The sampling frequency and the word width are the most important values when it comes to choosing converters. The basis for digital audio signal processing is the conversion of an analog signal into a digital, time-discrete signal, which, after digital processing, is turned back into an analog, time- continuous signal. The sampling theorem states that an audio signal band-limited to fmax must be sampled at a sampling frequency of at least fA= 2 x fmax . Accordingly, before the digitalization, the audio signal must be bandlimited by an analog low pass filter.
Our suppliers in this category ADC/DAC:
AKM provides customers with optimum solutions based on a wide variety of sensing devices using compound semiconductor technology and IC products featuring analog/digital mixed-signal technology. AKM is also one of the Key Players in the High End Audio IC market.
In order to keep the filter specification and the expenditure for realization associated with this low, most audio applications use a multiple of the sampling frequency actually required. In order to reduce the computing effort for the downstream digital processing needed due to the higher sampling frequency, as a rule a multi-stage decimation of the sampling frequency is applied, with anti-aliasing filters, which are often realized as FIR (Finite Impulse Response). FIR filters have the advantage that they have a linear phase, and, by polyphase splitting, with the incorporation of the decimation, the number of filter operations can be reduced by the decimation factor. An analogous procedure is fol owed before the digital-analog conversion; in other words, a step-up sampling of the audio signal takes place, with subsequent anti-aliasing filtering, whereby the reduction in the filter operations is likewise achieved by the interpolation factor using the same splitting process. Thanks to the step-up sampling, the requirements for the draft design of the downstream analog low-pass filter are minimized in the same way as is the case with the analog input filter. Over-sampling therefore helps keep the BOM costs in analog filter design to a minimum. And there is one more important property for the combination of over-sampling and low-pass filtering. If the filter operations are carried out in an adequately high word width, then the noise power produced by the quantization will be reduced by the factor 1/fA, as a result of which the SNR (Signal to Noise Ratio) wil be improved, and the quality of the audio signal increased. Besides the sampling frequency, the word width is the second major value to be considered when choosing converters. It determines the proportion of quantization noise in the audio signal. The higher the word width selected, the lower the proportion of noise power, and the better the SNR becomes. The two values, sampling frequency and word width, can be brought together in the SNR as fol ows: 2∗f max∗6 f mit SNR ¿ 2−N+1 , N = Number of bits = A q2 In order to meet the highest demands in digital audio processing, COIDICO offers a wide selection of converters with word widths of up to 24 bits, and sampling frequencies of 192 kHz and above.
Apart from differential inputs and outputs, and integrated amplifiers, converters are also available in multi-channel formats and with digital signal processing.