Online battery EIS using cell voltage balancing circuit and PRBS excitation

Abstract

Effective energy storage solutions have become essential as the world increasingly turns to renewable energy. Research into lithium-ion batteries, which offer excellent performance, is crucial for this transition. A key technology in this area is electrochemical impedance spectroscopy (EIS), which is used to assess the state of charge (SOC) and the health of batteries and fuel cells. Traditionally employed in research labs, EIS is gaining attention for real-time online monitoring integrated into management systems. It works by injecting a perturbation current into a battery, measuring the resulting voltage (or vice versa), and calculating impedance, which correlates with SOC and state of health (SOH). Various excitation signals, like sinusoidal and pseudo-random binary sequences (PRBS), can be utilised. There are two approaches to the implementation of real-time EIS: using additional dedicated EIS circuits [1]; or using existing circuits such as power electronic converters [2] or cell voltage equalisation circuits [3].
This paper introduces a method for conducting real-time EIS utilising the battery pack cell equalisation circuit as illustrated in Fig. 1. However, instead of using single sine excitations as reported in [3], the paper investigates using PRBS with different values of cell equalisation resistors and PRBS number of bits. The excitation is achieved by applying a PRBS signal to the gate of the MOSFET in the cell equalisation circuit. [2]
The frequency spectrum of a PRBS signal is related to its number of bits and clock frequency. The number of bits determines the breadth of the spectrum, while the clock frequency determines its minimum and maximum frequency values.
The excitation spectrum amplitude and power reduce as the frequency increases, i.e., the signal-to-noise ratio (SNR) reduces at high frequencies. For this reason, the clock frequency needs to be about 2.5 times higher than the maximum frequency we aim to observe to achieve a good SNR in the target frequency range. The lower the number of bits, the narrower the observable frequency range of the PRBS spectrum, which means that the excitation needs to be repeated with different clock frequencies to cover the entire battery impedance spectrum frequency range, typically 10 mHz – 1 kHz. A higher number of bits would enable fewer repeats, but it has the disadvantage of spreading the excitation power over a broader range of frequencies, which means that the amplitude of the excitation signal at a particular frequency decreases as the number of bits increases, thus worsening the SNR. This may be mitigated by reducing the cell equalisation resistor to increase the level of the excitation current at the expense of increasing energy losses by the EIS measurement.
In this paper, we investigate the effect of the PRBS number of bits and equalisation resistance value on the quality of the EIS of Efest 18650 lithium-ion cells in comparison with offline EIS measurement obtained using an Ivium-n-stat workstation. The PRBS cell voltage and current are recorded using a digital storage oscilloscope for a period that is at least 3 times the PRBS lowest frequency period, which is equal to the clock frequency divided by the number of bits. Using a MATLAB script, the collected voltage and current waveforms are de-trended, i.e., the declining DC offset is removed before carrying out Fast Fourier Transform (FFT) analysis and then calculate the impedance spectrum. Fig. 3 and 4 show an example of the impedance spectrum of an Efest cell at 100% SOC using 6-bit PRBS excitation repeated 3 times at different clock frequencies, in comparison with the spectrum measured using the Ivium-n-stat workstation for various values of cell equalisation resistor, 2 and 10 Ω, respectively.
There is a satisfactory agreement between the offline and proposed methods. More data scatter and variance are observed at 10 Ω due to the lower value of the excitation current and its SNR. Increasing the number of bits increases the data scatter further due to the spread of PRBS power over a broader range of frequencies, which reduces SNR. Results for higher numbers of bits will be presented at the conference.

References
[1. Islam, S.M.R. and S.Y. Park, Precise Online Electrochemical Impedance Spectroscopy Strategies for Li-Ion Batteries. IEEE Transactions on Industry Applications, 2020. 56(2): p. 1661-1669.
2. Liu, K.P., et al. Online Broadband Electrochemical Impedance Spectroscopy within Direct Power Control of a Neutral Point Clamped Inverter. in 2024 IEEE International Conference on Electrical Systems for Aircraft, Railway, Ship Propulsion and Road Vehicles & International Transportation Electrification Conference (ESARS-ITEC). 2024.
3. Blömeke, A., et al., Balancing resistor-based online electrochemical impedance spectroscopy in battery systems: opportunities and limitations. Communications Engineering, 2024. 3(1): p. 62.

More information

Identifiers

ORCID for Hendras Dwi Wahyudi: orcid.org/0009-0001-3940-8244

ORCID for Kai-Ping Liu: orcid.org/0009-0003-4092-4558

ORCID for Richard Wills: orcid.org/0000-0002-4805-7589

ORCID for Andrew Cruden: orcid.org/0000-0003-3236-2535

ORCID for Suleiman Sharkh: orcid.org/0000-0001-7335-8503

Catalogue record

Export record