Hello all!,
After you have configured the pump and motor in CSI software, and we are now using the feature "fault frequencies";
lets say there is a ball bearing problem, are the peaks in the spectrum supposed to align itself PERFECTLY with the fault frequencies that the software exhibits, to say for certain that there is a problem?,I hardly ever get them to be aligned exactly, what accounts for this? and what does this mean.
Many thanks!

Several things can cause this.
Wrong turning speed
Wrong bearing used for fault frequeny
Bearing geometry has changed due to wear or some other reason
FFT is only a "close" mathamatical representation of the true signal which could make the fault off slightly

Pattern reconizition is a better judge of a bearing race defect. IMO.

It's been a few years since I used RBMware, but as I recall even for a single manufacturer, you may have several different versions of the same bearing in the database. I.e. one may have an extra roller. This is usually addressed with different suffix letters on the bearing, but how are you supposed to know what version is actually in your machine. After all, they all fit on the same shaft. Then if you consider that it might be a different manufacturers bearing, all the geometry can be different again.

Plus all the stuff that Ralph said.

We are using RBMware and have never seen perfect fault frequncy alignment. To improve the alignment you need to put the cursor on the exact speed and save this value as the RPM. As the speed varies slightly with load you will need to do this for each measurement. Even then the alignment will not be perfect because there may be a small amount of skidding in the bearing as well as the ther problem mentioned already.

As has been said pattern recognition is the best way along with time waveform particularly for inner race defects which give impacts at shaft speed spacing. See attached PDF.

Regards

MDE

4AID.pdf (107 Kb, 24 downloads)

MDE,

Never?

I agree that there usually isn't a perfect match, but I have seen plenty of examples where there was an almost perfect match. (I only consider my wife to be a perfect match and that is only for me.)

Rather than correcting the RPM for each reading base on the apparent running speed from each spectrum, I usually correct it based on a low fmax reading. That is unless there is some speed reference that is constantly changing. The increased resolution gives a more accurate running speed indication.

Anyway, like others have implied-if it walks like a duck and quacks like a duck, it is probably a duck.