Noise Shaping
High-resolution signal conversion topologies trade spatial quantization accuracy for temporal sampling frequency. A delta sigma adc oversamples input signals at rates far above the Nyquist limit while pushing quantization noise out of the measurement band. Feedback loops within the modulator subtract the quantized output from the analog input, integrating the error over time.
System selection depends on achieving high effective number of bits for low-bandwidth physical measurements.
Quantization Resolution
Modulator feedback structures act as high-pass filters for quantization noise while presenting a low-pass characteristic to input signals. Higher-order modulators increase noise shaping steepness, shifting more quantization energy toward half the sampling frequency. A digital decimation filter subsequent to the modulator removes out-of-band noise and reduces the sample rate to the Nyquist frequency.
The resulting signal achieves high dynamic range without requiring precision-matched analog components inside the silicon die.
Sampling Bandwidth
Idling tones generate spurious spectral peaks when constant direct-current inputs interact with quantization feedback loops. Pattern noise arises from short periodic limit cycles within the modulator loop, corrupting low-level sensor measurements. Adding pseudo-random dither signals at the modulator input breaks periodic limit cycles, spreading tone energy into a continuous noise floor.
Input buffer amplifiers must drive high-frequency capacitive switching currents generated by the internal switched-capacitor sampling network.
Oversampling Limit
Effective resolution specifications define operational bounds across specified decimation rates and output bandwidths.