a noise consideration

[wateror]

when we design the amplifier which suppress the noise, we should choose big load capcitance C according to the kT/C equation on razavi's book.
but I'm confused because the the amplified signal is also related to C because it's a part of load. and the signal power should be inversely proportional square of C when it's high frequency signal.
so it seems the SNR drops when we increase C. but razavi says it increases, why?

 

[leo_o2]

Maybe razavi means for reference voltage?

 

[PaloAlto]

I may be wrong, but I'm having the impresion that you are mixing two things:

Caps don't create noise. If you just have an amplifier and a load, the signal power is decreased indeed with C, since the BW increases.

If you are talking about SC circuits, then switches create noise because of their R, and this is integrated in the C, together with signal. The difference is that BW is not set with this C, it is just transfering charge. Therefore, since noise is KT/C, larger C means same signal power but smaller noise power, so more SNR.