Random Vibration Spec

[Transient1]

I have a unit that is one large box (12'' by 9'' by 5'') with two smaller boxes(2'' x 4'' x 3'') on top of it side by side. There is one internal shelf in the larger box and PCBs within.

The smaller boxes have been redesigned. Without testing the whole assembled unit, they (the powers that be) want to test the smaller box. I have a random vibration input profile for the the whole unit, but I need to come up with one that has been translated through the unit to the small box mounting feet.

I have the following thoughts:

(1) Ideally, I would have had the original unit tested and measured the PSD off the top cover near the mounting feet of the smaller boxes. This would have been my input.

(2) I only have the transmissibility(3-axis) at the top cover a slight distance away from the mounting feet from a .5G sine survey. I basically know that it is at these resonances the initial random vibration curve will be amplified. What kind of correlation could I use to compare sine transmissibility to random vibe? I know I'd have to make some big assumptions, but a conservative value would be of help.

 

[monkeydog]

That's a tough one.

You have the response accel data from a sine sweep for the lower box top cover away from the mounting feet of one of the 2 boxes that are mounted on the lower box?
And the top box was redesigned - - - how? Structurally? Mass? Anything that could change the response accels on the lower box? I assume you are hard mounting the two boxes on the lower box?

That makes it one system.

It will be interesting if anyone has a good solution.

 

[Transient1]

Two circuit boards were combined into one in the top box, but the mass remains approximately the same (~2% difference). The natural frequency of the PCB that remained is about the same. so basically, the dynamic properties of one small board were lost. However, the natural frequency of the board is estimated to be >1900 Hz and the test is 20Hz to 2000Hz (it rolls off to .04 g^2/HZ at 2000 Hz).

I may be able to get a unit tested at 1/2 the qual level. then I can just double the measured PSD at the top box input for a qual test.