Why does Express PSD give a different amplitude result than FFT PSD?

[Donners]

Hi,

When I use the express Spectral Measurements VI to plot the Power Spectral Density I get a different result than when I use the FFT PSD VI. The data is the same but the amplitudes are different in the PSD plot. Has anyone come across this?

Thank you,

[johnsold]

Donners,

I have not dug into the VIs or their help files.  Often their are slight differences in scaling or normalization conventions. Also the Express VI may use some windowing not in the othre version. It is quite difficult to troubleshoot something like this from an image.

One thing I noticed is that the horizontal sclaes are different in the graphs.  Is that cosmetic or are there different numbers of points in the two graphs?

Please post your VI with some typical data saved as default.

Lynn

[Donners]

Hi Lynn,

It's not finished but here is what I have already attached,

Thank you,

[Donners]

Sorry this won't let me attach a TDMS file. Either Internet Explorer or Firefox.

[johnsold]

Donners,

You can Zip the file and attach the compressed version.  There is a size limit also.  If it is too big you may need to make a smaller example file.

Lynn

[Donners]

Cheers,

Here there are:

[johnsold · Accepted solution]

Donners,

The PSD2 and PSD2 graphs are different because you are feeding two extra points in the arrays (t0, df). When graphs get input data in arrays, they do not automatically interpret the t0 and df information.  In fact they do not even get that information.  You have to program the scales with property nodes.  If you just plot the magnitudes, the spectra look very much alike.

 

                                           ^^^ This just confuses the graph!!!

The spectral shapes differ slightly because the FFT Power Spectrum and PSD.vi uses a Hanning window by default.

Lynn

[Donners]

Hi Lynn,

I get the Hanning window aspect, and thanks for that, I wouldn't have figured that out alone. I plotted the graph using the magnitude alone and the frequency scale is still different when I plot the magnitude to when I plot the waveform.

There has to be some way to duplicate the x-axis from the upper power spectral density graph to the lower one, which is being saved in a TDMS file.

I tried building an array with the magnitude and the 'active x scale' property but it didn't work.

Sorry about this I should open a new post as you have already answered the subject question of this one.

Thank you,

[johnsold]

Donners,

Whether to start a new thread is a somewhat gray area.  In this case I think you are OK here since I first suggested a property node in this thread.

The property you want is X Scale >> Offset and Multiplier >> Multiplier.  When I wire df to that the x-axis scale chages to 0 .. 600.  Figuring out which property node will do what you want and finding it can be a challenge.  The help files can be useful but sometimes are rather cryptic.

Lynn