Bearing vibration signs - fundamental and harmonics frequencies

[MarcelloSilverio]

I'm working with monitoring of bearing vibrations and I need to specify a A/D conversor for aquisition data. I need to know the total amplitute of the frequency harmonics for a fundamental frequency about 1 kHz (like ball bearings).
Thanks
Marcello

 

[hechengli]

so, what are your questions?
if you want to experiment, you have to mearsure the response of the system to some kind of input, say, impuse input.
then doing fft, you can find the magnitude of frequency response.

 

[MarcelloSilverio]

Thank's for the answer. Sorry for my english. I was talking about the total frequency band for the measure, becouse I'm specify the A/D conversor, the stage after the acelerometer signal conditioning. In my case, the bearing used is SKF 2309 (number of balls: 12, ball diameters: 15,87 mm, diametral pitch 71,89 mm), rotating at 4800 rpm. I'm looking for any near experience becouse I can't do a preview test to measure. In another experience I found a 10 kHz as a total frequency band to mesasure. All of that harmonics were in this frequency band. Do you have some case like this or know some literature related?
Thanks,
Marcello

 

[MikeyP]

For decent time resolution, I would recommend a minimum of eight time points at the maximum frequency of interest. Whether you consider the 3rd 5th or even 7th harmonic frequency to be the maximum frequency of interest is up to you.

M

 

[ronfrend]

The fundamental frequencies you will get from this bearing at this speed are as follows:

cage: 31.52 Hz
ball spin: 172.96 Hz
outer race: 377.92 Hz
inner race: 582.087 Hz

You should be looking to at least the fourth harmonic of the highest fundamental so I would set the Fmax to an absolute minimum of 2kHz in FFT acceleration or velocity. However there will be a lot more information further up the frequency scale but it will depend on how you mount the accelerometer and what type of accelerometer you are using.

A typical low cost accel will have a natural frequency of between 10 and 20 kHz - and that is only when perfectly mounted on a good flat surface and tightened to the correct torque. If you are using hand-held you should not look at any data above 1kHz. A good 2 bar magnet will give a natural frequency of about 2 to 3 kHz (varies a lot). A "real-world" permanently mounted accel will have a natural frequency of about 5-10 kHz (again depending on how good the installation is).

I would suggest that you permanently mount the accelerometer and take a lot of care when carrying out the installation. Make sure you have a good flat surface and that the threaded hole to accept the accel stud is absolutely perpendicular.

Now you can set your Fmax at 5kHz for FFT (remembering that the sampling rate is about twice the FFT Fmax). I would then record my FFT with 800 lines to get reasonable resolution. I would also take a time domain reading at 10kHz with enough samples to record about 3 revs of the shaft. 1024 samples would give a sampling rate of 10.6kHz. This will let you see the individual impacts from the deteriorating bearing.

If you are looking at a centrifugal compressor you should also be thinking about the gear meshing frequencies - but that is another story.



Ron Frend
tel: 011 44 1253 400541
ron.frend@predicon.net

http://www.predicon.net