Acquiring different frequency bands

[AnibalM]

Hi All,

I need to measure vibration in velocity (till 500 Hz) and acceleration (till 5kHz). I wonder how this could be done with only one sampling frequency.

Tkx.

[GerdW]

Hi Anibal,

Nyquist says: use a sampling frequency of atleast 10kHz…

[mtwar]

Hello,

and practice says use at least 4 times the highest frequency - just in case 😉

If the measured frequency is higher than half of the sampling frequency you will see aliasing. Wikipedia has great article about this phenomenon:

https://en.wikipedia.org/wiki/Aliasing

Marcin

[AnibalM]

Marcin,

Sorry, I probably didn't express myself correctly.

I have one signal, this is beeing split to calculate a 5 kHz band FFT (acceleration vibration) and a 500 Hz band FFT (integration and then FFT, velocity vibration). I suppose I have to sample it very fast to obey Nyquist criterion, say 20 kHz. This is OK to acceleration FFT.

But how do I downsample it to get a smaller sample rate to calculate the 500 Hz band FFT ?

Do I need to go through decimation ?

[GerdW]

Hi Anibal,

why do you need two FFT operations? One FFT will give you all frequencies from 0 to half of sampling rate!

What is a "band FFT"? Can you attach an example VI?

How do you "split" one signal?

[AnibalM]

Hi GerdW,

It is a gear box. Important gearing frequencies are expected at 3 kHz. Also, important low frequencies are expected in the range of 3 to 1000 Hz.

The gear box can only run for minutes so I think it is not possible to change the frequency band and get other spectra. It has to be done once. Before calculating FFT we monitor vibration levels (velocity)  to see if the machine is OK.

We need to measure acceleration, envelope and velocity FFT's for machine evaluation. Accelerometer signals will be integrated and treated to get those results.

So we need a FFT for high frequency and another one for low frequency because of the resolution required to see the running speeds and deffects.

See attached file.

Regards,

AnibalM