Hi,

Does anyone know the lower limit on THD that one can achieve by using simple schemes for reducing signal-dependent charge-injection.

For instance, if one uses proper timing to cutoff the signal-independent switches before cutting off the signal-dependent ones (also known as bottom-plate sampling), and the design is fully-differential, then how good can the THD be for a unity-gain SC amplifier?

Of course, it depends on the choice of capacitors, choice of switches etc., but what are the other factors that introduce nonlinear distortion and how can a design be evaluated to assess if the performance obtained is the best possible for choice of components (switch size and cap size), or if there is still room for improvement?

Are there any references which provide such information? Most articles just explain techniques for reducing nonlinear distortion due to charge-injection, but apart from bottom-plate sampling, the methods described are largely unreliable as they are based on certain assumptions (like 50-50 splitting of charge as assumed in dummy compensation techniques).

How does one achieve the fabulously high-linearity specs that are common these days (14-16 bits or more) without burning prohibitively large amounts of power.

Thanks in advance for any useful comments or suggestions.

Vivek