Study On Signal Detection Technology For Giant Magnetoresistance Sensors

1/f noise is the major noise of giant magnetoresistive (GMR) sensors, which causesintrinsic detection limit at low frequency. Our research team supported by NationalNatural Science Fund has proposed vertical motion flux modulation (VMFM) structureswhich can suppress1/f noise Magnetic sensors effectively, and the magnetostaticdetection ability is largely improved. To realize pico-Tesla detection based on GMRsensors, the1ppm resolution of detection circuit should be required. Therefore, it issignificant to investigate the signal detection technology for GMR sensors.Presently, many attentions have been paid on low-noise pre-amplifier circuits andsteady signal processing methods, but seldom on the combination of two techonolgies.According to the discussions about overseas research state, several key techniques forGMR sensor signal detection have been inveatigated in this thesis. The main contentsare involved as follows:Technologies for low noise pre-amplifier were researched. Beginning withstructure design, how to regulate the quiescent operation point and suppress theequivalent input voltage noise were explored. Meanwhile, the circuit performaces weresimulated and the results showed that our design requests could be satisfied.The Signal detection algorithm based on spectrum correction technology wasstudied. Using LMS methods to fit triangular self-convolution window, we theoreticallyanalyzed the spectrum analysis ability of triangular self-convolution window andrectangle window under random noise influence, and provided the performance contrastof the two windows. A spectrum correction method based on moving average windowhas been proposed. The amplitude measure error and frequency measure error of thethree detection algorithms evaluated respectively by means of simulation.Also, the signal detection algorithm based on digital Lock-In Amplifier (DLIA)was investigated. By multi-sampling rate signal processing theory, a narrow band FIRfilter was designed. And the frequency response characteristics of its component filterswere researched. After resolving the realization difficulties due to high-order caused bydirect FIR design, we worked out a DLIA for practical requirements.Finally, three experiments were performed for testing the low-noise pre-amplifiercircuit and the performances of three above-mentioned detection algorithms anddetection system. The results verified the reasonablity of circuit design and theeffectiveness of detection algorithms.