Mixed-Signal Analog to Digital (ADC) Converter Anti-Alias Filter Criteria

Analog-to-Digital Converters (ADC) are used to sample and quantize an analog signal for processing in the digital domain.  According to the Nyquist sampling theorem, frequency components above one-half f_s will be aliased into the baseband (first Nyquist zone).  This aliased noise is added to the ADC quantization noise, which is evenly distributed across the Nyquist bandwidth, DC to f_s/2 (assuming the baseband signal and noise are uncorrelated).  ADC quantization noise is an inverse function of the ADC effective number of bits (ENOBS). Generally, ADC noise is detrimental to system performance. System requirements can be developed to allocate ADC noise into a quantization-noise budget and an alias-noise budget based upon practical considerations of ADC sample rate versus ENOBS versus cost.  The value of aliased noise is a function of the input noise-power spectral density, or NPSD.  This NPSD must be estimated; its source may be EMC, crosstalk, etc. An analog Anti-Alias Filter (AAF) preceding the ADC can be employed to reduce the NPSD to tolerable levels prior to ADC sampling. The AAF is often a multi-order filter with pass band designed to optimize the system signal-to-noise ratio and system performance.

Design of the analog Anti-Alias Filter can be quite involved.  Over-sampling ADCs can reduce the cost and complexity of the analog Anti-Alias Filter in certain applications, and reduce the pass band quantization noise due to noise shaping and processing gain.

The analog experts at Analog Advice can assist you with mixed-signal requirements and design, including the Anti-Alias Filter and ADC.