TL;DR
This paper introduces a novel fiber optic oxygen sensor that leverages the All-Phase Fast Fourier Transform (apFFT) algorithm for highly accurate and stable phase detection. This approach overcomes limitations of traditional methods, offering a more compact, cost-effective, and robust solution for oxygen sensing in various environments, particularly excelling at low oxygen concentrations.
What Problem Does This Paper Address?
Accurate and reliable oxygen sensing is crucial across numerous fields, from environmental monitoring and industrial processes to medical diagnostics and agricultural applications. Traditional oxygen sensors, such as Clark electrodes, often have drawbacks like oxygen consumption, slow response times, and susceptibility to interference. Fiber optic oxygen sensors, particularly those utilizing fluorescence quenching, offer significant advantages, including non-consumption of oxygen, fast response, and high sensitivity. However, accurately demodulating the phase information from the fluorescence signal, which is key to precise oxygen concentration measurement, has been a challenge. Existing methods, like lock-in amplifiers, can be bulky, expensive, and prone to noise, limiting their application in complex or remote environments. This paper aims to address these limitations by developing a fiber optic oxygen sensor with an improved phase detection method.
Methods
The core innovation in this paper is the application of the All-Phase Fast Fourier Transform (apFFT) algorithm for phase detection in a fiber optic oxygen sensor. Unlike traditional FFT methods that can suffer from spectral leakage and phase invariance issues, apFFT, developed by professors at Tianjin University, offers improved performance.
The sensor's operation relies on the principle of fluorescence quenching. An excitation light source causes a fluorescent material to emit light. Oxygen molecules in the environment interact with the fluorescent material, quenching its fluorescence. This quenching affects both the intensity and the lifetime of the fluorescence. The paper focuses on measuring the phase shift of the fluorescence signal relative to the excitation light, as this phase shift is directly related to the fluorescence lifetime and, consequently, the oxygen concentration.
The apFFT algorithm is employed to process the electrical signals from the excitation light and the fluorescence. This involves:
- Signal Acquisition: Sampling the electrical signals corresponding to the excitation light and the fluorescence.
- Full-Phase Preprocessing: A novel preprocessing step is applied to the sampled data. This involves a convolutional window and phase shifting to create a new array. This step is crucial for reducing noise and suppressing spectral leakage, which are common issues in FFT-based analysis.
- Fast Fourier Transform (FFT): The preprocessed data is then subjected to an FFT to obtain the frequency components, including their real and imaginary parts.
- Phase Calculation: The phase of the excitation and fluorescence signals is calculated using the arctangent of the ratio of the imaginary to real parts.
- Phase Difference Determination: The lag phase shift between the fluorescence and excitation signals is determined by subtracting their respective phases. This phase difference is then converted into an oxygen concentration value based on a pre-established calibration curve.
The system design integrates a central processing module (STM32F103 microcontroller), a signal generation module (AD9833), an optical path component, and a photoelectric conversion module, all integrated onto a printed circuit board for compactness.
Datasets
The paper does not explicitly mention the use of pre-existing datasets for training or evaluation. Instead, it describes experimental setups and measurements conducted by the authors. The "dataset" in this context refers to the data collected from the fiber optic oxygen sensor under controlled oxygen concentrations (0%, 5%, 15%, 21%, and 50%) and varying phase shifts.
Key Insights and Ideas
- apFFT for Robust Phase Detection: The apFFT algorithm proves to be a powerful tool for accurately and stably measuring phase shifts in optical sensor signals, overcoming limitations of traditional lock-in amplifiers and standard FFT methods.
- Noise Reduction and Stability: The full-phase preprocessing step within the apFFT algorithm significantly reduces noise and improves the stability of the phase measurement, which is critical for reliable oxygen sensing.
- Compact and Cost-Effective Design: By replacing bulky lock-in amplifiers with a microcontroller-based apFFT implementation, the sensor design becomes more compact, portable, and potentially less expensive.
- Superiority at Low Oxygen Concentrations: The sensor demonstrates higher accuracy and better performance in environments with low oxygen concentrations, making it particularly valuable for applications where precise detection of subtle changes is required.
- Intrinsic Parameter Measurement: Focusing on fluorescence lifetime (indirectly via phase shift) rather than intensity offers better resistance to external factors like light source fluctuations or transmission path variations.
Key Results
- High Linearity and Accuracy: The apFFT algorithm achieved excellent linearity in phase detection, with a goodness of fit reaching 0.9999999457. The accuracy of the apFFT itself was reported to be 0.05°.
- Fast Response and High Resolution: The experimentally measured response time was less than 0.2 seconds, and the resolution reached 0.0001°.
- Sensor Performance: The developed fiber optic oxygen sensor exhibited high stability. The phase shift variation was within 0.1° for long-term measurements at a constant oxygen concentration.
- Detection Errors: The detection errors for different oxygen concentrations were:
- 0%: 0.0447%
- 5%: 0.1271%
- 15%: 0.3801%
- 21%: 1.3426%
- 50%: 12.6316%
- Low Noise: The noise level for the sensor was reported to be within 0.01° for phase differences, and the phase shift change range was within 0.05° for long-term measurements.
Limitations and Future Work
- Accuracy at High Oxygen Concentrations: The paper explicitly states that the sensor is not suitable for detecting high oxygen concentrations (above 21%) due to significantly increasing detection errors. The accuracy decreases sharply as oxygen concentration increases.
- Noise Impact at High Concentrations: The impact of noise becomes more pronounced at higher oxygen concentrations (21% and 50%), affecting stability.
- Future Work: While not explicitly detailed as future work, the authors imply that the sensor's suitability for low oxygen concentrations and its overall stability and compactness make it a promising candidate for field applications and long-term monitoring in gaseous or liquid environments. Further research could focus on extending the accurate measurement range to higher oxygen concentrations or developing complementary methods for such scenarios.
Summary updated