A Noise-shaping Analog-to-Digital Converter using a ΔΣ Modulator Feedforward Network.

A noise-shaping analog-to-digital converter (ADC) using a ∆Σ modulator network is proposed, and signal-level simulations are carried out as a proof of concept. The present architecture is based on a feedforward artificial neural network, where an N -bit digital output is generated through N channels con- taining one ∆Σ modulator per channel. A moving average taken from each ∆Σ modulator is optimized to obtain a multi-level feedforward signal. Simulation results show proper noise-shaping characteristics for both the first-order and second-order ∆Σ modulators. The effective number of bits (ENOB) increases as the number of channels increases up to six. A non-binary conversion scheme suggests a further advance in the ENOB. Finally, the present ADC is compared with conventional multi-bit ∆Σ modulators.

Authors

• Takao Waho

Keywords

No keywords are indexed for this paper.