PSD magnitudes

[ella96]

Hi,

We are trying to study the PSD of a sine signal. We are finding that the magnitude  of the output of the PSD VI does not correspond to the value given by the math equation "A²/4" at the frequency of the sine (and A being the amplitude of the sine).

Moreover the magnitude of the output seems dependent on the number of points generated by the sine wave VI.

Can someone help us understand where this problem is coming from?

Thanks

Ella

[ella96]

Hello dougm,

We are having the same problem as you and we  are wondering if you managed to solve it?

Not only does the amplitude of our PSD signal change when we generate a sine with more points, but also it is never correct (the amplitude should be "A²/4", with A the amplitude of the sine.

Thanks for your help,

[mcduff]

I have not used the PSD feature from the Sound & Vibration toolkit, but the help for it says:

SVFA Power Spectral Density VI

Computes the power spectral density of scaled signal [EU] and returns the result in the selected units.

You need to use the SVL Scale Voltage to EU VI before the PSD VI and enter the correct information in the channel info control.

mcduff

[dougm]

I never did get a satisfactory answer to my issue.

If you figure it out, please share it with the group here.

[Bob_Schor]

Did you try the (helpful?) suggestion from mcduff?  Did it make any difference?  I suspect most of us have not used the Sound and Vibration Module (I certainly have not, though I have done FFTs and other Spectral stuff using "regular LabVIEW).  What happens when you use Transforms from the Signal Processing Palette?  Are those values more "understandable"?

Bob Schor

[ella96]

Hi,

We seem to have a similar problem. We would like to generate the PSD of a sine signal. We notice that the frequency of the peak of the PSD seems correct but the value of the amplitude of this peak is far from what we expect for a sine. We notice that this amplitude changes when we change the number of samples of our sine.

We have also tried simulating a PSD by doing the autocorrelation + FFT on the sine and we have the same problem.

Please could you help us find a way to have a correct amplitude?

Thanks a lot

Ella

[ella96]

Hi,

No we tried mcduff's suggestion but it didn't change anything. The problem is not linked to the units of the amplitude but the way it varies depending on the amount of samples of the sine signal.

We have tried using the autocorrelation and FFT VIs instead but we have the same problem (illustrated in the VI attached).

If you have any suggestions we would be very grateful,

Ella

[mcduff]

The integral of the PSD should give you the total power in the signal. When you increase the number of points while keeping your sampling the same, you increase your time acquisition. This means the frequency bins are smaller. Thus some peaks will increase, some peaks will decrease, but the integral of the spectrum remains constant.

Different number of points different size bins. If you want the same magnitude for each peak irrespective of number of points, assuming same sampling rate, divide the magnitude by the number of points. To get the correct answer when you integrate you will need to add a scaling factor.

mcduff

EDIT: