FFT analysis

[azimuddin]

vibration peak value during FFT obtained at 1 x rpm means unbalance or misalignment.
overall vibration (frequency range 10 to 10,000Hz) obtained as 5 mm/s rms. FFT analysis showing only one pick of the same measuring point as 0.5 mm/s rms at 1 x rpm. obviously the magnitude of the vibration obtained at a particular frequency is very insignificant. what shall be the conclusion. can we say that the machine has unbalance or misalignment due to which the overall vibration is 5 mm/s rms.

please any expert can help me with his valuable analysis.

 

[GregLocock]

Your rms signal includes all the noise (from all sorts of sources) from 0-10000 Hz. Try to reduce your frequency range to say two or four times the first order frequency, then it'll make more sense.

Cheers

Greg Locock

 

[GregLocock]

Sorry, also make sure that you have AC coupling switched on to eliminate the DC component.

What speed (or frequency) is your machine running at? Cheers

Greg Locock

 

[VibrationSpecialist]

Dear Azimuddin,
*) Your vibration measurement is in mm/s which means that your out put parameter is velocity. Now if you are using an accelerometer to have this high frequency range (10-10,000 Hz), you need to integrate your acceleration signal to velocity. Now you need to check the type of integration method used in the FFT. Is it digital integration of analog integration?
*) What do you see in the time domain? If the frequency spectrum is showing only one component at 1X then your time waveform should look almost sinusoidal. Is this what you see.
*) You can e-mail me your data if you need further analysis.
Take Care
;-)

 

[electricpete]

No expert here.

But if you have ruled out measurement problems, the problem may be a high noise floor, at least in part of the spectrum. This can sometimes be cavitation of a pump, lubrication problems on a bearing, flow noises, or other random-type noises without a periodic nature.

Also a rolling bearing in final stages of failure starts to exhibit multiple fault frequencies and sidebands which tend to raise the noise floor.

A logarithmic scale can help you look at what frequencies are in the grass within your noise floor. Look at the highest among them and see if you can recognize the frequencies (multiples of running speed, multiples of bearing fault frequencies). Also look for regularly-spaced peaks and identify the spacing between them.