Pl. see through the attached spectrum. Near 1X(975 RPM), there is another frequency of 956 RPM. I conclude this is resonance. Two frequencies in the vicinity . One i,e, 1X may be due to unbalance, looseness , or any other thing causing some component to vibrate at its natural freq. 956 RPM as can be clearly seen in the attached spectrum. I have a single channel analyser which I think is not sufficient for bump test. Any opinions?
Regards.

You might be correct, but don't discount the single channel analyzer. You would be surprised what you can learn through a bump test with a little improvising, especially if the analyzer has "pre-trigger" and the typical peak hold.

John from PA

Why would you think a single channel machine is not sufficient for a "bump" test?

Dave

We certainly do our bump tests with single channel analyser. It doesn't give the fancy transfer function analysis for detailed modal analysis, but still pretty effective in looking for resonance.

If the 956 cpm is in fact resonance, it's not likely being excited by unbalance or looseness. It would have to be something that causes broadband excitation such as rub or flow noise.

Another possibility would be that the 956 rpm is a sideband of 1x due to process oscillation or perhaps pole pass frequency oscillation if driven by induction motor with a rotating assymetry.

By the way, aare there any belts or gears or VFD on this machine? What is it driving?

Gentlemen,
I am very close to retirement. I have been preaching for over 30 years for people to use that little known feature on their analyzer (ANY ANALYZER) called a logrithmic scale. Change your amplitude scale to log and you will clearly see the excitation of the resonance resting at the base of those peaks. I can see it in your linear display, but then that's only due to being as old as the hills and having looked at thousands of them.
You might also consider getting a little bit more low end resolution and don't forget to take more averages (32).

One other thing. More resolution will also resolve there are probably sidebands around running speed. Remember that natural frequencies don't just generate peaks in the fft. They need to exciting by some forcing function. If the energy is broadband and low, the result is an broad 'hump'. The sharp narrow peak in the middle of resonance energy will always represent a forcing function.

quote:

Originally posted by electricpete:
We certainly do our bump tests with single channel analyser. It doesn't give the fancy transfer function analysis for detailed modal analysis, but still pretty effective in looking for resonance.

If the 956 cpm is in fact resonance, it's not likely being excited by unbalance or looseness. It would have to be something that causes broadband excitation such as rub or flow noise.

Another possibility would be that the 956 rpm is a sideband of 1x due to process oscillation or perhaps pole pass frequency oscillation if driven by induction motor with a rotating assymetry.

By the way, aare there any belts or gears or VFD on this machine? What is it driving?

There are no belts and it is not a VFD. this pattern is visible only on motor NDE and DE bearing.

quote:

Originally posted by Ron Brook:
Gentlemen,
I am very close to retirement. I have been preaching for over 30 years for people to use that little known feature on their analyzer (ANY ANALYZER) called a logrithmic scale. Change your amplitude scale to log and you will clearly see the excitation of the resonance resting at the base of those peaks. I can see it in your linear display, but then that's only due to being as old as the hills and having looked at thousands of them.
You might also consider getting a little bit more low end resolution and don't forget to take more averages (32).

Logarithmic scale amplitude is attached herewith and shows mounds of energy at base. May be resonance.

quote:

Originally posted by RRS_Dave:
Why would you think a single channel machine is not sufficient for a "bump" test?

Dave

Most of the times, the result is different if we repeat it at the same location. Amplitude is dependent on applied force which is not being measured. I don't have standard hammers for frbump test. There is no measurement of PHASE in single channel analyser. so Bode plot is not available.

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

quote:

Originally posted by John from PA:
You might be correct, but don't discount the single channel analyzer. You would be surprised what you can learn through a bump test with a little improvising, especially if the analyzer has "pre-trigger" and the typical peak hold.

John from PA

My analyser has got peak hold facility but what do you mean by pre-trigger?
Regards.

pre-trigger -- take data before the trigger.

I agree the log plot is helpful. It could be resonance, but it does not scream "resonance" for me in this case. The base of that peak on log scale starts to widen at a factor of 100 below the peak.... that is probably normal leakage due to the finite sample length.
If you zoom-in and tell us your bin width and the type of window, we might be able to estimate the normal expected spectral leakage at 40db down to see if the expected leakage is comparable to the width of your peak.

I would certainly also like to see a higher resolution plot as Ron suggested. Maybe Fmax=10,000 cpm, 3200 lines (would take about 20 seconds) or better yet 6400 lines (would take about 40 seconds). Not only does it give you a better ability to accurately estimate those frequencies (which might provide a clue)... but the higher frequency resolution will push the noise floor down even further and might reveal something about the pattern (maybe additional peaks are part of the pattern... for example if similar spaced peaks appear around the running speed harmonics, that would be new info). Also I guess if that broad peak stays the same width at higher resolution would point toward resonance... if got narrower in frequency with higher resolution than points toward normal leakage. (And also of course bump test is worth trying). And by the way.... from what we have here we can't even really see that what you have is two separate peaks... zoom-in and better frequency resolution would clear up the picture.

I still think it could be process oscillation. The fact that it only shows up on motor bearings may reflect the fact that the motor provides the only means to couple torque oscillations into radial vibrations. The torque supplied on the motor rotor is also seen equal/opposite on the motor stator. If the stator support is assymetric (like most horizontal motors), it can respond to the time-varying component of that torque with a time-varying rocking motion including radial (horizontal) movement. By the way does the peak show up similarly in both H and V? Do we know what this motor is driving?

Also returning to pole pass frquency. If I ASSUME that this is a 6-pole motor running on 50hz, then your sync speed is 1000cpm, slip speed is 25 cpm, pole pass sidebands would be spaced at 150 cpm... would put them at 9850 and 1050... doesn't match. Do you happen to know how close your grid frequency runs to 50hz? I have heard there can be pretty big deviations in some parts of the world.

A current signature analysis might be an interesting piece of info. It will show you exactly what is your line frequency as well as your sideband. If you see sidebands spaced at approx 20 cpm around line frequency in your current spectrum, I think it would narrow down to rotating asymmetry or load oscillation. Rotating asymmetery (including possible rotor bar defect) if that sideband spacing frequency matches pole pass and process oscillation if it doesn't.

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

A really silly question - have you zoomed in to verify these are really two different peaks? Without knowing your frequency resolution and looking at the plot we have, certainly looks like it could be just one peak and you are showing us the frequencies of two adjacent bins?

[QUOTE]Originally posted by electricpete:
I agree the log plot is helpful.

Pete,
I was away for 15 days. So I could not respond. a better resolution spectrum (Fmax=9000CPM with 3200 lines of spectrum is attached herewith.
Also following are the observations regarding solo run of motor(spectrum attached.)
1.For Fmax=9000 CPM and 3200 lines, the frequency band around peak decreases.
2.In case of solo run of motor, waveform is sinusoidal indicating unbalance and overall value is around 4.1 mm/sec rms.During coupled run, it is around 10.2 mm/sec rms.
3.I conclude, it is unbalance without resonance.
Any comments.

Sinusoidal waveform indicating unbalance.