Dear Michelangelo,

You are right, you can obtain the peak-to-peak value by multiplying the VRMS value by 2,828.

Regards.

Hi Michel. So you don't see any problem in converting the mVrms values from the PSD in mV peak to peak, right?

Thanks.

The conversion between VRMS to V peak-to-peak is right.

However, in your question, you said: "...in a power spectrum with y units in mV2 (squared), coming from the FFT analysis...", the PSD presents how the Power of the signal is distributed over frequencies. If you use the values from the PSD you will be dealing with Power,  from such values you can calculate the total Power in a given frequency band.

How could I calculate the total power for each band? By integration? Which units it will have? Thanks again.

You are right, this will be an integral of the PSD values in a frequency interval, you can look at the section Power spectral density in the following link, see the equation Pbandlimited.

https://en.wikipedia.org/wiki/Spectral_density

Sorry to interrupt, but I'm not sure about the "mV2" resp. "RMS values" statement.

I'd rather expect the FFT to result in "amplitude values". Tzis is no big issue but the factor would be 2 then. And you will have to adapt the formulas for "power" accordingly.

If you want to validate: perform the operation for a pure sine input and decide then.

Regards

Sure you can make some tests using a synthetic signal with different frequencies and evaluate your algorithm for estimating the Power on a specified band.

Thanks U. Dreher for completing the discussion.

If you need more information you can take a look at the MIT Course on Signals and Systems from Prof. Alan V. Oppenheim, it is on the link below.

https://ocw.mit.edu/resources/res-6-007-signals-and-systems-spring-2011/lecture-notes/

Thank you very much both for your time.