Problem with FFT

[deepen_solanki]

Hello. I am using LabVIEW 2013 with myRIO.  We acquired Accelerometer data on various positions of the engine and need to convert the time-domain accelerometer values to frequency domain to calculate vibration of the engine. I used the One Time Signal FFT function in signal analysis in DIADEM and i see that  the data above 6000RPM is not included in the FFT. 
While running the VI I had a timer placed in the loop and i have logged the values according to that timer at a freuqnecy of 500Hz. 

Why cant the FFT of RPM values above 6000 be displayed ? 

P.S need urgent help. 

Thank You 🙂

[ThiCop]

Hello,

To be able o better understand the issues you're seeing I have a few preliminary questions:

- Can you share your data set you load in DIAdem and the used VBScripts?
  If you are not using a script, then please share screenshots of the settings you are using for the Analysis function FFT (One Time Signal).

- Can you share the code you used on the myRIO to generate the data sets? (depending on what the data looks like this might be useful.
- General Information:
  Which version of DIAdem are you using?

[deepen_solanki]

Hello.

Sorry for the delayed reply.

Attached below is my data file. I'm taking a FFT of G-Force vs Seconds(my time channel during acquisiton). Then i try to plot the RPM v/s the obtained frequency.

As you will see from my data file, the RPM goes upto 10200 but the RPM vs frequency plot is shown only for RPM <8000. 

[deepen_solanki]

Continued. 

Im using DIADEM 2015 Evaluation version

[Brad_Turpin]

Hi deepen,

Your "Seconds" channel has an average sampling period of 0.014 seconds, which correspons to an average sampling rate of about 72 Hz.  This is why the FFT extends only to about 36 Hz-- that's the Nyquist frequency limit imposed by your choice of sampling rate.  If you want to correlate the instantaneous engine speed with your measured values, you need to perform an order analysis, which you can also find in the ANALYSIS panel.  This will correlate the full range of measured RPM values with your measured acceleration/vibration values.

Brad Turpin

DIAdem Product Support Engineer

National Instruments

[deepen_solanki]

Hello Brad, could you please elaborate on that 'order analysis'. I need to make sure that all the RPM's are covered and all the frequencies are displayed. What optimum settings should I choose ? ( There are three windows, Analysis Method, Parameters, FFT Window) 
I have my acquired time channel(not equidistant), RPM and Signal Channel (G-Force).

Finally I need the engine vibrations. 

[Brad_Turpin · Accepted solution]

Hi deepen,

I think it's easier to explore your data with order analysis using this prepackaged application:

http://www.ni.com/tutorial/3549/en/

I took a peek at your data with it, though, and it looks like there's no correlation between your vibration signal and the engine speed.

Brad Turpin

DIAdem Product Support Engineer

National Instruments

[deepen_solanki]

Thank You so much Brad.. This really helped 🙂 

[deepen_solanki]

If you could help me with one last thing ... Im only getting the frequencies till like 40Hz.. I think it should be higher.. I am attaching my data set, could you please take a look..  

[Brad_Turpin]

Hi deepen,

Nope.  As I said earlier, your "Seconds" channel has an average sampling period of 0.014 seconds, which correspons to an average sampling rate of about 72 Hz.  This is why the FFT extends only to about 36 Hz-- that's the Nyquist frequency limit imposed by your choice of sampling rate.  This is why your frequencies give out near 40 Hz.  If you want your FFT to give you amplitudes at higher frequencies, you'll have to sample faster.  You need at least 2 data points per period.

Brad Turpin

DIAdem Product Support Engineer

National Instruments