Will the amplitude collected by a sensor above it's roll off still be trendable even if it's not accurate?

Not a good idea...

Since your subject relates to accelerometer I'll address the issue with respect to that type of device only. You also say "above its roll off" so you are talking in general about operating the accelerometer above its mounted resonance, probably something in the 10 kHz region or even upwards depending on the transducer specifications. You are then operating the device in a region where mounting becomes very critical. As an example take 12 kHz and 3 g's peak. The displacement is 0.0004 mil, a very small amount.

Bruel & Kjaer once expressed guidelines that you should only use an accelerometer for quantitative measurements (actual number means something) up to 1/5 the mounted resonance and for qualitative measurements (trending) up to 1/3 its mounted resonance.

John from PA

I assumed as much, but wanted to hear it from someone else on the board. The reason I'm asking is due to the fact that I have around six machines with gear meshing frequencies around 13.5 kHz. I'd like to see upwards of 3.25 time this GMF for trending gear wear, but the sensors I have state the frequency response to be .5 to 15 kHz. These sensors are from an older building that we just decommissioned here at my plant and I was considering permanently mounting them to the machine so I wouldn't have to climb all over it to get my data. Guess I need to come up with another plan.....thanks....

In my personnel opinion you are being trapped by old and not applicable information. Years ago, people put forth the idea that 3X of mesh was an early indicator of gear wear and they backed this up with data from paper mill type gearing where gear mesh was typically 1 kHz or less. It is my understanding that Rockwell still has this recommendation in some of their literature, but without mentioning the qualification of how and where this data was obtained.

From a practical standpoint you can't measure out to 3*13.5 kHz without specialized accelerometers and extreme care in mounting. You should also accept the fact that measureing out to 2X of mesh (+ some for roll off) is the general recommendation of most gear manufacturer's and will also give early warning of wear related issues. That in itself can be a challenge with most accelerometers. Lastly, and this is based on experience with thousands of high speed gear units (I was with PGC many years ago) the human ear should never be discounted an an instrument. Very often airborne noise will be one of the first things detected.

John from PA

I purchased a high frequency accelerometer for these type measurements. It will allow me to get the 3.25 harmonics I need. It is made by PCB electronics. 623CXI is the unit id. It has a drop of 3db @ 900k cpm. Be sure you have a good mount.

Billy stated his mesh frequency is 13.5 kHz and he would like to look out to 3X of that or roughly 49.5kHz. The 623 accelerometer falls a bit short at "Frequency Range: (±3dB) 48 to 900000 cpm (0.8 to 15000 Hz)."

Hence my reason for saying from a practical standpoint, in most high speed gearing, you can't "look" out to 3X of mesh without special transducers and very special precautions in mounting.

John from PA.

What about just paying very close attention to sidebands around the primary gearmesh frequency. Take high resolution acceleration spectra using permanently mounted accels and pay close attention to amplitude and numbers of sidebands may work as well.

From a cost standpoint, I think I will permanantly mount these sensor I have left over and just monitor out to the 15kHz that they're good for. It's better than nothing, and I will get to see the first harmonic of GMF....thanks for the input guys...

Can you trend this point with an ultrasonic measurement? Just a thought!

Hooch

I'm not extremely familiar with ultrasound in general. What could be gained? Would you expect to see gear damage develop and be trended from this technology?