I am trying to measure the PSD(Power Spectral Density) of a signal, but don't understand why the results are so different when different windows are used.
Eg., for a signal y=2*sin(2*pi*50t), sampled at 1kHz for 20 seconds (simulated in software), I use NI's "SVFA Power Spectral Density.vi" from theN&V Suite to measure. Singal are broken into 10 segments of 2 seconds each and sent to the NI PSD module, calculating useing RMS averang, linear, number of averages=10.
When I choose "None" for the window selection, I got 4 for the max PSD amplitude, which is totally reasonable for the original signal with amplitude 2. However, if I choose "Hanning" window, the PSD max amp becomes 2.6. If I choose "Flattop", the PSD max becomes 1.06. I assume choosing different window should not have such a drastic effect on the PSD, since usually in FFT the window selection will not affect the result this much.
So, what's the reason for such big difference? Is this what PSD is supposed to be like, or there is some problem in my vi program?
Thanks for helping!
