A novel architecture of a bandpass sigma-delta converter based on signal prediction

This paper presents a new design of a sigma-delta band-pass converter using a feedback loop associated with a signal prediction function. Thanks to a model of the signal to digital, i.e. the target signal, and a Kalman state estimator, the system gives a predicted signal. An analog error summer is used to compare the predicted signal with the target signal and creates an error signal. A negative feedback loop is thus formed by this system to drive the error signal toward zero. Consequently, operating on the relatively small error signal makes it possible to enhance the performance of the sigma-delta converter. This architecture shows an increase of more than 25 dB in signal quantization noise ratio compared to conventional second order band-pass sigma-delta converter and has the advantage of being a completely digital solution.