Odd Observation in Vibration Response Data

[tc7]

I'm looking for a possible or plausible explanation for the following occurence:
Conducting Mil-Std-167 exploratory sine test on an equipment rack (the UUT is not important to the question). As we progress to each frequency of the lower test range (4-15 Hz) with table displacement of 0.020-inch (double displacement, vertical direction) for each frequency we see an exponentially increasing response on the test points mounted inside the cabinet. So far, so good - no problem.
When we move to the next phase of the test, we are to increase amplitude to 0.060-inch double displacment and progress through the same frequencies and we see at one and only one frequncy (12 Hz) a severe drop off in response in one of the accels. However at frequncies immediately on either side of this 12 Hz we see the expected expontially increasing response (of course somewhat higher responses than at .002 DA input).

We see no other anomolies on the particular accel in any other direction or at any other frequency. It is hard for me to believe that the response measured is actually changing from amplification to attenuation and back to a higher amplification within a band of two Hz.
Can anyone think of an explanation for this ? By the way, we cannot repeat the test or substitue another accel.

Thanks for any thoughts.
Best regards,
Tom

 

[GregLocock]

Were all your other responses exactly three times as high in the 0.06 test compared with 0.02? (ie is the system linear on displacement)



Cheers

Greg Locock

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

 

[cincibcats]

I suppose it is possible if a mode at 12Hz were sufficiently lightly damped and you happened to place an accelerometer directly at a node of that mode you could see a response similar to this. Although it is odd that you only see this response at a higher amplitude, which would point to a non-linearity as Greg was pointing to.

 

[Tmoose]

http://aemes.mae.ufl.edu/~uhk/HARDSPRING2.jpg

 

[tc7]

Greg-
I had a typo in my initial question - the anomoly is occuring at 14 Hz, not 12 Hz as stated.

However, it looks like the data is near (but not perfect) multiples of three times as high in the 0.06 test compared with 0.02 tests ONLY up to 13 Hz. Then at 14 Hz, the .002 DA response is .29 g and the .006 DA response is .5 g; then at 15 Hz, the .002 DA response is .35 g and the .006 DA response is 1.75 g.

F .002 Response .006 Response Multiple

9 Hz .096g .28g 2.9
10 Hz .125g .375g 3
11 Hz .160g .470g 2.9
12 Hz .180g .600g 3.3
13 Hz .240g .740g 3.1
14 Hz .290g .500g 1.7
15 Hz .350g 1.75g 5

Some slight non-linearity can be seen at and below 13 Hz. Now I should mention that the equipment rack is mounted on coil type isolators and the test point in question is mounted on a bracket within the rack, so I can see this situation offers lots of opportunity to introduce nonlinearities of all sorts. But my hardship is understanding the large swing in responses within a 2 Hz bandwidth.

I wonder if Tmoose's plot is somehow explaining my observation? Can you help relate it to the data above?

Thankyou all for your interest in this. It's a good problem.
T

 

[GregLocock]

The response plot does not look unreasonable for a metal assembly, so I'm inclined to think you've got a hardening system, which could be as simple as a small gap. At low amplitude there is no contact, so the system is soft, at high amplitude the part bangs on the stops, so it is effectively stiffer.

Alternatively if those results are in the time sequence they were taken in, maybe that channel crapped out for the last two readings.



Cheers

Greg Locock

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