The input impedance of the Tayloe circuit varies widely with frequency. For signals at the clock frequency it goes very high, infinite in the ideal.
?
As the input signal deviates from the clock frequency the QSD input impedance falls rapidly, ultimately reaching a low value limited by the MUX channel resistance. It follows the bandwidth curve that the QSD develops?
?
This behavior is not intuitively obvious. Far from it! It was confirmed in my simulations and also via Steve Dick's correspondence with Dan Tayloe himself.
?
IMO a resistive pad could be used to limit the impedance excursions and make the performance of a preceding filter more predictable.
?
Good question, OM ..JZ KJ4A?