a DC/DC converter. The required output voltage is up to 55V, and the output current is up to 10A (the battery capacity is 20Ah, and the charging current is set to 0.5C for longer life). Therefore, the maximum output power is about 550W. Since the power exceeds 75W, PFC is mandatory to reduce harmonic pollution and improve the power factor.
4.2.1 PFC Circuit Selection
For the PFC stage, I chose a boost converter operating in continuous conduction mode (CCM) with average current-mode control. This topology is the most common and provides near-unity power factor. The control scheme uses a voltage error amplifier, a multiplier, a current error amplifier, and a PWM modulator to force the inductor current to follow the rectified input voltage waveform. Figure 12 in the original design illustrates this scheme. The average current control method has very low total harmonic distortion (THD), is insensitive to noise, and is suitable for any topology.
4.2.2 DC/DC Converter Selection
For the DC/DC stage, I chose a parallel single-ended forward converter (also called a dual-transistor forward converter), which is an evolution of the single-ended forward topology. It consists of two forward converters connected in parallel at the output. With two switches driven simultaneously, it can deliver twice the output power. Alternatively, when driven alternately, the effective output ripple frequency is doubled, which reduces the output filter size. Compared with a full-bridge, this topology has simpler gate drive requirements and higher reliability.
4.2.3 PWM Controller IC Selection
I used the Texas Instruments UCC28517 as the combined PFC and PWM controller. The UCC28517 provides rising-edge PFC gating and falling-edge PWM gating, which reduces the ripple current in the PFC output capacitor. This mixed control scheme simplifies the overall design.
4.3 Design of the PFC Stage
I selected a PFC switching frequency of 100 kHz. The input voltage is the AC mains (typically 220V RMS). The maximum DC input voltage after rectification is approximately 375V. The maximum input DC current is calculated by considering the output power (550W) and an assumed efficiency of 0.85:
$$I_{\mathrm{DC\_MAX}} = \frac{\sqrt{2} P_{\mathrm{OUT\_MAX}}}{U_{\mathrm{AC\_MIN}}\,\eta}$$
Using the minimum AC input voltage (say 170V), I calculated $I_{\mathrm{DC\_MAX}} \approx 7.04$ A. The rectifier bridge was selected as KBPC1010, rated at 1000V and 10A.
The boost inductor was calculated using a current ripple factor $K=0.2$. The maximum duty cycle occurs at minimum AC input voltage:
$$D_{\mathrm{MAX}} = 1 – \frac{\sqrt{2}V_{\mathrm{AC\_MIN}}}{V_C}$$
With $V_C = 380$ V, $D_{\mathrm{MAX}} \approx 0.52$. The inductance is given by:
$$L = \frac{V_{\mathrm{AC\_MIN}}^2 D_{\mathrm{MAX}}}{\eta K P_{\mathrm{OUT\_MAX}} f_{\mathrm{PFC}}}$$
Substituting the values, I obtained $L \approx 2.44$ mH. Using a ferrite core with gap and the AL-value, I calculated the required number of turns as approximately 110.
For the PFC switch, I selected a power MOSFET 2SK1249 (500V, 15A). The PFC diode was chosen as MUR860 (600V, 8A) because of its fast reverse recovery, which is critical in CCM operation. The PFC output capacitor was computed based on a hold-up time of 60 ms:
$$C = \frac{2 P_{\mathrm{OUT\_MAX}} t_{\mathrm{HU}}}{V_C^2 – V_{C_{\min}}^2}$$
I obtained approximately 580 μF and selected an electrolytic capacitor of 600 μF/400V, paralleled with a 0.22 μF film capacitor for high-frequency decoupling.
4.4 Design of the DC/DC Stage
The DC/DC switching frequency was set to 200 kHz, giving a switching period of 5 μs. The maximum duty cycle must be less than 50% for a forward converter; I chose $D_{\mathrm{MAX}} = 44$%, so the maximum on-time is:
$$t_{\mathrm{on\_MAX}} = T D_{\mathrm{MAX}} = 5\,\mu s \times 0.44 = 2.2\,\mu s$$
The secondary winding voltage $U_s$ must be sufficient to produce the maximum output voltage (55V) at maximum duty cycle. Considering the output diode forward drop ($U_F=0.5$V) and the winding voltage drop ($U_L=0.2$V), I calculated:
$$U_{s_{\min}} = \frac{(U_{o_{\max}} + U_L + U_F) T}{t_{\mathrm{on\_MAX}}}$$
This gives $U_{s_{\min}} \approx 126.6$ V. The turns ratio between the secondary and the primary is determined by the minimum DC bus voltage (about 200V at the undervoltage lockout point):
$$N = \frac{U_{s_{\min}}}{U_{dc_{\min}}} \approx 0.633$$
Since each of the two forward converter transformers handles half of the total power (275W), I selected an EI(E)-40 core with an effective cross-sectional area of $S = 1.48$ cm² and maximum flux density $B_m = 0.2$ T. The number of secondary turns was calculated using:
$$N_s = \frac{U_{s_{\min}} t_{\mathrm{on\_MAX}}}{B_m S} \times 10^4$$
After substituting the values, I obtained $N_s = 9.40$ turns, rounded up to 10 turns. The primary turns are then:
$$N_p = \frac{N_s}{N} = \frac{10}{0.633} \approx 15.8$$
rounded up to 16 turns.
The primary and secondary winding currents were also estimated. The switch current is:
$$I_{DS} = \frac{N_s}{N_p} I_o = \frac{10}{16} \times 10 = 6.25\,\mathrm{A}$$
The primary RMS current is $I_p = I_{DS} \sqrt{D_{\max}} \approx 4.15$ A. The secondary RMS current is $I_s = I_p \frac{N_p}{N_s} \approx 6.64$ A. Because of the skin effect at high frequency, I chose multiple strands of AWG 22 wire for the primary (4 strands) and AWG 24 wire for the secondary (10 strands), meeting the current density requirements.
The output filter inductor was designed for a current ripple of 1A. With the maximum on-time and the condition $U_{s_{\min}}$, the inductance is:
$$L_o = \frac{(U_{s_{\min}} – U_o – U_F)}{I_{\mathrm{ripple}}} t_{\mathrm{on\_MAX}}$$
I calculated $L_o \approx 71\,\mu$H. Using a core with an AL-value of $200 \times 10^{-9}$ H/N², the number of turns is
$$N = \sqrt{\frac{L_o}{AL}} \approx 19$$
The output filter capacitance was calculated for a ripple voltage of 10 mV:
$$C_o = \frac{U_s T}{8 L_o \Delta U_o}$$
which gave approximately 111 μF; I selected a 120 μF/110V capacitor, paralleled with a small ceramic capacitor.
For the DC/DC power switches, I selected IRF460 MOSFETs (500V, 20A), which provide sufficient margin.
4.5 Feedback Control Design
4.5.1 PFC Voltage and Current Loops
The PFC voltage loop must be slow enough to attenuate the 100 Hz ripple from the rectified AC line, while the current loop must be fast to shape the input current. Using the design equations (as in the original thesis), I determined the voltage loop compensation components: $C_f \approx 110$ nF, $R_{f1} \approx 16.7$ kΩ, and $C_Z \approx 1.1$ μF. For the current loop, the measured inductor current is compared with the reference from the multiplier. The current error amplifier compensation resulted in $R_{cp} = 2$ kΩ, $R_{f} = 27$ kΩ, $C_{p} = 100$ pF, and $C_{Z} = 470$ nF. These values ensure adequate phase margin and bandwidth.
4.5.2 DC/DC Voltage and Current Loops
The DC/DC converter uses a current-mode control scheme with a voltage outer loop and an inner current loop. For battery charging, two regulation loops are required: constant current (CC) and constant voltage (CV). The reference values for both loops are provided by the microcontroller via a DAC. As shown in Figure 20 of the original design, the outputs of the voltage error amplifier and the current error amplifier are ORed through diodes to the PWM feedback pin. When the battery voltage is low, the current loop dominates and the supply operates in CC mode. When the battery voltage reaches the target, the voltage loop takes over and operates in CV mode. This seamless transition is essential for LiFePO4 charging.
4.6 EMC Design of the Switching Power Supply
Electromagnetic compatibility (EMC) is a critical aspect of switch-mode power supply design, especially for high-power chargers. I addressed EMI by considering the three basic elements: source, path, and receptor. For conducted EMI, a two-stage filter (power filter and EMI filter) was designed. The power filter includes a differential-mode choke and common-mode chokes, as shown in Figure 21 of the original. The EMI filter uses X and Y capacitors to attenuate differential and common-mode noise, respectively.
For the high-frequency transformer, I reduced leakage inductance by choosing an appropriate core shape, minimizing the number of primary turns, and using interleaved winding techniques. I also added a shield to prevent magnetic field coupling to nearby circuits. The power switches are equipped with RC snubbers to damp the voltage spikes caused by the transformer leakage inductance. Figure 23 in the original illustrates a typical RC buffer.
Printed circuit board (PCB) layout guidelines were followed: minimize loop areas, keep sensitive lines away from high-current paths, use thick traces for power and ground, place high-frequency decoupling capacitors close to the switches, and avoid parallel signal lines. These measures significantly reduce radiated and conducted EMI.
4.7 Simulation Results
I used PSPICE to simulate the switching power supply. The simulation included the PFC stage and the DC/DC stage. Figure 24 shows the PFC gate drive signal, Figure 25 shows the inductor current waveform, Figure 26 shows the inductor voltage, Figure 27 shows the PFC output voltage settling at about 380V, Figure 28 shows the DC/DC gate drive, and Figure 29 shows the DC/DC output voltage. The simulation results verified that the charger meets the requirements for the 48V LiFePO4 battery pack: the output voltage is stable, the output current ripple is within acceptable limits, and the power factor is high.
5. Design of the Protection (Control) Part
5.1 Hardware Design of the Protection Circuit
5.1.1 Microcontroller Selection
I selected the Silicon Laboratories C8051F020 as the core controller. This is a fully integrated mixed-signal system-on-chip. It contains a 12-bit multi-channel ADC with an internal temperature sensor, an 8-bit ADC, two 12-bit DACs, comparators, a voltage reference, a programmable counter array (PCA), five general-purpose 16-bit timers, 64 I/O pins, and multiple serial buses including I²C/SMBus, SPI, and UART. It has 64KB of flash memory, 4KB of RAM, and an internal oscillator, plus a JTAG debug interface. The C8051F020 is instruction-set compatible with the MCS-51 but runs at up to 25 MIPS. Its digital crossbar allows flexible mapping of peripherals to I/O pins.
5.1.2 Voltage Detection Circuit
The pack voltage is sensed through a resistive divider, filtered by a capacitor, and fed to the ADC. I set the protection thresholds as follows:
| Parameter | Value | Remark |
|---|---|---|
| Overcharge protection voltage | 54.5V ± 0.1V | 15 cells × 3.6V = 54V, margin 0.5V |
| Overcharge release voltage | 53.5V ± 0.1V | Below overcharge voltage |
| Overcharge protection delay | 2 s | To avoid false triggering |
| Overdischarge protection voltage | 31.5V ± 0.3V | Single cell cutoff 2.1V × 15 = 31.5V |
| Overdischarge release voltage | 37.5V | After charging, voltage rises |
| Overdischarge protection delay | 500 ms | To avoid transient triggering |
| Short circuit protection voltage | 5V | Much lower than normal voltage |
| Short circuit protection delay | 10 μs | Immediate cutoff |
| CC/CV transition voltage | 52.5V | Set below overcharge release |
5.1.3 Current Detection Circuit
For current sensing, I used a magnetic balance Hall current sensor instead of a shunt resistor or a current transformer. A Hall sensor offers galvanic isolation, wide bandwidth, good temperature stability, and high accuracy. The Hall voltage is compensated by a secondary winding current so that the net magnetic flux is zero. The output voltage across the measuring resistor is proportional to the sensed current. The direction of the current (charging or discharging) is determined by the polarity of the output voltage.
The protection parameters are:
- Charging overcurrent protection: 20A
- Discharging overcurrent protection: 20A
- Overcurrent protection delay: 50 ms
- Full-charge cutoff current: 200 mA with a delay of 3 s to avoid false cutoff.
In addition, a fuse is placed in series with the battery as a last-resort protection device.
5.1.4 Internal Resistance Detection Circuit
Internal resistance is a useful indicator of battery health and SOC. DC measurement (applying a large current for 2–3 s) is accurate but invasive. AC measurement (applying a 1 kHz, 50 mA signal) is non-invasive and faster but less accurate and sensitive to noise. Since the main focus of this design is protection and SOC estimation, I did not implement a dedicated internal resistance measurement circuit in hardware, but the concept is discussed as future work.
5.1.5 Temperature Detection Circuit
Temperature is measured using an integrated temperature sensor, e.g., AD590, which outputs a current proportional to absolute temperature. The AD590 operates over -55°C to 150°C with an accuracy of ±0.5°C. The sensor is placed in close contact with the battery pack, and its output is converted to a voltage and fed to the ADC.
5.1.6 Single Cell Voltage Detection Circuit
To monitor individual cells in the 15-cell series pack, I used a voltage-controlled current source circuit based on an operational amplifier. This circuit converts the differential cell voltage into a current signal, which is then passed through a resistor to produce a voltage compatible with the microcontroller ADC. The circuit uses a differential amplifier to reject common-mode voltages from the series-connected cells. Diodes protect the op-amp from reverse voltage transients. This method provides high accuracy and immunity to noise over long leads.
5.2 Software Design of the Protection (Control) Part
5.2.1 SOC Estimation
Accurate estimation of the remaining capacity (state of charge, SOC) is one of the most challenging tasks in battery management. I first analyzed the existing methods: open-circuit voltage (OCV) method, internal resistance method, neural networks, and ampere-hour (Ah) integration. OCV is only accurate when the battery is at rest and cannot be used dynamically. Internal resistance is not monotonic with SOC and is difficult to measure accurately. Neural networks require large training datasets and high computational effort. The ampere-hour method is simple and suitable for real-time monitoring, but it suffers from accumulating errors due to temperature variations, discharge rate, self-discharge, and aging. Therefore, I proposed an enhanced ampere-hour integration method with compensation terms, combined with OCV correction during rest periods.
The basic ampere-hour integration is:
$$Q_{\mathrm{use}} = \int_0^t i(t)\,dt$$
In discrete form:
$$Q_{\mathrm{use}} = \sum_{k=1}^{n} i_k \Delta t$$
Then the SOC is:
$$SOC = \frac{E_E – Q_{\mathrm{use}}}{Q_E}$$
where $E_E$ is the full charge capacity and $Q_E$ is the rated capacity.
To compensate for losses, I introduced a parasitic current $g_i$ that depends on the battery voltage and temperature:
$$g_i = \eta U + \mu (T – T_0)$$
where $\eta$ and $\mu$ are experimentally determined coefficients. During charging, the total effective current is $i_t = i – g_i$; during discharging, $i_t = i + g_i$.
Temperature compensation was applied as a linear factor:
$$SOC_T = SOC \times \left[1 – \nu (T – T_0)\right]$$
where $\nu$ is the temperature coefficient.
Aging is accounted for through an aging factor $A_F$:
$$A_F = \frac{Ah_{\mathrm{ref}} – Ah_{\mathrm{pre}}}{Ah_{\mathrm{ref}}}$$
The Peukert equation describes the effect of discharge current on the effective capacity:
$$I^n t = K$$
where $n$ and $K$ are constants obtained by discharging at two different rates:
$$n = \frac{\lg(t_2/t_1)}{\lg(I_1/I_2)}$$
and
$$K = I_1^n t_1$$
For a varying discharge current, I divided the current profile into small constant-current segments. The SOC in segment $j$ is computed using the effective capacity for that current:
$$SOC = 1 – \sum_j \frac{I_j \Delta t_j}{Q_{Ej}}$$
Combining all compensations, the final SOC expression is:
$$SOC = \left[1 – \sum_j \frac{I_j \left(1 + \frac{g_{ij}}{I_j}\right) \Delta t_j}{K I_j^{1-n}}\right] \times \left[1 – A_F\right] \times \left[1 – \nu (T – T_0)\right]$$
where $g_{ij}$ is the parasitic current at segment $j$.
To prevent error accumulation, I used a correction mechanism based on the open-circuit voltage when the battery has been at rest for a sufficiently long time. If the difference between the Ah-integration SOC and the OCV-derived SOC exceeds a threshold, the calculated SOC is reset to the OCV value. This combines the advantages of both methods and ensures long-term accuracy.
5.2.2 SMBus Communication
For external communication, I adopted the System Management Bus (SMBus), a two-wire serial bus (SDA and SCL) derived from I²C. The SMBus operates at 100 kbps and is widely supported by battery management ICs and microcontrollers. The protocol includes START and STOP conditions, a 7-bit slave address, a read/write bit, and ACK/NACK bits. Figure 34 of the original shows a typical data transfer. The SMBus supports arbitration between multiple masters in a non-destructive manner, as illustrated in Figure 35. Bus timeout is defined as any low level on SCL exceeding 25 ms, which triggers a communication reset.
The C8051F020 has a dedicated SMBus controller, which simplifies the implementation. The battery pack can communicate its SOC, voltage, current, temperature, and alarm status to a host charger or an electric vehicle controller. Because of the wide availability of bridge chips, SMBus can easily be converted to USB, RS232, or RS485 for PC communication.
6. Conclusion and Outlook
In this thesis, I designed a charging and protection circuit for a 48V/20Ah LiFePO4 power battery pack intended for use in electric bicycles, electric mopeds, and other light electric vehicles. The charging circuit consists of a boost PFC preregulator and a parallel single-ended forward DC/DC converter, controlled by a single UCC28517 controller. The design achieves high power factor, high efficiency, and low output ripple. PSPICE simulations confirmed that the output voltage and current meet the battery requirements.
The protection and control circuit is based on the C8051F020 microcontroller. It provides comprehensive monitoring of the battery pack, including pack voltage, current, temperature, and individual cell voltages. Protection functions include overvoltage, undervoltage, overcurrent, short circuit, and overtemperature. I proposed an enhanced SOC estimation algorithm using ampere-hour integration with temperature, discharge rate, aging, and self-discharge compensation, combined with open-circuit voltage correction during rest. The SMBus interface enables communication with external devices.
Future work may include:
- Optimization of the compensation coefficients through extensive experiments to improve SOC accuracy.
- Design of cooling and ventilation systems for the charger.
- Application of soft-switching techniques in the DC/DC converter to reduce switching losses and electromagnetic interference.
- Implementation of active thermal management to maintain the battery in its optimal operating temperature range.
- Development of intelligent cell equalization strategies to further extend the pack life.
With the continuous improvement of lithium power battery technology and the maturation of battery management systems, the electric vehicle battery industry will overcome the current bottlenecks, and we will see more reliable, safer, and more energy-dense electric vehicle battery systems in the near future.
This work was supported by my university and all the teachers and students who helped me during my research. I sincerely thank the reviewers for their valuable comments.