Just ran into another pattern of cage fault and cage is always scary. This time it is impactive. Amplitude high in PeakVue. Normal spectrum that I took with high resolution also shows cage frequency. it has been this way at least for 6 months.

How bad is it? Could cage be fractured in your opinion? This is a 250HP critical motor.

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

Cage_mill4_motor.ppt (86 Kb, 109 downloads)

This is the pattern I would expect for a cracked cage. My concern is you are seeing a impact like event occuring at 0.41 orders. The 6319 is (I believe) a deep groove ball type bearing. In the fault frequency table, the SKF 6319 bearing is an eight roller bearing with a cage frequency of 0.387 orders. If one assumes the speed is slightly off and the impacting is at cage, then the motor would be running well in excess of 1200 RPM, (Is this possible)? Is there something else which could be introducing this periodic activity at 0.41 orders?

If it is postulated that this is cage, then I would suggest monitoring 'frequently' to avoid running when the cage does 'break' and permits the rollers to come together (would no longer observe the periodic activity at cage when this occurs).

Dr. Robinson,

Thanks for your reply. The motor can't run above 1200RPM.

Sorry, there was an error in the DB. For this Toshiba motor the IB ( the plate says LS ) bearing is NU322C3 (cyl. roller).
FAG NU322 with 14 rollers FTF=0.409
SKF NU322E with 14 rollers FTF=0.404
SKF NU322 with 13 rollers FTF=0.409

So, it looks like it is truly a cage fault !!!

Surprizingly, this is an 1 year old motor which was sitting under an opened shed without being rotated for an year. But a crack, I think, would rather develop as a result of internal misalignment. Could it be a factory installation defect?

The OB ( the plate says OS) bearing is #6318
( ball bearing) with FTF=0.387

David

David,
We have seen similar data here at our plant before. What we were finding out, is that the bearings were not being greased properly. Our previous regime in their infinite intelligence did away with our lubricators and left the lubrication in the hands of operations. I can't explain why I was not seeing haystacking the velocity data or elevated peakvues, but I was seeing alot of cage frequecies with high waveform amplitudes. I would monitor this in the live mode as well as listen with my headphones while I greased it and would be able to see the cage frequencies "magically" disappear. This may or may not be the case in your instance, but it may be worth checking out.

Scott

Thanks for the clarification on which bearing is in the motor. The PV spectra agrees with a 14 roller postulate. What I have observed for cases where the beqaring was installed in a bind (race 'cocked' relative to either shaft or housing) has been a sinusoidal shape in the autocorrelation data (postulated to be the rollers rubbing the inner race with varying intensity repeating sat cage frequency). I do not see this pattern in your case. Your correlation factor is rather high (around 0.5) suggesting the waveform has a reasonable small component from random sources (about 30%).

Your g-level in the PV waveform is on the high side (10.5 g's) for this type fault. Has it been clearly trending up since you started seeing it 6 months ago or has it been stable?

With a correct bearing applied here is more data. Discussion follows at the end.

David

cage_mill4_dual_fault.ppt (242 Kb, 38 downloads)

I don't have much experience with cage viv symtpoms or cage defect. And the patterns expected are puzzling to me in general. I have seen low-magnitude fundamental FTF without the harmonics a few times and I think I heard on the board a few others have seen the same thing. Don't know what could cause that.

Today I we finished tearing down the motor described in "monitoring a degraded bearing". The bearing had a lot of problems, most notably a severe inner race spall. The phenolic cage was alos completely broken at one point. But since the rest of the cage was intact, it still seemed to perform it's function... balls were regularly spaced and relatively undamaged. The spectrum had shown no hint of cage frequencies.

btw - it's not a coincidence that BFPO/FTF=14.

quote:

Originally posted by electricpete:
btw - it's not a coincidence that BFPO/FTF=14.

That is absolutely correct, BPFO / FTF = N

This known fact somehow slipped off my mind. But in presence of FTF it makes it hard ( although feasable) to discern BPFO from the spectrum.

Thanks for the additional data. If you expand the autocorrelation data, I believe the higher rate of impacting will occur at 14 times per cage event. I think the fault being experienced is most probably a defect in the outer race (such as that introduced by the motor remaining stationary a long time). This defect would have 14 rollers passing over it per cage revolution (since the BPFO fault is 14 times the FTF fault), If a specific roller then impacts the cage harder than the others each cage revolution, then the pattern you are seeing would be seen. Your g-level has not increased in the past six months, hence there are no indication the fault is progressing in severity at a high rate. Given you are seeing the cage in the velocity spectrum, it does suggest the fault is not 'minor'.