Summary: | 博士 === 國立臺灣大學 === 電機工程學系 === 86 === The interference sources and three intrinsic noise sources of bioelectric
signal measurement are discussed in detail in this thesis. From the analysis
of the circuit model for bioelectric signal ,easurement, an equation, which
describes the equivalent noise contributed by the amplifier, is deduced to
estimate the performance of the designed circuit.
Three different approaches for interference removal are also proposed.
The first one is an active comb filter as well as a group of 138 bandpass
filters for its implementation. This proposed comb filter can be used to
eliminate the power-line interference and its harmonics. One popular bandpass
filter is chosen to implement the proposed active comb filter and its perfor-
mance is demonstrated by sinusoidal and triangular wave tests. The second one
is a novel preamplifier with multiple-feedback structure that can be used to
suppress the low-frequency interference. The proposed preamplifier with its
front end being implemented by an instrumentation amplifier enjoys the follow-
ing advantages: differential highpass filtering, high input impedance, high
CMRR and low passive sensitivity. The stability analysis is also given to
verify the inherent stability of this preamplifier. Experimental test results
demonstrate its feasibility. The power-line interference is assumed to be a
sinusoidal wave. This LMS-style algorithm updates the amplitude and the phase
of the sinusoid iteratively, and the interference frequency is estimated by
chirp-Z transform (CZT) successively. The simulation results show that this
method can eliminate sinusoid with its frequency keeping fixed over an interval
longer than the time needed for numerical convergence.
|