Hi~
I simulate noise of a fully differential OPA switch cap structure using spectre  Pnoise .
I have some problems
First i use PSS and Pnoise analysis , then i want to see input referred noise , and it is impractical small , just about 20nV/Hz (better than IEEE papers so imust some where wrong.......)
my clock  2MHz so i set PSS maxacfreq=240M and Pnoise max sideband =100 , i think it is enough.
i had read  Simulating switched-capacitor filters with SpectreRF , so i think other settlings is correct.
input reffered noise = (output noise)/(Transfer function)  is right? i see output noise , input noise,transfer function of Pnoise i think the former statement is right.
and transfer fuction of Pnoise is the same as PAC analysis.
So where i get wrong? How to measure right input referred noise?

Second i also using chopper to eliminate 1/f noise(chopper clock is 1MHz).
The same simulation step like above.
But the ttransfer function is wrong , it very small (minus dB)
So the input refered noise is larger than output noise.
My circuit is correct.
could some body give me some suggestions about it ~~~
thanks~~

Dear Ken, could you please explain why for a chopper, we should observe sideband 0? Thanks!

Hi,

20nV/sqrt(Hz) does not sound to me like ridicously small. State of the art chopper amplifiers show about 8nv/sqrt(Hz).
The relationship between input and ouput noise is (for a linear system)
So(f)=|H(f)|^2*Si(f)

where Si and So are the power spectral densities for both input and output noise respectively.


Tosei.