CaptainCrunch
Mechanical
- May 8, 2002
- 31
Hello all,
Read through an archive thread (thread384-6080) on putting components on angled fixtures (oriented in the vector sum direction) to simulate three directional vibration on a single axis shaker.
Currently this is the preffered way of testing components where I work. The benefit is running a single test versus three sequential tests to reproduce vibration data from a tri-axial accelerometer. I understand there is some simplification of the 3-D spectral virbation content, but I personally think this is a minor simplification.
My question is to how to combine three vibration signals, Ax, Ay, and Az, that have similar spectral content. The applications I work on are small engines and the frequency content is strongly related to engine orders, although vibration as a function of frequency in one direction is not a scalar mutliple of the other two signals (i.e. the y-direction 1st order may be twice the z-direction and the y-direction 2nd-order may be half the z-direction 2nd-order).
I am currently using square root sum of squares (SRSS) to combine the 3 direction vibration into a vector sum magnitude. I use the g RMS of the field data signal to find this magnitude. The highest vibration (again in G RMS over time signal) is then amplified to have the vector sum magnitiude.
I was hoping to get some opinions on the strengths and weaknesses of this method, and if the square root sum of squares is the proper way to scale the magnitude of the vector sum vibration.
Can random phasing of the 3-signals be assumed, and would it give something other than SRSS as the vector sum magnitude?
The applications I test occur in a wide variety of operating environments and the goal here is to come up with a good representative, slightly conservative test, and not 100% accuracy (not achievable anyway
)
I have searched the FAQs and previous posts but did not find a detailed thread on this, If you can point me to these I would appreciate it.
Thanks in advance for any help.
George Vandyke
Read through an archive thread (thread384-6080) on putting components on angled fixtures (oriented in the vector sum direction) to simulate three directional vibration on a single axis shaker.
Currently this is the preffered way of testing components where I work. The benefit is running a single test versus three sequential tests to reproduce vibration data from a tri-axial accelerometer. I understand there is some simplification of the 3-D spectral virbation content, but I personally think this is a minor simplification.
My question is to how to combine three vibration signals, Ax, Ay, and Az, that have similar spectral content. The applications I work on are small engines and the frequency content is strongly related to engine orders, although vibration as a function of frequency in one direction is not a scalar mutliple of the other two signals (i.e. the y-direction 1st order may be twice the z-direction and the y-direction 2nd-order may be half the z-direction 2nd-order).
I am currently using square root sum of squares (SRSS) to combine the 3 direction vibration into a vector sum magnitude. I use the g RMS of the field data signal to find this magnitude. The highest vibration (again in G RMS over time signal) is then amplified to have the vector sum magnitiude.
I was hoping to get some opinions on the strengths and weaknesses of this method, and if the square root sum of squares is the proper way to scale the magnitude of the vector sum vibration.
Can random phasing of the 3-signals be assumed, and would it give something other than SRSS as the vector sum magnitude?
The applications I test occur in a wide variety of operating environments and the goal here is to come up with a good representative, slightly conservative test, and not 100% accuracy (not achievable anyway
)I have searched the FAQs and previous posts but did not find a detailed thread on this, If you can point me to these I would appreciate it.
Thanks in advance for any help.
George Vandyke
