# Is it right to talk about dynamic range on frequency domain?

I'm comparing three frequency response's plots from simulations and experiments and I observed that two of them look like clamped to a DC component, where the max dB value is i.e. 100 and the min is 40 dB, while on the third one the max is i.e. 120 dB and the min is 10 dB. I want to use the correct term to talk about these intervals.

I've used the term dynamic range to express an amplitude variation interval on time domain, but is it right to use it on frequency domain too?

• so you would like to say that the first has a "spectral dynamic range" of 60 dB and the latter has a spectral dynamic range of 110 dB? what might you want to do about holes in the spectrum? like what if the output is from a notch filter and there are some frequencies where it's $-\infty$ dB level? what would the dynamic range be for that? – robert bristow-johnson Oct 22 '18 at 23:36

the dynamic range is the difference between the largest and the smalles value of a quantity can asume, so yes, you can have a dynamic range for the magnitud of your frequency components, I'm not aware if there's another terminology when dealing with the frequency domain. One thing to consider is not only the dynamic range of your signals but also the dynamic range of the variables you'rr using to store it and to calculate the frequency response, altough in the frequency domain (and specially when working with dB) it's not a big issue.

From a power spectral point of view, the dynamic range can indicate a ratio of signal power to noise power, hence, can be a useful metric similar to SNR. It will be useful to consider when detecting presence of periodic, sinsoidal components burried in noise.

Additive white noise on a given time domain signal will be reflected in the spectral floor of the respective PSDs. And this floor will be limiting the dynamic range per signal power level, and is a direct indicator of detectability.