Cancellation of Quantization Noise in Sigma Delta Modulators Through Modeling of Non-Linearities

The linearity of Sigma Delta Modulators (SDMs) is typically evaluated by performing a Fourier Transform of the output bitstream. However, the presence of quantization noise in the output bitstream complicates this task: many samples and lengthy simulations are needed to lower this noise floor and evaluate the signal harmonics. We present here a method to estimate the quantization noise in the presence of non-linearity, remove it from the SDM bitstream, and reduce the number of samples needed to quantify the SDM linearity. By applying the proposed technique to a behavioral model of a 4^th-order SDM, the accuracy of the linearity estimation is kept unchanged using 5000x less bitstream samples. This in turn can significantly reduce the SDM verification time.