CIC Interpolator Architecture Using Precomputed Integrator States
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Higher order Cascaded Integrator Comb (CIC) filters face issues with offset accumulation and quantization errors due to word retention, leading to distorted output when attempting to reset integrators, which is not effective for non-linear interpolation.
Innovation Solution
Pre-computation of intermediate integrator states in a Pth order cascaded integrator comb architecture, allowing for offset-free operation by applying outputs from each comb to pre-computed states at a multiple of the sampling frequency, thereby reducing errors and maintaining data integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If integrators are reset to reduce word retention and quantization errors, then error accumulation is reduced, but data loss occurs and output distortion increases in higher order interpolators
Solution Approach 1:
The patent pre-computes intermediate integrator states before the actual integration process. By calculating expected intermediate values in advance and storing them in lookup tables, the system can directly retrieve pre-computed states instead of performing real-time integration that would require resetting. This eliminates both the error accumulation problem and the data loss/distortion problem associated with resets.
2Manufacturing precision
If CIC filter order is increased to improve interpolation accuracy, then manufacturing precision improves, but device complexity and computational requirements increase
Solution Approach 1:
The patent divides the high-order CIC filter into multiple lower-order stages, each handling a portion of the overall interpolation task. By segmenting the filter into manageable stages with intermediate lookup tables, the system achieves high-order interpolation accuracy while keeping each individual stage relatively simple. This segmentation reduces the computational burden on any single unit and makes the overall system more implementable.
Solution Approach 2:
The patent pre-computes and stores intermediate integrator states in lookup tables before runtime. This allows the filter to retrieve pre-calculated values instead of performing complex real-time computations for each high-order operation. The preliminary computation shifts the complexity from runtime processing to offline preparation, significantly reducing the device complexity during actual operation.
3Speed
If real-time computation is performed without pre-computation, then processing speed is maintained, but power consumption and computational load increase
Solution Approach 1:
The patent pre-computes intermediate integrator states and stores them in lookup tables during an offline phase. During real-time operation, the system simply retrieves these pre-computed values from memory rather than performing complex calculations. This approach maintains real-time processing speed while dramatically reducing power consumption during actual operation, as memory access consumes far less power than computational operations.
Data Source
AI summary
A digital filtering method includes receiving a digital signal, and passing the digital signal through a Pth order comb cascade. The method includes beginning pre-computing of intermediate integrator states of a Pth order integrator cascade as a function of the digital signal, prior to receiving output from a last comb of the Pth order comb cascade. The outputs from each comb of the Pth order comb cascade are then applied to the pre-computed intermediate integrator states to thereby produce a filtered version of the digital signal. The Pth order comb cascade may operate at a sampling frequency, and the pre-computing of the intermediate integrator states is performed at the sampling frequency, while the application of the outputs from each comb of the Pth order comb cascade to the pre-computed intermediate integrator states is performed at a multiple of the sampling frequency.