The rapid advancement of new energy technologies has placed lithium-ion batteries at the forefront of energy storage solutions. Ensuring the safety, reliability, and efficient management of these power sources is paramount. Traditional battery management systems (BMS) primarily monitor basic parameters such as voltage, current, and temperature. However, to gain deeper insights into the complex electrochemical processes within a lithium-ion battery and enable precise state characterization, monitoring internal electrochemical characteristics is essential.
Among these characteristics, Electrochemical Impedance Spectroscopy (EIS) stands out as a powerful, non-invasive diagnostic tool. The impedance spectrum of a lithium-ion battery encapsulates rich information about its internal state. It has been shown to correlate strongly with critical states including the State of Charge (SOC), State of Health (SOH), and internal temperature. Different frequency bands within the EIS spectrum respond distinctly to these various states. For instance, the high-frequency region often relates to charge transfer and ohmic resistance, while the low-frequency tail is influenced by mass transport (diffusion) processes. Therefore, the ability to monitor the AC impedance of a lithium-ion battery online across a wide frequency range is of significant importance for advanced battery management, enabling real-time state assessment and early fault warning.
Conventional EIS measurement relies on bulky, expensive, and laboratory-bound potentiostats or frequency response analyzers. While highly accurate, these instruments are unsuitable for online, embedded applications. They often require the cell to be at a steady-state (resting) condition, which does not reflect its dynamic behavior under real operating loads. Several research efforts have aimed to develop portable EIS monitors using discrete components, data acquisition cards, or FPGAs. These solutions, however, often face challenges related to system complexity, computational load, power consumption, or integration difficulty with standard BMS architectures. There remains a need for a system that balances accuracy, integration simplicity, low power consumption, and real-time visualization capabilities.
This article presents the design and experimental validation of an embedded, online EIS monitoring system tailored for lithium-ion batteries. The system is built around a mainstream microcontroller, a dedicated battery management integrated circuit (IC) with EIS functionality, and a custom-developed PC software for data visualization and storage. It demonstrates accurate impedance measurement across a broad frequency spectrum, from 0.12 Hz to 7.8 kHz, and showcases its practical application in tracking dynamic impedance changes during charge-discharge cycles for state evaluation and over-discharge预警.
1. System Architecture and Design Principles
The primary objective of the online monitoring system is to inject a small-amplitude, multi-frequency AC current signal into the lithium-ion battery under test, simultaneously measure the resulting AC voltage response, and calculate the complex impedance at each frequency point. The core challenge lies in performing this signal generation, synchronous acquisition, and complex number calculation efficiently and accurately within an embedded platform.
1.1 Overall System Design
The system is architected into four main functional modules, as illustrated in the block diagram below. This modular approach ensures clarity in design and ease of integration.
| Module | Key Components | Primary Function |
|---|---|---|
| Data Acquisition & Control | Microcontroller (STM32), Battery Management IC | Generates control signals, performs EIS measurement via the dedicated IC, and processes raw data. |
| Power Supply | Voltage Regulators, DC-DC Converters | Provides stable, clean power to all active components (MCU, ICs, communication modules). |
| Data Communication | Bluetooth Low Energy (BLE) Module, RS-485 Transceiver | Facilitates bidirectional data transfer between the embedded system and the host PC/display. |
| Host PC Software (GUI) | Custom Application (e.g., LabVIEW, Python) | Provides user interface for control, real-time data visualization (Nyquist, Bode plots), and data logging. |
The workflow begins with system initialization. The host PC software sends a command to initiate an EIS scan. The microcontroller configures the battery management IC accordingly. The IC then applies the programmed sequence of AC current perturbations to the lithium-ion battery, measures the voltage response, and performs onboard digital signal processing (like Discrete Fourier Transform – DFT) to extract the impedance at each frequency. The calculated impedance data (real and imaginary parts, or magnitude and phase) is relayed back to the microcontroller, which packages it and transmits it via the chosen communication link to the host PC for display and storage.
1.2 Hardware Implementation: The Embedded Monitor
The heart of the embedded hardware is the combination of a general-purpose microcontroller and a specialized analog front-end (AFE) or battery management IC. The microcontroller (e.g., an ARM Cortex-M series chip) handles system control, communication protocols, and high-level data management. Its critical role is to interface with the EIS-capable AFE via a serial peripheral interface (SPI).
The selected battery management IC is pivotal. It integrates the necessary circuitry for galvanostatic (current-controlled) EIS excitation and synchronous voltage measurement. Typically, it contains a waveform generator (for creating the AC current signal), a precision analog-to-digital converter (ADC) for voltage sampling, and a digital processing core to compute the Fast Fourier Transform (FFT) or DFT. The excitation current amplitude (Iac) is set to be a small fraction of the C-rate of the lithium-ion battery (commonly less than 0.05C) to ensure linearity and avoid perturbing the battery’s state. For a standard 3.5 Ah lithium-ion battery, a 100 mA AC current is a suitable choice. The relationship between the set current and a reference resistor is often given by:
$$ I_{ac} = \frac{V_{ref}}{R_{set}} $$
where $V_{ref}$ is an internal reference voltage and $R_{set}$ is an external precision resistor.
The communication module offers flexibility. A BLE module provides wireless convenience for laboratory testing or mobile applications, characterized by its low power consumption. An RS-485 interface offers robust, long-distance wired communication, ideal for industrial environments with potential electromagnetic interference. The system can be designed to support both, selected via hardware configuration.
1.3 Software Implementation: Host Computer Interface
The host computer software serves as the user’s window into the system. Its design focuses on intuitiveness and functionality. The front panel provides real-time displays including:
- Numerical displays for battery voltage, and impedance parameters (Z’, Z”, |Z|, θ) at the current frequency.
- Dynamic plots: A Nyquist plot ( -Z” vs. Z’ ) and Bode plots ( |Z| and θ vs. Frequency ).
- Control inputs: Communication port selection, EIS scan parameters (start/stop frequency, number of points, interval time).
The software’s back-end handles data stream management. It implements a communication protocol with error-checking (e.g., using headers, checksums, and footers) to ensure data integrity. Upon receiving a valid data packet, the software parses it, updates the graphical displays, and appends the data with a timestamp to a file (e.g., .csv or .xlsx format) for post-analysis. This seamless integration of control, visualization, and logging is crucial for effective experimentation and analysis of the lithium-ion battery’s behavior.
2. Experimental Validation and Performance Analysis
To evaluate the performance of the developed online EIS monitoring system, a series of tests were conducted on commercial 18650 cylindrical lithium-ion cells. The cell specifications are summarized below:
| Parameter | Value |
|---|---|
| Nominal Capacity | 3500 mAh |
| Nominal Voltage | 3.7 V |
| Standard Charge Voltage | 4.2 V |
| Discharge Cut-off Voltage | 2.75 V |
| Approximate Internal Resistance | 22 mΩ |
2.1 Repeatability and Accuracy Assessment
A fundamental requirement for any measurement system is repeatability. The online monitor was used to perform five consecutive EIS scans on a single lithium-ion battery at a fixed state (e.g., 50% SOC, 25°C). The frequency range was set from 0.12 Hz to 7812.52 Hz with 40 logarithmically spaced points. The results, particularly the magnitude at key frequencies, show excellent consistency.
| Test Cycle | |Z| @ 0.48 Hz (mΩ) | |Z| @ 10.5 Hz (mΩ) | |Z| @ 1068 Hz (mΩ) |
|---|---|---|---|
| 1 | 39.52 | 31.88 | 22.15 |
| 2 | 39.48 | 31.85 | 22.14 |
| 3 | 39.61 | 31.91 | 22.16 |
| 4 | 39.45 | 31.82 | 22.13 |
| 5 | 39.56 | 31.89 | 22.15 |
| Mean | 39.524 | 31.870 | 22.146 |
| Std. Dev. (σ) | 0.061 | 0.034 | 0.011 |
| Relative Deviation | 0.15% | 0.11% | 0.05% |
The maximum standard deviation was observed at the lowest frequency (0.48 Hz), with a value of only 0.061 mΩ, corresponding to a relative deviation of 0.15%. This high degree of repeatability validates the stability of the hardware and the consistency of the measurement algorithm.
Accuracy was benchmarked against a professional electrochemical workstation (e.g., Solartron 1400A). The lithium-ion battery was conditioned to different States of Charge (30%, 50%, 70%) at a constant temperature of 25°C. EIS spectra were acquired using both the online monitor (galvanostatic mode, 100 mA AC) and the potentiostat (potentiostatic mode, 10 mV AC). The Nyquist plots from both systems showed remarkable overlap across all SOC levels. A quantitative comparison at the characteristic high-frequency intercept (which approximates the ohmic resistance, RΩ) reveals the level of agreement.
| Battery SOC | RΩ (Online System) (mΩ) | RΩ (Potentiostat) (mΩ) | Absolute Error (mΩ) | Relative Error |
|---|---|---|---|---|
| 70% | 22.08 | 22.01 | 0.07 | 0.32% |
| 50% | 22.21 | 22.15 | 0.06 | 0.27% |
| 30% | 22.40 | 22.31 | 0.09 | 0.40% |
The maximum relative error for the ohmic resistance was below 0.5%. Slightly larger discrepancies were sometimes noted at very high frequencies (>5 kHz), often attributable to differences in cabling inductance and connection methods between the two setups. Overall, the accuracy of the embedded system is deemed sufficient for online monitoring purposes of lithium-ion batteries.
2.2 Dynamic Impedance Monitoring and Application
The true advantage of an online system is the ability to track impedance changes under dynamic operating conditions, which is impossible with benchtop instruments requiring steady-state conditions. To demonstrate this, the lithium-ion battery was subjected to a standard 0.2C charge-discharge cycle while the online EIS monitor periodically performed rapid, single-frequency impedance measurements.
First, a suitable characteristic frequency needed to be selected. Analysis of full EIS spectra at different static SOCs showed that the low-to-medium frequency range (e.g., 0.1 Hz to 10 Hz) is most sensitive to SOC variations in this type of lithium-ion battery. A frequency of 0.95 Hz was chosen as a compromise between sensitivity and measurement speed (lower frequencies require longer sampling time).
During the charging process (Constant Current – Constant Voltage, CC-CV), the impedance magnitude at 0.95 Hz showed a general decreasing trend. This can be attributed to increasing temperature and the changing electrode thermodynamics. A noticeable acceleration in the impedance drop was observed during the transition from the CC to the CV phase, indicating a shift in internal dynamics.
The discharge process provided more critical insights. During the constant current discharge, the impedance initially decreased slightly (likely due to warming) but remained relatively stable through the mid-phase. However, as the lithium-ion battery approached the end of its discharge capacity, a sharp, exponential rise in the impedance magnitude was observed. This dramatic increase serves as a clear precursor to the voltage reaching the cut-off threshold.
This phenomenon can be leveraged for early fault warning, such as preventing over-discharge. By calculating the rate of change (first derivative) of the impedance magnitude, a more sensitive indicator is obtained:
$$ \frac{d|Z|}{dt} \approx \frac{|Z|_{t} – |Z|_{t-1}}{\Delta t} $$
| Discharge Phase | Battery Voltage | |Z| @ 0.95 Hz (mΩ) | Rate of Change (d|Z|/dt) (mΩ/s) | Observation |
|---|---|---|---|---|
| Early/Mid | ~3.7 V to ~3.3 V | ~39.5 to ~38.5 | ~ -0.01 to +0.05 | Stable, low variation rate. |
| Approaching Cut-off | ~3.0 V to 2.75 V (Nominal Cut-off) | ~39.0 to ~45.0 | Rises from ~0.1 to >0.6 | Clear upward trend begins. |
| Over-discharge | 2.75 V to 2.3 V (Deep Discharge) | >45.0 to >55.0 | >0.78 up to >4.0 | Rate of change escalates drastically. |
The data suggests that a threshold on the impedance magnitude rate of change, for example, **0.61 mΩ/s**, can be set as an early warning trigger. This trigger activated approximately **180 seconds before** the cell voltage reached the nominal discharge cut-off voltage of 2.75V, and significantly earlier before entering a damaging over-discharge region. This provides a valuable time buffer for the BMS to take protective action, showcasing a practical application of online EIS monitoring for lithium-ion battery safety.
3. Conclusion and Perspectives
This article detailed the design and implementation of a practical, embedded system for online Electrochemical Impedance Spectroscopy monitoring of lithium-ion batteries. The system successfully addresses key challenges by combining a dedicated analog front-end IC for accurate impedance measurement with a flexible microcontroller and a user-friendly host interface. Experimental validation confirmed its excellent repeatability (relative deviations < 0.2%) and good accuracy compared to a laboratory potentiostat, particularly in the frequency range most relevant for state diagnostics.
The core advancement demonstrated is the system’s capability for dynamic impedance tracking. By monitoring a characteristic low-frequency impedance point (0.95 Hz) during charge-discharge cycles, the system can capture the complex internal state changes of the lithium-ion battery in real-time. The most significant finding is the identification of a sharp rise in impedance and its rate of change as a reliable precursor to the end of discharge. Implementing a threshold-based warning on this parameter can enable proactive battery management, potentially preventing over-discharge and extending the lifespan of the lithium-ion battery.
The presented system offers a compelling balance of performance, cost, size, and power consumption, making it a strong candidate for integration into next-generation BMS. Future work will focus on:
- Multi-frequency Tracking: Implementing algorithms to rapidly track not just one, but several key frequency points to build a more robust state estimation model for SOC, SOH, and internal temperature of the lithium-ion battery.
- Algorithm Integration: Embedding machine learning or equivalent circuit model fitting algorithms directly onto the microcontroller for real-time, onboard state diagnostics without relying on a host PC.
- Multi-cell Monitoring: Scaling the architecture using the daisy-chain capability of modern battery management ICs to monitor impedance across all cells in a large lithium-ion battery pack simultaneously.
In conclusion, online EIS monitoring technology represents a significant step forward in intelligent battery management. By providing a window into the internal electrochemical processes, it empowers BMS with predictive capabilities, enhancing the safety, reliability, and efficiency of energy storage systems built on lithium-ion batteries.