Y-axis Units of Power Spectrum

[wisjaf12]

Hello

I am acquiring acceleration from an IEPE accelerometer. Units of the y-axis of the time domain signal are in (g). The time domain signal is then fed into the Spectral Measurements Express VI to get the power spectrum of the signal. In the Spectral Measurements Express VI, I set the results to Linear and the window to (Hanning). My very basic question is: what will be the Units of the y-axis of the power spectrum with linear option and how can convert them back to the unit (g). If you provide an example calculation that would be very appreicated. A shot of my simple code is attached.

Regards

[STASHkal]

Wisjaf,  

It has been a while since I took a physics class.  However, the VI performs an FFT so I think the units of the power spectrum are g^2/hz.  You had better check me on that though.

[wisjaf12]

John-St

Thank you for your reply. The application note '' The Fundamentals of FFT-Based Signal Analysis and Measurements'' attached mentions that the power spectrum is A^2/2 where: A is the peak amplitude of the time domain signal. Based on the, the power units should be g^/2. Am I right? Correcting my comment if I am wrong would be appreciated.

Regards

[STASHkal]

I dont see the attachemnt.

also, can you clarify "g^/2"

[ToeCutter]

Acceleration (g) is not proportional to root(power) so passing it as input to a power spectrum doesn't yield anything meaningful unless I'm missing something. How about using a plain old (unsquared) FFT to get the frequency components of the acceleration?

[wisjaf12]

John-St

I apologize for the couple mistakes in my previous posts. Here is the attachment. Also I meant that the power units based on the attached notes should be acceleration squared divided by two (g^2/2). 

[wisjaf12]

ToeCutter

Thanks for your interest. Because I am really a beginner in signal processing, I dont understand what do you mean by ''using a plain old (unsquared) FFT'' and how to do that. If just know what will be the units of acceleration signal after converting it from time domain to frequency domain that would be great.

[ToeCutter]

No worries.

Type 'FFT' which is a 'Fast Fourier Transform' into the search box on the functions palette and you will find these functions. A 'power' spectrum is just essentially the output of an FFT (which gives you the spectrum of a signal) squared. So you should just be able to use an FFT in place of the power spectrum.

[wisjaf12]

ToeCutter

Can you please attach a very simple example VI. I am originally using the spectrum measurements express VI and I chose the output to be power spectrum. How this can be diffrent that the FFT.

[ToeCutter]

Sorry no can do- I'm off work for the week with no access to LV.