harmonic

[kancil]

Can anyone out there explain for me why the bearing defect frequency and its harmonics amplitude decreased with frequency. Thanks.

 

[bulldawg]

Kancil,

What I think you are asking is why amplitudes of bearing defects decrease with a decrease in machine speed?

There is a rule of thumb for that when the speed of a machine is doubled the vibration squares. The inverse would thus hold true

 

[GregLocock]

I think he's actually asking why at a given speed the amplitude of the harmonics tend to reduce with increasing frequency.

Well, firstly it depends what you are measuring, if you keep on differentiating then you can change that round. However, if we limit ourselves to velocity and acceleration it is generally true.

I'd answer it by saying that the response to an impact can be calculated as the sum of the modes at that point (modal superposition), and that by its nature a low frequency mode represents a bigger proportion of the signal than HF modes. But I'm not wildly happy about that.

Another observation that I don't necessarily agree with is that damping of high frequency modes tends to be greater than for low frequency modes.

Probably the real reason this is hard to answer is that it is a general trend, rather than a truth, and so any theory can be knocked down by referring to a specific case.





Cheers

Greg Locock

 

[vanstoja]

A decreasing amplitude spectrum of a noise/vibration fundamental source and its harmonics is generally a sign of fluidborne vibratory source (waterborne or airborne) that either originates in the fluid continuum or somehow couples into the fluid continuum. Impeller blade passing noise is a example where fluidborne (pressure transducer spectra) broadband levels drop off independently of the signal to noise ratio of the individual harmonic peaks whose S/N levels depend on impulsiveness of rotating blade-stationary vane interations producing lift fluctuations on blade or vane airfoils. Structureborne noise/vibration, on the other hand has a generally increasing broadband level with increasing frequency. Sometimes you can have non-fluidic sources coupling into a fluid continuum such as electromagnetic line frequency peaks in water cooled motors which appear to have a real affinity for coupling into and travelling in the cooling water.

 

[electricpete]

Good comments all.

One thing that should be thrown into the mix of course is a consideration of which unit we are talking about: displacement, velocity or acceleration.

The exact same signal can be described as either
decreasing with frequency (in displacement)
constant with frequency (in velocity)
OR
increasing with frequency (in accelration)

 

[micjk]

Agreeing with electricpete good comments from all ( good point on units pete). Another way to envision this ( and correct me if I am wrong) is that as bearing degradation progresses the higher frequency fault indications will diminish and become more like "broadband noise" in appearance. These fault frequencies will migrate towards the lower end of the spectrum. The wave form will show impacting with increasing force and then may actually go from a very defined pattern, impact wise, to a "garbled" less force intensive appearance as more faults ( BPFO, BPFI. BSF, FTF) come into the picture and the original "bad apple" begins changing its geometry (smoothing out) appearing as if the problem is "healing" itself, when in fact catastophic failure is "just around the corner".
Regards,
MICJK

 

[electricpete]

were you looking for anything different than your previous question (which seemed to focus on demod measurements):

thread384-34375

Addressing normal spectrum (not demod), I'll repeat my statement that the linear model (as interpretted by electricpete) predicts the spectrum of periodic impacts at a fault frequency is the product of three other spectra:
#1 - Fourier transform of a single impact measured as force at the point of impact
#2 - Transfer function of the system reflecting attenuation of certain frequencies as they travel through the damped/mass spring system from point of impact to transducer
#3 (obvious) - A function which is 1 at harmonics of fault frequency and zero everywhere else.


#1 and #2 will both decrease toward zero as frequency approaches infinity for any real-world signals and systems. For #1 no signal has infinite frequency content unless it has zero rise time. For #2 I guess I was under the impression that most real-world linear systems have limited bandwidth and will provide increasing attenuation as frequency approaches infinity.

I think vanstoja was addressing some kind of attenuation as well in his comments about waterborn vibs. I didn't quite understand the comments about structureborne noise.

Sorry to ramble.

 

[electricpete]

One other minor point: If I view my system #2 as simple mass spring system, then "increasing" or "decreasing" depends on which part of the spectrum I'm looking at (below or above resonance). Far below resonance it may look like increasing with frequency. Above resonance it will always look like decrasing with frequency.