Accuracy of measured frequency using accellerometer 1

[desimontreal]

I was trying to determine if a set or frequencies and amplitude peaks corresponds to rotor out of balance. However, there is a 13% error on one of frequency peak.
Can anyone with more experience in using accellerometers elaborate on frequency measurement error of accellerometers...I would would have thought that the accuracy would be very good.....If the error is too high then the frequency would not be corresponding with the out of balance. I only have frequency spectrum..plot.

 

[SomptingGuy]

I can't see how an accelerometer can have a frequency error, given that it is a time-domain device that has no concept of frequency. I would be looking at the analyser. BTW, how do you know it's 13% out? Maybe the measured frequency is correct but your expectation is wrong.

Check your signal processing parameters too. For good estimate of frequency, use a hanning window. For good estimate of amplitude, use an (amplitude corrected) flattop window.

What is the frequency resolution of the spectrum? Is it comparable withyour 13% perceived error?

 

[GregLocock]

That peak wouldn't happen to be at multiple of your AC power frequency would it?

As SG says, there isn't really a way that an accelerometer can generate the wrong frequency, it is much more likely to be due to setup, analysis or so on.





Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[SomptingGuy]

I once had a colleague who came back from a Japanese trip with a box load of tapes all corrupted by mains.

The other easy mistake to make (and one I've made before) is aliasing. Are you sure your sample rate is high enough and you have a suitable anti-aliasing filter?

 

[sms]

Or your lines of resolution may be too course. What is your Fmax and how many lines did you select?

As has been said, it is not likely the accel. It is very likely the setup. May also be an interaction between the machine speed, and line frequency if you have a ground loop in the set up.

Finally how are you determining machine speed? It may be that if you are comparing the spectrum against a tacho that the tach is wrong...

What kind of machine are you measuring? Where are you placing the transducer? How are you mounting the transducer (mag base? Is the mag base rattling?)



"Why don't you knock it off with them negative waves? Why don't you dig how beautiful it is out here? Why don't you say something righteous and hopeful for a change?" Oddball, "Kelly's Heros" 1970

Please see FAQ731-376 for tips on how to make the best use of the Eng-Tips Forums.

 

[IRstuff]

Another rather silly problem might be if your accel data was sampled at a different frequency than what your canned FFT program expects. This would cause a scaling error in frequency.

TTFN

FAQ731-376

 

[GregLocock]

"a set or frequencies and amplitude peaks corresponds to rotor out of balance. However, there is a 13% error on one of frequency peak"

Implies that there is an extra peak, rather than a wholesale shift in frequencies.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[desimontreal]

Thanks,

Hard to say at this point...the sampling rate is sufficient for the required frequency range but requested to have those check out.
This is on OEM accelerometers on top of the casing at the front bearing position of the turbofan engine. The machine speed is accurate on these FAA approved airline turbofan engines however anything is possible if one of the teeth is not being pick up. Only way to really verify is on engine strip. I guess at the moment, i would request more frequency spectrums to see the behavior of the peaks...to determine if it tracks with RPM. I was curious though, would an out of balance vibration show up not exactly 1x rotor speed if there was no accel error. Im aware some misallignment might show up with higher engine order 1x 2x 3x etc in parrallel misallignment. Physically i can imagine what would cause the higher order frequencies or subharmonics..Can anyone explain ...thanks..

 

[GregLocock]

I'll try again.

An accelerometer can have many faults, but it CANNOT move a frequency peak. It can generate harmonics and so on, but it can't MOVE the peak.

OK?

I agree that getting a proper waterfall or Campbell's diagram is the easiest way to diagnose the problem.





Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[SomptingGuy]

Me too. You can infer almost anything you like from a single spectrum. A colormap/waterfall tells the whole story.

- Steve

 

[sms]

"would an out of balance vibration show up not exactly 1x rotor speed if there was no accel error"

No....

Unbalance drives 1X...

I am still not certain that I picked what is going on. Exactly what is the frequency of the vibraiton you are concerned about? And what is the running speed of the rotors?

I emphasis rotors, as these machines are generally 3 shafts, and depending on load you may be seeing running speed of one of the other rotors.



"Why don't you knock it off with them negative waves? Why don't you dig how beautiful it is out here? Why don't you say something righteous and hopeful for a change?" Oddball, "Kelly's Heros" 1970

Please see FAQ731-376 for tips on how to make the best use of the Eng-Tips Forums.

 

[desimontreal]

This is typical two shaft turbofan engine. The only 3 shaft engine i know are the RB211, and Trent engines. This one is a two spool engine. I have worked on some minor design on the the ITrent though.

The broadband frequency was not accurately described once the frequency spectrum was available, there were 4 high peaks at
1x+13% N1, 1.8x NI, 1,21 x N2 and high frequency peaks at about 3.5 NI .....was not able to get tracking or waterfall diagram to see how the peaked varied with RPM. Due to lack of time ...the whole fan was replaced and vibration was eliminated. These left the problems out in the air. The high peak in the order of 1.6 inch/sec is very high and i would think definitely related to rotor unbalance,misallignments, skew etc. The fan blades had mid-span snubbers and if worn uneven all around i can only say its possible wave pattern ie nodal diameter frequency response. Or perhaps just the blade leaning unevenly causing unbalance. Other potential is skew of bladed disc splined attachment to shaft if spline pilot diameters are worn.

Thank you for the response this forum definitely have a lot to offer and build our knowledge. I have been out of the field for awhile due to accident and only began getting back into things..I appreciate all your response. If i find out more i will share the results.

 

[Rob45]

The *only* time I've seen an error in the frequency analysis it was due to an error in the analyzer itself.

Accelerometers can have all sorts of errors, but as Greg says, frequency is not one of them.

 

[desimontreal]

Thank you for additional information. I was not100 percent certain if that was the case especially when you have to use frequency spectrum to determine engine vibration sources. Theory and practice are sometime not exact...