It sounds like running the analysis is something you are not asking about. Do you already have a software package selected for running the analysis?

I'm trying to get a grasp of the situation before moving onto the running the analysis part... I don't have much chance to choose a package; it is most likely to be Patran or Workbench.

Hi,

First you must confirm what kind of pcb you are putting under test? weather it shoud go for two dimensional test or three dimentional test as per its requirements. i.e. where is it used, in a stationary platform or in a moving object?

regards

Dilip

Hi Archmagus,

Johan Gouws, that made LinkedIn.com, has profound knowledge in deriving transfer functions from first principles (that contains your resonant frequencies) and analyzing control systems and did significant work involving defense systems. His name, profile and contact details are freely available on the net and, as "mycr4id" personnally knows him, he will gladly direct you in the direction to a sound solution-finding-answer for your question.

From my side: Shock vibration isolation would be necessary between the board and its tray if the resonant frequency of the board and the tray (which is assumed to be ONE with the holding cabinet) are such that the superinposed natural frequencies (of the board and box) could have destructive/damaging effect on the PCB.

A thorough analysis would suggest that you have to determine the natural frequency of the PCB (assumption is that the PCB is not 1/2' x 1/2' with surface mounted micro components) and the cabinet and compare the two. The two frequencies should be multiples of one another to have a neutralizing or controlable resultant.

Next is the source of vibration outside the cabinet. That range of frequencies is the determining factor for scaling your board and cabinet combination to the optimum size to have the smallest resultant effect. Well, if you forgot Laplace and Fourier transforms and related mathematics, it is suggested that you seriously dig into this again if the reliability of your PCB is imperrative for the sustenance of life.

I hope this is helpful.

First of all, thank you all for the wonderful advice, I could use every single letter of it. mycr4id, I'll definitely try to contact Johan Gouws, thanks for pointing me towards him. As to clarify the problem; I'll state it again with more information I received lately...

--> The Situation

There is a PCB board, square in shape (about 330mmx330mm in size), which will be inside a tray, which in turn will be mounted on a vertical panel. The system will be in motion, so vibrations are expected and feared of. The resonance frequencies of the panel are known. The PCB is mounted in its tray via screws from 20 or so points with no isolation; and the walls of the tray directly encircle the PCB. There is a cross-shaped support under the PCB padded by soft material (which I don't know what).

We are more concerned on the Z axis of the board, the direction normal to the board. The other directions are very stiff, and are considered to be out of scope. The modeling can be a simple one; they told me I could model only the board itself and place relatively heavy components (20gr, 50gr and such) on it as point masses.

--> The Goal

* Find coarsely the resonance frequencies of the board and see if they are close to the resonance frequencies of the panel it is mounted on.

* Find the loads the PCB experiences for a given excitation spectrum, and see if it breaks or anything.

--> The constraint

I'm not the design guy for the PCB. I have the drawings and the design, and I have to analyse it. So even if I can have a better idea for the PCB, the only thing I can do is suggest (and I don't think it will be accepted, the design might have been frozen.)

I am actually going to be responsible of the testing operations for it, but I was recently given the task to analyse it too (to make me gain some analysing skills, I suspect. So the better for me :) ) So I don't have much analysing knowledge, but I have a basic grasp on theory and FEM/FEA.

I have access to MSC Patran/Nastran, for that's what the company has. Workbench might be a very far-streched another option.

Thanks again in advance, for any comment you might have.