I designed and evaluated an energy management system for an electric vehicle battery pack based on series-connected lithium-ion cells. The purpose of my work was to improve the safety, reliability, usable capacity, and cycle life of an electric vehicle battery pack while also providing an accurate estimate of state of charge. I treated the electric vehicle battery pack as a complete electrochemical and electronic system rather than as a collection of individual cells. This perspective is important because a practical electric vehicle battery pack must simultaneously satisfy voltage, current, temperature, protection, balancing, communication, and display requirements. In my design, I combined a microcontroller-based measurement platform, a dedicated protection architecture, a charge-balancing strategy, and a compensated state-of-charge algorithm. I also verified the main ideas through charging and discharging experiments on a seven-cell lithium-ion electric vehicle battery pack.
Motivation and system boundary. The electric vehicle battery pack is one of the most critical subsystems in an electric vehicle because it determines driving range, acceleration capability, thermal behavior, and replacement cost. Compared with lead-acid, nickel-cadmium, and nickel-metal-hydride technologies, lithium-ion cells offer higher nominal voltage, higher specific energy, lower self-discharge, longer cycle life, and no memory effect. However, these advantages do not automatically transfer to a large electric vehicle battery pack. Cell-to-cell variation, overcharge sensitivity, overdischarge sensitivity, high-current stress, and thermal gradients can all reduce performance. Therefore, I focused on an energy management system that can monitor the electric vehicle battery pack in real time, protect it from abusive conditions, balance the cells during charging, and estimate the remaining energy with acceptable accuracy.
Comparative position of lithium-ion technology. I first compared the major candidate chemistries for an electric vehicle battery pack. Table 1 summarizes the qualitative and quantitative trade-offs that guided my technology selection. The comparison shows why lithium-ion cells are attractive for a modern electric vehicle battery pack, even though cost and safety management remain challenging.
| Battery chemistry | Nominal cell voltage | Specific energy | Cycle life | Memory effect | Environmental profile | Suitability for an electric vehicle battery pack |
|---|---|---|---|---|---|---|
| Lead-acid | About 2 V | Low | Moderate | Some | Contains lead | Low-cost but heavy and large |
| Nickel-cadmium | About 1.2 V | Moderate | High | Strong | Contains cadmium | Limited by toxicity and memory effect |
| Nickel-metal-hydride | About 1.2 V | Moderate to high | High | Low | Relatively benign | Used in early hybrid electric vehicle battery packs |
| Lithium-ion | About 3.6 V to 3.7 V | High | High | Negligible | No cadmium or lead | Strong candidate for a high-performance electric vehicle battery pack |
The higher cell voltage of lithium-ion cells means that fewer cells must be connected in series for a given electric vehicle battery pack voltage. This reduces the number of interconnections and the number of monitored nodes, but it does not eliminate the need for cell-level management. In fact, because lithium-ion cells are sensitive to overcharge and overdischarge, the management system becomes more important, not less important. I therefore treated protection and balancing as first-class design requirements for the electric vehicle battery pack.
Functions required in my energy management system. I organized the required functions of the electric vehicle battery pack management system into four groups: safety protection, real-time monitoring, charge equalization, and thermal supervision. Table 2 presents these functions and the corresponding design actions I took. The table also emphasizes that every high-level function depends on accurate measurement of voltage, current, and temperature.
| Function group | Purpose | Key measured variables | Design action in my system |
|---|---|---|---|
| Safety protection | Prevent overcharge, overdischarge, overcurrent, and short circuit | Cell voltage, pack current, temperature | Dedicated protection ICs with MOSFET disconnects |
| Real-time monitoring | Track the operating state of the electric vehicle battery pack | Pack voltage, cell voltage, current, temperature | Microcontroller-based sampling and display |
| Charge equalization | Reduce cell-to-cell imbalance during charging | Individual cell voltage | Energy-dissipative bypass balancing |
| State estimation | Estimate remaining capacity and state of charge | Current, voltage, temperature, history | Compensated ampere-hour plus open-circuit voltage method |
| Thermal supervision | Keep the electric vehicle battery pack within safe temperature limits | Surface or cell temperature | Digital temperature sensor and temperature display |
State of charge and remaining capacity. I defined the state of charge of the electric vehicle battery pack as the ratio of remaining capacity to available capacity under specified conditions. A simple definition is given by Equation (1):
$$SOC = \frac{Q_c}{C_I}$$
where \(Q_c\) is the remaining capacity and \(C_I\) is the capacity that the cell or electric vehicle battery pack can deliver at a specified constant current. An equivalent definition based on discharged capacity is given by Equation (2):
$$SOC = 1 – \frac{Q_d}{C_I}$$
In these definitions, \(SOC = 1\) corresponds to a fully charged electric vehicle battery pack, while \(SOC = 0\) corresponds to the discharge cutoff condition. However, this simple definition is not sufficient for a practical electric vehicle battery pack because the available capacity depends on discharge rate, temperature, aging, and recovery effects. I therefore treated state of charge as a dynamic variable that must be corrected using measured operating conditions and stored history.
Factors that influence remaining capacity. I identified several factors that directly affect the accuracy of any state-of-charge estimate for an electric vehicle battery pack. Table 3 summarizes these factors, their physical origins, and the compensation approach I adopted. The table shows that state-of-charge estimation cannot be reduced to a single voltage measurement or a single current integration.
| Influencing factor | Physical origin | Effect on an electric vehicle battery pack | Compensation in my design |
|---|---|---|---|
| Discharge current | Electrochemical reaction rate and internal resistance | Higher current reduces delivered capacity | Current-dependent efficiency factor |
| Temperature | Electrolyte conductivity, reaction kinetics, diffusion | Low temperature reduces usable capacity; high temperature accelerates aging | Temperature correction coefficient |
| Self-discharge | Parasitic side reactions and leakage | Capacity decreases during storage | Long-term correction and initial state reset |
| Cycle life | Active material loss, impedance growth, lithium inventory loss | Rated capacity decreases with aging | Cycle-life correction coefficient |
| Recovery effect | Diffusion and relaxation after load removal | Open-circuit voltage does not stabilize immediately | Stable or unstable initial-state selection |
Common state-of-charge estimation methods. I reviewed the main methods used for state-of-charge estimation and compared them for use in an electric vehicle battery pack. Table 4 summarizes the discharge test method, ampere-hour counting, open-circuit voltage, internal resistance, and Kalman filtering. Each method has strengths and weaknesses. For my electric vehicle battery pack, I needed a method that could work online, tolerate variable current, and use a small microcontroller.
| Method | Principle | Advantages | Limitations for an electric vehicle battery pack |
|---|---|---|---|
| Discharge test | Fully discharge at a controlled current and measure released charge | Direct and accurate in the laboratory | Offline, slow, and unsuitable for real-time operation |
| Ampere-hour counting | Integrate charge and discharge current over time | Simple, online, and independent of detailed electrochemical modeling | Needs a good initial value; current errors accumulate |
| Open-circuit voltage | Relate stabilized open-circuit voltage to state of charge | Useful for initialization and low-current conditions | Requires relaxation; sensitive to temperature and history |
| Internal resistance | Relate impedance or resistance to state of charge | Useful near the end of discharge | Weak sensitivity in the middle range; difficult online measurement |
| Kalman filtering | Use a model and noisy measurements for recursive estimation | Can handle dynamic current and provide error information | Higher computational load; model and noise parameters are critical |
My selected estimation model. I chose a hybrid method for the electric vehicle battery pack. The method uses open-circuit voltage to initialize the state of charge when the electric vehicle battery pack has relaxed to a stable condition. It uses ampere-hour counting to track charge transfer during operation. It also applies corrections for charge-discharge efficiency, temperature, cycle life, and the recovery state. The accumulated charge is calculated by Equation (3):
$$Q(t) = \int_0^t \eta i(\tau) d\tau$$
where \(i(\tau)\) is the instantaneous current, positive for charging and negative for discharging, and \(\eta\) is the charge-discharge efficiency factor. The overall state-of-charge estimate is given by Equation (4):
$$SOC = K_0 K_t K_c SOC_0 + (1 – K_0) SOC’_0 + \frac{Q(t)}{C_N}$$
In Equation (4), \(SOC_0\) is the state of charge obtained from the stabilized open-circuit voltage, \(SOC’_0\) is the last termination state of charge when the electric vehicle battery pack was previously used, \(C_N\) is the rated capacity, \(K_t\) is the temperature correction coefficient, \(K_c\) is the cycle-life correction coefficient, and \(K_0\) is a binary recovery indicator. I set \(K_0 = 1\) when the open-circuit voltage has stabilized and \(K_0 = 0\) when the electric vehicle battery pack is still recovering. This formulation allows the estimator to switch between a voltage-based initialization and a history-based initialization without losing continuity.
For digital implementation, I discretized the charge accumulation. At sample \(k\), the accumulated charge is given by Equation (5):
$$Q_k = Q_{k-1} + i_k \Delta t$$
and the updated state of charge is given by Equation (6):
$$SOC_k = SOC_{k-1} + \frac{\eta i_k \Delta t}{C_N}$$
The open-circuit voltage initialization is represented by Equation (7):
$$SOC_0 = f^{-1}(V_{OC})$$
where \(f^{-1}\) is the inverse of the experimentally measured relationship between open-circuit voltage and state of charge. I built this relationship for the electric vehicle battery pack through low-current discharge and relaxation tests. I also included an average filtering step in the measurement software to reduce random noise. For \(N\) samples, the filtered value is given by Equation (8):
$$\bar{x} = \frac{1}{N} \sum_{j=1}^{N} x_j$$
Hardware architecture. I based the hardware architecture on a PIC18F452 microcontroller because it provides an internal 10-bit analog-to-digital converter, sufficient program memory, multiple timers, and low-power operating modes. The electric vehicle battery pack management board measures pack voltage, cell voltage, current, and temperature; drives a digital display; scans function keys; and interfaces with protection and balancing circuits. Table 5 lists the main hardware modules and their roles.
| Hardware module | Main device or method | Measured or controlled quantity | Role in the electric vehicle battery pack |
|---|---|---|---|
| Microcontroller | PIC18F452 | Digital processing and control | Core of the energy management system |
| Power supply | LM7805, LM7812, ICL7662 | Regulated +5 V, +12 V, and -12 V rails | Supplies the control and analog circuits |
| Pack voltage sensing | Precision resistor divider | Total electric vehicle battery pack voltage | Provides scaled voltage for the ADC |
| Cell voltage sensing | Operational amplifier and P-channel MOSFET | Individual cell voltage | Supports balancing and protection decisions |
| Current sensing | Precision shunt and absolute-value amplifier | Charge and discharge current | Provides charge transfer data for SOC estimation |
| Temperature sensing | DS18B20 digital sensor | Cell or pack temperature | Supports thermal supervision and compensation |
| Display | MAX7219 and LED indicators | Voltage, current, temperature, SOC, and capacity bars | Human-readable status of the electric vehicle battery pack |
| Keypad | Independent push buttons | User commands | Selects displayed variables |
Voltage measurement. For the total voltage of the electric vehicle battery pack, I used a precision resistor divider. The ADC input voltage is given by Equation (9):
$$V_{ADC} = V_{pack} \frac{R_2}{R_1 + R_2}$$
Because the ADC reference is the regulated 5 V supply, I selected the divider ratio so that the maximum electric vehicle battery pack voltage remains within the ADC input range. For individual cell voltage measurement, I used an operational amplifier and a P-channel MOSFET to translate the high-side cell voltage into a ground-referenced signal. This approach avoids the common-mode problem that appears when several cells are connected in series. It also provides a scalable measurement path for a larger electric vehicle battery pack.
Current measurement. I used a low-value precision shunt resistor in the main current path. The shunt voltage is given by Equation (10):
$$V_s = I R_s$$
Because the shunt voltage may be positive during discharge and negative during charge, I designed an absolute-value amplifier stage. The output is given approximately by Equation (11):
$$V_0 = 5 |V_I|$$
where \(V_I\) is the conditioned shunt voltage. This stage converts both charge and discharge currents into a positive ADC input while preserving magnitude information. The sign of the current is inferred in software from the operating state and the raw polarity before the absolute-value stage. Accurate current measurement is essential because the ampere-hour component of state-of-charge estimation for the electric vehicle battery pack depends directly on it.
Temperature measurement. I used a DS18B20 digital temperature sensor. It communicates over a one-wire bus and provides a direct digital output, which reduces analog noise sensitivity. The sensor supports a measurement range from \(-55^\circ C\) to \(+125^\circ C\), with resolution down to \(0.0625^\circ C\). I placed the sensor close to the electric vehicle battery pack surface to capture the thermal condition that most strongly affects capacity and aging. Temperature is not only displayed; it is also used in the state-of-charge compensation and in the protection logic.
Display and user interface. I used a MAX7219 serial display driver to control a multi-digit LED display. The MAX7219 reduces the number of microcontroller pins required and provides brightness control, scan limiting, and decoding modes. I also implemented a four-level LED capacity indicator for rapid visual assessment of the electric vehicle battery pack. The user can press keys to view pack voltage, current, temperature, and state of charge. I added a short display hold time so that the reading remains visible after the key is released. The display logic is intentionally simple, because the electric vehicle battery pack may be used in a noisy electrical environment.
Protection design. I regarded protection as mandatory for an electric vehicle battery pack. Overcharge can cause electrolyte decomposition, gas generation, and thermal runaway. Overdischarge can cause irreversible capacity loss. Overcurrent and short circuit can damage both the electric vehicle battery pack and the connected load. I designed the protection system around single-cell protection ICs and power MOSFETs. Each cell is monitored locally, and the protection outputs are combined through isolation and logic. Table 6 lists the protection thresholds and timing values I used in the design.
| Protection parameter | Value or condition | Purpose in the electric vehicle battery pack |
|---|---|---|
| Overcharge protection voltage | About \(4.28 \pm 0.025\) V per cell | Prevents lithium-ion cell overcharge |
| Overcharge delay | Less than about 1.3 s | Avoids false tripping from noise |
| Overdischarge protection voltage | About \(2.3 \pm 0.050\) V per cell | Prevents deep discharge |
| Overdischarge delay | Less than about 150 ms | Provides fast cutoff while ignoring transients |
| Charge voltage limit | Below about 17 V for the tested four-cell subsection | Protects the charging interface |
| Continuous charge current | Below about 2 A | Limits thermal stress |
| Continuous discharge current | Below about 2 A | Limits thermal and impedance stress |
| Overcurrent protection | About \(5 A \pm 10\%\) | Protects against excessive load current |
| Overcurrent delay | Less than about 10 ms | Allows short transients but trips on faults |
| Short-circuit delay | Less than about 3 ms | Provides very fast disconnect |
| Quiescent current | Below about \(200 \mu A\) | Reduces parasitic drain on the electric vehicle battery pack |
The protection architecture separates the charge path and the discharge path. This separation allows the electric vehicle battery pack to continue discharging through the body diode of the charge-control MOSFET when charging is disabled, and it allows charging through the body diode of the discharge-control MOSFET when discharging is disabled, provided the conditions are safe. I also included a lower-limit latch so that the discharge MOSFET does not repeatedly turn on and off near the overdischarge threshold. This latch improves reliability and reduces switching stress in the electric vehicle battery pack.
Balancing strategy. I compared energy-dissipative and non-dissipative balancing methods for the electric vehicle battery pack. Table 7 summarizes the trade-offs. Non-dissipative methods can transfer energy between cells and can be efficient, but they require more components, more complex control, and careful magnetics or capacitor networks. Energy-dissipative methods are simpler and lower in cost. For the electric vehicle battery pack in my study, I selected a charging-oriented dissipative balancing method because the main goal was to prevent overcharge of the weakest or earliest-full cell during series charging.
| Balancing category | Example | Energy flow | Advantages | Limitations for an electric vehicle battery pack |
|---|---|---|---|---|
| Dissipative | Resistor bypass with protection IC | Excess energy is converted to heat | Simple, low cost, easy to extend | Heat generation and lower efficiency |
| Capacitive | Switched capacitor network | Charge is shuttled between adjacent cells | Low loss and bidirectional | Slow for distant cells and many cells |
| Inductive | Buck-boost or transformer modules | Energy is transferred through magnetics | Fast and efficient | Complex, costly, and bulky for many cells |
| Converter-based | Parallel DC-DC modules | Energy is converted between cells or pack | Fast and flexible | High component count and control complexity |
In my charge-balancing circuit, the balance current through the bypass resistor is given by Equation (12):
$$I_{bal} = \frac{V_{cell}}{R_{bal}}$$
and the power dissipated in the resistor is given by Equation (13):
$$P_{bal} = I_{bal}^2 R_{bal}$$
With a balance threshold near 4.275 V and a bypass resistor of 68 \(\Omega\), the maximum balance current is approximately 64 mA, and the maximum resistor dissipation is approximately 0.28 W. I selected this level because it provides useful balancing without creating excessive heat inside the electric vehicle battery pack enclosure. The balancing module for each cell is nearly identical, so the design can be extended to a larger electric vehicle battery pack with relatively little change to the control logic.
Software architecture. I implemented the software in a modular manner so that each function could be tested independently. Table 8 lists the main software modules. The main program initializes the microcontroller, the timers, the ADC, and the display. It then enters a loop that manages data acquisition, state-of-charge estimation, key scanning, and display updating. Timer interrupts are used for periodic sampling and for timekeeping.
| Software module | Main task | Key inputs | Key outputs |
|---|---|---|---|
| Initialization | Configure I/O, timers, ADC, and display driver | Reset state | Ready system |
| Voltage acquisition | Sample pack and cell voltage channels | ADC channels | Filtered voltage values |
| Current acquisition | Sample shunt-conditioned signal | ADC channel | Charge and discharge current |
| Temperature acquisition | Read DS18B20 one-wire data | One-wire bus | Temperature in degrees Celsius |
| SOC calculation | Combine OCV initialization and ampere-hour integration | Voltage, current, temperature, history | State-of-charge estimate |
| Key processing | Debounce and interpret key presses | Push-button inputs | Display selection |
| Display | Write data to MAX7219 and LED indicators | SOC, voltage, current, temperature | User-visible status |
Data acquisition and filtering. For the ADC, I used a polling method with a 10-bit result. I sampled each analog channel multiple times and averaged the results using Equation (8). This reduced the influence of ripple on the 5 V supply and other random noise sources. The current channel is especially important because any offset or gain error directly biases the ampere-hour integration. I therefore calibrated the current channel with known load currents before using the state-of-charge estimator on the electric vehicle battery pack.
Temperature acquisition. The DS18B20 follows a strict one-wire protocol. I implemented reset, presence detection, ROM command, function command, and read/write time slots. The conversion time depends on the selected resolution. I used a sufficient delay before reading the temperature result. The measured temperature is used for display, for protection, and for the temperature correction coefficient in Equation (4).
State-of-charge software flow. The state-of-charge routine first determines whether the electric vehicle battery pack has been at rest long enough for the open-circuit voltage to stabilize. If it has, the routine obtains the initial state of charge from the stored open-circuit voltage table. If it has not, the routine uses the last termination state of charge. During operation, the routine integrates the measured current with the selected efficiency factor. It then applies the temperature and cycle-life corrections. Finally, it limits the result to the valid range between zero and one. The output is written to the display and can be used by higher-level energy management functions.
Key scanning and display. I used independent push buttons with a 10 ms software debounce. When a key is pressed, the display switches to the corresponding variable. When the key is released, the display returns to the default state-of-charge screen after a short delay. The MAX7219 receives 16-bit packets containing register address and data. I used the built-in BCD decoding mode for numeric display and set the scan limit to match the number of digits. I also used a separate LED bar for a quick indication of the electric vehicle battery pack capacity.
Experimental electric vehicle battery pack. I tested the system on a seven-cell series lithium-ion electric vehicle battery pack. Table 9 lists the main parameters of the tested pack. The nominal voltage was 25.9 V, the maximum charge voltage was 29.4 V, and the discharge cutoff was 21 V. The rated capacity was 2 Ah. These values are representative of a small electric vehicle battery pack or a scaled test platform for larger electric vehicle battery pack concepts.
| Parameter | Value |
|---|---|
| Cell type | Lithium-ion |
| Number of series cells | 7 |
| Rated capacity | 2 Ah |
| Nominal voltage | 25.9 V |
| Maximum charge voltage | 29.4 V |
| Discharge cutoff voltage | 21 V |
| Charging method | Constant current then constant voltage |
| Maximum continuous charge current | 2 A |
| Maximum continuous discharge current | 2 A |
| Operating temperature range for charging | \(0^\circ C\) to \(45^\circ C\) |
| Operating temperature range for discharging | \(-20^\circ C\) to \(45^\circ C\) |
Open-circuit voltage and state-of-charge relationship. I first measured the relationship between open-circuit voltage and state of charge for the electric vehicle battery pack. I discharged the pack at a low current of 0.05 C, which corresponds to about 100 mA. At selected intervals, I stopped discharging and allowed the pack to relax until the open-circuit voltage became stable. I then recorded the remaining state of charge based on the integrated discharge charge. Table 10 gives the representative points I used to build the lookup relationship. The relationship is nonlinear, especially near the end of discharge, which is why a simple linear voltage-to-SOC mapping would be inaccurate for the electric vehicle battery pack.
| Open-circuit voltage per cell | Estimated state of charge | Region |
|---|---|---|
| 4.20 V | 1.00 | Fully charged |
| 4.05 V | 0.90 | Upper plateau |
| 3.90 V | 0.75 | Main operating region |
| 3.75 V | 0.60 | Main operating region |
| 3.65 V | 0.45 | Main operating region |
| 3.55 V | 0.30 | Lower operating region |
| 3.45 V | 0.15 | Lower operating region |
| 3.35 V | 0.08 | Near discharge cutoff |
| 3.20 V | 0.03 | Deep discharge region |
| 3.00 V | 0.00 | Discharge cutoff |
Charging experiment. I charged the electric vehicle battery pack using a constant-current then constant-voltage profile. The constant-current phase was set to 0.3 C, about 600 mA. When the pack voltage reached 29.4 V, the charger switched to constant-voltage mode. Charging stopped when the current fell below 50 mA. The initial state of charge was about 8%, based on the minimum cell voltage of 3.35 V. Table 11 summarizes the charging stages and the observed behavior. The charging process lasted approximately four hours. The voltage and state of charge rose with a relatively stable slope during the constant-current phase. During the constant-voltage phase, the current decayed and the state-of-charge curve flattened, which is favorable for full charging without excessive stress on the electric vehicle battery pack.
| Charging stage | Control condition | Observed behavior | Implication for the electric vehicle battery pack |
|---|---|---|---|
| Constant current | About 0.3 C until 29.4 V | Voltage and SOC rise steadily | Efficient bulk charge |
| Constant voltage | Hold 29.4 V until current falls below 50 mA | Current decays; SOC approaches full | Reduces overcharge risk |
| Termination | Current threshold reached | Charging stops | Protects cycle life |
Constant-current discharge experiments. I discharged the electric vehicle battery pack at 0.2 C, 0.4 C, and 0.5 C. Table 12 lists the measured discharge times and capacities. The discharge time decreased as the current increased. The delivered capacity also decreased with higher current. This is a well-known behavior of lithium-ion cells and is one reason that the state-of-charge estimator must account for discharge rate. If I had used a simple voltage threshold without rate compensation, I would have overestimated the remaining energy at high current for the electric vehicle battery pack.
| Discharge rate | Approximate current | Discharge time to cutoff | Delivered capacity | Observation |
|---|---|---|---|---|
| 0.2 C | 400 mA | About 3.6 h | About 1420 mAh | Highest delivered capacity |
| 0.4 C | 800 mA | About 2.2 h | About 1380 mAh | Moderate reduction |
| 0.5 C | 1000 mA | About 1.3 h | About 1229 mAh | Lower delivered capacity |
Variable-load discharge experiment. I also discharged the electric vehicle battery pack into a 40 \(\Omega\) power resistor. The current remained near 0.2 A for most of the discharge, so the behavior was close to constant-current discharge. The state-of-charge curve decreased almost linearly. This experiment confirmed that my estimator behaves well when the load is stable and the current does not vary rapidly. In a real electric vehicle battery pack, however, the load current can change quickly during acceleration and regenerative braking. The ampere-hour integration with efficiency compensation is therefore necessary, but the current measurement bandwidth and offset stability remain important.
Error analysis. I evaluated the state-of-charge error using Equation (14):
$$e_{SOC} = SOC_{est} – SOC_{ref}$$
where \(SOC_{est}\) is the estimated value and \(SOC_{ref}\) is the reference value obtained from the controlled discharge data. The main error sources were current measurement offset, voltage measurement quantization, temperature variation, cell imbalance, and the open-circuit voltage relaxation time. Table 13 summarizes these error sources and the mitigation methods I used. The largest practical error source was the accumulation of current measurement error over time. This is unavoidable in ampere-hour counting, so I reset the initial state whenever the electric vehicle battery pack had sufficiently relaxed.
| Error source | Effect on state-of-charge estimation | Mitigation |
|---|---|---|
| Current offset | Slow drift in integrated charge | Calibration and zero-current correction |
| ADC quantization | Small random error | Averaging multiple samples |
| Temperature change | Capacity and efficiency variation | Temperature correction coefficient |
| Cell imbalance | Pack behavior differs from average cell | Cell monitoring and charge balancing |
| OCV relaxation | Incorrect initial state of charge | Stable or history-based initialization |
| Cycle aging | Rated capacity decreases | Cycle-life correction coefficient |
Design observations. Several practical lessons emerged from my work on the electric vehicle battery pack. First, the protection circuit must be reliable even when the microcontroller is not operating. I therefore used dedicated protection ICs rather than relying only on software. Second, the charge-balancing method should match the size and cost target of the electric vehicle battery pack. Dissipative balancing is simple and effective for charge equalization, but it generates heat and should not be used as a substitute for good cell matching. Third, state-of-charge estimation is not a single algorithm; it is a measurement and correction process. The open-circuit voltage method gives a good anchor, and ampere-hour counting gives dynamic tracking, but both require calibration and compensation. Fourth, the display and keypad are not merely conveniences. They allow the user to observe the electric vehicle battery pack behavior and detect abnormal trends before a fault occurs.
Scalability and future electric vehicle battery pack requirements. The architecture I developed can be scaled from a small electric vehicle battery pack to a larger one by repeating the cell monitoring and balancing modules. The main limitations are the isolation requirements of the high-voltage measurement network and the computational load of the state estimator. For a larger electric vehicle battery pack, I would add a communication bus, distributed measurement nodes, and a more powerful state estimator. I would also add cell-level thermal sensors and a more advanced thermal management strategy. The protection thresholds would remain cell-specific, while the pack-level controller would coordinate charge and discharge limits based on the weakest cell. This hierarchical structure is well suited to a high-voltage electric vehicle battery pack.
Comparison with alternative designs. In my assessment, the most important difference between my design and a simple battery monitor is that my system treats the electric vehicle battery pack as a managed energy system. A simple monitor may display voltage and current, but it does not necessarily prevent overcharge, does not balance cells, and does not compensate state of charge for temperature and aging. My design includes all of these functions. The cost is increased hardware and software complexity, but the benefit is improved safety and longer service life for the electric vehicle battery pack. Table 14 compares the functional coverage of a basic monitor and my energy management system.
| Function | Basic monitor | My energy management system |
|---|---|---|
| Pack voltage display | Yes | Yes |
| Current display | Sometimes | Yes |
| Temperature display | Sometimes | Yes |
| Cell voltage monitoring | No or limited | Yes |
| Overcharge protection | No | Yes |
| Overdischarge protection | No | Yes |
| Overcurrent and short-circuit protection | No | Yes |
| Charge balancing | No | Yes |
| Compensated SOC estimation | No or simple voltage lookup | Yes |
| User-visible status | Limited | Voltage, current, temperature, SOC, capacity bars |
Limitations. My design also has limitations. The dissipative balancing method is not ideal for a very large electric vehicle battery pack because it wastes energy and produces heat. The state-of-charge estimator uses a simplified temperature and cycle-life correction rather than a full electrochemical model. The current measurement is based on a shunt and an analog amplifier, so it is sensitive to offset drift and common-mode noise. The protection thresholds are fixed by the protection ICs, which limits flexibility. Finally, the experimental validation was performed on a small electric vehicle battery pack rather than a full-scale vehicle pack. These limitations define the direction for future work.
Future work. I would extend this work in four directions. First, I would replace the dissipative balancing circuit with a switched-capacitor or inductor-based active balancing circuit for a larger electric vehicle battery pack. Second, I would implement a model-based state estimator, such as an extended Kalman filter, using a reduced-order equivalent-circuit model of the electric vehicle battery pack. Third, I would add distributed temperature sensing and a thermal management controller to reduce thermal gradients. Fourth, I would integrate a communication interface so that the electric vehicle battery pack can exchange state, limit, and diagnostic information with the vehicle controller. These improvements would make the electric vehicle battery pack more suitable for high-power and high-energy applications.
Conclusion. I designed, implemented, and tested an energy management system for an electric vehicle battery pack. The system uses a PIC18F452 microcontroller to acquire voltage, current, and temperature data. It uses a dedicated protection architecture to prevent overcharge, overdischarge, overcurrent, and short circuit. It uses a charge-balancing circuit to reduce cell imbalance during charging. It estimates state of charge using a hybrid of open-circuit voltage initialization and ampere-hour integration with temperature, cycle-life, and recovery compensation. The experimental results show that the electric vehicle battery pack management system can monitor the operating state, protect the cells, support charge equalization, and provide a practical state-of-charge indication. Although further improvements are needed for large-scale electric vehicle battery pack applications, my work provides a complete and extensible foundation for safe and reliable management of a lithium-ion electric vehicle battery pack.
The central conclusion of my study is that the electric vehicle battery pack cannot be treated as a simple voltage source. It is a dynamic, safety-critical, and aging-sensitive system. The more accurately the electric vehicle battery pack is measured, protected, balanced, and modeled, the more usable energy can be extracted from it, and the longer it will last. I believe that this principle will remain central to the development of future electric vehicle battery pack management systems.