PSD magnitudes

[dkriparos]

I'm having a problem in trying to determine the RMS value of a waveform (noise in this case) from a PSD plot. I am expecting LabView's "Basic DC/RMS" utility to correlate to a calculated RMS value. The calculated RMS value from a PSD should be ___  Vrms= ((V^2/Hz)*BW)^.5_____ where V^2 is the yaxis value read from the graph and BW is the bin width in Hz. I've included my program (which was copied from a previous discussion (http://forums.ni.com/ni/board/message?board.id=170&message.id=328715&requireLogin=False) and modified slightly). I basically have an unfiltered and filtered waveform of which I would like LabView to give me the RMS values. Any help would be appreciated.

[seVEMes]

What values are you expecting? I don't see how the equation that you mentioned "___  Vrms= ((V^2/Hz)*BW)^.5_____" is not being satisfied. What have you done to verify that the numbers being output on the PSD unfitered and RMS value 2 indicators are not the correct RMS values?

[dkriparos]

For example. Using the PSD Plot (unfiltered) on the right.

1. I would expect the bin width to be the entire display. i.e. ((1e-9)*(500Khz))^.5 but this equals .022Vrms and not the 1.00e-9

So 500Khz is obviously not the bin width.

Where I'm hoping to get is: Using the 2nd (PSD filtered plot) (set it up for bandpass, butterworth, order=30, F lower=1Khz, F upper=10Khz, I would like to be able to predict the RMS value of each part of the spectrum. (in this example approx. 100Hz to 1Khz, 1Khz to 10khz and then 10Khz to the end.

If I'm not being clear can I give you a call to discuss this?

Thanks

dan

[seVEMes]

The Basic Averaged DC-RMS VI does not know the difference between a waveform that is filtered or not. The RMS equation is not unique to LabVIEW and can be found anywhere. It is Vrms = ((1/T)*∫_to^to-T(V^2(t)dt))^1/2. For a sine wave the result of the integral is equal to T*(Vpk^2)/2. This gives a final result of Vrms = Vpk*(1/2)^1/2 for a sine wave.

Looking at your waveform PSD Plot it is clearly centered near 1E-9 which means that the RMS value should be near 1E-9 and it is. The Basic Averaged DC-RMS VI does not know that this is the frequency domain but preassumes that it is in the time domain because that is what it is made to measure. The value I get when I check the RMS of the time signal is near 7E-7 which is below but fairly close to the expected sine wave RMS value for asine wave with a similar amplitude.

[dkriparos]

Thanks Vince,

I'm still not sure how you calculate the RMS value using values as read from the Y axis of the PSD.  I'm assuming this is done using Vrms = ((V^2psd/Hz)*Hzbinwidth)^.5. This should correlate to Vrms from the Basic Dc-RMS VI.

The reason for these questions is that ultimatley, I have to be able to determine the RMS value of a particular frequency band given a PSD plot. This band will most likely be filtered.

Thanks

Dan

[seVEMes]

Well that doesn't sound too hard to program but that is not the same equation used for the DC-RMS VI. It might be a good idea to look at the DC-RMS code as an example to implement your equation. If we do have a VI that implements the equation that you are trying to use it would take searching through the palettes to see if any of them output the value that you expect.

Good Luck.