- Muchas notas - Fran Acién

20230509 - Quantization noise

It is defined a similar magnitude as in SNR, the Signal-to-quantization-noise ratio:

$$ SQNR = \frac{P_{signal}}{P_{noise}} = \frac{E[x^2]}{E[\tilde x^2]} \\ E[\tilde x^2] = \sigma_{eq}^2 = \text{Quantization noise power} = \frac{\Delta^2}{12} \\ \Delta = \text{step size} \\ SQNR \approx 6,0206b + 1,7609 $$ That means that every bit reduce the quantization noise for 6 dB, because the step size is half with every bit. The 1,7609 is a constant that is defined between the relations of the maxium power of the pure sine wave.

It is also introduced the concept of dynamic range that is the relation between the largest magnitude and the lowest magnitude, like:

$$ DR = 20 \cdot \log_{10} (\frac{Max}{Min}) \\ DR_{ADC} = 20 \cdot log_{10} \big ( \frac{2^b}{1} \big ) = (6.02 \cdot b ) \ dB $$