Binary dithered oversampling analog-to-digital converters

Journal Title
Journal ISSN
Volume Title
University of Delaware
In today’s world, analog-to-digital converters (ADC) play a major role. Our modern society requires and depends on electronic devices that process the analog input data in the digital domain, such as cellphones, audio and video systems, and even domestic appliances. Therefore, the development of fast and accurate ADCs have become a key issue for the industry. In general, a good ADC is the one that achieves high resolution (low quantization noise) with low complexity. One of the most popular techniques to decrease the quantization noise is to digitalize the incoming signal with a sampling frequency many times higher than twice the signal bandwith. This technique is called oversampling. Probably the most popular oversampling converters are the Sigma-Delta (SD) modulators which use a very high sampling frequency and a binary quantizer (for first order SD). In SD, the quantizer is embedded into a feedback loop in such a way that the quantization noise is not only spread over the spectrum (because of oversampling), but it is also shaped to the upper frequencies (this is called noise shaping). The big problem with SD is that the quantization noise spectrum presents undesired harmonics caused by the nonlinear nature of the quantizer. To avoid this, the solution is to add an independent signal before the quantization stage called dither. It was proven by Lipshitz and Vanderkooy in [1] that dithering a first order SD modulator is ineffective as it turns the modulator in constant overload. In addition to this problem, because of the feedback loop, higher order SD modulators can be unstable. That being said, the aim of this work is to present a simple oversampling ADC without feedback capable of generating the minimum uncorrelated quantization noise that yields the maximumpossible SNR at the output. For that purpose, this work develops the statistical characteristics of the optimum dither that achieves the mentioned goal for different types of input signals.