6/1/16: signal bandwidth is spread across both wideband and 1/f

Some literature says to break the calculation up into the 1/f noise below the corner frequency and the white noise above the corner frequency and add them together. The easier and slightly more accurate way is to add (as root-sum-of-squares) the 1/f noise and the white noise contribution assuming that there is 1/f noise and white noise everywhere, which gives you the following equation:

rms noise = e_n,wideband*√f_H - f_L + f_C*ln(f_H/f_L)

If you start playing with the numbers, you will find that white noise quickly starts to dominate. Most of the literature will tell you to ignore 1/f if the -3dB bandwidth of your circuit is significantly more than 10 times the corner frequency, f_H >> 10*f_C.

You may also be able to do the rms noise calculation faster with a SPICE simulator. See the webcast slides for the full equation.

Some literature says to break the calculation up into the 1/f noise below the corner frequency and the white noise above the corner frequency and add them together. The easier and slightly more accurate way is to add (as root-sum-of-squares) the 1/f noise and the white noise contribution assuming that there is 1/f noise and white noise everywhere, which gives you the following equation:

rms noise = e_n,wideband*√f_H - f_L + f_C*ln(f_H/f_L)

If you start playing with the numbers, you will find that white noise quickly starts to dominate. Most of the literature will tell you to ignore 1/f if the -3dB bandwidth of your circuit is significantly more than 10 times the corner frequency, f_H >> 10*f_C.

You may also be able to do the rms noise calculation faster with a SPICE simulator. See the webcast slides for the full equation.