Raised (numerical) noisefloor FFT between sine/square

Dear all,

I have taken the FFT of a sine wave and the spectrum shows a nice, single tone at the expected frequency (32MHz) with the numerical noise floor going down as low as -250dB.

It has to be noted that this sine wave was output from a Simulink model/simulation (structure with value/time points).

Next, I wanted to model the clipping of this sine wave (or actually the conversion to digital levels). The sine wave has an amplitude of 1 and is centered around 0.

Thus, I used the statement: output = round(input) which turns my sine wave into square. When I now take the FFT of this square, I can see the harmonics (which makes sense) but I also notice that the 'noise floor' has risen from -250dB to -100dB.

I don't understand this from a signal processing (FFT) point of view, so my guess is this is related to the Simulink solver's accuracy (which was the same for both simulations)?

Has anyone any insight on this?

Thanks in advance!

Peter Start a New Thread