Ph.d. degree
Chine University of Geosciences (Wuhan), China, 2108

Huan Liu

Supervisor: Haobin Dong

Research on an improved Overhauser magnetometer

Keywords:

• Optimization of Overhauser magnetometer
• free induction decay signal
• noise modeling
• secondary tuning

Abstract:

Background: The Overhauser magnetometer based on dynamic nuclear polarization (DNP) effect is a high-precision device for weak magnetic field measurement. The Larmor precession of excited protons around the geomagnetic field can generate a free induction decay (FID) signal. The strength of the magnetic field can be obtained by measuring the frequency of this signal. However, with the advances in the exploration technology, there is a need for the Overhauser magnetometer to have higher accuracy and higher sensitivity.

Methods: The goal of this work is trying to research on an improved Overhauser magnetometer with higher accuracy and sensitivity. Our approach is characterized by four key components: 1) the noise model of FID signal is established and different types of noises are investigated. Moreover, the measurement error has been analyzed and the measurement accuracies effected by different signal to noise ratios (SNRs) are determined; 2) a multi-channel frequency measurement algorithm is developed to improve the accuracy of the measured magnetic fields in a broader dynamic range and in a noisier environment; 3) a secondary tuning algorithm based on singular value decomposition (SVD) and shot time Fourier transform (STFT) is designed, which cannot only improve the accuracy of the sensor's tuning but also shorten the time of tuning process; 4) numerous laboratory and field tests on the developed prototype including noise level, frequency measurement accuracy, magnetic field measurement accuracy, geomagnetic observation and ferromagnetic target localization compared with one of the commercially available Overhauser magnetometers in the world market are conducted.

Results: The relationship between the total noise model of FID signal and the error of frequency measurement is formulated in a quantitative way. The simulation results show that when the SNR reached up to 10 dB, the improvement of accuracy is not obvious while in the field tests, the SNR is about 30 dB. As a whole, the trend of the simulation and the field test results are approximately consistent. The absolute error measured by the proposed frequency measurement algorithm is about 7.75 times smaller than that measured by the original method with a single channel. In addition, the sensitivity is reduced to 0.036 nT, which is improved by 11.4 times. The proposed tuning algorithm has higher accuracy and higher speed than the current commonly used methods, peak detection and auto-correlation, even in the interferential environment. It makes up for the insufficiency of the existing tuning methods by improving the magnetometer's ability to adapt to the environment, thereby solving the detuning phenomenon in the process of measurement. The longtime geomagnetic observational data recorded by the prototype and a commercial Overhauser magnetometer confirm that, the proposed device has a mean square error (a little larger than 0.3 nT) that is close to one for the commercial magnetometer (no more than 0.3 nT). Furthermore, the prototype shows a strong ability in magnetic anomaly detection and localization. The integrative structure is adopted, which can ensure the convenience and reliability of operation in the field.