Noise Fourrier Transform (FFT)
 

Is the following correct?

IF
[list=1][*]an op-amp has a listed voltage noise of 3 nV/rqrtHz[*]used with a gain of 100[*]the conditions are such that the voltage noise dominates[*]further ignore 1/f and just consider wideband white noise[*]I take the FFT of the output noise[/list=1]

THEN

the noise floor should be observed at 3nV * 100 = 300nV = -130 dB

AND

the rms voltage is equal to 3nV* sqrt(bandwidth)

there is an additional BW effect in the fft definition, if the bin size is ~ 1 Hz then your estimate is roughly correct - figure bin size from time duration of data set/#points in fft

also be aware that various math packages use different normalizations; you may see 1/n or sqrt(n) used

input refered noise is en*sqrt(bw) output noise will be 100x larger

it is quite an accomplishment to totaly remove power line fundamental and harmonics at these levels