i have a basic switched cap amplifier (see the attachement pls). i did pss+pnoise simulation (time-domain noise performed by an ideal sample-hold circuit following the SC-amp. the timing of sample signal is also shown). the noise level is too high. from the noise summary, over 90% of the noise is from the opamp.

therefore i thought the opamp is the dominant noise source, and a continues-time configuration of the phase 2 would also give an informative picture of the noise performance. the circuit is also in the attachment, and noise simulation of spectre is used. a sp1tswitch is used to set the operating point. however, such simulation shows the power spectral density is 100 times lower than the pnoise result (in V^2/Hz).

i know the two simulations certainly cannot give the same results, but i expect the noise simulation can at least give some basic guideline for the design. such big difference is inexplainable to me.  the opamp has a GBW at about 2*fsample, so i dont expect too much noise to be folded back.

could somebody tell me what is the reason? which simulation is reliable? thanks in advance.