ADC Dither Adjustment Using Analog Noise Subtraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High performance analog-to-digital converters (ADCs) face challenges with differential non-linearity (DNL) errors, which degrade spurious free dynamic range (SFDR) and cause false or missed detections, especially when dealing with large signals and smaller signals simultaneously. Conventional dithering techniques either increase noise floor or are impractical for providing uncorrelated random dither to multiple ADCs in systems like phased array radar.
Innovation Solution
A data conversion system that generates a random, non-deterministic analog noise signal, adds it to the RF signal, and uses two quantizers to produce a linearized digital representation by subtracting the digital noise signal, thereby reducing spurious tones without significantly increasing the noise floor. This system includes an analog noise generator, an adder circuit, and a digital dither adjustment module, with the second quantizer having lower bit resolution than the first, and can be applied to multiple ADCs in phased arrays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional dithering techniques are used to improve SFDR, then spurious tones are reduced, but noise floor increases
Solution Approach 1:
The patent segments the dithering process into two independent quantization paths: a first quantizer processes the dithered signal while a second quantizer processes only the dither signal. This segmentation allows separate handling of signal and dither components, enabling subsequent subtraction to remove dither-related noise from the final output.
Solution Approach 2:
The patent extracts the dither signal component from the overall system by routing it through a dedicated second quantizer. This extracted digital dither signal is then subtracted from the first quantizer's output, effectively removing the dither contribution and its associated noise floor elevation.
2Adaptability or versatility
If digital PRN generation is used to create random dither, then dither can be generated, but it is expensive and cannot provide uncorrelated dither for multiple ADCs
Solution Approach 1:
The patent replaces complex digital PRN generation systems with a simple analog thermal noise source. Thermal noise is inherently random and uncorrelated, providing the desired dither characteristics without requiring expensive digital random number generators or complex correlation management for multiple ADCs.
3Manufacturing precision
If dither is added to linearize quantization, then DNL errors are reduced, but resolution and linearity improvement comes at the cost of increased noise
Solution Approach 1:
The patent converts the harmful effect of added dither noise into a benefit by using the same dither signal that linearizes the quantization to also serve as a reference for subtraction. The dither's randomizing effect improves linearity, while its known characteristics allow it to be removed from the final output, converting the noise introduction into a controllable and removable element.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The approach effectively reduces spurious tones by about 10 dB while minimizing noise floor increase, achieving near-perfect performance in SFDR and maximizing signal-to-noise ratio (SNR) by nearly eliminating DNL errors' contribution to overall noise, as demonstrated in comparative test results.
Implementation Method 1
A well-known technique called dithering is often required to maximize SFDR. Dithering is the process of adding an uncorrelated signal, such as pseudo random noise (PRN) or broadband noise, to a desired analog signal prior to the analog input gate of the ADC. A common approach to creating dither is to use a noise or thermal diode whose output is summed with the wanted signal prior to digitization.
Data Source
AI summary
A method and system for data conversion includes an analog noise generator to generate a random, non-deterministic, analog noise signal. An adder adds the analog noise signal to an analog RF signal to produce a dithered analog signal. A first quantizer converts the analog noise signal to digital to produce a digital noise signal. A second quantizer converts the dithered analog signal to a digital equivalent signal. A digital dither adjustment module removes amplitude measurements of the digital noise signal from the digital equivalent signal to obtain a linearized digital representation of the analog RF signal.