To All,

The question is spelled out in the attachment.
Please pay attention that within the data block not all signals are periodic.
Thanks,
Dave

Roller_defect.doc (31 Kb, 83 downloads)

Series of BSF harmonics with sidebands of FTF.
Or more commonly series of 2*BSF harmonics with sidebands of FTF.

The time signals may not be strictly periodic until you reach a time period which is an integral number of both periods (1/FTF and 1/BSF).. which perhaps may never occur. But the longer the duration of the sample, the more the true character of the signal shows up in the spectrum. The true character of the signal includes a sinusoidal modulation at FTF of a signal that is periodic in BSF (or 2*BSF) and gives the spectrum described.

Is my computer messed up or is this the way you displayed the doc?

It wasn't garbled when I viewed it. Must be just your computer. Try downloading it and opening it in word.

Thanks Pete,

I don't seem to have an option for downloading it. It just opens when I click on it. Never seen a doc file on this forum do this. How do I download it?

I agree with Pete. In addition the new attachement offers an explanation (my version) and some conclusions for discussion.

Roller_defect2.doc (46 Kb, 36 downloads)

Ralph,

Did you try Right Clicking on file name and selecting "Save As"?

Walt

Hi Walt,

I did right click but it did not give me the option to save as. Hmmmm!!!!

Wonder what is going on. I'll try it on my laptop, maybe something is wrong with my desktop.

Nope! Same thing. Must be something to do with my service provider or something. Hard to believe 2 computers would do the same thing.

I don't recall ever having more than two FTF sidebands. Had anyone seen that? I guess if it did occur it's indication the modulation is nonsinusoidal (like your square wave window), but what would be the reason for that?

More harmonics of BSF generally means a sharper peak (higher true peak acceleration).

I agree with what you said about severity. Based in part on what I have learned from people like yourself here in the past, I would judge a bearing severity based on combination of perhaps three factors:
1 – High true peak acceleration. Roughly indicates the sharpness of the worst peak. High true peak normally equals lots of BSF (or 2*BSF) harmonics.
2 – Indication of diverse defects: ball, outer, inner, lots of sidebands. (although the FTF sidebands aroud 2*BSF are expected and don't say much).
3 – Worsening trend.
And of course that list goes on (I'm sure someone will add to it).

I wanted to return to one mathematical aspect that may or may not be interesting (depends on your interests). You showed a time waveform that looked very periodic. But it could just as easily be that we have two big peaks during one time through the load zone, and then one bigger peak plus two smaller peaks the next time through the load zone... changing a little bit each time through the load zone since those three impacts don't line up at the same time within the load zone "window" each subsequent revolution of the cage. So the signal is not periodic. So perhaps it seems counterintuitive that it can be represented by a fairly simple family of discrete peaks, each of which represents a periodic sinusoid. It leads to another related question: Can I add together two periodic signals and end up with a non-periodic signal? The answer is............. yes, definitiely. Attached is a plot of y(t) = sin(t) + sin(sqrt(2)*t). It is the sum of two simple periodic signals (and would show as two peaks in the spectrum). But the sum itself is not periodic. If you plotted that function y(t)forever, it would never repeat. A prerequisite for this type of behavior is that the ratio of the component frequencies is an irrational number. That would be the case for BSF and sidebands at BSF+/-FTF, so we also expect that time waveform will not be periodic. On the other hand, if you had a simple series of components with frequencies 1F, 2F, 3F etc... the frequencies ratios are rational numbers and the resulting sum is periodic.

btw Ralph - maybe there is some kind of internet security setting designed to block downloads hiding within your internet explorer security options tab. Or else mabye it's just a Mississippi thing

This message has been edited. Last edited by: electricpete,

SumOfTwoPeriodicsIsNonperiodic.xls (687 Kb, 24 downloads)

Pete hit the nail over the had and confirmed the point I was trying to make in my example, namely, FTF sidebandes around BSF are expected. Their presence do not signify any problems with the cage.
It is likely that number of sidebands correlate in some manner to number of defective rollers.

Concerning non-periodic TWFs produced in a way Pete has desctribed it.... Yes, it is a likely real life scenario and FFT interprets it with infinite number of harmonics, but if non-periodicity is insignificant the fundamental frequency will be standing out. In some rare (academic) cases where frequency is an irrational number, FFT will even produce finite number of harmonics (two, in Pete's example)