Non-integer CIC Filter Eliminates Integrator Coupling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional CIC interpolation filters face challenges with non-integer sample/hold ratios, leading to undesirable tones close to the signal band and high computational complexity due to integrator coupling in the output domain.
Innovation Solution
A non-integer CIC interpolation filter design that eliminates the need for integrator coupling in the output domain by using an N-th order differentiator and multiplier network, with integrators operating at the output sampling rate, simplifying the structure and reducing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional CIC interpolation filters are used with non-integer sample/hold ratios, then interpolation can be achieved, but undesirable tones appear close to the signal band and computational complexity increases due to integrator coupling
Solution Approach 1:
The filter is divided into two independent domains: an input sample domain containing differentiators and multiplier stages, and an output sample domain containing uncoupled integrators. This segmentation eliminates the need for integrator coupling while maintaining non-integer interpolation capability.
Solution Approach 2:
A multiplier network acts as an intermediary between the differentiator section and the integrator section. This intermediary processes the signal in the input sample domain before passing it to the integrators, enabling non-integer interpolation without requiring coupling between integrators.
2Adaptability or versatility
If conventional CIC filters use integrator coupling in the output domain, then non-integer interpolation is achieved, but computational complexity increases
Solution Approach 1:
The computational process is segmented into two distinct domains: input sample domain operations (differentiation and multiplication) and output sample domain operations (integration). This segmentation allows each domain to be optimized independently, with uncoupled integrators improving computational efficiency.
Solution Approach 2:
The conventional approach of using coupled integrators in the output domain is replaced by using a multiplier network in the input sample domain. This substitution eliminates the computational overhead of integrator coupling while achieving the same non-integer interpolation effect.
3Adaptability or versatility
If non-integer sample/hold ratios are used in conventional systems, then USB system compatibility is achieved, but alias attenuation performance deteriorates
Solution Approach 1:
Alias suppression is performed in advance in the input sample domain through the differentiator and multiplier stages before the signal reaches the integrators. This preliminary action prevents alias tones from being generated in the first place, rather than attempting to filter them out later.
Solution Approach 2:
The multiplier network serves as an intermediary that processes the signal with appropriate coefficients before it reaches the integrators. This intermediary action ensures that the signal is properly conditioned to prevent aliasing while maintaining compatibility with non-integer sample/hold ratios required by USB systems.
Data Source
AI summary
A non-integer CIC interpolation filter is provided for use in sigma-delta digital-to-analog systems, which realizes non-integer interpolation but eliminates the need for coupling of the integrators in the output domain. The present non-integer interpolation filter provides for more attenuation to all of the aliases of the input signal and has eliminated the need of complex computations.