Oversampled Audio DAC Path With Low-Ringing Nyquist Emulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital audio processing systems introduce audio distortions due to high-order filters and oversampling, leading to unnatural sound reproduction and size, power, and cost constraints in consumer devices like mobile phones.
Innovation Solution
Implementing a digital filter-based system that upsamples audio data and applies a digital low-pass filter to emulate the response of a Nyquist digital-to-analog converter (DAC) and analog low-pass filter, optimizing time-domain response rather than frequency-domain response, which reduces pre- and post-ringing and phase distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional oversampling DAC with high-order digital low pass filters is used, then audio distortion from aliased images is reduced, but pre- and post-ringing and group delay increase causing unnatural sound reproduction
Solution Approach 1:
The patent changes the filter corner frequency parameter from the conventional Nyquist rate (22.05 kHz for 44.1 kHz sampling) to a higher frequency (e.g., 44.1 kHz or higher). This parameter change allows the use of lower-order filters that produce less pre- and post-ringing while still effectively removing aliased images through the combination of oversampling and the elevated filter corner frequency.
Solution Approach 2:
The patent applies oversampling before the low pass filtering stage, which preliminary removes the aliased images from the signal. This preliminary action allows subsequent filtering to be performed with lower-order filters that have more favorable time-domain characteristics with reduced ringing artifacts.
2Object-generated harmful factors
If Nyquist-rate (non-oversampling) DAC is used, then pre- and post-ringing is reduced, but additional audio distortion is introduced and up-sampling images are not properly removed
Solution Approach 1:
The patent performs oversampling as a preliminary action before digital-to-analog conversion, which creates redundant samples that allow for effective low pass filtering to remove up-sampling images. This preliminary oversampling enables the system to achieve both reduced ringing (compared to conventional filtering) and proper removal of imaging artifacts.
Solution Approach 2:
The patent changes the filter corner frequency to a value higher than the Nyquist rate of the input digital audio data. This parameter change allows the low pass filter to pass the audio band frequencies with minimal phase distortion while still attenuating the up-sampling images, achieving both goals simultaneously.
3Reliability
If high-order filters with narrow transition bands are used, then aliased images are effectively removed, but impulse response energy spreads in time causing unnatural sound
Solution Approach 1:
The patent changes the filter corner frequency parameter to a higher value, which relaxes the transition band requirements. This allows the use of lower-order filters with shorter impulse responses that still achieve effective aliased image removal when combined with oversampling.
Solution Approach 2:
The patent performs oversampling before filtering, which preliminary separates the signal spectrum from the aliased images. This preliminary spectral separation reduces the filtering burden, allowing lower-order filters with shorter impulse responses to achieve the same aliased image rejection as high-order filters would provide alone.
4Reliability
If conventional oversampling system is used, then audio distortion is reduced, but device size, power consumption, and cost increase
Solution Approach 1:
The patent changes the filter corner frequency to a higher value, which enables the use of lower-order filters. Lower-order filters require fewer computational operations, directly reducing the power consumption of the digital signal processing while maintaining effective aliased image removal through oversampling.
Solution Approach 2:
The patent performs oversampling before filtering, which simplifies the subsequent filtering requirements. This preliminary action reduces the computational complexity of the filter stage, leading to lower power consumption in the overall system while maintaining audio distortion reduction benefits.
Data Source
AI summary
The behavior of a NOS DAC and an analog filter may be emulated by electronic components of an integrated circuit (IC) by upsampling data and applying a digital filter to the upsampled data. For example, the IC may include a zero-order-hold circuit that upsamples data from a first input sample rate to a second, higher input rate. The upsampled data may be passed to an Asynchronous Sample Rate Converter (ASRC) that performs further upsampling (e.g., from 8*Fs-64*Fs). The upsampled data may be passed to a digital low pass filter. The digital low pass filter may emulate, for example, a response of a fifth order Butterworth analog filter to mimic the effect of analog processing. The IC may integrate the upsampling circuit, the low pass digital filter, a digital-to-analog converter (DAC) and an amplifier to provide an audio solution for playing high-fidelity music in a mobile device.