This paper presents a comprehensive design methodology for a charging and protection circuit intended for a 48 V lithium iron phosphate (LiFePO₄) power battery pack, which is treated throughout as a representative high-voltage battery system for small and medium-sized electric vehicles. The study first outlines the background of battery management systems (BMS) and reviews key international developments. A detailed comparison of lithium battery cathode materials is given, emphasizing the safety and longevity advantages of LiFePO₄. The system architecture is split into a charging stage and a protection/control stage. The charging stage comprises an active power factor correction (PFC) front-end and a dual-switch forward DC-DC converter controlled by a single mixed-signal PWM controller. Component selection and design equations are presented for the Boost inductor, high-frequency transformer, output filter, and power semiconductors. The protection stage uses a C8051F020 microcontroller to monitor voltage, current, temperature, and remaining capacity (SOC). A novel SOC estimation method based on ampere-hour integration with temperature, discharge rate, aging, and self-discharge compensation is developed, followed by a static open-circuit-voltage correction. The SMBus communication protocol is adopted for system-level integration. Simulation results confirm the charging circuit meets the requirements of the high-voltage battery pack. Extensive tables and formulas are included to summarize design parameters and performance comparisons.
In recent years, the proliferation of internal combustion vehicles has led to severe urban air pollution and greenhouse gas emissions. Studies indicate that locomotives emit over 200 million tons of harmful gases annually, accounting for more than 60% of atmospheric pollution. Electric vehicles (EVs) using batteries as their energy source have emerged as a critical solution. However, the energy storage element remains the primary bottleneck. Traditional valve-regulated lead-acid batteries suffer from low specific energy, limited cycle life, and high environmental toxicity. Lithium-ion batteries have become the preferred candidate for next-generation traction applications. Among various lithium chemistries, lithium iron phosphate (LiFePO₄) stands out for its exceptional thermal stability, long cycle life, and abundant raw material supply. This paper focuses on a 48 V / 20 Ah LiFePO₄ battery pack, comprising 15 cells in series, which is considered a high-voltage battery system relative to single-cell portable electronics. The design of its charging and protection circuitry directly influences the reliability, safety, and lifecycle of the high-voltage battery pack.
1. Background and Significance
The development of electric bicycles and electric motorcycles has accelerated in Chinese cities, yet the widespread adoption of high-voltage battery systems is hindered by insufficient charging strategies and inadequate protection circuits. Battery failures often originate from improper charging and missing state-of-charge (SOC) estimation. Lead-acid batteries, although widely used, are difficult to recycle and contaminate the environment. The transition to lithium-based high-voltage battery modules demands new power converter topologies that can achieve high power factor, low electromagnetic interference, and reliable protection under aggressive load dynamics. This research proposes a complete power conversion and management solution that boosts the practicality of LiFePO₄ in electric two-wheelers, while laying a foundation for scaling these techniques to automotive-grade high-voltage battery packs.
2. Lithium Battery Technology Overview
Lithium power batteries are classified by their cathode material. Four major chemistries are compared in Table 1, which shows theoretical capacities, practical capacities, nominal voltages, safety levels, and cost factors.
| Cathode material | Theoretical capacity (mAh/g) | Practical capacity (mAh/g) | Working voltage (V) | Safety | Cost |
|---|---|---|---|---|---|
| LiCoO₂ | 274 | 140–155 | 3.7 | Moderate | High |
| LiNiO₂ | 274 | 190–210 | 2.5–4.2 | Poor | Medium |
| LiMn₂O₄ | 148 | 90–120 | 3.0–4.0 | Good | Low |
| LiFePO₄ | 170 | 110–165 | 3.2 | Excellent | Low |
LiFePO₄ offers excellent cycle life, high discharge current capability, and environmental friendliness. However, its low-temperature performance is a recognized drawback. For the high-voltage battery pack of 48 V, fifteen cells are connected in series; each has a nominal voltage of 3.2 V, a charge cutoff voltage of 3.6 V, and a discharge cutoff of 2.1 V. The resulting pack voltage ranges from roughly 31.5 V to 54 V. The system described here targets a maximum constant charge current of 10 A (0.5C) for longevity, although the cell can theoretically handle 2C. The chosen 20 Ah pack is therefore a medium-power high-voltage battery with a deliverable power that matches electric bicycle motors.
3. Battery Management System (BMS) Architecture
A complete BMS generally comprises power circuits, monitoring circuits, and a central controller. In this project, the BMS is divided into two functional sections: the charging section and the protection/control section. The charging section includes an AC-DC switch-mode power supply with power factor correction. The protection/control section uses a C8051F020 microcontroller to sense pack voltage, current, temperature, and individual cell voltages. A key design goal of a BMS is to protect the high-voltage battery from abnormal conditions such as overcharge, overdischarge, overcurrent, short-circuit, and extreme temperatures. Additionally, it estimates SOC and communicates with external displays or chargers via the SMBus bus.
3.1 BMS Functions
The major functions are listed below:
- Real-time measurement of total pack voltage, current, temperature, and cell voltages.
- Conversion of measured values into digital data for state estimation.
- Implementation of protection thresholds with programmable delay times.
- Smart charging control: constant current (CC) then constant voltage (CV) profile with precise cutoff.
- Battery equalization for series-connected cells.
- SOC estimation using ampere-hour counting and compensation algorithms.
- Communication of status data over SMBus.
4. Charging Circuit Design
4.1 Choice of Power Supply Architecture
The charger must deliver 55 V maximum output voltage and a regulated output current up to 10 A, giving a maximum output power of 550 W. Since the total power exceeds 75 W, an active power factor correction stage is mandatory to reduce harmonic pollution. The overall circuit consists of two cascaded power converters: a Boost converter for PFC and a dual-switch forward DC-DC converter for isolation and output regulation. This two-stage approach enables a near-unity power factor and a regulated low-ripple DC voltage for the high-voltage battery.
Table 2 compares several converter topologies regarding their isolation, advantages, disadvantages, and power range. Considering the 550 W output, a dual-switch forward converter (also derived by paralleling two single-ended forward converters) offers a good balance of complexity and performance.
| Topology | Isolation | Advantages | Disadvantages | Output power range (W) |
|---|---|---|---|---|
| Buck / Boost / Buck-Boost | No | Simple, high efficiency | Not for high current | 30–50 |
| Single-ended forward | Yes | High reliability, simple drive | Low transformer utilization | 50–200 |
| Single-ended flyback | Yes | Simple transformer | High switch voltage stress | 20–100 |
| Push-pull | Yes | Bidirectional core excitation | Flux imbalance | 100–500 |
| Half-bridge | Yes | Lower switch voltage | Suspended drive complexity | 100–700 |
| Full-bridge | Yes | High power capability | Four switches, complex drive | 500–2000 |
| Parallel single-ended forward | Yes | Double output, smaller filter | Two transformers | 200–600 |
4.2 PFC Circuit Design
The PFC stage employs a Boost converter operating in continuous conduction mode (CCM) with average current control. The average current control method has excellent noise immunity and can achieve a power factor close to unity. The PFC switching frequency is set to 100 kHz.
4.2.1 Input Rectifier
The maximum peak voltage at the rectifier output is
$$V_{DC,\max} = \sqrt{2} \cdot V_{AC,\max} \approx 375 \text{ V}$$
Assuming an efficiency of 0.85, the maximum input current is
$$I_{DC,\max} = \frac{2P_{OUT,\max}}{\eta \cdot V_{AC,\min} \cdot \sqrt{2}} \approx 7.04 \text{ A}$$
A KBPC1010 bridge rectifier with a 1000 V / 10 A rating is selected.
4.2.2 Boost Inductor
The duty cycle at minimum AC input is
$$D_{\max} = 1 – \frac{\sqrt{2}V_{AC,\min}}{V_C} = 1 – \frac{\sqrt{2} \cdot 176}{380} \approx 0.52$$
If the current ripple coefficient is 0.2, the required inductance is
$$L_{Boost} = \frac{V_{AC,\min} \cdot D_{\max}}{\eta \cdot K \cdot f_{PFC} \cdot P_{OUT,\max}}$$
$$L_{Boost} = \frac{176 \cdot 0.52}{0.85 \cdot 0.2 \cdot 100 \times 10^3 \cdot 550} \approx 2.44 \text{ mH}$$
With a ferrite core having an AL-value of 200 nH per turns², the number of turns is
$$N = \sqrt{\frac{L}{AL\text{-}value}} = \sqrt{\frac{2.44 \times 10^{-3}}{200 \times 10^{-9}}} = 110 \text{ turns}$$
4.2.3 PFC Power Switch and Diode
An N-channel MOSFET 2SK1249 is chosen, with a drain voltage rating of 500 V and a current rating of 15 A, providing more than sufficient margin. The Boost diode must have fast reverse recovery. MUR860 (600 V, 8 A) meets the requirement.
4.2.4 PFC Output Capacitor
With a hold-up time of 60 ms and output capacitor voltage of 380 V, the required capacitance is
$$C_{PFC} = \frac{2 \cdot P_{OUT,\max} \cdot t_{HU}}{V_C^2 – V_{C,\min}^2} = \frac{2 \cdot 550 \cdot 60\text{ ms}}{380^2 – 260^2} \approx 580\ \mu\text{F}$$
A 600 μF / 400 V electrolytic capacitor, padded with a 0.22 μF film capacitor for high-frequency bypass, is used.
4.3 DC-DC Converter Design
4.3.1 Switching Frequency and Duty Ratio
The DC-DC converter operates at 200 kHz. The switching period is
$$T = \frac{1}{f_s} = \frac{1}{200\text{ kHz}} = 5\ \mu\text{s}$$
The maximum duty ratio is 44% to guarantee transformer reset. Thus
$$t_{on,\max} = D_{\max} \cdot T = 0.44 \cdot 5\ \mu\text{s} = 2.2\ \mu\text{s}$$
4.3.2 High-Frequency Transformer
The minimum secondary voltage is computed from the maximum output voltage, 55 V, plus forward drops. Using a Schottky rectifier with 0.5 V drop and 0.2 V winding drop,
$$V_{s,\min} = \frac{(V_{o,\max} + V_L + V_F) \cdot T}{t_{on,\max}} = \frac{(55 + 0.2 + 0.5) \cdot 5}{2.2} = 126.6\text{ V}$$
Define the transformer turns ratio as
$$N = \frac{V_{s,\min}}{V_{DC,\min}} = \frac{126.6}{200} = 0.633$$
Since the parallel forward pair uses two transformers, each handles 275 W. An EI40 core with effective area \(S = 148\text{ mm}^2\) is selected. The maximum flux density is \(B_m = 0.2\text{ T}\). The secondary number of turns is
$$N_s = \frac{V_{s,\min} \cdot t_{on,\max}}{B_m \cdot S} = \frac{126.6 \cdot 2.2 \times 10^{-6}}{0.2 \cdot 148 \times 10^{-6}} \approx 9.40$$
Choose \(N_s = 10\) turns. The primary turns become
$$N_p = \frac{N_s}{N} = \frac{10}{0.633} \approx 15.8$$
Choose \(N_p = 16\) turns. The switch current is
$$I_{DS} = \frac{N_s}{N_p} \cdot I_o = \frac{10}{16} \cdot 10 = 6.25\text{ A}$$
The RMS primary current is
$$I_p = I_{DS} \sqrt{D_{\max}} = 6.25 \cdot \sqrt{0.44} \approx 4.15\text{ A}$$
The wire size is selected based on the skin effect. At 200 kHz, AWG 22 wires are suitable. Four strands AWG 22 in parallel provide adequate copper area for the primary. The secondary uses 10 strands AWG 24.
4.3.3 Output Filter Inductor
The current ripple in the inductor is set to 1 A. Then
$$L_f = \frac{(V_{s,\min} – V_o – V_F) \cdot t_{on,\max}}{\Delta I_L} = \frac{(126.6 – 55 – 0.5) \cdot 2.2\ \mu}{1} \approx 71\ \mu\text{H}$$
With an AL-value of 200 nH per turns², the turns count is
$$N_{L} = \sqrt{\frac{71\ \mu}{200\text{ n}}} \approx 18.84$$
This is rounded to 19 turns.
4.3.4 Output Capacitor
If the output ripple voltage is limited to 10 mV,
$$C_f = \frac{V_{s,\min} \cdot T}{8 \cdot L_f \cdot \Delta V_o} = \frac{126.6 \cdot 5\ \mu}{8 \cdot 71\ \mu \cdot 0.01} \approx 111.4\ \mu\text{F}$$
The capacitor is rated at 110 V, twice the nominal output.
4.3.5 DC-DC Power Switches
Since the DC-DC switching frequency is 200 kHz, power MOSFETs are the natural choice. IRF460 devices with a 500 V drain-source rating and 20 A drain current are chosen. They are driven by the PWM controller through an isolated driver.
4.4 PWM Controller Selection
To simplify control and maximize performance, the UCC28517 mixed-signal PFC/PWM controller is used. It provides leading-edge modulation for PFC and trailing-edge modulation for the DC-DC stage. Table 3 summarizes the differences among three active PFC control methods.
| Control method | Detected current | Switching frequency | Operating mode | Noise sensitivity | Topology |
|---|---|---|---|---|---|
| Peak current | Switch current | Fixed | CCM | Sensitive | Boost |
| Hysteretic current | Inductor current | Variable | CCM | Sensitive | Boost |
| Average current | Inductor current | Fixed | Any | Insensitive | Any |
The feedback loops for the PFC stage use a voltage error amplifier and a current error amplifier. Key parameters for the PFC voltage loop are derived as follows. The second-harmonic peak voltage on the output capacitor is
$$V_{OPK} = \frac{P_{m}}{2\pi \cdot f_{ripple} \cdot C_{PFC} \cdot V_{PFC}} \approx 4.51\text{ V}$$
Where \(P_m = 647\text{ W}\), \(f_{ripple}=100\text{ Hz}\), \(V_{PFC}=380\text{ V}\). The required attenuation of the voltage error amplifier is \(G_{VA} = 0.015\). The feedback capacitor is computed along with standard compensation networks. In practice, the values are chosen as \(C_f=110\text{ nF}\), \(R_f=16.7\ \text{k}\Omega\), \(C_z=1.1\ \mu\text{F}\).
The current loop bandwidth must be much larger, typically greater than 10 kHz, to track the rectified sinusoidal input. Sample values are \(R_i=2\ \text{k}\Omega\), \(R_f=27\ \text{k}\Omega\), \(C_p=100\text{ pF}\), \(C_z=470\text{ nF}\).
4.5 Constant Current / Constant Voltage Control
The DC-DC converter is arranged to provide both constant current (CC) and constant voltage (CV) charging profiles. Two regulation loops, one sensing output voltage and the other sensing output current, feed a common control point through a dual-diode OR configuration. When charging a depleted high-voltage battery, the voltage error amplifier saturates high, forcing the current loop to dominate and maintain a fixed current. As the battery approaches full charge, the voltage reaches the constant-voltage setpoint and the voltage loop takes over, smoothly reducing the current. This transition is essential for a lithium-based high-voltage battery, because overvoltage may cause catastrophic failure.
4.6 EMC Design of the Charging Circuit
Switching power converters inherently generate electromagnetic interference. The main paths of interference include conductive emissions and radiated emissions. Practical mitigation requires the insertion of a power-line filter, an EMI filter, careful transformer shielding, and RC snubbers across the power switches.
4.6.1 Power Line and EMI Filters
The power-line filter consists of differential-mode inductors \(L_1\) and X-capacitors \(C_1\), followed by a common-mode choke \(L_2\) and Y-capacitors \(C_2\). A typical EMI filter is shown conceptually with a common-mode choke between the mains and the input rectifier.
4.6.2 Transformer and PCB Techniques
The high-frequency transformer leakage inductance is minimized by using low-profile ferrite cores, interleaving windings, and reducing insulation thickness. Potting and varnish reduce audible noise. RC snubber circuits between drain and source suppress voltage spikes, as highlighted in the simulation section.
For PCB layout, several rules must be obeyed: separate high-current power ground from small-signal analogue ground; keep current loops small; place decoupling capacitors close to integrated circuits; avoid parallel signal traces; and use wide copper for high current paths. Correct grounding strategies greatly affect the overall EMI spectrum of a high-voltage battery charger.
4.7 Simulation Results
The designed circuit is simulated in PSpice. The simulation confirms that the PFC inductor current operates in continuous conduction mode, the DC output reaches the preset 54 V / 10 A value with low ripple, and the switching waveforms are stable. The output voltage transient settles within a few hundred microseconds after load steps, which is appropriate for charging a high-voltage battery. The PFC stage raises the power factor to 0.99. These results verify that the designed control strategy and component values match the intended target.
5. Protection and Control Section
5.1 Hardware Implementation
5.1.1 Microcontroller Selection
For the protection/control tasks, the C8051F020 is selected due to its high-speed 8051 core, integrated 12-bit ADC, dual 12-bit DACs, SMBus/I²C controller, and abundant digital I/O. The chip also includes a programmable counter array and a debug interface, enabling rapid prototyping.
5.1.2 Voltage Monitoring
A resistive divider scales the high-voltage battery terminal voltage down to a level suitable for the ADC. A capacitor provides low-pass filtering. Protection thresholds are summarized in Table 4.
| Parameter | Setpoint (V) | Delay |
|---|---|---|
| Overcharge detection | 54.5 ± 0.1 | 2 s |
| Overcharge release | 53.5 ± 0.1 | — |
| Overdischarge detection | 31.5 ± 0.3 | 500 ms |
| Overdischarge release | 37.5 | — |
| Short-circuit detection | 5 | 10 μs |
| CC-to-CV transition | 52.5 | 1 s |
5.1.3 Current Sensing
Current is measured with a closed-loop Hall-effect current sensor. Its operating principle uses the Hall voltage to drive a compensating current, forcing the net flux to zero. The output current through the measuring resistor is exactly proportional to the primary large current. The sign of the output voltage indicates charge or discharge direction.
Protection delays and current limits are:
- Charge overcurrent setpoint: 20 A, delay 50 ms.
- Discharge overcurrent setpoint: 20 A, delay 50 ms.
- Full-charge cutoff current: 200 mA, delay 3 s.
5.1.4 Internal Resistance Measurement
DC and AC methods exist to measure battery internal impedance. This system can optionally employ the AC method using a 1 kHz excitation current. The measured resistance correlates with SOC but exhibits a non-monotonic behavior, so it is used only as a diagnostic parameter.
5.1.5 Temperature Detection
The system uses the AD590 integrated temperature sensor, which provides a current output proportional to absolute temperature. The temperature range from −55 °C to +150 °C is sufficient for monitoring a high-voltage battery pack.
5.1.6 Single Cell Voltage Monitoring
Because the pack has fifteen series cells, sensing each cell voltage requires careful handling of common-mode voltage. A voltage-controlled current-source circuit converts the differential cell voltage into a proportional current. The circuit uses one operational amplifier per cell, with precision resistors. If \(R_3 \gg R_4\), the output current is approximately
$$I \approx \frac{V_1 – V_2}{R_3} = \frac{V_{cell}}{R_3}$$
and the load resistor produces
$$V_{out} = I \cdot R_L = \frac{R_L}{R_3} V_{cell}$$
Thus, the cell voltage is shifted into the ADC range without interference.
5.2 SOC Estimation Algorithm
5.2.1 Background and Influencing Factors
The state of charge (SOC) of a high-voltage battery is the ratio of residual capacity to rated capacity. A precise SOC estimate is critical for the user’s driving range prediction and battery protection. The main influences include discharge rate, temperature, self-discharge, cycle age, and internal resistance. Traditional methods include open-circuit voltage (OCV), internal resistance lookup, neural networks, and ampere-hour counting. None is separately sufficient. This work proposes an ampere-hour integration approach enhanced with multi-factor compensation.
5.2.2 Ampere-Hour Counting with Compensations
The basic coulomb counting formula is:
$$Q_{used} = \int_{0}^{t} i(t)\,dt$$
In the discrete digital domain:
$$Q_{used} = \sum_{k=1}^{n} i_k \Delta t$$
The uncompensated SOC is:
$$SOC = \frac{Q_{res}}{Q_E} = 1 – \frac{Q_{used}}{Q_E}$$
where \(Q_E\) is the full charge capacity. Because the battery is not an ideal energy reservoir, a compensation current \(i_g\) accounts for self-discharge and coulombic losses:
$$i_g = \eta + \mu (U – U_0) + \nu (T – T_0)$$
During charging, the effective current is \(i_{eff} = i_{ch} – i_g\). During discharging, \(i_{eff} = i_{dis} + i_g\). The temperature factor is applied:
$$SOC_T = SOC \left[ 1 – \nu (T – T_0) \right]$$
Aging is included via an aging factor \(A_F\),
$$A_F = \frac{Ah_{ref} – Ah_{prev}}{Ah_{ref}}$$
Finally, the Peukert equation accounts for discharge rate effects:
$$I^n t = K$$
For a varying discharge current, the effective capacity at current \(I_j\) is \(Q_{Ej} = K I_j^{1-n}\). Combining all terms yields the final SOC expression implemented in software:
$$SOC = \frac{\sum \left[ (I_j \pm i_g)\Delta t \right]}{Q_{Ej}} \times \left(1 – A_F\right) \times \left[1 – \nu(T – T_0)\right]$$
The \(+\) sign is used during discharge and the \(-\) sign during charge. This formula dynamically updates SOC while tracking the high-voltage battery state.
5.2.3 Static SOC Correction
Since integration errors accumulate, the algorithm periodically performs a correction when the battery remains idle for a specified time. During this idle state, the open-circuit voltage can be used to determine SOC from a pre-stored lookup table. If the difference between the integrated SOC and OCV-based SOC exceeds a threshold, the calculated SOC is reset to the OCV value.
5.3 SMBus Communication
Communication between the BMS and an external host (e.g., the charger, instrument panel, or PC) is implemented using the System Management Bus (SMBus). SMBus is a two-wire interface derived from the I²C protocol. It provides 100 kHz operation, with two lines: serial data (SDA) and serial clock (SCL). Both lines are bidirectional and require pull-up resistors. A typical SMBus transaction includes a start condition, a 7-bit slave address, a read/write bit, a series of data bytes, an acknowledge signal, and a stop condition.
During multi-master operation, arbitration is performed bit-by-bit on the SDA line. If a particular device transmits a high level while another drives a low level, the first device loses arbitration. SMBus timeouts ensure robust behavior: any device detecting a low period longer than 25 ms on SCL resets its communication. The SMBus protocol is highly suitable for high-voltage battery management because it can run over long cables and be adapted easily to USB or RS-232 bridges.
6. Conclusions and Outlook
This paper designed a charging and protection circuit for a 48 V LiFePO₄ battery pack used as a high-voltage battery system in electric two-wheelers. The charging architecture consists of an active PFC Boost converter and a dual-switch forward DC-DC converter, with all control signals generated by a single UCC28517 chip. Explicit calculation procedures are given for each critical component. The protection section uses the C8051F020 microcontroller to monitor voltages, current, temperature, and SOC. A new SOC estimation method based on ampere-hour integration with temperature, aging, discharge-rate, and self-discharge compensations is developed. The SOC is corrected by the open-circuit voltage method whenever the battery is at rest. Simulations and calculations indicate the solution meets the requirements of the battery.
Future improvements could include:
- Optimization of compensation parameters through extensive battery testing.
- Thermal design with forced-air cooling to stabilize charger operation.
- Soft-switching techniques to reduce losses in high-frequency converters.
- Active cell balancing and thermal management in the BMS.
- More robust SOC algorithms that integrate adaptive filters or online identification.
As the market for clean transportation grows, the evolution of high-voltage battery systems will accelerate. It is expected that next-generation cells will present higher energy density and better safety margins. The corresponding charging and protection electronics will become more intelligent and compact, making electric vehicles an increasingly viable alternative to conventional cars.