As a general rule I have always been told when setting up for gearbox vibration evaluation you need an Fmax atleast 3.25 x the gear mesh frequency.
My question is what is the actual reasoning behind this Fmax at at least 3x GMF?
I was once told that this is related to tooth engaging, and disengagment on the gears.
Any takers?
Thanks in advance

Robbie,

I believe it is just so you go far out enough to see all the paterns that a gear problem may show itself by. For instance, There may be very little vibration or sidebands at GMF for gear misalignment, but 2X or 3X GMF may be larger with more sideband activity.
Therefore you want your F-max to be out past at least 3X GMF. 3.25 sounds OK to me.

Dave

Robbie,
Also see to that the sensor mounting is either stud or epoxy coupled and the sensor used should cover the range of 3.25 of GMF.

Khalid

I believe the reasoning is that the tooth engagement, power transfer, and tooth disengagement can produce the 3xGMF peak. You add the .25xGMF to capture the upper sidebands.

Every tooth engagement represents a push-pull as the teeth roll through what is termed approach and recess. In these areas there is sliding of the teeth but at the theoretical pitch line there is pure rolling. The push-pull causes a 2X of mesh when the lubrication conditions become marginal. For example, degraded lubricant or a plugged spray nozzle may cause an early on change in the 2X of mesh since the friction characteristics in approach and recess are likely to change.

I've never understood why the need to look out to 3X(+) except to say that what goes on at the mesh is a very nonlinear phenomenon. My experience is 2X(+) is ample, the (+) being sufficient margin to insure you aren't being subject to a filter roll off frequency.

In petro chemical gearing where mesh frequency is typically in the 3 to 5 kHz region, 3X of mesh pushes you out to a 9 to 15 kHx region of interest. These high frequencies bring about another issue of accelerometer mounting methods and accelerometer selection. B&K years ago put forth the rule of thumb that for quantitative measurements (i. e., your concerned about the raw number delivered) your frequency range of interest shouldn't be above 1/5 the mounted resonance. For qualitative measurements the range of interest can be out to 1/3 the mounted resonance. You can readily see that this often pushes mounted resonances out to 50-60 kHz. That can be an expensive accelerometer.

Of course if you are looking a small industrial gearing where mesh frequency is often well below 1000 Hz the considerations of mounting method and mounted resonance are far less critical.

John from PA

John,

Thanks for the explanation of 2 x gmf.

Can you explain what effect load will have on the amplitudes at 1 and 2 x gearmesh frequency and their behavior relative to one another?

Thanks

Hi Guys
hope this helps, this has come from the Rockwell training manual for the VA1 course.
"One of the most common frequencies at which gear wear is 1st detected is at 3 x GMF. This is likely due to the fact that each mesh event includes 3 seperate events:
.A sliding action as one tooth enters mesh with another
.A rolling action as they approach the root of the mating gear
.A sliding action as the teeth disengage
Any interuption to the meshing action can generate frequencies at 3 pulses per mesh or 3 x GMF.
My 2 pennies worth hope it helps explain the 3 x GMF.
Cheers
Perry

That seems to be the explanation that was given to me before in the past.
I was looking for opinions from others who either agree or disagree.
Thanks,

Danny asked "Can you explain what effect load will have on the amplitudes at 1X and 2x gearmesh frequency and their behavior relative to one another?"

This isn't any easy question to answer in a forum like this because of the complexity of the answer being so very dependent on the type of gearing. Let me give you an example of one answer that might apply to a high horsepower high quality hardened and ground single helical gearset as might be used in a high speed gearbox in a refinery.

First of all such a gearbox has rotors that will operate with varying temperatures across the face width of the gear since oil moves across the teeth picking up heat as it goes. The teeth will actually be hotter on one end than the other. Very often you can find evidence of this in the form of discolored paint on the interior of the casing but only one one side. The gearing has a geometry modification introduced at the time of manufacturer to compensate. The modification is based on the full load of the gearset.

Next, the gearing undergoes torsional twist, more in the pinion than the gear but nevertheless a factor that is also accounted for in terms of a geometry modification at time of manufacture. Again, like the temperature effect, it is based on the full load of the gearset.

OK, what happens to that actual tooth - it bends, again as a function of load, and the amount it bends has to be taken into account. This may require another geometry modification.

While on bending, both rotors also undergo bending deflection due to the tangential and separating forces. Again, a modification might be introduced to account for this effect. In a high ratio gearset where the pinion gets quite skinny compared to the gear this modification can be substantial. It is based on the full load of the gearset.

OK, we now run this gearset under no load or light load all painted up with bluing to check tooth contact. Guess what, the contact stinks, it is all over toward one end of the helix. But guess what, it MAY be perfectly correct because we are seeing the effect of all those geometry corrections introduced to compensate for things that occur at full load. Even if we operate the gearset at 1/4 load, we aren't seeing full tempertures, full torsional twist of the gearset, full bending deflections, and thus the contact won't be optimum. You need to know what modifications have been done to that gearing before you judge contact at reduced loads.

Also, the 1X of mesh and 2X of mesh may actually be higher at reduced load than that obtained at full load. And the relative amplitudes between the components will vary as the load changes.

When you look at double helical gearing similar geometry modifications also occur, and now we introduce also the potential of external thrust causing uneven load split between the two helices.

Understand that in high speed gearing the modifications and accuracies are often measured
to millioniths of an inch. The machines that do this are priced as well in the millions of dollars per copy.

In relatively small horsepower industrial gearing where accuracies are not critical, the gearing is relatively soft and a lot of wearing in occurs. I'm speculating here a bit as my expertise is high speed gearing, but I would think that wear would tend to make the approach and recess areas more dominant in the meshing action and would increase the 2X relative to the 1X. Perhaps others with experience may have more to offer based on their actual observations.

John from PA

quote:

Originally posted by John from PA:
Every tooth engagement represents a push-pull as the teeth roll through what is termed approach and recess. In these areas there is sliding of the teeth but at the theoretical pitch line there is pure rolling.

John,

CSI, Technical Associates, James Taylor interpret 2xGMF as misalignment, Vibration Institute - as result of looseness and clearance.

Push-pull theory is new at least to me. Why would it occur within one meshing cycle? It is known that sliding above and below the pitch circle occurs only because of CL distance and involute inperfections, thus causing vibration. If this is true how does it fit into push-pull theory?

Apparently you meant that forcing function is being modulated two times per meshing cycle as teeth approach and recess. If so, why? Do you think misalignment results in this modulation?

David

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

It is my understanding that this could indicate a loose fit of the gear on the shaft.
I've only been doing this 3 years and this is my first post. This forum has been a valuable asset for me.Thanks.

43501_loose_shaft.rtf (131 Kb, 16 downloads)

Hello,
I have been working a lot with slow to medium speed gearing most of which were herringbone. I read Talor and find myself analysing more and more the waveform. I thing we should put a lot more effort analysing gear waveform than spectrum. A spectrum of 3½ times the speed of the pinion’s meshing on every shaft with a waveform that contains 8 completes revolution at the best resolution possible had been sufficient so far for me. Gear misalignment may show various pattern depending on their build and operation and knowing the whole situation of operation and maintenance maybe essential for diagnosis.
My opinion is that we should argue about length and resolution of the waveform not the spectrum.

Best regard, Marcel