I have a frequency in a motor that I don't understand. It is on a motor driving a boiler feed pump and it only shows in the motor so I am sure it is coming from there. I was taking a regular route spectrum on a 3580 rpm motor and noticed a very high reading at 184 Hz. This is at 3.1 orders, please see attached spectrum. I was taking the other measurement points when I noticed the vibration decreased significantly so I did an aquired spectrum with all parameters the same and the amplitude of the 184 Hz peak dropped from .39 in/sec to .17 in/sec. I looked at the history on this motor and the 184 Hz pead was there last month but was on .02 in/sec. I also aquired a spectrum at 200 Hz fmaz with 3200 lines so I could zoom in on it. I did a current signature analysis which had some very minute sidebands at 2 x slip freq but the rotor bar program said no cracked or broken bars. In the high resolution spectrum I show sidebands of .61 Hz and 5.9 Hz. around the 184 Hz peak. This motor is less that one year old and had some resonance issues with the base but stiffening took care of most of that. I also went into monitor mode and watched it for ten minutes and the high vibration never repeated. Any insight into this would be appreciated.

Thank You,
Ronnie Lynn

_3_Boiler_Feed_Pump.pdf (61 Kb, 108 downloads)

I have no idea. Just a few random thoughts that pop into mind.

- 3.1x - BPFO from an 8-ball bearing? Although I suspect these are sleeve and even if not I would think you would notice other symptoms at this level.

- 0.6 hz could be pole pass frequency.

- 6hz I have no idea. Maybe feedwater regulating valves are oscilalting at this frequency?

Does 184hz have harmonics? If not - could be a component resonance which some times gets modulated by periodically varying excitation source amplitude. Resonance frequency amplitude may also drop once excitation amplitude level has dropped.

Pete,
It is ball bearing but I don't know the bearing ID. Waveform doesn't have any impacting and peak is awful clean to suspect bearing frequency. You are right, .61 sidebands are very close to the pole pass frequency. I just don't have the confirmation of rotor bar defects around either 1 x turning speed in the vibration spectrum or 60 Hz in my current analysis.
David,
The 184 Hz peak does have harmonics but they are very low. I am at home now and my laptop is at work so I can't tell you the amplitudes of the harmonics.
One other thing I didn't mention earlier is a pulsating sound in the motor. That is part of the reason I wanted to do current analysis.
I do appreciate the feed back. I am at a loss right now but I will be checking on this one again next week. Any other thoughts or suggestions are more than welcome.

Thanks,
Ronnie Lynn

I agree it sounds like bearing defect is not likely. But as a longshot if you can get AFBMA number off nameplate we could likely rule it out with even more confidence.

You have a peak at 184hz. The peak to left would be in the neighborhood of 178hz i.e. 3x running speed. So maybe the main action is not 3.1x but just a 6hz sideband of 3x. As David says maybe that particular sideband just heavily amplified by resonance.... with that many frequencies you are bound to excite some resonances. Also looking at the way the floor rises a bit around that frequency on HIM also seems to suggest resonance somewhere at that frequency. But doesn't do that on HEM. By the way is HEM outboard?

None of that explains anything, just a different way to look at the 3.1x as possible sideband of 3x.

I really have no idea but can only think of some shotgun stuff to try. If others have some more focused ideas you'd be better off to listen to them.

Is current oscillating at 6hz? If it were you probably would have missed it on your current signature analysis if you have a narrow band around 60hz which would not include oscillation sidebands at 54hz and 66hz. At that very low frequency if current were oscillating you might even be able to pick it up on a panel meter or with handheld clampon ammeter.

Poke around the motor and base checking for anything vibrating at 6hz.

Check pump flow and pressure for oscillations at 6hz (I realize this is the motor but maybe some torque oscillations are passed back to the motor).

Look at the shaft and coupling with a strobe (if you freeze it at 1x you might see something moving around 6 times per second).

With a speed of 3580 I assume it's not a vfd ...is it?

There is a bubble of energy at the base of the frequency in question suggesting resonance amplification.

What does data look like at the other motor bearing housing positions?

Cross channel data would be useful to determine amplitude and phase of the vibration frequency of the motor base, feet, bearing housing centerline position to investigate for resonance. Consider investigating for possible motor mount/support problem.

Pete,
HEM and HIM are the same measurement point. HEM was setup to separate 2 x turning speed from 2x line frequency ( 400 Hz fmax 1600 Lines). I just randomly selected HIM to zoom in with 200 Hz fmax 3200 lines. And no this is not on a VFD.
Ken,
I am going to do more tests next week and I will see what cross channel phase readings will show. I am also going to do some bump tests to see if I can excite the resonance.

Thanks for the input,
Ronnie

Looks like the 6 Hz sidebands have sidebands of .61 Hz of their own and also looks like the 184 Hz peak has 1x sidebands, huh?

Is it possible the 184 Hz peak, having such a high amplitude and such low amplitude harmonics, is an outer race "slide" defect, where the balls are not rolling when they enter the load zone or maybe the outer race is being "squeezed" out of round in the housing and then the sidebands are from the motor having a not so strong casing design?

Ralph,
Good point on the sidebands. There are .61 Hz sidebands around the 6 Hz sideband. And there are 1 x running speed sidebands around the 184 Hz peak. But upon close inspection and raising the floor up, there are 5.07 Hz sidebands around the turning speed sidebands of the 184 Hz peak. Man it is hard to even say that without getting tongue tied. Any thoughts on what would cause that. Also, Ralph, would the scenario you discribe account for the sharp increase in amplitude for a few seconds and then drop to half the level and not repeat at least during the 20 or so minutes I stood by the motor aquiring spectrum and monitoring for changes. I do appreciate all the feed back.

Thanks,
Ronnie

Ronnie,

You have not addressed the rest of the peaks: looks like at around 238hz, 298hz, 358hz. Are they harmonics of 1x? Are they having 5.9hz sidebands? If this is the case then, as Pete has mentioned, it may be related to cyclic torsional load and will be detectable on the motor ONLY. Cyclic variable load will generate SBs around harmonics of 1x.

On the top of this trouble you may have another one: a resonance at 184hz excited by one of the sidebands. This idea was also suggested previously by other members.

Good clarifications requested by Ralph and David - in particular that peak in the neighborhood of 240hz - is it 4x running speed (tilts the diagnosisis towards oscillation) or is it exactly 1x running speed above that 3.1 peak (tilts the diagnosis torwards bearing)? Same question for peaks near 300 and 360hz.

David & Pete,
I have posted 4 more spectrum. The first and third are the same only one has text in it stating harmonics of turning speed. The second shows the sidebands of the three frequencies of interest of David;238, 398, 258, with 238 being marked and the other two frequencies have the same spacing of sidebands. The last spectrum is another aquired spectrum of 600 Hz fmax with 6400 lines and as you can see the harmonics of turning speed are out to 10 orders but everything is below .006 in/sec. With the amplitude of so many of the peaks being so low and the resolution needed to see them, does that take away from the significants of them?

Thanks,
Ronnie

Harmonics.pdf (78 Kb, 37 downloads)

Pete,
Here is one other spectrum from the 600 Hz fmax 6400 Lines showing the 1x sidebands around 184 Hz peak. Hope that is what you were asking about.

Thanks,
Ronnie

1x_sidebands_around_184_Hz.pdf (52 Kb, 33 downloads)

Thanks, that's what I was asking for. That one just to the right of 364.05 looks like 2*184?
It would seem to tilt toward a bearing pattern as Ralph suggested with BPFO (184) and harmonics of BPFO (368) and 1x sidebands around BFPO and harmonics.

Still remains a mystery why it would come and go and why your waveform didn't show impacting though. Thinking about stuff coming and going, maybe you had contaminants being moved around within the bearing? Whatever the explanation if it really is a bearing it sure sounds liek trouble.

If we assume that 184hz is BPFO and considering that the spectral pattern does not have higher BPFO harmonics to speak of, then in conjuction with absence if impacting in the TWF, leads me to believe that you are having a deformed outer race rather the a spalling in the race. I have arrived to the same conclusion as Ralph did (in different way, though). Not sure about "slide" effect.

The 5.9hz issue is still in the air. Electrical current analysis may identify a cyclic process related load at this frequency.

Another way to view all that is as harmonics of 1x having 5.9hz sideband amplitudes greater then that of center frequency. It is feasable, but I don't remember now its interpretation.

Ronnie,

I would like to see a spectrum from 0 to 1500 Hz and a waveform also.
Even better, if you could email me the extracted data from the motor in a small database so I could put it on my screen and display in setups I am used to seeing, . No two people display data the same way.

Pete,

quote:

Still remains a mystery ................ why your waveform didn't show impacting though.

A slide defect usually doesn't show much of an impact because of the smoothness of the "hollowed" out place on the race. It is sort of like a skateboard track a skateboarder uses, smooth in and smooth out.

It actually is not a spalding defect, yet.

Ronnie,

Did you try taking PeakVue data. It will respond to impacting if it is there.

Thanks guys,
Ralph,
I will build a database with one machine in it and send it to you. But it will be next week before I can do that. I am not sure what you mean by extracting it? I think that is in database utilities but never used it.I would need your email address and If you do not want to post it, email me at ronniecl@aol.com. Do you need a specific resolution or is 1600 lines enough?
David,
I did not take Peakvue but will do so next week and post it. I am not going to be back in the office until next week so I can't post any more data other than what has already been collected.
Pete,
I will look at that peak and let you know if it is 2x 184 Hz.

Thanks,
Ronnie

quote:

Ralph,
I will build a database with one machine in it and send it to you. But it will be next week before I can do that. I am not sure what you mean by extracting it? I think that is in database utilities but never used it.I would need your email address and If you do not want to post it, email me at ronniecl@aol.com. Do you need a specific resolution or is 1600 lines enough?

Ronnie,

Did you get my email?

Forgot to answer the 1600 question. Yes 1600 lines are plenty, depending on your Fmax.

I sent some help on the extraction of the machine. It is easier to extract, in a case like this, than build a new database.

Looking forward to the seeing the data.

Thanks Ralph,
I just got your email and forwarded it to my work address. And I sent you a reply but in case you see it here first, thanks.

Thank You,
Ronnie